Promotion of Feeder-Independent Self-Renewal of Embryonic Stem Cells by Retinol (Vitamin A)


  • Liguo Chen,

    1. Department of Microbiology and Molecular Genetics, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
    Search for more papers by this author
  • Jaspal S. Khillan Ph.D.

    Corresponding author
    1. Department of Microbiology and Molecular Genetics, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
    • Department of Microbiology and Molecular Genetics, 200 Lothrop Street, University of Pittsburgh, Pittsburgh, Pennsylvania 15213, USA. Telephone: 412-383-6987; Fax: 412-648-8695
    Search for more papers by this author


Retinol, the alcohol form of vitamin A, maintains pluripotency of mouse embryonic stem cells (ESCs) by the overexpression of Nanog, which is a key transcription factor for their self-renewal. ESCs represent the most promising source of all types of cells for regenerative medicine and drug discovery. These cells maintain pluripotency through a complex interplay of different signaling pathways and transcription factors including leukemia inhibitory factor (LIF), homeodomain protein Nanog, and Oct3/4. Nanog, however, plays a key role in maintaining the pluripotency of mouse and human ESCs. Overexpression of nanog by heterologous promoters can maintain pluripotency of ESCs in the absence of LIF. Also, Nanog alone is sufficient for the self-renewal of ESCs while maintaining the Oct4 levels. Normally, mouse and human ESCs are cultured over mouse embryonic fibroblasts as feeders to maintain pluripotency. Although feeder cells provide important growth-promoting factors, their use involves several cumbersome and time-consuming steps. Here we demonstrate that retinol can support feeder-independent self-renewal of ESCs in long-term cultures without affecting their pluripotency. The effect of retinol is independent of the strain background, and the cells maintain complete potential to differentiate into all the primary germ layers in embryoid bodies and in chimeric animals. Self-renewal of ESCs by retinol is not mediated by retinoic acid. The studies demonstrate for the first time that a physiologically relevant small molecule has growth-promoting effect on the self-renewal of ESCs by activating the endogenous machinery to overexpress a critical gene for pluripotency.

Disclosure of potential conflicts of interest is found at the end of this article.


Author contributions: J.S.K.: conception and design, financial support, manuscript writing; L.C.: molecular analysis, cell culture.

Pluripotent embryonic stem cells (ESCs) derived from the inner cell mass (ICM) of mammalian blastocysts have indefinite potential for growth while maintaining their pluripotency. After injection into blastocyst, ESCs contribute to all the primary germ layers such as endoderm, mesoderm, and ectoderm [1, 2]. It is therefore, hoped that ESCs will have future applications in treating degenerative diseases such as Alzheimer's disease, Parkinson's disease, diabetes, spinal cord injury, and cardiovascular diseases in humans [3].

ESCs express a specific transcriptional program that suppresses differentiation to maintain their self-renewal. Leukemia inhibitory factor (LIF), a member of interleukin-6 family of lymphokines, supports self-renewal of mouse ESCs by activating the Janus kinase/signal transducer and activator of transcription 3 (Stat3) pathway [4, [5], [6], [7], [8]–9], via a heterodimer receptor (LIF receptor) and via glycoprotein 130, which activates downstream genes [10]. Although LIF signaling is critical for mouse ESCs [11], it is not sufficient and, therefore, requires additional signals such as those provided by the bone morphogenic proteins (BMPs), which activate inhibitors of differentiation (Id) genes [12]. Human ESCs do not require LIF [13], suggesting that this pathway is not fundamental to the pluripotency of ESCs. Wnt/β-catenin signaling has also been implicated in maintaining the pluripotency of both human and mouse ESCs [14].

POU domain-containing transcriptional factor oct3/4, homeobox genes nanog and sox2, and fibroblast growth factor 4 (fgf4) also play a key role in ESC self-renewal [15, [16], [17], [18], [19]–20]. Oct3/4 and nanog are specifically expressed in ICM and epiblast of the early embryos [17, [18], [19]–20]. Null mutations of oct3/4 [17] and nanog [19, 20] result in early embryonic lethality. Nanog alone, however, is sufficient for pluripotency of mouse ESCs independent of LIF/Stat3 pathway [20]. Downregulation of nanog causes differentiation of ESCs into extraembryonic endoderm [19], whereas its overexpression allows self-renewal of ESCs in the absence of LIF [21].

Mouse and human ESCs are generally cocultured with mitotically inactive mouse embryonic fibroblasts (MEFs) to prevent their spontaneous differentiation [3, 22]. However, the feeder cells involve several cumbersome and time-consuming steps, such as generation of MEFs from mouse embryos, inactivation of cells by mitomycin or by γ-irradiation, and finally removal of feeder cells from ESCs before using for subsequent studies. In addition, coculturing ESCs with different batches of feeder cells may cause variability in their growth. Coculture of human ESCs with MEFs can potentially result in xenobiotic contamination, rendering them unsuitable for clinical applications. Therefore, the conditions that support feeder-independent culture of ESCs will be highly desirable.

Earlier we reported that retinol, the alcohol form of vitamin A, prevents differentiation of mouse ESCs by the overexpression of Nanog [23]. Retinol is normally associated with the differentiation during embryonic and fetal development via its highly potent metabolite retinoic acid [24]. Contrary to the published reports, our studies demonstrate that in ESCs, retinol prevents differentiation via yet unknown mechanism that does not involve retinoic acid and overcomes the requirement for feeder cells in long-term cultures. The cells express oct4, sox2, and nanog and retain complete potential for differentiation into cells of all the primary germ layers such as endoderm, mesoderm, and ectoderm in vitro in embryoid bodies as well as in chimeric animals after microinjection into blastocysts.

Materials and Methods

Embryonic Stem Cell Culture

Approximately 2.0 × 104 cells were cultivated over 6-well tissue culture plates coated with 0.1% gelatin using ESC medium containing Dulbecco's modified Eagle's medium (DMEM) with 15% fetal bovine serum, 1 mM l-glutamine, 1% nonessential amino acids, 0.1 mM β-mercaptoethanol, and 1,000 U/ml LIF [22]. Normal and green florescent protein (GFP)-positive R1, KBL2, FVB/N13, and embryonic day 14 (E14) ESCs derived from 129Sv, C57BL6, FVB/N, and 129Ola strains of mouse, respectively, were used for the studies. ES medium was supplemented with 0.5 μM retinol within 10–12 hours after the cells had settled down. Medium was changed every day using fresh retinol. The confluent cultures were trypsinized with 0.25% trypsin-EDTA usually after 5–7 days. All trans-retinol and all trans-retinoic acid purchased from Sigma-Aldrich (St. Louis, were dissolved in 100% ethanol as 100 mM stock solution. The experiments were performed at least three times.

Embryoid Body Formation

Embryoid body formation was initiated by plating dissociated ESCs in suspension culture in the absence of LIF using DMEM supplemented with 15% fetal calf serum (FCS), 1 mM l-glutamine, 1% nonessential amino acids, 0.1 mM β-mercaptoethanol, and penicillin-streptomycin [25]. Three days later as the embryoid body-like spheres began to form, the cells were transferred onto bacterial dishes for an additional 3 days in the same medium. The medium was renewed every other day.

Reverse-Transcription Polymerase Chain Reaction Analysis

The cells were collected after trypsinization with 0.25% trypsin EDTA, and total RNA was isolated using STAT 60 solution (TEL-TEST, Friendswood, TX, following instructions from the manufacturer. The RNA was converted into cDNA using oligo-dT and avian myeloblastosis virus reverse transcriptase using kit purchased from Invitrogen (Carlsbad, CA, Polymerase chain reaction (PCR) was carried out in a total reaction volume of 50 μl using specific primers. The PCR conditions used denaturation at 94°C for 45 seconds; extension at 72°C for 2 minutes; and annealing at temperature as specified for each primer pair for 23–30 cycles. Products were resolved by agarose gel electrophoresis and visualized by ethidium bromide staining using hypoxanthine-guanine phosphoribosyltransferase (HPRT) primers as control. Each reverse-transcription (RT)-PCR analysis was performed three times.

RT-PCR Primer Sequences


Western Blot Analysis

Anti-β-actin and anti-Oct4 antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA,, and Nanog antibody was purchased from Chemicon (Temecula, CA, Total protein was extracted with radioimmunoprecipitation assay buffer (Sigma-Aldrich catalog no. R0278). Fifty micrograms of protein was separated by 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred onto nylon membrane (Bio-Rad, Hercules, CA, The membranes were incubated with antibodies to specific protein followed by incubation with horseradish peroxidase-conjugated goat antibody to mouse IgG or rabbit antibody to goat IgG (Santa Cruz Biotechnology), and developed with chemiluminescence reagent (Pierce, Rockford, IL, The protein level was measured by Bio-Rad Fluor-S-MultiImager using Quantity one-4.2.3 software.

Immunostaining of ESCs

The cells were fixed with 4% paraformaldehyde and permeabilized with cold methanol. After washing and fixing, nonspecific binding was blocked with 5% FCS in phosphate-buffered saline for 1 hour at room temperature. ESCs were labeled with goat anti-mouse Nanog antibody, Oct-4 antibody (Santa Cruz Biotechnology), and Nestin antibody (BD Biosciences, San Diego,, and Texas Red-conjugated donkey anti-goat antibody was used as secondary antibody. Cells were examined by fluorescence microscopy using low-power Leica (Allendale, NJ, microscope.

Alkaline Phosphatase Assay

The cells were fixed with 4% paraformaldehyde for 2 minutes at room temperature. Staining for alkaline phosphatase (AP) was performed using a kit from Chemicon following protocols provided by the manufacturer.

Production of Chimeric Animals

ESCs at passage 15 of retinol treatment were either microinjected into blastocysts or cocultured with 2.5-day morulas [26] isolated from C57BL6 mice. The microinjected blastocysts and overnight-cocultured embryos were transferred to the uterine horns of 2.5-day pseudopregnant CD1 foster mothers, and the embryos were collected at specified stage. Embryos were also allowed to develop to obtain chimeric animals. All animal procedures were carried out according to the University of Pittsburgh-approved Institutional Animal Care and Use Committee protocols.


Retinol Supports Self-Renewal of ESCs in the Absence of MEF Feeder Cells

Earlier studies from our laboratory showed that retinol prevents the differentiation of ESCs cocultured with MEF feeder cells [23]. Therefore, to investigate its effect in the absence of feeder cells in long-term cultures, approximately 20,000 cells from R1 cell line (from A. Nagy) were cultured over 6-well gelatin-coated plates. The cells were allowed to attach overnight, after which the ES medium supplemented with 0.5 μM retinol was added. The medium was changed every day using fresh retinol. After 4 days, approximately 50% of the cells in control cultures showed differentiated phenotype as observed by the presence of flat colonies. However, the staining for AP after 7 days revealed that almost all the cells were differentiated (Fig. 1A, left panel). Retinol-treated parallel cultures, on the other hand, showed almost 100% colonies that were phase bright with sharp boundaries and stained strongly positive for AP (Fig. 1A, right panel), indicating that retinol can also prevent differentiation of ESCs in the absence of feeder cells.

Figure Figure 1..

Alkaline phosphatase (AP) staining of feeder-independent R1 embryonic stem cells (ESCs). (A): AP staining of control (left panel) and 6-day retinol-treated (right panel) cells. All the cells in the control panel are differentiated. Magnification, ×20. (B): AP staining of retinol-treated R1 ESCs at passages 1, 7, 10, and 15 shows positive staining for alkaline phosphatase. Magnification, ×20. Abbreviation: ps, passage.

Confluent cultures of retinol-treated cells were trypsinized, and 20,000 cells were plated on fresh plates coated with gelatin using medium with 0.5 μM retinol. The cells at this stage were designated as passage 1 cells. The medium was changed every day using fresh retinol. More than 90% of the cells developed into well-formed colonies within 4 days. After 7 days, the confluent cultures were again trypsinized and plated onto fresh gelatin-coated plates at passage 2. This process was repeated for 15 passages using 20,000 cells each time. The cells were trypsinized approximately every 5–7 days as soon as they appeared confluent. Figure 1B shows AP-stained cells at passages 1, 7, 10, and 15. As shown in the figure, ESCs stained positive for AP at all the passages. More than 90% of the colonies exhibited phase bright and well-defined boundaries typical of undifferentiated ESCs. No obvious difference was observed in the morphology or number of colonies at different passages.

Retinol Maintains Pluripotency-Specific Genes in ESCs

To study the gene expression, total RNA isolated from the retinol-treated R1 ESCs at passages 9, 10, 11, and 12 was analyzed by semiquantitative RT-PCR. ESCs at all the passages expressed undifferentiated cell-specific genes such as nanog, oct4, fgf4, rex1, stat3, and sox2 (Fig. 2A) similar to the normal ESCs (lane N), proving that retinol can maintain pluripotency of ESCs in long-term cultures in the absence of feeder cells. The cells were further cultured to passage 15 and as observed in earlier passages, the cells at passage 15 also expressed the undifferentiated cell-specific genes (Fig. 2B). AP staining (Fig. 1B) and RT-PCR analysis, therefore, indicate that ESCs retain their undifferentiated properties in long-term cultures for more than 3 months. Further analysis revealed that the cells do not express differentiation-specific genes such as nestin and α-fetoprotein at any of the passages (data not shown), supporting our earlier results that retinol suppresses the differentiation of ESCs [23]. The cells, however, expressed differentiation-specific genes such as Brachyury, GATA4, and nestin similar to normal ESCs within 96 hours after the withdrawal of retinol as observed in Figure 2C explained in the following section.

Figure Figure 2..

Semiquantitative reverse-transcription–polymerase chain reaction analysis. (A): RNA isolated from retinol-treated R1 embryonic stem cells (ESCs) at passages 9, 10, 11, and 12 and control ESCs (N). (B): RNA isolated from retinol-treated cells at passage 15. (C): RNA from ESCs and embryoid bodies (EBs) from untreated (−RE) and retinol-treated (+RE) cultures.

Retinol Maintains Differentiation Potential of ESCs in Embryoid Bodies

To study the potential of retinol-treated cells for different lineages, normal and retinol-treated cells were cultured under conditions that promote differentiation to form embryoid bodies (i.e., without LIF in the medium). After 6 days, total RNA was analyzed by RT-PCR analysis. The embryoid bodies from both retinol-treated and normal ESCs expressed follistatin, GATA4, and Brachyury, the genes specific for ectoderm, endoderm, and mesoderm lineages, respectively (Fig. 2C, last two lanes), indicating that retinol-treated cells retain their potential to differentiate into different cell types. As a control, parallel cultures treated with retinol were also tested. As expected, no expression of differentiation-specific genes was observed in the 6-day feeder-independent retinol-treated cells, whereas untreated ESCs expressed follistatin, GATA4, and Brachyury (Fig. 2C, first two lanes).

Retinol-Treated ESCs Contribute to Different Organs in Chimeric Animals

To test in vivo potential, ESCs that contain a gene for GFP were cultured in the presence of retinol for 15 passages in the absence of feeder cells. At passage 15, the cells were cocultured with 8–16 cell morulas in microwell plates [26]. After overnight culture, the embryos were screened for GFP. The blastocysts developed from coculture of morula and ESCs showed integration of GFP-positive cells (Fig. 3, top left panel). Separately, the blastocysts microinjected with GFP-positive cells were transferred to the pseudopregnant females. The mothers were sacrificed at E18.0 to recover fetuses. UV-light scanning of the fetuses showed high integration of ESCs throughout the body (Fig. 3). A high contribution of GFP-positive cells was also observed in the organs such as kidney, heart, lung, and testis. A strong GFP florescence in testis indicates potential of retinol-treated cells for germ line transmission. The recipient mothers were allowed to deliver the pups, which also showed high integration of ESCs, as seen by the agouti patches over the black background of a 15-day-old pup (Fig. 3, bottom panel). These data prove that long-term treatment with retinol does not compromise the potential of ESCs for contributing to different lineages. Feeder-independent ESCs cultivated with LIF medium for 5 days, however, failed to produce chimeras, indicating that these cells lose their pluripotency.

Figure Figure 3..

Integration of green fluorescent protein (GFP)-positive embryonic stem cells (ESCs) into embryos and chimeric animal. Blastocysts from overnight aggregation of morula and ESCs show integration of GFP-positive ESCs. (Magnification, ×100.) A composite of embryonic day 18.0 (E18.0) fetus with high ES cell contribution; kidney, heart, lung, and testis from E18.0 fetus. A 15-day-old chimeric pup with agouti patches representing ESCs on C57BL6 background (bottom panel).

Retinol-Treated ESCs Exhibit High Expression of Nanog

To study the effect of retinol at protein level, feeder-free R1 ESCs were treated with retinol for 6 days. Western blot analysis of the total proteins showed that retinol-treated cells maintain high levels of Nanog (Fig. 4A, lane +Re), whereas Nanog disappeared almost completely in the untreated cells (Fig. 4A, lane C). Although untreated cells showed expression of Oct4, its level was almost fourfold lower than the retinol-treated cells.

Figure Figure 4..

Reverse-transcription–polymerase chain reaction (RT-PCR), Western blot, immunostaining, and alkaline phosphatase (AP) staining of retinol and retinoic acid-treated cells. (A): Six-day feeder-independent retinol-treated (+Re) and normal (C) R1 embryonic stem cells (ESCs). (B): Feeder-dependent 24-hour normal (C), retinol-treated (+Re), and retinoic acid-treated (+RA) R1 ESCs. (C): Immunostaining of 4-day retinol-treated cells (+Re), retinoic acid-treated (+RA) R1, and normal (C) cells for Nanog (top panel), Oct4 (middle panel), and Nestin (bottom panel). (D): RT-PCR analysis of adult liver cells (lane 1), control ESCs (lane 2), retinol-treated ESCs (lane 3), retinoic acid-treated ESCs (lane 4), and primary fibroblasts (lane 5) for retinol-metabolizing enzymes alcohol dehydrogenase 1 (Adh1), Adh4, and retinaldehyde dehydrogenase 2 (RALDH2). (E): AP staining of 4-day retinol-treated (+Re), retinoic acid-treated (+RA), and normal (C) R1 ESCs. (F): RT-PCR analysis of control ESCs (lane 1) and retinol-treated ESCs (lane 2) with leukemia inhibitory factor. (Lanes 3–6) ESCs without LIF for 24, 48, 72, and 96 hours.

To determine the relative expression of Oct4, R1 ESCs cultured with MEF feeder cells were treated with retinol for 24 hours (Fig. 4B). Earlier we reported that retinol treatment results in an up to sixfold increase in Nanog within 24 hours [23]. An elevation of approximately 3.5-fold Nanog level (an average of fivefold from three independent experiments) was observed in retinol-treated cells (Fig. 4B, lane +Re) compared with untreated cells (Fig. 4B, lane C). Retinol-treated cultures contain almost the same level of Oct4 even after 7 days, supporting our earlier observations that Oct4 expression is not affected by retinol in undifferentiated ESCs (compare Fig. 4A and 4B).

The cells were further analyzed by immunostaining using specific antibodies for Nanog and Oct4. Immunostaining of the cells treated with retinol for 4 days showed a significantly enhanced staining for Nanog (Fig. 4C top; middle panel) that was obvious from sharp and localized staining in ES colonies compared with normal cells, which showed only diffused staining pattern (Fig. 4C top; left panel). Most of all, no staining was observed in retinoic acid-treated cells, suggesting differentiation of the cells (Fig. 4C top; right panel). Retinol-treated cells did not exhibit any staining with the antibodies for Nestin (Fig. 4C bottom; middle panel), which is a marker for differentiated cells, further supporting that retinol prevents the differentiation of ESCs. Retinoic acid, however, caused differentiation as observed by staining with Nestin antibodies (Fig. 4C bottom; right panel). The absence of Nestin in control cells may be due to the transient nature of this protein.

No major difference, on the other hand, was observed in the staining for Oct4 between treated and untreated control cells. The expression of Oct4 observed in control and retinoic acid-treated cells may be due to the reason that differentiated cells also contain this protein [27], which is also supported by the Western blot results in Figure 4B. Altogether, these data prove that retinol enhances the expression of Nanog in ESCs.

Retinol-Mediated Activation of Nanog Is Independent of Retinoic Acid

In somatic cells, the retinol is metabolized to retinoic acid, which causes differentiation by binding to retinoic acid receptors and activating retinoic acid response elements containing genes [28]. In blood plasma, retinol complexes with a specific 21-kDa retinol binding protein (RBP) and transthyretin protein (TTY) for delivery to the different organs [29]. Two independent mechanisms have been proposed for the transport of retinol into target cell: through a RBP receptor stimulated by retinoic acid 6 [30] and via passive diffusion. In the cytoplasm, retinol is metabolized by two families of alcohol dehydrogenases (Adh1 and Ad4) into retinaldehyde, which is then converted into retinaldehyde by RALDH2, which binds to cellular retinoic acid binding proteins (CRABPI and CRABPII) [28]. These proteins shuttle retinoic acid from cytoplasm to the nucleus, where it interacts with specific receptors and activates the gene transcription [31].

To investigate the potential mechanism of retinol function, ESCs were analyzed for the enzymes that cause metabolism of retinol into retinoic acid. RT-PCR analysis shown in Figure 4D revealed that normal ESCs (lane 2) and retinol-treated ESCs (lane 3) do not express Adh1 and Adh4 enzymes. Differentiated cells such as adult liver cells (Figure 4D, lane 1) and primary mouse fibroblasts (Figure 4D, lane 5) on the other hand, express both of these enzymes. ESCs treated with retinoic acid (Figure 4D, lane 4) clearly showed the expression of both Adh1 and Adh4.

Further analysis revealed that undifferentiated ESCs (Figure 4D, lane 2) and retinol-treated ESCs (Figure 4D, lane 3) also do not express RALDH2, the enzyme that carries out the terminal step to metabolize retinol into retinoic acid. The cells treated with retinoic acid (Figure 4D, lane 4) and adult liver cells (Figure 4D, lane 1) also did not show any expression of this enzyme. Absence of RALDH2 in liver cells (Figure 4D, lane 1) supports the earlier published results [32]. Therefore, based on the observations above, it can be concluded that undifferentiated ESCs cannot metabolize retinol into retinoic acid. Recently it was shown that ESCs do not contain CRABP I and II, the critical proteins that shuttle retinoic acid into the nucleus [31], which further supports our conclusion that self-renewal of ESCs by retinol occurs via a pathway that is independent of retinoic acid.

Retinoic Acid Induces Differentiation of ESCs

To confirm that retinol-mediated self-renewal does not involve retinoic acid, feeder-independent R1 ESCs were treated with 0.5 μM retinol and retinoic acid separately for 4 days followed by AP staining. Treatment of cells with retinoic acid caused complete differentiation of ESCs as observed by the flat and differentiated AP-negative colonies (Fig. 4E, +RA), whereas all the colonies in retinol-treated cultures (+Re) showed strong positive AP staining. Untreated cells (C) also showed flat colonies that were negative for AP staining. Treatment of cells with retinol and retinoic acid together also caused differentiation similar to retinoic acid alone, suggesting that retinol cannot prevent retinoic acid-induced cell differentiation (data not shown).

Rapid Loss of Pluripotency in LIF-Depleted Feeder-Independent ESC Cultures

Earlier reports have shown that retinol treatment of feeder-independent ESCs cultured for 96 hours in the absence of LIF resulted in differentiation [33]. However, in all the studies presented above, retinol was added within 10–12 hours after plating the cells once they had settled down. To address the critical nature of this requirement, ESCs were cultured in LIF-free medium for different periods of time followed by RT-PCR analysis. As shown in Figure 4F, the cells expressed differentiation markers nestin and follistatin within the first 24 hours of culture (lanes 3, 4, 5, and 6; 24, 48, 72, and 96 hours, respectively). Normal cells (Figure 4F, lane 1) and retinol-treated cells (Figure 4F, lane 2) in LIF-containing medium for 24 hours did not express these genes. Importantly, addition of retinol to the cells that exhibited differentiation markers failed to reverse the differentiation (data not shown). These observations explain the reason for observing cell differentiation in the earlier report [33]. The effect of origin of cells (i.e., CCE ESCs used by these investigators), however, cannot be ruled out.

Retinol-Mediated Self-Renewal Is Independent of ESC Background

R1 ESCs used above are derived from 129Sv strain of mice (A. Nagy). To evaluate whether retinol has similar effect on the cells derived from other strains of mice, feeder-independent cultures were established using 20,000 cells from FVB/N, C57BL6, and 129Ola mice. The cells were treated with retinol for 6 days followed by AP staining (Fig. 5A). As observed with R1 ESCs, more than 90% of the colonies showed strong positive AP staining in the retinol-treated (+Re) cells compared with untreated cells (−Re). The untreated cells differentiated completely as noticed by flat colonies in all the cell lines. The cells also maintained pluripotency over multiple passages (data not shown).

Figure Figure 5..

Alkaline phosphatase staining and Western blot analysis of FVB, C57BL, and 129Ola embryonic stem cells (ESCs). (A): Retinol-treated (+Re) FVB, C57BL, and 129Ola cells show undifferentiated colonies with strong positive staining for alkaline phosphatase compared with untreated cells (−Re). (B): Western blot analysis of 6-day retinol-treated (+RE) and untreated (−RE) FVB and C57BL6 ESCs for Nanog and Oct4.

To confirm data from AP staining, feeder-independent cultures of FVB/N and C57BL6 ESCs were also analyzed by Western blot analysis (Fig. 5B). Six-day retinol-treated FVB/N and C57BL6 cells (+Re) showed high expression of Nanog and Oct4, whereas the untreated cells showed complete absence of Nanog (−Re), indicating differentiation in the absence of feeder cells. These data and our observations with R1 cells clearly demonstrate that retinol-mediated self-renewal is independent of the strain background of the ESCs.


Pluripotent ESCs offer great promise to serve as an unlimited source of therapeutic material for regenerative medicine. The undifferentiated ESCs express genes such as oct4, sox2, stat3, and nanog, which disappear as the cells differentiate. Mouse and human ESCs exhibit special requirements for feeder cells to maintain their pluripotency and self-renewal. Use of feeder cells, however, involves several cumbersome and time-consuming steps such as generation of MEFs from mouse embryos, inactivation by mitomycin C or by γ-irradiation, and removal from ESCs before use for subsequent studies. Variations in MEFs can also influence the growth of ESCs (our unpublished observations). The conditions that support feeder-independent culture of ESCs, therefore, can greatly simplify the culture procedure and the interpretation of results.

ESCs require a circuitry of different signaling pathways, including LIF/Stat3 pathway, BMP-mediated activation of ids (inhibitor of differentiation genes), Wnt/β-catenin signaling, oct3/4, and sox2 and nanog genes [34], for maintaining their pluripotency. Gene knockout studies have demonstrated that oct3/4 and nanog are essential for self-renewal of ESCs [17, 19, 20]. Although the expression of nanog is unaffected by the absence of oct3/4, a sustained expression of oct3/4 is essential for the self-renewal of ESCs. Maintenance of self-renewal by nanog proves the critical importance of this gene for pluripotency of ESCs [19, 20]. Mouse and human ESCs require MEFs for their growth and self-renewal. Many laboratories have tried alternate procedures such as chemically defined medium [35], small molecules [36, 37], and conditioned media from the feeder cells [38, 39]. Our studies present strong evidence that retinol can effectively support the self-renewal of mouse ESCs in the absence of feeder cells in long-term cultures. The cells maintained high expression of pluripotent cell-specific genes oct3/4, nanog, sox2, stat3, and rex1 at all the passages. However, similar to normal ESCs, retinol-treated cells also differentiate spontaneously within 72–96 hours after withdrawal of retinol, the effect of which cannot be reversed.

Retinol-treated cells maintain complete potential to differentiate into all types of cells in long-term cultures. The embryoid bodies developed from these cells expressed genes such as nestin, Brachyury, and GATA4 as markers for ectoderm, mesoderm, and endoderm, respectively. In addition, the cells remain fully competent to form high-degree chimeric animals after microinjection into blastocysts as observed by their contribution to different organs in E18.0 fetus and in newborn pups (Fig. 3). Specifically, high contribution of cells into testis indicates germ line potential of these cells. Integration of cells into compacted morulas further supports the undifferentiated characteristics of these cells. Therefore, the presence of cells at all the stages, including blastocyst formed after coculture with compacted morula, clearly indicates that retinol does not affect the pluripotency of ESCs. Further, the chimeric animals born from retinol-treated cells did not exhibit any apparent abnormalities or lethality.

Retinol, a small molecule of 286 dalton, is an important dietary component. It has diverse effects on biological functions including cell differentiation, embryonic development, and reproduction. It is transported from liver to the target cells bound to a heteroduplex of RBP and TTY [29]. In the cytoplasm, it binds to cellular RBP and is metabolized to retinoic acid by enzymes such as Adh1, Adh4, and RALDH2. Retinoic acid then forms a complex with retinoic acid receptors (RAR and RXR receptors) in the nucleus and binds to the retinoic acid response element to activate responsive genes [28]. Our data (Fig. 4D) demonstrate that ESCs do not contain enzymes to metabolize retinol into retinoic acid. In addition the proteins, CRABP I and II, that shuttle retinoic acid into the nucleus are also not expressed in undifferentiated ESCs [31], which indicates that retinol function in ESCs does not involve retinoic acid.

The absence of retinoic acid involvement suggests the possibility of some novel retinol signaling mechanism unique to the ESCs. Our recent studies have shown that retinol-mediated overexpression of nanog is linked to phosphoinositide-3 kinase (PI3K) signaling pathway (unpublished data). In the light of a recent report that PI3K signaling regulates nanog expression [40], it will be exciting to find the potential connection of retinol with this pathway. Since other metabolites of retinol are shown to cause differentiation [33], we speculate that retinol may be directly involved in this function. The studies are currently in progress to test this hypothesis.

Feeder-independent cultures supplemented with retinol offer many advantages, such as (a) high percentage of colonies remain undifferentiated; (b) pure population of ESCs can be easily obtained for creating chimeric mice or for lineage-restricted differentiation in bioreactors for industrial-scale production; (c) retinol is less likely to have detrimental effect on the cells since it is an important dietary component; (d) the optimal effective dose is far below the 1–2 μM physiological concentrations [41]; (e) the effect of retinol is independent of the strain background and, therefore, may have universal application; and (f) most of all, retinol induces cells to turn on their own machinery to overexpress an essential gene for self-renewal. Since nanog is highly conserved between human and mouse [19], we expect that these observations may have application for human ESCs as well. Studies are currently in progress to test whether human ESCs also respond to the retinol in similar fashion.


ESCs maintain their pluripotency by a complex interplay of different signaling mechanisms and maintenance of genes such as stat3, oct4, sox2, and nanog, of which nanog plays a key role in maintaining the pluripotency of the ESCs. Coculture of ESCs with mouse embryonic fibroblast feeder cells can prevent spontaneous differentiation. The use of feeder cells in ESC cultures, however, involves several cumbersome and time-consuming steps. Therefore, feeder-independent cultures can greatly simplify the recovery of pure population of cells for subsequent studies. The data presented here demonstrate for the first time that retinol can support feeder-independent self-renewal of mouse ESCs in long-term cultures while maintaining the high expression of nanog. The cells maintain complete potential for in vivo and in vitro differentiation. The effect of retinol is independent of the ESC background and does not involve retinol. The studies may have significance for maintaining the pure population of ESCs for linage-restricted differentiation and drug discovery studies.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.


We thank Dr. A. Nagy for R1 ESCs. E14 cells were purchased from BayGenomics (San Francisco, We also thank Ms. Joyce Dawes and Ms. Minying Yang for technical assistance. We thank Drs. George Michalopoulos, Devjani Chatterjee, and Qingde Wang for their critical comments. We also thank Dr. A Levine, Dean, University of Pittsburgh School of Medicine, for supporting these studies.