Embryonic stem (ES) cells have the potential to differentiate into all three germ layers, providing new perspectives not only for embryonic development but also for the application in cell replacement therapies. Even though the formation of an embryoid body (EB) in a suspension culture has been the most popular method to differentiate ES cells into a wide range of cells, not much is known about the characteristics of EB cells. To this end, we investigated the process of EB formation in the suspension culture of ES cells at weekly intervals for up to 6 weeks. We observed that the central apoptotic area is most active in the first week of EB formation and that the cell adhesion molecules, except β-catenin, are highly expressed throughout the examination period. The sequential expression of endodermal genes in EBs during the 6-week culture correlated closely with that of normal embryo development. The outer surface of EBs stained positive for α-fetoprotein and GATA-4. When isolated from the 2-week-old EB by trypsin treatment, these endodermal lineage cells matured in vitro into hepatocytes upon stimulation with various hepatotrophic factors. In conclusion, our results demonstrate that endodermal cells can be retrieved from EBs and matured into specific cell types, opening new therapeutic usage of these in vitro differentiated cells in the cell replacement therapy of various diseases.
Embryonic stem (ES) cells derived from the inner cell mass of mammalian blastocysts are well known for their potential to maintain the undifferentiated state throughout an extended number of passages [1, 2]. Upon proper stimulation, ES cells differentiate into cells of various lineages, which can be used in cell replacement therapy . Transplantation of neural cells derived from mouse ES cells successfully rescued defective neurons in the central nervous system, proving their potential value in the treatment of neuronal diseases [4, 5]. More recently, ES cells have also been shown to differentiate into insulin-secreting β cells treating diabetic animal [3, 6, 7]. The development of human ES cell lines has widened the potential usage of ES cells even further, providing an excellent source for cell replacement therapy in various human diseases [8, 9]. Along with the technical progress in the cloning of mammalian cells, it is now quite conceivable to generate patients' own ES cell lines to develop stem cells of desired lineages .
To prepare ES-derived cells relevant to clinical situations, mouse ES cells have been experimentally differentiated via in vitro suspension culture into the embryoid body (EB), a cell clump comprised of all three germ layers. Various types of differentiated cells, such as neural cells, cardiac and skeletal muscle cells, hematopoietic cells, adipocytes, chondrocytes, and osteoclasts, are found in the EB [11–17]. Because differentiation of ES cells has been known to recapitulate changes in the embryonic development, factors that partake essential functions during early embryogenesis are also expected to be involved in the formation of EBs. Yet because of the short cultivation time and dearth of information about the cellular characteristics of EBs, little is known about the process guiding the development of three germ layers and specific lineage of cells within the EB.
Cell–cell contact mediated by various adhesion molecules and apoptosis play an important role in the early stage of embryonic development. However, the expression or function of catenins and cadherins has not been examined in EB cells. Apoptosis is also expected to be associated with the formation of EBs as well as the formation of three germ layers; however, definite histological location of apoptosis has not been determined within the EB.
Among specific markers of germ layers, GATA-4 and α-fetoprotein are considered endodermal markers initially expressed in the primitive endoderm during early postimplantation stages and are maintained in the visceral and parietal endoderm of the yolk sac during gastrulation. Nestin is expressed in the neuroectodermal area. Desmin is expressed in the mesodermal area, especially muscle fibers. These markers are useful for identifying three germ layers in EBs.
In this study, we confirmed the relative location and the expression of marker genes of three germ layers in EBs, especially the cells of endoderm, which thus far have been the least characterized of the three. Furthermore, those endodermal cells were differentiated into hepatocytes by adding hepatotrophic factors. Our findings provide grounds for developing EB-derived cells into various stem cell lineages that can be used in the cell replacement therapy of hard-to-cure diseases.
Materials and Methods
The 129/SvJ ES cell line used in this study was a subcell line of J1 ES cells originally established by Dr. Rudolph Jaenisch (Whitehead Institute for Biomedical Research, Cambridge, MA). Cells were maintained on the feeder of mitomycin C–treated primary mouse embryonic fibroblasts in Dulbecco's modified Eagle's medium (DMEM) supplemented with 15% fetal bovine serum (FBS) (Gibco BRL, Gaithersburg, MD, http://www.gibcobrl.com), 0.1 mM β-mercaptoethanol, 1,000 U/ml leukemia inhibitory factor (LIF) (Gibco BRL), and nonessential amino acids. The ES cells were obtained at passage 10 and were passed four more times before freezing in liquid nitrogen. Frozen vials of ES cells were thawed and passed three more times before differentiation. The feeder fibroblasts were prepared from C57BL/6J embryos at 13.5–16.5 days postcoitum. Briefly, after removal of soft tissues, including heart, liver, and other viscerae, the remains of embryos were minced in 2 ml of trypsin-EDTA solution (0.05% trypsin and 0.5 mM EDTA) and passed through a 200-μm mesh. After centrifugation, the cell pellets were resuspended in DMEM with 10% FBS, cultured again until confluent, and resuspended in 2 ml freezing medium. The frozen feeder aliquots were thawed and cultured until confluent and treated with 10 μg/ml mitomycin in DMEM with 10% FBS for 2 hours in 37°C, washed three times with phosphate-buffered saline (PBS), and plated onto a gelatin-coated dish as a feeder layer.
Differentiation of ES Cells
Undifferentiated ES cells were cultivated in DMEM supplemented with 15% FBS (Gibco BRL), 0.1 mM β-mercaptoethanol, 1,000 U/ml LIF (Gibco BRL), and 1 × nonessential amino acid without feeder layer for 1 week to make feeder-free ES cells. LIF (2,000 U/ml) was added to ES cells during the culture period. To promote differentiation, LIF was removed and ES cells were cultivated in suspension for 6 weeks on bacterial dishes.
EBs were collected at weekly intervals, washed three times with PBS, fixed overnight in 10% neutral-buffered formalin, dehydrated in a series of alcohol gradients (70%–100%), embedded in paraffin, and examined for general histomorphology. Sections of 4 μm were stained with hematoxylin and eosin (H&E).
The terminal deoxynucleotidyl transferase–mediated dUTP-digoxigenin nick-end labeling (TUNEL) procedure kit (Apop-Tag; Oncor, Gaithersburg, MD, http://www.qbiogene.com) was used to evaluate the apoptosis of ES cells in EBs, according to the manufacturer's instructions.
Undifferentiated ES cells were harvested by treating with trypsin-EDTA solution and washed three times with PBS by centrifugation. EBs were harvested not by centrifugation sedimentation but by gravity and washed three times with PBS. Total RNAs were extracted by using RNeasy kit (Qiagen, Hilden, Germany, http://www.qiagen.com), according to the manufacturer's instructions. RNA preparation was treated with 2 units of RNase-free DNase (Promega, Madison, WI, http://www.promega.com) for 1 hour at 37°C, and the DN ases were in activated at 65°C for 10 minutes. cDNA was synthesized from 4 μg of total RNA in 50 μl reaction mixture, containing oligo (dT) primers (0.8 μM) and Moloney murine leukemia virus reverse transcriptase (400 units; Promega), for 1 hour at 42°C. Polymerase chain reaction (PCR) amplifications were carried out using 2 μl cDNA product with 1.5 units of Extaq (Takara, Otsu, Japan, http://www.takara.co.jp) in 25 μl reaction mixture, according to the manufacturer's instructions. Oligonucleotide primers and PCR conditions used for reverse transcription (RT)–PCR reaction are listed in Table 1. PCR products were analyzed on a 1% agarose gel and visualized by ethidium bromide staining.
Table Table 1.. Polymerase chain reaction primers and conditions in this study
The rabbit polyclonal antibody against human α-fetoprotein was purchased from Quartett (Berlin, http://www.quartett.com), and the mouse monoclonal antibody for desmin was purchased from Signet (Dedham, MA, http://signetlabs.com). The goat polyclonal antibody for GATA-4 and the goat ImmunoCruz staining system were purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, http://www.scbt.com). The rabbit polyclonal antibody against nestin was purchased from BD Pharmingen (San Diego, http://www.bdbiosciences.com/pharmingen). The desmin antibody was purchased from Immunotech (Luminy, France, http://www.immunotech.com). Horseradish peroxidase–conjugated secondary antibody against mouse immunoglobulin G (IgG) was purchased from DAKO (Copenhagen, Denmark, http://www.dakocytomation.com). Cell adhesion sampler kit was purchased from Transduction Laboratories (Lexington, KY, http://www.translab.com). Immunostaining was carried out using the EnVision kit (DAKO) and the goat ImmunoCruz staining system, following the manufacturer's instructions. Briefly, for the EnVision kit, the deparaffinized sections were microwaved in target retrieval buffer for 20 minutes and blocked for the endogenous peroxidase by incubating in the peroxidase block solution for 10 minutes. The sections were incubated with primary antibody for 1 hour, followed by incubation in EnVision solution for 30 minutes, washed again, and treated with chromogenic substrate solution for 10 minutes. After 30-second counterstaining with hematoxylin, the sections were dehydrated, cleared, and mounted for 30 minutes at room temperature.
In the case of the goat ImmunoCruz staining system used for GATA-4 antibody, the use of the biotinylated anti-goat IgG and peroxidase-conjugated streptavidin was the only difference from the EnVison kit.
Isolation and Characterization of Endodermal Cells
Two-week-old EBs were treated with 2 ml of the trypsin-EDTA solution for 1, 3, 6, and 9 minutes, respectively, at room temperature to isolate external endodermal cells. Upper-layer after-gravity sedimentation was harvested and loaded on a 12-mm cover-slip placed in 24-well plates for isoelectric focusing or in six-well plates for RT-PCR. For immunofluorescence, cells on the cover-slips were briefly rinsed in PBS, fixed with 4% paraformaldehyde in PBS for 20 minutes, and permeabilized with 0.1% Triton X-100 in PBS for 5 minutes at room temperature. The fixed cells were preincubated with 0.1% bovine serum albumin (BSA) in PBS for 1 hour, incubated with primary antibody for 1 hour at room temperature, rinsed three times with BSA in PBS, and then further incubated for 1 hour with an appropriate Cy3-conjugated antibody. After rinsing, coverslips were fixed with mounting medium (Vecta, Burlingame, CA, http://www.vectorlabs.com).
In Vitro Maturation of Endodermal Cells
Endodermal cells isolated by trypsin treatment for 3 minutes were plated on 0.2% gelatin-coated dishes and cultured in the presence of acidic fibroblast growth factors (aFGF) (100 ng/ml) for 2 days. On the second day, hepatocyte growth factor (HGF) (20 ng/ml) was added, and cells were replated at day 5 on matrigel matrix (0.362 mg/ml, BD Biosciences, Bedford, MA, http://www.bdbiosciences.com) containing HGF, oncostatin M (OSM) (10 ng/ml), dexamethasone (Dex) (10−7 M), and a mixture of insulin, transferrin, and selenious acid (ITS) (5 mg/ml insulin, 5 mg/ml transferrin, and 5 μg/ml selenious acid [BD Biosciences]) and incubated for 3 days. To monitor hepatic development, expression of glucose-6-phosphatase (G-6-P), phenylalanine hydroxylase (PAH), and tyrosine aminotransferase (TAT) was assessed by RT-PCR.
Confirmation of Endodermal Cell–Derived Hepatocytes
Periodic Acid-Shiff Histochemical Staining
Differentiated hepatocytes were immersed in periodic acid solution for 5 minutes at room temperature and rinsed three times with distilled water. Cells were treated with Schiff's reagent (Sigma, St. Louis, http://www.sigmaaldrich.com) for 15 minutes at room temperature, washed in running tap water for 5 minutes, and examined with light microscope.
Indocyanine Green Uptake Staining
Indocyanine green (ICG) (25 mg; Dai-ichi Pharm. Co., Ltd., Tokyo, http://www.daiichipharm.co.jp) was dissolved in 5 ml of solvent in a sterile vial and then added to 20 ml of DMEM containing 10% FBS. The final concentration of the resulting ICG solution was 1 mg/ml. The ICG solution was added to the cell culture dish and incubated at 37°C for 15 minutes. After the dish was rinsed three times with PBS, the cellular uptake of ICG was examined with a stereomicroscope. After the examination, the dish was refilled with DMEM containing 10% FBS. ICG was completely eliminated from the cells 6 hours later.
Morphological Changes and Apoptosis in the EB
To examine the process of cell differentiation during EB formation, undifferentiated ES cell colonies were detached and grown in suspension for 6 weeks. As shown in Figure 1, H&E staining of EBs formed in the suspension culture revealed massive morphological changes accompanied by differentiation of diverse cell types during this period. As early as 1 week, cells within the EB started to show the typical characteristic of large nucleus and scanty cytoplasm outside of it. Chromatin condensation, cytoplasmic vacuolization, disruption of the nuclear membrane, and nuclear fragmentation were also observed in EB cells (Fig. 1A). Various phases of apoptosis were detected with TUNEL assay (Fig. 1G). Starting from the second week, cells with distinct characteristics could be discerned in EBs (Figs. 1B, 1C). At the same time, primitive neural tube–like structure (Fig. 1D), a cylinder-like structure imitating gut tube (Fig. 1E), and squamous epithelium-like structure (Fig. 1F) appeared. Central apoptotic areas were first seen in 1-week-old EBs (Fig. 1G) and then shrunk down to an undetected level by the sixth week (Figs. 1H–1L).
Expression of Cell Adhesion Molecules During the Development of EBs
Adhesive characteristics of EB cells were assessed by analyzing the gene expressions of cell adhesion molecules (Fig. 2). E-cadherin, α-catenin, γ-catenin, and desmoglein-2 were all continuously expressed in undifferentiated ES cells as well as in 6-week-old EBs. On the other hand, the message level of N-cadherin and β-catenin dwindled along with the progression of differentiation. Interestingly, paxillin mRNA, which was abundant in undifferentiated ES cells, showed a transient decrease during the second and fourth weeks but became abundant again by the sixth week.
We also examined the expression of these adhesion molecules at the protein level by immunohistochemical staining. The expression of α-catenin and β-catenin remained strong throughout the experimental duration (Figs. 3A–3F). γ-Catenin was scarce in the 1-week-old EB but gradually increased in the following weeks (Figs. 3G–3I). In accordance with the level of their mRNAs, E-cadherin and N-cadherin proteins showed sustained expression throughout the examination period (Figs. 4A–4F). The neural tubes, which were strongly stained with N-cadherin antibody, were not stained with E-cadherin antibody (Figs. 4B, 4E).
Expression of Genes Specific to the Neural Ectoderm and Mesoderm
The expression of ectoderm and mesoderm markers revealed distinct changes during the process of EB development (Fig. 5). In RT-PCR analysis, mRNAs of nestin, a protein specific to neural stem cells, were abundant in undifferentiated ES cells but slightly decreased from the third week and thereon. The level of Pax-6 transcript, another neural stem cell marker, showed a peak in the 1-week-old EB. Expression of Flk-1 and platelet endothelial cellular adhesion molecule, both known to be mesodermal specific, was differentially regulated during the examination period. The level of collagen IV, a cartilage component, increased during the first week and remained strong throughout the rest of the culture period.
When examined by immunohistochemical staining, nestin protein was abundantly expressed all over the 1-week EB except in the apoptotic areas (Fig. 6A). In later EBs, nestin proteins were detected in focal areas in 2- to 3-week-old EBs (Figs. 6B, 6C) and then in the primitive neural tube–like structures until the fifth week (Figs. 6D, 6E). By 6 weeks, nestin-positive neural tubes were destroyed and lost their cell–cell contact (Fig. 6F).
We also examined the localization of desmin, a mesoderm marker specific to striated and smooth muscle. Desmin proteins were not detectable for the first 3 weeks of culture but appeared in the 4-week-old EB (Figs. 6G, 6H). At 5 weeks, cystic areas in EBs showed internal desmin-positive areas, imitating the typical structure of muscle fibers inside endodermal layers (Figs. 6I, 6J). Desmin proteins were also present in areas outside of the cysts in the 5-week-old EB (Figs. 6K, 6L).
Expression of Endoderm-Specific Genes
Next we compared the expression of endodermal genes in the EB, in 16.5-day embryos, and in the adult liver (Fig. 7). The expression of Oct-4, a specific marker for undifferentiated ES cells, showed a continuous decrease throughout EB development and was absent in the E6.5-day embryo or in the adult liver. GATA-4, which is expressed in visceral endoderm, showed increasing expression in 1- to 2-week-old EBs but decreased thereafter. mRNAs of α-fetoprotein and transferrin were not detectable in the ES cells, continuously increased between 1 and 3 weeks, and started to decrease after 5 weeks. Both transcripts were highly expressed in the E16.5-day embryo, but α-fetoprotein was absent in the adult liver. Three liver-specific molecules, that is, transthyretin (TTR), aldolase, and albumin, were highly expressed in early weeks and continuously decreased throughout the rest of the culture period.
By immunohistochemical analyses, α-fetoprotein protein was localized to the outer layer of the EB (Fig. 8) until the fourth week. After 5 weeks, small areas inside of the EB were positive for α-fetoprotein (Figs. 8E, 8F). GATA-4 was present in both outer layers and inner areas of the EB (Figs. 8G–8L) but was avoided in the area around the neural tube–like structures (Fig. 8J).
In Vitro Differentiation of Endoderm-Like EB Cells into Hepatocytes
To further characterize cells from the outer endodermal layer, 2-week-old EBs were treated with trypsin, and retrieved cells were analyzed by RT-PCR for the expression of endoderm-specific genes such as α-fetoprotein, Apo2, and TTR. The expression level of these endodermal genes increased with the duration of trypsin treatment (Figs. 9, 10). Immunohistochemical staining revealed that α-fetoprotein is strongly expressed in cells isolated by 3-minute trypsin treatment (Fig. 11).
Next we examined the developmental potential of the dissociated endoderm-like cells by cultivating them with various hepatotropic factors, based on previous report of Hamazaki et al.  on hepatic differentiation. The endodermal cells were cultured in gelatin-coated dishes for 2 days with aFGF and then with HGF for another 2 days. After replating on matrigel matrix–coated dishes, cells were treated with HGF, OSM, and Dex for 4 days, followed by incubation with ITS for 3 days. After this 7-day stimulation regimen of hepatic differentiation, total RNAs were extracted and analyzed for the expression of liver metabolic enzymes. RT-PCR confirmed that the combined stimulation with growth factors, cytokines, and matrigel matrix strongly induced the expression of liver metabolic enzymes, such as G-6-P, PAH, TAT, and cytochrome P450s (P450-2B10 and P450-cb), in these in vitro–maturated endoderm cells (Fig. 12). With this maturation condition, polygonal hepatocyte-like cells were seen after a 7-day maturation period (Figs. 13A, 13B). Binucleated cells suggesting hepatocytes were seen (Fig. 13C). Those cells were stained with albumin, indicating that hepatocyte-like cells were possessing functional hepatic protein (Fig. 13D). Furthermore, these hepatocyte-like cells were stained with periodic acid-Shiff and ICG (Figs. 14A, 14B).
With their pluripotency and continuous proliferation, ES cells are the most convenient source of donors in cell replacement therapy. ES cells have been found to differentiate in vitro to many clinically relevant cell types [11–17], and the ES-derived endoderm cells have eagerly been tested for their applicability to the cell-mediated therapy of chronic human diseases. Along with the previous success in generating ES-derived neurons [4, 5] and pancreatic β cells [3, 6, 7], the next hurdle is to differentiate ES cells into fully functional hepatocyte for the treatment of chronic liver diseases. Expression of hepatocyte-related genes has been demonstrated in ES cells by RT-PCR [18–21], and markers for fully differentiated hepatic cells were identified in ES cells differentiated in dexamethasone and OSM containing culture medium . Recently, several independent reports showed definite differentiation of hepatocytes from ES cells in vitro and in vivo [22–27]; however, basic mechanisms that underpin the differentiation of ES cells into endodermal cells and hepatocytes have not been elucidated.
Although the generation of EBs in the suspension culture of undifferentiated ES cells has been adopted by many scientists as an initial differentiation strategy, it accompanies several inherent problems. Many cells located inside the EB are in apoptotic status, indicating that most ES cells die instead of differentiating and producing healthy ES-derived cells. Therefore, it seems that an alternative differentiation method should be found to retrieve viable cells from EBs. Another setback in using EBs as the source of differentiated ES cells is the dearth of information about their characteristics. In the present study, we analyzed the expression of lineage-specific markers in a time-course examination of the EB and assessed the possibility of differentiation trypsin-treated endoderm cells into a specific cell type via in vitro stimulation with hepatotropic agents.
Cell–cell contact mediated by various adhesion molecules plays an important role in the early stage of embryonic development. For example, the E-cadherin–catenin complex is crucial in the embryonic compaction and the maintenance of epithelial layers  and is highly expressed throughout the cleavage stage. Null mutant embryos depleted of β-catenin or E-cadherin fail to survive, suggesting that cell adhesion molecules play essential parts in early development [27, 28]. However, the expression or function of catenins and cadherins has not been examined in EB cells. Cell adhesion molecules are important markers for ES cell differentiation. Our data showed that E-cadherin was strongly expressed not only in undifferentiated ES cells but also in most parts of the EB, except primitive neural cells. These neural tube–like cells expressed N-cadherin instead, confirming the requirement for N-cadherin shown previously in the in vivo development of neural tube using N-cadherin–deficient ES cells . However, homogeneous expression of cell adhesion molecules hampered the purification of differentiated ES cells by these molecules using the fluorescence-activated cell scanner method.
Apoptosis is a natural biological process of development and is regulated by the intricate balance between cell-cycle proteins, tumor suppressor genes, and protooncogenes [30, 31]. Several studies have revealed that these apoptosis-related genes are involved in the differentiation of ES cells into EBs. For example, the antiapoptotic protein Bcl-2 inhibited retinoic acid–induced cell death during the differentiation of ES cells into neurons , whereas K-ras and LIF prevented apoptosis in the early differentiation of ES cells [32, 33]. Apoptosis is also expected to be associated with the formation of EB as well as the formation of three germ layers; however, definite histological location of apoptosis has not been determined within the EB.
Our study also revealed that if provided with proper culture condition, ES cells not only can differentiate within the EB into fully differentiated cells but also can form primitive tissues. An example of this is squamous epithelial tissue in 6-week-old EBs. Also, structures resembling the primitive neural tube and the intestinal tube-like cysts  appear in the 4-week-old EB. Because numerous distinct cell types can be generated in vitro depending on the procedures used to induce differentiation [7, 23, 35], it is quite possible to generate and isolate clinically relevant cell types other than the ones reported here from EBs.
Among specific markers of germ layers, GATA-4 is initially expressed in the primitive endoderm during early postimplantation stages and is maintained in the visceral and parietal endoderm of the yolk sac during gastrulation. Derivatives of the definitive endoderm, including cells associated with heart formation, also express GATA-4 later in development . Relationship between the endoderm and GATA-4 was confirmed by knockout experiments, in which GATA-4–depleted EBs failed to form outer visceral endoderm [36, 37]. Formation of endodermal layers in the surface of EBs has been confirmed by the presence of prealbumin (also called TTR) , yet the expression of GATA-4 has not been determined. Other than endodermal cells, EBs have been shown to contain cells of ectodermal lineage, including neural cells, fully differentiated neurons, astrocytes, oligodendrocytes, and dopaminergic and serotonergic neurons [4, 5, 35]. Many studies have also identified mesodermal cells in EBs [11, 14, 17, 38].
Unlike in normal embryos, endodermal cells are located in the outer surface of EBs [12, 36]. We observed cells of relatively large size in the outer layer of EBs that were stained positive with endodermal markers GATA-4 and α-fetoprotein. However, these endoderm-like cells start to disappear after the fourth week of the culture. Taking advantage of the external localization of endoderm layer in EBs, we attempted to retrieve relevant cells while they existed by brief trypsinization. Our results showed that differentiated hepatocytes producing liver-specific metabolic enzymes were generated after maturing the EB-derived endoderm cells with hepatotropic factors. The attached endoderm cells to gelatin-coated dish proliferated well under hepatotropic factors aFGF and HGF, but after replating on matrigel matrix–coated dishes, they did not seem to proliferate. To determine the clinical relevance of this procedure, it should be reevaluated with human ES cells.
In conclusion, our results demonstrate the potential of mouse ES cells to differentiate into endodermal cells expressing proper markers on the outer surface of EBs and the enrichment of these cells not by genetic manipulation but by mechanical dissociation and maturation with desired growth factors. This procedure will help researchers use human ES cells in the generation of donor cells with pure and discrete lineage for the cell replacement therapy of various human maladies.
D.C. and H.-J.L. contributed equally to this study. This work was supported by research grants from the Korea National Institutes of Health and from the Korea Health 21 R&D Project, Ministry of Health and Welfare, Republic of Korea (00-PJ1-PG1-CH5-0004) and Soonchunhyang University Research Fund (20040161). We thank Dr. Won-Ki Paik for the critical reading and discussion of this manuscript. We are grateful to Dr. Se Jin Jang for technical advice on histological analyses. We also thank Cheol Yong Song, Jae Hyeong Kim, Kuk Hyeon Lee, Tai Ho Im, and Byung-Seok Park for their financial support.