Pluripotent Stem Cells Induced from Mouse Somatic Cells by Small-Molecule Compounds

See allHide authors and affiliations

Science  09 Aug 2013:
Vol. 341, Issue 6146, pp. 651-654
DOI: 10.1126/science.1239278

Promoting Pluripotency

A specialized mammalian cell can be set back to the pluripotent state either by transfer of the somatic cell nucleus into an oocyte or by delivery of exogenous pluripotency-associated transcription factors. Hou et al. (p. 651, published online 18 July) developed an approach to induce pluripotency in somatic cells using a cocktail of small molecules. The ability to generate such chemically induced pluripotent stem cells may provide an alternate route for therapeutic cloning and for drug development in regenerative medicine.


Pluripotent stem cells can be induced from somatic cells, providing an unlimited cell resource, with potential for studying disease and use in regenerative medicine. However, genetic manipulation and technically challenging strategies such as nuclear transfer used in reprogramming limit their clinical applications. Here, we show that pluripotent stem cells can be generated from mouse somatic cells at a frequency up to 0.2% using a combination of seven small-molecule compounds. The chemically induced pluripotent stem cells resemble embryonic stem cells in terms of their gene expression profiles, epigenetic status, and potential for differentiation and germline transmission. By using small molecules, exogenous “master genes” are dispensable for cell fate reprogramming. This chemical reprogramming strategy has potential use in generating functional desirable cell types for clinical applications.

Pluripotent stem cells, such as embryonic stem cells (ESCs), can self-renew and differentiate into all somatic cell types. Somatic cells can be reprogrammed to become pluripotent via nuclear transfer into oocytes or through the ectopic expression of defined factors (14). However, exogenous pluripotency-associated factors, especially Oct4, are indispensable for establishing pluripotency (57), and previous reprogramming strategies have raised concerns regarding the clinical applications (8, 9). Small molecules have advantages because they can be cell permeable, nonimmunogenic, more cost-effective, and more easily synthesized, preserved, and standardized. Moreover, their effects on inhibiting and activating the function of specific proteins are often reversible and can be finely tuned by varying the concentrations. Here, we identified small-molecule combinations that were able to drive the reprogramming of mouse somatic cells toward pluripotent cells.

To identify small molecules that facilitate cell reprogramming, we searched for small molecules that enable reprogramming in the absence of Oct4 using Oct4 promoter-driven green fluorescent protein (GFP) expression (OG) mouse embryonic fibroblasts (MEFs), with viral expression of Sox2, Klf4, and c-Myc. After screening up to 10,000 small molecules (table S1A), we identified Forskolin (FSK), 2-methyl-5-hydroxytryptamine (2-Me-5HT), and D4476 (table S1B) as chemical “substitutes” for Oct4 (Fig. 1, A and B, and figs. S1 and S2). Previously, we had developed a small-molecule combination “VC6T” [VPA, CHIR99021 (CHIR), 616452, Tranylcypromine], that enables reprogramming with a single gene, Oct4 (6). We next treated OG-MEFs with VC6T plus the chemical substitutes of Oct4 in the absence of transgenes. We found that VC6T plus FSK (VC6TF) induced some GFP-positive clusters expressing E-cadherin, a mesenchyme-to-epithelium transition marker, reminiscent of early reprogramming by transcription factors (10, 11) (Fig. 1C and fig. S3). However, the expression of Oct4 and Nanog was not detectable, and their promoters remained hypermethylated, suggesting a repressed epigenetic state (fig. S3).

Fig. 1 Generation of CiPSCs by small-molecule compounds.

(A and B) Numbers of iPSC colonies induced from MEFs infected by SKM (A) or SK (B) plus chemicals or Oct4. Error bars, mean ± SD (n = 3 biological repeat wells). (C) Morphology of MEFs for chemical reprogramming on day 0 (D0) and a GFP-positive cluster generated using VC6TF on day 20 (D20) after chemical treatment. (D) Numbers of GFP-positive colonies induced after DZNep treatment on day 36. Error bars, mean ± SD (n = 2 biological repeat wells). (E to G) Morphology of a compact, epithelioid, GFP-positive colony on day 32 (D32) after treatment (E), a primary CiPSC colony on day 40 (D40) after treatment (F), and passaged CiPSC colonies (G). (H) Schematic diagram illustrating the process of CiPSC generation. Scale bars, 100 μm. For (D), cells for reprogramming were replated on day 12.

To identify small molecules that facilitate late reprogramming, we used a doxycycline (DOX)–inducible Oct4 expression screening system, adding DOX only in the first 4 to 8 days (6). Small-molecule hits, including several cAMP agonists (FSK, Prostaglandin E2, and Rolipram) and epigenetic modulators [3-deazaneplanocin A (DZNep), 5-Azacytidine, sodium butyrate, and RG108], were identified in this screen (fig. S4 and table S1B).

To achieve complete chemical reprogramming without the Oct4-inducible system, these small molecules were further tested in the chemical reprogramming of OG-MEFs without transgenes. When DZNep was added 16 days after treatment with VC6TF (VC6TFZ), GFP-positive cells were obtained more frequently by a factor of up to 65 than those treated with VC6TF, forming compact, epithelioid, GFP-positive colonies without clear-cut edges (Fig. 1, D and E, and fig. S5). In these cells, the expression levels of most pluripotency marker genes were elevated but were still lower than in ESCs, suggesting an incomplete reprogramming state (fig. S6). After switching to 2i-medium with dual inhibition (2i) of glycogen synthase kinase-3 and mitogen-activated protein kinase signaling after day 28 posttreatment, certain GFP-positive colonies developed an ESC-like morphology (domed, phase-bright, homogeneous with clear-cut edges) (Fig. 1F) (12, 13). These colonies could be further cultured for more than 30 passages, maintaining an ESC-like morphology (Fig. 1, G and H). We refer to these 2i-competent, ESC-like, and GFP-positive cells as chemically induced pluripotent stem cells (CiPSCs).

Next, we optimized the dosages and treatment duration of the small molecules and were able to generate 1 to 20 CiPSC colonies from 50,000 initially plated MEFs (fig. S7). After an additional screen, we identified some small-molecule boosters of chemical reprogramming, among which, a synthetic retinoic acid receptor ligand, TTNPB, enhanced chemical reprogramming efficiency up to a factor of 40, to a frequency comparable to transcription factor–induced reprogramming (up to 0.2%) (fig. S8 and table S1B). Furthermore, using the small-molecule combination VC6TFZ, we obtained CiPSC lines from mouse neonatal fibroblasts (MNFs), mouse adult fibroblasts (MAFs), and adipose-derived stem cells (ADSCs) with OG cassettes by an efficiency lower by a factor of ~10 than that obtained from MEFs (fig. S9 and table S3). Moreover, we induced CiPSCs from wild-type MEFs without OG cassettes or any other genetic modifications by a comparable efficiency to that achieved from MEFs with OG cassettes (fig. S9). The CiPSCs were also confirmed to be viral-vector free by genomic polymerase chain reaction (PCR) and Southern blot analysis (fig. S10).

The established CiPSC lines were then further characterized. They grew with a doubling time (14.1 to 15.1 hours) similar to that of ESCs (14.7 hours), maintained alkaline phosphatase activity, and expressed pluripotency markers, as detected by immunofluorescence and reverse transcription (RT)-PCR (Fig. 2, A and B, and fig. S11). The gene expression profiles were similar in CiPSCs, ESCs, and OSKM-iPSCs (iPSCs induced by Oct4, Sox2, Klf4, and c-Myc) (Fig. 2C and fig. S12). DNA methylation state and histone modifications at Oct4 and Nanog promoters in CiPSCs were similar to that in ESCs (Fig. 2D and fig. S13). In addition, CiPSCs maintained a normal karyotype and genetic integrity for up to 13 passages (fig. S14 and table S2).

Fig. 2 Characterization of CiPSCs.

(A and B) Pluripotency marker expression as illustrated by immunofluorescence (A, clone CiPS-25) and RT-PCR (B). Scale bars, 100 μm. (C) Hierarchical clustering of global transcriptional profiles. 1-PCC, Pearson correlation coefficient. (D) Bisulfite genomic sequencing of the Oct4 and Nanog promoter regions. MNF-CiPS-7, MNF-derived CiPSC line no. 7; MAF-CiPS-1, MAF-derived CiPSC line no. 1.

To characterize their differentiation potential, we injected CiPSCs into immunodeficient (SCID) mice. The cells were able to differentiate into tissues of all three germ layers (Fig. 3A and fig. S15). When injected into eight-cell embryos or blastocysts, CiPSCs were capable of integration into organs of all three germ layers, including gonads and transmission to subsequent generations (Fig. 3, B to E, and fig. S16). Unlike chimeric mice generated from iPSCs induced by transcription factors including c-Myc (14), the chimeric mice generated from CiPSCs were 100% viable and apparently healthy for up to 6 months (Fig. 3F). These observations suggest that the CiPSCs were fully reprogrammed into pluripotency (table S3).

Fig. 3 Pluripotency of CiPSCs.

(A) Hematoxylin and eosin staining of CiPSC-derived teratoma (clone CiPS-30). (B to D) Chimeric mice (B, clone CiPS-34), germline contribution of CiPSCs in testis, (C, clone CiPS-45) and F2 offspring (D, clone CiPS-34). Scale bars, 100 μm. (E) Genomic PCR analyzing pOct4-GFP cassettes in the tissues of chimeras. (F) Survival curves of chimeras. n, total numbers of chimeras studied.

We next determined which of these small molecules were critical in inducing CiPSCs. We found four essential small molecules whose individual withdrawal from the cocktails generated significantly reduced GFP-positive colonies and no CiPSCs (Fig. 4, A to C). These small molecules (C6FZ) include: CHIR (C), a glycogen synthase kinase 3 inhibitor (15); 616452 (6), a transforming growth factor-beta inhibitor (16); FSK (F), a cAMP agonist (fig. S17) (17); and DZNep (Z), an S-adenosylhomocysteine (SAH) hydrolase inhibitor (figs. S18 and S19) (18, 19). Moreover, C6FZ was able to induce CiPSCs from both MEFs and MAFs, albeit by an efficiency lower by a factor of 10 than that induced by VC6TFZ (fig. S20 and table S3).

Fig. 4 Stepwise establishment of the pluripotency circuitry during chemical reprogramming.

(A and B) Numbers of GFP-positive (A) and CiPSC (B) colonies induced by removing individual chemicals from VC6TFZ. The results of three independent experiments are shown with different colors (white, gray, and black). (C) Structures of the essential chemicals. (D and E) The expression of pluripotency-related genes (D) and Gata6, Gata4, and Sox17 (E) as measured by real-time PCR. (F) Gene expression heat map at the single colony level. The value indicates the log2-transformed fold change (relative to Gapdh and normalized to the highest value). (G and H) Oct4 activation (G) and numbers of GFP-positive and iPSC colonies (H) induced by the overexpression of Sall4 and Sox2, with C6F removed from VC6TFZ. (I to K) The expression of pluripotency-related genes (I), DNA methylation (J), and H3K9 dimethylation (K) states of the Oct4 promoter in the presence and absence of DZNep on day 32. (L) Schematic diagram illustrating the stepwise establishment of the pluripotency circuitry during chemical reprogramming. Error bars, mean ± SD (n ≥ 2 biological repeats).

To better understand the pluripotency-inducing properties of these small molecules, we profiled the global gene expression during chemical reprogramming and observed the sequential activation of certain key pluripotency genes, which was validated by real-time PCR and immunofluorescence (fig. S21). The expression levels of two pluripotency-related genes, Sall4 and Sox2, were most significantly induced in the early phase in response to VC6TF, as was the expression of several extra-embryonic endoderm (XEN) markers Gata4, Gata6, and Sox17 (Fig. 4, D to F, and figs. S22 to S24). The expression of Sall4 was enhanced most significantly as early as 12 hours after small-molecule treatment, suggesting that Sall4 may be involved in the first step toward pluripotency in chemical reprogramming (fig. S22B). We further examined the roles of the endogenous expression of these genes in chemical reprogramming, using gene overexpression and knockdown strategies. We found that the concomitant overexpression of Sall4 and Sox2 was able to activate an Oct4 promoter–driven luciferase reporter (fig. S25) and was sufficient to replace C6F in inducing Oct4 expression and generating iPSCs (Fig. 4, G and H, and fig. S26). The endogenous expression of Sall4, but not Sox2, requires the activation of the XEN genes, and vice versa (fig. S27). This suggests a positive feedback network formed by Sall4, Gata4, Gata6, and Sox17, similar to that previously described in mouse XEN formation (20). Moreover, knockdown of Sall4 or these XEN genes impaired Oct4 activation and the subsequent establishment of pluripotency (fig. S28), inconsistent with our previous finding that Gata4 and Gata6 can contribute to inducing pluripotency (21). Taken together, these findings revealed a Sall4-mediated molecular pathway that acts in the early phase of chemical reprogramming (Fig. 4L). This step resembles a Sall4-mediated dedifferentiation process in vivo during amphibian limb regeneration (22).

We next investigated the role of DZNep, which was added in the late phase of chemical reprogramming. We found that Oct4 expression was enhanced significantly after the addition of DZNep in chemical reprogramming (Fig. 4D), and DZNep was critical for stimulating the expression of Oct4 but not the other pluripotency genes (Fig. 4I). As an SAH hydrolase inhibitor, DZNep elevates the concentration ratio of SAH to S-adenosylmethionine (SAM) and may thereby repress the SAM-dependent cellular methylation process (fig. S18) (18, 19). Consistently, DZNep significantly decreased DNA and H3K9 methylation at the Oct4 promoter, which may account for its role in Oct4 activation (Fig. 4, J and K) (23, 24). As master pluripotency genes, Oct4 and Sox2 may thereby activate other pluripotency-related genes and fulfill the chemical reprogramming process, along with the activation of Nanog and the silencing of Gata6, in the presence of 2i (12, 13, 25, 26) (Fig. 4F and fig. S29). In summary, as a master switch governing pluripotency, Oct4 expression, which is kept repressed in somatic cells by multiple epigenetic modifications, is unlocked in chemical reprogramming by the epigenetic modulator DZNep and stimulated by C6F-induced expression of Sox2 and Sall4 (Fig. 4L).

Our proof-of-principle study demonstrates that somatic reprogramming toward pluripotency can be manipulated using only small-molecule compounds (fig. S30). It reveals that the endogenous pluripotency program can be established by the modulation of molecular pathways nonspecific to pluripotency via small molecules rather than by exogenously provided “master genes.” These findings increase our understanding about the establishment of cell identities and open up the possibility of generating functionally desirable cell types in regenerative medicine by cell fate reprogramming using specific chemicals or drugs, instead of genetic manipulation and difficult-to-manufacture biologics. To date, the complete chemical reprogramming approach remains to be further improved to reprogram human somatic cells and ultimately meet the needs of regenerative medicine.

Supplementary Materials

Materials and Methods

Figs. S1 to S30

Tables S1 to S5

References (2737)

References and Notes

  1. Acknowledgments: We thank X. Zhang, J. Wang, C. Han, Z. Hou, J. Liu, and L. Ai for technical assistance. This work was supported by grants from the National 973 Basic Research Program of China (2012CB966401 and 2010CB945204), the Key New Drug Creation and Manufacturing Program (2011ZX09102-010-03), the National Natural Science Foundation of China (90919031), the Ministry of Science and Technology (2011AA020107, 2011DFA30730 and 2013DFG30680), the Beijing Science and Technology Plan (Z121100005212001), the Ministry of Education of China (111 project), and a Postdoctoral Fellowship of Peking-Tsinghua Center for Life Sciences. Microarray and RNA-seq data are deposited in the Gene Expression Omnibus (GEO) database (accession no. GSE48243). The authors have filed a patent for the small-molecule combinations used in the chemical reprogramming reported in this paper.
View Abstract

Navigate This Article