Cdh2 stabilizes FGFR1 and contributes to primed-state pluripotency in mouse epiblast stem cells

The cell adhesion molecule Cadherin 2 (Cdh2) plays important roles in somatic cell adhesion, proliferation and migration. Cdh2 is also highly expressed in mouse epiblast stem cells (mEpiSCs), but its function in these cells is unknown. To understand the function of Cdh2 in mEpiSCs, we compared the expression of pluripotency-related genes in mEpiSCs and mouse embryonic stem cells (mESCs) after either Cdh2 knockdown or Cdh2 over-expression. Introduction of specific siRNA against Cdh2 led to attenuation of pluripotency-related genes. Pluripotent gene expression was not recovered by over-expression of Cdh1 following Cdh2 knockdown. Western blot analysis and co-immunoprecipitation assays revealed that Cdh2 stabilizes FGFR1 in mEpiSCs. Furthermore, stable transfection of mESCs with Cdh2 cDNA followed by FGF2 supplementation accelerated cell differentiation. Thus, Cdh2 contributes to the establishment and maintenance of FGF signaling-dependent self-renewal in mEpiSCs through stabilization of FGFR1.

Scientific RepoRts | 5:14722 | DOi: 10.1038/srep14722 of mESCs 17 . Cdh1 is a single transmembrane glycoprotein with five extracellular domains that participate in calcium-dependent homophilic cell-cell adhesion 18 . The intracellular domain of Cdh1 interacts with the actin cytoskeletal through catenin proteins 19 . While Cdh1 expression is robust in mature epithelial cells, it is also appears during the compaction phase of mouse early embryonic development in morula stage embryos 20 . Interestingly, recent studies have shown that Cdh1 stabilizes STAT3-mediated signaling by binding to LIF/GP130 and subsequently activating pluripotency-related genes such as Nanog in mESCs 21 . These facts suggest that cadherins are involved in stem cell development.
We previously reported that Cadherin2 (N-cadherin, neuronal-cadherin; Cdh2) is the predominant cadherin expressed by mEpiSCs 22 . We also observed that the conversion from mESCs to mEpiSCs coincides with cadherin-switching from Cdh1 to Cdh2. However, the function of Cdh2 in mEpiSCs and the significance of cadherin-switching are still unknown. In this study, we investigate the expression status, function and significance of Cdh2 expression in mEpiSCs.
Cdh2 is important for maintaining mEpiSCs in an undifferentiated state. To determine the relationship between Cdh2 and mEpiSCs pluripotency, we either disrupted Cdh2 function using ADH-1 (also known as Exherin), which is a selective inhibitor for Cdh2, or suppressed Cdh2 expression using specific siRNA to Cdh2 (siCdh2). ADH-1 and siCdh2 had no effect on colony size and viability (data not shown), but both treatments induced differentiation in mEpiSCs, as indicated by a decrease in SSEA-1, which is a pluripotent state-specific glycoprotein in mouse cells ( Fig. 2A). We next analyzed the expression levels of other pluripotency-related genes in the ADH-1-and siCdh2-treated cells. A knockdown efficiency of ~80% occurred for siCdh2 compared with scrambled siRNA-transfected cells, as determined by qRT-PCR. The ADH-1 and the siCdh2 treatments both suppressed the expression of the Pou5f1, Sox2 and cMyc genes at both the RNA and protein levels. Although Nanog mRNA expression did not decrease in siCdh2-treated cells, Nanog protein levels were lower in both ADH-1-and siCdh2-treated cells compared to non-treated cells. Expression of Eomes, which is an important transcription factor for neuronal and trophoblast differentiation, was upregulated by the ADH-1 and siCdh2 treatments (Fig. 2B). To elucidate the mechanism by which pluripotency-related genes are down-regulated by inhibition of Cdh2, we observed phosphorylation of Erk and Akt, both of which are important molecular mediators of pluripotency in primed-state stem cells [23][24][25] . WB analysis revealed that disruption of Cdh2 inhibited Erk and Akt phosphorylation (Fig. 2C,D).
Overexpression of Cdh1 does not recover Cdh2 function. To identify the specific function of Cdh2 and its significance in determining primed state pluripotency, we examined if Cdh1 could compensate for a loss of Cdh2 function in mEpiSCs. For this experiment, we produced a mEpiSC line (Cdh1-EpiSC) that stably expressed Cdh1 and maintained pluripotency in standard mEpiSC medium. Cdh1 localized to the surface of Cdh1-EpiSCs (Fig. 3A). Interestingly, even though Cdh1-EpiSCs strongly expressed cell surface Cdh1, some cells lost SSEA-1 expression after treatment with either ADH-1 or siCdh2 (Fig. 3B). To investigate this phenomenon in detail, we performed qRT-PCR and WB assays for pluripotency-related genes. ADH-1 suppressed the expression of Nanog, Sox2 and cMyc, although Cdh1 expression rescued Pou5f1. Furthermore, consistent with the results from wild-type mEpiSCs, Eomes was upregulated by ADH-1 treatment. Treatment with siCdh2 had similar effects on Sox2, cMyc and Eomes expression (Fig. 3C). In addition, we found that siCdh2 treatment suppressed Erk and Akt phosphorylation in Cdh1-EpiSCs (Fig. 3D,E). These results suggest that Cdh2 is important for Erk and Akt phosphorylation and that suppression of Cdh2 decreases the expression of pluripotency-related genes.
Cdh2 stabilizes FGFR1. Since Cdh2 disruption suppresses Akt and Erk phosphorylation, we focused our attention on the FGF signaling cascade, a pathway in which both Akt and Erk act as mediators. Since it is most likely for other molecules on the cell surface to interact with Cdh2, we hypothesized that Cdh2 may regulate FGF receptors (FGFRs). To test this hypothesis, we observed expression levels of three FGFRs in mEpiSCs. FGFR1 showed the highest expression in both mESCs and mEpiSCs. FGFR2 and FGFR3 expression levels were significantly lower than FGFR1 expression levels in mEpiSCs. The expression levels of FGFR2 and FGFR3 were 0.1-fold that of the FGFR1 level and were almost undetectable (Fig. 4A). Next, we examined the significance of FGFR1 in mEpiSCs pluripotency by introducing specific siRNA for Fgfr1 (siFgfr1). According to qRT-PCR, the siFgfr1 achieved a knockdown efficiency of 90% compared with scrambled siRNA. Treatment with siFgfr1 resulted in reduced expression of pluripotency-related genes (Fig. 4B). We next performed co-immunoprecipitation to confirm the interaction of Cdh2 with FGFR1 and to demonstrate that Cdh2 bound to FGFR1 in mEpiSCs (Fig. 4C). Furthermore, we determined if Cdh2 stabilized FGFR1 during FGF2 stimulation by analyzing FGFR1 degradation efficiency in cells treated with cycloheximide (CHX), which is an inhibitor of de novo protein synthesis. FGFR1 protein levels decreased in a time-dependent manner following CHX treatment, while Cdh2 expression did not change. The siCdh2-treatment accelerated FGFR1 protein degradation in these cells (Fig. 4D,E).
Cdh2 supports FGF2-mediated differentiation in ESCs. FGF-Erk signaling induces differentiation of mESCs 26 . Thus, we examined if the overexpression of Cdh2 can accelerate FGF2-induced differentiation in mESCs. FGF2 is an important mitogenic cytokine for various types of cells including mEpiSCs 27,28 . We observed cell proliferation of wild-type and Cdh2-overexpressing ESCs under FGF2 stimulation 24 h, 48 h and 72 h after changing the medium. In LIF-supplemented medium, there was no difference between wild-type mESCs and Cdh2-ESCs (Fig. 7A). In contrast, there was a significant increase in cell number at each time-point in Cdh2-ESC compared with wild-type ESCs when they were cultured in FGF2-supplemented medium (Fig. 7A). This suggests that Cdh2 enhances the response of mESCs to FGF2. Next, we observed the efficiency of FGF2-induced differentiation using mESCs transfected with an Oct4 deltaPE-GFP reported plasmid, which contains a distal enhancer and the Oct4 promoter for naïve-state-specific GFP expression. In the LIF-supplemented medium, GFP-positive mESCs were present at the same ratio in wild-type ESCs as in Cdh2-ESCs (Fig. 7B). However, in the FGF2-supplemented culture medium, the number of GFP-positive cells significantly decreased in Cdh2-ESCs (Fig. 7B,C). These results confirm that Cdh2 contributes to FGF2-mediated differentiation of mESCs.

Discussion
Cadherins are expressed in tissue-and cell-lineage-specific patterns that give rise to specific cell and tissue functions [29][30][31][32][33][34] . In this study, we found that mESCs express Cdh1 and Cdh3, while mEpiSCs express Cdh1 and Cdh2. The expression of Cdh1 mRNA was not consistent with its protein expression, suggesting posttranscriptional regulation of Cdh1, as reported previously 35,36 . Consistent with a previous study 37 , we observed that EpiSCs predominantly expressed Cdh2, suggesting a functional role of Cdh2 in regulating the pluripotency of mEpiSCs. Cdh2 knockdown decreased phosphorylated Erk and Akt, both of which are important for maintenance of pluripotency in primed-state stem cells [23][24][25] , and modified the expression of pluripotency-related genes. Interestingly, Cdh1 expression could not compensate for lost Cdh2 function, which prompted us to examine the molecular function of Cdh2 in pluripotency. Others have suggested that cadherins interact with various proteins, with partner protein binding varying by cadherin family member 38,39 . We focused our analysis on the phosphorylation status of Erk and Akt in siCdh2-treated mEpiSCs. FGF signaling regulates Erk and Akt phosphorylation in pluripotent stem cells 40 . We confirmed that suppression of FGFR1 attenuated pluripotency-related gene expression in mEpiSCs. This led us to hypothesize that Cdh2 contributes to a pluripotency-related network through its relationship with FGFR1, which we tested by observing protein-protein interactions between Cdh2 and FGFR1 in a co-IP assay. Suyama et al. reported that extracellular domain 4 of Cdh2 binds to Ig domains 1 and 2 of FGFR1, which inhibits internalization of FGFR1 and prevents FGFR1 degradation 41 . Hazan et al. reported that stabilizing Cdh2 on the cell membrane causes continuous tyrosine phosphorylation of FGFR1 by FGF2, with downstream activation of MAPK/Erk 42 . Consistent with these studies, we observed that Cdh2 stabilizes FGFR1 protein expression in CHX-treated mEpiSCs. Importantly, the observation that Cdh2 contributes to stabilization of FGFR1 and downstream Erk and Akt phosphorylation was reproducible in Cdh2-expressing mESCs. In Cdh2-ESCs, cMyc expression increased at both the mRNA and protein levels. This reasonable observation was supportive of our hypothesis, since it is well documented that cMyc expression is positively regulated by FGF2-Erk signaling 25,43 . Forced expression of Cdh2 enhanced FGF2-dependent cell proliferation in mESCs that were originally not responsive to FGF2. Furthermore, Cdh2-overexpression promoted FGF2-induced differentiation in mESCs. It has been reported that FGF signaling is detrimental to naïve-state pluripotency. However, our results suggest that Cdh2 is important for signal transduction of FGF2. Interestingly, Debiais et al. observed that FGF2 enhances the expression of Cdh2 in human calvaria osteoblasts through activation of protein kinase C and the Src-kinase pathway 44 . This suggests that there is a positive feedback-loop involving FGF and Cdh2.
On the other hand, the fact that our results indicate that Cdh2 contributes to pluripotency maintenance in mEpiSCs raises an important question about the similarity of EpiSCs to other primed-state stem cell types, such as human ESCs (hESCs) and rabbit ESCs (rbESCs). Until now, researchers have described primed-state pluripotent stem cells, including hESCs and rbESCs, according to their high Cdh1 expression to indicate their undifferentiated status, rather than by their Cdh2 expression level 45,46 . Furthermore, hESCs can lose Cdh1 and alternatively obtain Cdh2 expression during early differentiation processes 45 . This apparent contradiction may be accounted for if Cdh1 is not functional in the hESCs and rbESCs. Using mESCs, del Valle et al. demonstrated that Cdh1 can interact with β catenin and STAT3, and promote the LIF-dependent signaling cascades that are vital for naïve-pluripotency 21 . However, the LIF-dependent pathway does not function in hESCs and rbESCs. Thus, only low levels of Cdh2 may exist in hESCs and rbESCs, because it is dispensable for these cells. As we observed, mEpiSCs lacking Cdh2 maintain pluripotency-related gene expression, although the expression levels are attenuated compared with controls. Since Cdh2 supports FGFR1 stabilization, it may not be an essential co-factor/protective agent against the protein-degradation machinery. The importance of Cdh2 is thus likely to depend on the cellular status, for example, the type and level of cadherin that is expressed. In support of the importance of cellular differences between hESCs and mEpiSCs, Brons et al. reported significantly lower Rex1 expression in mEpiSCs compared to hESCs 15 . Rex1 is an important pluripotency-related gene in mESCs and hESCs 47 . There is also evidence that the primitive endodermal differentiation marker Gata6 48,49 is not expressed in mESCs and hESCs, but is highly expressed in mEpiSCs 14,15 . Therefore, it is possible that mEpiSCs display a more differentiated state than other primed-state ESCs. These differences may result in altered cadherin expression patterns.
In this study, we only focused on Cdh2 function and its interaction with FGFR1. However, we cannot ignore other possible functions of cadherins. Recent research has elucidated direct interactions between cadherins and transcription factors that influence gene expression. For example, the intercellular-domain of Cdh2 from the presenilin1/gamma secretase complex binds to a transcriptional co-activator CREB-binding protein (CBP) and induces proteasome-mediated CBP degradation 50 . These reactions can directly suppress CBP/CREB-mediated transcription, leading to Cdh1 modulation of cell growth and epithelial-mesenchymal transition through interaction of β -catenin transcriptional activity in cancer cells and epithelial cells 51,52 . Furthermore, Cdh13 suppresses activation of c-Jun/JNK and induces G2/M cell cycle arrest in hepatocellular carcinoma cells 53 . Recently, Sancisi et al. reported that Cdh6 interacts with Runx2 in thyroid tumor cells 54 . These results suggest functional diversity in the cadherin proteins and the possibility that cadherins contribute to significant changes in cell fate over long periods. Further research on cadherin expression patterns in various stem cells and a comparison of cadherin expression levels according to differentiation status will be essential for a complete understanding of stem cell pluripotency.

Materials and Methods
Animal use and care. This project and all related experimental protocols were approved by the Kinki University Animal Experimental Ethics Committee for Laboratory Experimentation (project number: KAME-26-043) and were carried out in accordance with the approved guidelines. mESC and mEpiSC Cell Culture. Mouse wild-type ESCs (C57BL/6J), Oct4 deltaPE-GFP expressing ESCs and wild-type EpiSCs (C57BL/6J) were produced as described previously 22  ESGRO (Millipore, MA, USA). The mEpiSCs were cultured in mEpiSC medium, which consisted of Knockout DMEM with 1× MEM nonessential amino acids, 0.1 mM 2-mercaptoethanol, 200 mM L-glutamine and 20% Knockout Serum Replacement, supplemented with 5 μ g/ml FGF2 (Wako Pure Chemical Industries, Ltd., Osaka, Japan). To passage the cells, 10 μ M Y-27632 (Wako) was added to the medium. ADH-1 (Exherin ™ ; MedKoo Biosciences, Inc., NC, USA), which is a competitive inhibitor of N-cadherin 55 , was dissolved in PBS (− ) and added to the medium at a concentration of 0.2 mg/ml.

Production of Stable Cdh1-Expressing mEpiSCs and Cdh2-Expressing mESCs.
To create the pCAG-Ncadherin-IRES-Puro plasmid, the Ncadherin ORF sequences from the Ncadherin-GFP plasmid, which was purchased from Addgene (number #18870), were cloned into the EcoRI site of the pCAG-IRES-Puro plasmid, which was kindly provided by Dr. Hirofumi Suemori. Quantitative RT-PCR (qRT-PCR) analysis. Total RNA was collected using a TRI Reagent ® (Molecular Research Center, Inc., OH, USA) and a PrimeScript ® RT Master Mix Kit (TAKARA Bio Inc., Shiga, Japan). Quantitative RT-PCR for total cDNA was performed using Perfect real-time SYBR green II (Takara). PCR amplifications were performed using a Thermal Cycler Dice ® Real Time System Single at 95 °C for 20s followed by 40 cycles of 95 °C for 5s and 60 °C for 30s. To quantify the relative expression of each gene, the Ct (threshold cycle) values were normalized by Lamina (Δ Ct = Ct target − Ct Lamina ) and compared with a calibrator using the "Δ Δ Ct method" (Δ Δ Ct = Δ Ct sample − Δ Ct calibrator ). All values are means ± SD of 3 experiments. Statistical significance was evaluated with JMP software version 10.0.0 (SAS Institute, Cary, NC, USA) using either Student's t-test or the Tukey-Kramer HSD test. The primer sequences are given in Supplementary Table S1.
Western Blot (WB) Analysis. Samples were homogenized in SDS buffer (4% SDS, 125 mM Trisglycine, 10% -mercaptoethanol, 2% bromophenol blue in 30% glycerol) and centrifuged at 10,000 rcf for 10 min at 4 °C to remove debris. Aliquots were subjected to polyacrylamide gel electrophoresis in the presence of SDS (SDS/PAGE) followed by electrotransfer onto a PVDF membrane (Hybond-P; GE Healthcare Japan, Tokyo, Japan). The blotted membranes were blocked overnight with Block Ace (Dainippon Sumitomo Pharma) and then probed with primary antibody overnight at 4 °C. Detection was performed with horseradish peroxidase (HRP)-conjugated secondary antibodies (all antibody were purchased from Santa Cruz Biotechnology) and either the ECL prime Western blotting detection system (GE Healthcare Japan) or Immunostar ® LD (Wako). The lumino-labeled membranes were analyzed using a CCD-based chemiluminescent analyzer (Amersham ™ Imager 600, GE Healthcare Japan). To assess the expression level, relative band intensities were estimated using ImageQuant ™ TL (GE Healthcare Japan).
All values are represented as means ± SD of 3 experiments. Statistical significance was evaluated by Student's t-test using JMP software. Antibodies are descried in Supplementary Table S3.
Co-Immunoprecipitation (Co-IP). Samples were lysed in IP extraction buffer (25 mM Tris-HCl pH7.5, 100 m NaCl, 0.5% Triton X-100) for 30 min on ice. The samples were then centrifuged to collect the supernatant. An anti-FGFR1 antibody (Cell Signaling Technology, Inc., MA, USA) with Dynabeads Protein G (Veritas Corporation., Tokyo, Japan) in 0.02% tween TBS buffer was used for the Co-IP reaction. After Co-IP, proteins were eluted in SDS buffer for WB analysis. The antibodies are described in Supplementary Table S3.
Scientific RepoRts | 5:14722 | DOi: 10.1038/srep14722 Flow cytometry (FACS) analysis. Wild-type-and Cdh2-ESCs were dissociated into single cells by Trypsin-EDTA after 48 hours of treatment, washed twice and diluted in the FACS buffer (2% FBS and 10 mM HEPES in DMEM). The Oct4 deltaPE-GFP positive cell fractions were analyzed by flow cytometry with a FACS Caliber cytometer (Becton Dickinson, Franklin Lakes, NJ). All values are means ± SD of 3 experiments. Statistical significance was evaluated by Student's t-test using JMP software.
Cell counts. A total of 5 × 10 4 wild-type, empty-or Cdh2-ESCs were plated per well of a 24 well plate. After 24, 48 and 72 hour, the cells were trypsinized into single cell suspensions and automatically counted using a TC20 ™ Automated Cell Counter (Bio-Rad Laboratories).

Inhibition of Protein Synthesis by Cycloheximide.
We observed the expression of various cadherin types and the degradation of FGFR1 protein by activation of FGFR1/Mek-Erk following FGF2 stimulation. After FGF2 stimulation and cycloheximide treatment (final concentration is 25 μ m) for 0, 5 min, 30 min, 1 hr, 2 hr, 4 hr and 8 hr, all samples were eluted in 2× SDS sample buffer and analyzed by Western blot using anti-Cdh2, anti-FGFR1 and anti-Actin antibodies (detailed information in Supplementary Table S3).