The brain is a complex tissue, comprising myriad cell types and more than 109 cells and 1015 connections in humans. Through cell division, commitment, and differentiation, cells are generated and become the final cell types observed in the mature tissue. We are studying these processes in rodents and chicks, animals that exhibit patterns of development and final structures very similar to those of humans.
Retinal Cell Fate Determination
More than 20 years ago, we developed a lineage marking technique that we used to discover that the mitotic progenitor cells that generate the various neuronal and glial cell types of the retina are multipotent. Products of a final division, which produces two postmitotic daughter cells, can be as different as a specialized sensory neuron, a rod photoreceptor, and a nonneuronal cell type, a Müller glial cell. Recently we found an exception to this rule. A specialized type of interneuron, the horizontal cell, was found to be generated by a committed mitotic cell that made two horizontal cells of a particular subtype in a final division. This finding might be relevant to retinoblastoma, a tumor of the eye, where cells with horizontal cell properties appear to be the cell of origin. Other than this exception, however, it appears that most retinal neurons are made by multipotent progenitor cells. Thus, there is not a simple pattern of inheritance in which each type of neuron is produced by a particular type of progenitor cell dedicated to making only that type of neuron (or glial cell).
We have also done experiments that show that the environment can influence cell fate choices, but only in a limited sense. Moreover, we have found that the progenitor cells themselves change over time, in terms of their ability to divide and their ability to make different types of retinal cells. To determine what kind of gene expression changes underlie these changes, we have used genomics methods to profile the expression within single developing retinal cells. These data have revealed an enormous complexity in the gene expression patterns of single progenitor cells. We are investigating this complexity to try to understand how it reflects, or drives, the production of the many types of retinal neurons.
In addition to profiles from single progenitor cells, we have also profiled single developing ganglion cells, amacrine cells, bipolar cells, and Müller glia. These data have provided a wealth of new markers for these cell types and suggestions for transcription factors, and other genes, that are important in the differentiation of these cell types.
Patterning of the Retina
The retina is not a uniform sheet of cells, but exhibits various types of patterns. In humans and chicks, for example, there is a central region devoid of rod photoreceptors, referred to as the fovea in humans and rod-free zone in chicks. We have identified several genes that may control some aspects of retinal pattern. These genes are expressed asymmetrically across the retina of mice and chicks early in development when the patterns are set up. For example, Wnt2b, a secreted molecule, is expressed very early in development in the periphery of the eye. This expression pattern dictates formation of the most peripheral eye structures, the ciliary body and iris.
In addition to examining transcription factors and Wnt, we have been exploring the role of two small molecules, retinoic acid and thyroid hormone, which also appear to be regulators of pattern. Retinoic acid receptor signaling is upstream of the expression of a receptor that regulates the targeting of ganglion cell axons. It also regulates photoreceptor development, in part through cooperation with the thyroid hormone receptor. Thyroid hormone–dependent signals control multiple aspects of retinal development, including the overall timing of differentiation of photoreceptor cells. To visualize cells that respond to these molecules, we have developed plasmid-based reporters. These reporters are delivered in vivo or in vitro, where their spatial and temporal activation is monitored.
Photoreceptors in Development and in Disease
To characterize the development of photoreceptor cells, we have applied comprehensive methods for gene expression profiling to developing retinal tissue. We have done this with normal tissue, as well as with tissue from several mouse mutants in which photoreceptor transcription factors were knocked out. Our data have provided insight into the transcriptional networks that regulate photoreceptor development. These data are being used to model the cis regulatory elements that enable photoreceptor-specific gene expression. Many of the presumptive target genes are also disease genes that cause retinal degeneration in humans.
We are also using retinal microarrays to elucidate the mechanism of retinal disease. In many forms of human retinal degeneration, rod photoreceptors express a mutant gene, while cone photoreceptors do not. Interestingly, the cones still die. This implies a nonautonomous process in the death of cones. Humans rely most heavily on cones, and thus we would like to understand why the cones die, so that a pharmaceutical or genetic intervention can be developed. To this end, we have used microarrays to examine several mouse models of human disease to discover the nonautonomous process leading to cone death. These studies, and several other approaches, have revealed that the cones are starving, subsequent to the death of rods. We are now developing viral vectors to deliver genes that promote better nutrition within cones to promote their survival in a degenerating retina.
We have also found that delivery of the histone deacetylase gene, HDAC4, to degenerating rods will promote their survival. We have discovered that one of the targets of this deacetylase is the hypoxia-inducible factor-1α (HIF1α), a transcription factor that regulates many genes. Delivery of a stabilized version of HIF1α was found to promote survival of rods. We are now including these geneswithin viral vectors to determine if they can promote long-term survival of rods and/or cones in vivo in models of degeneration.
Electroporation as Means to Enable Studies of Gene Regulation and Function
Becausethe approaches described above identify genes that potentially play important roles in development and disease, we have developed methods for relatively rapid alteration of the expression levels of genes in vivo. Electroporation of plasmid DNA is an efficient and rapid method that allows one to overexpress genes, map regulatory elements, and/or use RNAi to reduce gene expression in vivo. We have created promoter constructs for most of the retinal cell types and have shown them to be specific for their cognate cell type following electroporation. In addition, we have extended the electroporation method to the embryonic period of development in rats and mice. As many as five plasmids have been shown to be delivered simultaneously at a high frequency, allowing us to investigate epistatic relationships. Recently these methods have allowed us to map discrete regulatory elements for cell-type–specific expression and have given insight into the core program that leads to bipolar cell-specific expression. They have also allowed us to develop reagents that mark discrete stages in photoreceptor development. These reagents can be used to probe these developmental stages, as well as to supply regulatory elements for viral vectors for gene expression in specific cell types.
