Animals were housed in the animal facility of the Guangzhou Institutes of Biomedicine and Health (GIBH). Nupr1^fl/fl mice were provided by Beijing Biocytogen Co., Ltd. CD45.1, Vav-cre, Mx1- cre, and Mdm2^+/- mice were purchased from the Jackson Laboratory. All the mouse lines were maintained on a pure C57BL/6 genetic background. All experiments were conducted in accordance with experimental protocols approved by the Animal Ethics Committee of GIBH.

We first labeled the HSC with (CD2, CD3, CD4, CD8, Ter119, B220, Gr1, CD48)-Alexa Fluor700, Sca1-Percp-cy5.5, ckit- APC-cy7, CD150-PE-cy7, CD34-FITC and CD135-PE. The cells were then fixed using 4% paraformaldehyde. After washing, the fixed cells were permeabilized with 0.1% saponin in phosphate-buffered saline together with Ki-67-APC staining for 45 min. Finally, the cells were resuspended in DAPI solution for staining for 1 h. The data were analyzed using Flowjo software.

Nupr1^-/- mice and WT littermates were injected with 1 mg bromodeoxyuridine (BrdU) on day 0. They were then allowed to drink water containing BrdU (0.8 mg/mL) ad libitum. On days 3, 4, and 5 after the injection of BrdU, four mice of each group were sacrificed. The rate of BrdU incorporation was analyzed by flow cytometry according to the BD Pharmingen^™ APC BrdU Flow Kit instructions.

The HSC culture protocol has been described elsewhere.³² Briefly, 50 HSC were sorted into fibronectin (Sigma)-coated 96- well U-bottomed plates directly and were cultured in F12 medium (Life Technologies), 1% insulin-transferrin- seleniumethanolamine (ITSX; Life Technologies), 10 mM HEPES (Life Technologies), 1% penicillin/streptomycin/glutamine (P/S/G; Life Technologies), 100 ng/mL mouse thrombopoietin, 10 ng/mL mouse stem cell factor and 0.1% polyvinyl alcohol (P8136, Sigma-Aldrich). Half the medium was changed every 2-3 days, by manually removing medium by pipetting and replacing it with fresh medium, as indicated.

For limiting dilution assays,³³ cells cultured for 10 days were transplanted into lethally irradiated C57BL/6-CD45.1 recipient mice, together with 2×10⁵ CD45.1 bone-marrow competitor cells. Recipients were analyzed every 4 weeks. Limiting dilution analysis was performed using ELDA software.³⁴ based on 1% peripheral blood multilineage chimerism as the threshold for positive engraftment.

One day before bone marrow transplantation, adult C57BL/6 recipient mice (CD45.1, 8-10 weeks old) were irradiated with two doses of 4.5 Gy (RS 2000, Rad Source) at a 4- hour interval. Bone marrow nucleated cells (BMNC; 2.5×10⁵) from Nupr1^-/- mice (CD45.2) and their WT (CD45.1) counterparts were mixed and injected into irradiated CD45.1 recipients by retro-orbital injection. Control BMNC (CD45.2), Mdm2^+/-Nupr1^-/- BMNC (CD45.2) or Mdm2^+/- BMNC (CD45.2) were also mixed with the same number of competitors (CD45.1) and transplanted into recipients. For Nupr1^fl/flMx1-cre transplantation, cre expression was induced through intraperitoneal injection of polyinosinic-polycytidylic acid (pI-pC, 250 mg/mouse) every other day 1 week before transplantation. The same number (2.5×10⁵) of Nupr1^fl/flMx1-cre and WT (CD45.1) BMNC were mixed and transplanted into the lethally irradiated CD45.1 recipients. Mx1-cre⁺ mice were taken as the experiment control. Mx1-cre and WT (CD45.1) BMNC (2.5×10⁵) were used for the transplant control. The transplanted mice were maintained on trimethoprim-sulfamethoxazole-treated water for 2 weeks. For secondary transplantation, BMNC (1×10⁶) were obtained from primary competitive transplanted recipients and injected into irradiated CD45.1 recipients (2 doses of 4.5 Gy, 1 day before transplantation). Donor-derived cells and hematopoietic lineages in peripheral blood were assessed monthly by flow cytometry.

The majority of long-term HSC are quiescent under homeostasis, which is a key mechanism for maintaining the HSC pool for life-long steady hematopoiesis. We hypothesized that genes preferentially expressed in HSC but immediately downregulated in multipotent progenitors (MPP) might form an intrinsic regulatory network for maintaining HSC quiescence. To test our hypothesis, we explored candidate factors by RNA-sequencing analysis of sorted HSC (Lin^– CD48^– Sca1⁺ c-kit⁺ CD150⁺) and MPP (Lin^– Sca1⁺ c-kit⁺ CD150^–). Analysis of differentially expressed genes showed a pattern of transcription factors preferentially present in HSC, including Rorc, Hoxb5, Rarb, Gfi1b, Mllt3, and Nupr1. By literature search, we found that most of the candidate genes except Nupr1 were reportedly not involved in regulating HSC homeostasis. Thus, we focused on the Nupr1 gene, the role of which in hematopoiesis has not been reported. The expression of Nupr1 in HSC was significantly higher (>25-fold, P=0.002) than that in MPP (Figure 1A, left). Real-time polymerase chain reaction (PCR) analysis further confirmed the same expression pattern (P<0.001), indicating an unknown role for Nupr1 in HSC (Figure 1A, right).

To study whether Nupr1 has any potential impact on the hematopoiesis of HSC, we created Nupr1 conditional knockout mice by introducing two loxp elements flanking exons 1 and 2 of the Nupr1 locus using a C57BL/6 background mESC line (Figure 1B). The resultant Nupr1^fl/fl mice were further crossed to Vav-Cre mice to generate Nupr1^fl/fl; Vav-Cre compound mice (Nupr1^-/- mice). The deletion of Nupr1 was confirmed by PCR in HSC (Online Supplementary Figure S1A-C). Adult Nupr1^-/- mice (8-10 weeks old) had a normal percentage of blood lineage cells in peripheral blood, including CD11b⁺ myeloid, CD19⁺ B, and CD90.2⁺ T lineage cells (Online Supplementary Figure S2). We further investigated the potential alterations of HSC homeostasis in the absence of Nupr1. Flow cytometry analysis demonstrated that the Nupr1^-/- HSC pool was comparable to the wild-type counterpart in terms of ratios and absolute numbers (Online Supplementary Figure S3). Subsequently, we examined the cell cycle status of Nupr1^-/- HSC using the proliferation marker Ki-67 and DAPI staining and found that the ratio of Nupr1^-/- HSC in G0-status was reduced significantly (P<0.001). Compared with WT HSC (median value: Nupr1^-/- HSC =73.67%, WT HSC = 87.15%), more Nupr1^-/- HSC entered G1-S-S2 and M phases (Figure 1C, D). To further confirm this novel phenotype, we performed a BrdU incorporation assay, which is conventionally used to assess the turn-over rates of blood cells in vivo.³⁵ The 8-week-old Nupr1^-/- mice and littermates were injected intraperitoneally with 1 mg BrdU on day 0, followed by continuous administration of BrdU via water (0.8 mg/mL) for up to 5 days (Figure 1E). After 3 days of BrdU labeling, ~50% of Nupr1^-/- HSC became BrdU⁺ compared with ~35% of WT HSC. The BrdU incorporation rates in HSC differed between the two mouse models (WT and Nupr1^-/-, P<0.001), and the dynamics changed along with time elapsed (P=0.012, two-way analysis of variance [ANOVA]). Kinetic analysis of BrdU incorporation from day 3 to day 5 revealed that Nupr1^-/- HSC contained a 1.5- fold larger BrdU⁺ population over WT HSC (Figure 1F, G). Collectively, these data indicate that the Nupr1-deletion drives HSC to enter the cell cycle and accelerates their turnover rates in homeostasis.

To confirm whether Nupr1^-/- HSC have a repopulating advantage or disadvantage in vivo, we performed a typical HSC competitive repopulation assay. BMNC (2.5x10⁵) from Nupr1^-/- mice (CD45.2) were transplanted into lethally irradiated recipients (CD45.1) along with the same number of WT (CD45.1) competitor cells. Bone marrow cells from littermates (Vav-Cre⁺, CD45.2⁺) were mixed with WT (CD45.1) competitors and transplanted into the recipients as the experiment control (Figure 2A). Sixteen weeks later, 1x10⁶ BMNC from the primary recipients were transplanted into lethally irradiated recipients to assess long-term engraftment. We observed that donor Nupr1^-/- cells accounted for ~70% of cells in the primary recipients, while the control cells accounted for 50%-60% in the recipients from the transplantation control assay. Nupr1^-/- cells gradually dominated in peripheral blood of recipients over time after transplantation (Figure 2B). In the chimeras, ~70% of myeloid cells and B lymphocytes were Nupr1^-/- donorderived cells, while ~60% of T lymphocytes were from CD45.1 competitive cells (Figure 2C). To further explore whether Nupr1^-/- HSC dominated in the HSC pool, we sacrificed the recipients and analyzed HSC 16 weeks after transplantation. The proportion and absolute number of Nupr1^-/- HSC were significantly higher (~1.5-fold) than those of the control HSC in primary recipients (Figure 2D, E). Previous research documented that HSC proliferated rapidly at the expense of their long-term repopulating ability. ^36-40 Interestingly, consistent with the dominating trend in the primary transplants, Nupr1^-/- cells continuously dominated in the peripheral blood of secondary recipients (Figure 3A). Nupr1^-/- HSC further occupied up to 90% of the total HSC in the bone marrow of secondary recipients. However, the control HSC accounted for less than 10% in the secondary recipients (Figure 3B, C). Collectively, these results indicate that the deletion of Nupr1 promotes the repopulating ability of HSC without impairing their longterm engraftment ability.

HSC in the cell cycle were reported to be more sensitive to irradiation damage.^41,42 To explore whether Nupr1^-/- HSC with a fast turn-over rate are more sensitive to irradiation, WT mice and Nupr1^-/- mice were exposed to a single dose of total body irradiation (4 Gy dose, 1 Gy/min). Apoptosis and cell cycle status were analyzed 6 h (early stage) and 24 h (later stage) later. As expected, Nupr1^-/- HSC showed a significantly enhanced sensitivity to irradiation: only ~40% of Nupr1^-/- HSC lived 6 h after irradiation, whereas ~70% of WT HSC were still alive (Online Supplementary Figure S4A). The proportion of radiation-induced apoptosis (annexin V⁺) of Nupr1^-/- HSC was significantly higher (~2-fold, P=0.02) than that of WT HSC (Online Supplementary Figure S4A). Furthermore, ~60% of the residual Nupr1^-/- HSC were in G1-S-G2-M proliferative phases compared with ~50% of the residual WT HSC (P=0.01) (Online Supplementary Figure S4B), indicating an accelerated replenishment rate in response to irradiation damage. At a later stage (24 h) after irradiation, we observed more living Nupr1^-/- HSC (WT vs. Nupr1^-/-: 74% vs. 86%) and less irradiation-induced apoptotic Nupr1^-/- HSC compared with WT HSC (Online Supplementary Figure S4C). The proportion of cycling cells (G1-S-G2-M proliferative phases) in the residual Nupr1^-/- HSC was still significantly higher (P<0.001) than the WT HSC 24 h after irradiation (WT vs. Nupr1^-/-: 29% vs. 48%) (Online Supplementary Figure S4D). Thus, Nupr1^-/- HSC were susceptible to irradiation-induced damage, but the surviving HSC proliferated, resulting in a fast recover of the HSC pool.

We next examined whether the deletion of Nupr1 could enhance HSC expansion in vitro. Fifty HSC sorted from WT and Nupr1^-/- mice were cultured in vitro for 10 days following a recently described protocol³² (Figure 4A). After 10 days of culture, the WT input cells yielded more than 2.2×10⁴ cells, while Nupr1^-/- HSC produced approximately 5×10⁴ total cells (P<0.001) (Figure 4B). The colonies derived from Nupr1^-/- HSC were much larger than those from WT HSC (Figure 4C). Furthermore, we analyzed the phenotypic HSC populations in the expanded cells and found that the absolute number of phenotypic HSC in individual Nupr1^-/- colonies was 3 times more than WT HSC (P=0.005) (Figure 4D, E). To determine whether the quantitative expansion of phenotypic HSC contained net proliferation of functional HSC, we performed competitive repopulating-unit assays,³³ using serial doses of limiting dilutions of the in vitro-expanded cells. The WT HSC frequency in the 10-day expanded cells was 1 in 371 cells, which is equivalent to 61 functional HSC, while the Nupr1^-/- HSC frequency in the 10-day expanded cells was 1 in 190 cells (Figure 4F),³⁴ which is equivalent to 263 functional HSC (P=0.045). Therefore, the deletion of Nupr1 induced an approximately 4-fold expansion of functional HSC numbers over WT HSC. Thus, deletion of Nupr1 enhances the expansion ability of HSC in vitro.

To further investigate the underlying molecular mechanisms of Nupr1 in regulating HSC, we performed RNAsequencing analysis of Nupr1^-/- HSC from 8-week-old Nupr1^-/- mice. Analysis of gene expression indicated that there were 319 genes differentially expressed between WT and Nupr1^-/- HSC (>2-fold difference in expression; adjusted P value <0.05 [DESeq2 R package]). Gene-ontology analysis of these differentially expressed genes indicated enrichment of genes involved in regulation of mitotic cell cycle and negative regulation of cell cycle (Online Supplementary Figure S5A, B). In addition, the positive regulatory genes of cell cycle, such as Cdk4, Cdk6, Akt1 and Akt2, were upregulated in the Nupr1^-/- HSC. However, regulators of HSC quiescence, such as Gfi1, Pten, Hlf, Cdc42 and Foxo1 were downregulated in the Nupr1^-/- HSC (Online Supplementary Figure S5C).^43,44 Gene set enrichment analysis illustrated that genes related to p53 pathways feedback loops, including Trp53, Ccng1, Ctnnb1, Pten, and Pik3c2b, were enriched in WT HSC (Figure 5A). The p53 pathway regulates a series of target genes involving cell cycle arrest, apoptosis, senescence, DNA repair, and metabolism.⁴⁵ Interestingly, the expression of p53 was significantly reduced (P<0.001) to one-third of the control value in Nupr1^-/- HSC (Figure 5B). Therefore, we hypothesized that downregulation of p53 in Nupr1^-/- HSC might account for the competitive advantage of the HSC. MDM2 is a ubiquitin ligase E3 for p53, which is a key repressive regulator of p53 signaling.⁴⁶ Mdm2-deficient mice showed increased levels of active p53, which is an ideal substitute model of upregulating p53 since directly overexpressing p53 leads to cell death and blood malignancies in mice.^27,47 Nupr1^-/- mice were crossed with Mdm2^+/- mice to achieve upregulation of p53 expression in Nupr1^-/- HSC. As expected, the levels of p53 protein expression in Nupr1^-/- Mdm2^+/- HSC were comparable to those in WT HSC (P>0.05) but significantly higher than those in Nupr1^-/- HSC, as measured by indirect immunofluorescence (Figure 5C, D). In addition, most genes involved in the p53 pathway were upregulated in the Nupr1^-/-Mdm2^+/- HSC, indicated partial recovery of the p53 pathway (Online Supplementary Figure S6). We next examined phenotypic HSC in the Nupr1^-/-Mdm2^+/- mice. Flow cytometry analysis showed that the Nupr1^-/-Mdm2^+/- HSC pool was indistinguishable from the WT, Nupr1^-/-, and Mdm2^+/- counterparts in terms of ratios and absolute numbers (Figure 6A, B). Furthermore, we tested the competitiveness of Nupr1^-/-Mdm2^+/- HSC in parallel with WT, Nupr1^-/-, and Mdm2^+/- HSC. BMNC (2.5x10⁵) from WT, Nupr1^-/- Mdm2^+/- mice (CD45.2), Nupr1^-/- mice (CD45.2), or Mdm2^+/- mice (CD45.2) were transplanted into lethally irradiated recipients (CD45.1) along with the same number of WT (CD45.1) BMNC. In the recipients of Nupr1^-/-Mdm2^+/- donor cells, the contribution of Nupr1^-/-Mdm2^+/- cells was significantly reduced (P<0.001) to ~20%, which was far below the percentage of Nupr1^-/- cells in recipients of Nupr1^-/- donor cells, and Mdm2^+/- cells accounted for less than 10% in the peripheral blood of recipients 16 weeks after transplantation (Figure 6C). Sixteen weeks after transplantation, we also analyzed the Nupr1^-/-Mdm2^+/- HSC in the chimeras. Surprisingly, only a few Nupr1^-/-Mdm2^+/- HSC were present in the HSC pool of the recipients, while the Nupr1^-/- HSC dominantly occupied the HSC pool (Figure 6D, E). Overall, the reversal of p53 expression offset the competitive advantage of Nupr1^-/- HSC.

In the Nupr1^fl/fl Vav-Cre model, the Nupr1 locus was deleted at an embryonic stage. To exclude the possibility that the effects of loss of Nupr1 observed in adulthood is a consequence of an effect coming from the embryo, Nupr1^fl/fl mice were crossed with Mx1-Cre mice to generate induced Nupr1 knockout mice at an adult stage in the presence of polyinosinic-polycytidylic acid (pIpC). The deletion of the Nupr1 gene in the Nupr1^fl/flMx1-cre mice was verified by PCR in HSC (Online Supplementary Figure S1D, E). Consistent with the observation of loss of Nupr1 in the Nupr1^fl/fl Vav-Cre model, significantly more Nupr1^fl/flMx1-cre HSC entered the G1-S-S2 and M phases (median value: 26.35%) than their counterparts from littermate control mice (median value: 14.13%, P=0.07) (Figure 7A-C). The competitive transplantation result showed that donor Nupr1^fl/flMx1-cre cells were advantaged over WT competitors in the peripheral blood of recipients (60%-80%) (Figure 7D). To further investigate whether Nupr1^fl/flMx1-cre HSC dominantly occupy recipient bone marrow, we sacrificed the recipients and analyzed the HSC 16 weeks after transplantation. The proportion and absolute number of Nupr1^fl/flMx1-cre HSC were significantly greater (~2-fold) than the control HSC competitors in primary recipients (Figure 7E, F). Thus, in the Nupr1^fl/fl Mx1-cre model, deletion of Nupr1 in adulthood also promotes HSC engraftment.