Generation of a new trisomy 8 mouse model
To investigate the mechanisms by which trisomy 8 affects hematopoiesis in vivo, we attempted to generate a new trisomy 8 mouse model. Since homologous genes in human chromosome 8 are separately located in eight murine chromosomes (Supplementary Fig. 1A), we introduced human chromosome 8 as an extra allele of genes in murine diploid cells (Fig. 1A) [25]. Published datasets on MDS patients revealed that the incidence of trisomy 8 was higher in males than in females, and also that median survival was shorter in males (Supplementary Fig. 2), suggesting that sex differences may affect the pathogenesis of trisomy 8 MDS. Therefore, we transferred human chromosome 8 into murine male ES cells, successfully isolated four trisomy 8 ES clones, and generated one control ES clone that spontaneously lost both the long and short arms, but maintained the centromere region expressing EGFP, hereafter referred to as the WT control (Fig. 1B). Genomic PCR confirmed the absence of human genes on both arms in control cells (Supplementary Fig. 1B). Since trisomy 8 mosaic humans are rarely born and are predisposed to develop hematological malignancies in later life [11, 12], we injected an aggregate mixture of trisomy 8 ES cells and ICR morulae into pseudo-pregnant female mice (Fig. 1C). On E14.5, the number of alive chimeric embryos derived from distinct trisomy 8 ES clones was slightly lower than that from control ES clone (Fig. 1D and Supplementary Fig. 1C, D). The frequency of GFP+ cells in fetal liver (FL) blood cells was lower in trisomy 8 chimera embryos than in control embryos (Fig. 1E). A flow cytometric analysis of FL cells revealed the emergence of lineage-Sca-1+c-Kit+ (LSK) HSPC in both trisomy 8 chimera and control embryos (Fig. 1D), whereas trisomy 8 chimera contained significantly fewer GFP+ cells in CD150+CD48-LSK HSC than control embryos (Fig. 1F). In addition, under culture conditions using OP9 stromal cells, we generated blood cells using the CD31+CD41+ hemogenic endothelial (HE) cells differentiated from parent ES cell and these ES clones (Fig. 1G). While parent ES cells differentiated into blood cells, trisomy 8 ES clones differentiated into HSPC, myeloid cells, and B lymphoid cells at lower frequencies than control ES clone (Fig. 1H). Therefore, we successfully generated a new trisomy 8 chimera mouse model that was able to generate blood cells at lower frequencies under both in vivo and in vitro conditions.
Fig. 1: Generation of a trisomy 8 mouse model.
A Schematic illustration showing microcell-mediated human chromosome 8 transfer to generate trisomy 8 mouse ES cells. B FISH analysis of ES clones (WT control and trisomy 8 #7, #11, #23 and #30) showing chromosome 8 stained by digoxigenin-labeled human COT-1 probe with anti-digoxigenin antibody conjugated rhodamine (Red) and EGFP gene-labeled biotin with anti-biotin antibody conjugated fluorescein isothiocyanate (Green). C Schematic illustration showing trisomy 8 chimera mice generated by the aggregation of trisomy 8 ES cells and ICR morula embryos. D Representative pictures of E14.5 embryos and flow cytometric plots of GFP+ LSK cells of WT control and trisomy 8 FL cells (clones 7 and 23). E Frequencies of GFP+ cells in the E14.5 FL blood cells of WT (n = 11) and trisomy 8 (n = 4–8) chimera mice. Data were combined from four to eleven independent aggregate injection experiments. F Frequencies of GFP+ cells in the E14.5 FL HSC of WT (n = 8) and trisomy 8 chimera mice (n = 13). Data were combined from three independent experiments. G Schematic illustration showing the in vitro differentiation protocols of HSPC, myeloid cells, and B cells from ES cells through CD31+CD41+ hemogenic endothelium (HE) cells. H Relative frequencies of GFP+ trisomy 8 cells in HSPC, granulocytes, erythrocytes, and B cells (n = 3) after in vitro differentiation. The data were representative of three independent experiments. I Principal component analysis of expression levels of all genes in WT and trisomy 8 ES cells, HE cells, and FL HSC (number of clones; n = 2). Data information: E, F, H Data are presented as mean ± SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ns, not significant. P values were calculated by Student’s t test unless otherwise indicated.
To understand how trisomy 8 affected the transcriptome at different stages of development, we performed RNA sequencing of ES cells, HE cells, and fetal HSC. By using a combined reference of human chromosome 8 with the murine genome (mm10), principal component analyses (PCA) of RNA sequencing revealed the overall similar transcriptional profiles between WT and trisomy 8 cells during the development (Fig. 1I). In addition, human genes in chromosome 8 and their homologous murine genes showed a linear correlation of expression levels in trisomy 8 cells (Supplementary Fig. 1E), suggesting that the human chromosome was transcriptionally regulated in murine cells in a similar manner to homologous regions in murine chromosomes. Despite the overall similarity in the transcriptome, a hierarchical clustering revealed that trisomy 8 cells from different stages clustered independently (Supplementary Fig. 1F), suggesting that trisomy 8 induced transcriptional changes in blood cells.
Trisomy 8 impaired the self-renewal of HSC and failed to develop MDS in mice
To determine whether trisomy 8 affected adult hematopoiesis in a cell-intrinsic manner, we transplanted trisomy 8 FL blood cells into WT mice and assessed the hematopoietic function. We transplanted 5 × 105 GFP+ FL cells isolated from E14.5 embryos into lethally irradiated Ly5.1+ mice together with 2 × 105 WT BM cells (Fig. 2A). We used two trisomy 8 clones (clones 7 and 23) and the WT control (clone 21) for transplantation (Fig. 1D). We found that the frequencies of GFP+ cells in myeloid and lymphoid cells in PB and HSC and multipotent progenitor cells (MPP) in BM three months after transplantation were similar in trisomy 8 mice and control mice (Fig. 2B, C). We then investigated whether trisomy 8 HSC maintained their self-renewal capacity by performing a competitive secondary transplantation of LSK HSPC isolated from the primary recipients of trisomy 8 cells. The frequencies of GFP+ cells in both mature cells in PB and HSC and MPP in BM were lower in trisomy 8 mice than in control WT mice (Fig. 2D, E). In order to examine the possibility of human genes on the extra chromosome 8 to express xenoantigens, which may eliminate trisomy 8 cells in immunocompetent mice, we transplanted trisomy 8 HSPC into NSG immunodeficient mice together with Ly5.1+ WT BM cells (Supplementary Fig. 3A). NSG recipients showed that the frequencies of GFP+ cells in HSC and MPP in the BM and mature myeloid cells in the PB were markedly lower in trisomy 8 mice than in control WT mice (Supplementary Fig. 3B–D). Therefore, trisomy 8 impaired the self-renewal capacity of HSC in a cell-intrinsic manner.
Fig. 2: Trisomy 8 impaired the self-renewal capacity of HSC and the hematopoiesis in mice.
A Schematic illustration showing the transplantation of WT control and trisomy 8 FL cells (clones 7 and 23 in Fig. 1D) with Ly5.1+ wild-type BM cells into Ly5.1+ wild-type recipients, the HSPC of which were then serially transplanted into secondary recipients three months after transplantation. B Frequencies of GFP+ cells in peripheral blood of WT control and trisomy 8 primary recipient mice two to three months after transplantation (n = 5–9). C Frequencies of GFP+ cells in HSC, MPP2, MPP3, and MPP4 in the BM of WT control and trisomy 8 primary recipient mice three months after transplantation (n = 5–7). D Frequencies of GFP+ cells in peripheral blood of WT control and trisomy 8 secondary recipient mice four months after serial HSPC transplantation (n = 5–24). E) Frequencies of GFP+ cells in HSC, MPP2, MPP3, and MPP4 in WT control and trisomy 8 secondary recipient mice four months after transplantation (n = 9–18). Data information. B–E Data are presented as mean ± SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ns, not significant. P values were calculated by Student’s t test unless otherwise indicated. Data were combined from two or three independent experiments.
We next investigated whether trisomy 8 cells were sufficient to induce malignant disease in the primary recipients. While trisomy 8 mice mostly lost GFP+ blood cells in PB 3 to 4 months after primary transplantation, 3 out of 148 transplanted mice developed lethal disease, showing anemia and dysplastic blood cells in the maximum observation period of one year (Supplementary Fig. 4A–C). To assess the proliferative capacity of trisomy 8 disease cells in these mice, we performed the secondary transplantation of 1 × 106 BM cells isolated from two disease mice into sub-lethally irradiated WT recipients (Supplementary Fig. 4D). Consistent with the impaired self-renewal capacity of trisomy 8 HSC (Fig. 2D, E), trisomy 8 disease cells maintained the extra chromosome 8 and lost their repopulation capacity (Supplementary Fig. 4E–H), resulting in a few GFP+ cells in the PB of secondary recipients (Supplementary Fig. 4I), suggesting that trisomy 8 alone was not sufficient to develop full-blown MDS in vivo, although trisomy 8 induced transcriptional changes in blood cells and rarely developed lethal blood diseases in the primary recipients.
Trisomy 8 HSC dysregulated the transcriptional programs by stem-cell regulators
To elucidate the molecular mechanisms by which trisomy 8 impaired the self-renewal capacity of HSC and hematopoiesis, we performed RNA sequencing on HSC and granulocyte/monocyte progenitor cells (GMP) isolated from the primary recipients two months after transplantation. The expression of 261 genes was higher, and that of 20 genes was lower in trisomy 8 HSC than in control HSC, while the expression of 196 genes was higher and that of 41 genes was lower in trisomy 8 GMP than in control GMP (Fig. 3A, B and Supplementary Dataset 1). A hierarchical clustering map showed that the expression levels of genes in human chromosome 8 were higher in trisomy 8 HSC than in control HSC, and were not shared with up-regulated genes in ES cells and HE cells harboring trisomy 8 (Fig. 3C), supporting that transduced human genes followed tissue-specific transcriptional regulation (Fig. 1I). A gene set enrichment analysis (GSEA) revealed that in comparison with control HSC, trisomy 8 HSC (clones 7, 23, and 30) showed positive enrichments in both stem cell and myeloid signature genes (Fig. 3D). In addition, trisomy 8 HSC positively enriched the canonical target genes of stem cell regulators, such as Runx1, PU.1, Gata2, and Mecom TFs, and that of polycomb repressive complex 2 (PRC2), in which Ezh2 is critical for differentiation via the modulation of histone H3K27me3 modifications [33, 34] (Fig. 3D). In contrast, in comparison with control GMP, trisomy 8 GMP negatively enriched the target genes of PU.1 TF (Fig. 3D), suggesting that trisomy 8 activated both stem cell and myeloid signature genes in HSC, but repressed the transcription of myeloid signature genes at the GMP stage. Although the MYC gene is located in human chromosome 8, the canonical target genes of Myc were not activated in trisomy 8 HSC and GMP (Fig. 3D).
Fig. 3: Trisomy 8 HSC dysregulated transcription of stem cell and myeloid signature genes.
A, B Volcano plots showing differentially expressed murine genes in trisomy 8 (clone #7) HSC and GMP relative to those in WT HSC and GMP (number of samples: n = 2), respectively. The data were representative of two independent experiments. C Heatmap showing z-scores for the expression values of genes on human chromosome 8 in WT and trisomy 8 ES cells, HE cells, and adult HSC, forming three clusters based on K-means clustering. D GSEA of target genes of stem cell regulators and inflammatory pathways (linked in GSE22178, GSE50537, GSE237192 and https://doi.org/10.1371/journal.pone.0067134.s004) in trisomy 8 HSC and GMP relative to those in WT HSC and GMP (n = 2), respectively. Asterisks show *FDR < 0.1, **FDR < 0.01, and ***FDR < 0.001 with p < 0.05. E Expression levels of Runx1 in WT and trisomy 8 HSC and GMP (clone #7 and #23) examined by RamDA sequencing (n = 3). F Expression levels of the Runx1 protein in bone marrow mononuclear cells of 2-month-old WT and trisomy 8 primary mice. Actin was used as the loading control. The data are representative of two independent experiments. G Heatmap showing z-scores for the expression values of Runx1 target genes in WT and trisomy 8 HSC (n = 2), forming three clusters based on K-means clustering. H GO analysis of differentially expressed genes in three clusters in (F). P-values were calculated using the gene ontology functions of HOMER software. Data information: A, B Red dots and blue dots representing significantly changed genes with an expression fold-change >1.5 and P value < 0.01.
We found a mild elevation of Mecom expression in trisomy 8 HSC among those dysregulated stem cell regulators (Supplementary Fig. 5A). Among those dysregulated TFs, we focused on Runx1 TF because loss-of-function mutations in RUNX1 have been most frequently found in patients with trisomy 8 MDS [35, 36]. While trisomy 8 HSC increased the expression of Runx1-target genes identified in murine progenitor cells (Fig. 3D), trisomy 8 HSC and GMP showed similar Runx1 expression levels to those in WT HSC and GMP (Fig. 3E). We also found that trisomy 8 mononuclear cells (MNC) showed similar Runx1 protein expression levels to those in WT MNC (Fig. 3F), suggesting that the expression change of Runx1 may not account for the dysregulated expression of Runx1-target genes. In comparison with control HSC, among canonical Runx1-target genes [37], we found that down-regulated genes in trisomy 8 HSC were enriched in cluster 1 and up-regulated genes were enriched in clusters 2 and 3 (Fig. 3G). Gene ontology (GO) analyses revealed that cluster 1 genes were involved in biosynthesis and metabolism, while cluster 2 and 3 genes were involved in immunity, inflammation, and differentiation pathways (Fig. 3H). These results suggest that trisomy 8 HSC induced the activation of inflammatory genes and dysregulated the transcriptional programs by stem cell TFs, including Runx1, which appeared to impair the self-renewal function of HSC and hematopoiesis.
Trisomy 8 altered the chromatin structures in genome-wide and combinatory manners
Chromatin accessibility is a hallmark of active elements that regulate the transcription of genes in stem cells for their differentiation [38]. Based on the dysregulated programs for the differentiation by TFs, such as Runx1 and PU.1, in trisomy 8 HSC and GMP cells (Fig. 3D), we performed ATAC sequencing on WT and trisomy 8 HSC. ATAC sequencing revealed that trisomy 8 induced open chromatin and closed chromatin in HSC (3829 and 14371 out of 75685 peaks) (Fig. 4A). TF-binding motif enrichment analyses revealed that in comparison with WT HSC, trisomy 8 HSC enriched binding motifs of the RUNX family in both open and closed chromatin (Fig. 4B and Supplementary Dataset 2), which may have contributed to the activation or repression of individual Runx1-target genes (Fig. 3G). In addition, trisomy 8 HSC enriched binding motifs of SpiB/PU.1 related and CEBPE in closed chromatin, which may have impeded the transcription of PU.1-target genes in GMP (Fig. 3D). Therefore, trisomy 8 dysregulated chromatin accessibility at the binding regions of Runx1 and myeloid TFs in HSC, leading to the repressed expression of myeloid genes in GMP cells.
Fig. 4: Trisomy 8 changed chromatin structures in HSC.
A Scatter plot showing chromatin accessibility between trisomy 8 HSC (Y axis) and WT HSC (X axis) out of 75685 ATAC peaks. B Ranks of enriched TF-binding motifs in open and closed chromatin in trisomy 8 HSC relative to WT HSC. The data are representative of two independent experiments. C Scatter plots showing PC1 values in the genomic regions of intra-chromosomal structures between trisomy 8 LSK HSPC and WT HSPC (n = 2). D Normalized differentially regulated PC1 regions in each chromosome in trisomy 8 HSPC. E GSEA of corresponding genes in up- or down-regulated PC1 regions in trisomy 8 HSPC relative to WT HSPC. F Histograms showing frequencies of up-regulated PC1 regions in trisomy 8 HSPC relative to WT HSPC in open and closed chromatin in trisomy 8 HSC (62.31% versus 55.43%, p = 1.43 × 10–11).
Based on the alterations in chromatin accessibility and transcriptional programs mediated by TFs in trisomy 8 HSC, we investigated whether trisomy 8 changed the conformation of chromatins, since the conformation of chromatins facilitates the process of transcription [39]. To achieve this, we performed Hi-C sequencing on HSPC isolated from primary transplanted mice two months after transplantation. Nuclear compartments were captured by the value of the first principal component (PC1) in a principal component analysis of the Hi-C contact map. In comparison with control HSPC, trisomy 8 HSPC showed 1707 genomic regions with increased PC1 values and 2760 genomic regions with decreased PC1 values in intra-chromosomal structures (Fig. 4C). These chromatin conformation changes were observed in all chromosomes to similar degrees, but were lower in chromosome 9 (Fig. 4D). GSEA revealed that up-regulated PC1 regions correlated with the higher expression of the corresponding genes in trisomy 8 HSC than in WT HSC (Fig. 4E and Supplementary Dataset 3), suggesting that trisomy 8 activated chromatin compartments, accompanied with the increased expression of genes in HSC, consistent with the previous finding [39]. To determine whether those changes in the chromatin accessibility and chromatin conformation in trisomy 8 cooperated to dysregulate the transcription of genes, we analyzed Hi-C and ATAC sequencing data in combination. We found that trisomy 8 HSPC increased PC1 values in open chromatins more than closed chromatins, compared to control HSPC (Fig. 4F), suggesting that the formation of open chromatins may be facilitated by active chromatin compartments in trisomy 8 HSC. Therefore, trisomy 8 altered the chromatin structures in HSPC in genome-wide and combinatory manners, which may lead to the altered transcriptional programs for hematopoiesis.
Trisomy 8 up-regulated the transcription of Y chromosome genes and facilitated the activation of Uty-target genes
We attempted to determine which altered chromosome regions contributed to the impaired hematopoiesis by trisomy 8. Among the 73 genes with increased expression in active compartments (Fig. 4E), we found that only the Y chromosome containing Kdm5d, Eif2s3y, Uty/Kdm6c, and Ddx3y genes showed a significant enrichment by the DAVID analysis (p = 0.025). The Hi-C sequencing revealed that a centromere-proximal region in the Y chromosome containing those genes showed higher contact levels forming a topologically associating domain (TAD) and a more active compartment than other chromosome regions, relative to WT HSPC (Fig. 5A, B). As the expression levels of the Kdm5d, Eif2s3y, Uty/Kdm6c, and Ddx3y genes were increased in the active region (Fig. 5C), Q-RT-PCR confirmed that trisomy 8 HSC variably increased the expression levels of Uty and Kdm5d genes, relative to WT HSC (Fig. 5D). In addition, we performed Hi-C of SKK-1 human leukemic cells presenting 47, XY + 8 originated from a male patient with MDS [40] and THP-1 non-trisomy 8 leukemic cells. The Hi-C analysis also revealed a significantly stronger TAD formation containing the UTY gene (Supplementary Fig. 6A) in SKK-1 cells than in THP-1 cells, suggesting that both murine and human trisomy 8 blood cells induced an active TAD formation and increased expression of the UTY gene.
Fig. 5: Trisomy 8 up-regulated the transcription of the Y chromosome genes.
A Hi-C matrices of a region in the Y chromosome showing higher contact levels in trisomy 8 HSPC (clone #7) than WT HSPC. B Genomic view of a region in the Y chromosome showing the active compartment in trisomy 8 HSPC (clone #7 and #23) relative to WT HSPC. C Expression levels of Kdm5d, Eif2s3y, Uty, and Ddx3y in the active compartment in the Y chromosome in WT and trisomy 8 HSC (clone #7 and #23) examined by RamDA sequencing (n = 2). D Kdm5d and Uty mRNA expression levels in WT and trisomy 8 HSC examined by Q-RT-PCR (n = 4–6). E Genomic view of the proximal regions of SPI1/PU.1 genes showing UTY-binding peaks detected in SKK-1 cells and RUNX1-binding peaks, H3K27ac peaks, and H3K27me3 peaks detected in human CD34+ cells (linked in GSE45144, GSM772894, and GSM621431). F Rank of enriched TF-binding motifs in UTY-binding peaks in SKK-1 cells. G Venn diagram showing significant overlaps between UTY-target genes and Runx1-target genes or PRC2-target genes identified in murine HSPC (linked in GSE22178 and GSE50537). H GO analysis of 259 UTY/PRC2/Runx1-target genes in (G). P-values were calculated using the gene ontology functions of HOMER software. I GSEA of UTY-target genes and UTY/PRC2/Runx1-target genes in trisomy 8 HSC relative to those in WT HSC in mice and in MDS HSPC with trisomy 8 relative to those with disomy 8 in patients (linked in GSE58831). Asterisks show *FDR < 0.1 and **FDR < 0.01 with p < 0.05. J Q-RT-PCR showing a reduction of UTY mRNA in distinct shRNA-directed UTY-transduced SKK-1 cells, compared to shRNA-directed Luciferase-transduced cells (n = 3–4). K Reduced growth of UTY knocked-down SKK-1 cells in a liquid culture (n = 3). L Morphology of SKK-1 cells in Fig. 6K observed by May-Giemsa staining. Scale bars show 10 µm. Data information. C, D, J, K Data are presented as mean ± SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ns, not significant. P values were calculated by Student’s t test unless otherwise indicated. Two-way ANOVA was used for (K). J–L Data were representative of three independent experiments.
Based on that, trisomy 8 HSC did not change the expression levels of PRC2 components but positively enriched the target genes of PRC2 (Fig. 3D and Supplementary Fig. 5A). Among the up-regulated genes in the Y chromosome, we focused on Uty because human UTY is known to demethylate histone H3K27me3 in vivo despite its weak enzymatic activity compared to UTX, its homologue on the X chromosome [41]. We found that the expression of UTY in HSPCs was higher in a subset of male MDS patients than in healthy males, whereas UTX was not activated in murine trisomy 8 HSC or MDS HSPCs in male and female patients (Supplementary Fig. 7A, B). To identify the direct targets of UTY, we performed UTY ChIP sequencing on SKK-1 trisomy 8 cells and found that UTY directly bound to the promoter and enhancer regions of stem cell genes, including RUNX1 and SPI1/PU.1 (Fig. 5G and Supplementary Fig. 6B; Supplementary Dataset 4). UTY-binding regions were enriched by active histone modifications, such as H3K4me1, H3K4me3, and H3K27ac, and the RUNX1 peak was higher than the H3K27me3 repressive mark, which were identified in human CD34+ cells and monocytes using published datasets (Fig. 5E and Supplementary Fig. 6C). A motif enrichment analysis of UTY-binding peaks showed significant enrichments in the binding motifs of RUNX, GATA, and ETS family TFs (Fig. 5F). In addition, UTY-binding annotated genes markedly overlapped with the Runx1-binding genes and canonical PRC2-target genes identified in murine HSPCs [37, 42] (Fig. 5G). By utilizing these UTY- and Runx1-binding gene sets (Supplementary Dataset 4), we found that the expression of both UTY-target and common target genes, such as MECOM/EVI1, which are involved in stem cell regulation and leukemia-related pathways (Fig. 5H and Supplementary Dataset 5), was significantly higher in trisomy 8 HSC than in WT HSC (Fig. 5I). By analyzing published datasets of patients with MDS, the expression levels of the common target genes in HSPC were higher in trisomy 8 MDS than in disomy 8 MDS (Fig. 5I), suggesting that the Uty gene activated the expression of PRC2-target genes, leading to the impaired hematopoiesis and the leukemogenesis by trisomy 8.
Finally, in order to investigate the role of UTY in trisomy 8 MDS, we knocked down the expression of the UTY gene in SKK-1 cells using three distinct shRNA vectors and confirmed lower expression levels in these cells than control-vector transduced cells (Fig. 5J). In a culture condition, the knockdown of UTY markedly repressed the growth of SKK-1 cells despite the expression of UTX (Fig. 5K and Supplementary Fig. 7C) and showed a similar immature morphology to the control shLuciferase-transduced cells (Fig. 5L), suggesting that UTY drove the proliferation of trisomy 8 MDS cells, which may not be compensated for by UTX. Interestingly, the knockdown of UTX mildly repressed the growth of SKK-1 cells in a culture condition, compared to control shLuciferase-transduced cells (Supplementary Fig.7D, E). Therefore, trisomy 8 up-regulated the expression of Y chromosome genes, including Uty, thereby facilitating the activation of its target genes, which may drive the transformation and the proliferation of MDS cells.
Deletion of Runx1 facilitated the activation of Uty-target genes and drove a pre-MDS state in trisomy 8 HSC
Trisomy 8 induced chromatin conformation changes and increased the transcription of Runx1-target genes involved in inflammation, proliferation, and differentiation in HSC. Since trisomy 8 was insufficient to develop full-blown MDS in mice and the RUNX1 gene is most frequently mutated in MDS patients with trisomy 8 [35, 36], we investigated whether the deletion of the Runx1 gene completely transformed trisomy 8 HSC in mice. By using a CRISPR/Cas9 system to delete the Runx1 gene [30], we transduced sgRNA-directed Runx1 into 1000 HSPCs isolated from primary transplant recipient mice and transplanted them into lethally irradiated mice together with 2×105 WT BM cells (Fig. 6A). We confirmed a reduction in Runx1 protein levels in BM blood cells after transplantation (Fig. 6B). Four months after transplantation, sgRNA-Runx1 transduced WT mice showed a similar donor chimerism in whole cells in PB to that in sgRNA-control transduced WT mice, but with a slightly higher number of myeloid cells (Fig. 6C), which is consistent with the mild myeloproliferative phenotype in Runx1 mice [43]. In contrast, control trisomy 8 mice all showed markedly lower donor chimerism in both the myeloid and lymphoid fractions of PB, while the deletion of Runx1 increased donor chimerism in myeloid and B cells in PB in 4 out of 10 trisomy 8 mice (Fig. 6D). Control trisomy 8 mice showed very low chimerism in HSC and MPP in BM, while the deletion of Runx1 induced a detectable chimerism in HSC and increased that in MPP in the secondary transplant recipient mice (Fig. 6E); however, we did not find evident malignant diseases in compound mutant mice in the one-year observation period, suggesting a pre-malignant state.
Fig. 6: Deletion of Runx1 induced a pre-malignant state in trisomy 8 HSC.
A Schematic illustration showing the competitive transplantation of Runx1-deleted WT control and trisomy 8 HSPCs isolated from primary transplanted mice with Ly5.1+ WT BM cells into lethally-irradiated WT recipient mice. B Expression level of the Runx1 protein in BM blood cells isolated from the depicted genotype mice. Actin was used as the loading control. C, D Frequencies of GFP+ WT and trisomy 8 cells transduced with sgRNA-control (in black line) and shRNA-Runx1 (in red line) in myeloid cells and B cells in the PB of recipient mice (n = 8–9). E Frequencies of GFP+ cells in HSC and MPP2, MPP3, and MPP4 cells in the BM of recipient mice (n = 6–7). F Principal component analysis of expression levels of all genes in HSC isolated from the depicted genotype mice (n = 2). G Heatmap showing the z-scores for expression values in 339 differentially expressed genes between sgRNA-control-transduced trisomy 8 HSC and WT HSC, forming four clusters based on K-means clustering. H GO analysis of genes in clusters 1 and 4 in (G). P-values were calculated using the gene ontology functions of HOMER software. I GSEA of stem cell and myeloid signature genes and UTY/PRC2/Runx1-target genes in sgRNA-Runx1-transduced HSC relative to sgRNA-control-transduced HSC. Asterisks show *FDR < 0.1, **FDR < 0.01, and ***FDR < 0.001 with p < 0.05. Data information. C–E Data are presented as mean ± SEM. Data were combined from two independent experiments.
To elucidate the molecular mechanisms by which the deletion of Runx1 mitigated the impaired self-renewal capacity of trisomy 8 HSC, we performed RNA sequencing on HSC isolated from transplanted mice one year after transplantation. Principal component analyses revealed that sgRNA-Runx1 transduced trisomy 8 HSC was located between WT HSC and control trisomy 8 HSC (Fig. 6F). Among differentially expressed genes from control WT HSC, we found that up-regulated genes in control trisomy 8 HSC were enriched in cluster 4, in which the expression of genes was markedly repressed in sgRNA-Runx1 transduced trisomy 8 HSC (Fig. 6G). GO analyses revealed that cluster 4 genes were involved in cytokines, complements, and differentiation, while cluster 1 genes were involved in T cell-related signaling (Fig. 6H and Supplementary Dataset 6), suggesting that the Runx1 deletion repressed the expression of a subset of inflammatory genes activated in trisomy 8 HSC. Furthermore, GSEA revealed that the expression of stem cell signature genes and UTY-target genes was higher in sgRNA-Runx1-trisomy 8 HSC than in control trisomy 8 HSC (Fig. 6I). Therefore, the deletion of the Runx1 gene attenuated the activation of a subset of inflammatory genes and facilitated the activation of UTY-target genes in trisomy 8 HSC, which appeared to provide the competitive fitness of the mutant clone in BM and induce a pre-malignant state.
