Reprogramming mouse fibroblasts into engraftable myeloerythroid and lymphoid progenitors

Abstract

Recent efforts have attempted to convert non-blood cells into hematopoietic stem cells (HSCs) with the goal of generating blood lineages de novo. Here we show that hematopoietic transcription factors Scl, Lmo2, Runx1 and Bmi1 can convert a developmentally distant lineage (fibroblasts) into 'induced hematopoietic progenitors' (iHPs). Functionally, iHPs generate acetylcholinesterase⁺ megakaryocytes and phagocytic myeloid cells in vitro and can also engraft immunodeficient mice, generating myeloerythoid and B-lymphoid cells for up to 4 months in vivo. Molecularly, iHPs transcriptionally resemble native Kit⁺ hematopoietic progenitors. Mechanistically, reprogramming factor Lmo2 implements a hematopoietic programme in fibroblasts by rapidly binding to and upregulating the Hhex and Gfi1 genes within days. Moreover the reprogramming transcription factors also require extracellular BMP and MEK signalling to cooperatively effectuate reprogramming. Thus, the transcription factors that orchestrate embryonic hematopoiesis can artificially reconstitute this programme in developmentally distant fibroblasts, converting them into engraftable blood progenitors.

Figure 1. Generating induced hematopoietic progenitors from wild-type MEFs.

(a) Schema of experimental design. MEFs (P0) were purified by sorting out any contaminating hematopoietic cells and passaged to P2-3 before experiments. iHP cells induced from MEFs by reprogramming factors were used for further characterization and evaluation; negative control cultures were transduced with empty vector (EV) only. (b) Representative ‘cobblestone' colonies at 24 dpi induced by lentiviral FuW-TetO vectors carrying 7F. Representative of three independent experiments. Scale bar: 100 μm. (c) CFU colonies derived from 27 dpi FuW-TetO-7F-induced iHP cells, representative of three independent experiments. Scale bar: 100 μm for GM, 50 μm for GEMM and E colonies. (d) Frequency of different type of CFU colonies derived from 27 dpi FuW-TetO-7F-induced iHP cells, with dox added (or withheld) at the beginning of CFC assays as indicated. Data are shown as mean±s.d. of four biological replicates from two independent experiments. (e) Representative ‘cobblestone' colonies (27 dpi) induced by different combinations of factors: SL, SLB, SLHR or SLBR (with factors singly delivered in individual pMX vectors), representative of three independent experiments. Scale bar: 100 μm. (f) Different types of representative CFU colonies derived from 27 dpi SLHR-iHP in CFC assays (factors were singly delivered in individual FuW-TetO vectors), representative of two independent experiments. Scale bar: 100 μm for GM/G colonies; 50 μm for GEMM/E/M/Mk colonies. (g) Frequencies of different types of colonies derived from FACS-sorted Kit⁺CD41⁻, Kit⁺CD41⁺, Kit⁻CD41⁺ and Kit⁻CD41⁻ subsets of SLHR-iHP cells in CFC assays. Factors were singly delivered in individual FuW-TetO vectors. Data shown are mean±s.d. of biological triplicates. (h) Benzidine positive cells in a GEMM colony and adult (β-major) globin expression in CFU-E colonies as shown in f, representative of two independent experiments. BM: bone marrow cells. FL: E12.5 foetal liver cells. Data shown are mean±s.d. of technical triplicates. Scale bar, 100 μm. (i) Images of AchE⁺ megakaryocyte-containing colonies (CFU-mix and CFU-Mk) and phagocytic CD45⁺ cells, representative of three independent experiments. 27 dpi SLHR-iHPs (induced using pMX vectors) were used for CFU-MK assay. Scale bar, 200 μm for CFU-mix/Mk colonies; 100 μm for phagocytosis picture.

Figure 2. Runx1 together with HoxB4 or Bmi1 augments the CFU-S forming-ability of iHP cells.

(a) Similar to control bone marrow cells (BM Ctr, tdTomato⁺), SLHR-iHP cells (tdTomato⁺) form tdTomato⁺ nodules in the spleen at 12 days post-transplantation (dpt) into lethally irradiated C57Bl/6 mice (CFU-S12). Factors were singly delivered in individual pMX constructs. 1 × 10⁵ BM control cells, 5 × 10⁶ SLHR-iHP cells or 1 × 10⁶ Kit⁺ cells were transplanted per mouse. Mice analysed: BM Ctr, n=4, SLHR-iHP cells, n=20, SLHR-iHP kit⁺ cells, n=4. These are representative of three independent experiments. Scale bar, 2 mm. (b) Comparison of CFU-S12 potential of iHP cells induced by differing TF cocktails. SL-, SLB-, SLHR-, SLRH- or SLRB-iHP cells (tdTomato⁺) were transplanted into lethally irradiated SCID mice. Factors were delivered in pMXconstructs; SLHR denotes individual delivery of S, L, H and R; SLRH and SLRB denotes use of a polycistronic construct containing S, L and R in one pMX vector, together with individual delivery of either H or B in a separate construct. For SL-, SLB-, SLHR-iHP, 5 × 10⁶ cells were transplanted. For SLRH- and SLRB-iHPs, 2 × 10⁶ cells were transplanted. Mice analysed: for SL-, SLB-, SLHR-iHP, n=6, for SLRB- and SLRH-iHP, n=12 each. Scale bar: 2 mm. These are representative of three independent experiments. (c) Frequency of CFU-S12 of different iHP cells. iHP cells are named as in b. Data are shown as mean±s.d. per spleen. These data are from three independent experiments. (d) Representative FACS analysis of tdTomato⁺ cells in BM of SCID mice at 12–14 dpt. SLR factors were delivered in one polycistronic construct. For SLRB/H-iHP cells: 2 × 10⁶ cells per mouse were transplanted; for Ctr BM cells (tdTomato⁺), 2 × 10⁵ cells per mouse were transplanted. Percentage of tdTomato⁺ cells are shown as mean±s.d. (n=6 mice for each type of cells), representative from three independent experiments. (e) Representative FACS analysis of tdTomato⁺ cells stained with lineage markers (shown on the plots) in the BM of SCID mice transplanted with either SLRB/SLRH-HP cells and control BM cells (Ctr tdTomato⁺) at 12–14dpt. SLR factors were delivered in one polycistronic construct. Mice analysed: n=6 for each type of cells. Ery, Erythroid; Meg, Megakaryocytes. These data are representative of three independent experiments.

Figure 3. SLRB-iHP cells engraft for up to 4 months in vivo.

(a) Representative FACS plot of SLRB/SLRH-iHP (tdTomato⁺) cells in PB and SP at 16 wpt (Wpt: week post-transplant). SLR factors were delivered in one polycistronic construct. 5 × 10⁶ iHP cells were transplanted per mouse. Mice analysed: SLRB iHP (n=9), SLRH-HP (n=4). These are representative of three independent experiments. (b) Summary of SLRB/SLRH-iHP (tdTomato⁺) cell engraftment in PB at 5 and 16 wpt. SLR factors were delivered in one polycistronic construct. For SLRB-iHP, NOD-SCID (NS) mice n=6 (5 wpt) and n=9 (16 wpt) were analysed. For SLRH-iHP, different mouse strains were analysed. At 5wpt, NS mice n=5, NSG mice n=3, C57BL/6 (B6) mice n=3. At 16 wpt, NS mice n=3, NSG mice n=1, B6 mice n=2. Only tdTomato⁺ cell contribution ≥0.1% are plotted. Data drawn from three independent experiments. (c) Summary of SLRB/SLRH-iHP (tdTomato⁺) cell engraftment in SP at 5 and 16 wpt. SLR factors were delivered in one polycistronic construct. For SLRB-iHP cells, mice n=5 at 5wpt, and n=9 at 16 wpt. For SLRH-iHP cells, mice analysed: n=5 at 5wpt, n=4 at 16 wpt. Only tdTomato⁺ cells contribution ≥0.1% were plotted. These data are from three independent experiments. (d) Representative FACS plot of multilineage reconstitution of SLRB/H-iHP (tdTomato⁺) cells in SP at 16 wpt. SLR factors were delivered in one polycistronic construct. Percentage of cells are shown as mean±s.d. (for SLRB, n=6 mice; for SLRH, n=3 mice). These data are representative of three independent experiments. (e) SLRB-iHP (tdTomato⁺) cells in PB at 16 wpt. SLR factors were delivered in one polycistronic construct. tdTomato⁻ PB cells from respective CD45.1⁺ recipient mice shown as control (Ctr). These data are representative of three independent experiments.

Figure 4. SLRB-iHPs generate B cells in vivo.

V(D)J recombination events and transgene integration were detected in SLRB-iHP-derived B cells (iHP-B cells) at single-cell level. iHP-B cells are from 16 wpt spleens. Numbers at top of the panel are cell ID of individual single cells. Con (B6) denotes control (Con) cells from the SP of C57BL/6 (B6) mice. MEF (14 dpi) are cells from SLRB infected tdTomato⁺ MEF at 14 dpi. Sp (B6) and Sp (tdTomato⁺) stand for total spleen cells. V_H7183-DJ_H and D_HQ52-J_H are used to reveal V to DJ and D to J recombination events at heavy chain, respectively. V_k−J_k denotes V to J recombination at Kappa chain. WT (Wild type) and MT (tdTomato mutant) show the genotype of the cells. SLR-T and B-T stand for transgene integration of SLR (Polycistronic) and Bmi1. Polycistronic SLR was used for the generation of iHP-B cells and 14 dpi MEF. These are representative of two independent experiments.

Figure 5. Transcriptional dynamics and Lmo2 genomic binding during iHP reprogramming.

(a) Schema of the transdifferentiation procedure with days indicating when samples were collected for microarray and ChIP-Seq analyses. (b) Hierarchical clustering of D0, D4, iHP (Kit⁺ and CD45⁺ cells) and BM (Kit⁺ and CD45⁺ cells) with published HSPC microarray datasets. Red boxes denote the samples generated in this study whereas the other samples are HSPCs from published datasets. In brief, HSPCs from published datasets are VE-cadherin⁺CD45⁺ cells in AGM, Lin⁻Sca1⁺Kit⁺VE-cadherin⁺Mac-1^low cells in E12.5 foetal liver, CD45⁺Kit⁺CD34^mid cells in E12.5 placenta, Lin⁻Sca1⁺Kit⁺CD150⁺CD48⁻ cells in E13.5 and E14.5 foetal liver, CD41⁺Kit⁺CD34⁺ cells in yolk sac, Kit⁺CD41⁺ cells in day 6 embryoid bodies (EB), and CD41^brightCD45⁻CD34⁻ ESC-derived HSC-like cells. (c) Dynamic gene expression during the microarray timecourse, as assayed by GEDI. The GEDI plots demonstrate a transition state in 14 dpi cells and a major change of the gene expression profile at 26 dpi cells during reprogramming. (d) Motif analysis for Lmo2-bound peaks at 4 dpi shows an enrichment of motifs recognized by other hematopoietic transcription factors. (e) Clustered heatmaps of MEF H3K4me3, H3K27ac and H3K27me3 signals at genomic regions bound by Lmo2 at 4 dpi. The heatmaps indicate that 4 dpi Lmo2 binds both closed and open chromatin. (f) GO analysis of Lmo2-bound genes; bar plots indicate P values of the overrepresented pathways and biological processes.

Figure 6. Lmo2 binds to Hhex and Gfi1 to effectuate iHP reprogramming.

(a) Venn diagram demonstrating the numbers of 4 dpi Lmo2-bound genes that are upregulated at 4 dpi cells and GO analysis of the 82 genes that both Lmo2-bound and 4 dpi upregulated. (b) Mouse genome screenshots demonstrating binding of Lmo2 on the Gfi1 enhancer and the Hhex promoter at 4 dpi. (c) Gfi1 and Hhex were upregulated at early stages of SLRB-iHP reprogramming; SLR factors were delivered in one polycistronic construct. Data shown are mean±s.d. of technical triplicates, representative of two independent experiments. (d) shRNAs against Hhex or Gfi1 were introduced into fibroblasts 2 days before introducing the SLRB reprogramming factors; SLR factors were delivered in one polycistronic construct. Images are hematopoietic colonies representative of five independent experiments. Scale bar: 100 μm. (e) Knockdown of Gfi1 or Hhex perturbs iHP reprogramming. 50,000 (50 K) MEFs were infected with SLRB and hematopoietic colonies (including cobblestone colonies and clusters of round cells, as shown in d) were enumerated at 21 dpi. SLR factors were delivered in one polycistronic construct. Data shown are mean±s.d. of five independent experiments. GFP shRNA was used as a non-targeting control; *P<0.001, **P<0.003 (two-tailed Student's t-test).

Figure 7. BMP and MAPK/ERK signalling cooperate to drive iHP reprogramming.

(a) Venn diagram demonstrating the numbers of 4 dpi Lmo2-bound genes that were upregulated in 14 dpi cells. The box demonstrates enrichment of genes associated with various signalling pathways that both Lmo2-bound (4 dpi) transcriptionally upregulated (14 dpi). (b) Mouse genome screenshots demonstrating the binding of 4 dpi Lmo2 onto the promoters of various signalling components (for example, Bmp4, Smad5, BMP4 cascade; Gab1, MAPK-MEK cascade). These are representatives from a single experiment. (c) Addition of signalling pathway modulators during SLRB iHP reprogramming reveals that specific inhibition of either BMP or MEK pathways abrogates iHP formation; SLR factors were delivered in one polycistronic construct. LDN193189 (0.4 μM), DMH1 (1 μM), PD0325901 (0.8 μM) and GSK1120212 (0.4 μM) were added every other day from 1–21 dpi. These data are from four independent experiments. (d) BMP signalling is critical for iHP reprogramming at early stage (from 1–7 dpi), while MEK signalling is required throughout SLRB iHP reprogramming. SLR factors were delivered in one polycistronic construct. Pictures shown exemplify hematopoietic colonies observed at 21 dpi. The effects of various small-molecule inhibitors on hematopoietic colony formation was normalized to DMSO-treated control cultures (21 dpi). Images are representative of four independent experiments. Scale bar: 100 μm. Data shown were mean±s.d. of four independent experiments. Chemicals were added at different time windows (I, II, III, IV and V) as diagrammed in c.

Figure 8. Model of hematopoietic reprogramming.

Scl, Lmo2, Runx1 might act as ‘lineage commitment' regulators whereas Bmi1 or Hoxb4 might be ‘self-renewal' factors in HSC development. The hematopoietic reprograming activity of these transcription factors also jointly requires extracellular signals mediated through the BMP and MEK cascades. Finally, Lmo2 activates other hematopoietic transcription factors (for example, Gfi1 and Hhex) to drive iHP reprogramming.