mNanog Possesses Dorsal Mesoderm-Inducing Ability by Modulating Both BMP and Activin/Nodal Signaling in Xenopus Ectodermal Cells

Background In Xenopus early embryogenesis, various genes are involved with mesoderm formation. In particular, dorsal mesoderm contains the organizer region and induces neural tissues through the inhibition of bone morphogenetic protein (BMP) signaling. In our initial study to identify novel genes necessary for maintaining the undifferentiated state, we unexpectedly revealed mesoderm-inducing activity for mNanog in Xenopus. Methodology/Principal Findings The present series of experiments investigated the effect of mNanog gene expression on Xenopus embryo. Ectopic expression of mNanog induced dorsal mesoderm gene activity, secondary axis formation, and weakly upregulated Activin/nodal signaling. The injection of mNanog also effectively inhibited the target genes of BMP signaling, while Xvent2 injection downregulated the dorsal mesoderm gene expression induced by mNanog injection. Conclusions/Significance These results suggested that mNanog expression induces dorsal mesoderm by regulating both Activin/nodal signaling and BMP signaling in Xenopus. This finding highlights the possibly novel function for mNanog in stimulating the endogenous gene network in Xenopus mesoderm formation.


Introduction
Germ layer formation is one of the most important processes in the fundamental patterning of an embryo. In Xenopus early embryogenesis, mesoderm is induced by signals secreted from endodermal tissue during the blastula stage, and nodal-related (Xnr) genes are known to play important roles in this biological process. VegT and Wnt signaling induces ¥¥nr5/6, followed by the sequential upregulation of Xnr1/2 and Xnr4 [1][2][3], and consequently, various mesoderm gene activities. Activin A, a TGF-ß superfamily member, was first identified as a factor that could induce both ventral and dorsal mesoderm [4]. In dorsal mesoderm, also called the Spemann-Mangold organizer, several genes including chordin (chd), noggin (nog), goosecoid (gsc), and xlim-1 are expressed to induce neural tissues in the presumptive neuroectoderm [5][6][7][8].
Xenopus blastula ectodermal cells, or animal cap (AC) cells, possess multipotency and can differentiate into many types of tissues including mesoderm. However, the period for mesoderm induction in AC is limited until early gastrula. This phenomenon is known as ''loss of mesodermal competence'' (LMC) [9]. To identify novel factors involved with maintaining multipotency in Xenopus embryo, we first attempted to find genes involved in releasing LMC.
The first candidate gene we examined was mNanog, which encodes a homeodomain protein and is efficiently expressed in mammalian embryonic stem (ES)/induced pluripotent stem (iPS) cells [10][11][12]. Our preliminary experiments revealed that in the presence of Activin A treatment, mNanog injection promotes AC elongation and some mesodermal gene expression even at the late gastrula stage (data not shown). We also unexpectedly found that mNanog injection induces AC elongation without Activin A treatment and could promote the expression of dorsal mesoderm genes such as chd, gsc, and xlim-1 in AC. Further experiments revealed showed that mNanog also weakly promotes Activin/nodal signaling and inhibits BMP signaling. Together, these data indicated that mNanog modulates both these signaling pathways to induce the dorsal mesoderm cell fate in Xenopus AC, suggesting a novel function for mNanog in embryogenesis.

Microinjection
Microinjecion was performed using a picojector (Harvard Medical Instruments). RNA for injection was synthesized with the mMESSAGE mMACHINE SP6 kit (Ambion/Applied Biosystems). Injected embryo was obtained by artificial fertilization and dejellied with 4.6% L-cysteine hydrochloride solution. Injection was performed in 5% Ficoll/1 X Steinberg's Solution (SS). Injected embryos were cultured in 0.1 X SS solution. Xenopus maintenance was carried out in compliance with institutional regulations and all Xenopus experiments were approved by the institutional ethics committee noted above (#21-10 and #24-8).
Animal cap assay mRNA was injected into the animal pole region of 2-cell-stage embryos. ACs were dissected at the late blastula stage (Stage 9), and then cultured to the appropriate stage with/without treatment with 10 ng/ml of Activin A. The shape of treated ACs was observed at about 12 hours after treatment. Treated AC was also assessed by the expressions of several marker genes.

In situ hybridization
Embryos were bleached in hydrogen peroxide-methanol before fixation in MEMFA (formaldehyde-MOPS solution) and dehydration with ethanol. Rehydrated embryos were hybridized with DIG-labeled probe for 24 h at 60uC. Embryos were then incubated with 20006 anti-DIG antibody (Roche) for 12 h, washed 5 times, and then visualized by reaction in NBT/BCIP solution (Roche).

TUNEL staining
In situ TUNEL assay for detecting apoptotic cells were carried out by previous method [17]. Briefly, fixed and bleached embryos were incubated with TdT enzyme (Invitrogen) and DIG-dUTP (Roche) for 1day. After washing, embryos were incubated with anti-DIG antibody, washed with MAB and detected with BMpurple (Roche).

Cycloheximide (CHX) treatment
The procedure for CHX treatment was basically carried out as previously described [18]. Normal or Injected embryos were treated with 40 ng/ml of CHX in 16Steinberg's solution at Stage 7, and was homogenized at Stage 9.

Xenopus Nanog gene cloning
To clone the Xenopus homolog of Nanog gene, we carried out degenerated PCR with following primers:

mNanog injection stimulated mesoderm-inducing activity in AC
At first, we confirmed the expression of mNanog protein in Xenopus embryo. By Western blot analysis, we could detect a protein of 40 kDa, consistent with the molecular size of the mNanog protein (Fig. 1A). Immunohistochemistry with anti-mNanog antibody showed intense mNanog reactivity in the nuclei of mNanog-injected embryos (Fig. 1B, C). Next, we examined the effects of mNanog on Xenopus early embryogenesis. 200 pg of mNanog mRNA injected into the animal pole of 4-cell embryos caused a defect in the anterior region at the late neural stage (Fig. 1D, E), although no obvious developmental delay was observed (data not shown). In 3-day-old tadpoles, head defects with small eye vesicles could be seen (Fig. 1G, Table S1). This head defect was more intense and lethality was also strikingly increased by injection with 400 pg of mNanog (Table S1), although the lethality did not manifest until the neural stage (data not shown). To examine whether the head defect occurred by apoptosis, we carried out terminal deoxynucleotidyl transferasemediated deoxyuridine-triphosphate nick end-labeling (TUNEL) assays. mNanog injection increased the number of apoptosis-positive cells, suggesting that the head defect was due to apoptosis in the head region ( Fig. 1H-J). We then observed the AC shapes. Injection of 200 pg of mNanog slightly elongated the AC in the absence of Activin A treatment (Fig. 1K, L), but less so with Activin A treatment (Fig. 1M). This elongation was dependent on the dose of injected mNanog (Fig. 1O). On the other hand, elongation of AC by Activin A was suppressed by injection with mNanog ( Fig. 1N, O). Indeed, RT-PCR analysis of stage-18 AC revealed that mNanog injection decreased expression of mesoderm genes such as ms-actin [19] and Xbra [20] in Activin A-treated AC, whereas the expressions of notochord markers, chd and Xnot, were upregulated (Fig. 1P). Furthermore, injection of 200 pg of mNanog mRNA into the ventral hemispheres of 4-cell embryos induced a weak secondary axis formation (Fig. 1Q, R). This induced axis did not include a head structure with eye vesicles (Fig. 1S), suggesting that mNanog may function not as a positive regulator of canonical Wnt signaling like siamois (sia) [21], but instead as a BMP inhibitor like chd and truncated-type BMP receptor (tBR) [5,22]. Furthermore, HE staining of mNanog-injected tadpole revealed both neural structures and notochord (Fig. 1T, U). Together, these results raised the possibility that mNanog possesses dorsal mesoderminducing activity in Xenopus embryo. mNanog injection promoted expression of dorsal mesodermal genes, but inhibited ventral mesodermal genes in both AC and embryos The phenotypes of mNanog-injected embryos and their corresponding ACs suggested to us that mNanog could induce dorsal mesodermal tissues. We next performed RT-PCR analysis to examine the expression of mesodermal genes in earlier stages. When 200 pg of mNanog mRNA was injected into 2-cell embryos, the expression of dorsal mesodermal marker genes chd, gsc, and xlim-1 was increased in stage-11 ACs without Activin A treatment ( Fig. 2A 1st-3rd column; lane 3, 5), and 400 pg of mNanog mRNA injection further increased these gene expressions ( Fig. 2A  On the other hand, Xbra expression was not effectively induced by mNanog injection ( Fig. 2A 4th column, lane 3, 5, 7), and induction of Xbra expression by Activin A treatment was clearly inhibited by mNanog ( Fig. 2A 4th Fig. 2A 5th-8th columns). To assess whether the enhancement of dorsal mesodermal gene expressions was specific for mNanog function, we carried out RT-PCR with a deletion mutant of mNanog that produces a protein lacking the Cterminus domain including the W-repeat motif (mNanogDCD; Fig. 2B) [28,29]. Dorsal marker gene expression was not induced by mNanogDCD (Fig. 2B, 1st-3rd columns). Quantitative analysis of the mNanog mRNA also suggested that mNanog function in mesoderm induction requires dimerization of the mNanog protein (Fig. 2B).
To examine the effect of mNanog on endogenous mesodermal gene expressions, we performed in situ hybridization. Endogenous chd expression was observed in the dorsal lip region (Fig. 2C, black arrow), and only the control lacZ injection did not affect chd expression (Fig. 2C, white arrow). When mNanog was injected into the ventral marginal zone, ectopic chd expression was obviously induced (Fig. 2D, white arrow), suggesting that mNanog can induce chd expression in embryo, confirming the RT-PCR analysis. Xbra expression was seen around the yolk plug in normal embryo (Fig. 2E), but was specifically inhibited in the mNanog-injected area (Fig. 2F, white arrow), suggesting that mNanog negatively regulates Xbra expression. These data also indicated that mNanog affects the endogenous expression of mesodermal genes in Xenopus embryo.
To further profile the mechanism of mesoderm induction driven by mNanog, we next compared the expression of mesodermal marker genes between Activin A treatment and mNanog injection. AC from normal embryo did not express any mesodermal genes (Fig. 2G, lane 2), but following treatment with Activin A at the dose of 1-10 ng/ml, chd and gsc were expressed in a dosedependent manner (Fig. 2G, lane 6-7). Xbra was also efficiently expressed following both 1 ng/ml and 10 ng/ml Activin A treatment (Fig. 2G, lane 6-7). When mNanog was injected, gsc and chd expressions gradually increased (Fig. 2G, lane 3-5), as did Xbra expression, although the effect of mNanog injection on Xbra expression was less enhanced than that induced by Activin A treatment (Fig. 2G, 3rd column).
Several mesodermal genes including chd are induced by overexpression of canonical Wnt signaling and Xnr genes [5,30]. We therefore examined the expression of early canonical Wnt signaling target genes in our system. There was no increased expression of sia and Xnr3, known targets of canonical Wnt signaling, despite the injection of mNanog mRNA [21,31] (Fig. 2H). This result suggested that mNanog does not affect canonical Wnt signaling in the embryos stages we examined. mNanog subsidiary utilizes Activin-nodal signaling for dorsal-mesoderm induction Previously, it was shown that both mesoderm and endoderm formation requires activation of Activin/nodal signaling. Thus, we next examined whether the expression of Xnr genes is induced by mNanog. RT-PCR analysis indicated that Xnr1 and Xnr2 expressions were increased in a dose-dependent manner (Fig. 3A). On the other hand, expression of Xnr5/6 was not increased in mNanoginjected AC (Fig. 3B). From these results, we proposed that mesoderm induction by mNanog involves the upregulation of not Xnr5/6, but Xnr1/2. To assess whether mNanog overexpression promotes the nuclear transport of Smad2, we coinjected embryos with mNanog and Smad2GFP [32]. Without mNanog injection, GFP signal was observed in the cytoplasm of AC cells (Fig. 3C), whereas 10 pg of Xnr5 injection promoted a nuclear localization of the GFP signal (Fig. 3E). When 200 pg of mNanog was coinjected, Smad2GFP signal was occasionally observed in nuclei, although the efficiency was low (Fig. 3D). This result suggested that, at least in some cases, mNanog regulates Activin/nodal signaling through Xnr1/2.
Next, to clarify whether mesodermal gene induction was dependent on Activin signaling, coinjection experiments were performed with a truncated form of the type I Activin receptor (tALK4) [15], which acts as a dominant-negative mutant. Indeed,  tALK4 clearly suppressed Xnr1, Xnr2, gsc, and chd expressions, but not those of Xbra and xWnt8 (Fig. 3F). When tALK4 was injected, expression of Xnr1 and chd induced by mNanog was slightly suppressed, but Xnr2 and gsc expression was little changed (Fig. 3F,  lane 6, 7). We further analyzed the effect of cleavage mutants of Xnr1 and Xnr2 (cmXnr1 and cmXnr2) on mesodermal gene induction by mNanog. Although cmXnr1 and/or cmXnr2 were injected, chd expression was only slightly decreased (Fig. 3G, lane 3-5). With xlim-1, coinjection with both cmXnr1 and cmXnr2 slightly reduced their respective expressions (Fig. 3G, 3rd column, lane 2, 5). Together, these results suggested that mNanog weakly modulates Activin/nodal signaling, but that Activin/nodal signaling is not the main factor in mesoderm gene induction by mNanog.

mNanog modulated dorsal mesodermal marker genes by regulating BMP signaling via Xvent2
We finally examined other marker gene expressions. It is known that dorso-ventral specification in mesodermal tissue involves BMP signaling, and previous reports indicated that Xvent1 and Xvent2 facilitate BMP4 transcription, directing ventral mesodermal cell fate [16]. Our results also showed that Activin treatment induced Xvent1, Xvent2, and BMP4 expressions in AC (Fig. 4A), and when mNanog was injected, these gene expressions were obviously decreased (Fig. 4A, lane 4-6). These results suggested that dorsal mesoderm induction by mNanog is dependent on BMP signaling.
Thus, we next examined the effect of coinjecting mNanog and Xvent2. Xvent2 and 200 pg of mNanog gradually inhibited the expressions of chd, gsc, and xlim-1 in a dose-dependent manner (Fig. 4B, 1st-3rd columns). BMP4 expression was detected in normal AC, and mNanog injection inhibited such expression (Fig. 4B, 4th column, lane 2). Interestingly, coinjection of mNanog with Xvent2 rescued the BMP4 expression (Fig. 4B, 4th column, lane [3][4][5], suggesting that mNanog suppresses Xvent2 transcription, resulting in the decreased BMP4 signaling and promotion of dorsal mesodermal gene expression. To clarify whether mNanog function directly or indirectly affects the dorsal mesoderm gene expression, we used CHX treatment to block protein translation. Applying CHX to AC inhibited the expressions of chd, gsc, and Xnr2 (Fig. 4C, lane 3, 4), and decreased the expressions of Xvent1 and Xvent2 (Fig. 4C, lane 3, 4). These data suggested that both induction of mesoderm genes and inhibition of Xvent2 expression could be indirectly regulated by mNanog.

Discussion
In this study, we showed a novel function of the Nanog gene in Xenopus embryo. In the process of LMC analysis with mNanog, we found that mNanog induces elongation of AC and expression of mesoderm marker genes. Both RT-PCR and in situ hybridization showed that mNanog effectively induces dorsal mesoderm marker genes, but not ventral mesodermal genes. This is also shown in Fig. 2G as a difference in marker gene induction between mNanog injection and Activin A treatment.
On the other hand, elongation of Activin A-treated AC was suppressed by mNanog injection (Fig. 1L-M). The expression of ventral mesodermal genes in Activin A -treated AC was inhibited by mNanog in both stage-11 and stage-18 embryos (Fig. 1O and Fig. 2A). This inhibitory effect was also observed by in situ hybridization for Xbra (Fig. 2E, F). Recent study showed Nanog functions like xVent, supporting this result [33]. We think that this effect of mNanog would be due to upregulation of chd and gsc, resulting in downregulation of ventral mesoderm gene expression. In the case of Activin A -treated cap or whole embryo under mesoderm-inducing conditions, upregulation of chd and gsc may drive suppression of the ventral mesoderm gene expression (such as Xbra, Xwnt8, mix, mixer) via gsc and chd. Indeed, as shown in Fig. 2A, dorsal mesoderm gene expression was not decreased by mNanog injection. And, untreated AC showed upregulation of several meso/endoderm genes such as Xwnt8, Cer, and Sox17a. In Zebrafish embryo, depletion of Nanog-like caused inhibition of Sox17expression [34]. Furthermore, it is shown that Xvent1 could not substitute for Nanog function [35]. We think that, in AC cells (without Activin treatment), only upregulation effects could be observed because these ACs have no potential to become ventral mesoderm. In any case, Nanog function in mesoderm formation is thought to be complicated, thus further studies need to be done to clarify detail mechanisms.
The mNanog injection also caused head defect, and results from the TUNEL assay implicated cell death in the anterior (injected) region as an underlying cause. Injection with 400 pg of mNanog induced high lethality in 3-day tadpole (Table S1), confirming the severe effects in mNanog-injected regions. We also propose that ectopic expression of a gene possessing mesoderm-inducing activity could affect normal head development. Indeed, 0.25 pg of Xnr5 injection into animal pole regions caused a similar head defect (data not shown).
In this study, mNanog overexpression promoted neither sia/Xnr3 nor Xnr5/Xnr6 expressions (Fig. 2H, 3B), suggesting that mNanog could not affect early embryonic signaling such as canonical Wnt signaling and maternal Nodal signaling. On the other hand, both Xnr1 and Xnr2 expressions were enhanced by mNanog injection (Fig. 3A). The simplest idea to account for these findings is that mNanog upregulates Xnr1/2 transcription, promoting Activin/ nodal signaling and gsc/chd transcription. However, RT-PCR analysis with tALK4, cmXnr1, and cmXnr2 showed that these dominant-negative genes did not effectively inhibit dorsal mesoderm gene expression (Fig. 3F, G). Nevertheless, mNanog actually induced Xnr2, and tALK4 weakly suppressed Xnr1 and chd expression, thus it is suggested that mNanog, at least partially, modulates Xnr signaling and contributes to dorsal mesoderm gene induction.
In Fig. 4, we showed that dorsal mesoderm induction by mNanog is closely involved with inhibition of BMP signaling. Indeed, mNanog injection inhibited Xvent1, Xvent2, and BMP4 gene expressions (Fig. 4A), and coinjection of mNanog with Xvent2 clearly suppressed chd, gsc, and xlim-1 expression (Fig. 4B). Together with the CHX experiment, our data implicated the dorsal mesoderm-inducing activities of mNanog in the modulation of BMP signaling, possibly by indirectly regulating Xvent1/2 expression. Our results can be used to propose a model for the modulation and induction of mesoderm genes (Fig. 4D) In short, mNanog positively regulates Xnr2, but it inhibits expression of BMP factors such as Xvent1/2 and BMP4, resulting in induction of chd and gsc. This function is similar to that of Tsukushi (TSK), which modulates both nodal and BMP signaling [36], suggesting that mNanog might be involved with the regulation of TSK.
Even though our experiments were conducted in an artificial system, we think they are still important in clarifying a novel mechanism involving mNanog function, as well as suggesting a novel means of endogenous mesodermal induction in Xenopus. This proposed mNanog function of mesoderm induction in itself seems opposite to its role in maintaining the undifferentiated state. However, Nanog is a possible target gene of Activin signaling [37,38], and low doses of Activin A are important in maintaining the pluripotency of ES cells in some conditions [39,40]. Although our results indicated involvement of mNanog in Activin/nodal signaling, they also suggested that mNanog contributes, at least in part, to the gene regulation mechanism around Activin/nodal signaling that underpins mesoderm formation in Xenopus. We expect that other factors involved with pluripotency, like Oct3/4 and Sox2, could also induce activity similar to that observed with mNanog, although our preliminary findings showed no mesoderm gene induction following coinjection with xSox2 or Oct61 (data not shown).
This study sought to identify the Xenopus gene homolog of mammalian Nanog by using sequences of axolotl and newt [41,42]. Although we designed six primers in homeodomain and caspase domain ( Fig. S1 and M&M section) and performed seven rounds of degenerate PCR using combination of these primers, we failed to find any sequence identified as xNanog, although many identified were similar genes including Xvent1 (6/16) and Hoxd11 (6/16) (Fig.  S1). Moreover, whole genome analysis of Xenopus tropicalis revealed no known nucleotide sequence for the XtNanog gene. Further exploration of Xenopus Nanog or another factor that substitutes for Nanog is obviously needed.