Background Using genetically modified mouse lines, we ablate genes and investigate their function in the development of retinal neurons, or we specifically describe the anatomy of retinal cell types, and then seek to manipulate a specific cell population and learn about its function by investigating the consequences of its loss. Specifically, we had previously generated conditional knock-in reporter lines for a family of three transcription factors, Brn3a, Brn3b and Brn3c, which are expressed in Retinal Ganglion Cells (RGCs). We now describe the use of these lines towards the definition of RGC cell types, their development, and their role in neuronal circuits. Results a) We have described in what RGC types Brn3 transcription factors are expressed, and how their deletion influences the development of RGC dendritic arbors at single cell level. Our finding suggests the existence of a transcriptional combinatorial code used for RGC cell type differentiation, comprising but not limited to the three Brn3 transcription factors. Surprisingly, this code might work by encoding various features of the RGC (dendrite area, lamination depth, axon formation), with the final RGC cell type being built by the appropriate combination of these features and the transcription factors encoding them (1). b) We further refined the cell type composition of intrinsically photosensitive RGCs (ipRGCs) expressing the photopigment Melanopsin (Opn4). In collaboration with the Hattar group in Hopkins, we defined two distinct ipRGC cell populations: (I) Opn4 (+), Brn3b (-): (II) Opn4 (+), Brn3b (+). ipRGCs expressing Melanopsin but not Brn3b belong exclusively to the M1 morphological cell type and project to the suprachiasmatic nucleus (SCN), responsible for circadian photoentrainment. M1 ipRGCs that express both Brn3b and Melanopsin send very few projections to the laterocaudal aspect of the SCN, however innervate the Olivary Pretectal Nucleus (which relays information necessary for pupil constriction reflex PLR), and to parts of the lateral geniculate nucleus and superior colliculus. Taking advantage of this knowledge, together with S-K. Chen and S. Hattar we genetically ablated specifically the Brn3b positive, Melanopsin positive neurons in mice, and demonstrated that in these animals PLR is almost completely removed, whereas circadian photoentrainment is unaffected (2). These results demonstrate that one neuronal cell population, having the same dendritic arbor morphology and molecular marker (melanopsin), can be separated into two cell types, based on the distinction of axonal targets in the brain (SCN vs. OPN), and their roles in two distinctive circuits (Circadian Photoentrainment vs. PLR). This raises new complexities for anatomists and physiologists involved in classifying neuronal populations, as it becomes increasingly clear that the definition of a cell type may require physiological as well as several layers of molecular and anatomical evidence. This study further supports our previously stated hypothesis that Brn3b might be involved in the development of axonal processes in RGCs. In this particular instance, the distinction in axonal targeting of RGCs to separate nuclei is brought about by a distinction in Brn3b expression. It is quite interesting to note that the light information for circadian phototentrainment, a visual behavior integrating light information over extremely long time periods is transmitted to the brain by a different channel than pupil constriction, a reflex which can adapt pupil diameter to light conditions within seconds. c) Since Brn3a, Brn3b and Brn3c are expressed in distinct but overlapping RGC populations, and since they appear to have different functions in RGC development, we explored the genetic interactions between the three transcription factors, by generating double knock-out lines for Brn3a Brn3b, Brn3b Brn3c and Brn3a Brn3c, and analyzed the effects on RGC survival and neuronal arbor formation using our reporter alleles. These experiments, carried out by M. Shi, S.R. Kumar and O. Motajo, reveal Brn3b independent functions of Brn3a, and document the Brn3a and Brn3c RGC cell populations surviving Brn3b ablation. These results are currently in preparation for publication. d) Using sparse genetic recombination techniques, S. Sajgo and M. Ghinia are in the process of describing the early developmental stages of Brn3 positive RGCs with single cell resolution. e) To understand the molecular requirements for neuronal arbor formation, M. Ghinia and S. Sajgo, are conducting a large-scale screen for genes specifically expressed in RGCs expressing Brn3a, Brn3b or Brn3c at ages relevant for neuronal arbor formation. f) As part of a collaborative project with the Kolodkin lab at Johns Hopkins, we helped identifying extracellular/transmembrane Semaphorins and their Plexin ligands as potential regulators of retina IPL lamination. The results of this effort, published this year (4,5) demonstrate very clearly that Semaphorins function as repulsive cues that help to delineate the very sharp lamination boundaries of the IPL. They do so via signaling through their receptors, Plexins, expressed on the neurites of retinal neurons. Therefore, in mice lacking either Semaphorin ligands, or Plexin receptors, the stratification levels of different neuronal arbors are shifted, resulting in abnormal lamination of the IPL, and in some instances functional defects at circuit level. Specifically, removing Sema 6A and its ligand, Plexin A4 from retinal neurons results in the partial displacement of axon arbors for Dopaminergic Amacrine cells as well as their synaptic partners, the dendrites of M1 melanopsin cells (4). In addition, in mice lacking Sema5A/5B or their receptors PlexinA1/A3, the OFF strata of the IPL are severely disrupted, with neurites of many amacrine, bipolar and retinal ganglion cells being shifted sclerad, and invading the Inner Nuclear Layer (INL), essentially creating a novel, abnormal plexyform layer in the middle of the INL. In contrast, the ON-OFF and ON strata of the IPL look relatively normal. S. Kumar, has analyzed multielectrode array recordings of RGCs from Sema5A/5B mutant retinas and determined that OFF, but not ON-OFF and ON RGC responses are dramatically reduced. Direction selective responses, mostly mediated by ON and ON-OFF RGCs were also preserved, and so was the ability of the mutant mice to track moving gratings stimuli (OKR responses) (5). Together, these studies strongly suggest that negative cues play a major role in setting up the correct lamination of the retina, and the correct co-stratification of the neuronal arbors that are synaptically connected. g) We are generating mouse lines carrying Cre recombinases to the three Brn3 loci, in order to be able to manipulate target genes in a RGC population specific manner. Brn3b and Brn3a are the earliest known RGC markers appearing very soon during RGC birth at E11 E12, and Brn3c targets a specific set of RGCs (work of M. Shi, M. Ghinia and S. Sajgo). Significance Taken together, this past year we have made significant advances in understanding how retinal layers are constructed, and how RGC types are being defined and develop to serve specific functions. Our insights into circadian photo entrainment and pupilary light reflex circuitry have implications for the neurology of diseases affecting these reflexes. The new genetic mouse lines we are in the process of generating and the gene expression profiles of RGC populations expressing Brn3a, Brn3b and Brn3c will help us understand their development and further dissect visual circuitry, thus opening the way for accurately targeted strategies for cell and gene therapies and the design of prosthetic devices.
Showing the most recent 10 out of 25 publications
