**3. Inducing specific differentiation of stem cells using with cell co-culture system**

In our research, we introduce an efficient cell co-culture system for in vitro differentiation of specific cell type from ES cells in a serum-free condition that requires neither EBs nor compli‐ cated treatment with growth factors.

#### **3.1. Induction of midbrain dopaminergic neurons from primate embryonic stem cells by coculture with Sertoli cells**

In the first study, we have established a new method for generating dopaminergic neurons from primate ES cells by co-culture with Sertoli cells. Neurodegenerative diseases present severe problems due to the limited repair capability of the nervous system [26]. Stem cells have a capacity for unlimited self-renewal, along with the ability to produce multiple different types of terminally differentiated descendants. They are candidate therapeutic tools in neurodege‐ nerative disorders, such as Parkinson's disease, which is characterized by degeneration and death of midbrain neurons that produce dopamine. Transplantation of dopaminergic neurons taken from human fetuses into Parkinson's disease patients shows a remarkable, but incon‐ sistent, ability to replace endogenous degenerated dopaminergic neurons and to ameliorate some of the disease symptoms [27, 28]. However, since treatment of a single Parkinson's disease patient requires dopamine neurons from 6 to 10 human fetuses, replacement therapy is not routinely available. Other sources of dopamine-producing cells, including those from the adrenal medulla or carotid bodies, have been examined for their ability to alleviate Parkinson's symptoms, but these sources are also limited in numbers and/or are not as effective as fetal dopamine neurons [29]. ES cells can proliferate indefinitely and are able to differentiate into cell types of all three germ layers in vivo and in vitro. These unique properties of ES cells make them an excellent candidate for the treatment of Parkinson's disease [30,31].

Our ES cell-Sertoli co-culture system is technically simple, cost-saving, and the induction is efficient and speedy. This protocol only used both of cells and treated without any inducing factors. After 3 weeks of induction, immuno-staining revealed that 90%±9% of the colonies contained tyrosine hydroxylase-positive (TH+ ) neurons, and 60% ±7% of the tubulin IIIpositive (Tuj III+ ) neurons were TH+ (Fig2. A-E). At the same time, the possible roles of Sertoli cell in the differentiation of ES cells into Dopaminergic Neurons were investigated. When cultured on the gelatin-coated dish in the same medium but without Sertoli cells, ES cells differentiated into neurons at a low frequency compared with the rate obtained with ES cells cultured on Sertoli cells (Fig2. F, lanes 1 and 2). This suggested that Sertoli cells had an active role in the promotion of neural differentiation of ES cells. We tested whether direct physical contact between ES cells and Sertoli cells was essential for the induction. ES cells cultured on gelatin-coated dishes and separated from co-cultured Sertoli cells by a 0.22-um filter membrane were still able to induce significant neural differentiation of ES cells (Fig2. F, lane3; Fig2.H). When Sertoli-conditioned medium was used to culture ES cells without Sertoli cells as a feeder layer, neural differentiation was still induced compared with ES cells cultured alone on gelatincoated dishes (Fig2. F, lane4; Fig2.I)). Together these data suggest that Sertoli cells secrete one or more soluble neural-inducing factors.

Based on the known ability to promote neuronal differentiation [32], GDNF seemed to be a likely candidate for the Sertoli-induced differentiation of primate ES cells. We confirmed the presence of GDNF expression in Sertoli cells by immunohistochemistry (Fig. 2 J, Fig.2 K lane 1) and in the conditioned medium by Western blot analysis (Fig.2 K, lane 2). We therefore tested whether GDNF promoted dopaminergic neuron differentiation of primate ES cells. First, we added GDNF to ES/Sertoli cell co-cultures and compared results with ES cells on a gelatin

**3.1. Induction of midbrain dopaminergic neurons from primate embryonic stem cells by co-**

In the first study, we have established a new method for generating dopaminergic neurons from primate ES cells by co-culture with Sertoli cells. Neurodegenerative diseases present severe problems due to the limited repair capability of the nervous system [26]. Stem cells have a capacity for unlimited self-renewal, along with the ability to produce multiple different types of terminally differentiated descendants. They are candidate therapeutic tools in neurodege‐ nerative disorders, such as Parkinson's disease, which is characterized by degeneration and death of midbrain neurons that produce dopamine. Transplantation of dopaminergic neurons taken from human fetuses into Parkinson's disease patients shows a remarkable, but incon‐ sistent, ability to replace endogenous degenerated dopaminergic neurons and to ameliorate some of the disease symptoms [27, 28]. However, since treatment of a single Parkinson's disease patient requires dopamine neurons from 6 to 10 human fetuses, replacement therapy is not routinely available. Other sources of dopamine-producing cells, including those from the adrenal medulla or carotid bodies, have been examined for their ability to alleviate Parkinson's symptoms, but these sources are also limited in numbers and/or are not as effective as fetal dopamine neurons [29]. ES cells can proliferate indefinitely and are able to differentiate into cell types of all three germ layers in vivo and in vitro. These unique properties of ES cells

make them an excellent candidate for the treatment of Parkinson's disease [30,31].

contained tyrosine hydroxylase-positive (TH+

or more soluble neural-inducing factors.

) neurons were TH+

positive (Tuj III+

Our ES cell-Sertoli co-culture system is technically simple, cost-saving, and the induction is efficient and speedy. This protocol only used both of cells and treated without any inducing factors. After 3 weeks of induction, immuno-staining revealed that 90%±9% of the colonies

cell in the differentiation of ES cells into Dopaminergic Neurons were investigated. When cultured on the gelatin-coated dish in the same medium but without Sertoli cells, ES cells differentiated into neurons at a low frequency compared with the rate obtained with ES cells cultured on Sertoli cells (Fig2. F, lanes 1 and 2). This suggested that Sertoli cells had an active role in the promotion of neural differentiation of ES cells. We tested whether direct physical contact between ES cells and Sertoli cells was essential for the induction. ES cells cultured on gelatin-coated dishes and separated from co-cultured Sertoli cells by a 0.22-um filter membrane were still able to induce significant neural differentiation of ES cells (Fig2. F, lane3; Fig2.H). When Sertoli-conditioned medium was used to culture ES cells without Sertoli cells as a feeder layer, neural differentiation was still induced compared with ES cells cultured alone on gelatincoated dishes (Fig2. F, lane4; Fig2.I)). Together these data suggest that Sertoli cells secrete one

Based on the known ability to promote neuronal differentiation [32], GDNF seemed to be a likely candidate for the Sertoli-induced differentiation of primate ES cells. We confirmed the presence of GDNF expression in Sertoli cells by immunohistochemistry (Fig. 2 J, Fig.2 K lane 1) and in the conditioned medium by Western blot analysis (Fig.2 K, lane 2). We therefore tested whether GDNF promoted dopaminergic neuron differentiation of primate ES cells. First, we added GDNF to ES/Sertoli cell co-cultures and compared results with ES cells on a gelatin

) neurons, and 60% ±7% of the tubulin III-

(Fig2. A-E). At the same time, the possible roles of Sertoli

**culture with Sertoli cells**

122 Regenerative Medicine and Tissue Engineering

**Figure 2. A-E.** Sertoli-induced dopaminergic neural differentiation and of monkey embryonic stem cells. Characteriza‐ tion of monkey ES colony induced by Sertoli cells. Expression of NCAM (A), NeuN (B) confirmed the neural identity of cells in ES colony. After 3 weeks, approximately 97% of the ES cell colonies cultured with Sertoli cells were Tuj III+ **(D).** After 3 weeks of culture, immunostaining of Sertoli-induced neurons with anti-TH antibody **(E). (C):** Electron microscopy showed TH ImmunoGold particles (15nm) were associated with small vesicles presumably containing neurotransmitters located at the presynaptic terminal (arrow). **F-I.** The physical and chemical role of Sertoli cells in the induction of dopaminergic neurons. **(F):** Sertoli cells induced the expression of neuronal marker Tuj III in co-cultured monkey embryonic stem (ES) cells even when separated by a filter membrane. CM induced neural differentiation in ES cells cultured on gelatin-coated dish. However the differentiation rate was lower than with direct coculture with Sertoli cells. Tuj III (green) and TH (red) double staining of ES cells when separated from Sertoli cells by a filter **(H)** and cultured in conditioned medium alone **(I).** Scale bar = 50um. **J-O.** The role of GDNF in Sertoli-induced dopaminergic neuron differentiation**. (J):** Immunostaining of GDNF in Sertoli cells. **(K):** GDNF protein was detected by Western blot of lysed Sertoli cells (lane 1) and in conditioned me‐ dium (lane 2). However, it was not detected in conditioned medium blocked by anti-GDNF antibody (lane 3). **(L):** In em‐ bryonic stem (ES) cells cocultured with conditioned medium derived from Sertoli cells, the TH+ cell percentage was reduced significantly when GDNF was blocked by antibody. TH staining of ES cells cultured on Sertoli cells **(M),** on Sertoli cells treated with extraneous GDNF **(N),** and on blocked conditioned medium **(O).** Scale bars =50um.

coated substrate after 3 weeks in culture. We found that GDNF increased the number of TH+ cells in co-culture with Sertoli (Fig. 2 L, N) At the same time, we incubated the conditioned medium with GDNF-blocking antibodies and effectively removed GDNF as determined by Western blot (Fig. 2K, lane3). The number of TH+ cells supported by anti-GDNF-treated conditioned medium was reduced to 35% ±6% (*n*= 6,000; Fig. 2L), which was significantly less than that induced by Sertoli cells (\**p*<.05; Fig. 2L), but still more than in colonies grown on gelatin alone. GDNF antibody inhibited the TH+ neural induction activity of Sertoli cells, as shown by TH staining of ES/Sertoli cell co-cultures and ES cells cultured in conditioned medium blocked by GDNF antibody (Fig. 2 M-O).

Possible roles of Sertoli sells in the differentiation of ES cells into dopaminergic neurons were discussed. Sertoli cells have an active role in the promotion of neural differentiation of ES cells. The mechanism of dopaminergic neuron induction in co-cultures of ES cells with Sertoli cells remains to be understood. First, whether direct physical contact between ES cells and Sertoli cells was essential for the induction was tested through filter membrane to separate ES cells with Sertoli cells. The result showed that Sertoli cells were still able to induce significant neural differentiation of ES cells, indicating that Sertoli cells produce soluble inducing factors. However, Sertoli-conditioned medium could not elicit significant induction. It suggested two possibilities as to the molecular nature of neuron-inducing activity by Sertoli cells. One is that Sertoli cells secrete two different neuron inducing factors, a cell surface-anchored factor and a labile soluble factor. Another might be that the neuron-inducing activity is mediated by secreted factors. At present, we cannot exclude either possibility.

Some factors have been implicated in the regulation of dopaminergic differentiation [33], such as FGF8, Shh, interleukin (IL) 1, IL11, GDNF, and neutralizing antibodies of FGF8 and Shh. Among them, GDNF has the most potent neuroprotective and trophic effects on dopamine neurons in many model systems [34,35,36]. However, it is a large protein and has to be delivered directly to the brain rather than given peripherally. When successfully delivered, GDNF supports the survival and outgrowth of dopamine neurons following transplantation [37]. In addition, GDNF added to cell suspensions of embryonic ventral mesencephalic tissue improves the survival of dopamine neurons following grafting into the degenerative striatum [38]. Other studies have shown that intermittent injections of GDNF in the vicinity of intra‐ striatal nigral cell suspension grafts have similar effects on improving the survival and/or fiber outgrowth of transplanted dopamine neurons [39,40]. In a previous study [41], neurospheres modified to produce GDNF increased the survival of transplanted dopamine neurons in 6- OHDA-lesioned animals. GDNF is also capable of promoting differentiation of mesencephalic neurospheres towards the neuronal lineage, and more importantly, towards the dopaminergic development indicated by expression of NurrI and Ptx3. Buytaert-Hoefen et al. [42] proved that significant differentiation of dopaminergic neurons were not induced when cultured on PA6 stromal cells alone except for the presence of GDNF or striatal astrocytes. Sertoli cells secrete GDNF and promote the survival of transplanted dopaminergic neurons. In the present study, we showed that GDNF plays a role in dopaminergic neuron differentiation when primate ES cells were co-cultured with Sertoli cells. In our study, extraneous GDNF induced the differentiation of dopaminergic neuron. Sertoli-induced neural differentiation of ES cells was partly suppressed by a low-dose of anti-GDNF antibody. However, blocking GDNF did not completely inhibit the neural differentiation. We cannot be certain that GDNF activity was completely blocked, although it was not found by Western blot. Therefore, the diminished neuronal differentiation that occurred could have been in response to the remaining, un‐ blocked GDNF. Alternatively, if all of GDNF was blocked, then one or more other factors were present in the conditioned medium and promoted differentiation at a reduced rate. If these factors exist, they are more effective in the presence of GDNF. This confirmed the conclusion of Buytaert-Hoefen et al. [42] that GDNF is required but not necessary for the induction of dopaminergic neurons.

coated substrate after 3 weeks in culture. We found that GDNF increased the number of TH+ cells in co-culture with Sertoli (Fig. 2 L, N) At the same time, we incubated the conditioned medium with GDNF-blocking antibodies and effectively removed GDNF as determined by

conditioned medium was reduced to 35% ±6% (*n*= 6,000; Fig. 2L), which was significantly less than that induced by Sertoli cells (\**p*<.05; Fig. 2L), but still more than in colonies grown on

shown by TH staining of ES/Sertoli cell co-cultures and ES cells cultured in conditioned

Possible roles of Sertoli sells in the differentiation of ES cells into dopaminergic neurons were discussed. Sertoli cells have an active role in the promotion of neural differentiation of ES cells. The mechanism of dopaminergic neuron induction in co-cultures of ES cells with Sertoli cells remains to be understood. First, whether direct physical contact between ES cells and Sertoli cells was essential for the induction was tested through filter membrane to separate ES cells with Sertoli cells. The result showed that Sertoli cells were still able to induce significant neural differentiation of ES cells, indicating that Sertoli cells produce soluble inducing factors. However, Sertoli-conditioned medium could not elicit significant induction. It suggested two possibilities as to the molecular nature of neuron-inducing activity by Sertoli cells. One is that Sertoli cells secrete two different neuron inducing factors, a cell surface-anchored factor and a labile soluble factor. Another might be that the neuron-inducing activity is mediated by

Some factors have been implicated in the regulation of dopaminergic differentiation [33], such as FGF8, Shh, interleukin (IL) 1, IL11, GDNF, and neutralizing antibodies of FGF8 and Shh. Among them, GDNF has the most potent neuroprotective and trophic effects on dopamine neurons in many model systems [34,35,36]. However, it is a large protein and has to be delivered directly to the brain rather than given peripherally. When successfully delivered, GDNF supports the survival and outgrowth of dopamine neurons following transplantation [37]. In addition, GDNF added to cell suspensions of embryonic ventral mesencephalic tissue improves the survival of dopamine neurons following grafting into the degenerative striatum [38]. Other studies have shown that intermittent injections of GDNF in the vicinity of intra‐ striatal nigral cell suspension grafts have similar effects on improving the survival and/or fiber outgrowth of transplanted dopamine neurons [39,40]. In a previous study [41], neurospheres modified to produce GDNF increased the survival of transplanted dopamine neurons in 6- OHDA-lesioned animals. GDNF is also capable of promoting differentiation of mesencephalic neurospheres towards the neuronal lineage, and more importantly, towards the dopaminergic development indicated by expression of NurrI and Ptx3. Buytaert-Hoefen et al. [42] proved that significant differentiation of dopaminergic neurons were not induced when cultured on PA6 stromal cells alone except for the presence of GDNF or striatal astrocytes. Sertoli cells secrete GDNF and promote the survival of transplanted dopaminergic neurons. In the present study, we showed that GDNF plays a role in dopaminergic neuron differentiation when primate ES cells were co-cultured with Sertoli cells. In our study, extraneous GDNF induced the differentiation of dopaminergic neuron. Sertoli-induced neural differentiation of ES cells

cells supported by anti-GDNF-treated

neural induction activity of Sertoli cells, as

Western blot (Fig. 2K, lane3). The number of TH+

124 Regenerative Medicine and Tissue Engineering

gelatin alone. GDNF antibody inhibited the TH+

medium blocked by GDNF antibody (Fig. 2 M-O).

secreted factors. At present, we cannot exclude either possibility.

From these results, we can conclude that Sertoli cells may stimulate dopaminergic differen‐ tiation by a complex combination of growth factors or other factors, including other uniden‐ tified components. Among these, GDNF plays some role, but not a decisive role. It is also possible that the supporting environment provided by Sertoli cells, or an interaction between Sertoli cells and primate ES cells, plays a role in their neuron inducing activity.

In addition, isolated Sertoli cells enable survival and function of co-grafted foreign dopami‐ nergic neurons in rodent models of Parkinson's disease. They also promote regeneration of damaged striatal dopaminergic circuitry in those same Parkinson's disease models [43]. In our study, significant TH+ cells were found in the degenerative striatum when differentiated primate ES cells were co-transplanted with Sertoli cells. Moreover, 2-month survival of TH+ neurons derived from ES cells was observed. It is likely that the nutritive support of the Sertoli cells is responsible for this enhanced TH cell survival.

In common, dopaminergic neurons were generated from neural precursor cells amplified from EBs. One group [44] used a lengthy four-step method to produce TH+ neurons at an efficiency of approximately 7% of the Tuj III+ neurons. Kawasaki et al. [45] used the term stromal cellderived inducing activity to describe the dopaminergic neuronal-inducing defect of PA6 cells. The stromal feeder-mediated neural induction had been demonstrated for mouse, primate, and human ES cells [46-48]. In contrast to the previous protocols, our methods does not require growth serum, the formation of EBs, selection of neural precursor cells, retinoic acid, growth factors, or other special treatment. Thus, our method is more suited to detailed analyses of differentiation and transplantation therapy.

#### **3.2. Differentiation of primate ES cells into retinal cells induced by ES cell-derived pigmented cells**

In the second study, retinal cells were induced from primate ES cells by co-culturing with ESderived pigmented cells.

Photoreceptors are the primary sensory neurons residing in the outer nuclear layer (ONL) of the vertebrate retina. Photoreceptor degeneration is a common cause of human visual impairments resulting from light damage, genetic changes and aging. The unfortunate nonrenewable nature of photoreceptors has inspired that these degenerative diseases maybe treatable by transplantation of healthy fetal cells. Previous work has shown that freshly harvested retinal pigment epithelium (RPE) can be effective in rescuing photoreceptors in the Royal College of Surgeons (RCS) rat, an animal model of indirect photoreceptor degeneration [49]. However, obtaining a sufficient number of suitable donor cells remains a problem.

The isolation of embryonic stem (ES) cells has drawn much attention, given their potential to generate all adult cell types. As ES cells continue to proliferate in an undifferentiated state in vitro, an unlimited stem cell source or its derivatives may be secured. It is also a potential benefit that ES cells may be genetically manipulated to permit the selective differentiation and/ or isolation of a specific cell type.

Recently, several laboratories have devised hetero chronic co-culture experiments to charac‐ terize the nature of the interactions necessary for the differentiation of retinal cells [50,51]. Early embryonic mouse cells (E12) were shown to differentiate into rod photoreceptor cells with a higher frequency when co-cultured with either postnatal rat retinal cells or with later staged embryonic retinal cells that were competent to generate rod photoreceptors. The results suggested that the differentiation of retinal progenitor cells as rod photoreceptors is influenced by locally diffusible signals in the extra cellular environment that are developmentally regulated during the period of retinal histogenesis.

Although the specific factors that promote differentiation of cells as rod photoreceptors are not known, several classes of molecules have been shown to play a role in the developing retina, for example, small peptide growth factors, Shh, taurine, epidermal growth factor (EGF) and fibroblast growth factor (FGF) [52-54]. The vitamin A derivative retinoic acid has been shown to influence cell fate in the developing nervous system. In particular, all-trans retinoic acid has been shown to alter cell fate decisions in the developing limb bud, hindbrain and inner ear [55,56]. In addition, several studies have demonstrated that all-trans retinoic acid and at least one of the nuclear retinoic acid receptors (RARa) are present in the developing retina (56). RA caused a dose-dependent, specific increase in the number of cells that developed as photoreceptors in culture throughout the period of retinal neurogenesis [52,53].

Retinal pigmented epithelium (RPE) is a neuroectodermal derivative essential for the survival of photoreceptors. It supplies nutrition and provides several trophic factors that help maintain the normal physiology within the neuro sensory retina and photoreceptors [57,58].

In most of studies, differentiation of the retinal progenitors into photoreceptors was infre‐ quent in the absence of co-culture with embryonic retinal tissues [59,60]. Use of the fetal retina involves some ethical and practical consideration, and the cell supply is a problem. Osakada and colleagues [52,53]induced the generation of putative rod and cone photoreceptors from ES cells by stepwise treatments under defined culture conditions. However, the practical steps were complicated and time consuming. In this study, the monolayer of RPE derived from ES cells wasusedas an inducing feederlayer which couldreplicate themicroenvironment ofretina or sub-retinal to successfully induce photoreceptor cells. This is the first time that ES-derived cells are used to induce the differentiation of ES cells. Furthermore, retinoic acid was used to induce the differentiation of photoreceptors. To our knowledge, a few reports have produced photoreceptor cells in uncontaminated culture conditions except for Takahashi's group [52] Osakada [53] and lamba's [60,61,62,] research. The present study demonstrated the genera‐ tionofphotoreceptorsfromprimateEScellsunderconditionsfreeofanimal–derivedsubstances. During the study of ESC-Sertoli co-culture system, one unexpected finding was the appearance of epithelial cells with massive pigmentation from the second week and they grew at a constant rate. After monkey ES cells were cultured on Sertoli cells for 3 weeks, a patch of pigmented cells was mechanically isolated by using a tip after being loosened with trypsin and plated on a gelatin-coated dish without feeder cells in primate ES cell culture medium. The polygonal morphology with a compact cell-cell arrangement was reminiscent of the pigmented epithe‐ lium of the eye, and clearly distinct from pigmented melanocytes derived from neural crest (Fig3 A-I). In recent, transplantation of retinal pigment epithelium (RPE) has become a possible therapeutic approach for retinal degeneration. Meanwhile, RPE is a neural ectodermal derivative essential for the survival of photoreceptors. It serves as nutritional cells and provides several trophic factors that help maintain the normal physiology within the neural sensory retina and photoreceptors. Therefore, we investigated whether retinal cells could be induced by co-culturing ES cells together with RPE. At the same time, RA was treated in order to improve the differentiation efficiency. The results showed that after 10-day co-culture of ES cells and these RPE, some ES derivatives became immuno-positive for rhodopsin. RT-PCR analysis demonstrated expression of retina-related gene markers such as Pax6, CRX, IRBP, Rhodopsin, Rhodopsin kinase and Muschx10A. When RA was added, the distinct increase of photoreceptor specific proteins markers was found. Besides, the differentiation of bipolar, horizontal cells was demonstrated by protein and gene expression. RA treatment also altered more retinal cell differentiation (Fig3. J-M).

Royal College of Surgeons (RCS) rat, an animal model of indirect photoreceptor degeneration [49]. However, obtaining a sufficient number of suitable donor cells remains a problem.

The isolation of embryonic stem (ES) cells has drawn much attention, given their potential to generate all adult cell types. As ES cells continue to proliferate in an undifferentiated state in vitro, an unlimited stem cell source or its derivatives may be secured. It is also a potential benefit that ES cells may be genetically manipulated to permit the selective differentiation and/

Recently, several laboratories have devised hetero chronic co-culture experiments to charac‐ terize the nature of the interactions necessary for the differentiation of retinal cells [50,51]. Early embryonic mouse cells (E12) were shown to differentiate into rod photoreceptor cells with a higher frequency when co-cultured with either postnatal rat retinal cells or with later staged embryonic retinal cells that were competent to generate rod photoreceptors. The results suggested that the differentiation of retinal progenitor cells as rod photoreceptors is influenced by locally diffusible signals in the extra cellular environment that are developmentally

Although the specific factors that promote differentiation of cells as rod photoreceptors are not known, several classes of molecules have been shown to play a role in the developing retina, for example, small peptide growth factors, Shh, taurine, epidermal growth factor (EGF) and fibroblast growth factor (FGF) [52-54]. The vitamin A derivative retinoic acid has been shown to influence cell fate in the developing nervous system. In particular, all-trans retinoic acid has been shown to alter cell fate decisions in the developing limb bud, hindbrain and inner ear [55,56]. In addition, several studies have demonstrated that all-trans retinoic acid and at least one of the nuclear retinoic acid receptors (RARa) are present in the developing retina (56). RA caused a dose-dependent, specific increase in the number of cells that developed as

Retinal pigmented epithelium (RPE) is a neuroectodermal derivative essential for the survival of photoreceptors. It supplies nutrition and provides several trophic factors that help maintain

In most of studies, differentiation of the retinal progenitors into photoreceptors was infre‐ quent in the absence of co-culture with embryonic retinal tissues [59,60]. Use of the fetal retina involves some ethical and practical consideration, and the cell supply is a problem. Osakada and colleagues [52,53]induced the generation of putative rod and cone photoreceptors from ES cells by stepwise treatments under defined culture conditions. However, the practical steps were complicated and time consuming. In this study, the monolayer of RPE derived from ES cells wasusedas an inducing feederlayer which couldreplicate themicroenvironment ofretina or sub-retinal to successfully induce photoreceptor cells. This is the first time that ES-derived cells are used to induce the differentiation of ES cells. Furthermore, retinoic acid was used to induce the differentiation of photoreceptors. To our knowledge, a few reports have produced photoreceptor cells in uncontaminated culture conditions except for Takahashi's group [52] Osakada [53] and lamba's [60,61,62,] research. The present study demonstrated the genera‐ tionofphotoreceptorsfromprimateEScellsunderconditionsfreeofanimal–derivedsubstances.

photoreceptors in culture throughout the period of retinal neurogenesis [52,53].

the normal physiology within the neuro sensory retina and photoreceptors [57,58].

or isolation of a specific cell type.

126 Regenerative Medicine and Tissue Engineering

regulated during the period of retinal histogenesis.

We demonstrated that pigment epithelial cells can be generated, enriched, and expanded from primate ES cells when ES cells were co-cultured with Sertoli cells. These ES-converted pigment epithelial cells showed development of several of the characteristics of RPE cells and were able to be used to induce the differentiation of retinal cells. For the clinical application, methods for purifying large numbers of lineage specific cells should be developed. In the present study, RPE cells could be easily identified under a dissecting microscope and selectively expanded into a uniform single cell layer.

The RPE cells contain melanin granules, have a characteristic polygonal morphology, and play some critical roles including (i) forming a barrier separating the retina from the blood vessels of the choroid coat; (ii) regulating nutritive substance transport required for retinal progenitors to differentiate into retinal neurons and the maintenance of retinal cells; (iii) regenerating visual pigments; (iv) digesting the shed parts of photoreceptor cells after having phagocytized them [63]. Therefore, we used ES cell- derived RPE cells to provide epigenetic retinal neurons, not only the efficient induction of photoreceptors, but also other retinal cell lineages, such as bipolar, and horizontal cells. It indicated that the secreted/diffusible factors from RPE or direct cell-cell contact were sufficient to induce retinal cells from ES cells.

A monolayer of ESC-derived RPE cells, an inducing feeder layer that replicates the retinal microenvironment and RA were used to successfully induce photoreceptor differentiation. To our knowledge, this is the first time that ESC-derived cells have been used to induce ESC differentiation. ES/RPE co-culture system can serve as a promising method for therapeutic application and basic research on retinal degeneration disease, although we are still far from an established in vitro or in vivo source of retinal cells.

**Figure 3.** A-I. Analysis of pigmented epithelial cells derived from primate ES cells using LM (A-E), TEM (F), RT-PCR (G), immunofluorescence staining (H-I). (A) LM images showing pigmented cells present in the primate ESC colonies grown at a constant rate on Sertoli feeder cells for 2, 4, and 6 weeks (A,B and C, respectively). The polygonal morphol‐ ogy with a compact cell–cell arrangement (D) is similar to the pigmented epithelium of the eye and clearly distinct from pigmented melanocytes derived from the neural crest (E). (F) TEM images of primate ESC-derived pigmented cells displaying melanin granules. (G) Shown are the RT-PCR-amplified products of RPE cell-specific markers (C1): tyro‐ sinase-related protein-2(Trp-2), RPE65, CRALBP, MertK. (H-I) Immunostaining of RPE cells derived from ES cells with

anti-Pax6 and RPE65 antibody is also shown. J-M. Analysis of differentiated retinal cells induced by ESC-derived RPE cells by immunofluorescence staining. Immunopositive ESC derivatives for retinal cell-specific markers are shown as follows: (J) ES cells were immunopositive for rhodopsin after a 21-day co-culture with RPE cells, and an increased fre‐ quency of rhodopsin-positive cells was estimated after RA treatment, and some of them developed into rosettes. The eyes of a db/db mouse (2 weeks) were used as a positive control to identify the positive immunofluorescence staining results and the specificity of retinal antibodies. (K) PKC-positive cells appeared as early as day 3 after starting co-cul‐ tures. A drastic increase was observed after RA treatment. Positive control with mouse eyes and PKC antibody is also shown. (L) Recoverin-positive cells appeared after a 10-day o-culture with RPE cells. Expression increased at a constant rate and after RA treatment. Positive control with mouse eyes and recoverin antibody is also shown. (D) After a 10-day co-culture with RPE cells, ES cells became immunopositive for calbindin. Immunopositive ESC derivatives for calbindin are shown at day 2. The number of calbindin-positive cells increased after RA treatment (D2). Positive control with mouse eyes and calbindin antibody is also shown. Scale bars = 50 lm.

#### **3.3. Bone marrow stromal cells as an inducer for cardiomyocyte differentiation from mouse embryonic stem cells**

In the third study, bone marrow stromal cells (BMSCs) were used as an inducer to induce cardiomyocyte differentiation from mouse ES cells.

Several studies reported that different feeder layers induced cardiomyocytes from ES cells [64-67]. It proved that the visceral-endoderm-like cell line, END-2 induced mouse P19EC, mouse and human ES cells to aggregate in co-culture and give rise to cultures containing beating areas. For mouse P19 EC cells, it has been demonstrated that a diffusible factor secreted by the END-2 cells is responsible for the induction of cardiomyocyte formation.

Like END-2 cells, various cell types of stem cells remain in a mature body. Among them, bone marrow stem cells (BMSCs) are unique because of rich functional products. A wide array of cytokines including vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF), and insulin growth factor-1 (IGF-1) were detected in the BMSC cultured medium by ELISA [68]. Furthermore, they secreted hepatocyte growth factor (HGF), and transforming growth factors (TGF-beta) [69], which are all potent cardiomyo‐ cyte growth and survival factors, or play an important role in proliferation and differentia‐ tion of stem cells [70]. IGF-1 can promote angiogenesis in infracted myocardium, reduce the degree of myocardial necrosis, maintain the myocardial structure, stimulate prolifera‐ tion of cardiac fibroblast, and inhibit matrix degradation; thus preventing ventricular dilatation and reducing load capacity of the heart [71]. After binding to its receptor on ES cells, IGF-1 induces expression of a number of cardiac-specific transcription factors such as the zinc finger GATA proteins and Nkx-2.5, a co-activator of GATA-4. GATA- 4 and Nkx-2.5 are essential for heart development [72]. HGF is a multifunctional factor promoting cell mitosis. It can promote cell survival and regeneration, inhibit the apoptosis of stem cells, and increase the survival rate of transplanted cells. More importantly, HGF enhances growth and favors cell extracellular matrix interactions; the critical steps during myocardial regeneration after infarction. In the heart, high concentration and wide distribution of FGF has been identified from early embryonic stages. Basic FGF plays a vital role in the growth and differentiation of cardiac myocytes. FGF was found to induce DNA synthesis in adult rat myocytes, cells which were considered capable of regeneration [73]. Members of the

**Figure 3.** A-I. Analysis of pigmented epithelial cells derived from primate ES cells using LM (A-E), TEM (F), RT-PCR (G), immunofluorescence staining (H-I). (A) LM images showing pigmented cells present in the primate ESC colonies grown at a constant rate on Sertoli feeder cells for 2, 4, and 6 weeks (A,B and C, respectively). The polygonal morphol‐ ogy with a compact cell–cell arrangement (D) is similar to the pigmented epithelium of the eye and clearly distinct from pigmented melanocytes derived from the neural crest (E). (F) TEM images of primate ESC-derived pigmented cells displaying melanin granules. (G) Shown are the RT-PCR-amplified products of RPE cell-specific markers (C1): tyro‐ sinase-related protein-2(Trp-2), RPE65, CRALBP, MertK. (H-I) Immunostaining of RPE cells derived from ES cells with

128 Regenerative Medicine and Tissue Engineering

TGF-b superfamily play important roles in cardiac development during embryogenesis [74] as well as in various cardiac pathologies [75]. TGF-b1 has been shown to induce cardiac differentiation in vitro in embryonic explants and stem cells [76,77] as well as in adult bone marrow-derived cells [70]. Previous studies of non-conditional global VEGF or VEGF isoform gene deletion have demonstrated embryonic and early postnatal mortality associat‐ ed with severe cardiac abnormalities and abnormal vascularization; thus establishing that systemic VEGF expression is essential normal myocardial development [78]. From these reports BMSCs are expected to function as inducer for cardiac differentiation from ES cells.

To confirm whether BMSCs are a real cardiac inducer to differentiate ES cells into cardio‐ myocytes, mouse embryoid bodies (EBs) were co-cultured with rat BMSCs. After about 10 days, areas of rhythmically contracting cells in more solid aggregates became evident with bundle-like structures formed along borders between EB outgrowth and BMSC layer. ESCderived cardiomyocytes exhibited sarcomeric striations when stained with troponin I (Trop I), organized in separated bundles. Besides, the staining for connexin 43 was detected in cell–cell junctions, which demonstrated that ESC-derived cardiomyocytes were coupled by gap junction in culture (Fig4 A-F). In addition, an improved efficiency of cardiomyocyte differentiation from ESC-BMSC co-culture was found in the serum-free medium: 5-fold upregulation in the number of beating area compared with the serum medium. Effective cardiac differentiation was also recognized in transfer filter assay and in condition medium obtained from BMSC culture. A clear increase in the expression of cardiac genes and TropI protein confirmed further cardiac differentiation by BMP4 and Retinoic Acid (RA) treatment (Fig4 G,H). These results demonstrate that BMSCs can induce cardiomyocyte differentiation from ES cells through soluble factors and enhance it with BMP4 or RA treatment. Serum-free ESC-BMSC co-culture represents a defined in vitro model for identifying the cardiomyocyteinducing activity from BMSCs and, in addition, a straightforward experimental system for assessing clinical applications.

In our study, two experiments of separation from BMSCs by the filter and using CM without contact of BMSCs were performed. These results showed that both conditions were still able to induce significant cardiac differentiation (Fig4 I-M). As expected previously, these data have shown that it is due to varied soluble factors secreted by BMSCs, though effective soluble factors have to be determined in future. On the other hand, cardiomyocyte bundles were formed long the edge of EB outgrowth, which was a borderline with BMSC layer. These regular arrangements seemed to be associated with BMSC layer structures. Moreover, in fact, direct cell-to-cell interaction between BMSC and EBs was more effective to induce cardiac differen‐ tiation. It may be possible that the supporting environment provided by BMSCs or an interaction between BMSCs and EBs may play a role in their cardiac-inducing activity.

Compared with the inducing effects of low dose cytokines alone or with BMSCs alone, our coculture with BMSCs supplemented with cytokines could increase the differentiation of cardiomyocytes, which indicated that BMSCs had the ability to promote the induction and proliferation of ESC-derived cardiomyocyte and the addition of low dose cytokines had a synergistic effect on this ability.

TGF-b superfamily play important roles in cardiac development during embryogenesis [74] as well as in various cardiac pathologies [75]. TGF-b1 has been shown to induce cardiac differentiation in vitro in embryonic explants and stem cells [76,77] as well as in adult bone marrow-derived cells [70]. Previous studies of non-conditional global VEGF or VEGF isoform gene deletion have demonstrated embryonic and early postnatal mortality associat‐ ed with severe cardiac abnormalities and abnormal vascularization; thus establishing that systemic VEGF expression is essential normal myocardial development [78]. From these reports BMSCs are expected to function as inducer for cardiac differentiation from ES cells.

To confirm whether BMSCs are a real cardiac inducer to differentiate ES cells into cardio‐ myocytes, mouse embryoid bodies (EBs) were co-cultured with rat BMSCs. After about 10 days, areas of rhythmically contracting cells in more solid aggregates became evident with bundle-like structures formed along borders between EB outgrowth and BMSC layer. ESCderived cardiomyocytes exhibited sarcomeric striations when stained with troponin I (Trop I), organized in separated bundles. Besides, the staining for connexin 43 was detected in cell–cell junctions, which demonstrated that ESC-derived cardiomyocytes were coupled by gap junction in culture (Fig4 A-F). In addition, an improved efficiency of cardiomyocyte differentiation from ESC-BMSC co-culture was found in the serum-free medium: 5-fold upregulation in the number of beating area compared with the serum medium. Effective cardiac differentiation was also recognized in transfer filter assay and in condition medium obtained from BMSC culture. A clear increase in the expression of cardiac genes and TropI protein confirmed further cardiac differentiation by BMP4 and Retinoic Acid (RA) treatment (Fig4 G,H). These results demonstrate that BMSCs can induce cardiomyocyte differentiation from ES cells through soluble factors and enhance it with BMP4 or RA treatment. Serum-free ESC-BMSC co-culture represents a defined in vitro model for identifying the cardiomyocyteinducing activity from BMSCs and, in addition, a straightforward experimental system for

In our study, two experiments of separation from BMSCs by the filter and using CM without contact of BMSCs were performed. These results showed that both conditions were still able to induce significant cardiac differentiation (Fig4 I-M). As expected previously, these data have shown that it is due to varied soluble factors secreted by BMSCs, though effective soluble factors have to be determined in future. On the other hand, cardiomyocyte bundles were formed long the edge of EB outgrowth, which was a borderline with BMSC layer. These regular arrangements seemed to be associated with BMSC layer structures. Moreover, in fact, direct cell-to-cell interaction between BMSC and EBs was more effective to induce cardiac differen‐ tiation. It may be possible that the supporting environment provided by BMSCs or an interaction between BMSCs and EBs may play a role in their cardiac-inducing activity.

Compared with the inducing effects of low dose cytokines alone or with BMSCs alone, our coculture with BMSCs supplemented with cytokines could increase the differentiation of cardiomyocytes, which indicated that BMSCs had the ability to promote the induction and proliferation of ESC-derived cardiomyocyte and the addition of low dose cytokines had a

assessing clinical applications.

130 Regenerative Medicine and Tissue Engineering

synergistic effect on this ability.

**Figure 4.** Characterization of cardiomyocyte induced by BMSC co-culture. (A,B) Morphology of serum or non-serum treated EB outgrowth on BMSC layer on day 14. (A) 20% FBS: Multi-angular and flattened cells spread out from EBs. (B) Absence of FBS: fiber-like cells extended radially and formed bundle-like structures each other. Scale Bars = 25um. (C-F) Beating areas stained for Trop I (green) and CX43 (Red). (C) In EBs formed in the presence of FBS, cardiomyocyte fibers are distributed disorderly. (D) Cardiomyocyte fibers are organized regular in EBs formed in the absent of FBS. Radial-morpho‐ logical-like fibers surrounded or spread from EBs. Beating cells showed spindle, round and tri or multi-angular morpholo‐ gieswith characteristic organized in separated bundles. (E) CX43 staining among TropI-positive cardiomyocytes showed the presence of gap junctions. (F) The heart of C57BL/6 mice was used as a positive control to identify the positive immu‐ nofluorescence staining results and the specificity of retinal antibodies. Bars = 100um. **G-H** Examination of expression changes of cardiomyocytes from EBs after BMP4 and RA treatment. (G) Induction of cardiac-specific genes and BMP sig‐ naling molecules in EBs exposed to BMP4, as shown by RT-PCR. Lane1: undifferentiated ES cells; lane2: EB cultured on gel‐ atin-coated dish, FBS(−); lane3: EB co-cultured with BMSCs, FBS(−); lane4: EB co-cultured with BMSCs, FBS(−), and treated with BMP4. The expression of cardiac markers and BMP signals were increased markedly by BMP4 treatment. TropI pro‐ tein was also detected by Western of lysed EB on gelatin-coated dish (lane1), EB on gelatin-coated dish and treated with BMP4 (lane2); EB co-cultured with BMSCs (lane3); EB co-cultured with BMSCs and treated with BMP4 (lane4). Combined using BMP4 and BMSCs induced significant cardiac differentiation. (H) The effect of RA on the differentiation of cardio‐ myocytes. (D) RT-PCR analysis revealed that both early cardiac genes, MLC-2a and MLC-2v, were increased in RA-treated EBs. Lane1: undifferentiated ES cells; lane2: EB cultured on gelatin-coated dish, FBS(−); lane3: EB co-cultured with BMSCs, FBS(−); lane4: EB co-cultured with BMSCs, FBS(−), and treated with RA. TropI protein was detected by Western of lysed EB on gelatin-coated dish (lane1), EB on gelatin-coated dish and treated with RA (lane2), EB co-cultured with BMSCs (lane3); EB co-cultured with BMSCs and treated with RA (lane4). Combined using RA and BMSCs significantly induced cardiac dif‐ ferentiation. **I-M.** The physical and chemical role of BMSCs in cardiomyocyte induction. (I) BMSCs induced beating cardio‐ myocytes in co-cultured ES cells even when separated by a filter membrane. CM induced cardiac differentiation in ES cells cultured on gelatin-coated dish. However the differentiation rate was lower than with direct co-culture with BMSCs. (J-M) EB on gelatin-coated dish (J); EB co-cultured with BMSCs (K); TropI staining of ES cells when separated from BMSCs by a fil‐ ter (L) and cultured in CM alone (M). Scale Bars = 50um.
