In the womb, a ball of identical cells gives rise to varied cell types that ultimately form highly ordered structures and then the full panoply of organs in the human body. The process advances according to an internal biological script that directs each fold and crease of tissue to assume exactly the proper shape and dimension.

Scientists familiar with this progression from simple parts to a complex system have never stopped contemplating embryonic development with a sense of muted wonder and a concomitant desire to replicate early development on top of a laboratory bench—both to understand the biology better and to translate the information into ways of repairing and replacing damaged tissues. Their time may have come. Recent successes in deciphering the intricacies of development have raised the prospect of replacement organs grown outside the body arriving in surgical suites within as little as 10 years.

My own optimism about that prediction comes from recent studies in my lab on stem cells, which go on to diversify into other cell types. We showed that, even when grown in culture, stem cells can give rise to a retina, a key structure within the eye that translates light from the outside world into electrical and chemical signals that are then relayed to the rest of the brain. In other work, my colleagues and I have also grown cortical tissue and a part of the pituitary gland. In doing these experiments, we have taken advantage of our expanding understanding of the body's own innate signaling systems to coax a flat layer of disconnected cells in a petri dish to form a contoured, three-dimensional structure. Making use of chemical signals we provided, the stem cells took it on themselves, in essence, to build their own retina. This success spurs hope that retinal tissue produced by such methods can help treat several eye disorders, including macular degeneration.


When my lab began its attempts to grow a retina, we were trying to answer basic questions about how it forms. We knew that the retina emerges from a part of the embryonic brain called the diencephalon. During early embryonic development a segment of the diencephalon expands to form the optic vesicle, a balloonlike structure. The vesicle then folds inward to assume the shape of the optic cup, tissue that eventually morphs further to become the retina.

For more than a century biologists had debated the exact mechanism underlying the formation of the optic cup, a dispute that still lingers among scientists who study the developing brain. One looming question involves the role of neighboring structures, such as the lens and cornea. Some observers claim that the lens physically pushes the retina to bend inward, whereas others posit that the optic cup can form without the help of nearby lens tissues.

Seeing what is going on in live, developing animals is very difficult, and so about 10 years ago my group decided to see if we could learn more by isolating eye development—essentially by putting embryonic stem cells into a culture dish, exposing them to chemicals known to be involved in eye formation and watching what happened. Embryonic stem cells are the most immature type of stem cell and eventually go on to differentiate into all the body's diverse tissue types, from neurons to muscle cells.

No techniques existed for generating organs from stem cells in culture. Attempts to use these cells to build a new organ had seeded individual cells on an artificial scaffold shaped like a bladder or an esophagus. Tissue engineers had mixed success in growing actual organs with this technique. For that reason, we tried a different approach. As a prelude to this process, we devised a cell culture method in 2000 for turning mouse embryonic stem cells into various types of neurons. We then put a single layer of embryonic mouse stem cells on a culture dish, along with “feeder” cells that transmit chemical signals that prompt the stem cells to mature beyond their embryonic state. We understood that this flat sheet does not replicate the three-dimensional contours of actual human organs, but we wanted to see if the cells' own chemical signaling might be enough to prod them to generate special types of neurons characteristic of the eye's early development process.

We did not have much success at first, but in 2005 we achieved a technical breakthrough by inventing a way to move beyond the two-dimensional constraints of our lab's earlier stem cell technology and allow the stem cells to float in a culture solution. We started to use this three-dimensional culture, termed a floating culture, for a number of reasons. First, a three-dimensional aggregation of cells can better generate complex tissue topology than one formed in a flat sheet. Second, one cell needs to communicate with another to develop into a complex structure, and a three-dimensional culture is more suitable for promoting such communications because cells can more flexibly interact with one another.

Applying this new method, we suspended separate cells in a tiny amount of a liquid medium in wells in a lab dish and found that they began to bind together with their well mates. These small cell aggregates, typically 3,000 cells per well, could then be coaxed to differentiate into the same kind of neural progenitors (immature neural cells) that populate the front of the brain. The cells then started to signal to one another and, after three to four days, spontaneously organized into a hollow sphere formed by a single-layer cell sheet, a neuroepithelium. We called this method of making the sheets a SFEBq (serum-free floating culture of embryoid body–like aggregate with quick reaggregation) culture.

In the embryo, neuroepithelial cells eventually form specific brain structures after they receive external chemical signals from outside the cells. One of these signals triggers development of the diencephalon, which later gives rise to the retina and the hypothalamus (the brain region that controls appetite and a number of other basic bodily functions). Once we had gotten cells to form spheres in the lab, we attempted to induce the constituent cells to differentiate into retinal progenitors— precursors of mature retinal cells—by adding a cocktail of proteins (containing the chemicals that perform the same task in the embryo) to the SFEBq culture.

After these spheres remained in a floating culture for several more days, the retinal epithelial tissue spontaneously protruded outward, or evaginated, and formed optic vesicle–like structures. Moreover, the vesicles spontaneously changed shape: the epithelial part on the outside of the main body of the sphere folded inward. This movement generated a brandy glass–like shape resembling the optic cup of the embryonic eye. As seen in live animals, the optic cup derived from embryonic stem cells consisted of two walls: the outer epithelial wall and the inner wall of the actual retina.

In other words, aggregation of dissociated stem cells in a culture dish alone resulted in an ordered structure—a literally eye-popping result. Unlike in the embryo, no lens or cornea formed next to the optic cup. This finding gave a clear answer to the long-standing question of whether or not this protoretina requires external forces from neighboring tissues such as lens cells. Retinal formation, at least in vitro, is a self-organizing phenomenon based on an internal program that resides within these cells.

Layering a Retina

Normal development as seen in embryos continued to proceed in the lab dish. When we subjected the optic cup to an additional two weeks in a floating culture, the tissue grew to approximately two millimeters in diameter, with the single-layered retina epithelium becoming, as in the embryo, a stratified structure containing all six categories of cells found in the postnatal retina. The laminated material contained an outer photoreceptor layer and an innermost layer of ganglion cells, which, in the body, connect the retina to the brain. In between, as would be expected in a true retina, were several connecting layers of cells called interneurons. As before, the formation of multiple layers occurred through an internal program that guides what kinds of cells to make and how to arrange them in a three-dimensional space.

Our work is not over. Questions still persist as to how the optic cup forms and how a ball of cells generates a patterned structure. Spontaneous emergence of intricate shapes from a homogeneous clump of matter is known as symmetry breaking and occurs throughout embryonic development. If it were not for symmetry breaking, repeated cell divisions of the fertilized egg would not progress beyond an undifferentiated mass of cells. Our self-organizing embryonic stem cell culture appears to serve as an ideal experimental platform for understanding the elusive mechanisms of this process during mammalian embryogenesis.

Another looming question relates to how the retinal epithelium, initially just a sheet of cells, programs the shaping of the optic cup. In general, mechanical force and stiffness control changes in epithelial tissue. By measuring the direction of force and tissue stiffness in different parts of the epithelium during formation of the optic cup in vitro, we found three steps that lead to formation of tissue structure. As the optic cup forms, stiffness in the retina diminishes, increasing flexibility. At the same time, cells at the junction of the retina and the epithelium assume a wedge shape, and finally the retina begins to fold inward because of its rapid expansion. These three steps are critical for optic cup formation. In fact, when these conditions related to tissue mechanics were introduced in a computer simulation, the familiar brandy glass shape emerged.

To See Clearly

Of course, people who hear of our research want to know whether work on mouse embryonic stem cells will eventually help humans with eye ailments. We have made some progress in that direction. Notably my lab very recently reported successful formation of an optic cup and multilayered neural tissue derived from human embryonic stem cells. It is also expected that the same culture method should be applicable to human induced pluripotent stem cells—mature cells that were prodded to go through a reverse development process that allows them to behave like embryonic stem cells. We have also invented an improved cryopreservation method that can reliably store human embryonic cell–derived retinal tissue in liquid nitrogen.

All of this work will propel us toward medical applications of retinal tissue. For instance, we may be able to create artificial retinas that help researchers explore the pathology of common eye diseases, perhaps leading to the development of drugs and gene therapy to reverse retinal degeneration.

Three categories of retinal degeneration that might benefit from our research—macular degeneration, retinitis pigmentosa and glaucoma—affect millions of people worldwide. Each disease causes problems in different layers of the retina. In macular degeneration, the integrity of the epithelium is impaired by the breakdown of supporting tissue, and this breakdown leads to the deterioration of photoreceptors, particularly in the central region of the retina. In retinitis pigmentosa, the number of the photoreceptors called rods decreases gradually over many years. “Night blindness” appears as the first common symptom. Later, the patient loses most of the visual field except for a small area at the center. Finally, glaucoma damages ganglion cells, which connect the retina to the visual-processing center in the cortex at the back of the brain through projecting optic nerves.

Macular degeneration seems the most amenable of the three to being eased by cell-replacement therapy. Human embryonic stem cells and induced pluripotent stem cells can generate the support tissue, known as retinal pigment epithelium, relatively easily when grown in conventional culture as well as by our method, and cells can be retrieved directly from these cultures. Early small-scale clinical trials with these cells have already started in the U.S., and similar trials are planned in other countries. In these studies, stem cell–derived pigment epithelial cells are injected with a fine needle into the space between the pigment epithelium and photoreceptor layers to replace at least part of the damaged tissue.

Cell therapy for retinitis pigmentosa requires additional technical advances before it can be offered to humans. Our technique, unlike a conventional culture, can generate rod photoreceptors in a cell-dense sheet suitable for transplantation, but we need another critical tool before transplants of such sheets can improve vision. Unlike the simple support tissue of epithelium, photoreceptors need to integrate into the eye's neural circuitry; specifically, they need to reconnect to another type of sensory cell, a bipolar cell, and we do not yet know how to make that linkage happen efficiently. Transplantation of photoreceptors, if successful, would be expected to enable those with even advanced retinitis pigmentosa to recover at least some of their vision.

Glaucoma may be the most difficult of the three diseases to treat through cell therapy. Embryonic stem cell cultures are capable of generating ganglion cells needed for this endeavor. In the postnatal eye, however, optic nerve regrowth is suppressed, and no one has yet figured out how to induce their axons (the branches that send signals into the brain and that form the optic nerve) to reconnect with other cells.

We have learned that embryonic stem cell–derived tissues can do much more than we can currently achieve through artificial tissue engineering in which cells are placed on scaffolds shaped like a layer of skin or a bladder. As researchers, we must humbly and patiently uncover what developing cells can teach us about the intricate processes that lead from a single cell to an organ as complex as the eye.