UTX maintains the functional integrity of the murine hematopoietic system by globally regulating aging-associated genes

Covalent modifications of histone tails, such as methylation, acetylation, and ubiquitination play essential roles in appropriate cell fate decisions. Trimethylated H3K27 (H3K27me3) is regarded as a repressive histone mark that functions in gene silencing.¹  H3K27 methylation is mediated by polycomb repressive complex 2 (PRC2), which comprises the catalytic subunit EZH2 and at least 2 noncatalytic subunits, EED and SUZ12.²  On the other hand, demethylation of H3K27 is regulated by 2 distinct enzymes: ubiquitously transcribed tetratricopeptide repeat, chromosome X (UTX), also known as KDM6A, and Jumonji-C (JmjC) domain-containing protein-3 (JMJD3), also known as KDM6B,^3,4  both of which contain the JmjC domain, which exerts the demethylase activity.^3,4  However, unlike JMJD3, UTX possesses the tetratricopeptide repeat (TPR) domain that mediates protein-protein interactions and functions as a component of COMPASS (complex of proteins associated with Set 1)-like and SWI/SNF complexes.^5-8 

UTX is an X-chromosome–specific enzyme, and the male counterpart is on the Y chromosome, named ubiquitously transcribed tetratricopeptide repeat gene, Y chromosome (UTY), also known as KDM6C.^3,4  UTX and UTY have a high structural similarity, including in the TPR and JmjC domains⁴ ; however, UTY possesses very low demethylase activity.^4,9  Knockout studies for UTX demonstrated that Utx-deficient female mice died in utero but Utx-deficient male mice possessing the Uty gene survived embryogenesis, indicating that UTY can compensate for the function of UTX during development.^10,11 

UTX contributes to various biological processes: it regulates body patterning by binding to Hoxb1 promoter,^3,12  promotes myogenesis by demethylating muscle-specific genes,¹³  determines cell fate by controlling the retinoblastoma pathway,¹⁴  supports cardiac development during embryogenesis,¹⁵  regulates hematopoietic cell migration,¹⁶  plays essential roles in embryonic stem cell differentiation,^17,18  and coordinates steroid hormone-mediated autophagy.¹⁹  In addition, UTX plays demethylase-independent roles in T-box factor–mediated gene expression, mesoderm differentiation of embryonic stem cells, and prenatal development.^8,10,20-22 

UTX dysfunction has been identified in human diseases. Somatic inactivating mutations of Utx were identified in various human malignancies.^23-30  In addition, functional loss of Utx was reported to contribute to drug resistance and disease relapse.^31,32  These findings indicated that UTX functions as a cell-fate determinant, as well as a tumor suppressor; however, its role(s) in the hematopoietic system remains to be fully understood. In this study, by generating and analyzing Utx-deficient mice, we found that UTX plays essential roles in the functional integrity of hematopoietic stem-progenitor cells (HSPCs) by globally regulating aging-associated genes.

The procedures for construction of the targeting vector and generation of Utx conditional knockout mice are described in the supplemental Information, available on the Blood Web site. All the animal experiments were performed in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of Hiroshima University Animal Research Committee (permission no. 25-107) and Tokyo Women’s Medical University (permission no. GE 19-066).

Mouse survival curves were constructed by using the Kaplan–Meier methodology and compared by using the log-rank test, using GraphPad Prism software. Other statistical analyses were performed using the Student t test, unless otherwise stated.

A complete and detailed description of the methods is provided in the supplemental Methods.

To conditionally abrogate UTX function, we generated mice in which exons 11 and 12 of the Utx gene were flanked by 2 loxP sites and crossed them with tamoxifen-inducible ERT2Cre⁺ mice (supplemental Figure 1A). Utx is an X-chromosome–specific gene with a male counterpart, Uty; hence, the results of crossing were sex dependent, producing Utx^flox/flox, ERT2Cre⁺ females and Utx^flox/Uty, ERT2Cre⁺ males. Quantitative polymerase chain reaction (PCR) and western blot analysis of bone marrow (BM) cells showed that, in tamoxifen-treated females, Utx mRNA and UTX protein were almost completely absent in Utx^flox/flox, ERT2Cre⁺ (Utx^Δ/Δ) mice compared with Utx^flox/flox, ERT2Cre^– (Utx^+/+) mice, whereas in tamoxifen-treated males, Uty messenger RNA and UTY protein were comparable in Utx^flox/Uty, ERT2Cre^– (Utx^+/Uty) and Utx^flox/Uty, ERT2Cre⁺ (Utx^Δ/Uty) mice (supplemental Figure 1B; primer sequences for quantitative PCR are listed in supplemental Table 1). In addition, immunofluorescent staining of long-term hematopoietic stem cells (LT-HSCs) [for surface markers of hematopoietic stem-progenitor cell (HSPC) subfractions, see supplemental Table 2] of Utx^+/+, Utx^+/Δ, and Utx^Δ/Δ females and of Utx^+/Uty and Utx^Δ/Uty males exhibited graded and significant increases in H3K27me3 levels, along with Utx deficiency, confirming that UTX functions as a demethylase for H3K27 in primitive HSCs (supplemental Figure 1C).

Then, we analyzed hematopoietic parameters in the peripheral blood (PB), BM, and spleen in Utx^+/+, Utx^+/Δ, and Utx^Δ/Δ females and also in Utx^+/Uty and Utx^Δ/Uty males. Although no obvious changes were observed in Utx^+/+ and Utx^+/Δ mice, Utx^Δ/Δ mice exhibited a significant increase in white blood cell (WBC) count, mainly of myeloid-lineage (Mac1⁺). A significant decrease was noted in platelets in the PB (Figure 1A, top); significant increases in LT-HSCs, short-term HSCs (ST-HSCs), and Mac1⁺ cells and significant decreases in T-lineage (Thy1.2⁺) and erythroid-lineage (Ter119⁺) cells in the BM (Figure 1A, middle); significant increases in LSK cells (LSKs), LT-HSCs, common myeloid progenitor cells, granulocyte-monocyte progenitor cells, and Mac1⁺ cells and a significant decrease in Thy1.2⁺ cells in the spleen (Figure 1A, bottom). Compared with Utx^Δ/Δ mice, the changes in Utx^Δ/Uty mice were less evident, showing an increase in WBC count in the PB, an increase in common myeloid progenitor cells, and a decrease in Thy1.2⁺ cells in the BM, and increases in LT-HSCs, granulocyte-monocyte progenitor cells, and Mac1⁺ cells in the spleen (supplemental Figure 2A).

In addition, morphological abnormalities were observed in Utx^Δ/Δ mice (Figure 1B), which included WBCs with abnormal nuclei, including pseudo Pelger-Huët anomalies (Figure 1B1-2), neutrophils with hypersegmentation (Figure 1B3), giant platelets (Figure 1B4), erythrocytes with Howell-Jolly bodies (Figure 1B5), apoptotic cells (Figure 1B6), myeloid progenitor cells with abnormal nuclei (Figure 1B7,9-10), and micromegakaryocytes (Figure 1B8-9). These results indicate that acquired deficiency of UTX induces abnormal hematopoietic parameters with trilineage dysplasia, reminiscent of human myelodysplastic syndrome (MDS), as previously described.^16,33,34 

Macroscopically, the femurs of Utx^Δ/Δ mice were pale and the spleens were enlarged, compared with those of Utx^+/+ mice (supplemental Figure 3A, left). Pathological analyses revealed that the Utx^Δ/Δ BM contained markedly reduced erythroid cells and megakaryocytes and the Utx^Δ/Δ spleen contained an increased number of myeloid cells and megakaryocytes (supplemental Figure 3A, right). These phenotypes are characteristic of extramedullary hematopoiesis and suggest the transition of HSPCs from the BM to the spleen through the PB. To address this possibility, we subjected HSPCs in the PB to flow cytometry and colony formation assays. The results exhibited significant increases in c-kit⁺ and LSK cells in the Utx^Δ/Δ PB (supplemental Figure 3B), which generated a considerable number of trilineage hematopoietic cell colonies containing dysplastic cells, as observed in the Utx^Δ/Δ PB and BM (supplemental Figure 3C-D), indicating circulating HSPCs in Utx^Δ/Δ mice.

Analysis of Utx and Uty expression in HSPC subfractions showed similar patterns, being highest in the LT-HSC subfraction and decreasing as the cells differentiated (Figure 2A; supplemental Figure 2B). To investigate the repopulating ability of Utx^+/+ and Utx^Δ/Δ LSKs, competitive repopulation assays were performed (Figure 2B, top). In the first bone marrow transplantation (BMT), the total PB chimerism of Utx^Δ/Δ cells was similar to that of Utx^+/+ cells at 1 month, but became significantly lower at 2 and 3 months (Figure 2B, middle left). In the second BMT, the reduced PB chimerism of Utx^Δ/Δ cells was more pronounced, and almost no contribution was detected (Figure 2B, middle right). Analysis of the lineage-positive cells in the first BMT showed that the percentage of Utx^Δ/Δ cells was significantly reduced in all myeloid, B-cell (B220⁺), and T-cell (Thy1.2⁺) lineages (Figure 2B, bottom), indicating the severely impaired repopulation potential of Utx^Δ/Δ HSPCs. We also performed a rescue experiment in which Utx^+/+ LSKs were transduced with an empty vector (EV) (Utx^+/++EV), Utx^Δ/Δ cells with empty vector (Utx^Δ/Δ+EV), or Utx^Δ/Δ cells with Utx-expressing vector (Utx^Δ/Δ+Utx) were subjected to competitive reconstitution assays. Although the results were not statistically significant, Utx^Δ/Δ+Utx cells exhibited an enhanced engraftment capacity compared with Utx^Δ/Δ+EV cells, indicating that the impaired repopulating ability of Utx^Δ/Δ HSPCs was a direct effect of Utx deficiency (Figure 2C).

The efficiency of BM reconstitution correlates strongly with the homing ability of HSPCs. Thus, we performed homing assays and also measured the expression of CXCR4, a key chemokine receptor for HSC homing.³⁵  Consistent with the results of a previous report,¹⁶ Utx^Δ/Δ LSKs exhibited a significantly reduced homing ability, despite comparable expression of Cxcr4 (Figure 2D).

Utx is frequently mutated in human cancers^23,28  and is a tumor suppressor in mice.^33,34,36  To investigate the effect of UTX deficiency on leukemogenesis, we performed retrovirus-insertional mutagenesis using MOL4070A, a retrovirus capable of inducing leukemia.³⁷ Utx-expressing (Utx^+/+ and Utx^+/Uty) and Utx-deficient (Utx^Δ/Δ and Utx^Δ/Uty) mice were infected with MOL4070A as neonates, and their hematopoietic parameters were analyzed (Figure 3A, top).

During a 250-day observation period, although no disease was observed in Utx^+/++MOL4070A and Utx^+/Uty+MOL4070A mice, all the Utx^Δ/Δ+MOL4070A and Utx^Δ/Uty+MOL4070A littermates developed leukemia; the former died within 150 days and the latter within 250 days (Figure 3A, middle; supplemental Table 3). To determine the lineage of the leukemias, enlarged spleens containing infiltrating leukemic cells, were subjected to flow cytometric and gene rearrangement analyses. Leukemias were mostly acute myeloid leukemia (AML) or T-cell acute lymphoblastic leukemia (T-ALL), along with B-cell ALL in a few cases (Figure 3A, bottom; supplemental Table 3), closely corresponding to the phenotypes of human leukemias with UTX deficiency.^23,28 

Next, we investigated virus integration sites by inverse PCR.³⁷  Six common integration sites (Sox4, Mecom, Osbpl1a, Notch1, Ikaros, and Tax1bp1) were identified (supplemental Figure 4; supplemental Table 4), most of which were reported as leukemia-associated genes.^38-40  Because Sox4 was highly expressed in tumors (supplemental Figure 4) and reported as a cooperative gene in leukemia development,^41,42  we examined the cooperative role of Sox4 overexpression with Utx deficiency (Figure 3B, top). Sox4-expressing Utx^Δ/Δ (Utx^Δ/Δ+Sox4) cells generated significantly increased colony-forming and replating abilities (Figure 3B, second row) and developed leukemia, classified as AML, at significantly higher frequency than Sox4-expressing Utx^+/+ (Utx^+/++Sox4) cells (Figure 3B, third row and bottom left). RNA-sequencing (RNA-sea) and gene set enrichment analysis (GSEA) identified the most positively enriched pathway in Utx^Δ/Δ+Sox4 cells as “Hedgehog signaling” (Figure 3B, bottom right), which was reported to contribute to hematopoietic malignancies.⁴³ Utx deficiency rendered hematopoietic cells susceptible to leukemia and additional gene alterations, such as Sox4 overexpression, and promoted acute transformation, possibly through activation of aberrant signal(s).

To clarify the molecular mechanisms underlying the Utx deficiency–induced hematopoietic abnormalities, we analyzed the gene expression profiles of Utx^+/+ and Utx^Δ/Δ LSKs. GSEA indicated that the most positively enriched gene set in Utx^Δ/Δ LSKs was “Oxidative phosphorylation” characterized by the upregulation of ATPases, NADH dehydrogenases, and cytochrome c oxidases (Figure 4A, left; supplemental Figure 5A, left; supplemental Table 5). In contrast, the most negatively enriched gene set was “TGF-β signaling,” with downregulation of TGF-β receptors, Activin receptors, and Smad proteins (Figure 4A, right; supplemental Figure 5A, right; supplemental Table 5).

Notably, the phenotypes of Utx^Δ/Δ mice, such as myeloid skewing with dysplasia, extramedullary hematopoiesis, impaired reconstitution activity, and leukemia susceptibility, are characteristics of hematopoietic aging.^44-46  In addition, upregulation of oxidative phosphorylation and downregulation of TGF-β signaling are hallmarks of an aged hematopoietic system.^47-49  Therefore, we compared the gene expression profiles of Utx^Δ/Δ LSKs with those reported for young and aged HSPCs.⁴⁷  Interestingly, genes with enhanced expression in Utx^Δ/Δ LSKs correlated significantly with the top 200 upregulated genes in aged HSPCs⁴⁷  (Aging Up Top200; supplemental Table 6; Figure 4B, left; supplemental Figure 5B, left). In addition, genes with reduced expression in Utx^Δ/Δ LSKs correlated significantly with the top 200 downregulated genes in aged HSPCs⁴⁷  (Aging Down Top200; Supplemental Table 6; Figure 4B, right; supplemental Figure 5B, right). Moreover, the gene expression patterns of Utx^Δ/Δ LSKs correlated conversely with molecular signatures of young HSPCs^50-52  (Figure 4C, top) and correlated positively with the molecular signature of myeloid cells⁵²  (Figure 4C, bottom). These results collectively indicate that the acquired loss of UTX converts the gene expression patterns of young HSPCs to those of aged cells.

We then analyzed Utx^Δ/Δ HSPCs from the perspective of aging. We first examined changes in Utx expression with aging. The expression levels of Utx in aged (20 months) LT-HSCs, ST-HSCs, and multipotent progenitor cells decreased significantly compared with those in young (2 months) cells, confirming the aging-associated decline in Utx expression in HSC subfractions (Figure 5A). Similarly, the expression levels of Uty were significantly reduced in aged HSCs (supplemental Figure 2C).

A previous study reported that the expression of CD41, a megakaryocyte/platelet marker, increases with age.⁵³  After confirming that aged (18 months) LSK-gated c-kit⁺ cells ex‐pressed CD41 at a high level, we measured CD41 expression in young (2 months) Utx^+/+ and Utx^Δ/Δ cells and found that Utx^Δ/Δ cells expressed CD41 at a significantly higher level than Utx^+/+ cells (Figure 5B). In addition, the marked enrichment of the oxidative phosphorylation in Utx^Δ/Δ LSKs (Figure 4A, left) strongly suggested an increase in reactive oxygen species (ROS), which is implicated in the aging process.⁴⁹  In fact, measurement of ROS in Utx^+/+ and Utx^Δ/Δ LSKs and LT-HSCs demonstrated a significant accumulation of ROS in Utx^Δ/Δ cells (Figure 5C). We also found positive enrichment of glutathione metabolism pathway in Utx^Δ/Δ cells (supplemental Figure 6A), which would be activated to scavenge the accumulated ROS. Another aspect of aged HSPCs is impaired recovery from the DNA damage response (DDR).^54,55  The analysis of the behavior and localization of DNA repair proteins in irradiated Utx^+/+ and Utx^Δ/Δ LT-HSCs showed that, although a similar number of γH2AX/53BP1-overlapping foci developed at 2 hours and finally cleared by 24 hours, Utx^Δ/Δ LT-HSCs exhibited significantly delayed repair kinetics and DDR remnants at 18 hours (Figure 5D). Because expression levels of DDR-associated genes were comparable between Utx^+/+ and Utx^Δ/Δ LSKs, although there was a slight decrease in Xrcc5 in Utx^Δ/Δ cells after irradiation^56,57  (supplemental Figure 6B), this phenotype may be related to Utx deficiency–induced altered chromatin accessibility at the DNA damage sites, as reported in a previous study.³⁴ 

We investigated whether the reexpression of Utx can rescue the functional defects of aged HSPCs in terms of reconstitution ability. To examine the contribution of the demethylase activity of UTX, we transduced aged (20 months) LSKs with EV-, wild-type Utx (Utx^WT)-, or demethylase-dead Utx (Utx^DD)¹⁰ -IRES-EGFP vector (referred to as EV-, Utx^WT-, or Utx^DD-IRES-EGFP), and c-kit⁺, EGFP⁺ cells were subjected to competitive repopulation assays (Figure 6A, top). Whereas all transplant-recipient mice receiving EV-IRES-EGFP–transduced or Utx^DD-IRES-EGFP–transduced aged LSKs (Aged+EV and Aged+Utx^DD) exhibited failed reconstitution (<1% PB chimerism),⁵⁸  several recipients transplanted with Utx^WT-IRES-EGFP-transduced aged cells (Aged+Utx^WT) were successfully reconstituted (>1% PB chimerism)⁵⁸  (reconstituted mice numbered 1-3; Figure 6A, bottom left). Interestingly, the degree of reconstitution correlated well with increased levels of Utx expression and decreased myeloid-biased differentiation (Figure 6A, bottom right). In addition, we found that the enforced Utx expression in aged cells significantly reduced the H3K27me3 levels and also significantly reversed the expression of P-selectin (Selp) and clusterin (Clu), the hallmark genes of hematopoietic aging (Figure 6B).⁵²  These findings indicate that Utx restores the aging-related phenotypes, at least in part, through reprogramming the H3K27 methylation status.

To analyze aging-associated genes regulated by UTX-mediated demethylation, Utx^+/+ and Utx^Δ/Δ LSKs were subjected to chromatin immunoprecipitation sequencing (ChIP-seq) for H3K27me3. The number of H3K27me3 peaks was higher in Utx^Δ/Δ LSKs than in Utx^+/+ LSKs (Figure 7A). The genomic regions with H3K27me3 peaks in Utx^Δ/Δ LSKs were subjected to the biological processes of DNA binding and transcriptional regulation in gene ontology enrichment analysis (Figure 7B). Of note, motif and pathway analyses revealed that the regions with H3K27me3 peaks in Utx^Δ/Δ LSKs were significantly enriched in the expected binding sites of SMAD, SP1, and EGR1, which are downstream transcription factors of TGF-β and are expected to functionally decline during aging^47,48  (Figure 7C-D). To investigate the direct regulation of UTX at TGF-β signaling-associated gene loci, we searched for genes with decreased expression and increased H3K27me3 levels around the transcription start site (TSS) ± 5 kb. The results identified Hnf4a and Col1a1, which are TGF-β signaling-associated genes^59,60  and are included in Aging Down Top200,⁴⁷  suggesting that these are direct targets of UTX (Figure 7E).

UTX regulates gene expression in a demethylase-independent manner as a component of COMPASS-like and SWI/SNF complexes.^5-8  To analyze demethylase-independent roles of UTX in the expression of aging-associated genes, we picked up commonly upregulated and downregulated genes between RNA-seq data of Utx^Δ/Δ LSKs and Aging Up/Down Top200 (Figure 7F, named Up genes and Down genes, respectively) and searched for the binding of KMT2D and BRG1/SMARCA4, the key components of COMPASS-like and SWI/SNF complexes, respectively, to the regulatory regions (TSS ± 5 kb) of the genes, by referring to published data sets.^61-68  We found KMT2D peaks only in Down genes (∼20%; Figure 7G). In addition, although SMARCA4 binds to both Up and Down genes, the total peak number was much higher in Down genes than in Up genes (430 vs 132) and co-occupancy of KMT2D and SMARCA4 was observed in 13 genes (Figure 7H). These results collectively suggest that UTX mainly regulates the expression of genes downregulated with aging, through both demethylase-dependent and -independent mechanisms.

There has been increasing interest in how stem cells maintain tissue homeostasis and in how their dysfunctions induce aging.^69,70  Epigenetic deregulation has been reported to contribute to the aging process of HSPCs.^71,72  For example, deletion of sirtuin (Sirt) family genes encoding histone deacetylases, such as Sirt1, Sirt3, Sirt6, and Sirt7, was reported to induce hematopoietic aging through accumulating DNA damage, reduced resistance to oxidative stress, aberrant activation of WNT signaling, and impaired mitochondrial biogenesis.^73-77  Deregulation of histone methylation is also deeply implicated in aging^47,78 ; however, the underlying molecular mechanisms remain less understood.

In this study, we demonstrated that deficiency of UTX, a demethylase for H3K27 and a component of the COMPASS-like and SWI/SNF complexes, induces phenotypes characteristic of hematopoietic aging. Utx^Δ/Δ mice exhibited abnormal hematopoietic differentiation with myeloid skewing, MDS-like morphological changes, extramedullary hematopoiesis, impaired repopulation ability, and high susceptibility to leukemia (Figures 1,²-3). RNA-seq analysis revealed that Utx deficiency converted the gene expression profiles of young HSPCs to those of aged cells (Figure 4). The expression levels of Utx in HSCs decline with age, and Utx^Δ/Δ HSPCs exhibited an increase in the expression of an aging-associated marker, accumulation of ROS, and impaired DDR after irradiation (Figure 5). These findings collectively indicate that Utx deficiency genetically compromises various metabolic and signaling pathways and phenotypically induces hematopoietic aging.

It remains controversial whether UTX functions as an oncogene or antioncogene.⁷⁹  In T-ALL cases, a previous report showed that UTX acts as an oncogene in TAL1-driven T-ALL.⁸⁰  In contrast, we and others^36,81  have provided evidence that UTX functions as an antioncogene in T-ALL. The results seem contradictory. However, interestingly, in the report by Van der Meulen et al,⁸¹  all the T-ALL cases with UTX mutations were associated with TLX3 and/or NOTCH1 mutations, and none of them cooccurred with TAL1 mutations. In addition, in our mutagenesis study, Tal1 was not detected as a cooperative gene (supplemental Table 4). Thus, whether UTX functions as an oncogene or antioncogene may depend on the driver mutation, and its tumorigenic role may vary in different cellular contexts.

It remains to be clarified whether UTX maintains the functional integrity of HSPCs via demethylase-dependent and/or -independent mechanisms. H3K27me3 staining of Utx^+/+ and Utx^Δ/Δ LT-HSCs showed graded and significant increases in H3K27me3 levels, along with Utx-deficiency (supplemental Figure 1C). In addition, in our experimental system, demethylase-dead Utx did not rescue the aged phenotype (Figure 6A). Moreover, motif and pathway analyses of the H3K27me3 ChIP data identified transcription factors involved in TGF-β signaling (Figures 7C-D), the most downregulated pathway in Utx^Δ/Δ HSPCs (Figure 4A). These data support demethylase-dependent mechanisms. However, the overlap between promoter regions with gains of H3K27me3 by Utx deficiency and Aging Down Top200 was found in a small subset (Figure 7E). In addition, comparison of RNA-seq data of HSPCs deficient in Ezh2, encoding an H3K27 methyltransferase of PRC2,^82,83  with those of our Utx^Δ/Δ HSPCs showed that EZH2 targets were classified into 2 groups that would represent methylation-dependent and -independent genes (supplemental Figure 7A). Moreover, comparison of the same data with Aging Up/Down Top200 exhibited much less enrichment than Utx^Δ/Δ HSPCs (supplemental Figure 7B). These findings suggest that there are demethylase-independent mechanism(s). Indeed, the search for binding of KMT2D and SMARCA4, the key components of COMPASS-like and SWI/SNF complexes, to the regulatory region of aging-associated genes revealed that KMT2D selectively and SMARCA4 preferentially bind to Down genes with frequent co-occupancy (Figure 7F-H). These results collectively suggest that UTX contributes to hematopoietic homeostasis, mainly by maintaining the expression of genes downregulated with aging, through both demethylase-dependent and -independent epigenetic programming. The upregulation of oxidative phosphorylation (Figure 4A) may be a secondary effect, because previous studies demonstrated that TGF-β signaling suppresses oxidative phosphorylation.^84,85 

Whether stem cell aging undergoes similar genetic and/or epigenetic pathways in the different tissues is an intriguing question. An increase in H3K27me3 levels at the transcription start sites was reported, not only in aged HSPCs but also in muscle stem cells (MSCs),⁸⁶  suggesting that the functional decline of UTX contributes to MSC aging as well as HSC aging. This idea is supported by a study reporting that UTX promotes MSC regeneration through the demethylation of H3K27me3.⁸⁷  Therefore, we compared changes in signaling pathways between Utx^Δ/Δ LSKs and aged MSCs.⁸⁸  Interestingly, we found a substantial overlap between positively and negatively enriched pathways. In upregulated pathways, 10 of 17 positively enriched pathways of aged MSCs⁸⁸  overlapped with the top 50 positively enriched pathways of Utx^Δ/Δ LSKs, and in downregulated pathways, 26 of the top 50 negatively enriched pathways of aged MSCs⁸⁸  overlapped with the top 50 negatively enriched pathways of Utx^Δ/Δ LSKs (supplemental Figure 8; see also supplemental Tables 5 and 7). Moreover, we compared the changed pathways not only in aged MSCs but also in aged fibroblasts and aged induced neurons (iNs)⁸⁹  to those in Utx^Δ/Δ LSKs. We found that all of the statistically significantly enriched pathways in aged fibroblasts and aged iNs⁸⁹  were included in the top 30 significantly downregulated pathways in Utx^Δ/Δ LSKs (supplemental Figure 9), strongly suggesting that the deregulation of specific sets of pathways may underlie stem cell aging in different tissues. Of note, the statistical change of calcium pathway was commonly observed in all the cell types examined (indicated by an asterisk in supplemental Figure 9), suggesting that this pathway is pivotal in maintaining the homeostasis of different stem cells, as suggested in previous studies.^90,91 

Interestingly, although only minimal changes were observed in the hematopoietic parameters of Utx^Δ/Uty males (supplemental Figure 2), retroviral insertional mutagenesis induced leukemia in Utx^Δ/Uty males (Figure 3A). Our findings, together with the results of previous studies,^10,11,16,81  indicate that UTY compensates for the absence of UTX in steady-state hematopoiesis but cannot compensate for UTX function as a tumor suppressor. UTY exerts very low demethylase activity but possesses the TPR domain^4,9 ; hence, it is intriguing to clarify the compensatory roles of UTY for UTX deficiency in different physiological and pathological conditions.^34,92,93 

In summary, our findings demonstrate that UTX plays a pivotal role in the functional integrity of HSPCs and maintenance of hematopoietic homeostasis by globally regulating aging-associated genes. Further studies are necessary to clarify how UTX recognizes aging-related genes and whether the enhanced expression of UTX can prevent aging.

The authors thank Yuki Sakai, Sawako Ogata, Rika Tai, and Yoshio Mitamura for providing animal care, genotyping, and performing molecular experiments and Junji Takeda and the RIKEN BioResource Center for providing KY1.1 ES cells and B6-Tg(CAG-FLPe)36 mice (RBRC01834), respectively.

This work was supported by Japanese Society for the Promotion of Science (JSPS) KAKENHI grant 17J05696 (Y. Sera) and grants 19H03693 and 19K22546 (H.H.); a research grant from The Japanese Society of Hematology (Y. Sera); and funding from the Medical Research Institute (MRI), Tokyo Women’s Medical University.

Contribution: Y. Sera, Y.N., T.U., T. Inaba, Y. Sotomaru, T. Ichinohe., M.K., Y.M., Z.-i.H., T.S., K.T. and H.H. designed the research, generated the genetically engineered mice, and wrote the manuscript; Y. Sera, Y.N., T.U., N.Y., H.K., K.-i.I., K.K., M.I., A.N., and K.T. performed hematopoietic cell analyses; A.K. analyzed the RNA-seq data; S.K. and A.I. analyzed the ChIP data; H.O. examined pathological specimens; L.W. generated the MOL4070A retrovirus; and all authors reviewed and agreed on the final version of the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Hiroaki Honda. Institute of Laboratory Animals, Tokyo Women’s Medical University, 8-1 Kawada-cho, Shinjuku-ku, Tokyo 162-8666, Japan; e-mail: honda.hiroaki@twmu.ac.jp.