|Home | About | Journals | Submit | Contact Us | Français|
Pluripotent stem cells, including induced pluripotent and embryonic stem cells (ESCs), have less developed mitochondria than somatic cells and, therefore, rely more heavily on glycolysis for energy production.1-3 However, how mitochondrial homeostasis matches the demands of nuclear reprogramming and regulates pluripotency in ESCs is largely unknown. Here, we identified ATG3-dependent autophagy as an executor for both mitochondrial remodeling during somatic cell reprogramming and mitochondrial homeostasis regulation in ESCs. Dysfunctional autophagy by Atg3 deletion inhibited mitochondrial removal during pluripotency induction, resulting in decreased reprogramming efficiency and accumulation of abnormal mitochondria in established iPSCs. In Atg3 null mouse ESCs, accumulation of aberrant mitochondria was accompanied by enhanced ROS generation, defective ATP production and attenuated pluripotency gene expression, leading to abnormal self-renewal and differentiation. These defects were rescued by reacquisition of wild-type but not lipidation-deficient Atg3 expression. Taken together, our findings highlight a critical role of ATG3-dependent autophagy for mitochondrial homeostasis regulation in both pluripotency acquirement and maintenance.
Pluripotent stem cells (PSCs), including induced pluripotent and embryonic stem cells (ESCs), hold great promise for regenerative medicine due to their unlimited self-renewal capability and their ability to differentiate into any cell types of 3 germ layers in our bodies. PSCs are characterized by their ability to self-renew, pluripotency, immortalization in culture, high proliferation rate, and short G1 phase.4,5 The maintenance of intracellular homeostasis in ESCs requires timely clearance of damaged organelles and toxic proteins, as well as quick synthesis of total biomass associated with their high proliferation rate. Autophagy is a conserved degradation pathway by which organelles and cytoplasm are sequestered and subsequently delivered to lysosomes for hydrolytic digestion.6,7 Autophagy serves as a protective strategy to eliminate toxic cytoplasmic contents, preventing cellular damage in response to stress.8 In addition, autophagy also serves as a dynamic recycling system that produces new building blocks and bioenergy to fuel cellular renovation and homeostasis during normal development and differentiation.6,9
Mitochondria are dynamic double-membrane organelles that play critical roles in multiple biological processes, such as energy production, apoptosis, and signal transduction.10,11 To meet the different bioenergetic demands of distinct cell types, cells regulate mitochondrial homeostasis through either biogenesis and degradation or dynamic fusion and fission.12 Nuclear reprogramming induces both structural and functional remodeling of parental mitochondria in somatic cells to a state typical of ESCs.13,14 However, how mitochondrial remodeling matches the requirements of somatic reprogramming and how ESCs harness their mitochondria to regulate pluripotency and self-renewal have not been clearly defined. Herein, we have determined that ATG3-dependent autophagy is an executor for both mitochondrial remodeling during reprogramming and mitochondrial homeostasis regulation in ESCs, and is pivotal for pluripotency acquirement and maintenance.
During the acquisition of pluripotency, specific mitochondrial remodeling must occur to meet the energetic and anabolic requirements of the pluripotent states. To examine how mitochondria are remodeled during reprogramming, we first monitored changes in mitochondrial morphology and number using transmission electron microscopy and mitochondrial mass using MitoTracker staining of mouse embryonic fibroblasts (MEFs) and PSCs. While mitochondria in MEFs possess well-developed morphology with dense cristae and an elongated shape, PSC mitochondria have an immature morphology with a globular shape and low-density cristae (Fig. 1A). Furthermore, the number of mitochondria in PSCs was significantly lower than that in MEFs (Fig. 1B). Accordingly, the total mitochondrial mass in PSCs was significantly lower than that of MEFs (Fig. 1C). Then, we isolated FUT4/SSEA-1-positive cells at reprogramming d 10 and iPSCs for mitochondrial mass and mtDNA copy number determination (Fig. 1D). The established iPSCs have normal karyotype, express pluripotent markers, form teratomas and contribute to chimeric mice, supporting their pluripotency (Fig. S1). The total mitochondrial mass of reprogramming cells gradually decreased in parallel with reprogramming progression (Fig. 1E), indicating mitochondria are removed during reprogramming. In support of these observations, a quantitative polymerase chain reaction (PCR) assay using mitochondrial DNA (mtDNA) as a template was performed to track mtDNA copy number changes during reprogramming. As expected, genomic mtDNA gradually decreased as reprogramming progressed, confirming removal of mitochondria during this process (Fig. 1F).
Because autophagy degrades both proteins and organelles, we next assessed whether autophagy was involved in mitochondrial removal during reprogramming. The autophagy inhibitors 3-methyladenine (3-MA) and bafilomycin A1 (Baf-A1) were added at the beginning of reprogramming for 3 d to block autophagic flux. Interestingly, both 3-MA and Baf-A1 treatments significantly inhibited mitochondrial removal without affecting proliferation and apoptosis, indicating autophagy contributes to mitochondrial removal during reprogramming (Fig. 1G; Fig. S2A and S2B). As a result, reprogramming efficiency was significantly reduced by 3-MA and Baf-A1 treatments (Fig. 1H).
ATG3 is a membrane-curvature sensor and plays a key role in autophagosome formation.15,16 To further investigate whether autophagy directly contributes to mitochondrial remodeling during reprogramming, we isolated both Atg3+/+ and atg3−/− MEFs (Fig. 2A). Atg3 deletion did not affect MEF proliferation and apoptosis (Fig. S2C and S2D). Unlike wild-type Atg3+/+ MEFs, we did not observe autophagosomes containing mitochondria in atg3−/− knockouts upon treatment with carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP), indicating a lack of autophagy-mediated mitochondrial removal in atg3−/− MEFs (Fig. S3A). We then transfected Atg3+/+ and atg3−/− MEFs with Yamanaka factors and monitored mitochondrial removal during reprogramming. Upon reprogramming initiation, mitophagy activation in Atg3+/+ but not atg3−/− MEFs was observed (Fig. 2B, C; Fig. S3B). Although a decrease in the number of mitochondria occurred during reprogramming in both Atg3+/+ and atg3−/− MEFs, the relative decrease of mitochondria in wild-type MEFs was significantly higher than that of atg3−/− MEFs, indicating ATG3-dependent mitophagy contributes to mitochondrial remodeling during reprogramming (Fig. 2D; Fig. S3C). Accordingly, the reprogramming efficiency was significantly inhibited (Fig. 2G).
An amphipathic helix domain at the N-terminus of ATG3 is responsible for membrane-curvature sensing and is critical for lipidation of the MAP1LC3/GABARAP family of autophagy proteins.15 Mutation of this domain leads to failure of autophagosome formation.15 To further confirm that impaired reprogramming of atg3−/− MEFs results from a defect in autophagy, gain-of-function assays were performed by overexpression of both wild-type and lipidation-deficient Atg3V8D in atg3−/− MEFs (Fig. 2E). We found ectopic expression of wild-type Atg3 but not lipidation-deficient mutant Atg3V8D in atg3−/− MEFs can rescue both defective mitochondrial clearance and reprogramming efficiency, supporting the view that ATG3-dependent autophagy is essential for mitochondrial remodeling and reprogramming (Fig. 2F, G).
We next investigated whether lack of ATG3-dependent autophagy affected the established iPSCs. By transmission electronic microscopy, we found that the abnormal mitochondria were accumulated in Atg3-deficient iPSCs and the mitochondrial cristae showed a blurred morphology (i.e., lack of clear definition) compared to Atg3+/+ iPSCs (Fig. 2D, H). Further investigation identified significantly lower mitochondrial membrane potential and oxygen consumption, and compromised ATP production in atg3−/− versus Atg3+/+ iPSCs, indicating the unsuccessful metabolic reprogramming of atg3−/− knockouts (Fig. 2I, J, K). Taken together, these data support the conclusion that ATG3-dependent autophagy contributes to both reprogramming efficiency and reprogramming quality and is required for successful reprogramming.
To investigate whether autophagy is involved in the regulation of mitochondrial dynamics and/or affects ESC identity, we first designed specific small interfering RNA (siRNA) targeting the autophagy regulator Atg3 and found transient inhibition of this gene leads to ESC differentiation and loss of pluripotent gene expression (Fig. S4A-C). Then, we treated ESCs with 3-MA and found autophagy inhibition significantly affected ESC self-renewal (Fig. S4D). These data suggest the involvement of autophagy in ESC identity maintenance. To further investigate how autophagy regulates mitochondrial dynamics in ESC and thereby affects ESC identity, Atg3+/+ and atg3−/− ESCs were isolated from the blastocyst of mice at embryonic d 3.5 (Fig. 3A). Upon FCCP treatment, we detected mitochondria associated with autophagic structures in wild-type Atg3+/+ but not atg3−/− ESCs (Fig. 3B; Fig. S5B and S5C), indicating a lack of ATG3-dependent mitophagy in atg3−/− ESCs. Meanwhile, abnormal accumulation of mitochondria was detected in atg3−/− ESCs, as indicated by the existence of increased mtDNA copy number, blurred-cristae mitochondria accumulation, enhanced ROS production and decreased ATP generation in atg3−/− ESCs (Fig. 3C–G). These data suggest ATG3-dependent autophagy plays critical roles to maintain mitochondrial homeostasis in ESCs. As a result, the clonogenic survival of atg3−/− ESCs was significantly impaired, indicating ATG3-dependent autophagic mitochondria removal is pivotal to ESC self-renewal (Fig. 3H).
To further confirm whether the abnormal accumulation of mitochondria and aberrant self-renewal ability of atg3−/− ESCs was directly caused by the loss of ATG3-mediated mitophagy, gain-of-function assays were performed by introducing Atg3 expression into atg3−/− ESCs. We established stable atg3−/− ESC lines carrying an empty vector, wild-type Atg3, and mutant Atg3V8D. As expected, MAP1LC3B-II was only detected in atg3−/− stable ESC lines carrying wild-type Atg3 but not the empty vector or Atg3V8D mutant upon starvation, demonstrating recovery of autophagy in atg3−/− ESCs by expression of wild-type Atg3 (Fig. S5A). Meanwhile, mitophagosomes were again detected in atg3−/− ESCs expressing wild-type Atg3 upon FCCP treatment (Fig. S5B). Consequently, the increased mtDNA copy number in atg3−/− ESCs was significantly reduced by gain of wild-type Atg3 expression, indicating accumulation of mitochondria in Atg3-deficient ESCs was directly caused by defects in ATG3-dependent mitochondrial autophagy (Fig. 3E). Furthermore, ROS production and ATP generation were also rescued by reacquisition expression of wild-type but not V8D mutant Atg3 (Fig. 3F, G). Accordingly, the abnormal self-renewal of Atg3-deficient ESCs was compensated by reintroducing wild-type Atg3 but not lipidation-deficient Atg3V8D mutant (Fig. 3H). Together, these data suggest that ATG3-dependent canonical autophagy maintains ESC mitochondrial homeostasis and self-renewal.
Differentiation of PSCs requires a specific remodeling process involving an increase in mitochondrial numbers. To test whether autophagy is involved in this process, we performed an embryonic body (EB) differentiation assay. The number of mitochondria gradually increased during EB differentiation as expected (Fig. 4A). We next assessed EB differentiation using Atg3+/+ and atg3−/− ESCs. Surprisingly, contrary to the relatively larger number of mitochondria in atg3−/− ESCs, the mitochondrial increase was delayed in atg3−/− ESCs compared to the wild-type during EB differentiation, indicating atg3−/− ESC mitochondria are abnormal and ATG3-dependent autophagy is crucial for mitochondrial remodeling during EB differentiation (Fig. 4B, C; Fig. S6A and S6B). Furthermore, decreased pluripotent gene expression was detected in atg3−/− vs. wild-type ESCs, suggesting elimination of ATG3-dependent autophagy leads to the compromised pluripotency in ESCs (Fig. 4D). In support of this assumption, atg3−/− ESCs showed abnormal EB differentiation, characterized by delayed expression of certain endodermic and mesodermic marker genes (Fig. 4E). Furthermore, a teratoma formation assay was employed to investigate the autophagy contribution to ESC differentiation. While both Atg3+/+ and atg3−/− ESCs formed teratomas, the average weight of teratomas formed by Atg3+/+ ESCs is significantly larger than that of atg3−/− ESCs, supporting the view that ATG3-dependent autophagy is critical for differentiation of pluripotent stem cells (Fig. 4F). Together, these data suggest ATG3-dependent autophagy is involved in mitochondrial remodeling during ESC differentiation and regulates ESC pluripotency and differentiation.
In somatic cells, damaged or superfluous mitochondria are selectively recognized and sequestered by phagophores, followed by mitophagosomal fusion with lysosomes and autophagic degradation, a process called mitophagy.17-19 However, how mitochondrial homeostasis is regulated in PSCs is largely unknown. By using an Atg3 knockout model system, we demonstrate that ATG3-dependent autophagy is required for mitochondrial homeostasis regulation in both pluripotency acquirement and maintenance, and plays pivotal roles for ESC differentiation.
Somatic cell reprogramming remodels cellular components and organelles, as well as metabolism patterns. Transition of somatic cell oxidative phosphorylation to glycolysis through downregulation of mitochondrial respiratory chain complexes and upregulation of glycolytic genes has been identified as a critical event during reprogramming.20 In agreement with this model, a PDPK1/phosphoinositide-dependent kinase-1 activator, which facilitates the metabolic conversion from oxidative phosphorylation to glycolysis, in combination with POU5F1/OTF3 and a couple of small molecules could successfully induce pluripotency.21 However, how mitochondria remodel themselves to facilitate the metabolic switching in somatic cell reprogramming is not clear.
Mitochondria number and structure in PSCs were dramatically changed compared to somatic fibroblasts, suggesting a total mitochondrial remodeling process must occur during reprogramming. Although mitochondrial removal has been proposed to be required for nuclear reprogramming, the detailed mechanisms driving this remodeling process have not been clearly defined. Mitochondrial fission and fusion have been proposed to be responsible for mitochondrial remodeling during reprogramming.22,23 Inhibition of mitochondrial fission regulator DNM1L (dynamin 1-like) expression significantly decreased reprogramming efficiency, indicating the potential involvement of mitophagy in somatic cell reprogramming.23 Conversely, a recent study has demonstrated an ATG5-independent noncanonical mitophagy involvement in somatic reprogramming through ULK1 (unc-51 like kinase 1) and RAB9 pathway.24 In contrast, we have demonstrated that ATG3-dependent canonical autophagy contributes to mitochondrial remodeling and regulates reprogramming. Although Atg3 deletion did not completely block mitochondrial clearance and iPSC colony formation, the abnormal mitochondria accumulated in established atg3−/− iPSCs compared to Atg3+/+ iPSCs, leading to enhanced ROS generation and defective metabolic reprogramming. These data support the hypothesis that ATG3-dependent canonical autophagy regulates both mitochondrial quantity and quality and is critical for successful reprogramming. Consistent with this notion, recent studies have shown that transient activation of autophagy through MTOR (mechanistic target of rapamycin [serine/threonine kinase]) suppression either by SOX2 (SRY [sex determing region Y]-box 2) or rapamycin is a critical early step for successful reprogramming.25
Mitochondrial dynamics (fusion and fission) and mitophagy have been proposed to be important players for mitochondrial homeostasis regulation in somatic cells.12 Recent studies have shown that the mitochondrial fusion protein GFER (growth factor, erv1 [S. cerevisiae)-like (augmenter of liver regeneration]), which is highly expressed in ESCs, modulates the mitochondrial fission GTPase DNM1L to preserve mouse ESC mitochondrial homeostasis and function, and maintain pluripotency in ESCs.26 Knockdown of Gfer in ESCs leads to decreased pluripotent marker gene expression accompanied by excessive mitochondrial fragmentation and mitochondrial autophagy, indicating the involvement of mitochondrial dynamics in mitochondrial homeostasis and pluripotency regulation in mouse ESCs.26 By using atg3−/− ESCs, we demonstrate here that ATG3-dependent autophagy is required for ESC mitochondrial homeostasis and ESC identity maintenance. ESCs are endowed with high rates of catabolic processes to rapidly generate multiple building blocks for cellular reconstruction to meet their high proliferation rates. Furthermore, to maintain their identity, rapidly dividing ESCs have to efficiently remove damaged mitochondria to avoid the oxidative damage to their genomes. Atg3 null mouse ESCs accumulate aberrant mitochondria accompanied by enhanced mitochondrial mass, and generate increased levels of ROS and decreased levels of ATP, leading to compromised self-renewal and pluripotency ability. Thus we propose ESCs employ autophagy, by which dysfunctional mitochondria and the resulting excessive ROS can be rapidly removed, to protect their genome from oxidative damage and thus maintain their self-renewal and pluripotency.
In somatic cells, mitochondria can be eliminated either through selective mitochondrial autophagy or nonselective autophagy.17 The selective autophagic elimination of mitochondria has been extensively reported to be regulated by PARK2 and PINK1 (PTEN induced putative kinase 1).17,27-34 We did not observe defective mitochondrial clearance and impaired reprogramming efficiency when we used park2−/− MEFs for the reprogramming assay, indicating PARK2-mediated selective mitochondrial autophagy was not involved in mitochondrial remodeling during reprogramming, and other unidentified mitophagy pathways may play a role in this process.
Our findings significantly contribute to the current understanding of reprogramming and/or ESC identity maintenance in terms of mitochondrial homeostasis regulation, and may provide new strategies for directed differentiation of ESCs by manipulating autophagy. Furthermore, our data highlight the contribution of mitochondria in ESC identity maintenance. In the future, it would be of great interest to investigate how genes that mediate mitochondrial homeostasis are regulated by core pluripotency transcription factors in PSCs upon differentiation or, conversely, during reprogramming.
B6D2-Tg (CAG/Su9-DsRed2, Acr3-EGFP) RBGS002Osb (RBRC03743),35 MAP1LC3B-GFP (RBRC00806),36 and Atg3+/− (RBRC02761)16 mice were purchased from Riken BioResource Center. All protocols used for animal manipulation were approved by the Institutional Animal Care Committee. JC-1 (40705ES03), MitoTracker Green (40742ES50), and MitoTracker Red (40743ES50) were purchased from Yeasen. The anti-MAP1LC3B antibody (Medical and Biological Laboratories Co., PM036) was used at 1:200; Alexa Fluor® 488 donkey anti-rabbit IgG (H+L) (Invitrogen Thermo Fisher Scientific, A21206) was used at 1:500; anti-ACTB (1:5000) was obtained from Sigma Aldrich (A5441); anti-SQSTM1/p62 (1:2000) were purchased from Abcam (ab56416); anti-ATG3 (1:500) was purchased from Cell Signaling Technology (3415S); SSEA-1 were bought from Santa Cruz Biotechnology (SC-21702AF488); 3-MA, chloroquine, Hoechst 33342, Baf-A1, and anti-MAP1LC3B (1:2000) were obtained from Sigma Aldrich (M9281, C6628, B2261, B1793, L7543). pMXs-Pou5f1, pMXs-Sox2, pMXs-Klf4, and pMXs-cMyc were purchased from Addgene (13366, 13367, 13370, 13375; Deposited by Shinya Yamanaka lab). Atg3 and its lipidation-deficient V8D mutant were cloned into pMXs and pCDH-CAG-RFP lenti-vectors as described previously.15
Atg3+/+ and atg3−/− ESCs were isolated at embryonic d 3.5 and cultured on feeder layers for 5 d using 2i medium (ESC medium with 1 μM PD0325901 [Stemgent, 040006], 3 μM CHIR99021 [Stemgent, 040004] and 1000 U/ml leukemia inhibitory factor [Merck Millipore, ESG1107]). Then, ESC colonies were selected and cultured for 3 to 5 passages. After, isolated ESCs were routinely maintained in ESC medium (knockout Dulbecco's modified Eagle's medium [Gibco Life Technologies,10829018] with 15% fetal bovine serum [GE Healthcare Life Sciences, sh30070.03], 2 mM glutamine [Gibco Life Technologies, 25030164], 1 mM sodium pyruvate [Gibco Life Technologies, 11360070], 0.1 mM nonessential amino acids [Gibco Life Technologies, 11140050], 100 μg/ml streptomycin and 100 U/ml penicillin [Gibco Life Technologies, 15140122], 0.055 mM β-mercaptoethanol [Gibco Life Technologies, 21985023], and 1000 U/ml leukemia inhibitory factor [Merck Millipore, ESG1107]). For iPSC generation, MEFs were seeded at 50,000 cells/well in a 6-well plate and infected with a retrovirus cocktail expressing Pou5f1/Oct4, Sox2, Klf4, and Myc; iPSC colonies were picked 14 d after infection as described previously.37
For mtDNA copy number determination, total DNA was extracted by a TIANamp Genomic DNA Kit (Tiangen Biotech [Beijing] Co., DP304-03). The mtDNA copy number was detected by quantitative real-time PCR using genomic DNA as a loading control. The primers used were published previously.38 Primers for genomic DNA (H19) were 5′-GTCCACGAGACCAATGACTG-3′ (reverse) and 5′-GTACCCACCTGTCGTCC-3′ (forward); mtDNA (CytB) primers were 5′-ATTCCTTCATGTCGGACGAG-3′ (reverse) and 5′-ACTGAGAAGCCCCCTCAAAT-3′ (forward). For total mitochondrial mass detection, cells were digested into single cell suspensions by trypsin (Gibco Life Technologies, 25200072), stained with 100 nM MitoTracker Green/Red for 30 min at 37°C, then washed and analyzed by a FACS calibur flow cytometer (BD Biosciences, Shanghai).
Measurement of intact cellular respiration was performed using a Seahorse XF24 analyzer (Seahorse Bioscience Asia, Shanghai). Briefly, ESC or MEF cells were seeded at 60,000 cells/well 6 h before measurements. Then respiration was measured and determined using a standard protocol in the manual. The cellular ATP content was determined using a CellTiter-GloLuminescent Cell Viability Assay kit (Promega Corporation, 0000092970). The cellular ROS was measured by flow cytometry using HDCF-DA from Sigma Aldrich (D6883).
The immunostaining was performed as previously described.39 Briefly, MEFs or ESCs cultured on gelatin-coated glass slides were fixed with 4% paraformaldehyde for 20 min, washed with Dulbecco's PBS (Corning Inc., R21-031-CV), permeabilized by 0.2% Triton-X100 (Sigma Aldrich, X100) for 0.5 h, blocked with 2% bovine serum albumin (Sigma Aldrich, A1933) for 1 h, stained with appropriate primary antibodies overnight at 4°C, and then incubated with secondary antibodies for 2 h at room temperature. Cell nuclei were counterstained with Hoechst 33342.
Total RNA was extracted from MEFs, ESCs, iPSCs, and EBs with an RNeasy Total RNA Isolation Kit (Qiagen, 74104). Total RNA (1 µg) was reverse transcribed into cDNA using a SuperScript™ III First-Strand Synthesis System (Invitrogen Thermo Fisher Scientific, 18080051). The primers used were as reported previously.40 Pou5f1: 5′-AGAGGATCACCTTGGGGTACA-3′ (forward), 5′- CGAAGCGACAGATGGTGGTC-3′ (reverse); Nanog: 5′-TCTTCCTGGTCCCCACAGTTT-3′ (forward), 5′-GCAAGAATAGTTCTCGGGATGAA-3′ (reverse); Sox2: 5′-GCGGAGTGGAAA CTTTTGTCC-3′ (forward), 5′-CGGGAAGCGTGTACTTATCCTT-3′ (reverse); Klf4: 5′-GTGCCCC GACTAACCGTTG-3′ (forward), 5′-GTCGTTGAACTCCTCGGTCT-3′ (reverse); Esrrb: 5′-CAG GCAAGGATGACAGACG-3′ (forward), 5′-GAGACAGCACGAAGGACTGC-3′ (reverse); Tbx3: 5′-TTGCAAAGGGTTTTCGAGAC-3′ (forward), 5′-TGGAGGACTCATCCGAAGTC-3′ (reverse); Tcl1: 5′-AAATTCCAGGTGATCTTGCG-3′ (forward), 5′-TGTCCTTGGGGTACAGTTGC-3′ (reverse); Fbxo15: 5′-TCGTGGGACTGAGCACAACTA-3′ (forward), 5′-TGACAGATGAGCCT CTAACAAAC-3′ (reverse); Rexo1: 5′-CCCTCGACAGACTGACCCTAA-3′ (forward), 5′-TCGGG GCTAATCTCACTTTCAT-3′ (reverse); Cdh1: 5′-CAGGTCTCCTCATGGCTTTGC-3′ (forward), 5′-CTTCCGAAAAGAAGGCTGTCC-3′ (reverse); Nes: 5′-CCCTGAAGTCGAGGAGCTG-3′ (forward), 5′-CTGCTGCACCTCTAAGCGA-3′ (reverse); Fgf5: 5′-CTGTATGGACCCAC AGGGAGTAAC-3′ (forward), 5′-ATTAAGCTCCTGGGTCGCAAG-3′ (reverse); Otx2: 5′-TATC TAAAGCAACCGCCTTACG-3′ (forward), 5′- AAGTCCATACCCGAAGTGGTC-3′ (reverse); Tal1: 5′-CTGGCCTCCAGCTACATTTCT-3′ (forward), 5′-GTCACGGTCTTTGCTCAACTT-3′ (reverse); Gata2: 5′-CACCCCGCCGTATTGAATG-3′ (forward), 5′-CCTGCGAGTCGAGATGGTTG-3′ (reverse); T: 5′-GCTTCAAGGAGCTAACTAACGAG-3′ (forward), 5′-CCAGCAAGAAAGAG TACATGGC-3′ (reverse); Afp: 5′-CTTCCCTCATCCTCCTGCTAC-3′ (forward), 5′-ACAAA CTGGGTAAAGGTGATGG-3′ (reverse); Gata4: 5′-CCCTACCCAGCCTACATGG-3′ (forward), 5′-ACATATCGAGATTGGGGTGTCT-3′ (reverse); Sox17: 5′-GATGCGGGATACGCCAGTG-3′ (forward), 5′-CCACCACCTCGCCTTTCAC-3′ (reverse); Actb: 5′-GGCTGTATTCCCCTC CATCG-3′ (forward), 5′-CCAGTTGGTAACAATGCCATGT-3′ (reverse).
No potential conflicts of interest were disclosed.
We thank G.W., at our institute, T.Z., J.L. and L.C.at IM-CAS for technique supports.
This work was supported by grants from the Strategic Priority Research Program of the Chinese Academy of Sciences XDA01040108, the China National Basic Research Program 2012CB966901, the National Natural Science Foundation of China Program 31271592, 31570995, and National Thousand Young Talents Program to T.Z.