Activin-Nodal signaling is involved in propagation of mouse embryonic stem cells

One of the most important characteristics of stem cells is their ability to self-renew. Self-renewal is achieved by suppression of differentiation and stimulation of proliferation. Embryonic stem (ES) cells are self-renewing cells derived from the inner cell mass (ICM) of blastocysts (Niwa, 2001). They have the ability to maintain pluripotency to differentiate into all types of cells of the three germ layers, and are expected to be of great use in regenerative medicine. The signaling instructions that govern these characteristics are provided by growth factors in the stem cell niche microenvironment (Schofield, 1978). Identification of these growth factors and extending such knowledge to control ES cell propagation would improve understanding of the basic biology of ES cells and may yield therapeutic benefits in regenerative medicine. However, the signaling pathways that govern the proliferation of ES cells remain largely unknown.

At present, mouse ES (mES) cells can be propagated in medium containing fetal calf serum (FCS) and cytokine leukemia inhibitory factor (LIF) without the support of feeder cells (Smith et al., 1988; Niwa, 2001). The effect of LIF is mediated through a cell-surface complex composed of LIFRβ and gp130. Upon ligand binding, gp130 activates Janus-associated tyrosine kinases (JAK) and their downstream component signal transducer and activator of transcription (STAT)-3. Although activation of STAT3 is necessary and sufficient for suppression of differentiation of mES cells (Niwa et al., 1998; Matsuda et al., 1999), LIF does not appear to regulate the proliferation of mES cells directly (Raz et al., 1999; Viswanathan et al., 2002). These findings suggest that unidentified growth factors provided by serum or feeder cells and/or in an autocrine fashion by ES cells could contribute to self-renewal of ES cells.

Several lines of evidence suggest that the signaling pathways mediated by the members of the transforming growth factor β (TGFβ) superfamily play important roles in the biology of epiblasts and ES cells. The TGFβ superfamily includes nearly 30 proteins in mammals, e.g. TGFβ, activin, Nodal and bone morphogenetic proteins (BMPs), and its members have a broad array of biological activities. Members of the TGFβ superfamily signal via heteromeric complexes of type I and type II receptors (Heldin et al., 1997). Upon ligand binding, the constitutively active type II receptor kinase phosphorylates the type I receptor which, in turn, activates intracellular signaling cascades including Smad pathways. Activins, TGFβs and Nodal bind to type I receptors known as activin receptor-like kinase (ALK)-4, ALK-5 and ALK-7, respectively. In addition, Cripto serves as a co-receptor for Nodal in conjunction with ALK-4. The activated type I receptors phosphorylate receptor-regulated Smad proteins (R-Smads). Smad2 and Smad3 transduce signals for TGFβ, activin and Nodal, whereas Smad1, Smad5 and Smad8 are activated by BMP type I receptors (Massague, 1998). Activated R-Smads form complexes with common-partner Smad (Co-Smad, i.e. Smad4), translocate into the nucleus, and regulate the expression of target genes in cooperation with various transcription factors such as those of the FAST-FoxH family. Smad6 and Smad7 are inhibitory Smads (I-Smads) (Imamura et al., 1997; Nakao et al., 1997), and have been reported to exhibit significant differences in the manner of inhibition of TGFβ superfamily signaling. BMP signaling is inhibited by both Smad6 and Smad7, whereas activin-Nodal-TGFβ signaling is more potently inhibited by Smad7 (Hata et al., 1998; Itoh et al., 1998; Hanyu et al., 2001).

Genetic studies have shown that embryos deficient in Smad4 display defective epiblast proliferation and retarded ICM outgrowth (Sirard et al., 1998), and that Nodal null mice display very little Oct3/4 expression and substantial reduction in size of the epiblast cell population (Conlon et al., 1994; Robertson et al., 2003). Furthermore, large-scale gene profiling of embryonic and adult stem cells has revealed that TGFβ signaling networks are likely to play important roles in maintenance of the unique properties of ES cells (Ramalho-Santos et al., 2002; Ivanova et al., 2002; Brandenberger et al., 2004). Nodal signaling, in particular, has been speculated to be active in undifferentiated human ES (hES) cells, since components of Nodal signals (human orthologs of Cripto and FAST1) and a target gene (a human homolog of Lefty2) are transcriptionally enriched (Brandenberger et al., 2004). Moreover, phosphorylation and the nuclear localization of Smad2/3 were detected in undifferentiated hES cells and shown to play important roles in maintenance of their pluripotency (James et al., 2005). However, the precise roles of Smad2/3 signaling mediated by activin-Nodal-TGFβ in self-renewal of mouse and human ES cells have yet to be elucidated.

In the present study, we investigated the effects of TGFβ superfamily members on mES cell self-renewal. When activin-Nodal-TGFβ signaling was inhibited by Smad7 expression or the specific inhibitor SB-431542, mES cell propagation was dramatically decreased, whereas inhibition of BMP signaling by Smad6 expression did not. In clonal cultures with serum-free medium, supplementation of recombinant Nodal and activin increased the ES cell proliferation ratio with maintenance of the pluripotent state, but supplementation with BMP-4 did not. These findings indicate that Nodal and activin signaling promotes mES cell propagation, and that mES cells themselves produce this activity.

To study the roles of TGFβ superfamily signaling in the proliferation of ES cells cultured in FCS-containing medium, we first used various natural inhibitors such as Smad6 and 7. We confirmed the specificity of TGFβ superfamily signaling, which is inhibited by Smad6 and Smad7 in MGZ5 ES cells, with luciferase reporter assays using BMP-specific Id-1-luc and activin-Nodal-specific activin responsive elements (ARE)-luc (Chen et al., 1996; Korchynskyi and ten Dijke, 2002). Transient expression of Smad6 inhibited BMP-dependent reporter activity, whereas that of Smad7 inhibited both BMP-dependent and activin-Nodal-TGFβ-dependent reporter activities (Fig. 1A). These results suggested that the pathway specificity of inhibitory Smads is conserved in mES cells, and that BMP and activin-Nodal-TGFβ signaling is autonomously activated in mES cells in FCS-containing medium.

To examine the effects of inhibition of BMP and activin-Nodal-TGFβ signaling on self-renewal of mES cells, we overexpressed mouse Smad6 or Smad7 using an episomal vector system, which allows efficient transfection and strong expression of transgenes. Smad6 and Smad7 cDNAs were introduced into pCAG-IP supertransfection vector and then transfected into MGZ5 ES cells. Quantitative RT-PCR analysis showed that expression of Smad7 was nine times higher in the Smad7-transfected cells than in the control (supplementary material Fig. S1A). Forced expression of Smad7 significantly reduced cell number and colony size after culture in FCS-containing medium, whereas that of Smad6 yielded a smaller reduction of cell number and colony size compared with empty transfectants (Fig. 1B-D). We also calculated the cell number per colony to determine the relative proliferation. Forced expression of Smad7 decreased the ES cell proliferation ratio by about 75%, whereas that of Smad6 decreased ES cell proliferation ratio by only 10%, which was not significant (Fig. 1D).

To examine whether the growth-inhibitory effect of Smad7 is due to inhibition of activin-Nodal-TGFβ signaling, we blocked the same signal pathway with SB-431542, a synthetic molecule that inhibits the kinases of receptors for activin-Nodal-TGFβ but not those of BMPs (Laping et al., 2002; Inman et al., 2002). The size of colonies without inhibition of TGFβ signaling was larger than that of colonies of mock transfectants in episomal transfection, possibly because of the absence of drug (puromycin) selection (Fig. 1B and Fig. 2A). Addition of 10 μM SB-431542 to FCS-containing medium significantly inhibited mES cell proliferation (Fig. 2). However, SB-431542 did not reduce the cell number as strongly as forced expression of Smad7, which may inhibit the plating efficiency of mES cells. These results strongly suggest that autonomously activated activin-Nodal-TGFβ signaling, not BMP signaling, contributes to proliferation of mES cells in FCS-containing medium.

We further examined whether growth inhibition by Smad7 expression affects the characteristics of mES cells, using a reversible Smad7 expression system. As shown in Fig. 3A, fSmad7-10⁺ ES cells were generated by stable integration of the floxed-Smad7 cDNA transgene into EB3 ES cells. fSmad7-10⁺ ES cells expressing five times as many Smad7 transcripts as the control cells (supplementary material Fig. S1B), but not the DsRed transgene, were cultured in FCS-containing medium with 1 μg/ml puromycin for 1 month, and then transfected with pCAGGS-Cre. Cre-mediated recombination resulted in the generation of fSmad7-10^- ES cells, in which the Smad7 transgene had been excised and the DsRed transgene activated (Fig. 3A,B). Excision of the Smad7 transgene was confirmed by RT-PCR analysis (supplementary material Fig. S1B), which showed that the elevated expression of Smad7 reverted to normal in Smad7-10^- ES cells. We measured the growth ratios of fSmad7-10⁺, fSmad7-10^- and EB3 ES cells by determining cell numbers at 1, 3 and 5 days after seeding. Proliferation and plating efficiency of fSmad7-10⁺ ES cells was significantly decreased by expression of Smad7 transgene, as found with the episomal expression system (Fig. 3C-E). However, upon removal of the Smad7 transgene, proliferation of fSmad7-10^- ES cells recovered completely to the level of wild-type EB3 ES cells (Fig. 3C-E). Essentially similar inhibitory effects of Smad7 expression on mES cell propagation were observed using another clone fSmad7-7 (data not shown). Although stable expression of Smad7 interferes with mES growth, cell-cycle distribution of fSmad7-10⁺, fSmad7-10^- and EB3 ES cells was not significantly altered (data not shown). These results clearly indicate that the inhibitory effect of Smad7 on proliferation of mES cells is completely reversible.

To determine whether the undifferentiated state of mES cells was retained during reversible expression of the Smad7 transgene, we prepared RNA from these ES cells and investigated the expression of three stem-cell-specific genes, Oct3/4 (Niwa et al., 2000; Niwa et al., 2002), Sox2 (Yuan et al., 1995) and Zfp42/Rex1 (Rogers et al., 1991) by northern blot analysis. All stem cell marker genes were strongly expressed in both Smad7-10⁺ and Smad7-10^- ES cells (Fig. 3F) although there were differences in their expression levels (supplementary material Fig. S2). To confirm the maintenance of pluripotency in these cells, DsRed-expressing fSmad7-10^- ES cells were injected into blastocysts obtained from C57BL/6 strain mice. Embryos with DsRed fluorescence were successfully obtained, suggesting that the pluripotency of Smad7-10^- ES cells contributing to embryogenesis (Fig. 3G, left, compared with control littermates). These results indicate that temporary forced expression of Smad7 does not affect the pluripotency of ES cells.

We showed that activin-Nodal-TGFβ signaling is autonomously activated in ES cells in FCS-containing medium (Fig. 1A). We next examined whether this activation results from the agents included in FCS-containing medium. We recently demonstrated that ES cells, when cultured without FCS in serum-free medium supplemented with KSR, LIF, and ACTH, can propagate even from single cells (clonal density <25 cells/cm²) with proper pluripotency (Ogawa et al., 2004). Since serum-free medium contains no TGFβ-related molecules, we used it to investigate whether the negative effect of forced Smad7 expression depends on TGFβ-related activity in FCS. In serum-free medium, although proliferation of fSmad7-10⁺ cells was slower than that of fSmad7-10^- cells and wild-type EB3 ES cells, the inhibitory effect of forced Smad7 expression was reduced compared with that in FCS-containing medium (compare Fig. 3E with Fig. 4C). These results indicate that the growth-inhibitory effect of forced Smad7 expression is partly due to the blockade of signaling from soluble TGFβ-related molecules in FCS that promote ES cell propagation.

We next attempted to enhance ES cell proliferation by stimulating activin-Nodal-TGFβ signaling using exogenous ligands. To determine whether ES cells are capable of transducing TGFβ superfamily signals, we examined the expression of various TGFβ superfamily signaling components in mES cells by RT-PCR analysis (supplementary material Fig. S3). We detected transcripts of all of the TGFβ superfamily ligands except activin-βA and TGF-β3, suggesting that ES cells produce TGFβ superfamily ligands by themselves, at least at the transcriptional level. We also detected transcripts of type I and II receptors for activin-Nodal and BMPs but not ALK-7, which is known to be a specific receptor for Nodal, in MGZ5 and EB5 ES cells. However, we detected transcripts of Cripto-l in these ES cells, suggesting that Nodal signaling can be transduced in ES cells via ALK-4 (Yan et al., 2002). By contrast, transcripts of type II TGFβ receptors were not detected, as has been previously reported (Goumans et al., 1998), suggesting that ES cells are not capable of responding to TGFβ ligands. However, transcripts of all types of Smads were detected, suggesting that ES cells are capable of responding to activin, Nodal and BMP.

We next examined whether exogenous TGFβ superfamily ligands are capable of phosphorylating Smad2 protein. When ES cells were treated with recombinant activin or Nodal for 1 hour in serum-free medium, Smad2 protein was strongly phosphorylated (Fig. 5A). By contrast, neither BMP-4 nor TGFβ phosphorylated Smad2 protein in ES cells (Fig. 5A), indicating that the TGFβ signal cannot be integrated into ES cells, as suggested above. In addition, expression of Lefty-1 and Lefty-2, known target genes of Nodal (Juan and Hamada, 2001), was upregulated by treatment with recombinant Nodal (supplementary material Fig. S4), indicating that Nodal signaling was transduced into the nucleus in ES cells, as reported for other cells.

We then examined the effects of TGFβ superfamily ligands on ES cell proliferation in serum-free medium. Exogenous 30 ng/ml activin increased ES cell propagation ratio by 85% and 20% on Days 3 and 7, respectively. Furthermore, 1 or 2 μg/ml Nodal increased the ES cell propagation ratio on both Days 3 and 7 (Fig. 5B-D). Although 30 ng/ml exogenous BMP-4 appeared to slightly enhance ES cell propagation on Day 3, BMP-4 inhibited ES cell proliferation on Day 7 (Fig. 5B-D). These results suggest that activin and Nodal promote ES cell proliferation in serum-free medium via the canonical signal.

We previously found that serum-free medium supplemented with KSR and ACTH can maintain proper cellular pluripotency (Ogawa et al., 2004). In order to examine whether the undifferentiated state of ES cells was maintained when they were clonally expanded in serum-free medium supplemented with exogenous activin or Nodal, we determined the expression of ES cell markers such as Oct3/4 (Niwa et al., 2000; Niwa et al., 2002), Nanog (Chambers et al., 2003; Mitsui et al., 2003), Sox2 (Yuan et al., 1995), Zfp42/Rex1 (Rogers et al., 1991) and Utf1 (Okuda et al., 1998). As shown in Fig. 6A, these pluripotent state-specific genes were strongly expressed in each ES cell. To further examine the pluripotency of ES cells treated with Nodal, we subcutaneously injected Nodal-treated ES cells into nude mice. These cells generated teratomas that grew to a few centimeters in size in 3-4 weeks. Histological examination revealed that they consisted of derivatives of all three germ layers, including hair-follicle-like structures with keratohyaline granules (ectoderm), cartilage (mesoderm), and ciliated or mucus-producing epithelia (endoderm) (Fig. 6B). Moreover, these cells produced overt chimeras with ES-derived agouti-chinchilla coat color (Fig. 6C). These results suggest that activin and Nodal enhance mES cell proliferation without affecting their pluripotency.

Although the growth-stimulatory activity of Nodal and activin was detected in serum-free medium, it was weaker than the proliferative activity inhibited by overexpression of Smad7 or addition of SB-431542 to FCS-containing medium. To test for the presence of autocrine activin-Nodal activity, which might mask the effect of exogenous Nodal on ES cell proliferation, we performed luciferase reporter assay in serum-free medium without activin-Nodal supplementation. The relative activity of Id-1-luc was dramatically reduced in serum-free medium compared with that in FCS-containing medium (Fig. 7A). By contrast, removal of FCS resulted in the reduction of ARE-luc activity, which is consistent with the observation that Smad7 overexpression was less effective in repressing proliferation in serum-free culture, although it was less than that of Id-luc activity (Fig. 7A). We next investigated the effect of SB-431542 on ES cell proliferation in serum-free medium. Addition of 3 μM SB-431542 inhibited ES cell proliferation, as assessed by colony staining, cell number and proliferation intensity (Fig. 7B-D). The difference in proliferation ratio between fSmad7-10⁺ and fSmad7-10^- cells was less in serum-free medium than in FCS-containing medium. These results indicate that activin-Nodal signaling is autonomously activated in ES cells in serum-free conditions, and that activin-Nodal activity produced by ES cells masks the effects of exogenous Nodal or activin. Furthermore, signaling by soluble TGFβ-related molecules in FCS, presumably activin-Nodal, might increase endogenous activin-Nodal activity in ES cells.

In the present study, we demonstrated, to our knowledge for the first time, that Nodal and activin promote mES cell proliferation with maintenance of pluripotent state in serum-free conditions. The findings presented here strongly suggest a contribution of Nodal signaling to maintenance of rapid proliferation of mES cells.

Many homozygous mutations of Nodal-related genes in mice result in reduction of pluripotent cell propagation in early embryogenesis. ICM outgrowth is markedly reduced in Smad4^-/- blastocysts (Yang et al., 1998), and the size of the epiblast cell population is substantially reduced in Nodal^-/- or proprotein convertases for Nodal, Spc1^-/-/Spc4^-/- embryos (Conlon et al., 1994; Beck et al., 2002; Robertson et al., 2003). In Smad2^-/- mutant embryos, Oct3/4 expressing regions are diminished by 8.5 days post coitus (d.p.c.), suggesting that pluripotent epiblast cells are prematurely lost (Waldrip et al., 1998). In the peri-implantation stage of the mouse embryo, the proliferation of pluripotent stem cells is tightly regulated. In the blastocyst, pluripotent stem cells exhibit very slow proliferation, with a doubling time of 64-65 hours (Copp, 1978). However, their growth accelerates after implantation, and doubling time is shortened to 11-12 hours at embryonic day 5.5 (E5.5), 9-10 hours at E6.0 and 4-5 hours at E6.5 (Snow, 1977). In serum-free medium, mES cells exhibit very slow proliferation, as pluripotent cells in ICM, at the beginning of culture, but their growth accelerates after the formation of a small colony about 5 days later (Ogawa et al., 2004). Furthermore, since the doubling time of ES cells cultured in FCS-containing medium is about 12 hours, their physiological characteristics resemble those of the epiblast around E5.0, suggesting that growth stimulation of mES cells after the formation of a small colony might correspond to the event triggering rapid epiblast growth after implantation. Interestingly, in mouse embryos, high levels of Smad7 expression are detected during pre- and postimplantation stages when the epiblast cells are slow growing and gradually decrease from E6.5 to E7.5 while epiblast cell proliferation accelerates (Zwijsen et al., 2000), although the function of Smad7 in mouse development has not yet been reported. These findings, together with those of the present study, suggest that the dramatic changes that occur in the proliferation of pluripotent cells during implantation stages may be regulated by Nodal, as a positive regulator, and Smad7, as a negative regulator.

Consistent with our findings, several lines of evidence have suggested that activin-Nodal signaling is autonomously activated in human and mouse ES cells. The transcriptome of undifferentiated and differentiated hES cells was characterized to elucidate the signaling networks that play a role in the maintenance of the unique characteristics of hES cells (Brandenberger et al., 2004). These authors speculate that Nodal signaling is activated in undifferentiated hES cells, because components of Nodal signals (human orthologs of Cripto and FAST1) and a target gene (a human ortholog of Lefty2) are highly expressed in undifferentiated hES cells. Furthermore, phosphorylation of Smad2/3, but not of Smad1/5, was observed in undifferentiated hES cells, and decreased upon their differentiation (James et al., 2005).

We found that activin-Nodal signaling is autonomously activated in serum-free-cultured mES cells (Fig. 7), suggesting that mES cells produce activin-Nodal ligands, as supported by supplementary material Fig. S3. Notably, the target genes of Nodal include its positive regulators, such as the Nodal gene itself, and its negative regulators, such as Lefty1 and Lefty2 (Hamada et al., 2002). We failed to observe any effects of exogenous Nodal on proliferation of serum-free cultured ES cells when these cells were grown to confluence (data not shown), suggesting that Nodal ligands and/or Lefty1 and Lefty2 produced by autocrine loops of Nodal signaling in ES cells might mask the effects of exogenous Nodal or activin.

We found that exogenous addition of Nodal or activin increased mouse ES cell proliferation, whereas addition of BMP or TGFβ had no significant effects (Fig. 5B-D). The finding that TGFβ failed to affect ES cell proliferation can be explained by the lack of expression of type II receptors for TGFβ (supplementary material Fig. S3). By contrast, the failure of exogenous BMP-4 to stimulate proliferation of mES cells is inconsistent with previous reports (Ying et al., 2003a; Qi et al., 2004). BMP sustains self-renewal of mES cells in concert with LIF, and the crucial contribution of BMP to mES cell self-renewal is to induce expression of Id genes (Ying et al., 2003a). In N2B27 medium, which these authors used to test the effects of BMP, mES cells tend to differentiate into neural precursors and exhibit decreased proliferation even in the presence of LIF (Ying et al., 2003b). An autocrine BMP loop has been described in mES cells (Monteiro et al., 2004) and BMP provided by feeder cells has also been identified as a factor required for the maintenance of ES cell self-renewal (Qi et al., 2004). Since embryoid body formation studies have shown that BMP-4 inhibits neural differentiation, as found in many other vertebrate models (Wilson et al., 1995; Finley et al., 1999; Wilson and Edlund, 2001), the growth-stimulatory effects of BMP in the study by Ying et al. (Ying et al., 2003a) might have been due to its inhibitory effects on neural differentiation of mES cells intrinsically induced in serum-free conditions with N2B27 medium, in which B27 supplement contains vitamin A, a precursor of a potent differentiation inducer retinoic acid. By contrast, our serum-free culture conditions allow mES cells to maintain an undifferentiated state in the presence of LIF (Ogawa et al., 2004). To examine the possibility that endogenously activated BMP signals in our culture condition might have masked the effects of exogenous BMP-4 on mES cell proliferation, we studied the levels of Smad1/5/8 phosphorylation in the absence and presence of Noggin and BMP-4 (supplementary material Fig. S5A). Endogenous phosphorylation of Smad1/5/8 was barely detected whereas BMP-4 significantly induced their phosphorylation. While Noggin decreased Smad1/5/8 phosphorylation induced by BMP-4, it did not change the basal level of Smad1/5/8 phosphorylation, nor inhibit the mES cell propagation (supplementary material Fig. S5), suggesting that endogenous BMP signals did not play important roles under our culture conditions. We suspect that this is the reason why the present study, which focused on the effects of TGFβ superfamily members on the proliferation of undifferentiated mES cells, failed to detect growth-stimulatory effects of BMPs.

There is evidence that, during early embryogenesis, Nodal signaling plays a role in patterning of the anteroposterior body axis, formation of mesoderm and endoderm, and left-right axis patterning (Schier and Shen, 2000; Brennan et al., 2002; Eimon and Harland, 2002; Nonaka et al., 2002), suggesting that Nodal signaling induces cellular differentiation. Finding that the expression of Zfp42/Rex-1 in Smad7-10 ES cells was higher than that in Smad7-10-1 and EB3 (supplementary material Fig. S2) suggests that endogenous activin-Nodal-TGFβ signaling might also be involved in the promotion of differentiation to primitive ectoderm of mES cells. However, in human ES cells, Nodal or activin-Nodal-TGFβ signaling plays a role in the inhibition of ES cell differentiation (Vallier et al., 2004; James et al., 2005; Beattie et al., 2005). Although addition of activin or Nodal to our serum-free medium in the absence of LIF did not significantly induce differentiation of mES cells (data not shown), it remains to be determined whether the growth-stimulatory activity of activin-Nodal signaling is associated with its potential to inhibit differentiation of mES cells.

TGFβs have been shown to inhibit proliferation of many types of cells through upregulation of cell-cycle inhibitors, such as p21, p27, and p15 (Massague et al., 2000). Recently, Nodal was also shown to inhibit proliferation and induce apoptosis of human trophoblast cells through ALK7 and Smad2/3 (Munir et al., 2004). Therefore, our finding that Nodal enhances proliferation of ES cells seems exceptional. However, during formation of the anteroposterior body axis of mouse embryos, Nodal signaling provides the driving force for distal visceral endoderm (DVE) migration by stimulating the proliferation of visceral endodermal cells (Yamamoto et al., 2004), suggesting that Nodal is capable of enhancing the proliferation of certain types of cells.

To further investigate how Nodal promotes ES cell growth, we studied its effects on cell-cycle distribution and apoptosis. While activin did not affect the apoptosis of ES cells, addition of activin to ES cell culture resulted in a slight decrease in G1 ratio and subsequent increase in G2-M ratios (supplementary material Fig. S6). The overall distribution of mES cells at different cell-cycle stages was not dramatically affected, suggesting that Nodal signaling may positively regulate the cell cycle. We examined whether the stimulation of activin or Nodal might influence the expression of ERas, Utf1, B-Myb, Myc and CDK4, which are reported to be components involved in propagation of mouse ES cells (Savatier et al., 1996; Iwai et al., 2001; Takahashi et al., 2003; Cartwright et al., 2005; Nishimoto et al., 2005). We found that both Nodal and activin induced a significant increase in the expression of ERas and Utf1, whereas they induced a decrease in B-Myb expression (supplementary material Fig. S7). Clarification of the mechanisms whereby exogenous activin-Nodal signaling regulates the propagation of mouse ES cells will require further biochemical analyses using activin- or Nodal-null ES cells in which autocrine loops of activin-Nodal signal are absent.

Various signaling cascades, such as LIF-STAT3 (Smith et al., 1988; Niwa et al., 1998; Matsuda et al., 1999) and Wnt (Sato et al., 2004; Ogawa et al., 2006), and transcription factors, such as Oct3/4 (Niwa et al., 2000; Niwa et al., 2002; Niwa et al., 2005) and Nanog (Chambers et al., 2003; Mitsui et al., 2003), are thought to play roles in the maintenance of self-renewal and/or pluripotency of ES cells. Further investigation is needed to determine the molecular mechanisms by which activin-Nodal signaling cooperates with such signaling or transcriptional networks to regulate ES cell proliferation.

We previously reported that ES cells produce an activity to support their clonal propagation in serum-free medium (Ogawa et al., 2004). Stem cell colonies were never formed from single cells in low-density culture (<100 cells/cm²) in medium supplemented with 10% KSR and 1000 U/ml LIF (KSR medium). However, addition of a small amount (0.3%) of the final volume of FCS or 1-10 μM ACTH to KSR medium caused clonal ES cell propagation. We therefore hypothesized that a factor possessing ACTH-like activity is secreted by ES cell themselves, which we designated stem-cell autocrine factor (SAF). Since activin and Nodal enhance ES cell proliferation in an autocrine fashion, we questioned whether activin-Nodal activities mimic SAF activity. However, the addition of either activin or Nodal did not promote clonal propagation in KSR medium (data not shown), suggesting that at least two autocrine activities, those of SAF and activin-Nodal, support ES cell propagation in serum-free conditions. Based on our previous findings (Ogawa et al., 2004), we hypothesize that the survival of ES cells cultured in serum-free medium was supported by SAF activity at the start of culture, and that the growth of the attached ES cells is accelerated by upregulated autocrine Nodal signaling after cells have formed small colonies in low-density culture.

ES cells have attracted the interest of many researchers partly because of their promise for applications in regenerative medicine. However, such use requires that risks of pathogenic contamination caused by the addition of serum to medium be avoided. While the present findings obtained by mouse ES cells need to be carefully examined to be applicable for human ES cells, we hope that the combination of previously determined serum-free culture conditions and use of recombinant activin and/or Nodal will yield a safe and efficient protocol for propagation of ES cells in a completely chemically defined medium.

MGZ5 (derived from CCE), EB5 and EB3 (derived from E14tg2a) ES cells were maintained on feeder-free, gelatin-coated plates in LIF-supplemented medium as described previously (Niwa et al., 2002). Serum-free culture medium with knockout serum replacement (KSR, Invitrogen) and 10 μM adrenocorticotrophic hormone (ACTH; American Peptides) was described previously (Ogawa et al., 2004). To assess the clonal propagation of ES cells, EB5 cells were seeded at 300 cells/well in 12-well plates and cultured in serum-free medium supplemented with various growth factors. SB-431542 (Sigma) was prepared as previously described (Watabe et al., 2003). BMP-4 (30 ng/ml), TGFβ1 (1 ng/ml), activin (30 ng/ml), Nodal (1 or 2 μg/ml) and Noggin (90 ng/ml) were purchased from R&D Systems and used in each experiment at these concentrations unless specifically noted otherwise. The colonies that appeared were stained with Leishman's reagent (Sigma) or BCIP/NBT solution (Sigma) for alkaline phosphatase staining. Transfection of episomal expression vectors [pCAG-IRESpuropA (pCAG-IP)] into MGZ5 cells (supertransfection) was performed using Lipofectamine 2000 (Invitrogen) as described previously (Niwa et al., 2002).

pCAG-floxed-DsRed-IRESpacEGFP expression vector was constructed by ligation of loxP fragment, Discosoma red fluorescent protein (DsRed) T4 (Bevis and Glick, 2002) expression cassette derived from pBluescript SK^-DsRed T4, and enhanced green fluorescent protein (EGFP) expression cassette derived from pEGFP (Clontech) into pCAG-IP. Smad7 cDNA was introduced into the XhoI and NotI sites of pCAG-floxed-DsRed-IRESpacEGFP to obtain pCAG-fSmad7R-IRESpacEGFP.

In stable integration experiments, 2×10⁷ EB3 ES cells were electroporated with 100 μg of linearized pCAG-fSmad7R-IRESpacEGFP DNA at 800 V and 3 μF in a 0.4-cm cuvette using a Gene Pulser II (Bio-Rad Laboratories) and cultured in the presence of 1 μg/ml puromycin. After 7-10 days, colonies were isolated for clonal expansion, and one of them, named fSmad7-10⁺, was maintained in the presence of 1 μg/ml puromycin for 1 month. fSmad7-10^- ES cells were generated by Cre recombination using pCAGGS-Cre and maintained in the presence of 1 μg/ml puromycin.

MGZ5 cells were transiently transfected with an appropriate combination of reporter vector and expression plasmids together with Renilla luciferase control reporter vector (pTK-RL or pRL-CMV) using Lipofectamine 2000 (Invitrogen) as previously described (Chen et al., 1996; Yagi et al., 1999; Korchynskyi and ten Dijke, 2002; Kahata et al., 2004). Luciferase activities were normalized to the luciferase activity of co-transfected pTK-RL or pRL-CMV.

Total RNAs were prepared with ISOGEN reagent (Nippongene) or TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Oligo(dT)-primed cDNAs were prepared from 1 μg of total RNA using SuperScript reverse transcriptase (Clontech) or ReverTra Ace (Toyobo). For northern blot analysis, 5 μg of total RNA were separated on a denaturing agarose gel, and then blotted onto Hybond-N membrane (Amersham Biosciences). Probes used for northern blot analyses have been previously described (Ogawa et al., 2004). Analysis was performed with GeneImage (Amersham Biosciences) according to the manufacturer's instructions. Expression of various signaling components was compared by RT-PCR analysis. PCR products were separated by electrophoresis in 1% agarose gel and visualized with ethidium bromide. Quantitative real-time RT-PCR analysis was performed using the GeneAmp 5700 Sequence Detection System (Applied Biosystems) or MyiQ Real-Time PCR Detection Systems (Bio-Rad Laboratories). Primer sequences used for PCR reactions are described in supplementary material Table S1.

Antibodies for phospho-Smad2, Smad2/3 and phospho-Smad1/5/8 for western blot analysis were obtained from Cell Signaling. Antibodies for α-tubulin were obtained from Sigma. Western blot analysis was performed as described (Kawabata et al., 1998).

EB5 cells were cultured in serum-free medium supplemented with Nodal protein at clonal density for two weeks. Cells suspended in phosphate-buffered saline were subcutaneously injected into the flank of nude mice. After 3 weeks, the teratomas were excised, fixed in 10% paraformaldehyde, and subjected to histological examination with hematoxylin and eosin staining. fSmad 7-10^- cells were cultured in medium supplemented with 1 μg/ml puromycin for more than 1 month. EB3 cells were cultured in serum-free medium supplemented with 2 μM recombinant Nodal at clonal density for more than 1 month. Microinjection of fSmad7-10^- and EB3 ES cells into C57BL/6J blastocysts was performed according to standard procedures (Hogan et al., 1994).

EB5 cells were stimulated with 30 ng/ml Activin for 24 hours. All cells (attached and detached) were collected, washed once in PBS, and the cellular DNA was stained with propidium iodide (25 μg/ml; Sigma). The cellular DNA content was analyzed by FACS (Beckman Coulter, Fullerton, CA).

We thank R. Watanabe (Kyoto University) for providing mouse Smad6 and Smad7 cDNAs, Tetsutaro Hayashi (RIKEN, CDB) for technical support of FACS analyses and Arisa Mita (University of Tokyo) for technical support of teratoma formation assays. This work was supported in part by a Grant Aid for Scientific Research from the Ministry of Education, Science, Culture of Japan award (to H.N. and K.M.), and Leading Project (to H.N.).