SEARCH

SEARCH BY CITATION

Keywords:

  • human ES cells;
  • transplantation;
  • neural retina;
  • pigmented epithelium;
  • epiblast stem cells

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Correctly inducing the differentiation of pluripotent hESCs to a specific lineage with high purity is highly desirable for regenerative cell therapy. Our first effort to perform in vitro differentiation of hESCs resulted in a limited recapitulation of the ocular tissue structures. When undifferentiated hESCs were placed in vivo into the ocular tissue, in this case into the vitreous cavity, 3-dimensional retina-like structures reminiscent of the invagination of the optic vesicle were generated. Immunohistochemical analysis confirmed the presence of both a neural retina-like cell layer and a retinal pigmented epithelium-like cell layer, possibly equivalent to the developing E12.5 mouse retina. Furthermore, mouse epiblast-derived stem cells, which are reported to share some characteristics with hESCs, but not with mouse ESCs, also generated retinal anlage-like structures in vivo. hESC-derived retina-like structures present a novel therapeutic possibility for retinal diseases and also provide a novel experimental system to study early human eye development. Developmental Dynamics 238:2266–2279, 2009. © 2009 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Embryonic stem (ES) cells have the potential of differentiating into cells of various lineages both in vivo and in vitro (Evans and Kaufman,1981). Human embryonic stem cells (hESCs; Thomson et al.,1998) and recently invented methods to initialize some differentiated somatic cell types into ES cell-equivalent iPS cells (Takahashi and Yamanaka,2006; Takahashi et al.,2007) hold promise for an unlimited source of cells for therapeutic use (Itskovitz-Eldor et al.,2000; Reubinoff et al.,2000). In addition, direct research on early development and organogenesis of the human embryo is so restricted that the ES cells as a substitute for the human embryo are expected to partially overcome this situation (Heins et al.,2004; Pera et al.,2000; Prelle et al.,2002; Trounson,2006).

For the induction of specific cell lineages from ES cells, purified growth and/or differentiation factors or factor-producing feeder or stromal cells have been used (Chadwick et al.,2003; Cohen et al.,2007; Green et al.,2003; Kaufman et al.,2001; Kehat et al.,2001; Lavon et al.,2006; Levenberg et al.,2002; Mummery et al.,2003; Rambhatla et al.,2003; Sottile et al.,2003; Vodyanik and Slukvin,2007; Zhang et al.,2001). In some cases, very complex culture strategies have been used, including the sequential addition and withdrawal of factors and purification of specific cell populations during the culture period (Fang et al.,2006; Wang et al.,2005). As for the induction of ocular tissues, lens and several retinal cell types were reported to have been differentiated from ES cells (Gong et al.,2008; Lamba et al.,2006; Motohashi et al.,2006; Osakada et al.,2008). Factors regulating the differentiation of the retinal tissues, such as Wnt2b, Wnt3a, Notch, Hes1, and RA (Jadhav et al.,2006; Kubo et al.,2003; Luo et al.,2006; Osakada et al.,2007; Tomita et al.,1996), were sequentially added to induce specific retinal cell lineages such as photoreceptor cells (Osakada et al.,2008). Co-cultures of ES cells and stromal cell lines (Aoki et al.,2008a,b; Hirano et al.,2003; Kawasaki et al.,2002; Ooto et al.,2003; Ueno et al.,2006) or developing chick retina (Ikeda et al.,2005; Sugie et al.,2005) were reported to successfully form differentiated photoreceptor cells, amacrine cells, horizontal cells, retinal ganglion cells, lens cells, and retinal pigmented epithelium from ES cells.

In the present study, we first set up co-cultures of hESCs and PA6 stromal cells to induce retinal or lens cells by using the same procedures reported to induce eye-like structures from mouse ES cells (mESCs; Aoki et al.,2007; Hirano et al.,2003). Although we could obtain some retinal cells, the efficiency and total number of these cells in the cultures were very low and disadvantageous for further analysis.

As the retina itself is a multi-layered structure with various cell types, it is to be expected that recapitulation of this structure in a simplified two-dimensional culture would be difficult. To overcome the refractory nature of complicated organs such as retina for the stem cell–based regenerative techniques and induction of retinal organ-like structures useful for the analysis of early human eye development, we tested here the cultivation of hESCs in the adult mouse eye, expecting the in vivo environmental cues to be favorable for eye development. Earlier we had found the developing chick eye to provide a suitable environment for the development of retinal and lens cells (Aoki et al.,2006), although we had used ES cells that had differentiated in culture. Other environments have also been evaluated, e.g., developing rat embryonic kidneys were used to direct proper nephrogenic development of human bone marrow–derived mesencymal stem cells (Yokoo et al.,2006). In spite of our use of adult mouse eyes instead of developing embryonic or neonatal organs as a guide for retinal development, we constantly and effectively obtained structures including both a retina-like layer and a pigmented layer from hESCs. In contrast, no production of structures of this kind occurred when the cells were transplanted into subcutaneous tissues.

Although later processes in retinal development, for example those leading to the formation of mature photoreceptor cells, were not evident at this point, the expression of retina-specific proteins revealed a close similarity between the in vivo–induced structures differentiated from hESCs and the developing retina. Reportedly, reconstituted chick E5 retina formed concentric layered structures indicative of retina in vitro in the presence of Wnt (Nakagawa et al.,2003). In this case, however, reconstitution starts from a mixture of cells dissociated from the whole developing retina. By using the intravitreous cavity of the adult mouse eye, we showed that the undifferentiated hESCs were induced to form early retinal structures composed of retinal cells and pigmented epithelial cell layers.

This simple and efficient model indicates that a mature organ may provide each environmental cue necessary for the development of the retina-like organ from undifferentiated cells. Since in vitro studies so far have not successfully obtained retinas as multi-layered structures from a single type of stem cell, the strategy to use the in vivo environment may provide a model for the developmental study of the human retina as well as transplantable retinal tissue potentially useful for regenerative therapy for retinal diseases.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Generation of Retina-Like Structures From Human Embryonic Stem Cells In Vitro and In Vivo

We previously reported the generation of eye-like structures from mESCs in vitro by using a PA6 feeder layer (Hirano et al.,2003). By using this co-culture system, we tried to generate such structures from hESCs. From the HES-1 cell line (Fig. 1A), obtained from ES Cell International Pte Ltd., we rarely observed the differentiation to pigmented epithelium-like cells in this culture system after day 30 to 40 (a total of 3 pigmented epithelial cell colonies out of 250 colonies observed, as described in “patch” in Supplemental Table1, which is available online, and see Fig. 1B and C; Motohashi et al.,2006). These pigmented epithelium-like cells contained melanin granules (Fig. 1E), with characteristic polygonal morphology (Fig. 1F), and expressed Pax6 (Fig. 1G and H), which is a retinal pigmented epithelium marker as well as one of retina and lens. For another human embryonic stem cell line khES-1 cell line (Fig. 1D) obtained from Kyoto University, 18.5 ± 7.6% of the colonies generated from hESCs differentiated into colonies categorized as “patch,” which consisted of pigmented epithelium-like cells (See Supp. Table 2 and Fig. 1I). However, unlike the mESCs, well-organized eye-like structures such as those from mECSs were formed at only a very low frequency (1.1 ± 0.99%, Fig. 1K, enlarged in “L”), and most of the pigmented cells appeared as patchy pigmented spots (Fig. 1J). The infrequently observed eye-like structures contained a reflective cell cluster in their central area surrounded by non-pigmented neuroepithelial tissue-like structures (Fig. 1M, N) and pigmented cells that formed an independent cell layer (Fig. 1O, P).

thumbnail image

Figure 1. Differentiation of hESCs into eye-like structures in vivo and in vitro. A: Morphology of the undifferentiated hESCs, HES-1. B, C: Thirty-five days after induction of the differentiation of HES-1 co-cultured with PA6 stromal cells. D: Morphology of the undifferentiated hESCs, khES-1. EH: Morphology of the pigmented epithelial cells derived from HES-1 hESCs 35 days after induction. Bright-field image (E), phalloidin staining (F), and Pax6 immunocytochemical staining (G) are shown; F and G are merged in H. I: A representative photograph of a khES-1 hESC culture well at day 35. J: Representative morphology of a khES-1 hESC-derived patch-like structure consisting of pigmented epithelial cells. K: Representative morphology of a khES-1 hESC-derived eye-like structure consisting of pigmented epithelial cells and non-pigmented cell clusters. L: Higher magnification image of K showing the pigmented epithelial cells at the outer side and the non-pigmented ones at the inner side. MP: HE-stained khES-1 hESCs after 35 days of having been co-cultured with PA6 cells. Unpigmented region (M) and pigmented region (O) of the differentiated khES-1 hESCs are shown. N and P are higher magnification views of M and O, respectively. QV: Histology of an eye from a nude mouse at day 35 after khES-1 hESCs had been transplanted into the vitreous cavity. Results of HE staining (Q) and anti-human Ki67 immunostaining (S) of the host mouse eye are shown. R and T are higher magnification views of the region marked by the black square in Q and S, respectively. U and V show higher magnification views of the regions marked by the white star and the black star area, respectively, in S. Arrowheads in U show some human ES cell–derived cells reactive with anti-human Ki67 (MIB1). Note that Ki67 only stains the nuclei of proliferating human cells. Scale bars = (B) 100 μm; (A, C–V) 50 μm.

Download figure to PowerPoint

To investigate a more efficient method for the ocular differentiation and/or regeneration, we transplanted khES-1 directly into adult host mouse eyes, expecting the microenvironment or local cues to be favorable for the properly controlled development of hESCs into ocular tissues or structures. Approximately 30 days after transplantation, about two-thirds of the eyes transplanted with hESCs formed teratoma-like cell masses (n = 16/24; see Fig. 4 Aa and Ab). However, in more than 90% of these teratoma-like masses, we found three-dimensionally organized optic cup–like structures similar to the embryonic eye, reminiscent of the invaginating optic vesicle that forms the optic cup (n = 11/12, Fig. 1Q and R).

thumbnail image

Figure 4. hESC-derived teratoma formed in vivo 35 days after transplantation. Morphology of the undifferentiated hESC-derived teratomas formed after subcutaneous transplantation (Aa, Ab) or intravitreous transplantation (Ac, Ad). Ab and Ad show higher magnification views of the pigmented areas inside of Aa and Ac. From 2 representative independent teratomas generated in subcutaneous area, areas differentiated into neural tissue with pigment granules were dissected and further analyzed in B and C. Sections were stained for human Ki67 (a), HMB45 (b), nestin (c), Mitf (d), synaptophysin (e), Hu C/D (f), GFAP (g), RPE65 (h), Pax6 (i), Rax (k), Oct3 (l), and Tuj1 (m) or stained with HE (j). Dotted lines in each panel of B and C show the boundaries between inner layer and outer layer in the bi-layered neural tissues in the teratoma-like structures. Arrows, Marker-positive cells; arrowheads, Oct3-negative cells. Scale bars = 50 μm.

Download figure to PowerPoint

To confirm that these optic cup-like structures formed in vivo were exclusively composed of hESC-derived cells, we performed immunohistochemical analysis by using human-specific antibody. Many of the cells in experimentally induced organized optic cup–like structures formed in the vitreous cavity of an adult nude mouse were stained by anti-human Ki67 (MIB1; Fig. 1S and T). Figure 1U clearly shows the nuclear staining of these experimentally induced organized optic cup–like structures at the area indicated by the white star in Figure 1S. Weak, non-specific staining of the non-nuclear region of a host adult nude mouse retina indicated by the black star in Figure 1S is enlarged in Figure 1V. Therefore, it is convincing that these experimentally induced organized optic cup–like structures were derived from transplanted hESCs.

Ocular Development During Mouse Embryogenesis

Apparently, these hESC-derived optic cup–like structures formed in vivo were not a close counterpart of the fully developed eye (adult retina sections shown in Fig. 2 Fa to Fj). To gain insight into the developmental stage of these partially-organized optic cup–like structures generated in the host adult mouse eyes, we investigated the expression pattern of eye-related marker proteins in the developing ocular tissue prepared from E10.5 to E17.5 mouse embryos (Fig. 2 Aa–Ej). Negative controls of these stainings are shown in Supplemental Figure 3A–F.

thumbnail image

Figure 2. Immunohistochemical analysis of the mouse eye during development. Ocular tissues were prepared from embryonic day (E) 10.5 (A), E12.5 (B), E14.5 (C), E15.5 (D), E17.5 (E), and from the adult (F) mouse. Ocular tissues at E10.5, E12.5, and E15.5 embryos were obtained from the C57Bl/6 strain, and those at E14.5 and E17.5 embryos and from the adult mouse, from the ICR strain. Results of immunohistochemical staining for mouse Ki67 (a), HMB45 (b), nestin (c), Tuj1 (d), synaptophysin (e), Hu C/D (f), GFAP (g), RPE65 (h), Pax6 (i), and HE staining (j) are shown. Black or white arrows and arrowheads, Marker-positive cells as described in the text. In Bb and Db, we did not indicate any landmarks due to the indistinguishably colored immunohistochemical staining and their original pigmentation. Inset in Ec shows a higher magnification of the nestin-positive area of neural retina close to the lens side. Scale bars = 50 μm.

Download figure to PowerPoint

As shown previously (Ikeda et al.,2005), most of the cells in the early neural retina were proliferating progenitors at E10.5 (Fig. 2 Aj), as confirmed by the mitotic marker Ki67 (arrowheads in Fig. 2 Aa), whereas in the retinal pigmented epithelium, only some of the cells were proliferating (arrows in Fig. 2 Aa); and most of them were post-mitotic cells expressing the melanosome marker HMB45 (arrows in Fig. 2 Ab). Unlike most of the neural progenitors in the central nervous system, neural retina progenitors at this stage did not express nestin (Fig. 2 Ac), a marker of proliferating neural progenitors, and they were negative for Tuj1 (Fig. 2 Ad), a post-mitotic neural marker, and synaptophysin (Fig. 2 Ae), a marker for synaptic connections between neurons. Another neural marker, Hu C/D, which is expressed in the ganglion cells of the adult retina, was detected only in lens cells at this stage (open arrowheads in Fig. 2 Af). We observed no GFAP-positive glial cells (Fig. 2 Ag), and also there were no rhodopsin-positive rod-photoreceptor cells (data not shown) or RPE65-positive functional RPE cells in the retina (Fig. 2 Ah). All of these eye fields including lens, neural retina, and retinal pigmented epithelium were clearly positive for Pax6 (open arrowheads, arrowheads, and arrows in Fig. 2 Ai, respectively).

At E12.5 (Fig. 2 Ba–j), most of the cells in the outer layer of the neural retina were still proliferating (arrowheads in Fig. 2 Ba), whereas the retinal pigmented epithelial cells had turned into a thin layer of heavily pigmented cells, as indicated by Figure 2 Bb and Supplemental Figure 3B. In the inner neuroblastic layer in the neural retina, the Ki-67-negative post-proliferating cells were slightly positive for nestin (arrowheads in Fig. 2 Bc), and clearly positive for Tuj1 (arrowheads in Fig. 2 Bd). Furthermore, axons from these cells, forming the optic nerve, were also positive for synaptophysin (arrowheads in Fig. 2 Be). Also, Hu C/D was expressed in the innermost neuroblastic layer containing a few longitudinal axonal processes (arrowheads in Fig. 2 Bf) and in the proliferating lens cells (open arrowheads in Fig. 2 Bf). We observed no GFAP-, rhodopsin-, or RPE65-positive cells in the retina at this stage (Fig. 2 Bg and 2 Bh and data not shown). The eye field–specific marker Pax6 was clearly expressed in most of the ocular cells including lens, neural retina, and retinal pigmented epithelium (open arrowheads, arrowheads, and arrows in Fig. 2 Bi, respectively).

At E14.5 and15.5 (Fig. 2 Ca–j and Da–j), most of the neural retinal cells except for those in the innermost layer, which is the presumptive ganglion cell layer, and some of the retinal pigmented epithelial cells marked by HMB45 were proliferating (arrowheads and arrows in Fig. 2 Ca, respectively; arrowheads in Da and arrows in Cb). The neural retina cells had nestin-positive axons extending from the inner layer to the outer layer (arrowheads in Fig. 2 Cc and Dc), and Tuj1-positive axonal processes were observed throughout the neural retinal layers (areas between the arrowheads in Fig. 2 Cd and Dd). On the other hand, synaptophysin-positive regions were restricted to the presumptive nerve fiber layer (open arrowheads in Fig. 2 Ce and Fig. 2 De), inner plexiform layer (open arrows in Fig. 2 De), and outer side of the outer nuclear layer (arrowheads in Fig. 2 Ce and De). Hu C/D was expressed in the inner layer of the neural retina (arrowheads in Fig. 2 Cf and Df) and the surface cells of the lens (open arrowheads in Fig. 2 Cf and Df), and by comparison with Figure 2 Ca and Da, these Hu C/D-positive cells in the inner layer of the neural retina were post-mitotic. No expression of GFAP, rhodopsin, or RPE65 was detected (Fig. 2 Cg, Dg, Ch, and Dh, and data not shown). All nuclei were clearly positive for Pax6 in these eye fields including lens, neural retina, and retinal pigmented epithelium (open arrowheads, arrowheads, and arrows in Fig. 2 Ci and Di, respectively).

At E17.5 (Fig. 2 Ea–j), most of the neural retinal cells, except for the presumptive ganglion cells in the innermost layer (arrowheads in Fig. 2 Ea), and a few retinal pigmented epithelial cells marked by HMB45 were proliferating (arrows in Fig. 2 Ea and Eb). Nestin-positive axons extended throughout the inner layer to the outer layer (arrowheads in Fig. 2 Ec and magnified view in the inset), and Tuj1-positive axonal processes were observed throughout all neural retinal layers (areas between the arrowheads in Fig. 2 Ed). Synaptophysin-positive regions seemed to be restricted to the presumptive inner plexiform layer (open arrowheads in Fig. 2 Ee) and the outer side of the outer nuclear layer (arrowheads in Fig. 2 Ee). Also, Hu C/D was still expressed in the inner layer of the neural retina (arrowheads in Fig. 2 Ef) and the surface cells of the lens (open arrowheads in Fig. 2 Ef), and by comparison between Figure 2 Ea and Ef, these Hu C/D-positive cells in the inner layer of the neural retina were post-mitotic (Fig. 2 Ef). GFAP, rhodopsin, and RPE65 were still not detected (Fig. 2 Eg and Eh, and data not shown). Nuclei in the eye field including lens, neural retina, and retinal pigmented epithelium were still clearly positive for Pax6 (open arrowheads, arrowheads, and arrows, respectively in Fig. 2 Ei).

In the adult ICR mouse retina (Fig. 2 Fa–j), cells in the choroids were positive for HMB45 (arrows in Fig. 2 Fb). The retinal ganglion cells were stained by anti-nestin (arrowheads in Fig. 2 Fc). As shown in the later embryonic stages, Tuj1-positive axonal processes were observed throughout all neural retinal layers (areas between the arrowheads in Fig. 2 Fd), as were synaptophysin-positive ones (areas between the arrowheads in Fig. 2 Fe). Tuj1-positive cells located outside the retinal pigmented epithelium seemed to be neurons in the choroids (arrows in Fig. 2 Fd). Hu C/D was expressed in the ganglion cell layer more heavily (open arrowheads in Fig. 2 Ff) and in some cells in the inner nuclear layer (open arrows in Fig. 2 Ff). We observed the expression of GFAP in the inner layer of the neural retina (open arrowheads in Fig. 2 Fg), rhodopsin in the outer layer of the neural retina consisted of the photoreceptor layer and in the retinal pigmented epithelium (data not shown), and RPE65 in the retinal pigmented epithelium (arrows in Fig. 2 Fh). Pax6 was expressed in the ganglion cell layer (open arrowheads in Fig. 2 Fi) and in the inner side of the inner nuclear layer (open arrows in Fig. 2 Fi).

These results show that: (1) while retinal pigmented epithelial cells express HMB45 and Mitf (Mitf expression was reported previously; Ikeda et al.,2005) at an early stage in embryogenesis, RPE65 is not expressed during embryogenesis; (2) the development of the neural retina proceeds gradually from the inner part: at first, the inner neuroblastic layer starts to actively proliferate and supposedly differentiates into Hu C/D post-mitotic cells whose axons form the optic nerve, and then the proliferating cells in the outer layer start to form synaptic connections; (3) ocular-field nuclei including those of the neural retinal and retinal pigmented epithelial cells continue to express Pax6 throughout all developmental stages.

Characterization of the Optic Cup–Like Structures Formed After Transplantation of hESCs Into the Mouse Eye

As shown in Figure 1, we successfully generated hESC-derived optic cup–like structures in the host albino adult nude mouse eye. Next, we closely investigated these regenerated structures immunohistochemically. These structures formed in the host albino adult nude mice eye were characteristically bi-layered, composed of 2 immunohistochemically different types of cells, as typically depicted in Figure 3 Aa and Ba. To well illustrate these bi-layered structures, dotted blue lines were inserted in the boundary of the bi-layered optic cup–like structures. Immunohistochemically stained cells in the inner layer (located to the left of the dotted line) and those in the outer layer (to the right of the dotted line) are indicated by the arrowheads and arrows, respectively (Fig. 3 Ba to Cl). We also show their negative controls in Supplemental Figure 3H to J. Cells forming the hESCs-derived optic cup–like structure (arrowheads and arrows in Fig. 3 Ba and Ca) were still proliferating as judged from the expression of the mitotic marker Ki67 (Fig. 3 Aa, Ba, Ca), and only the outer cell layer in the optic cup–like structure indicated by the arrows was positive for the melanosome marker HMB45 (Fig. 3 Ab and arrows in Fig. 3 Bb, Cb). Furthermore, these HMB45-positive cells were also positive for Mitf (Fig. 3 Ad and arrows in Fig. 3 Bd, Cd), which is another early pigmented epithelial cell marker; and this cell layer was relatively thin. Except for this region, the remainder of the human Ki67-positive area (located to the left of the dotted line) was relatively thicker and positive for nestin (Fig. 3 Ac and arrowheads in Fig. 3 Bc, Cc) and Tuj1 (Fig. 3 Ak and arrowheads in Fig. 3 Bk and Ck). In the nestin-positive area, whereas synaptophysin expression was negative (Fig. 3 Ae, Be, Ce), Hu C/D-positive cells were localized distal to the HMB45-positive area of the inner cell layer side (left side of the dotted line) but not detected around the boundary area between the neural retina-like layer and pigmented epithelium-like cell layer (Fig. 3 Af and arrows in Fig. 3 Bf and Cf). Furthermore, this presumptive neural retina-like structure was composed of not only a Hu C/D-positive presumptive retinal ganglion-like cell layer but also additional layers possibly corresponding to the plexiform layer and nuclear layer located between the presumptive ganglion cell layer and pigmented epithelium-like cell layer. GFAP (Fig. 3 Ag, Bg, and Cg), rhodopsin (data not shown), and RPE65 (Fig. 3 Ah, Bh, and Ch) were still negative in all specimens tested (n = 10) as in the corresponding mouse embryonic eyes described above (Fig. 2), whereas RAX (Fig. 3 Al and arrowheads and arrows in Fig. 3 Bl and Cl) and Pax6 (Fig. 3 Ai and arrowheads and arrows in Fig. 3 Bi and Ci) were expressed throughout the area.

thumbnail image

Figure 3. In vivo differentiated hESC-derived eye-like structures at 55 days after transplantation. Undifferentiated hESCs were transplanted into the vitreous cavity of nude mice. By 55 days after transplantation, optic cup–like structures had formed in hESC-derived cell masses in vivo. Whole optic cup–like structure (A), enlarged photo of the solid boxed area in A (B), and the transition area from the neural retina-like layer to the retinal pigmented epithelium prepared from another optic cup–like structure (C) are shown. Sections were stained for human Ki67 (a), HMB45 (b), nestin (c), Mitf (d), synaptophysin (e), Hu C/D (f), GFAP (g), RPE65 (h), Pax6 (i), Tuj1 (k), or Rax (l) or with HE (j). Dotted blue lines in each panel of B and C show the boundaries between neural retina-like inner layer and pigmented epithelium-like outer layer in the optic cup–like structures. Note that Ki67 only stains the nuclei of proliferating human cells. Arrows, Marker-positive cells in the pigmented epithelium-like outer layer; arrowheads, those in the neural retina-like inner layer as described in the text. Scale bars = 50 μm.

Download figure to PowerPoint

The morphology of the whole optic cup–like structure and the marker expression pattern except for that of synaptophysin suggest that the optic cup–like structure 55 days after transplantation closely resembled the E14.5 to 15.5 mouse retina. Each marker expression pattern at an earlier post-transplantation stage, i.e., 35 days after transplantation, was almost identical to that seen at 55 days after transplantation (Supp. Figs. 1A–K and 3G). These results indicate that the hESC-derived optic cup–like structure generated in vivo formed a presumptive retinal pigmented epithelium and a presumptive neural retina.

Differentiation of hESCs Transplanted Into Intracutaneous Tissue

Next, to investigate whether these experimentally induced hESC-derived optic cup–like structures were generated specifically in the ocular tissue or not, we subcutaneously injected hESCs into the back skin of nude mice. The transplanted hESCs formed typical teratoma-like structures when transplanted into the subcutaneous area (Fig. 4 Aa to Ad) compared to the marker-positive cells observed after transplantation in the eye (Fig. 3B).

Among these experimentally induced teratoma-like structures, we focused on the bi-layered or folded neural tissues (Fig. 4B and C). We marked the boundary as in Figure 3, and marker-positive cells were marked by arrowheads (Fig. 4 Ba to Cm and Supp. Fig. 3K and L). Cells composing the bi-layered neural tissues in the teratoma-like structures were proliferative (arrowheads in Fig. 4 Ba and Ca) as those in the hESC-derived optic cup–like structures in the eye. HMB45-positive cells in the structure were present in a small cluster (arrowheads in Fig. 4 Bb and Cb), and this cluster was slightly positive for Mitf (arrowheads in Fig. 4 Bd and Cd) and positive for Pax6 (arrowheads in Fig. 4 Bi and Ci). Outside of this region, cells were positive for nestin (arrowheads in Fig. 4 Bc and Cc), Tuj1 (arrowheads in Fig. 4 Cm), and RAX (arrowheads in Fig. 4 Bk and Ck). We never observed synaptophysin- (Fig. 4 Be and Ce), Hu C/D- (Fig. 4 Bf and Cf), GFAP- (Fig. 4 Bg and Cg), rhodopsin- (data not shown), or RPE65- (Fig. 4 Bh and Ch) positive cells in these cell clusters. (Note that uniform staining for HuC/D in Fig. 4 Bf and Cf is background, because HuC/D is a nucleus-localizing protein.) However, Oct3/4-positive cells were frequently found in the Ki67-positive area (arrowheads in Fig. 4 Bl and 4 Cl) except for the small cluster that included the HMB45-, nestin-, Mitf-, and Pax6-positive cells described above (arrows in Fig. 4 Bl). The negative controls are shown in Supplementary Figure 3K and L.

After close inspection of the skin bearing the subcutaneously transplanted hESCs, we never observed the three-dimensional eye-like structure or any other organized cellular structures resembling the neural retina (n = 7/7 negative). These results indicate that the hESC-derived optic cup–like structures were the specifically induced structures in the ocular environment when undifferentiated hESCs were transplanted into it.

Neuroepithelial Structures Formed in the Teratoma of a Human Patient

To investigate whether the retina- or optic cup–like structures formed in the experimentally induced hESC-derived teratoma were the common features of spontaneous human teratomas, we then analyzed various areas of some surgically resected human immature teratoma generated in human patients by examining the expression of the markers used above. Among the various differentiated cells, we focused the neural lineage cells, especially those forming the layered structures (Fig. 5A and B). Most of the cells in such layered structures of a surgically resected human immature teratoma were proliferative (arrowheads in Fig. 5 Aa and Ba) as those observed in the experimentally induced hESC-derived teratoma. HMB45-positive cells were observed in a part of the teratoma that had formed a neural tissue-like structure (arrowheads in Fig. 5 Ab and Bb). Most of these HMB45-positive cells also weakly expressed Mitf (arrowheads in Fig. 5 Ad and Bd). On the other hand, Pax6 was expressed broadly (Fig. 5 Ai and Bi). The Pax6-positive regions contained nestin- (arrowheads in Fig. 5 Ac and Bc), Tuj1- (arrowheads in Fig. 5 Ak and Bk), and RAX- (arrowheads in Fig. 5 Al) positive cells. Synaptophysin-positive regions were observed in the HMB45-negative areas (arrowheads in Fig. 5 Ae and Be), and in some cases, these synaptophysin-positive regions were clearly localized on the opposite side of the nestin- and HuC/D-positive area. In some cases, Hu C/D-positive cells were co-localized in a part of the nestin and/or Tuj-1 positive area (arrowheads in Fig. 5 Af and Bf). The structures shown in Figure 5 seem to be partly similar to those found in hESCs-derived optic cup–like structures formed in the mouse eye; however, we could not find any bi-layered, optic cup–like structures (as shown in Fig. 3A) after the inspection of 25 independent areas of the specimen from different parts of the original teratoma. We never observed GFAP- (Fig. 5 Ag and Bg), rhodopsin- (data not shown), or RPE65- (Fig. 5 Ah and Bh) positive cells in these structures. GFAP-positive cells were present in another area of the same section shown in Figure 5 Bm. Their negative controls are found in Supplementary Figure 3M and N.

thumbnail image

Figure 5. Immunohistochemical analysis of a teratoma generated in the subcutaneous area of the back skin of a human patient. A, B: Areas that had differentiated into neural tissue with pigment granules in the human immature teratoma were selected. Sections were stained for human Ki67 (a), HMB45 (b), nestin (c), Mitf (d), synaptophysin (e), Hu C/D (f), GFAP (g, m; m is a GFAP-positive area outside of the neural tissue with pigment granules and is shown as a positive control for GFAP immunostaining), RPE65 (h), Pax6 (i), Tuj1 (k), and Rax (l) or with HE (j). Arrows, Marker-positive cells. Scale bars = 50 μm.

Download figure to PowerPoint

Thus, in a surgically resected spontaneous human immature teratoma, we never observed the well-organized retina- or optic cup–like structures like those formed in the experimentally induced hESC-derived teratoma. These results indicate that eye-like structures were not commonly generated in tetatomatous tissues, and suggest that the well-organized retina- or optic cup–like structures were likely to have been generated specifically from the undifferentiated hESCs.

Differentiation of EpiSCs But Not mESCs Into Retinal Anlage-Like Structures In Vivo

Finally, we investigated the characteristics of structures derived from mouse epiblast-derived stem cells (mEpiSCs) in vivo, because we had not successfully observed the retina- or optic cup–like structures generated in vivo from mESCs. mEpiSCs were recently reported to be multipotent stem cells derived from the epiblast, a tissue of the post-implantation embryo that generates the embryo proper and shares patterns of gene expression and signaling responses with hESCs (Brons et al.,2007; Tesar et al.,2007).

To investigate the characteristics of teratomas derived from mEpiSCs, we established mEpiSCs from the epiblast of a pregastrula-stage mouse [E5.5, (B6xDBA2)F1 mated inter se] embryos (Supp. Fig. 2A and B). Our established mEpiSCs expressed E-Ras and Oct3/4, both markers for undifferentiated cells such as mESCs (Supp. Fig. 2C and D). As previously reported (Brons et al.,2007; Tesar et al.,2007), our established mEpiSCs showed no or very low expression of Rex1, which is another marker for undifferentiated cells such as mESCs but not mEpiSCs, and high expression of Fgfr5, which is expressed in mEpiSCs but not in mESCs (Supp. Fig. 2E and F). The mEpiSCs could differentiate into all 3 germ layers in vitro, such as αSM-positive muscle cells, GFAP-positive glial cells, Tuj1-positive neurons, and CD105 (Endoglin)-positive endothelial cells (Supp. Fig. 2H–P).

mEpiSCs transplanted into either subcutaneous tissue or the vitreous cavity generated teratomas (Fig. 6A; n = 8/10 injected into subcutaneous tissue and n = 5/10 into the vitreous cavity). Again, we observed the differentiation of these cells into all 3 germ layers in teratomas generated in subcutaneous tissue at 45 days after transplantation, such as skin-like structures (Fig. 6B), ganglion cells (Fig. 6C), rosette-like neural structures (Fig. 6D), cartilage (Fig. 6E), bone (Fig. 6F), gastrointestinal structures (Fig. 6G), and melanocytes (Fig. 6H). However, we did not observe any retina- or optic cup–like structures in these teratomas. On the other hand, in teratoma-like cell masses generated in the vitreous cavity at 45 days after transplantation, we observed the co-localization of neural cells and pigmented epithelial-like cells (Fig. 6I and J). These structures were never observed when mESCs were transplanted into the vitreous cavity (data not shown). As described in our previous report (Aoki et al.,2007), when mESCs were transplanted into the vitreous cavity by the same protocol used for the mEpiSCs, most of the teratomas were generated at 10 to 15 days after transplantation, whereas mEpiSCs never generated teratomas in such a short period. Thus, we showed mEpiSCs to have the ability to differentiate in vivo into retinal anlage-like structures, in which the neural cell cluster and the pigmented epithelium-like cell layer were closely placed, although the induced structures were deformed compared with those induced from hESCs.

thumbnail image

Figure 6. Histological analysis of the differentiation of EpiSCs into cells of all 3 germ layers. A: Teratomas generated from EpiSCs in subcutaneous tissue at 45 days after transplantation. Skin-like structures (B), ganglion cells (C), rosette-like neural structures (D), cartilage (E), bone (F), gastrointestinal structures (G), and melanocytes (H) were observed. Teratomas generated from EpiSCs in vitreous cavity at 45 days after transplantation (I, J). Co-localization of neural cells and pigmented epithelium-like cells were observed in these teratomas.. Scale bars = 50 μm.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Efforts to induce various kinds of cells and tissues from stem cells including ES cells have been pursued to comply with patients looking to regenerative medicine for a cure. After the discovery of iPS cells (Takahashi and Yamanaka,2006; Takahashi et al.,2007), methodological improvements enabling the safe and effective use of hESCs or iPS cell derivatives have become the immediate tasks imposed upon the medical and biological science communities. In addition, our knowledge about mammalian development, which is definitely important to correctly manage the basic strategies required for the therapeutic use of ES cells or iPS cells, is restricted to mice. Thus, hESCs are expected to substitute for the human fetus, at least in part, to unravel the mystery of human-specific developmental mechanisms (Cauffman et al.,2006,2009; Choudhary et al.,2007; Dvash et al., 2004,2006; Gjerstorff et al.,2008; Pera and Trounson,2004; Rugg-Gunn et al.,2005). Even restricted to the ocular cells or tissues, only cultured cornea cell sheets induced from epithelial stem cells have been used for therapy (Nozaki et al.,2008; Sekiyama et al.,2006), and many induced cell types including those of retina and retinal pigmented epithelium have been used only in animal models (Kubota et al.,2006; MacLaren et al.,2006; Wang et al.,2008). Still, ocular cells good for therapeutic use may possibly be prepared from ES cells or iPS cells; however, the complex layered structure of the retina would be hard to regenerate as a tissue in present culture systems.

In this current study, we sought to induce structures closely resembling the neural retina, including the retinal pigmented epithelium, from hESCs and succeeded in generating embryonic optic cup–like structures by the transplantation of undifferentiated hESCs into the adult mouse eye. Although we did not test the physiological function of the regenerated cells in these structures, the morphology and expression of retina-specific markers suggested the organized induction of optic cup–like tissues in vivo from hESCs. It should be noted that after a prolonged time post-transplantation, eyes were deformed and sometimes spontaneously lost, indicating the progression of teratoma-like tumor formation from the transferred hESCs. Thus, simple transplantation of hESCs into the human eye would obviously not be suitable for clinical application, though the formation of the 3-dimensional ocular tissues could be created based on the present study.

As indicated by various investigators (Gerecht-Nir et al.,2004; Gertow et al.,2004; Yokoo et al.,2008), developing tissues or organs have the ability to induce implanted stem cells to differentiate into the same types of tissues or organs into which the stem cells were transplanted. An impressive case for renal promodial-like ducts induced from mouse ES cells was reported (Yamamoto et al.,2005). They allowed the outgrowth of ES cell–derived embryoid bodies for a week in vitro, then the harvested cells were transplanted to adult nu/nu host retroperitoneum. After 14 to 28 days, in the teratoma-like structures formed, structures intrinsic for mesonephric ducts and ureteric buds based on their gene expression pattern and morphological features were frequently observed. After survival of 28 days, some ducts were accompanied by mesonephric nephron-like convoluted tubules. In this case, a retroperitoneum environment might have guided or restricted teratoma-like development of ES-derived cells to the renal developmental pathway.

It is easily speculated that stem cells may follow the developmental process of the target tissues or organs after transplantation, influenced by numerous factors spatially and temporally similar to those found in the developmental milieu. As shown in Figure 4A, neural retina-like layers formed in the host eyes were closely associated with the retinal pigmented epithelium layer in most cases, as is observed in the early development of the optic cup. This finding suggests some specific influence of the adult eye on the recapitulation of the ocular developmental process in the transplanted stem cells. Our future goal is to elucidate not yet identified environmental cue(s) present in the adult eye that forcefully directs the differentiation of transplanted hESCs.

mESCs transplanted in vivo did not form the retina- or optic cup–like structures found with hESCs; however, mEpiSCs established from the late epiblast layer of post-implantation mouse embryos were shown to form retinal anlage-like structures associated with the co-induced neural cells and pigmented epithelium-like cells. Based on the similarity of expression of various markers and the fact that the mEpiSCs were established by using chemically defined, activin-containing culture medium, which is sufficient for long-term maintenance of hESCs, we consider mEpiSCs to be the key to understanding the distinctions between mouse and human ES cells. Very recently, introduction of a single reprogramming factor, Klf4, into EpiSCs induced mESC-like pluripotent stem cells named Epi-iPS cells (Guo et al.,2009), strongly suggesting different developmental states of EpiSCs and ES cells. In this respect, the tendency to develop optic cup–like or retinal anlage-like structures in vivo, common to both hESCs and mEpiSCs, would be expected from the similarity of these two types of cells.

In addition to the therapeutic potential of hESCs, the present regenerative system may provide a novel experimental system to study human eye development especially in its early stage.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Cell Cultures

The hESCs were used in accordance with the human ES cell research guidelines of the Japanese government. Human embryonic stem cell line HES-1, obtained from ES Cell International Pte Ltd., was maintained according to the supplier's protocol (Version 2 February 2002. See http://www.escellinternational.com/). Another human embryonic stem cell line, khES-1, was obtained from Kyoto University and maintained as previously described (Suemori et al.,2006; Ueno et al.,2006) and supplementary methods. The human ES cell lines expressed the undifferentiated ES cell markers Oct-3/4 and Nanog. These hESCs were immunonegative for pan-neural markers such as nestin, Tuj-1, and GFAP, consistent with hESCs remaining undifferentiated in our cultures.

For transplantation, human ES colonies were partially dissociated into clumps (5–10 cells per clump) by treatment with 0.25% trypsin and 0.1 mg/ml collagenase IV in PBS containing 1 mM CaCl2 and 20% Knock-out Serum Replacement (KSR) and then pipetted gently. The cells were then maintained as referenced above.

Differentiation of hESC In Vitro

With minor modification, the method for the induction of eye-like structures was the same as previously described (Hirano et al.,2003). Briefly, HES1 colonies were dissected with glass needles in the same manner as used for the passage procedure. About 15–20 pieces of the colonies were inoculated on PA6 cells previously seeded on 60-mm cell culture dishes (BD Falcon). khES-1 colonies were partially dissociated into clumps (5–10 cells per clump) by treatment with 0.25% trypsin and 0.1 mg/ml collagenase type IV in PBS containing 1 mM CaCl2 and 20% KSR, and then gently pipetted. About 50 hESC clumps were inoculated on a monolayer of PA6 cells previously seeded on 6-well plates (BD Falcon). hESCs were induced to differentiate in alpha-minimum essential medium (α-MEM: Invitrogen Co., Carlsbad, CA) supplemented with 10% fetal calf serum (FCS: EQUITECH-BIO, Inc., Kerrville, TX). They were maintained in an atmosphere of 5% CO2 and 95% air at 37°C. Then, 40 pM bFGF was added to the medium on day 0, and 10 nM dexamethasone (Sigma-Aldrich, St. Louis, MO) was added on day 6. The cultures were also exposed to 20 pM cholera toxin (Sigma) from days 0 to 6, and the medium was changed every 3 days.

Animals

All animal experiments were conducted in accordance with the ARVO Statement for Use of Animals in Ophthalmic and Vision Research and were approved by the Animal Research Committee, Graduate School of Medicine, Gifu University. Five-week-old female BALB/C (nu/nu) nude mice were obtained from Chubu Kagakusizai Co. (Aichi, Japan). We also used ICR and C57Bl/6 pregnant female mice to obtain embryos.

They were housed in standard animal rooms with food and water ad libitum under controlled humidity and temperature (22° ± 2°C). The room was illuminated by fluorescent lights that were on from 8:00 to 20:00 hr.

Transplantation Procedures

The surgical procedures used for the transplantation were described in detail earlier (Timmers et al.,2001) and in the Supplementary Materials. A suspension of about 105 hESCs was injected into the vitreous cavity (Aoki et al.,2008b). In control eyes, the same volume of medium with or without the mitomycin C–treated mouse embryonic fibroblasts (mEFs) used in hESC cultures, but not including seeded hESCs, were injected into different groups of animals. After these injections, 1 drop of ofloxacin ophthalmic solution (Tarivid topical solution, Santen) was applied to the host eyes.

Then, we also transplanted hESCs intracutaneously into the same mice that had received hESCs transplanted into their vitreous cavity. In this case, hESCs were prepared at a concentration of 105 cells/μl, and 50 μl of this hESC suspension was mixed with the same amount of Matrigel (BD Biosciences, CA). Then, the total 100-μl hESC suspension was injected intracutaneously into the back skin.

Histology and Immunohistochemistry

Thirty-five or 55 days after the transplantation, the mice were killed by an overdose of pentobarbital sodium (200 mg/kg). The eyes and the teratoma-like cell cluster generated under the skin were removed and fixed by immersion in 10% formalin in phosphate buffer (pH 7.2) overnight. For immunohistochemistry, embryonic day (E) 10.5, 12.5, 14.5, 15.5, and 17.5 mouse embryos were collected. The embryos were fixed by immersion in 10% formalin in phosphate buffer (pH 7.2) for 4, 6, 9, 12 hr, and overnight, respectively, depending on their size. Specimens of human immature teratoma used in the present study were prepared from surgically resected tumors.

The methods used for immunohistochemistry were described in detail elsewhere (Aoki et al.,2006,2007) and in supplementary methods. Briefly, specimens were dehydrated with ethanol, soaked in xylene, and embedded in paraffin. Horizontal serial sections were prepared at a 3-mgr;m thickness by using a Leica RM2125RT microtome (Leica RM2125RT, Leica Microsystems Inc., Bannockburn, IL). Sections were stained with hematoxylin and eosin (HE) or used for immunohistochemical analysis described in the Supplementary Materials. Negative controls were prepared to test for the specificity of the antibodies used. The preparation of the negative control for each specimen is described in supplementary methods, and the results are shown in Supplementary Figure 3.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

We thank Kyoko Takahashi for her excellent technical assistance. We also thank Dr. Tsutomu Motohashi for his thoughtful advice. This work was supported by JSPS Research Fellowships for Young Scientists (H.A.).

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
DVDY_22008_sm_SupFigS1.tif8540KSupplemental FigureS1.
DVDY_22008_sm_SupFigS2.tif16477KSupplemental FigureS2.
DVDY_22008_sm_SupFigS3.tif23224KSupplemental FigureS3.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.