The in vitro differentiation of mouse embryonic stem cells into different somatic cell types such as neurons, endothelial cells, or myocytes is a well-established procedure. Long-term culture of rat embryonic stem cells is known to be hazardous, and attempts to differentiate these cells in vitro so far have been unsuccessful. We herein describe stable long-term culture of an alkaline phosphatase-positive rat embryonic stem cell-like cell line (RESC) and its differentiation into neuronal, endothelial, and hepatic lineages. RESCs were characterized by typical growth in single cells as well as in embryoid bodies when cultured in the presence of leukemia inhibitory factor. RESC expressed stage-specific-embryonic antigen-1 and the major histocompatibility complex class I molecule. For neuronal differentiation, cells were incubated with medium containing 10−6 M retinoic acid for 14 days. For endothelial differentiation, RESCs were grown on Matrigel for 14 days, and for induction of hepatocyte-specific antigen expression, RESCs were grown in medium supplemented with fibroblast growth factor-4. Differentiated cells exhibited typical morphological changes and expressed neuronal (nestin, mitogen-activated protein-2, synaptophysin), glial (S100, glial fibrillary acid protein), endothelial (panendothelial antibody, CD31) and hepatocyte-specific (α-fetoprotein [αFP], albumin, α-1-antitrypsin, CK18) antigens. In addition, expression of hepatocyte-specific genes (αFP, transthyretin, carbamoyl-phosphate synthetase, and coagulation factor-2) was detected by reverse transcription polymerase chain reaction. We were able to culture RESCs under stable, long-term conditions and to initiate programmed differentiation of RESCs to endothelial, neuronal, glial, and hepatic lineages in the rat species.
Embryonic stem (ES) cells are blastocyst-derived cells that are characterized by pluripotency, the expression of alkaline-phosphatase , stage-specific embryonic antigen (SSEA-1) , octamer-binding protein 4 (Oct-4) , the ability of germline formation upon injection into early embryos , and their capacity for somatic differentiation. Several attempts have been made to isolate and culture pluripotent ES cell lines from different species including mouse , chicken , hamster , rabbit , pig , bovine , and human . But so far, only ES cells from mouse and chicken have been found to have germline competence . Attempts to establish a stable pluripotent ES cell line from rats have only been reported once .
Long-term maintenance of mouse ES cells (MESC) in their pluripotent undifferentiated state is provided by leukemia inhibitory factor (LIF) . MESC spontaneously form three-dimensional cell aggregates termed embryoid bodies, which are composed of ectodermal, endodermal, and mesodermal tissue. Directed differentiation of MESC occurred in vitro when growth factors were added in limited amounts. A number of protocols have been described to induce neuronal cells [14–16], adipocytes , smooth muscle myocytes , and endothelial cells [19, 20]. However endoderm-derived cells (gut, liver, and exocrine and endocrine pancreas) have proven difficult to obtain. Schwartz et al.  recently reported the differentiation of mouse, rat, and human multipotent adult progenitor cells from bone marrow into functional hepatocyte-like cells using fibroblast growth factor 4 (FGF-4).
In the present study we describe the establishment and characterization of rat embryonic stem cell-like cells (RESCs) from Wistar Kyoto (WKY) [RT1.l] rat blastocysts with respect to alkaline-phosphatase, Oct-4, major histocompatibility complex (MHC), SSEA-1 expression, and their proliferative capacity.
We recently showed that C12 RESCs were able to induce tolerance for second-set allogeneic heart transplantation . After intravenous injection, these cells could be found in the thymus, spleen, lymph nodes, and heart of nonimmunosuppressed allogenic recipient rats. Morphological and immunohistochemical studies showed that RESCs differentiated into monocytes, macrophages, and lymphocytes. However, so far, in vitro differentiation of these cells has not been successful.
We now describe the differentiation of RESCs into cells of the neuronal and glial lineage by treatment with retinoic acid as described for MESC [14–16]. Endothelial differentiation is generally induced by treatment with vascular endothelial growth factor (VEGF), erythropoietin, FGF-2, and interleukin (IL)-6 . Since Matrigel is known to induce tube formation of endothelial cells , we tested whether growing pluripotent cells in Matrigel can supply the stimulus for endothelial differentiation. Furthermore, RESCs were grown on Matrigel and treated with FGF-4 since this was shown to be the most potent stimulus for hepatic differentiation in bone marrow-derived cells .
Materials and Methods
Isolation, Growth, and Differentiation of RESCs
RESCs were isolated from the inner cell mass of 4- to 5-day-old blastocysts that were derived from pregnant WKY rats as described in Faendrich et al. . We used a single cell-cloned line termed C12-WKY, which contains approximately 70% euploid cells. For in vitro expansion of RESCs, a feeder cell line was established from embryonic fibroblasts of 14-day-old mice. C12 RESCs were grown on a mitomycin-treated feeder layer  or on gelatin-coated dishes (Sigma; Germany; http://www.sigmaaldrich.com). The stem cell medium was composed of high-glucose Dulbecco's modified Eagles medium (Sigma), 10% heat inactivated fetal bovine serum (FBS, GIBCO/Invitrogen; Eggenstein, Germany; http://www.invitrogen.com), 1% 200 mM L-glutamine (GIBCO/Invitrogen), 1% penicillin/streptomycin solution (50 IU/50 μg, GIBCO/ Invitrogen), 1% nonessential amino acids (GIBCO/Invitrogen), 5 ml nucleoside solution (self-made stock: 80 mg adenosine, 85 mg guanosine, 73 mg cytidine, 73 mg uridine, 24 mg thymidine dissolved in 100 ml ddH2O [Sigma]), insulin (0.09 mg/l) (Sigma), and 1,000 U/ml LIF (Chemicon International; Temecula, CA; http://www.chemicon.com). Undifferentiated RESCs were stained with antibodies to SSEA-1 (MC 480, Developmental Studies Hybridoma Bank; University of Iowa; http://www.uiowa.edu/∼dshbwww), alkaline phosphatase (DAKO; Hamburg, Germany; http://www.dako.dk), and KiS3R (kind gift of Professor Parwaresch, Department of Pathology, University of Kiel; Kiel, Germany).
RESCs were cultured without LIF in six-well plates on coverslips and treated with 1 × 10−6 M retinoic acid (Sigma) in basal medium (RPMI 1640; GIBCO/Invitrogen) with 10% heat inactivated FBS, 1% 200 mM L-glutamine, and 1% penicillin/streptomycin solution (as described above). After 5, 10, and 20 days, coverslips were washed in phosphate-buffered saline and fixed with methanol (Merck; Darmstadt, Germany; http://www.merck.com) for 10 minutes. Immunohistochemical staining was performed against mitogen-activated protein 2 (MAP2; Sigma), S100 (DAKO), glial fibrillary acidic protein (GFAP; Sigma), nestin (DAKO), and synaptophysin (Dianova; Germany; http://www.dianova.com) antigens.
Endothelial Differentiation on Matrigel
Twenty-four-well dishes were coated with a 5-mm Matrigel layer (Becton Dickinson; Hamburg, Germany; http://www.bd.com), a basal lamina extract. RESCs were cultured within the Matrigel. Immunohistochemistry against CD31 (Sigma) and panendothelial antibody (Serotec; Oxford, UK; http://www.serotec.co.uk/asp/index.html) was performed, and tube formation was photodocumented on days 5, 10, and 14.
Differentiation of Cells with Hepatocyte-Specific Properties
RESCs were cultured on coverslips precoated with Matrigel, and FGF-4 (3 ng/ml; Sigma) was added to their basal medium. Cells were fixed on days 5, 10, and 20. Immunohistochemical staining was performed against αFP (Sigma), albumin (DAKO), CK18 (Dianova), and α-1-antitrypsin (Dianova) antigens.
RT-PCR Analysis of Hepatocyte-Specific Markers
Total RNA was prepared from cultures of C12 cells using peqGOLD RNAPure (Peqlab; Erlangen, Germany; http://www.peqlab.de). cDNA was synthesized from 5 μg of total RNA from untreated control cultures or from FGF-4-treated cells cultured on Matrigel with oligo(dT)12–18 and Superscript II reverse transcriptase (Invitrogen GmbH) as described previously . For polymerase chain reaction (PCR), cDNA corresponding to 100 ng RNA was amplified with Taq-polymerase (Invitrogen GmbH) using oligonucleotide primers (sequences are listed below) in a hot-start touch-down program with the following conditions: denaturation at 95°C for 1 minute, annealing at 65°C for 1 minute, and extension at 72°C for 1 minute. Every fourth cycle, the annealing temperature was reduced by 2°C until a temperature of 55°C was reached at which time a varying number of cycles (depending on the abundance of the target mRNA, αFP) and coagulation factor II (CFII): 15, carbamoyl-phosphate synthetase I (CPSI): 25, transthyretin (TTR): 21) were performed. The number of cycles was chosen as to stay within the linear phase of the assay. All amplification products were separated on 1.5% agarose gels and visualized by staining with ethidium bromide. In all cases a single band of the size predicted from the published cDNA sequence was obtained. PCR amplification of αFP, CFII, CPSI, and TTR was performed with the following oligonucleotide primers: αFP: sense, 5′-GTCCTTTCTTCCTCCTG GAGAT-3′ and antisense, 5′-CTGTCACTGCTGAT TTCTC TGG-3′; CFII: sense, 5′-ACTACATTCACCCCGTGTGCT TGC-3′ and antisense, 5′-CACAAACCTATCTATGCTGA TCAATGAC-3′; CPSI: sense, 5′-CTATTCTGAGATGTGA GATGGCTTC-3′ and antisense, 5′-AGCGCTGTACTGCCT GTAGTGGAA-3′; TTR: sense, 5′-CAGCAGTGGTGCTGT AGGAGTA-3′ and antisense, 5′-GGGTAGAACTGGACAC CAAATC-3′ .
Cells were fixed with methanol (Merck) for 10 minutes and subsequently incubated with the primary antibody for 45 minutes. The incubation time of the second antibody was 45 minutes. The antigens were visualized with the APAAP protocol (DAKO) . For DAB staining, a biotinylated second antibody goat anti-mouse (Sigma) was used followed by a streptavidin complex (DAKO). After washing, cells were incubated with DAB reagent (DAKO). Reaction was stopped with tap water. Hemalum was used for nuclear counterstain. Stained material was mounted with Kaiser's glycerin gelatin (Merck).
For immunofluorescence, Alexa Fluor 450 nm Fab fragment rat anti-mouse antibody (Molecular Probes; Eugene, Oregon; http://www.probes.com) or fluorescein isothiocyanate coupled with secondary goat anti-mouse (Pharmingen; Hamburg, Germany; http://www.bdbiosciences.com/pharmingen) was used. Nuclei were counterstained with DAPI (Molecular Probes).
Typically, C12 RESCs grow in smooth round clumps. Dense established cultures stained almost uniformly for alkaline phosphatase (Fig. 1A), a commonly used ES cell marker. Staining for the carbohydrate antigen SSEA-1, a marker of MESC, showed a mosaic expression pattern in most of the clumps (Fig. 1B). C12 RESCs showed an expression of MHC class I antigen (Fig. 1C), but lacked MHC class II antigen (Ox-3) (Fig. 1D) or the expression of costimulatory molecules such as B7.1 (CD80−) and B7.2 (CD86−) (data not shown). In addition, immunofluorescence staining for B-cell (CD20), T-cell (CD3), natural killer cell (3.2.3), and macrophage (CD14) markers was negative (data not shown). The C12 cell line was highly proliferative, as determined by the rat-specific proliferation nucleus cell antigen marker KiS3R (Fig. 1E).
C12 RESCs Treated with Retinoic Acid
After 5–7 days of treatment with retinoic acid (RA), the first morphological changes could be observed. Similar to untreated cultures, C12 RESCs started to form embryoid bodies (EBs). The expression of S100 was detectable within the embryoid body before morphological changes occurred (Fig. 2A) and remained positive during the culture period. Hence, C12 RESCs lost their characteristic round shape, attached to the bottom of the culture dish, and developed a bipolar morphology characteristic for immature neurons (Fig. 2B). Between day 5 and 10, C12 RESCs started to exhibit long slender dendritic or neurite-like processes (Fig. 2C). Staining of RA-treated cells after 5 days showed the expression of nestin, which is expressed early in neuronal differentiation (Fig. 2D), whereas untreated C12 RESCs did not express nestin (data not shown). With morphological changes occurring, neuronal and glial markers were simultaneously upregulated. Glial cells could be identified after 5 days as having a polygonal shape and expressing the GFAP antigen (Fig. 3A). Cells having a bipolar morphology with dendritic processes appeared after 8–10 days of culture period, weakly expressing MAP2 (Fig. 3C) and synaptophysin (Fig. 3E). Untreated C12 RESCs lacked expression of late neuronal markers such as MAP2, synaptophysin, or the glial marker GFAP (Fig. 3B, D, F). During treatment with RA, the percentage of cells expressing neuronal markers (proportion of positive cells) increased up to 70% after 14 days, whereas longer culture periods did not yield a greater percentage of positive cells. The percentage of cells expressing glial antigens, however, was highest after 10 days of RA treatment (40%).
Endothelial Differentiation of C12 RESCs on Matrigel
After 5 days of culture on Matrigel, C12 RESCs developed a cord-like morphology. Several cells clustered together and began to form tube-like structures (Fig. 4A). After 10 days, C12 RESCs showed a three-dimensional growth pattern on the gel, forming a complex network of vascular-like structures (Fig. 4B). In the absence of Matrigel, C12 RESCs grew either on the bottom of the gelatin-coated dish or as EBs (Fig. 4C). Their morphology was comparable to the network that is formed by human umbilical vein endothelial cells (HUVEC) when cultured on Matrigel (Fig. 4D).
After a culture period of 10 days on Matrigel, C12 RESCs exhibited a strong expression of CD31 and panendothelial antibody in tube-like structures (Fig. 4E). C12 RESCs cultured without Matrigel did not show expression of endothelial cell markers or the typical tube-like morphology (Fig. 4C).
C12 RESCs Acquire a Hepatocyte-like Phenotype When Cultured with FGF-4
Morphological changes of C12 RESCs occurred after 3 days of treatment with FGF-4. Cells acquired a round morphology, attached to the bottom of the culture dish, and their cell diameters increased (Fig. 5A). A subpopulation of FGF-4-treated cells became diploid, and the cytoplasm of a fraction of FGF-4-treated RESCs darkened, resembling the cytoplasm of primary hepatocytes (Fig. 5B). Immunohistochemical staining revealed expression of hepatocyte markers such as CK18 (Fig. 5C), albumin (Fig. 5D), αFP (Fig. 5E), and α-1-antitrypsin (Fig. 5F), whereas untreated cells were devoid of these antigens (data not shown).
FGF-4 Induces Hepatocyte Differentiation
We confirmed hepatocyte-like differentiation by semiquantitative RT-PCR for a set of markers for hepatocyte differentiation, namely αFP, CFII, CPSI, and TTR. On day 0, cultures of C12 cells expressed low levels of CFII, CPSI, and TTR, while αFP expression was almost undetectable. In cultures grown on Matrigel and treated for 21 days with FGF-4, αFP, and TTR, expression was strongly upregulated (Fig. 6). The mRNA for CFII and CPSI also increased, albeit to a lesser extent (Fig. 6). However, it should be mentioned that some late markers of hepatocyte differentiation (CK18, albumin, hepatic nuclear factor 1 [HNF-1]α, and CYP2B1) were expressed but not upregulated at the transcript level in FGF4-treated cells (data not shown).
Properties of RESCs
In this study, a rat embryonic stem cell-like cell line (C12) was used, which has not yet been described or characterized by others. C12 RESCs express alpaline phosphatase and SSEA-1 but are immunonegative for Oct-4 (unpublished observations). Vassilieva et al. were the first to establish a stable, long-term rat embryonic-like cell line , which shares several properties with the C12 RESC line. Both cell lines grow in a morphologically similar manner forming tightly compacted aggregates that are difficult to dissociate . Both RESC lines are significantly different than MESC in their behavior towards enzymatic digestion, as RESCs tend to differentiate immediately thereafter . For this reason, RESCs were passaged by mechanical dissociation. Both cell lines required LIF in order to inhibit differentiation as described for MESC , and they could be kept in their undifferentiated state in the absence of feeder layer. In contrast to the cell line described by Vassilieva et al., C12 RESCs did not express Oct-4 (unpublished findings). Restricted expression during the germline cycle of mouse development [27–29] suggests that Oct-4 expression is associated with an early undifferentiated state of ES cells. The reason for the lack of Oct-4 expression in RESCs could be that mouse-specific antibodies were used, or that the highly compact C12 RESC colonies were more mature than the cells described by Vassilieva et al. These cells express Oct-4, albeit at a much lower mRNA level compared with MESC . Undifferentiated C12 RESCs expressed the MHC class I molecule (I1.69), whereas there was no detectable expression of the MHC class II molecule (O×3). However, we observed that after intraportal injection of C12 RESCs into allogeneic hosts  30%-40% of the injected cells stained positive for MHC class II antigen (O×3). Implicitly, the persistence of C12 rat embryonic stem cell progenitor cells in the allogeneic host was associated with rapid differentiation from MHC class II negative to MHC class II positive, which is likely induced by the allogeneic host environment.
Differentiation of C12 RESCs
Vassilieva et al. , though demonstrating stable long-term cultivation of rat embryonic-like stem cells, did not describe any directed differentiation of rat embryonic stem cells towards specific cellular fates. We show here that RESCs can be programmed to differentiate into cells of endodermal and ectodermal origin under appropriate culture conditions.
Cellular differentiation by the vitamin A derivative RA has been studied with undifferentiated pluripotent embryonic carcinoma cells [30, 31] and mouse ES cells [32, 33]. In vivo, RA was identified as a morphogenic, teratogenic compound and, furthermore, as a signaling molecule regulating gene expression. Neuronal and glial differentiation of MESC has been described by Bain et al.  and Guan et al.  with standardized protocols. Nevertheless, no studies have been published so far that report on successful induction of neuronal differentiation in rat embryonic stem cells. We show here that treatment of C12 RESCs with RA can induce neuronal and glial differentiation after 7 days of culture. As part of the differentiation process, C12 RESCs changed their morphology and expressed several neuronal markers (MAP2 and synaptophysin), comparable with neurons derived from MESC and native neurons derived from the central nervous system of mouse and other mammalian species.
In contrast to the directed differentiation, which yields approximately 70% neuronal or glial cells, we observed that a fraction of cells within the cellular aggregates of EBs cultured in the absence of LIF expressed neuronal and glial markers. This is due to undirected spontaneous differentiation as described for human ES cells, which were shown to form EBs containing a mixture of differentiated and undifferentiated cells resembling early postimplantation embryos .
In embryonic carcinoma cells, neurons and glial cells developed simultaneously during RA treatment . This is consistent with our observations that during a culture period of 10–12 days in the presence of RA, C12 RESCs developed long, slender bipolar cells that were immunohistochemically identified as neurons and glial cells.
For a standard protocol of endothelial differentiation, mesodermal cells have to be isolated from EBs or embryos [36, 37], allowing spontaneous differentiation of other lineage-positive cells. Homogenous endothelial differentiation of ES cells was achieved after treatment with VEGF, erythropoietin, FGF-2, and IL-6 . For endothelial differentiation of C12 RESCs, we used Matrigel since it is known to induce tube formation in endothelial cells such as HUVEC [23, 37]. C12 RESCs underwent changes in morphology and antigenicity, suggesting that only the contact with Matrigel induced phenotypical changes in C12 RESCs. Culturing cells on Matrigel may thus represent a simple method for induction of endothelial differentiation.
Differentiation of RESCs into Hepatocyte-like Cells
We have presented evidence that RESCs can differentiate into an endodermal cell type with a hepatocyte phenotype. C12 RESCs cultured under hepatocyte differentiation conditions expressed primitive and mature hepatic genes as shown by immunohistochemistry and reverse transcription-PCR. The protein expression was hepatocyte specific and not spurious because nonhepatic markers were not coexpressed, and C12 RESCs cultured under standard conditions showed no or very low expression of hepatocyte-specific antigens. We extended results from immunohistochemistry by RT-PCR, which corroborated upregulation of early (αFP) and late (CFII, CPSI, TTR) markers of hepatocyte differentiation. Schwartz and colleagues  described hepatic differentiation of mouse, rat, and human multipotent adult progenitor cells from bone marrow by treatment with different growth factors including FGF, hepatocyte growth factor, basic FGF, FGF-4, FGF-α, and FGF-7, all of which are thought to be important in embryological hepatic development. We found that FGF-4 alone induced and promoted hepatic differentiation in RESCs. The fact that some late markers of hepatocyte differentiation (CK18, albumin, HNF-1α, and CYP2B1) were not upregulated in FGF-4-treated C12 cells could be explained by the rather short period of differentiation (21 days), whereas Schwartz et al. reported results from longer culture periods . Alternatively, additional differentiation signals might be required for upregulation of late hepatic markers.
In agreement with results from Schwartz et al. , hepatic differentiation in C12 RESCs was only observed when cells were grown on extracellular matrix components such as Matrigel or fibronectin.
Our observations suggest that C12 RESCs are a stable pluripotent rat embryonic stem cell-like cell line. These cells can be induced to differentiate into specific cell types if the signals that trigger the cell's inherent program for differentiation are provided.
This work was supported by a grant of the Interdisziplinäre Forschergruppe Transplantationsmedizin (IFTM) of the University of Kiel. We thank P. Boll, I. Berg, I. Schellhorn, P. Krüger, and M. Hauberg for their excellent technical support.