Haematopoietic stem and progenitor cells from human pluripotent stem cells

Abstract

A variety of tissue lineages can be differentiated from pluripotent stem cells by mimicking embryonic development through stepwise exposure to morphogens, or by conversion of one differentiated cell type into another by enforced expression of master transcription factors. Here, to yield functional human haematopoietic stem cells, we perform morphogen-directed differentiation of human pluripotent stem cells into haemogenic endothelium followed by screening of 26 candidate haematopoietic stem-cell-specifying transcription factors for their capacity to promote multi-lineage haematopoietic engraftment in mouse hosts. We recover seven transcription factors (ERG, HOXA5, HOXA9, HOXA10, LCOR, RUNX1 and SPI1) that are sufficient to convert haemogenic endothelium into haematopoietic stem and progenitor cells that engraft myeloid, B and T cells in primary and secondary mouse recipients. Our combined approach of morphogen-driven differentiation and transcription-factor-mediated cell fate conversion produces haematopoietic stem and progenitor cells from pluripotent stem cells and holds promise for modelling haematopoietic disease in humanized mice and for therapeutic strategies in genetic blood disorders.

Extended Data Figure 1. Induction of haemogenic endothelium from hPSCs

a, Embryoid bodies formed from hPSCs. Subsequent panels show FACS analysis of day 8 embryoid bodies before magnetic cell isolation. b, CD34⁺FLK1⁺ population indicates haemogenic endothelium cells.CD235A and FLK1 plots show percentage of FLK1⁺CD235A⁻ cells in embryoid bodies, further gated with CD43 and CD34 plots to detect haemogenic endothelium (CD34⁺FLK1⁺CD235A⁻CD43⁻). c, d, Haemogenic endothelium isolated at day 8 (Supplementary Fig. 1) was further cultured in EHT medium for the indicated number of days. qRT–PCR showed (c) downregulation of endothelial genes and (d) upregulation of haematopoietic genes. HSPC genes (RUNX1, SCL/TAL1) peak on day 3 of EHT culture; consequently, this time point was chosen for introducing transcription factors followed by transplantation in subsequent experiments. e, Microscopy and FACS analysis on day 7 of EHT showed the appearance of haematopoietic cells (CD34⁺CD45⁺). Data shown as mean ± s.d.

Extended Data Figure 2. Rationale for selecting candidate transcription factors for library screening

a, Heatmap of the expression profile of HSC-specific transcription factors (TFs) in haemogenic endothelium (CD34⁺FLK1⁺CD43⁻CD235A⁻) versus fetal-liver HSCs (CD34⁺CD38⁻CD90⁺CD45⁺). Twelve HSC-specific transcription factors enriched in fetal-liver HSCs relative to haemogenic endothelium (blue box) were cloned individually into a Dox-inducible lentiviral vector. b, The expression level of SOX17, a marker of haemogenic endothelium, was 2.4-fold higher in haemogenic endothelium (N = 7) than fetal-liver HSCs (N = 10). *P < 0.001. c, The library was supplemented with genes identified in previous screens. Candidates in red were drawn from a previous screen by the Rafii group; blue from the Rossi group; green from the Daley group. The final library of 26 candidates is shown. d, Diagram of the Dox-ON pInducer-21 lentiviral vector used in this study (top). rtTA3 and eGFP are driven by EF1α -promoter; infection efficiency is indicated by the GFP signal. e, GFP analysis by FACS and fluorescence 3 days after infection of haemogenic endothelium cells. Routinely, over 50% transduction efficiency was achieved. For transplantation, haemogenic endothelium cells infected at day 3 EHT were incubated for 24 h and injected into mice. Dox was provided for 2 weeks in vivo after transplantation into sub-lethally irradiated immune-deficient NSG mice. f, Scheme for screening the 26 transcription factors library, and resulting haematopoietic chimaerism. hPSC-derived haemogenic endothelium was cultured for an additional 3 days in EHT medium, then infected with the library of 26 transcription factors. Infected cells (100,000) were injected intrafemorally into sub-lethally irradiated (250 rad) NSG mice, which were treated with doxycycline for 2 weeks to induce transgene expression in vivo. g, FACS analysis of bone marrow and thymus of an engrafted recipient is shown. Human CD45⁺ cells from bone marrow were analysed for CD33⁺ myeloid cells, CD19⁺ B cells, and CD3⁺ T cells as indicated. Thymic cells were analysed for human CD4 and CD8, with percentages of single and double-positive cells indicated. A photograph of an engrafted thymus is shown (bottom right). Data shown as mean ± s.d.

Extended Data Figure 3. Compiled data from transplantation experiments performed so far

a, b, Total engrafted mice assessed in primary transplantation at various time points. a, Histogram of total mice injected with haemogenic endothelium cells and haemogenic endothelium cells infected with indicated library or constituent transcription factors. Haemogenic endothelium only: 2 out of 30 mice engrafted, none of which were multi-lineage (myeloid cells, erythroid cells, B cells and T cells); library: 11 out of 40 mice engrafted, 5 multi-lineage; 7 transcription factors: 33 out of 76 mice engrafted, 9 multi-lineage; 5TFpoly: 15 out of 30 mice engrafted, 5 multi-lineage. Engraftment was assessed from bone marrow at 4–16 weeks. b, Tables 1 and 2 show compiled data. c, Representative photomicrographs and FACS plots of robust EHT (top) and failed EHT (bottom). The appearance of round cells budding from adherent haemogenic endothelium cells by direct microscopic visualization was assessed routinely. RUNX1c+24 reporter positivity of haemogenic endothelium cells at day 3 of culture likewise reflected robustness of EHT (20% versus 5%).

Extended Data Figure 4. Identification of transcription factors that confer multi-lineage haematopoiesis in vivo

a, SNP analysis of engrafted blood cells compared with hiPSC and cord blood MNCs. SNP array genotyping was conducted to confirm the origin of engrafted human cells from hiPSCs in representative mice. SNP genotypes for human CD45⁺ cells taken from bone marrow of engrafted mice, original hPSCs and reference cord blood MNCs were clustered, as shown, showing concordance of original hiPSCs and human cells recovered from bone marrow of engrafted mice. b, Tabular presentation of SNP data: concordance > 99% indicates identity between cell types. Green highlight shows that original hPSC line (34hiPSC or H9 hESC) corresponds with human CD45⁺ cells in engrafted mice. Comparison with different hPSC lines or cord blood MNCs did not achieve 99% concordance, validating the SNP array as a means of defining origin of cells. c, Transgene detection in engrafted cells of primary recipients. CD33⁺ myeloid cells, CD19⁺ B cells, and CD3⁺ T cells were isolated from the human CD45⁺ population of bone marrow at 10 weeks from five independent mice. Genomic DNA of each cell type was analysed by PCR to detect integrated lentivirus. Identification number of recipients is shown (numbers 1, 5, and 6 were engrafted with hiPSC-derived haemogenic endothelium; numbers 2 and 3 were engrafted with hESC-derived haemogenic endothelium). L, left femur (injected side); R, right femur; +, positive control (lentiviral vector with each transcription factor). –, negative control (lentiviral vector without transcription factor). d, Overlap between transcription factors that conferred in vivo engraftment from hiPSC- and hESC-derived cells injected into mice and in vitro multi-lineage CFU potential. Transcription factors detected from genomic DNA PCR from in vitro colony screening and in vivo engraftment screening are shown. For in vivo screening, multiple cell lines (iPSC and ESC) were used. Factors detected by in vivo screening were overlapped with those detected by in vitro screening (RUNX1 and LCOR). Consistently, MYB and RORA were detected by in vitro screening as previously reported. Overall, RUNX1, LCOR, SPI1, ERG, HOXA5, HOXA9, and HOXA10 (defined seven transcription factors) were identified in individual experiments with different PSC lines in vivo. e, Factor-minus-one approach to define essential transcription factors for engraftment. Haemogenic endothelium was infected with combination of seven transcription factors minus one each, as indicated, then transplanted into NSG mice. At the 8 week time point, engraftment of human CD45⁺ cells in bone marrow was determined by FACS. Each panel indicates a representative result of GFP vector: all seven transcription factors, and seven minus RUNX1, ERG, SPI1, LCOR, HOXA5, HOXA9, or HOXA10 as indicated in the panel. Reduction of chimaerism was seen when RUNX1, ERG, LCOR, HOXA5, or HOXA9 were removed. In contrast, omitting SPI1 or HOXA10 had a negligible effect on engraftment.

Extended Data Figure 5. FACS analysis after secondary transplantation of HE-7TF cells

Human CD34⁺ cells were obtained by magnetic cell isolation from bone marrow of primary recipients of HE-7TF cells, then 3,000 cells (a, b) or 1,000 cells (c) were intrafemorally transplanted into secondary recipients. Multi-lineage engraftment in bone marrow from a representative recipient mouse is shown; a, 8 weeks; b, 14 weeks; c, 16 weeks. Specific mice analysed are indicated in Fig. 2b. d, Limiting dilution assay of HE-7TF cells after secondary transplantation. CD34⁺ cells were isolated from bone marrow of primary recipients, and either 1,000 or 3,000 cells were transplanted into secondary recipients. Multi-lineage engrafted recipients were counted as response. Confidence interval of 1/(stem-cell frequency) was calculated by ELDA (http://bioinf.wehi.edu.au/software/elda/) according to Poisson distribution. A limiting dilution assay of cord blood was used as reference. e, Transgene detection in engrafted cells of secondary recipients of HE-7TF cells. Recipient numbers are from Fig. 2b (numbers 31 and 35). f, Bone marrow chimaerism of primary mouse engrafted with 5-TF^Poly at 12 weeks. Human CD45⁺ bone marrow of engrafted NSG was analysed for HSPCs (CD34⁺CD38⁻), nucleated erythroid (GLY-A⁺SYTO60⁺), enucleated erythroid (GLY-A⁺SYTO60⁻), neutrophils (PECAM⁺CD15⁺), B cells (IgM⁺CD19⁺), B progenitor cells (IgM⁻CD19⁺), B lymphocytes (IgM⁻CD19⁺CD38⁺⁺), and T cells (CD3⁺/CD4, CD8). g, Representative FACS plots of bone marrow engrafted with human cord blood HSCs are shown at 10 and 12 weeks.

Extended Data Figure 6. Identification of the source of engraftable cells within the HE-7TF population

a, Conversion of haemogenic endothelium (HE) into haematopoietic stem/progenitor cells by seven transcription factors requires EHT. Haemogenic endothelium or haemogenic endothelium grown in EHT medium for 3 days was transduced with seven transcription factors and transplanted into mice followed by bone marrow analysis at 4 weeks. Cells grown after 3 days of EHT showed CD45⁺ cells while those that were not grown under EHT conditions did not show CD45⁺ cells. b, Human umbilical vein endothelial cells were transduced with seven transcription factors or GFP lentiviral vectors, then cultured in EHT medium with Dox for a week. Flow cytometry analysis of PECAM (EC marker) and CD34 (haematopoietic marker) is shown. Human umbilical vein endothelial cells transduced with seven transcription factors fails to produce robust CD34⁺ cell population. c, Twenty-five thousand CD34⁺CD43⁺CD45⁺ (triple positive, TP) or CD34⁺CD43⁻CD45⁻ (single positive, SP) cells were FACS-isolated and transplanted (N = 5 mice per group). Engraftment of human CD45⁺ cells was assessed in peripheral blood (PB) at 8 weeks. FACS plots of human CD45 and mouse CD45.1 of TP (left) and single positive (right) transplanted mice at the 8 week time point are shown. d, Multi-lineage engraftment of bone marrow and spleen from primary recipient mouse at 6 weeks is shown. CD34⁺CD43⁺CD45⁺ cells were intravenously injected.

Extended Data Figure 7. Molecular features of HE-7TF cells

a, Correlation matrix of RNA-seq data from HSPC populations (CD34⁺CD38⁻CD45⁺) from HE-7TF cells or cord blood HSC engrafted for 12 weeks, iPS-haemogenic endothelium, and publicly available gene expression data (PubMed identifiers 26541609 and 26502406). RNA-seq samples from this study were RSips_7F_Average, RS_HE_Average, and RS_HSC_CB. Samples from ref. 23 were BCP_Average, CLP_Average, Thy_Average, HSC_Average, LMPP_Average, and HSC_CB_Average. The remaining samples are from ref. 24. b, Gene set enrichment analysis signature of HE-7TF cells compared with haemogenic endothelium cells. HE-7TF cells show gene expression signatures that positively correlate with LMO2 targets, TEF1 targets, HOXA4 targets, chemokine receptors and chemokines, TCF4 targets, integrin signal, GATA3 targets, and HSPCs. P < 0.05 and FDR q value < 0.25 were considered significant conditions. All gene set enrichment analysis plots satisfied these conditions, except the HSPC signatures, which had a q value of 0.121 and a P value of 0.06, suggesting that transcriptional differences remain between HE-7TF and bona fide HSCs/HSPCs. HSPC signature taken from ref. 56. c, RPKM values of seven transcription factors and HOXA target genes in indicated cell types are shown. HOXA target genes from ref. 28. PubMed identifier 27183470. d, Heatmap depiction of relative expression levels of the 26 transcription factors in the library in the following samples: HE, HE-7TF cells (engrafted), cord blood HSCs (engrafted), and fresh HSCs and progenitors. Notably, HE-7TF cells show high expression of HOXA family genes, GATA2, TGIF2, SOX4, and EVI1. e, The t-SNE of in-droplet single-cell RNA-seq of HE-7TF cells and cord blood HSCs (engrafted CD34⁺CD38⁻CD45⁺ cells from bone marrow at 12 weeks). The t-SNE from the top 500 most variable genes is presented in the top panel. f, Expression value of 8 haematopoietic genes in the same plot. Notably, the middle population (a subpopulation of HE-7TF cells) shared similar expression values, and degrees of heterogeneity of RUNX1, MEIS1, CD34, TAL1 with cord blood HSCs.

Extended Data Figure 8. Characterizations of differentiated cells: analysis of definitive erythropoiesis by relative quantification of globin transcripts

a, Human GLY-A⁺ cells were isolated from lysed bone marrow (to exclude enucleated cells) and analysed by qRT–PCR to quantify (b) HBE, (c) HBG, and (d) HBB genes. CB, GLY-A⁺ erythroid cells from cord-blood-engrafted in NSG bone marrow; 7 TF HSPCs, GLY-A⁺ erythroid cells from seven transcription factor HSPC-engrafted in NSG bone marrow; 5F, GLY-A⁺ erythroid cells from hPSCs transduced with ERG, RORA, HOXA9, SOX4, and MYB. Analysis of T-cell receptor diversity in engrafted T cells. e, Flow cytometric phenotyping of T cells from engrafted HE-7TF cells. Thymus was collected at 8 weeks and analysed for T-cell markers (CD4, CD8, CD3, TCRαβ, and TCRγδ). TCR phenotyping of the CD3⁺ population is shown on the right. One out of three recipients showed the presence of TCRγδ. Three thymic engrafted mice from independent experiments each. f–h, TCR rearrangement of thymocytes from cord blood CD34⁺ and HE-7TF engrafted in NSG. CD3⁺ T cells were isolated from NSG mice engrafted with (f) cord blood HSCs or (g) HE-7TF cells. Purified DNA was subjected to next-generation sequencing of the CDR3 using immunoSEQ (Adaptive Biotechnology) and analysed with the immunoSEQ Analyzer software (Adaptive Biotechnology). A high degree of combinatorial diversity in the V-gene segment usage was observed in CDR3 length, following a standard Gaussian distribution. h, Frequency of clonotype of T cells. i, Flow cytometric phenotyping of spleens from engrafted HE-7TF cells versus cord blood (j). Spleens were collected at 8 weeks and human CD45⁺ cells were analysed for T-cell markers (TCRβ, CD4, CD45RO, and CD45RA), and B-cell marker (CD19).

Extended Data Figure 9. Integration sequencing analysis of engrafted myeloid cells, B cells, and T cells from two individual animals

a, Left: from 8 weeks, indicated in Fig. 2b; right: from 5 weeks. Clonally expanded populations are shown for each lineage, with common clones among three lineages represented by colour. The smallest proportional coloured segments for each animal represent the unit value of two clonal sequences. b, Overlap of commonly expanded integration sites from an animal shown on the left in a. Genomic DNA-sequencing of CD33⁺ myeloid cells, CD19⁺ B cells, and CD3⁺ T cells from bone marrow detected common expanded integration sites. c, Agarose gel electrophoresis of adaptor-ligation of engrafted cells. Cells were obtained independently from those used in a at the 1 week time point. d, Integrated loci mapped in genome. Nearby genes of common integrated sites are described. These data are from an animal indicated in a at the 8 week time point.

Extended Data Figure 10. Representative FACS plots

a, b, FACS plot of cord blood HSCs (a) and HE-7TF cell-engrafted NSG mice at the 8 week time point (b). Each panel shows human CD45⁺ engraftment, HSPCs (CD34⁺CD38-), myeloid cells (CD33⁺), B cells (CD19⁺), T cells (CD3⁺CD4⁺CD8⁺), and erythrocytes (GLY-A⁺). Spleens were collected at 8 weeks, and human CD45⁺ cells were analysed for T-cell markers (TCRβ, CD4, CD45RO and CD45RA), and B-cell marker (CD19).

Figure 1. In vivo screening identifies transcription factors that enable engraftment from PSCs

a, Percentage of human CD45⁺ cells detected in peripheral blood of injected mice at indicated number of weeks. b, Multi-lineage contribution of human cells in bone marrow of engrafted mice. Bone marrow of NSG mice engrafted with haemogenic endothelium cells infected with the transcription factor library was analysed at 12 weeks for myeloid cells (M; CD33⁺), erythroid cells (E; GLY-A⁺), B cells (CD19⁺), and T cells (CD3⁺) within the human CD45⁺ population. Recipients 1, 5, and 6 were engrafted from hiPSCs; recipient 2 left (L) femur and right (R) femur, recipient 3 left (L) femur and right (R) femur were engrafted from hESCs; recipients CB 1 and CB 2 were engrafted with cord blood HSPCs. c, Bone marrow of primary NSG mouse engrafted with HE-7 transcription factor was analysed at 12 weeks for human CD45⁺ HSPCs (CD34⁺CD38⁻), nucleated erythroid cells (GLY-A⁺SYTO60⁺), enucleated erythroid cells (GLY-A⁺SYTO60⁻), neutrophils (PECAM⁺CD15⁺), B cells (IgM⁺CD19⁺), and B progenitor cells (IgM⁻CD19⁺). The thymus was analysed for T cells (CD3⁺/CD4, CD8) (bottom right). d, In vivo factor-minus-one analysis of defined seven transcription factors to identify necessary and redundant factors. Bone marrow of engrafted NSG was analysed at 8 weeks for human CD45⁺ population. The absence of RUNX1 (0.33-fold, P = 0.037), ERG (0.40-fold, P = 0.056), LCOR (0.23-fold, P = 0.020), HOXA5 (0.37-fold, P = 0.056), or HOXA9 (0.26-fold, P = 0.026) reduced chimaerism. Lentiviral vector with green fluorescent protein (GFP) was used as negative control. N = 2 mice analysed in two independent experiments with three mice each (two mice each for GFP). *P < 0.05. Average lineage distribution from each group is shown (right). Data shown as mean ± s.d.

Figure 2. Defined transcription factors confer multi-lineage engraftment

a, Engraftment of human CD45⁺ cells in bone marrow (BM) of recipient mice detected by flow cytometry at 12 weeks for haemogenic endothelium cells infected with transcription factor library (n = 12 mice analysed in two independent experiments with six mice each), seven transcription factors (n = 15 mice analysed in five independent experiments with three mice each), and 5-TF^Poly (n = 9 mice analysed in three independent experiments with three mice each). b, Bone marrow lineage distribution of myeloid cells (CD33⁺), erythroid cells (GLY-A⁺), B cells (CD19⁺), and T cells (CD3⁺) is shown as a bar graph for individual mice (numbers 1–29, primary recipients; numbers 30–49, secondary recipients). Each number indicates independent mice at time of bone marrow analysis after being euthanized (w, weeks). Multi-lineage capacity was assessed from bone marrow samples. Right: CD34⁺ cord-blood-engrafted recipients (numbers 50–64) injected with indicated numbers of cells. Secondary recipient numbers 30–34 were transplanted with cells from mouse number 9; numbers 35–39 were transplanted with cells from number 11; numbers 40–44 received cells from number 16; numbers 45–49 received cells from number 26). The dose and route of injected cells is shown (i.f., intrafemoral; i.v., intravenous). Mice numbers set in bold sloping type indicate multi-lineage engraftment with myeloid cells, erythroid cells, B cells, and T cells. Data are shown for a subset of mice. Compiled experience to date is reported in Extended Data Fig. 3. c, Limiting dilution assay of HE-7TF cells during secondary transplantation. Three thousand or 1,000 CD34⁺ cells isolated from bone marrow of primary recipients were transplanted to secondary recipients. Multi-lineage engraftment was counted as response. Confidence intervals of 1/(stem cell frequency) were calculated by ELDA (http://bioinf.wehi.edu.au/software/elda/) according to Poisson distribution. Limiting dilution assay of cord blood was used as reference. d, t-Distributed stochastic neighbour embedding (t-SNE) plot of inDrops single-cell RNA-seq data for CD34⁺CD38-CD45⁺ HSPC populations recovered from bone marrow of mice engrafted with HE-7TF cell- or cord blood HSC-engrafted at 12 weeks. Analysis is anchored on 62 canonical haematopoietic genes (see Methods). Data shown as mean ± s.d.

Figure 3. Characterization of differentiated haematopoietic cells in engrafted mice

a, Adult β -globin expression in engrafted erythroid cells. Human GLY-A⁺ cells were isolated from bone marrow of NSG mice engrafted with HE-7TF cells at 8 weeks, and analysed by quantitative PCR with reverse transcription (qRT-PCR) to quantify embryonic (HBE), fetal (HBG), and adult β -globin (HBB) transcripts. CB, GLY-A⁺ erythroid cells from cord-blood-engrafted in NSG bone marrow; 7F cells, GLY-A⁺ erythroid cells from HE-7TF cell-engrafted in NSG bone marrow; iPS-HPC, GLY-A⁺ erythroid cells from hPSCs transduced with ERG, RORA, HOXA9, SOX4, and MYB. b, Enucleation of engrafted erythroid cells. Bone marrow of NSG mice engrafted with HE-7TF cells at 8 weeks was analysed for human GLY-A and SYTO60. Representative cytospin images of red blood cells from GLY-A⁺ populations separated by SYTO60 nuclear staining are shown. N = 9 measurements from three independent experiments performed each with three technical replicates. c, Neutrophils. Human CD45⁺ cells from bone marrow of NSG mice engrafted with HE-7TF cells at 8 weeks, analysed for human PECAM and CD15. Myeloperoxidase activity of isolated CD45⁺ PECAM⁺ CD15⁺ neutrophils was measured with or without PMA stimulation. Neutrophils from NSG engrafted with cord blood HSCs were used as reference. The basal level of MPO of haemogenic endothelium was 0.40-fold less than cord blood (P = 0.036). PMA stimulation increased MPO production 2.5-fold (P = 0.010) (cord blood) and 3.0-fold (P = 0.10) (seven transcription factor). Stimulated MPO production of HE-7TF was 0.47-fold (0.049) versus cord blood. *P < 0.05. N = 6 measurements from two independent experiments performed each with three technical replicates. d, Human immunoglobulin. Serum was isolated from NSG mice engrafted with cord blood HSCs or HE-7TF cells at 8 weeks and 14 weeks. Production of IgM (8 weeks) and IgG (14 weeks) was measured by enzyme-linked immunosorbent assay (ELISA). Serum from mock transplant and NSG engrafted with cord blood HSCs was used as reference. Human ovalbumin-specific IgM and IgG after immunization with ovalbumin (ova) in serum of mice engrafted with HE-7TF cells, as indicated (right two bars of each panel). *P < 0.05. N = 6 measurements from two independent experiments (three for cord blood versus HE) performed with three technical replicates. e, Production of IFN-γ from human CD3⁺ cells isolated from bone marrow of NSG mice engrafted with cord blood HSCs and HE-7TF cells at 8 weeks, and cultured with or without PMA/ionomycin stimulation for 6 h. CD3⁺ T cells from NSG engrafted with cord blood HSCs were used as reference. The basal level of IFN-γ of haemogenic endothelium was 0.53-fold (P = 0.073) versus cord blood. PMA stimulation increased IFN-γ production 4.4-fold (P = 0.17) (cord blood) and 3.0-fold (P = 0.16) (haemogenic endothelium). Stimulated IFN-γ production of haemogenic endothelium was 0.36-fold (0.039) versus cord blood. IFN-γ production from cord blood HSCs and haemogenic endothelium themselves are shown as reference. *P <0.05. N = 6 measurements from two independent experiments performed with three technical replicates. f, TCR repertoire of engrafted T cells. Human CD3⁺ thymocytes of NSG mice engrafted with HE-7TF cells at 8 weeks were analysed by immunoSEQ to detect TCR rearrangement. CD3⁺ thymocytes from cord blood HSC-engrafted NSG were used as a reference.