Epithelial Cell Adhesion Molecule (EpCAM) Complex Proteins Promote Transcription Factor-mediated Pluripotency Reprogramming*

Background: EpCAM is highly expressed in ESCs. However, the role of EpCAM complex proteins in pluripotency reprogramming is still unknown. Results: Overexpression of EpCAM complex proteins significantly repressed the expression of p53 and enhanced reprogramming efficiency in MEFs. Conclusion: EpCAM signaling enhance reprogramming through suppression of the p53-p21 pathway. Significance: EpCAM signaling enhance reprogramming through suppression of the p53-p21 pathway. Epithelial cell adhesion molecule (EpCAM) is a transmembrane glycoprotein that is highly expressed in embryonic stem cells (ESCs) and its role in maintenance of pluripotency has been suggested previously. In epithelial cancer cells, activation of the EpCAM surface-to-nucleus signaling transduction pathway involves a number of membrane proteins. However, their role in somatic cell reprogramming is still unknown. Here we demonstrate that EpCAM and its associated protein, Cldn7, play a critical role in reprogramming. Quantitative RT-PCR analysis of Oct4, Sox2, Klf4, and c-Myc (OSKM) infected mouse embryonic fibroblasts (MEFs) indicated that EpCAM and Cldn7 were up-regulated during reprogramming. Analysis of numbers of alkaline phosphatase- and Nanog-positive clones, and the expression level of pluripotency-related genes demonstrated that inhibition of either EpCAM or Cldn7 expression resulted in impairment in reprogramming efficiency, whereas overexpression of EpCAM, EpCAM plus Cldn7, or EpCAM intercellular domain (EpICD) significantly enhanced reprogramming efficiency in MEFs. Furthermore, overexpression of EpCAM or EpICD significantly repressed the expression of p53 and p21 in the reprogramming MEFs, and both EpCAM and EpICD activated the promoter activity of Oct4. These observations suggest that EpCAM signaling may enhance reprogramming through up-regulation of Oct4 and possible suppression of the p53-p21 pathway. In vitro and in vivo characterization indicated that the EpCAM-reprogrammed iPSCs exhibited similar molecular and functional features to the mouse ESCs. In summary, our studies provide additional insight into the molecular mechanisms of reprogramming and suggest a more effective means of induced pluripotent stem cell generation.

In summary, our studies provide additional insight into the molecular mechanisms of reprogramming and suggest a more effective means of induced pluripotent stem cell generation.
Epithelial cell adhesion molecule (EpCAM) 2 (1), a type I transmembrane glycoprotein protein consisting of two epidermal growth factor-like extracellular domains, one single transmembrane domain, and one small intercellular domain (EpICD), is involved in homotypic cell-cell adhesion (2). EpCAM expression can also affect the cell-cell interactions mediated by classic cadherins (3) that link with the cytoskeleton (4). Recently EpCAM has also been demonstrated (5) to have a critical role in the transmission of proliferative signals into the nucleus via regulated intramembrane proteolysis to relieve the EpCAM intercellular domain (EpICD), which can then associate with FLH2, Lef-1, and ␤-catenin and form a complex capable of activating the transcription of cell proliferation-related genes, such as c-MYC (6), cyclin-encoding genes (6), and other cell-cycle regulators (7). The release of the EpICD is sequentially accomplished by two enzymes (5), a TNF-␣-converting enzyme, ADAM17 (also known as TACE), and a ␥-secretase complex containing presenilin-2 (PSEN2). Moreover, it has been demonstrated that EpICD alone is sufficient to induce proliferation signals both in vitro and in vivo (5).
Human EpCAM has been shown to associate with a protein complex consisting of a group of membrane proteins including tight-junction protein CLDN7 (8,9), CD44v6 (10), and at least two tetraspanins, TSPAN8 (11) and CD9 (12) in rat carcinoma * This work was supported by an intramural grant from Academia Sinica, cells and human colon cancer cells. CLDN7 contributes to the formation of the complex by recruiting EpCAM into the tetraspanin-enriched membrane microdomain (9). The role of CLDN7 in regulating EpCAM function has been further demonstrated by the observation that an EpCAM-CLDN7 complex, rather than EpCAM itself, can promote proliferation, apoptosis resistance, migration, and tumorigenicity (13). However, whether CLDN7 and these tetraspanin proteins form a functional complex with EpCAM in other cell types or tissues remains to be elucidated.
In addition to its basolateral localization in some normal epithelial tissues, EpCAM is also known to be highly expressed in many epithelial carcinomas (1,14), cancer stem cells (15,16), and mouse (17) and human (18 -21) ESCs. Mouse EpCAM has been shown to be essential to the maintenance of pluripotency of mouse ESCs (mESCs) (17). In human ESCs (hESCs), Ng et al. (20) demonstrated that EpCAM expression is down-regulated during differentiation and EpCAM knockdown decreases cell proliferation and increases gene expression in the endoderm and mesoderm lineages. In addition, Lu et al. (21) showed that EpCAM knockdown diminishes the expression of pluripotency genes such as OCT4, NANOG, SOX2, and KLF4; and they proposed that an EpCAM nuclear complex may directly regulate the promoters of these pluripotency genes.
Recently, both mouse (22) and human (23,24) somatic cells have been successfully reprogrammed into iPSCs with defined transcription factors such as OCT4, SOX2, KLF4, and MYC (OSKM). Previous studies indicated that iPSCs resemble ESCs in morphology, growth requirements, proliferation, expressions of markers and genes, and epigenetic and functional features (25). Thus, iPSCs have been regarded as an attractive contender for patient-specific regenerative medicine as they circumvent the immunorejection problem and the ethical issues associated with ESCs. Despite these advantages, the low efficiency of iPSC derivation is one of the major challenges in the field of somatic cell reprogramming. Recently, several studies have put forward methods by which reprogramming efficiency can be improved, including: 1) the use of chemicals modifying chromatin structures such as valproic acid (26), 5-azacytidine (27), and a combination of BIX and BayK (28); 2) the use of inhibitors of signaling pathways such as TGF-␤ inhibitors (29), and a combination of MEK and GSK 3 inhibitors (30,31); 3) the use of antagonists of senescence, such as suppressors of p53 (32,33) and vitamin C (34); 4) hypoxia (35); and 5) the use of Cdh1 (E-cadherin) (36). These diversified approaches have revealed that reprogramming is a complex process involving many levels of subcellular biochemical and molecular events, and implicated that efficient reprogramming may be achieved by effective manipulation at several of these levels.
The purpose of this study was to determine whether membrane proteins that form a complex with EpCAM in some cancer cells also form a complex with EpCAM in ESCs and iPSCs, and whether components of this complex play a role in enhancing the efficiency of fibroblast reprogramming in the hope of advancing the search for a more effective means of generating iPSCs.

EXPERIMENTAL PROCEDURES
Cell Culture-Mouse embryonic fibroblasts (MEFs) and 293T cells were cultured in DMEM supplemented with 15% fetal bovine serum (FBS), 1ϫ nonessential amino acid (Invitrogen), 2 mM L-glutamine (Invitrogen), and 1ϫ penicillin/streptomycin (Invitrogen). mESC D3 and mouse iPSCs (miPSCs) were cultured in DMEM containing 15% FBS, 1ϫ nonessential amino acid, 2 mM L-glutamine, 0.0007% ␤-mercaptoethanol, and 103 unit/1 ml of leukemia inhibitory factor (Millipore) on a monolayer of mitomycin C-treated MEFs (feeder cells). Human ESCs and iPSCs were grown on MEF feeders (2 ϫ 10 4 cells/cm 2 ) in DMEM/F-12 medium plus 20% Knock-out Serum Replacement (Invitrogen) and 4 ng/ml of bFGF (Sigma). Human foreskin fibroblasts (HFs) from a 28-year-old male were obtained with written informed consent in accordance with the protocol approved by the Internal Research Board of National Taiwan University Hospital and Academia Sinica. Human granulose cells were obtained from donors of the In vitro Fertilization Center at National Taiwan University with signed informed consent. Both HFs and granulose cells were cultured in a medium similar to the MEF medium described above except that FBS was 10%.
Generation of miPSCs-Mouse iPSCs were generated using lentiviruses produced by TetO-FUW-mOSKM (Addgene) that contained mouse Oct4, Sox2, Klf4, and c-Myc cDNAs in one plasmid, and simultaneously by FUW-M2rtTA (Addgene). 293T cells were transfected with these two lentiviral vectors accompanied with pCMV⌬8.9 and pCMV-VSVG (Addgene) using FuGENE 6 transfection reagent (Roche Applied Science). Viral supernatant fractions were harvested at 60 and 84 h after transfection and filtered through a 0.45-m filter (Millipore). MEFs were then infected with two rounds of lentiviruses 24 h apart and incubated with viruses for another 24 h before the medium was changed to regular MEF medium. After 4 days, cells were transferred onto feeder cells and the medium was replaced with regular mESC medium. Doxycycline (2 g/ml) was added 24 h later to induce the expression of OSKM. iPSC colonies were subjected to in vitro and in vivo characterization, or manually picked and expanded 20 days after viral transduction. Generation of miPSCs using the retrovirus system was performed and characterized as described previously (22).
Alkaline Phosphatase Staining, Immunofluorescence Analysis, and Nanog Immunostaining-Alkaline phosphatase (AP) staining was performed using the Leukocyte Alkaline Phosphatase kit (Sigma) according to the manufacturer's instructions. Immunofluorescence (IF) staining was performed using the primary antibodies listed under supplemental Table S5. IF staining of the cells was observed under a Leica FW4000 confocal microscope (Leica Camera) or by epifluorescence microscopy with fluorescent optics. The details of IF analysis were described previously (37). The efficiency of reprogrammed iPSC colonies was evaluated by counting Nanog-positive colonies, which were visualized by immunostaining with anti-mouse Nanog antibodies using the diaminobenzidine method according to the manufacturer's instructions (Vector Labs).

Isolation of Human Foreskin Fibroblasts and Granulosa Cells
for iPSC Derivation-Human iPSCs were derived as described previously (2). Briefly, HFs or human granulose cells were plated at a density of 8 ϫ 105 cells/10-cm dish in DMEM with 10% fetal bovine serum. The next day, the lentivirus expressing the retrovirus receptor Slc7a1 (Addgene) and Virapower packaging mix (Invitrogen) were added to the fibroblast medium. Selection of fibroblasts expressing Slc7a1 was carried out by blasticidin (12 g/ml). Plat-E cells (Cell Biolabs) were seeded at a density of 8 ϫ 106 cells/10-cm dish with 1 g/ml of puromycin and 10 g/ml of blasticidin and transfected with 9 g of pMXs-hOCT4, pMXs-hSOX2, pMXs-hKLF4, or pMXs-hMYC (all from Addgene) using FuGENE 6 (Roche Applied Science). The blasticidin-selected HFs were transfected with supernatant from 4 different retroviruses (3 ml from each) and Polybrene (4 g/ml). Five days later, the fibroblasts were replated onto mitomycin C-inactivated MEFs. The next day, the fibroblast medium was replaced with ESC medium, as described above. ESC-like cell colonies emerged ϳ20 days after infection. They were picked up manually and cultured in conditions for ESC propagation. All experiments involving recombinant DNA were performed according to National Institutes of Health guidelines.
Characterization of miPSCs and Human iPSCs-Genomic DNA was extracted from iPSC clones with DNAEasy kit (Qiagen) and PCR analysis was performed with specific primers as described previously (22,23) to confirm the integration of retroviral transgenes. Total RNA of iPSC clones was extracted using RNAeasy Kit (Qiagen) and RT-PCR was performed with specific primers (22,23) to assay the expression of viral transgenes of Oct4 (OCT4), Sox2 (SOX2), Klf4 (KLF4), and c-Myc (c-MYC), and their endogenous counterparts.
Teratoma Formation-Approximately 1-2 ϫ 106 iPSC cells were injected into the rear leg muscles of 5-8-week-old NOD-SCID mice (National Laboratory Animal Center). Teratomas were allowed to develop for 10 to 12 weeks. They were then excised, fixed with 4% paraformaldehyde overnight at 4°C, and cryoprotected with 30% sucrose before embedding for cryostat sectioning. Samples were cut to 8 m in thickness, transferred to poly-D-lysine-coated slides, and stained with hematoxylin & eosin (H&E) for histological analysis. All animal experiments were approved by the Animal Care and Use Committee of Academia Sinica and performed in accordance with its guidelines.
Chimera Generation-The generation of iPSC chimeras was performed by laser-assisted microinjection at the 8-cell stage. Inducible iPSC chimeras were generated by injection of albino coat-colored ICR strain iPSCs into C57BL/6 ϫ DBA/2 F2 embryos (black, diluted-brown, or agouti). EpCAM-overexpressed iPSC chimeras were generated by injection of black coat-colored C57BL/6 iPSCs into ICR embryos.
Reverse Transcriptase PCR, Quantitative Real-time PCR, Lentivirus-mediated Short Hairpin RNA Knockdown, and Ectopic Gene Overexpression-Primers used are summarized under supplemental Table S6. All other details are described under supplemental "Methods").
Co-immunoprecipitation and Western Blotting-The antibodies used in co-immunoprecipitation and Western blotting are listed under supplemental Table S5). The details of co-im-munoprecipitation and Western blot assay are described under supplemental "Methods".
Luciferase Reporter Assay-Human OCT4 promoter-containing plasmid was a kind gift from Dr. Wei Cui (Imperial College London, UK). The 3-kb mouse Oct4 promoter fragment was PCR-amplified from the genomic DNA as described previously (38). Details are described under supplemental "Methods".
Statistical Analysis-All in vitro results were obtained from triplicate experiments. Results are presented as the mean Ϯ S.D. Student's t test was used to examine the significance of differences between group means, and differences with p values less than 0.05 were considered significant. For measurement of the strength of the association among the RNA expression levels of the EpCAM complex protein-encoding genes and pluripotency genes, Pearson correlation coefficients were determined using the Statistical Analysis System program (SAS Institute Inc.); p values less than 0.05 were considered significant.
To understand the time point at which EpCAM and Cldn7 are activated during reprogramming, and the relationship between their activation and the up-regulation of the pluripotency genes, we performed Q-RT-PCR to quantify the endogenous expression levels of these genes at different time points during the reprogramming of MEFs (Fig. 2, C and D). The results showed that the expression levels of both EpCAM and Cldn7 rose steadily from day 6 (a 10-and 6.2-fold increase for EpCAM and Cldn7, respectively) after the transduction of MEFs with the virus containing OSKM (Fig. 2C). The expression patterns of the two genes were very similar during this period with the expression levels of both genes reaching a peak at day 14 (Fig. 2C). In parallel, the expression levels of endogenous Oct4, Sox2, and Nanog also rose gradually, but more slowly than the expression of EpCAM and Cldn7 (Fig. 2D). Endogenous Oct4, Sox2, and Nanog expression levels also peaked at day 14 (Fig. 2D). Next, we explored the possibility that EpCAM regulates Oct4 expression through direct up-regulation of the Oct4 promoter activity. To this end, fragments of mouse and human Oct4 (OCT4) promoters (40), which have previously been shown to successfully direct the Oct4-specific pattern expression of GFP transgene (41), were cloned to upstream of the luciferase reporter gene in a pGL3-basic vector and co-transfected into 293T cells with the expression plasmid of either EpCAM or its intracellular domain, EpICD. Both EpCAM and EpICD could activate mouse Oct4 and human OCT4 promoter activities but EpICD resulted in a stronger activation (Fig. 2E). Taken together, these results suggest that initial ectopic expression of exogenous OSKM likely activates the expression of EpCAM complex proteins and this may con- (C) and Nanog, Sox2, and Oct4 (D). RNAs were extracted from OSKM-mediated reprogramming MEFs at various time points as indicated. Data are normalized to that of day 0 and presented as mean Ϯ S.E., n ϭ 3. E, luciferase assay for EpCAM (or EpICD)-mediated Oct4 promoter activation. Luciferase reporter assay showed that overexpression of EpCAM or EpICD activated the activity of mouse (left panel) and human (right panel) Oct4 promoterluciferase constructs, which contained 3-kb mouse and 4-kb human Oct4 (OCT4) promoter sequences, respectively. The luciferase activity was normalized to that of the control group (empty vector) and adjusted by the corresponding ␤-galactosidase enzyme activity of the co-transfected CMV-␤gal plasmid. Data are presented as mean Ϯ S.D., n ϭ 3.
tribute to the subsequent reactivation of endogenous pluripotency-associated genes, such as Oct4, in the reprogramming somatic cells.
Interruption of EpCAM or Cldn7 Expression in MEFs Reduces Reprogramming Efficiency-Because our results indicated that EpCAM and Cldn7 were highly expressed in both ESCs and iPSCs and activated during the course of pluripotency reprogramming, it was thus of interest to know whether disruption of EpCAM or Cldn7 expression in somatic cells would affect the efficiency of OSKM-mediated reprogramming. To this end, we knocked down the expression of EpCAM or Cldn7 by introducing EpCAM or Cldn7 shRNA into mouse MEFs using lentiviral vectors in conjunction with ectopic OSKM overexpression. The efficiency of knockdown was demonstrated using the same batch of viruses in the side-by-side infection of the Hepar 1-6 cell line, which expresses endogenous mouse EpCAM and Cldn7 (data not shown). At day 21 after doxycycline-induced expression of OSKM reprogramming factors (Fig. 3A), the reprogramming cells were harvested for Q-RT-PCR analysis (Fig. 3B), phase-contrast imaging (Fig. 3C), AP staining (Fig.  3D), and Nanog immunostaining (Fig. 3E). Quantitative RT-PCR analysis revealed that knockdown of either EpCAM or Cldn7 significantly reduced the expression levels of the majority of the pluripotency genes tested (Fig. 3B, 4

of 5 genes for
EpCAM knockdown, and 3 of 5 for Cldn7 knockdown), indicating that disruption of EpCAM or Cldn7 induction had a direct impact on the efficient activation of the endogenous pluripotency-related genes in the reprogramming cells. The morphology of either EpCAM-or Cldn7-knockdown colonies were irregular and loose unlike the typical compacted dome-like structure of mESC colonies in the control group (Fig. 3C). In addition, knockdown of either EpCAM or Cldn7 also significantly reduced the numbers of AP-positive (Fig. 3, D and F) and Nanog-positive (Fig. 3, E and G) colonies compared with the control groups. Taken together, these results suggest that disruption of either EpCAM or Cldn7 expression in the reprogramming MEFs significantly reduced the efficiency of OSKMmediated reprogramming.
Overexpression of EpCAM, EpCAM plus Cldn7, or EpICD Enhances Reprogramming Efficiency in MEFs-To test whether overexpression of EpCAM-associated proteins can promote OSKM-mediated reprogramming, we transduced MEFs with inducible lentiviruses encoding EpCAM, Cldn7, EpCAM plus Cldn7 or EpICD, in the presence of OSKM (Fig. 4A and supplemental Fig. S4) and the reprogramming efficiency was compared by various assays at day 21 after doxycycline induction. Q-RT-PCR analysis showed that transient overexpression of EpICD alone could significantly increase the expression level of endogenous Nanog (2.2-fold), Oct4 (1.9-fold), and Sox2 (2.2fold) (Fig. 4B); the overexpression of EpCAM or Cldn7 alone had less impact on pluripotency gene expression as only Nanog or Sox2 and c-Myc were significantly up-regulated in the reprogramming cells, respectively. Notably, simultaneous overexpression of EpCAM and Cldn7 resulted in a significantly higher level of pluripotency gene expression (Nanog (4.7-fold), Oct4 (3.8-fold), and Sox2 (4.3-fold)) than other groups, indicating that EpCAM and Cldn7 combined had a synergistic effect on the enhancement of reprogramming efficiency. This interpre-tation is also backed up by the significant increase in the numbers of AP-positive (Fig. 4, C and E) and Nanog-positive (Fig. 4, D and F) colonies in the EpCAM plus Cldn7 group. Together, these results suggest that ectopic expression of the EpCAM protein alone or together with Cldn7 in MEFs enhanced OSKM-mediated reprogramming efficiency.
Overexpression of EpCAM, EpICD, or EpCAM plus Cldn7 Suppresses the Expression of p53 during Pluripotency Reprogramming-As it has been reported that the activation of the EpCAM signaling pathway by anti-EpCAM antibody downregulated p53 expression (7), we next examined whether overexpression of EpCAM can down-regulate the expression of p53 and its downstream gene p21 in MEFs. The p53 expression in MEFs with or without overexpression of EpICD alone was analyzed by Q-RT-PCR and Western blot. The results showed that overexpression of EpICD did indeed have a significant repressive effect on the expression level of the p53 RNA (Fig. 5A) and protein (Fig. 5B) in MEFs when compared with the MEFs overexpressing the control vector. Furthermore, to prove that overexpression of EpCAM has a similar effect on p53 expression during the reprogramming of MEFs, we analyzed the expression levels of p53 and p21 in reprogramming MEFs infected with lentiviruses encoding OSKM and EpCAM, EpICD, Cldn7, EpCAM plus Cldn7, or control vector. The results showed that overexpression of all 4 groups had significantly down-regulated RNA levels of p53 and p21 at day 4 of reprogramming ( Fig. 5C) compared with vector control. However, after day 4, such an effect diminished and, in general, all the p53 and p21 RNA levels converged to similar levels at the end of reprogramming. Consistently, Western blot analysis also showed that both total and phosphorylated p53 protein levels were significantly decreased in the reprogramming MEFs that overexpressed EpCAM, EpICD, or EpCAM plus Cldn7 compared with those overexpressing the control vector 4 days after OSKM reprogramming (Fig. 5D); overexpression of Cldn7 alone had a less obvious effect. The above results indicated that during the early stage (day 4) of the reprogramming process, overexpression of EpCAM did down-regulate the expression of p53. The effect of EpCAM on p53 at later stages may therefore be masked by the up-regulation of OSKM, which are also strong down-regulators of p53. Taken together, these results suggest that EpCAM (or EpICD)-mediated p53 down-regulation at the early stage of reprogramming may contribute to the increased reprogramming efficiency seen in overexpression of EpCAM, EpICD, or EpCAM plus Cldn7.
Functional Characteristics of iPSCs Generated with Overexpression of EpCAM, Cldn7, EpCAM plus Cldn7, or EpICD-To determine the molecular and functional characteristics, we selected representative mouse iPSC clones generated by OSKM factors plus one of the following factors, EpCAM (iPSC-EP number 57), Cldn7 (iPSC-CL7 number 1), EpCAM plus Cldn7 (iPSC-EPCL7 number 7), or EpICD (iPSC-EPI number 24) for further characterization. RT-PCR and IF analyses demonstrated that miPSC-EP, CL7, EPCL7, and EPI robustly expressed pluripotency-related genes and proteins as well as EpCAM-associated proteins and their genes (Fig. 6, A and C,  and supplemental Fig. S5A). Because the expression levels of Nanog and Gtl2 are an important index of pluripotency status (42), we quantified the expression level of both genes in a number of clones of iPSC-EP (9 clones), -CL7 (10 clones), -EPCL7 (12 clones), and -EPI (13 clones) using Q-RT-PCR. Our results (Fig. 6D and supplemental Table S4) showed that the average Nanog expression levels in these clones were indistinguishable from that of control iPSCs (vector), whereas the average Gtl2 expression level of iPSC-EP clones, but not the other iPSC clones, was significantly higher (Fig. 6D, *, p Ͻ 0.05) than that of  Table S4) also indicated that iPSC-EP groups with higher numbers of clones had higher expression levels of Nanog (more than 1-fold greater than the expression in mESCs) than the control group. Furthermore, 2 representative cell clones from each of the iPS-EP, iPSC-CL7, iPSC-EPCL7, and iPSC-EPI groups were able to develop teratomas containing cell types of all three embryonic germ layers after injection into the muscle of NOD/SCID mice (Fig. 6E). The presence of three germ layer lineages in teratomas was further confirmed by ICC with specific antibodies against cells of the three embryonic germ layers (supplemental Fig.  S5B). By microinjection of cells from a representative iPSC-EPC clone into E3.5 mouse blastocysts, we were able to obtain adult chimeric mice with a high-grade of chimerism (Fig. 6F). Thus, our results demonstrated that iPSCs derived from OSKM alone or OSKM plus various EpCAM complex proteins, such as EpCAM, Cldn7, EpCAM plus Cldn7, or EpICD have similar in vitro and in vivo functional characteristics.

DISCUSSION
Although EpCAM has recently been shown to be expressed in mESCs (17) and hESCs (18 -21), as far as we know, no study has investigated the roles of EpCAM and its associated proteins in the somatic cell reprogramming process. In this study, we confirm and extend previous observations that EpCAM-associated proteins, including EpCAM, Cldn7 (CLDN7), Cd9 (CD9), Psen2 (PSEN2), Adam17 (ADAM17), are highly expressed and associated in ESCs and iPSCs with the exception of Adam17 (ADAM17). Our loss-and gain-of-function studies of EpCAM, Cldn7, and EpICD show that these proteins have a profound impact on modulating reprogramming efficiency. Thus the EpCAM nuclear signaling pathway is Our results showing that reprogramming efficiency can be enhanced by overexpression of EpICD alone suggest that the EpICD-mediated EpCAM nuclear signaling pathway may be the major contributor to the improvement in the reprogramming process. Importantly, the combination of EpCAM and Cldn7 had a synergistic effect on the enhancement of repro-gramming efficiency. The augmentation of the reprogramming efficiency by EpCAM and Cldn7 together may be explained by Cldn7-mediated modulation of the functional activity of EpCAM (13). Nevertheless, we should not rule out the possibility that Cldn7 may also be involved in the reprogramming process through an EpCAM-independent pathway. This possibility is supported by a previous report showing that knockdown of CLDN7 in cancer cells leads to loss of the expression of cdh1 . Data are presented as mean (n ϭ 3); S.D. was omitted for better visualization of data. b, further analysis of data in a indicated that at day 4, the RNA levels of both p53 and p21 were significantly lower in MEFs overexpressing EpCAM, EpICD, Cldn7, or EpCAM plus Cldn7 than those overexpressing vector control (*, p Ͻ 0.05). D, Western blotting (a) and the corresponding semiquantitative analysis (b) at day 4 of reprogramming showed that both total and phosphorylated p53 proteins were significantly decreased (*, p Ͻ 0.05) in MEFs overexpressing EpCAM, EpICD, or EpICD plus Cldn7 compared with control. Intensities of bands were first normalized to those of ␤-actin and then represented as the fold-relative to those of vector. Note that in b there was no significant difference in the level of total p53 protein between Cldn7 and vector. In both C and D, data are presented as mean (n ϭ 3) Ϯ S.D. (CDH1) (43,44), an important factor known to promote reprogramming (36). These observations therefore provide a further example, along with cdh1 (CDH1) (36,44), of the involvement of surface-to-nucleus regulation in regulating the somatic cell reprogramming process.
We reasoned that generation of EpICD in ESCs/iPSCs is accomplished by enzymatic cleavage by Psen2 (PSEN2) and Adam17 (ADAM17) as in colon cancer cells, although our coimmunoprecipitation results did not indicate that Adam17 (ADAM17) could be co-precipitated with EpCAM; this is probably due to the weak interaction between EpCAM and Adam17 (ADAM17) or low level of expression. Indeed, our result in this respect is consistent with a previous report in which the direct interaction of EpCAM and ADAM17 could be demonstrated by co-immunoprecipitation only when EpCAM was overexpressed in HEK293 cells (5).
Our study of EpCAM-associated protein expression showed that unlike EpCAM and Cldn7 (CLDN7), Cd9 (CD9), Psen2 (PSEN2), and Adam17 (ADAM17) were not only expressed abundantly in ESCs and iPSCs, but also in MEFs and HFs. This result is in agreement with previous reports (45)(46)(47)(48), which indicated that CD9 has ubiquitous distribution and is apparently present in fibroblasts (45,46), whereas both PSEN2 (47) and ADAM17 (48) are also widely expressed in many types of tissues and cells. Nevertheless, their roles in reprogramming should perhaps not be underestimated. For example, PSEN2 and ADAM17 may be directly involved in generating EpICD and thus they may be critical in enhancing the efficiency of reprogramming. The role of CD9 in reprogramming is less clear and awaits more studies, although there is a recent report (49) showing that Cd9 is dispensable for the maintenance of pluripotency in mESCs.
Recently, during the preparation of this manuscript, two studies have shown (44,50) that mesenchymal-to-epithelial transition is the hallmark of the initial step of somatic cell reprogramming, and is probably induced by the BMP signaling pathway and the downstream miRNAs (50). They also found that during mesenchymal-to-epithelial transition in the reprogramming of MEFs, the expressions of epithelial markers, such as Cdh1 and EpCAM, are up-regulated (44,50). Functionally, Chen et al. (36) and Li et al. (44) also showed that Cdh1-mediated cell-cell contact is critical for somatic cell reprogramming. These studies highlight the importance of signaling events from the extracellular milieu to the nucleus in the reprogramming process. In this study, we further addressed the role of EpCAM in iPSC generation. Consistent with previous reports (44,50), our Q-RT-PCR study showed that, like Cdh1, the expression of EpCAM was up-regulated during the OSKM-mediated reprogramming of MEFs. Moreover, we found that EpCAM played an important role in reprogramming, as manipulation of the expression level of EpCAM during the reprogramming of MEFs had profound effects on iPSC generation. A previous study showed that iPSCs established in the absence of adequate levels of Cdh1 have a status similar to so-called FAB-stem cells (36), however, in this study knockdown of EpCAM in the reprogramming cells resulted in nearly complete failure in iPSC generation. Thus, our observations indicate that EpCAM may have a more profound impact on somatic cell reprogramming than Cdh1.
Several possible mechanisms may explain how EpCAM enhances reprogramming. First, previous studies (6, 7) have shown that EpCAM or EpICD are able to increase the expression of cell cycle-promoting genes, which are known to enhance reprogramming efficiency (51). Second, because EpCAM is an adhesion molecule, it is capable of maintaining the proper cellcell contact that is required for mesenchymal-to-epithelial transition (44,50) and ultimately critical for derivation of pluripotent stem cells (17)(18)(19)(20)(21). Third, our current results (Fig. 2E) also provide evidence that one of the key pluripotency genes, Oct4, could be directly up-regulated by EpCAM and EpICD. This is consistent with a recent report (21) showing that EpCAM can bind to the promoter of human OCT4 in chromatin immunoprecipitation assays. Taken together, these results propose a model to address the relationship between the EpCAM complex and the pluripotency-related transcription factors (Fig. 7). The basic concept is that several positive feedback loops generate mutual augmentation of the expression of these genes that leads to complete reprogramming. In addition, Step 1, exogenous OSKM factors activate the expression of endogenous OSKM genes, and EpCAM as well as Cldn7.
Step 2, EpCAM and Cldn7 are transported to plasma membrane and associate with other members of the complex.
Step 3, EpCAM is cleaved by Psen2 and Adam17 to relieve EpICD, which is subsequently transported into the nucleus.
Step 4, EpICD forms a complex with other proteins to directly or indirectly enhance the expression of endogenous OSKM genes, such as Oct4 and c-Myc; on the other hand, EpICD-containing complex may down-regulate the expression of p53 and its downstream genes, such as p21. It has been shown that p53 can bind the promoter of EpCAM and inhibit its activity (53). Therefore, EpICD can positively regulate its own expression. Red lines, findings in this study; dashed lines and question marks, a possible but unproved effect.
EpCAM also interacts with several well known factors that can enhance reprogramming. For example, Yoshida et al. (35) showed that hypoxic conditions promote generation of iPSCs, whereas Trzpis et al. (52) demonstrated that the promoter activity of EpCAM in renal epithelial cells is increased under hypoxic conditions; it is thus possible that hypoxia may enhance reprogramming efficiency partly via activation of EpCAM expression. Another important EpCAM (or EpICD) interacting gene is p53 (P53), a well known factor that acts as a roadblock during the reprogramming process (32,33). Sankpal et al. (53) showed that p53 can bind to a regulatory element within the EpCAM gene and inhibit its activity (53). Furthermore, Maaser and Borlak (7) showed that activation of EpCAM signaling by an anti-EpCAM antibody down-regulated P53 expression. In line with these findings, our results show that overexpression of EpCAM (or EpICD) results in significantly lower levels of p53 RNA and phosphorylated p53 protein in both reprogramming MEFs and MEFs. It is, therefore, tempting to suggest that overexpression of EpCAM may promote reprogramming efficiency through the clearing of the p53 roadblock and thus further enhance EpCAM expression through a positive feedback loop (Fig. 7).
In summary, through the analysis of EpCAM-associated proteins in ESCs, iPSCs, and fibroblasts that underwent reprogramming, we have shown that EpCAM and Cldn7 were required for reprogramming of mouse somatic cells. Moreover, we found that overexpression of EpCAM or EpICD during the process of reprogramming further enhanced the reprogramming efficiency of MEFs, probably through EpICD-mediated activation of Oct4 and down-regulation of p53. Therefore, our results revealed that EpCAM is not merely a passenger gene that is up-regulated during reprogramming but a main player in a positive feedback loop among key factors that propel the reprogramming machinery.