Transient inhibition of mTOR in human pluripotent stem cells enables robust formation of mouse-human chimeric embryos

Abstract

It has not been possible to generate naïve human pluripotent stem cells (hPSCs) that substantially contribute to mouse embryos. We found that a brief inhibition of mTOR with Torin1 converted hPSCs from primed to naïve pluripotency. The naïve hPSCs were maintained in the same condition as mouse embryonic stem cells and exhibited high clonogenicity, rapid proliferation, mitochondrial respiration, X chromosome reactivation, DNA hypomethylation, and transcriptomes sharing similarities to those of human blastocysts. When transferred to mouse blastocysts, naïve hPSCs generated 0.1 to 4% human cells, of all three germ layers, including large amounts of enucleated red blood cells, suggesting a marked acceleration of hPSC development in mouse embryos. Torin1 induced nuclear translocation of TFE3; TFE3 with mutated nuclear localization signal blocked the primed-to-naïve conversion. The generation of chimera-competent naïve hPSCs unifies some common features of naïve pluripotency in mammals and may enable applications such as human organ generation in animals.

Fig. 1. Converting hPSCs from primed to naïve pluripotency.

(A to H) Localization of TFE3 in primed H9 hESCs treated with 10 μM Torin1 for the indicated durations. DAPI, 4′,6-diamidino-2-phenylindole. (I to L) Primed H9 (I) were converted with the protocol in (J) to dome-shaped colonies (K; red arrows), which were picked and passaged (L). (M to T) Expression of pluripotency markers in nH9 cultured in 2iLI medium. (U to X) In 2iLI medium, primed H9 differentiated (U) and lost the expression of pluripotency markers (V to X). White scale bars, 10 μm; red scale bars, 100 μm. (Y to AB) Spontaneous differentiation of nH9 in vitro through embryoid bodies (EB) (Y) to cells of all three germ layers (Z to AB). (AC to AE) Naive H9 grafted in kidney capsules of severe combined immunodeficient (SCID) mice produced teratomas.

Fig. 2. Cellular properties of naïve hPSCs.

(A to D) Clonal efficiencies of naïve and primed H9 and RUES2 cultured in 5% O₂ (A and B) or 21% O₂ (C and D) in the absence (A and C) or presence (B and D) of ROCK inhibitor. *P < 0.05 and #P < 0.05, n = 3, unpaired, two-tailed t test versus naïve H9 (nH9) or naïve RUES2 (nRU), respectively. (E) Growth curve and cell doubling time of primed and naïve H9 and RUES2. *P < 0.05 and #P < 0.05, n = 4, repeated-measures analysis of variance (ANOVA) versus primed H9 or primed RUES2 (RU), respectively. (F to I) Primed H9 and nH9 were live stained with TMRE to detect mitochondrial inner membrane potential (F and G) or MitoTracker to locate mitochondria (H and I). (J to M) Mitochondrial respiration in primed versus naïve H9 (J) or RUES2 (K) hESCs was quantified (L and M) in a Seahorse analyzer. *P < 0.05, n = 3, unpaired, two-tailed t test versus primed state. OCR, oxygen consumption rate; FCCP, carbonyl cyanide p-trifluoromethoxyphenylhydrazone; ATP, adenosine 5′-triphosphate.

Fig. 3. Transcriptomic profiles of naïve hPSCs.

(A) H3K27Ac chromatin immunoprecipitation assays showing the utilization of OCT4 distal enhancer (DE) and proximal enhancer (PE) in primed H9 and nH9. P < 0.05, n = 6 versus H9. (B and C) PCA (B) and clustering analysis (C) of RNA-seq data from naïve (Hu_N; blue triangles) and primed (Hu_P; blue dots) H9 and RUES2 against data on single cells from human late blastocysts (Ya_LB; black triangles) (27); human E5 to E7 embryos (Pe_E5, Pe_E6, and Pe_E7; brown pluses) (28); hESCs (Ya_ESC; black dots) (27); or bulk RNA-seq data from naïve hPSCs maintained in chemical inhibitors [Sa_N (red triangles) (29), Ta_N (green triangles) (8), Ch_N (cyan triangles) (7), and Gr_N (light blue triangles) (30)] and their parental primed hPSCs [Sa_P (red dots) (29), Ta_P (green dots) (8), Ch_P (cyan dots) (7), and Gr_P (light blue dots) (30)]. Primed hPSCs (represented by dots) are named by first two letters of the first author’s family name, followed by _P. Naive hPSCs (represented by triangles) are named accordingly with _N. Matching color of dot and triangle indicates the same genome. (D) Coding genes that were differentially expressed between our naïve (underlined green) and primed (underlined red) H9 and RUES2 as compared with expression patterns in other naïve and primed hPSCs. (E) Transposable elements differentially expressed between our naïve (underlined green) and primed (underlined red) H9 and RUES2 as compared with expression patterns in other naïve and primed hPSCs. (F) Increased expression of TEs such as HERVK and LTR5_Hs in our naïve cells (Hu_*; blue) and other naïve cells as compared with their corresponding primed hPSCs.

Fig. 4. Reactivation of X-inactivated genes in naïve hPSCs.

(A to H) X chromosome inactivation in primed RUES2 (A to D) and reactivation in naïve RUES2 (E to H) as revealed by costaining for H3K27me3 (A and E), NANOG (B and F), their merged images with DAPI (C and G), or XIST RNA FISH (D and H). (I) Ratio of the expression level of genes on each chromosome in naïve versus primed hPSCs. X_i, X-inactivated genes (37); X_e, X-escaped genes (37). (J and K) SNP-based allelic expression analysis of X-inactivated genes (J) and X-escaped genes (K) showed that monoallelically expressed X-inactivated genes in primed H9 were expressed biallelically in nH9 (nH9) (J), while genes escaped from X inactivation were always biallelically expressed in H9 and nH9 (K). The two colors represented expression levels of a gene from two different alleles in the three replicates in the RNA-seq data.

Fig. 5. Decreased DNA methylation in naïve hPSCs.

(A to B‴) Costaining for 5mC (A and B), 5hmC (A′ and B′), and DAPI (A″ and B″) in primed H9 (A to A‴) and nH9 (B to B‴). Scale bars, 10 μm. (C to F) Dot blots (C and E) and quantification (D and F) of 5mC (C and D) and 5hmC (E and F) levels in genomic DNA isolated from primed and naïve H9, and AB2.2 mESC. *H9 versus mESC; ^H9 versus nH9; #nH9 versus mESC, all at P < 0.05, n = 3, unpaired, two-tailed t test. au, arbitrary units. (G) PCA of genome-wide DNA methylation in primed and naïve H9 and RUES2 using β values of all probes in Infinium MethylationEPIC BeadChip. (H) Comparison of DNA methylation levels in the 128,383 tiling regions that were differentially methylated between primed and naïve H9 and RUES2. (I) Comparison of DNA methylation levels in imprinted regions (38) between primed and naïve H9 and RUES2. H9, primed H9; nH9, naïve H9; RU, RUES2; nRU, naïve RUES2.

Fig. 6. Robust formation of mouse-human chimeric embryos.

(A) Large amounts of GFP⁺ human cells were found in an E17.5 mouse embryo (nN004-2) derived from mouse blastocysts injected with GFP-labeled naïve N004 iPSCs (injection #6). (B to E) At a different z-level from (A), two neighboring sections of this embryo were DAB stained with anti-GFP (B) or stained with H&E (C). Boxes 1 (heart) and 2 (retina) in (B) correspond to boxes 1 and 2 in (C), which are enlarged in (D) and (E), respectively. Areas highlighted by arrows and box 1 contained GFP⁺ (B) RBCs (C and D). Box 2 contained GFP⁺ (B) retinal pigmented epithelium (C and E). (F to J‴) Different sections of this embryo were DAB stained with antibody against hRBC (F) or immunostained for GFP (G to J), hRBC (G′; mesoderm), the RBC marker Band 3 anion transporter (H′; mesoderm), AFP (I′, endoderm), and the photoreceptor marker recoverin (J′; ectoderm). (G) to (G‴) correspond to boxed areas of fig. S4 (A to A‴). Boxed area in (J‴) is enlarged in fig. S4B. Scale bars, 10 μm. (K to M) PCR detection of GFP (K) or human-specific DNA using DNA fingerprinting primers TPA-25 (L) or D1S80 (M) in genomic DNA isolated from E17.5 mouse embryos derived from mouse blastocysts injected with GFP-labeled naïve RUES2 (embryos 1 to 14; injection #12) or unlabeled naïve RUES2 (embryos i to iv; injection #14). GFP, GFP plasmid as positive control; RU, genomic DNA from RUES2. (N) qPCR measurements of hmtDNA normalized against ultraconserved noncoding element (UCNE) in genomic DNA isolated from the above samples (1 to 14 and i to iv; green bars), C57BL/6 mouse genomic DNA (Ms) and RUES2 human genomic DNA serially diluted in C57BL/6 mouse genomic DNA (red bars). Red dotted line highlights level equivalent to 1:1000 diluted standard. (O) Quantification of human and mouse 18S rDNA by NGS of PCR amplicons of the positive samples in (N) (green bars), C57BL/6 mouse genomic DNA (Ms), and RUES2 human genomic DNA serially diluted in C57BL/6 mouse genomic DNA (red bars). Sequences of the human and mouse amplicons are identical at the primer binding sites on both ends and diverge by 9 bp in the middle of the amplicons. This enables unbiased PCR amplification of human and mouse DNA and their absolute quantification by counting human and mouse reads in NGS.

Fig. 7. Nuclear localization of TFE3 is critical for primed to naïve conversion.

(A to F) Primed H9 hESCs stably expressing TFE3-GFP fusion proteins (A to C) or NLS-mutated TFE3-GFP (NLS-GFP) (D to F) were costained for GFP (A and D) and DAPI (B and E). (G to Q) Primed H9 hESCs stably expressing TFE3-GFP (G to J) or NLS-GFP (K to N) went through the conversion protocol in Fig. 1J and were stained at day 5 for GFP (G and K), NANOG (H and L), and hNA (I and M). Merged images (J and N) were quantified for the percentage of NANOG⁺ cells in human nuclear antigen (hNA⁺) cells (O). *P < 0.05, n = 4, one-way ANOVA versus control H9. Phase-contrast images of H9 expressing TFE3-GFP (P) or NLS-GFP (Q) were acquired at day 5 of conversion. Scale bars, 10 μm. (R to V) HEK293 cells transfected with MYC-TFE3 alone (R and T) or together with NLS-GFP (S and U′) were treated with vehicle (R to S′) or Torin1 (10 μM for 3 hours) (T to U′) and stained as indicated. Merged images (S′ and U′) highlighted cells transfected with NLS-GFP. Percentage of cells with MYC-TFE3 in nucleus was quantified for each condition. *P < 0.05, Student’s t test, n = 250 cells per condition. Scale bars, 10 μm.