BACKGROUND:The recent success in the derivation of differentiated cell types from stem cells has raised prospects for the application of regenerative cell therapy. In particular, embryonic stem cells are attractive sources for cell transplantation, due to their immortality and rapid growth. These cells, however, also possess tumorigenic properties, which raises serious safety concerns and makes biosafety testing mandatory. Our goal was to establish a highly sensitive animal model for testing the proliferative potential of stem-cell grafts.METHODS:BALB/c nude mice received cell grafts of non-neoplastic MRC-5 cells containing defined numbers of mouse embryonic stem cells. We either injected 1 million viable cells into the kidney capsule, or mixed 2 million cells with Matrigel for s.c. transplantation. To analyze the possible impact of an intact immune response on tumor development, we also transplanted the cells into immunocompetent mice. Animals were sacrificed when the tumors became >1 cm and were analyzed in detail.RESULTS:The nude mouse model reproducibly allowed detection of 20 tumorigenic cells, and even as few as 2 ES cells were found to form teratoma. Interestingly, the administration of cell grafts at two different application sites resulted in different growth kinetics and tumor phenotypes. The highest level of sensitivity (100% detection of 20 tumorigenic ES cells) was achieved by s.c. injection of cells mixed with Matrigel. The influence of the immune system on tumor-cell development was demonstrated by a higher tumor rate of transplants in immunodeficient nude mice compared with immunocompetent mice.DISCUSSION:We have established a reliable animal model for routine assessment of the biosafety profile of stem-cell-derived cell transplants. This model will facilitate the generation of homogenous non-tumorigenic cell populations, and will help to integrate standardized safety systems into the application of stem-cell-derived grafts for clinical purposes.

BackgroundCellular cardiomyoplasty is evolving as a new strategy to treat cardiac diseases. A prerequisite is a reliable source of pure cardiomyocytes, which could also help in the exploitation of recent advances in genomics and drug screening. Our goal was to establish a robust lab-scale process for the generation of embryonic stem (ES)-cell derived cardiomyocytes in suspension.MethodsA F1 ES cell clone carrying a construct consisting of the alpha-cardiac myosin heavy chain (alphaMHC) promoter driving the neomycin resistance gene was used for antibiotic-driven cardiomyocyte enrichment. Rotating suspension culture was established to initiate embryoid body (EB) formation. To track growth and differentiation kinetics, cell count and flow cytometry for SSEA-1. E-cadherin (stem-cell marker) and sarcomeric myosin (cardiomyocytes marker) was performed. Oct4 expression was measured via real time (RT)-PCR.ResultsCultures comprising 2.5-8 x 10(6) differentiating ES cells/mL were obtained after 9 days in rotating suspension. Upon G418 addition, vigorous contracting spheres, termed cardiac bodies (CB), developed. These cultures consisted of about 2.1 x 10(5) enriched cardiomyocytes/mL after 6-10 days of selection. Suspensions comprising 90-95% viable single cells were generated using an improved dissociation method. Seeding of cardiomyocytes with 7 x 10(4) cell/cm(2) resulted in a homogenous monolayer of synchronously contracting cells. Myocyte specific immunohistochemistry indicated purity of >99%.DiscussionWe have established a reliable lab-scale protocol to generate cultures of highly enriched cardiomyocytes in suspension. This will facilitate development of larger-scale processes for stem-cell based cardiomyocyte supply. An improved method is provided to derive vital suspensions of cardiomyocytes, which could be utilized for transplantation as well as for drug screening purposes.

We investigated the functional role of glia cells during retinogenesis using the rotation culture system. Reaggregating cells from the embryonic chick retina have the unique capacity to reassemble into laminated cellular spheres. These spheres are composed of several compartments holding the constituents of many retinal layers in a topologically correct, yet inverse orientation. However, when these spheres are cultured in the presence of conditioned media derived from monolayers of cerebellar glia cells, the reassembling retinal cells behave totally differently. The anlage of the originally reversed lamina polarity is progressively transformed within a week into a sphere with a compound and correctly laminated orientation. Conditioned media from fibroblasts, other glia cells (except Müller cells) or a set of already characterized retinogenetic factors are not able to produce this dramatic transformation. Additionally, we were able to show that only retinal cells are able to respond with a reorganization process. Reaggregating cells from the chick cerebellum also form spheroids; however, neither in the presence of cerebellar glia cell-derived conditioned medium nor their control counterparts are they able to reassemble histotypically. This indicates that cerebellar glia cells produce diffusible factors to which retinal cells can respond and that these factors can act as important determinants for the correct establishment of the retinal polarity. Since all types of laminar disorganization are of great clinical significance, the knowledge of factors which determine and sustain the normal retinal architecture are biomedically highly relevant.

Müller cells, that belong to the family of radial glia cells, have central functions during retinogenesis. They form a stabilizing scaffold, they are candidate targets for the mediation of extraneous retinogenetic factors, and they are an important source for retina-borne retinogenetic factors. Reaggregate cultures allow the analysis of retinogenesis from dispersed cells to fully laminated tissues. Reaggregating cells from the embryonic chick retina reassemble to reversed laminated cellular spheres including constituents of all retinal layers, yet the outer nuclear layer is represented by internal rosettes. Using spheroids, we tested whether Müller cells have a decisive function in establishing retinal polarity and in determining the lamination pattern. To this end, we established confluent monolayers of highly enriched Müller cells derived from E6 or E13 chicken retinas, and then let dispersed E5.5 retinal cells reaggregate either in the absence of these monolayers or on top of them. In the presence of Müller cells, the reversed lamina polarity of rosetted spheroids progressively transformed within a week into correctly laminated retinal spheres, whereas all initial rosettes vanished. Moreover, photoreceptors formed a regular outer nuclear layer, as visualized by the rod-specific CERN901 antibody. In correctly laminated spheroids, staining for vimentin and glutamine synthetase was much more pronounced than in rosetted spheroids; in particular, a well-established inner limiting membrane stood out wherever the retinal lamination was complete. Because these effects can be similarly achieved by supernatants derived from Müller cells, direct cell-cell contacts or cellular replenishment from the monolayer do not account for these effects. We conclude that Müller cells are involved in the establishment of a correct retinal lamination and in the arrangement of the cells in the reaggregate cultures. In particular, rosette formation is counteracted and the formation of an inner limiting membrane is induced. Because rosettes are objects of concern in several ophthalmological defects, these results are highly relevant, both biomedically and also for normal retinogenesis.

During eye formation, inductive phenomena occurring between retinal pigmented epithelium (RPE) and retina are not well understood. After briefly summarizing the normal development of retina and RPE, we present three-dimensional in vitro models of the chick embryonic retina which allows elucidation of RPE-retina interactions. In such retinospheroids, a complete arrangement of layers is achieved, provided that dispersed retinal cells are: (1) young enough; and (2) reaggregated on a monolayer of RPE. Thereby, the RPE extends cell proliferation, while differentiation is much delayed. These findings assign to the RPE a decisive role for the genesis and regeneration of a vertebrate retina.

The establishment of cell and fibre layers and the specification of different cell types are crucial processes during development of the central nervous system. Here we investigated the developmental architecture of radial glia cells in these processes using so-called spheroids that arise from dissociated chicken embryonic neural cells in rotation culture. We were able to produce retinal, tectal, and telencephalic spheroids from E6 embryos and cerebellar spheroids from E10 embryos. Cell and fibre differentiation can be observed in all types of spheroids, however, it is most abundant in retinal spheroids. Moreover, only in retinal spheroids a histotypic organization can be detected. Using immunohistochemistry and electron microscopy, we assign this -at least partially- to the capacity of Müller cells to form radial scaffolds, since we observe a congruency between these radial scaffolds and the presence of rosettes formed by photoreceptor precursors and Müller cells. Tectal, telencephalic and cerebellar spheroids do not show organized radial glia scaffolds, instead, the radial glia cells are randomly arranged and the spheroids do not show histotypical organization. The application of the specific gliotoxin 6-aminonicotinamide to growing retinal spheroids leads to a significant decrease in the number and size of the rosettes. Concomitantly, the degree of histotypical organization is also drastically reduced. This organizing capacity of Müller cells in vitro now strongly suggests the presence of a comparable function also in vivo. Moreover, since non-retinal radial glia cells are not able to re-organize an histotypic organization in vitro, Müller cells seem to be qualitatively different from other radial glia cells. In future studies we want to untangle these differences.

Reaggregation of dispersed retinal cells of the chick embryo leads to histotypic retinospheroids in which the laminar organization remains incomplete: photoreceptors form rosettes which are surrounded by constituents of the other retinal layers. Here, for the first time, a complete arrangement of layers is achieved in cellular spheres (stratoids), provided that fully dispersed retinal cells are younger than embryonic day E6, and are reaggregated in the presence of a monolayer of retinal pigmented epithelium (RPE). A remarkable mechanism of stratoid formation from 1 to 15 days in vitro is revealed by the establishment of a radial Muller glia scaffold and of photoreceptors. During the first two days of reaggregation on RPE, rosettes are still observed. At this stage immunostaining with vimentin and F11 antibodies for radial Muller glia reveal a disorganized pattern. Subsequently, radial glia processes organize into long parallel fibre bundles which are arranged like spokes to stabilize the surface and centre of the stratoid. The opsin-specific antibody CERN 901 detects photoreceptors as they gradually build up an outer nuclear layer at the surface. These findings assign to the RPE a decisive role for the genesis and regeneration of a vertebrate retina.

The neural cell recognition molecule F11 is a member of the immunoglobulin superfamily. During development it is expressed on axons in neuropils, and it is implicated in neurite outgrowth and in fasciculation. F11 is a multifunctional protein that interacts with L1/Ng-CAM, Nr-CAM, tenascin-C, tenascin-R (restrictin) and receptor protein tyrosine phosphatase beta. Here we show that F11 is not only expressed on neurons, but also on Müller cells and Schwann cells in vitro. Using two independent cell culture techniques, stratospheroids which represent an in vitro regeneration system of the chicken retina, and purified Schwann cells from chicken sciatic nerve, we show that Müller cells and Schwann cells express F11. On semithin cryosections of stratospheroids we were able to colocalize F11 with the Müller cell markers vimentin and 3A7. Expression of F11 by purified Schwann cells in demonstrated histologically and by immunoblot analysis. These in vitro data suggest the involvement of F11 in glia-neuron or glia-neurite interactions.

According to an earlier resented model (Layer and Willbold, Int. Rev. Cytol. 146: 1-47, 1993), growth of the retina can be conceived of as an areal increase of an epithelial tissue sheet ("lateralization") plus a concomitant establishment of the layered retina ("radialization"). To provide further support for this model, here we have reaggregated dissociated retinal plus pigmented cells from chick or quail embryos and observed their development into histotypic three-dimensional spheres in rotation culture. These so-called stratospheroids consist of a continuous fully laminated retinal part with a coiled-up pigmented epithelial core. Using BrdU-labeling, we show that radial growth, i.e. the sequential production of cell types in spheroids, is comparable to normal vitreal-scleral retinogenesis. The region next to the pigmented epithelial core represents a "lateral growth zone" (equivalent to an ora serrata in vivo), where mitotic cell numbers are highest, even when in the laminated part proliferation has already ceased. Gradients of lateral differentiation emanate from this growth zone into the retinal tissue, as revealed by immunostaining of the photoreceptor protein opsin and the cell recognition molecule F11. Moreover, we found that stratospheroids derived from older embryos consist only of a hollow monolayered neuroepithelium which develops in the absence of any radial growth. This indicates that cell production is sustained longer in lateral than in radial direction. These differently staged stratospheroids will be excellent models to characterize genes involved in the regulation of lateral and radial growth processes.

Radial columnar organization of cell clones is a characteristic feature of vertebrate retinae that is structurally not understood. Here we provide in vitro evidence that Müller glia processes stabilize cells within columns. Dissociated embryonic chick retinal plus pigmented cells regenerate in vitro into fully laminated stratospheroids. After reaggregating chick and quail cells, quail‐derived spheroid areas are detected as isolated sectors, as shown by a quail‐specific antibody. Each sector contains one or multiple cell columns. The radial borders separating chick and quail sectors are fully congruent with the extension of 3A7‐labelled Müller glia processes. While cell somata do not show any lateral interspecies mixing, quail‐derived neuropil extends within the inner plexiform areas far into chick sectors. After selective damage of Müller cells by the gliotoxin dl ‐α‐aminoadipic acid, the columnar organization is destabilized, as evidenced by a decrease in vimentin expression and by the migration of individual neurons out of their cell column. These data demonstrate that Müller cells actively stabilize cells within their columns, while neuritic growth is not hindered.