Dlk1 maintains adult mice long-term HSCs by activating Notch signaling to restrict mitochondrial metabolism

Dlk1 conditional KO mice was generated by Jennifer Schmidt, University of Illinois at Chicago, and we purchased from Jackson lab. ptprc mutant mice and Mx1-cre mice were purchased from Shanghai Model organisms. All mice were housed under specific pathogen-free conditions in the Laboratory Animal Center of Zhejiang University (ZJU). The genotype identification of mice offspring was performed 3–4 weeks after birth. The mice offspring were intraperitoneally injected every other day with 2 mg/kg b.w. poly I:C for 7 times after genotype identification. The control mice were treated with the same conditions with the Dlk1 knockout mice. The flow cytometric analysis was carried out at least 1 month after injection of poly I:C.

The bone marrow cells derived from mice were culture in vitro in 96 well plate in StemSpan™ SFEM II (Stem Cell Technologies, #09655) with mouse five growth factors (SCF, TPO, FLT-3L, IL3 and IL-6) (novoprotein) for 7 days. At the second day, a final concentration of 10 mM of N-acetylcysteine (NAC, sigma, C7352-25G) or 50 ng/mL JAG1 (Abcam, ab109346) was added and at the fifth day, the medium in 96 well plate was replaced with fresh medium containing 10 mM NAC or 50 ng/mL JAG1. At the seventh day, the cells were harvested, counted. The HSCs gated from LSK, CD150⁺, CD48⁻ was analyzed by flow cytometry for frequency and mitochondrial parameters.

Mice mononuclear cells were collected from bone marrow, spleen and peripheral blood that were further lysed to remove the red blood cells. For mouse HSCs identification, mononuclear cells were stained with antibodies against Sca-1 (D7, biolegend, #108114), c-Kit (2B8, eBioscience, #17-1171-83), CD34 (RAM34, eBioscience, #11–0341-85), Flk2 (A2F10, BD, #562898), CD48 (HM48-1, eBioscience, #48-0481-82), CD150 (TC15-12F12.2, BioLegend, #115904), along with lineage cocktail including CD3e (145-2C11, eBioscience, #17-0031-83), CD4 (RM4-5, eBioscience, #15-0042-83), CD8a (53–6.7, eBioscience, #15-0081-83), Mac-1 (M1/70, eBioscience, #15-0112-83), Gr1 (RB6-8C5, eBioscience, #15–5931-83), CD45R (B220, RA3-6B2, eBioscience, #15–0452-83), IgM (Il-41, eBioscience, #15-5790-82), Ter119 (TER-119, eBioscience, #15-5921-83), CD127 (A7R34, eBioscience, #48-1271-82), CD16/CD32 (93, eBioscience, #25-0161-82), CD45.2 (eFluor780, eBioscience, #47-0454-82), CD45.1 (A20, eBioscience, #15-0453-82). Antibodies against CD3, CD4, CD8, B220, IgM, Ter119, CD71, Mac1 and Gr1 were utilized for lineages clarifying. As for donor engraftment analysis, mononuclear cells isolated from peripheral blood were stained with antibodies against CD45.1, CD45.2, Mac1, Gr1, CD3, B220. To identify progenitors, cells were stained with antibodies against Sca-1, c-Kit, CD34, CD16/32, CD127 and lineage cocktail. Antibodies against lineage cocktail, Sca-1 and c-Kit were used to stain the LSK for sorting that is conducted on MoFlo (Dako).

Ptprc stain mice were bred by baytril water for 1 week before and after transplantation and irradiated with nine lethally irradiated (9 Gy) in twice at the previous day. For ELDA assay, 2 × 10⁵, 7.5 × 10⁴, or 2.5 × 10⁴ donor-derived bone marrow mononuclear cells (BMMCs) from WT or Dlk1 KO mice (CD45.2), together with 2 × 10⁵ rescue cells (CD45.1) were transplanted intravenously into ptprc recipient mice. For secondary transplantation, 1 × 10⁶ BMMCs derived from primary transplanted mice were mouse-to-mouse transplanted into secondary recipient mice. For competitive transplantation assay using donors under long-term Dlk1 knockout, the procedures were nearly same as ELDA assay, but it only transplanted 2 × 10⁵ donor cells and 2 × 10⁵ rescue cells. In terms of invert transplantation, ptprc stain mice were served as donor while WT and Dlk1 KO mice were considered as recipient, the pretreatment and experimental procedures are the same to the competitive transplantation. Briefly, 2 × 10⁵ BMMCs derived from ptprc mice and 2 × 10⁵ BMMCs isolated from CD45.2 mice were intravenously transplanted into WT and Dlk1 KO mice. For reciprocal transplantation, 500 HSCs from wild-type congenic CD45.1 mouse were transplanted into Mx-1 Cre induced Dlk1 WT or Dlk1 KO CD45.2 recipients. The recipients were intraperitoneally injected with poly I:C to induce the expression of Mx-1 Cre 1 month after birth, and were lethally irradiated at the previous day of transplantation. Overall donor engraftment was confirmed by flow cytometric analysis.

According to the manufacturer’s instructions of BD Pharmingen™ FITC Mouse Anti-Ki-67 Set (BD Pharmingen™, #556026), 5 million mononuclear cells stained with HSCs markers as described above were fixed and permeated with fixation/permeabilization buffer at 4 °C for 1 h, and then washed with PBS containing 2% fetal bovine serum. After permeabilization, cells were incubated with anti-Ki-67 at 4 °C for 1 h in the dark and then stained with DAPI (Sango, E607303) at room temperature (RT) for 10 min. As for apoptosis assay, cells also conducted with HSCs antibodies stained with Annexin-V (Yeasen, 40304ES60) or 7-AAD (Sango, A606804) at RT for 10–15 min.

Three replications of LSK cells were sorted and total RNA were extracted with TRizol (Takara, 9109). High quality total RNA was reverse transcript to cDNA for library preparation using the SMART-Seq v4 Ultra Low Input RNA Kit for Sequencing (Takara, 634832) and Nextera XT (Illumina, FC-131–1024) according to the manufacturers’ protocols.

Lineage negative cells were isolated from WT and Dlk1 KO mice and total RNA was extracted by TRizol (Takara, 9109) according to its instruction. 200 ng total RNA were used for reverse transcription with HiScript II 1st Strand cDNA Synthesis Kit (Vazyme, R212) and cDNA were used for qPCR.

For measuring protein level of Dlk1 and key components of Notch pathway, lineage negative cells were isolated and cellular protein were extracted with RIPA lysis buffer following the manufacturer’s protocol. Immunoblotting was performed with anti-Dlk1 mouse monoclonal antibody (Santa cruz, sc-376755), Notch 1 mouse monoclonal antibody (Santa cruz, sc-376403), Jagged 1 mouse monoclonal antibody (Santa cruz, sc-390177), Hes 1 mouse monoclonal antibody (Santa cruz, sc-166410) and β-actin mouse monoclonal antibody (Sangon, D191047), HEY1 rabbit pAb (abclonal, A16110), HES5 rabbit pAb (abclonal, A16237), Notch3 rabbit pAb (abclonal, A13522), Notch1 NICD rabbit mAb (abclonal, A19090). The goat anti-mouse or anti-rabbit secondary antibody was purchased from Sangon.

Comparison between two groups was analyzed by unpaired two tailed Mann–Whitney test (non-parametric test). Comparison between multiple groups was analyzed by Kruskal Wallis test. Data were shown as mean ± SD. Statistical significance was defined as p < 0.05. Majority of graphs were generated by GraphPad Prism 8 (GraphPad Software). For CRUs in ELDA assay, data were analyzed by ELDA software. Successful engraftment was defined as that the frequency of CD45.2⁺ CD45.1⁻ population is above 5% of total hematopoietic cells in peripheral blood.

In our previous study, 17 types of murine hematopoietic cells were sorted by fluorescence-activated cell sorting and used for RNA-seq [2]. This RNA-seq result displayed that Dlk1 was highly expressed in LT-HSCs (Lin⁻, c-Kit⁺, Sca1⁺, CD34⁻, Flk2⁻ and CD49b^low) at both mRNA and protein level, the latter of which was validated by flow cytometry (Fig. 1A, B). qRT-PCR was also exploited to validate the expression of Dlk1 and showed a similar result (Fig. 1C). Several online hematological databases were also examined to systematically analyze the expression of Dlk1 in diverse blood cells. Bloodspot database exhibited that Dlk1 presented highest expression in HSCs, with relatively moderate level in MPP and low level in progenitors and mature cells (Additional file 1: Fig. S1A). Haemosphere database also showed that Dlk1 has a highest expression in LT-HSCs and moderate expression in ST-HSCs and MPP (Additional file 1: Fig. S1B). Overall, these results hinted that Dlk1 might play a unique role in LT-HSCs.

The Mx1-cre/floxp platform was used to specifically knockout Dlk1 in adult mice HSCs, in which the last two exons of Dlk1 were flanked by loxP sites (Additional file 1: Fig. S1C) [23]. After poly I:C induction, the protein expression of Dlk1 was examined and showed that Dlk1 was almost completely knocked out (KO) in HSCs (Fig. 1D). Interestingly, we observed a significant increase in frequency and absolute number of LT-HSCs in Dlk1 knockout mice, although total number of bone marrow cells were almost unchanged (Fig. 1E–G, Additional file 2: Fig. S2A). Analysis of cell cycle of HSCs showed that Dlk1 knockout resulted in pronounced higher frequency of S-G2-M stage cells, suggesting that Dlk1 deficiency facilitated the entry of quiescent HSCs into the cell cycle (Fig. 1H). We further examined the committed progenitors and mature lineage cells, and found that Dlk1 deletion did not exhibit obvious alterations in frequency and absolute number of common lymphoid progenitor (CLP), common myeloid progenitors (CMP), granulocyte–macrophage progenitors (GMP), megakaryocyte-erythrocyte progenitors (MEP) (Fig. 1I, J, Additional file 2: Fig. S2B). We further observed that Dlk1 knockout did not result in obvious alterations in frequency and absolute number of myeloid, T cells, except that B cells were slightly decreased although with no significance, and erythrocyte (erythro as abbreviation in figures) showed a significant reduction (Fig. 1K, L, Additional file 2: Fig. S2C). Together, these results implied that Dlk1 KO could increase the number of phenotypical LT-HSCs,.

To address the concern that whether HSC expansion caused by Dlk1 deletion has any detrimental effects on normal hematopoiesis, we examined the HSCs in the Dlk1 knockout mice 8 months after injection of poly I:C. We showed that the spleen size, spleen total cell number, the frequency and absolute number of HSPCs, progenitors in the spleen were not evidently changed (Additional file 3: Fig. S3A–I). We also examined the hemograms and found that a modest decrease in number of white blood cells (WBC), number and frequency of lymphocytes, frequency of HCT, MCV and PCT, as well as an increase in frequency of granulocytes (Additional file 3: Fig. S3J), which were not the classic leukemia symptoms. Thus, we reckoned that Dlk1 knockout in hemopoietic system would not render malignant hematopoiesis.

To identify the functional roles of Dlk1 in hematopoietic reconstitution ability of HSCs, we performed extreme limiting dilution analysis (ELDA), as shown in Fig. 2A. Interestingly, we found that the repopulating potential of Dlk1 knockout HSCs was notably undermined, as manifested by reduction of the overall engraftment and the absolute number of donor cells (CD45.2) from Dlk1 knockout mice when compared with their littermates (Fig. 2B, C). We also observed a moderate decrease in T cells of Dlk1 knockout recipients after 16 weeks of transplantation, but with no obvious changes in B cells and myeloid (Fig. 2D). A more thorough analysis of the bone marrow of recipient mice showed that the absolute number and frequency of various donor cells, including LSK (Lin⁻, c-Kit⁺, Sca1⁺), LT-HSC (LSK, CD34⁻ and Flk2⁻), ST-HSC (LSK, CD34⁺ and Flk2⁻), MPP (LSK, CD34⁺ and Flk2⁺), CLP, CMP, GMP, MEP and mature lineages (myeloid, T cells, B cells and erythrocyte) were reduced with different degrees (Fig. 2E–J, Additional file 4: Fig. S4A–C). To further verify long-term outcomes of Dlk1 knockout in hemopoietic system, we conducted secondary transplantation (Fig. 2K). We also observed a significant decrease in chimerism rate of Dlk1 knockout group (Fig. 2L). ELDA software was utilized to assess the absolute number of functional HSCs in total transplanted cells [24]. We showed that competitive repopulating units (CRUs) in Dlk1 wild type group was 1/27036 while CRUs in Dlk1 knockout group was 1/86257, which was 3.2-fold lower than wild-type group (Fig. 2K). Also, analysis of the bone marrow of the secondary recipient mice revealed that absolute number of bone marrow cells and LT-HSC, ST-HSC, MPP, LSK from recipient mice were significantly reduced (Fig. 2M, O), although the frequency of LT-HSC, ST-HSC, MPP, LSK showed a descending trend with no significance (Fig. 2N, Additional file 4: Fig. S4D).

We further characterized the functional roles of Dlk1 in HSCs with competitive repopulation assay, which the donor mice were sustained for a longer time of 6 months after poly I:C inducement (Fig. 3A). In line with the results of the ELDA assay, we observed that Dlk1 knockout in HSCs dramatically decreased its hematopoietic repopulation potential after series transplantation (Fig. 3B) and significantly decreased the absolute number of donor cells with a greater degree compared with that of ELDA assay (Figs. 2C, 3C). We also observed an obvious myeloid lineage bias and inhibition of B cells and T cells in first transplantation recipients of Dlk1 knockout group after 12 weeks of transplantation (Fig. 3D). The frequency and absolute number of almost all kinds of donor-derived cells in first transplantation recipients were decreased to varying degrees (Fig. 3E–J, Additional file 5: Fig. S5A–C). The results of secondary transplantation were similar to that of first transplantation (Fig. 3K–P). Notably, the mature lineages of T cells, B cells and erythrocyte, as well as the chimerism rate were approaching to zero after 16 weeks of secondary transplantation (Fig. 3B, O, P, Additional file 6: Fig. S6A–C). Taken together, these data indicated that Dlk1 knockout in HSCs would impair the long-term hematopoietic repopulation potential of HSCs.

Given that the full-length Dlk1 could be cleaved to produce a secreted protein with biological functions [25]. It has been reported that the secreted form of Dlk1 could interact with its ligands to regulate cellular physiological activities in a paracrine mode [26]. Besides, bone marrow stromal cells were able to produce Dlk1 [27], which may have an impact on functions of HSCs in an extrinsic way. So, to distinguish whether Dlk1 knockout-mediated effects were due to extrinsic influence of soluble form of Dlk1 in bone marrow microenvironment or intrinsic changes of HSCs, we conducted a reciprocal transplantation assay, in which CD45.1 donor cells were transplanted into Dlk1 knockout or wild type recipients (CD45.2). We showed that Dlk1 knockout in bone marrow microenvironment did not notably change the hematopoietic repopulation potential of CD45.1 donor cells (Additional file 7: Fig. S7A–D). After 16 weeks of transplantation, the frequency and absolute numbers of LT-HSCs, ST-HSCs, IT-HSCs and MPP have no discernible differences between Dlk1 wild type and knockout group (Additional file 7: Fig. S7E, F), indicating that an intrinsic change in the Dlk1 knockout HSCs was the primary cause of the phenotypes.

To identify the mechanism by which Dlk1 knockout expanded the phenotypical long-term HSCs but impaired its repopulation potential, we sorted LSK cells from Dlk1 WT or KO mice and conducted RNA-sequencing (Additional file 8: Fig. S8A). Unbiased principal component analysis indicated that the expression pattern of Dlk1 wild type or knockout group was diacritical (Fig. 4A). The volcano map revealed that Dlk1 knockout in LSK led to 5147 differentially expressed genes (DEGs), in which 1533 genes were upregulated and 3614 genes were downregulated (Fig. 4B). Gene Ontology (GO) analysis revealed that most of the upregulated terms enriched in cell cycle, mitotic nuclear division, translation, ribosome biogenesis, rRNA processing and mitochondrial translation (Fig. 4C). KEGG analysis also showed that upregulated genes enriched in pathways related to metabolism, oxidation phosphorylation and ribosome (Fig. 4D). GSEA analysis showed that in Dlk1 knockout group, gene signatures related to mitochondrial function and metabolism were upregulated, including mitochondrial gene expression, mitochondrial translation, mitochondrial respiratory chain complex assembly, mitochondrial electron transport, ATP synthesis and oxidation phosphorylation (Fig. 4E, Additional file 8: Fig. S8C). Collectively, these data imply that Dlk1 knockout in adult mice HSCs could activate the cell cycle and enhanced the cellular translation and mitochondrial activities.

To verify the increased mitochondrial metabolism and ribosomal translation caused by Dlk1 knockout in HSCs, we used multiple fluorescence probes to experimentally measure cellular metabolism and mitochondrial function. We analyzed protein synthesis in Dlk1 knockout HSPCs using an O-propargyl-puromycin (OPP) incorporation assay. In line with the RNA-Seq results, we observed that OPP incorporation was significantly increased in only Dlk1 knockout LT-HSCs, but not in ST-HSCs, MPP and LSK (Fig. 4F). Ribosome function and translation efficiency are important for HSCs homeostasis. In regards to the steady state of adult HSCs, protein synthesis level is low compared with committed cells and is under strict regulation [28]. Dysregulation in ribosome assembly and translation could impair the function of HSCs and even cause hematopoietic disorders, including Diamond-Blackfan anemia (DBA) and MDS [29‐31]. Increased protein synthesis tends to require a higher energy demand. We measured the glucose uptake by using 2-NBDG in Dlk1 knockout and wild type HSPCs. We showed that Dlk1 knockout led to significantly increased glucose uptake in LT-HSCs (Fig. 4G). We further examined the mitochondrial mass, mitochondrial DNA copy number and mitochondrial membrane potential (MMP), and showed that these mitochondrial parameters were all significantly increased in Dlk1 knockout LT-HSCs (Fig. 4H–J). The increased mitochondrial mass and MMP represent a higher mitochondrial biogenesis and mitochondrial metabolic activity, and in consistent with this, ATP content also revealed an obvious increase after Dlk1 knockout (Fig. 4K). We speculated that the increased mitochondrial metabolic activity after Dlk1 knockout might produce excessive ROS, which could harm HSCs. Therefore, we examined the ROS level and showed that the overall ROS was significantly increased in LT-HSCs and LSK (Fig. 4L). HSCs, particularly the LT-HSCs have been demonstrated to primarily keep a quiescence state in peacetime, which is maintained by low mitochondrial activity and very low ROS level [32]. A previous study also found that HSCs with lowest mitochondrial activity exhibit greatest hematopoietic potency and stem cell potential [33]. Accordingly, our results indicated that Dlk1 knockout could increase the ribosomal translation and mitochondrial activity in LT-HSCs.

To test whether the enhanced mitochondrial metabolism was the direct cause of Dlk1 knockout-induced phenotype in HSCs, we used an ROS scavenger, N-acetyl-l-cysteine (NAC) to rescue the phenotypes of HSCs in the context of Dlk1 knockout (Fig. 5A). As shown in Fig. 5, it was revealed that Dlk1 knockout significantly decreased the frequency and absolute number of HSCs, and increased the mitochondrial mass, MMP and superoxide level, which however, could be recovered to a large extent by NAC (Fig. 5B–F). Thus, these results indicated that Dlk1 knockout could expand phenotypic HSCs, which is possibly due to activation of cell cycle and enhanced mitochondrial metabolic activities in LT-HSCs, and consequently led to accumulation of ROS and damage of long-term repopulation potential of LT-HSCs.

As mentioned above, Dlk1 has been regarded as a non-canonical Notch ligand, and previous studies suggested that Dlk1 could exert its activity through activation or inhibition of Notch signaling under different cellular contexts [8, 9]. Notch signaling has been reported to regulate mitochondrial metabolism in breast cancer cells and in proinflammatory macrophage [34, 35] and was essential in embryonic hematopoiesis process [36]. Thus, we wondered whether Dlk1 knockout-induced phenotypes in HSCs were mediated by Notch signaling. To this end, we thoroughly analyzed the changes in Notch signaling pathway in RNA-Seq data and found that it was notably inhibited by Dlk1 knockout (Fig. 6A). Some critical genes in Notch signaling pathway was distinctly downregulated (Fig. 6B). We further experimentally examined the expression of some of these genes, and we showed that the mRNA levels of Notch1, Hes1 and Hey1 and protein expression of Notch1, active form of Notch1 (NICD), Notch3, and Notch downstream genes Hes1, Hey1, Hes5 were distinctly decreased after Dlk1 knockout in HSCs (Fig. 6C, D, Additional file 9: Fig. S9A). To further verify the involvement of Notch signaling in Dlk1 knockout-induced changes in HSCs, we re-activated the Notch signaling with an exogenous Notch ligand JAG1 in the context of Dlk1 knockout in HSCs (Fig. 6E). We firstly confirmed that exogenous JAG1 could partly reverse the inhibitory effect on Notch downstream gene expression (Hes1, Hey1 and Hes6) by Dlk1 KO (Additional file 9: Fig. S9B–D), indicating that JAG1 indeed activated Notch signaling. We then found that re-activation of Notch signaling could largely rescue the alterations in frequency and absolute number of HSCs (Fig. 6F, G), and reverse the over-activated mitochondrial metabolic activities of HSCs induced by Dlk1 knockout (Fig. 6H, I). We further over-expressed (OE) Notch1 active form NICD in the Dlk1 WT and Dlk1 KO HSCs to activate the Notch signaling (Additional file 9: Fig. S9E) and examined the alterations in HSC frequency and number and mitochondrial ROS levels. We found that activating Notch signaling under Dlk1 knockout or not by NICD overexpression had similar rescue effects on HSC frequency and number and mitochondrial ROS level with adding exogenous Notch ligand JAG1 (Additional file 9: Fig. S9F–H). Overall, these data showed that in adult mice HSCs, Dlk1 knockout could expand phenotypic HSCs, but virtually undermine their hematopoietic repopulation potential, which was partially due to inhibition of Notch signaling leading to over-activation of mitochondrial metabolism (Fig. 6J, Additional file 10: Table S1).