B-NDG (NOD-Prkdc ^Scid IL2rg ^tm1 /Bcgen, CD45.1⁺) mice were purchased from Biocytogen Jiangsu Co., Ltd (Jiangsu, China). Mice were maintained in individually ventilated racks in the SPF-grade animal room of the Guangzhou Institutes of Biomedicine and Health, Chinese Academy of Sciences (GIBH, CAS, China) and given sterilized food.

To generate green fluorescent protein positive ESCs (GFP-ESCs), the CAG Pr-GFP-PGK Pr-PuroR cassette was inserted into the Hipp11 locus of mouse ESCs (C57BL/6 background, CD45.2⁺) by CRISPR/Cas9-mediated gene knockin system. The target clones (GFP-ESCs) were selected by puromycine (1 μg/ml, Thermo Fischer Scientific), and the expression of GFP was confirmed by microscopy analysis and flow cytometry. In our study, p2a derived from porcine teschovirus-1 was used between Runx1 and Hoxa9 and between Hoxa9 and Hlf, and e2a derived from equine rhinitis A virus was used between Hlf and Hoxa7. The genes of Runx1, Hoxa9, Hlf, and Hoxa7 following CAG promoter were directionally cloned into the expression vector based on ClonExpress® technology (Vazyme) to form a gene tandem co-expression cassette: Runx1-p2a-Hoxa9-p2a-Hlf-e2a-Hoxa7. Simultaneously, the rtTA, TRE, and hygromycin (Hygro B) resistance selection marker sequences were also integrated to form a complete expression vector with hygromycin (Hygro B) resistance selection function and the genes of Runx1, Hoxa9, Hlf, and Hoxa7 we inserted being inducible by doxycycline. Then the CAG Pr-rtTA-3 × Stop-TRE-Runx1-p2a-Hoxa9-p2a-Hlf-e2a-Hoxa7-PGK Pr-HygroR cassette we have constructed was knocked into the Rosa26 locus of GFP-ESCs by cre-loxp-specific homologous recombination to generate iRunx1-Hoxa9-Hlf-Hoxa7 (iR9F7) ESCs. Similarly, a CAG Pr-rtTA-3 × Stop-TRE-Runx1-p2a-Hoxa9-p2a-Hlf-PGK Pr-HygroR cassette was inserted into the Rosa26 locus of GFP-ESCs by homologous recombination to generate iRunx1-Hoxa9-Hlf (iR9F) ESCs, and a CAG Pr-rtTA-3 × Stop-TRE-Runx1-p2a-Hoxa9-e2a-Hoxa7-PGK Pr-HygroR cassette was inserted into the Rosa26 locus of GFP-ESCs to generate iRunx1-Hoxa9-Hoxa7 (iR9A7) ESCs. The target clones (iR9F7-ESCs, iR9F-ESCs, and iR9A7-ESCs) selected by hygromycin B (150 μg/ml, Invivogen) were further cultured in ES medium containing doxycycline (1 μg/ml, Sigma). The induced expression of Runx1, Hoxa9, Hlf, and Hoxa7 of iR9F7-ESCs were confirmed by qPCR.

Mouse embryonic fibroblasts (MEFs) were derived from embryonic day 13.5 C57BL/6 mouse embryos. MEFs were incubated in DMEM/high glucose (Hyclone) with 10% fetal bovine serum (FBS) (Natocor) supplemented with 1% nonessential amino acids (NEAA, Gibco). C57BL/6 mouse embryonic stem cells (Biocytogen) were seeded into 12-well plate containing feeder layers in ES medium with DMEM/high glucose, 15% FBS (Gibco), 1% NEAA, 1% GlutaMAX (Gibco), 1% Sodium Pyruvate (Gibco), 0.1 mM β-mercaptoethanol (Gibco), 1 μM PD0325901 (Selleck), 3 μM Chir99021 (Selleck) and 1000 U/mL LIF. The OP9-DL1 cells (GFP⁺) were incubated in α-MEM (Gibco) supplemented with 20% FBS (CellMax). The AFT024 cell lines (ATCC) each with the gene encoding mIL3, mIL6, mSCF, and hFlt3L respectively been knocked in, were cultured in DMEM/high glucose supplemented with 10% FBS (Natocor), 0.1 mM β-mercaptoethanol, and 1% Sodium Pyruvate.

Following the two-step method described in our previous study (Guo et al., 2020), mouse ESCs were detached by 0.05% Trypsin-EDTA (Gibco) and resuspended in the basic differentiation medium (BDM): IMDM, 15% FBS (Gibco), 200 μg/ml iron-saturated transferrin (Sigma), 0.1 mM β-mercaptoethanol (Sigma), 1% GlutaMAX, and 50 μg/ml ascorbic acid (Sigma). The digested ESCs and feeder cells were plated into the 0.1% gelatin-coated well for 30 min, then the floating cells but not the rapidly attached feeder cells were collected for the subsequent differentiation. For embryoid body (EB) generation, the ESCs were resuspended at 100,000 cells/mL in the BDM supplemented with 5 ng/ml BMP4 (Peprotech) and plated at 20 uL/drop for inverted culture in 15 cm dishes for 2.5 days, then replanted them into gelatinized plates in BDM supplemented with cytokines (5 ng/ml BMP4 and 5 ng/ml VEGF, Novoprotein) for 3.5 days. At day 6, the medium was changed to BDM supplemented with 2% conditioned medium derived from the supernatants of AFT024-mIL3, AFT024-mIL6, AFT024-hFlt3L, and AFT024-mSCF cell cultures. Doxycycline (1 μg/ml, Sigma) was added from day 6. The medium was replaced every other day. 2 × 10⁴ OP9-DL1 stromal cells were seeded into a 12-well plate 12 h prior to the hematopoietic maturation step in EM medium (α-MEM, 15% FBS (Hyclone), 200 μg/ml iron-saturated transferrin, 0.1 mM β-mercaptoethanol, 1% GlutaMAX, 50 μg/ml ascorbic acid, 2% conditioned medium derived from supernatants of AFT024-mIL3, AFT024-hFlt3L, and AFT024-mSCF cell culture and 1 μg/ml doxycycline. Then, 1,000–3,000 sorted iHECs were seeded into each well for hematopoietic maturation. The EM medium was half replaced every 2 days.

Eight-to-ten-week-old B-NDG mice were sub-lethally irradiated (2.25 Gy) by an X-ray irradiator (RS 2000, Rad Source Inc.). Three million ESC-derived iHPCs were injected into each irradiated B-NDG mouse via retro-orbital veins. Doxycycline (1 mg/ml) was maintained in the drinking water to continuously induce the expression of the inserted genes. We also sorted Lineage⁻Scal⁺Kit⁺ (LSK) cells from the bone marrow of the primary recipients 6 weeks post-transplantation and performed secondary transplantation using B-NDG mice as recipients, with 500 LSK cells per recipient mouse.

Single-cell suspensions were prepared in phosphate-buffered saline (PBS) containing 2% FBS and filtered by 70 μm filter. After blocked by Fc (CD16/32) (Biolegend) antibody, cells were stained with combinations of fluorochrome-conjugated monoclonal antibodies, including Kit (2B8, eBioscience), CD31 (390, eBioscience), CD41 (eBioMWReg30, eBioscience), CD45 (30-F11, eBioscience), CD201 (eBio1560,eBioscience), CD45.2 (104, eBioscience), CD2 (RM2-5, eBioscience), CD3 (145-2C11, eBioscience), CD4 (GK1.5,eBioscience), CD8a (536.7, eBioscience), B220 (RA3-6B2, eBioscience), Mac1 (M1/70, BioLegend), NK1.1 (PK136, BioLegend), Ter119 (TER-119, eBioscience), Gr1 (RB6-8C5,eBioscience), IgM (II/41, eBioscience), IgD (11–26c.2a, BioLegend), Sca-1 (D7,eBioscience), CD19 (eBio1D3, eBioscience), CD43 (eBioR2/60, eBioscience), CD71 (R17217, eBioscience), CD61 (2C9.G2, BioLegend), CD127 (A7R34, BioLegend), CD135 (A2F10, BioLegend), Ly-6D (49-H4, eBioscience), CD48 (HM48-1, eBioscience), CD150 (TC15-12F12.2, BioLegend), Ly51 (6C3, BioLegend), CD24 (M1/69, BioLegend), Streptavidin Alexa Fluor® 700 (Invitrogen), and Streptavidin PE-Cy7 (BioLegend). DAPI solution (Sigma), or PI solution (BioLegend) was used to exclude dead cells. Cells were analyzed with Fortessa cytometer (BD Biosciences) and sorted with Arial II cytometer (BD Biosciences). The flow cytometry data were analyzed with FlowJo software (BD Biosciences).

Genomic DNA was purified with the Discover-sc Single Cell WGA Kit (Vazyme) from the regenerated cells which were sorted from recipient spleens 16 weeks post-transplantation. DNA samples (200 ng of each) were amplified with a S1000^TM Thermal Cycler (BIO-RAD). In brief, we firstly used genomic DNA for the first round of PCR (low brightness), and then used the purified and recovered PCR product as a template for the second round of PCR, and eventually used this PCR products as a sample for agarose gel electrophoresis. Agarose gel electrophoresis was performed using 1.2% agarose gel, 1% TAE buffer. DL5000 DNA Marker (Tsingke) was used to indicate DNA molecular weight. PCR sequencing was performed using the primer pairs flanking the Runx1-p2a-Hoxa9-p2a-Hlf-e2a-Hoxa7 element (Supplementary Figure S1). We used Trizol (Sigma) to lyse cells, then extracted RNA by isopropanol precipitation. ReverTra Ace qPCR RT Master Mix with gDNA Remover (FSQ-301, TOYOBO) kit was used for reverse transcription experiments. Hieff qPCR SYBR Green Master Mix was used to prepare the PCR reaction mixture before performing the reaction on the fluorescence quantitative PCR instrument (Bio-Rad, CFX-96). The primer information for genomic PCR, RT-PCR, and sequencing is provided in Supplementary Figure S1. PCR products were separated by agarose gel electrophoresis and were visualized by ethidium bromide staining. All PCR products shown correspond to expected molecular sizes.

We previously achieved regeneration of hematopoietic progenitors with dominant T lineage potential using a Runx1-Hoxa9 knockin mouse ESC line (Guo et al., 2020). To evaluate the effects of Hlf and Hoxa7 transcription factors on hematopoietic regeneration from PSCs in vitro, we construct a doxycycline inducible Runx1-Hoxa9-Hlf-Hoxa7 knockin mouse ESC line (iR9F7-ESCs, C57BL/6 background) by inserting the Runx1-Hoxa9-Hlf-Hoxa7 polycistronic into the Rosa26 locus of mouse ESCs via homologous recombination. In addition, we added a sequence of constitutively expressing GFP fluorescent protein into the Hipp11 locus of mouse ESCs (GFP-ESCs, C57BL/6 background) to monitor the progeny cells (Figure 1A). The GFP protein signals in ESCs were identified in the GFP-ESCs (Figure 1B). The addition of doxycycline successfully drives the conditional expression of exogenous Runx1, Hoxa9, Hlf, and Hoxa7 (Figure 1C). Thus, we build a mouse ESC line for conditionally expressing the tandem Runx1, Hoxa9, Hlf, and Hoxa7.

We used a two-step differentiation protocol (Guo et al., 2020) to induce hematopoietic regeneration from mouse ESCs (Figure 2A). To promote mesoderm differentiation toward hematopoietic cells, embryoid bodies were cultured with BMP4 (days 0–2.5), BMP4 and VEGF (days 2.5–6). Cultured supernatants of AFT024 cell lines with mSCF/mIL3/mIL6/hFlt3L overexpression were used as conditioned medium for the in vitro induction of induced hemogenic endothelial cells (iHECs) and subsequent iHPCs. In the presence of doxycycline, embryoid bodies on days 6 to 11 generated iHECs showing an immunophenotype of CD31⁺CD41^midCD45⁻Kit⁺CD201⁺ (Figure 2B), which were phenotypically similar to embryonic pre-HSCs (Zhou et al., 2016). To promote the development of iHECs towards iHPCs, embryoid body-derived iHECs were plated on OP9-DL1 stroma, and further educated into Lin⁻Kit⁺Sca1⁺ iHPCs on day 11 to 21 in the presence of doxycycline (Figure 2C). Morphologically, we witnessed the gradually changed cell morphology along with hemogenic and hematopoietic induction (Figure 2D). Thus, iR9F7-ESCs produced iHPCs in dishes.

To further explore the synergistic action of Runx1-Hoxa9-Hlf-Hoxa7 in engraftment and lineage contribution, 3 million R9F7-ESC-derived-iHPCs on day 21 were injected into sub-lethally irradiated (2.25 Gy) B-NDG mice (8-week-old, CD45.1 strain), Doxycycline (1 mg/ml) was maintained in the drinking water after transplantation. Strikingly, R9F7-iHPCs continuously contributed to multi-lineage reconstitution up to 16 weeks. The frequency of GFP⁺ donor cells in peripheral blood of recipients started at around 25% at 6 weeks post-transplantation, persisted around 20% to week 8 and decreased to 10% at 16 weeks post-transplantation (Figure 3A). At 8 weeks post-transplantation, donor-derived CD19⁺ B cells, CD3/CD4/CD8⁺ T cells, and Gr1/CD11b⁺ myeloid cells were all detected in the peripheral blood of recipient mice (Figure 3B). Of note, the proportion of R9F7-iHPC derived CD19⁺ iB cells were sustained at high level from week 6 (∼75%) to week 16 (∼65%) post-transplantation. In comparison, the proportions of iT cells and iMyeloid cells in the regenerated cell population have been maintained in a relatively low level, with the cell ratio both remaining at 3–10%. This suggests that the R9F7-ESC-derived-iHPCs contributed to myeloid, B, and T cell lineages at various time (Figure 3C). We sacrificed the R9F7-iHPC recipients to check the repopulation in the bone marrow and spleen 16 weeks after transplantation. Flow cytometric analysis showed that there were abundant CD19⁺ iB cells, CD3/CD4/CD8⁺ iT cells, and Gr1/CD11b⁺ iMyeloid cells in these organs in addition to peripheral blood of the recipients (Figure 3D). We also detected different B cell subtypes in the spleen of the recipients 6 weeks post-transplantation (Figure 3E), and the contribution of the iHPCs to erythroid and megakaryocytic lineages in the spleen of the recipients 7.5 days post-transplantation (Figure 3F). Collectively, our data imply that R9F7-overexpressed iHPCs repopulated multi-lineage haematopoiesis in B-NDG mice.

To investigate whether the R9F7-iHPCs can produce lineage-committed progenitors in vivo, we investigated the hematopoietic progenitor components in the recipient mice. Firstly, we analysed the bone marrow of the recipients at as early as day 7.5 post-transplantation, and clearly detected donor-derived common lymphoid progenitors (CLPs), multi-potent progenitors (MPPs), and B lymphoid precursors (pre-B) (Figure 4A). We also detected lineage-committed progenitor populations, including CLPs, myeloid progenitors (MPs), and B lymphoid progenitors (pro-B), in the bone marrow of the recipient mice up to 6 weeks after transplantation (Figures 4B,C). We next performed secondary transplantation with donor-derived LSK cells from the primary recipients 6 weeks post-transplantation, but hardly detected chimerism in the peripheral blood of the secondary recipient mice (Figure 4D). Together with the finding showing nearly absence of the immunophenotypic HSCs in the bone marrow of the primary recipients (Figure 4A), we concluded that forced expression of Runx1-Hoxa9-Hlf-Hoxa7 only achieved transient multi-lineage hematopoiesis rather than HSCs regeneration from ESCs.

To further verify that the regenerated cells were of iR9F7-ESC origin, genomic PCR assays were performed with donor-derived lineage cells from the iHPC recipients (Figures 5A,B). The GFP⁺CD45.2⁺ B, T, and myeloid cells were isolated from the spleen of the recipients as these cells were abundant in the spleen (Figure 5C). Then, DNA of sorted 2,000 regenerated cell aliquots were extracted, amplified and used for PCR of Runx1-p2a-Hoxa9-p2a-Hlf-e2a-Hoxa7 sequence, with the plasmid containing the gene tandem co-expression cassette Runx1-p2a-Hoxa9-p2a-Hlf-e2a-Hoxa7 as the positive control (Figure 5D). Sequencing of PCR products confirmed that the reconstituted lineage cells in vivo were derived from iR9F7-ESCs, which carried the inserted Runx1-p2a-Hoxa9-p2a-Hlf-e2a-Hoxa7 element (Figure 5E). We also validated the iR9F7-ESC-derived cell identity of the GFP⁺ erythroid progenitor cells (GFP⁺Ter119^loCD71⁺) and megakaryocytes (GFP⁺Lin⁻Ter119^-CD41⁺CD61⁺) by genome PCR (Figure 5F). Taken together, these results showed that R9F7-iHPCs repopulate multi-lineage haematopoiesis in B-NDG mice.

We wondered whether a combination of three transcription factors would be sufficient to confer multi-lineage engraftment, so we constructed and characterized two inducible cell lines for tandemly expressing Runx1-Hoxa9-Hlf and Runx1-Hoxa9-Hoxa7 respectively (Figure 6A). The GFP reporter was inserted into the Hipp11 locus of mouse ESCs (GFP-ESCs, C57BL/6 background), which allowed tracking of cells with conditional expression of Runx1-Hoxa9-Hlf (R9F) and Runx1-Hoxa9-Hoxa7 (R9A7). Following the same two-step method of testing transcription factor combinations (Guo et al., 2020), we achieved in vitro regeneration of R9F/R9A7-iHECs and subsequently R9F/R9A7-iHPCs (Supplementary Figure S2). Furthermore, transplantation assays of R9F-ESC-derived iHPCs showed that the R9F-iHPC derivatives could reconstitute only B cell lineage with extremely low contribution (Figure 6B). R9A7 ESC-derived iHPCs somewhat recapitulated the engraftment phenotype of the R9F7-iHPC recipients, despite showing a lower contribution, and the regenerated myeloid cells and T cells could only be temporarily maintained in vivo (Figure 6C). Overall, Coordination of Runx1, Hoxa9, Hlf, and Hoxa7 confers T, B, and myeloid cell lineage potential on ESC-derived iHPCs.