Membrane translocation of t-SNARE protein syntaxin-4 abrogates ground-state pluripotency in mouse embryonic stem cells

Embryonic stem (ES) and induced pluripotent stem (iPS) cells are attractive tools for regenerative medicine therapies. However, aberrant cell populations that display flattened morphology and lose ground-state pluripotency often appear spontaneously, unless glycogen synthase kinase 3β (GSK3β) and mitogen-activated protein kinase kinase (MEK1/2) are inactivated. Here, we show that membrane translocation of the t-SNARE protein syntaxin-4 possibly is involved in this phenomenon. We found that mouse ES cells cultured without GSK3β/MEK1/2 inhibitors (2i) spontaneously extrude syntaxin-4 at the cell surface and that artificial expression of cell surface syntaxin-4 induces appreciable morphological changes and mesodermal differentiation through dephosphorylation of Akt. Transcriptome analyses revealed several candidate elements responsible for this, specifically, an E-to P-cadherin switch and a marked downregulation of Zscan4 proteins, which are DNA-binding proteins essential for ES cell pluripotency. Embryonic carcinoma cell lines F9 and P19CL6, which maintain undifferentiated states independently of Zscan4 proteins, exhibited similar cellular behaviors upon stimulation with cell surface syntaxin-4. The functional ablation of E-cadherin and overexpression of P-cadherin reproduced syntaxin-4-induced cell morphology, demonstrating that the E- to P-cadherin switch executes morphological signals from cell surface syntaxin-4. Thus, spontaneous membrane translocation of syntaxin-4 emerged as a critical element for maintenance of the stem-cell niche.

The decrease in ES cell clonogenic capacity is considered a consequence of temporal epithelial to mesenchymaltransition (EMT) 11,12 . EMT is characterized by downregulation of E-cadherin, a target of transcriptional repression by snail family proteins, and upregulation of N-cadherin 13,14 . E-cadherin downregulation is closely linked to inactivation of PI3K/Akt signals that regulates GSK3β , a key element of the Wnt/β -catenin signaling pathway in ES cells 15 . Thus, the functional modulation of these key elements is crucial for ES cell stabilization 16,17 . In addition, considerable attention has been given to P-cadherin, an alternative cadherin that is upregulated during EMT. Increased expression of P-cadherin has been detected in many cancer cells 18 , and P-cadherin appears to promote aggressive/invasive properties in several tumor cells 19 . Zscan4 family members have recently emerged as key factors that maintain ground-state pluripotency. These factors appear to play critical roles in the stability/integrity of the ES cell genome, and functional knockdown of cognate Zscan4 members resulted in decreased self-renewal potential in ES cells 20,21 . The Zscan4 family comprises six paralogs (Zscan4a-f) and three pseudogenes, the expression of which has been specifically detected in 2-cell embryos in vivo and in ES cells in vitro 22,23 . The expression of these proteins is also regulated by PI3K/Akt, suggesting the importance of this pathway for maintaining the ground state of pluripotency in ES cells 23 . Given the spontaneous appearance of aberrant cell populations, with altered morphology and differentiation potential, in clonal stem cell colonies, one might predict the existence of unidentified non-diffusible extracellular signals that can focally and temporally regulate the PI3K/Akt pathway to maintain stem cell plasticity.
In the present study, we detected local extrusion of syntaxin-4 in ES cell colonies even in the absence of differentiation stimuli. Using an inducible expression system, we found that extracellular expression of syntaxin-4 leads to dephosphorylation of Akt, morphological changes, an E-to P-cadherin switch, and the early onset of differentiation, with downregulation of Zscan4 genes, in ES cells. Thus, local membrane translocation of syntaxin-4 has emerged as a key element for stem-cell plasticity.

Expression of syntaxin-4 in mouse ES cells.
In certain cell types, a subpopulation reportedly extruded extracellular syntaxin-4 to locally propagate signals for morphogenesis and differentiation 27,30 . To determine whether this is also the case in ES cells, extracellular exposure to endogenous syntaxin-4 was investigated in cells grown in LIF-containing medium. GSK3β /MEK1/2, key elements of Wnt/mitogen-activated protein kinase (MAPK) signaling, have been proposed to play regulatory roles in ES cell stemness, and the simultaneous inhibition of these signaling pathways resulted in the maintenance of ES cells in the ground state of pluripotency 1 . We found that all cells expressed cytoplasmic syntaxin-4; however, certain cell populations extruded syntaxin-4 at lateral and apical cell surfaces when cultured without GSK3β /MEK1/2 inhibitors (2i) (Fig. 1a). The effect of 2i on the membrane translocation of syntaxin-4 was confirmed in lysate of ES cells transfected with a T7 tagged-syntaxin-4 transgene (T7stx4) ( Fig. 1b and Supplementary Fig. S1). The ratio of cells presenting extracellular syntaxin-4 was approximately 20%, which was confirmed in the transfected ES cells (Fig. 1c). These results demonstrated that clonal stem cells extrude syntaxin-4 in the absence of potent pluripotency sustaining reagents.

Extracellular syntaxin-4 results in morphological changes and mesodermal differentiation in ES cells.
To know the possible role of extruded syntaxin-4, we tested an antagonistic effect of the recombinant syntaxin-4 fragment prepared in the previous study 32 . As was the case in mammary epithelial cells, the recombinant protein of the SNARE domain (F3), but not of the helix c (F2) gave effect on ES cell morphology; ES cells formed more densely packed-colonies, indicative of restoration of the stemness (Fig. 2a). Intriguingly, a fragment of the N-terminal domain of syntaxin-4 (F1) also showed the effect similar to F3 (Fig. 2a), and this fragment clearly reduced the expression of a downstream element of cell surface syntaxin-4 in ES cells (see below, and Supplementary Fig. S2), indicating a critical role of the membrane translocation of syntaxin-4 in ES cell behaviors. To define the role of membrane-tethered extracellular syntaxin-4, we generated ES cells with inducible expression of cell surface syntaxin-4 (STstx4) using a PiggyBac transposon inducible expression system 33,34 . The transposon carried a cDNA encoding syntaxin-4 connected to cDNAs for a IL-2 signal peptide and T7-tag for extracellular localization and effective detection, respectively 27 . Cells expressed T7-tagged syntaxin-4 at the cell surface upon treatment with doxycycline (Dox) (Fig. 2b). In response to inducible expression of cell surface syntaxin-4, these cells lost their cell-cell adhesive property and showed an altered morphology, exhibiting active filopodial protrusions, after a few days, even in the presence of LIF (Fig. 2c). Such cell behaviors are indicative of EMT, which reportedly precedes cyto-differentiation in ES cells 11,12 . Parental ES cells without the syntaxin-4 transgene did not show similar morphological changes upon treatment with Dox (Fig. 2c). We found that ES cells with extracellular syntaxin-4 downregulated E-cadherin, a major mediator of epithelial cell-cell adhesion; however, its mRNA level was unchanged (Fig. 2d). Instead, these cells expressed P-cadherin, which has been shown to temporally increase in several invasive tumors to promote cellular mobility 19 (Fig. 2d), implying that an E-to P-cadherin switch might execute signals propagated by syntaxin-4. It is noteworthy that the expression of Snail and Slug, canonical transcription factors involved in tumorigenic transformation, was not increased, while the fibroblastic marker vimentin was apparently downregulated (Fig. 2e). However, these cells upregulated mesodermal markers including brachyury, a critical T-box transcription factor involved in gastrulation/mesodermal differentiation, α -smooth muscle actin, and myosin heavy chain (Fig. 2f). The stemness marker nanog was downregulated, whereas the oct3/4 level remained unchanged (Fig. 2f). We could exclude the possibility that the EMT-like cell behaviors were instructed as a consequence of the artificial gene manipulation; the phenotypical (a), Endogenous syntaxin-4 (stx4; green) was detected with anti-syntaxin-4 antibodies in ES cells before (Cell surface) and after membrane permeabilization (Total). The cells were precultured in the presence (+ 2i) or absence (− 2i) of GSK3β /MEK1/2 inhibitors for two days. As an internal control, cytoplasmic β -actin was simultaneously detected (red). Left lower column, cells were stained only with secondary antibodies (2 nd a.b.). All cells abundantly express β -actin and syntaxin-4 (left). Whereas a small population of the cells cultured in the absence of 2i display cell surface syntaxin-4 at the lateral (arrows) and apical (arrowhead) membrane, cells treated with 2i apparently decrease the extrusion of syntaxin-4 at the cell surface (right images). DMSO alone was used as a control (right images). Cell nuclei were counterstained with DAPI. Bars, 10 μ m. (b), ES cells were transfected with syntaxin-4 tagged only with T7 peptide (T7stx4) and surface expression of exogenous syntaxin-4 in cells cultured with (+ 2i) or without (+ DMSO) 2i for two days was analyzed. Upper panel, cell surface proteins were labeled with NHS-sulfo-biotin, and the cell lysate (Input) or immunoprecipitates with anti-syntaxin-4 antibodies (IP:stx4) was analyzed with HRP-labeled anti-T7 antibody (T7) or HRP-labeled streptavidin (St-Av). Lower panel, biotinylated cell surface proteins in the cell lysate (Input) or those retrieved with neutrAvidin-beads (Pull-Down) were analyzed with antibodies against syntaxin-4 or cytoplasmic β -actin. Cropped areas from the original blots ( Supplementary Fig. S1) are shown. Treatment with 2i clearly decreases the cell surface expression of T7 tagged-syntaxin-4 (*). (c), ES cells were transfected with T7 tagged syntaxin-4 (T7stx4), and exogenous syntaxin-4 was labeled with anti-T7 tag antibodies (green) after 24 h. As an internal control, cytoplasmic β -actin was simultaneously stained (red). Upper images, anti-T7 tag antibody clearly detects the syntaxin-4 transgene product, but not β -actin, at the cell surface. Bars, 10 μ m. Lower panel, the ratio of cells exhibiting syntaxin-4 at the cell surface in transfected cells. Approximately 20% of T7stx4-expressing cells (transfection efficiency, 28 ± 5%) extrudes T7stx4 at the cell surface. N = 18, ***p < 0.001.  features induced by cell surface expression of syntaxin-4 were clearly receded when the antagonistic fragment of syntaxin-4 (F1) was present in the medium ( Supplementary Fig. S2).

Identification of other downstream mediators of cell surface syntaxin-4.
Next, we comprehensively analyzed factors downstream of cell surface syntaxin-4 that could mediate morphogenic and/or differentiation signals. Among the genes whose transcripts were up-or downregulated in ES-STstx4 cells, we selected 157 genes with LogFC absolute values ≥ 2, and q-values < 0.01 (Fig. 3a, highlighted in red); this subset contained 19 genes whose expression was also changed in parental ES cells upon Dox treatment. Among thus selected 138 genes, 119 genes could be classified into several categories, using the PANTHER-GO-Slim system ( Fig. 3b and Supplementary Table S1). In addition to the dramatic upregulation of p-cadherin, we detected marked downregulation of all members of the zscan4 family (Fig. 3a,c), transient expression of which reportedly plays key roles in genomic stability, which is important for restoring ES cell pluripotency 21 . Upon siRNA-mediated silencing of the Zscan4 genes, p-cadherin expression was decreased in parental ES cells, albeit to a lesser extent than upon overexpression of syntaxin-4 ( Fig. 2d), suggesting that these genes partly mediate the syntaxin-4 signals (Fig. 3d). These Zscan4 genes are known to be regulated by the PI3K/Akt pathway 23 ; we found a significant reduction in Akt phosphorylation upon stimulation with syntaxin-4 ( Fig. 3e). Notably, LY294002, an inhibitor of the PI3K/ Akt pathway, suppressed the expression of zscan4 transcription and promoted that of p-cadherin in parental ES cells (Fig. 3f). However, this reagent showed significant toxicity and had little effect on the expression of brachyury (Fig. 3f). These results suggested that syntaxin-4 extruded at the cell surface affects the PI3K/Akt signaling cascade to elicit cell responses, in which Zscan4 genes play an important role, even in the presence of LIF.
Inhibition of GSK3β/MEK1/2 signals cannot completely abrogate functions of forcibly expressed cell surface syntaxin-4. Similar to the antagonistic peptide of syntaxin-4, we found that 2i apparently blocked the morphological changes induced by forcible expression of cell surface syntaxin-4 ( Fig. 4a). However, syntaxin-4-induced changes in p-cadherin, zscan4, and brachyury expression were not inhibited (Fig. 4b). These results suggested that whereas inhibitors of GSK3β /MEK1/2 blocked extracellular extrusion of endogenous syntaxin-4 ( Fig. 1b), they were not able to abrogate differentiation initiation triggered by forcibly expressed cell surface syntaxin-4.

Extracellular syntaxin-4 leads to an E-to P-cadherin switch and morphological changes in F9 cells.
To define the role of extruded syntaxin-4 in ES cell behavior, we used the simpler, stable embryonic carcinoma (EC) F9 stem cell system, in which ES cell-specific regulators of differentiation, including a set of Zscan4 proteins, are undetectable ( Supplementary Fig. S3). The F9 cell line expresses E-cadherin and forms compact colonies similar to ES cells, though these cells have lost their potential to differentiate into mesodermal cell lineages. F9 cells possessing the syntaxin-4 transgene (F9-STstx4) showed morphological responses similar to those of ES cells upon expression of cell surface syntaxin-4. Specifically, they lost cell-cell adhesiveness and showed a dramatically changed cell morphology, with active filopodial protrusions after a few days (Fig. 5a,b). However, significant changes in the expression of various differentiation markers, including those for canonical EMT (snail, slug, and zeb1), endodermal differentiation (gata4 and afp), and fibroblastic cell types (vimentin, cofilin, α-smooth muscle actin and myosin heavy chain) were not evident in this short period (Fig. 5c). We found that the E-to P-cadherin switch, accompanied by inactivation of Akt signaling, was induced by cell surface expression of syntaxin-4 in F9 cells, as was the case in ES cells (Fig. 5d,e). E-to P-cadherin switch as a key element to mediate signals from syntaxin-4. To examine the association between the E-to P-cadherin switch and morphological changes, parental F9 cells were transfected with a P-cadherin transgene and/or treated with an anti-E-cadherin blocking antibody ECCD1. We found that cells over-expressing P-cadherin exhibited active pseudopodium formation, similar to that observed in cells stimulated with extracellular syntaxin-4, which was substantially blocked by anti-P-cadherin antibody PCD1, confirming the causal role of P-cadherin in the morphological alternation (Fig. 6a). On the other hand, cellcell adhesion remained almost intact in cells with and without forcible expression of P-cadherin (Fig. 6a). In contrast, treatment with ECCD1 led to cellular dissociation without morphological changes (Fig. 6b), as reported previously 35 . We found that simultaneous over-expression of P-cadherin and functional inhibition of E-cadherin resulted in a phenotype quite similar to that observed with cell surface syntaxin-4 expression, revealing that the E-to P-cadherin switch mediates morphological changes associated with extracellular syntaxin-4 ( Fig. 6c). However, we could not exclude the possibility that the decline in E-cadherin expression was merely due to syntaxin-4-induced changes to the cellular context since the expression of E-cadherin was apparently altered solely by modification of the substrate adhesive property of F9 cells; cells adherent to the substrate (using poly-l-lysine coat) showed dramatically decreased E-cadherin expression as compared to floating cells on petri-dishes (Fig. 6d). We could not assess the impact on differentiation in F9 cells as they do not readily differentiate into mesodermal lineages 36 . In addition, these cells succumb to apoptosis after long-term treatment with Dox. However, with sustained expression of cell surface syntaxin-4 using another expression construct 27 , F9 cells upregulated gata4, which is a transcription factor that acts as an endodermal differentiation cue in this cell type (Fig. 6e). In these cells, the levels of cofilin, a molecular regulator for actin dynamics 37 , and vimentin, encoding an intermediate filament in fibroblasts 38 , were also increased (Fig. 6e), suggesting that long-term signals propagated by extracellular syntaxin-4 affect differentiation even in non-mesodermal lineages.

Extracellular syntaxin-4 induces mesodermal differentiation in P19CL6 cells.
Finally, we tested the effects of cell surface expression of syntaxin-4 in another EC cell line, P19CL6, which retains the intrinsic potential to differentiate into mesodermal lineages 39 (Fig. 7a). Upon induction of syntaxin-4, P19CL6 cells elicited similar morphogenetic responses and displayed a flattened fibroblastic morphology (Fig. 7b). Although they do not express detectable E-cadherin, cell surface syntaxin-4 led to the marked upregulation of P-cadherin also in this cell type (Fig. 7c). Whereas the mesodermal markers α-smooth muscle actin and brachyury were upregulated, the neuronal marker tuj1 and cardiac marker gata4 were unchanged (Fig. 7d,e). When P-cadherin was forcibly expressed, parental P19CL6 cells exhibited a morphology similar to that seen with cell surface syntaxin-4, whereas differentiation markers remained unchanged (Fig. 7f). It is noteworthy that exogenous P-cadherin did not accumulate at the cell-cell contact sites in both F9 and P19CL6 cells (Figs 6a and 7f).
Taken together, syntaxin-4, which is spontaneously extruded at the cell surface in response to the removal of stemness reagents 2i might play a critical role in the abrogation of ground-state pluripotency in ES cells, through inactivation of the PI3K/Akt signaling, induction of an E-to P-cadherin switch, and downregulation of several stemness factors including the Zscas4 family (Fig. 8).

Discussion
Here, we provide evidence that syntaxin-4 spontaneously translocates across the cell membrane to execute its latent extracellular functions, including morphological changes and the onset of differentiation in ES cells; this leads to instability in ES cell stemness. As this effect does not require additional differentiation stimuli and is achieved even in the presence of LIF, the molecular mechanisms of syntaxin-4 membrane translocation are unclear at present. However, given that 2i inhibited cell surface expression of syntaxin-4, upstream extracellular elements of GSK3β or MEK1/2 might be involved in the extracellular extrusion of syntaxin-4. Alternatively,   recent studies have revealed that epimorphin (syntaxin-2), another plasmalemmal syntaxin, is extruded as a component of synaptotagmin and annexin II protein complexes in response to apoptotic signals 40 . Thus, one could assume that the extracellular presentation of syntaxin-4 is a result of the cell competition in ES cell colonies, which spontaneously induces local apoptosis to eliminate unfit cells 41 . As 2i could prevent spontaneous differentiation in ES cells but not that by forcible expression of cell surface syntaxin-4, it is conceivable that one of the key roles of 2i might be to prevent membrane translocation of syntaxin-4.
Whereas ES cells treated with 2i succumbed to syntaxin-4-triggered differentiation, morphological changes induced by the syntaxin-4 signal were blocked. Considering that syntaxin-4 led to inactivation of Akt, which negatively regulates GSK3β , a potent regulator of E-cadherin expression 42,43 , GSK3β might mediate syntaxin-4 signaling to induce the morphogenic effect but not differentiation. GSK3β has been reported to play key roles in the perturbation of clonogenic activity in ES cells 44 ; however, activation of downstream elements equivalently depend on the phosphorylation and subcellular localization of this protein 16 . Thus, it is not surprising that the syntaxin-4 signaling pathway inactivates PI3K/Akt signaling, leading to the activation of GSK3β for morphological changes, but not differentiation initiation.
An E-to P-cadherin switch was identified as a key element to mediate the morphologic function of syntaxin-4. Given that the expression of E-cadherin is strictly regulated by PI3K/Akt in ES cells 43 , the downregulation of E-cadherin could be attributed to the syntaxin-4-induced dephosphorylation of PI3K/Akt. It is noteworthy that the inactivation of PI3K/Akt signaling by cell surface syntaxin-4 occurred even in the presence of LIF, suggesting that the effect of syntaxin-4 might be dominant over LIF-triggered PI3K/Akt activation. Moreover, as GSK3β is a component of Wnt signaling, its subsequent activation may also affect the fate of β -catenin 45 . Thus, the elimination of E-cadherin might be accompanied by changes in subcellular localization or stability of β -catenin, a potent transcriptional activator for P-cadherin 46 . Transcriptome analyses did not detect the upregulation of other transcription factors that regulate P-cadherin such as p63 and C/EBPβ 47,48 . All of the classical cadherins, including E-, P-, and N-cadherin, play central roles in intercellular adhesion systems 49,50 ; however, they exert different functions such as contact inhibition of locomotion via affecting the spatial distribution and activation of small GTPases 51,52 . A recent report demonstrated the regulation of cell migration by Cdc42, which is highly dependent on P-cadherin, but not E-cadherin 53 . In addition, P-cadherin fails to localize to the cell-cell contact sites and causes invasive cell behaviors in several aggressive cell types 54 . This study detected P-cadherin localization to the cell membrane rather than to cell-cell contact sites, suggesting that the P-cadherin-dependent morphological changes involved a function other than that in intercellular adhesion. Alternatively, certain cell surface molecules that can associate with P-cadherin but not E-cadherin-for example, integrin α 6β 4 18 -could propagate signals for morphological changes.
In addition to morphological changes, the syntaxin-4-triggered inactivation of PI3K/Akt signaling is a plausible differentiation cue, as this signaling pathway reportedly mediates the upregulation of genes involved in self-renewal 43 , retention of transcription factors for maintenance of stemness 23 , and inhibition of Wnt/MAPKinduced differentiation signals in ES cells 1,55 . However, ES cells failed to upregulate the differentiation marker brachyury in response to an Akt inhibitor, despite a significant downregulation of the stemness factor zscan4. Thus, whereas the activation of PI3K/Akt might be sufficient for maintenance of ES cell pluripotency 56 , its specific inactivation might be necessary but insufficient for activation of the differentiation program. Presumably, additional signaling pathways that could cooperate with PI3K/Akt signals to mediate differentiation in ES cells are propagated by cell surface syntaxin-4.
An important and contentious point is the causal link between morphological changes and differentiation in pluripotent ES cells. Although morphological changes and onset of differentiation are usually induced concurrently, an engineered substrate using recombinant E-cadherin caused a flattened morphology without differentiation 57 , and inhibition of GSK3β /MEK1/2 (2i-treatment) with forcible cell surface expression of syntaxin-4 resulted in the onset of differentiation without morphological changes. On the other hand, syntaxin-4-triggered inactivation of PI3K/Akt signaling subsequently downregulated zscan4 during differentiation and severely affected cadherin-mediated cell adhesion systems involved in cell morphology. Accordingly, one might assume that these phenomena could be studied independently if signals downstream of PI3k/Akt are of interest. However, cell surface syntaxin-4 rapidly elicited mesodermal differentiation in ES and P19CL6 cells, whereas its sustained stimulation ultimately led to endodermal differentiation in F9 cells that are unable to differentiate into mesodermal lineages. This indicates that syntaxin-4-triggered changes in cell shape could activate a differentiation ECCD1 causes the dissociation of cells without active pseudopodial protrusions. Bars, 20 μ m. (c), F9-P-cad cells at the periphery of the colony extend filopodia in response to P-cadherin expression (ON), and the number of cells with filopodia increases when cells are simultaneously dissociated by ECCD1. Left panels, phasecontrast images. Bar, 20 μ m. Right panel, quantification of cells with filopodia. N = 4, **p < 0.01, ***p < 0.001. (d), Substrate-adhesion affects E-cadherin expression at the protein level. Parental F9 cells adhering to and spreading on poly-l-lysine (PL-Lysine) show significantly decreased expression of E-cadherin (E-cad) protein, but not mRNA, compared to that of similar cells found in clusters on the Petri dish (petri). N = 3, ***p < 0.001. For middle panels, cropped areas from the original blots ( Supplementary Fig. S)   program for non-mesodermal lineages, albeit to a lesser extent. Thus, it is important to account for the fact that morphological changes and differentiation might have cause-effect relationships.
Although this study underlined the pernicious effect of syntaxin-4 signaling on the maintenance of ES cell stemness, the biological relevance to embryogenesis in vivo remains unknown. Disruption of syntaxin-4 reportedly resulted in early embryonic lethality; however, this could be due to the functional ablation of cytoplasmic syntaxin-4 given its important role as a t-SNARE protein in intravesicular fusion 58 . We surmise that signals from cell surface syntaxin-4 might exert their functions at early developmental stages in embryogenesis, since EMT-like cell behavior accompanied by a cadherin switch has been proposed to be essential for gastrulation or neural crest delamination 52,59,60 . Although in preliminary experiments we detected extracellular syntaxin-4 in early-stage mouse embryos, its localization was not restricted to cells of the primitive streak or other specific regions. Moreover, attempts were made to generate transgenic mice expressing signal peptide-and T7-tagged syntaxin-4; however, no pups expressing the transgene product have been obtained to date. Whereas one pup from 62 manipulated-fertilized eggs was found to possess the transgene, T7-tagged syntaxin4 was not expressed, implying that extracellular syntaxin-4 severely disturbed normal embryogenesis. Careful analyses of successive developmental processes in fertilized eggs are necessary to clarify this issue.  (α MEM; Wako) with 5% FCS. To induce transgene expression, ES-STstx4, F9-STstx4, F9-Pcadherin, or P19-STstx4 was treated with 5 μ g/ml Dox (Sigma-Aldrich) for two or three days. In some cultures, an inhibitor of PI3K (LY294002; Wako) was dissolved in DMSO and added at a concentration of 1.25 or 2.5 μ M.

Expression constructs and transfection.
To construct the plasmid for tetracycline (Tet)/Dox-inducible expression of syntaxin-4 or P-cadherin, cDNA encoding syntaxin-4 with an N-terminal fusion of the IL2 signal peptide and T7 peptide tag 61 or cDNA encoding P-cadherin (a generous gift from Dr. Takeich) was subcloned into the PiggyBac-TET transposon plasmid 34,62,63 . The plasmid harbored cDNA of the transgene and the neomycin-resistance gene connected by the IRES sequence. To generate cells with Dox/Tet-dependent expression of the transgene, one of these plasmids was transfected into cells, together with PB-CArtTA Adv and pCAG-Pbase 33 , using Lipofectamine 2000 (Life Technologies). To avoid a possible clonal artifact, a mixture of the clones that express the transgene product in response to Dox treatment (5 μ g/ml) was used for experiments. These transfected cells included ES-STstx4 (ES cells with signal peptide/T7 tag-connected syntaxin-4), F9-STstx4 (F9 cells with signal peptide/T7 tag-connected syntaxin-4), F9-P-cad (F9 cells with P-cadherin), and P19-STstx4 (P19CL6 cells with signal peptide/T7 tag-connected syntaxin-4). To assess the ratio of cells that extrude syntaxin-4 at the cell surface, ES cells were transfected with an expression plasmid for T7-tagged syntaxin-4 (without a signal peptide, T7stx4) 61 and stained with anti-T7 antibody before and after cell permeabilization.
Antibodies and immunocytochemistry. Primary antibodies against T7-tag (Novagen), α SMA (Sigma-Aldrich), β -actin, (Sigma-Aldrich), Akt, and Akt phosphorylated at (Ser473) (CST Japan) were used in this study. Anti-E-cadherin monoclonal antibodies (ECCD1 for functional blocking and ECCD2 for immunodetection) and anti-P-cadherin monoclonal antibody PCD1 were generous gifts from Dr. Takeichi. Affinity-purified polyclonal antibodies against syntaxin-4 were prepared in a previous study 61 . HRP-, Alexa 488-, and Cy3-labeled secondary antibodies were purchased from Sigma-Aldrich, Invitrogen and GE healthcare. To detect cell surface expression of endogenous syntaxin-4, non-permeabilized cells were stained simultaneously for syntaxin-4 and the negative control β -actin, as described previously 61 . In brief, living cells on chamber slides were incubated with a mixture of affinity-purified rabbit anti-syntaxin-4 antibody and mouse anti-β -actin monoclonal antibody. After washing twice with Tris-buffered saline (TBS) followed by fixation with methanol (having an initial temperature of − 20 °C) for 10 min, the cells were incubated with a mixture of Alexa488-labeled anti-rabbit IgG and Cy3-labeled anti-mouse IgG antibodies. For conventional immunocytochemistry, the cells were permeabilized with − 20 °C methanol for 15 min, and stained with primary and Alexa488-or Cy3-labeled secondary antibodies, as described above. For some samples, cells were permeabilized with 0.1% triton X-100 instead of methanol, and F-actin was stained directly with Alexa 488-phalloidin (Invitrogen). Nuclei were counterstained with DAPI (Sigma). Stained cells were photographed with a confocal microscope system A1 (Nikon) or a fluorescence microscope AXIOSHOP (Zeiss) with a CCD camera VB-7010 (Keyence, Japan). To detect cell surface expression of transiently introduced T7-tagged syntaxin-4 (T7stx4) on a blot, cell surface proteins were labeled with NHS-biotin and lysed in buffer containing 1% Triton X-100 and 1% NP-40. The cell lysate was immunoprecipitated with anti-syntaxin-4 antibodies, and analyzed with HRP-labeled anti-T7 antibody (Novagen) or streptavidin (Sigma-Aldrich), as described previously 40 . Concurrently, biotin-labeled extracellular proteins in the cell lysate were retrieved with neutrAvidin-beads (Thermo Scientific) and analyzed with antibodies against syntaxin-4 or cytoplasmic β -actin.

Reverse transcription quantitative RT-PCR (qRT-PCR).
Total RNA was extracted from ES-STstx4, F9-STstx4, F9-P-cad, and P19-STstx4 cells cultured with or without doxycycline was extracted using an RNeasy Mini kit (Qiagen) and reverse-transcribed with an RNA-PCR kit (Takara). To assess transgene expression, cDNA was amplified using Quick tag (Toyobo) and primer pairs for T7-syntaxin-4 (5′ -GGG GCG GCC GCA TGG CTA GCA TGA CTG GTG GAC-3′ and 5′ -TTT TAG CTG CGC CCG GAC C-3′ ) and gapdh (5′ -GGATTTGGCCGTATTGG-3′ and 5′ -TCATGGATGACCTTGGC-3′ ). qRT-PCR was then carried out using FastStart Essential DNA Green Master on a LightCycler Nano system (Roche) according to the manufacturer's protocol (PCR; 45 cycles). The primer pairs used in this study are listed in Supplementary Table S2. As the expression of both gapdh and β -actin was not changed in cells with and without exogenous syntaxin-4 ( Supplementary Fig. S4), the mRNA expression of differentiation markers was normalized to that of β -actin. qRT-PCR was conducted in triplicate.

Recombinant proteins.
Recombinant forms of syntaxin-4 fragment F1 (Met1~Glu110), F2 (Ala111~Arg197), and GFP were generated as 6 X histidine-tagged forms and purified as described previously 32 . These proteins were dialyzed against PBS, filtrated, and added to cultures at a concentration of 50 μ g/ml for three days.

Analysis of cell behaviors upon cell surface expression of syntaxin-4. The morphology of cells cultured in
the presence and absence of Dox for two (F9-STstx4 and F9-P-cad) or three (ES-STstx4 and P19-STstx4) days was analyzed. The number of cells extending more than three pseudopodia (10 μ m or longer) was counted in ES-STstx4, F9-STstx4 or F9-P-cad cells. The area occupied by a cell was calculated in P19-STstx4 cells using the ImageJ software 64 .

RNA-seq analysis.
To identify differentially expressed (DE) genes in ES cells upon stimulation with extracellular syntaxin-4, comparative transcriptome analyses were performed at the Phyloinformatics Unit of RIKEN CLST. To exclude genes affected by Dox itself, cDNAs from ES-STstx4 and parent ES cells, with and without doxycycline for three days, were analyzed. The transcriptome analysis identified 138 DE genes with an absolute LogFC value ≥ 2 and a q-value < 0.01. RNA-seq data sets are available at the Gene Expression Omnibus under accession number PRJDB4923.
Scientific RepoRts | 7:39868 | DOI: 10.1038/srep39868 RNA interference of Zscan4 genes. To suppress the expression of Zscan4 genes in ES cells, oligonucleotide siRNAs for which clear knockdown effects have been demonstrated 22 were obtained from Nippon Gene (Tokyo, Japan) ( Supplementary Fig. S5). The mixture of the siRNAs (20 nM in total) was transfected into parental ES cells in 12-well plate well using Lipofectamine ™ RNAiMAX Transfection Reagent (Invitrogen). The expression of zscan4 and p-cadherin was analyzed on day 3 by qRT-PCR. As a control, universal negative control NEGD/ NEGAS labeled with Hilyte 488 (20 nM) (Nippon Gene) was used; the transfection efficiency was more than 70% as judged by the fluorescence.

Statistical analyses.
Results are expressed as the mean ± SD of three independent experiments. Data were analyzed by a Student's t-test or Mann-Whitney U-test, and a p-value of < 0.05 was considered statistically significant.