Abstract

The most prominent feature of the human brain is its large and folded cortex. The cortex is a laminar sheet of tissue that is organized into several layers and believed to control high level cognitive functions. During evolution, the cortex of many mammals expanded laterally much more compared to its increase in thickness. Since the surface area of the surrounding skull expanded less, the only way to fit the large cortical surface into the limited space was to fold it. Thus, today we can see gyrencephalic (folded, e.g. human) and lissencephalic (non-folded, e.g. mouse) mammalian brains. The precise mechanisms guiding cortical folding are still largely unknown. During brain development many neu- rons travel long distances from their birthplace to their final destination. To form the six layered structure of the cortex, newborn neurons migrate from the ventricular and subven- tricular zone all the way out into the cortical plate. The timing, speed, travel distance and direction of migration determines the neuron’s final location and thus the overall morphology of the cortex. To date, all experimental evidence suggested an increased number of neural progenitors as the main cause of folding. However, several theoretical models proposed that cortical folding can be induced by rearranging the same number of cortical neurons. The distribution and organization of cells depend on a balance of intercellular adhesive and repulsive signalings. The Fibronectin Leucine-Rich Transmem- brane (FLRTs) family of proteins can provide both signals by functioning as adhesive molecules and as heterotypic chemorepellents. The FLRTs (1-3) are regulators of early embryonic vascular and neural development. Mice with deletions of Flrt1 and Flrt3 show evidence that, cortical folding can be induced without increasing the number of neurons or neural progenitor cells. Instead removal of Flrt1 and Flrt3 alters neuronal migration and distribution. We consistently found bilateral clustering of mutant neurons normally destined to express FLRT3 in posterior cortical regions, which mostly coincided with the location of folds. Interestingly most folds appeared unilaterally on the left side in Flrt1/3 DKO suggesting that cortical asymmetries favour folding on one side. Moreover, live imaging of embryonic cortex slices showed that neurons in Flrt1/3 DKO mice were more likely to reach high migration speeds. The higher migratory speed was confirmed in vivo where more neurons reached the upper cortical plate too early and failed to mature their dendritic trees. Based on the observation that both an increase and reduction in FLRT expression in the developing cortex induces neuron clustering, simulations of neuronal migration showed that a tight balance between cell adhesion and repulsion is required for concerted neuronal migration. Simulations of neuronal migration with reduced cell ad- hesion favours the formation of a wavy cortical surface and may thus occasionally induce folding. In addition, we found reduced levels of Flrt1/3 mRNA in humans and future sulcus regions of ferrets, suggesting that the fine-tuned balance between attractive and repulsive forces is a key regulator of folding. The mammalian retina has the remarkable ability to dissect the visual scene into distinct streams of information encoding color, luminosity, motion and contrast. Each stream is integrated by a subtype of cells in the ganglionic cell layer (GCL) and trans- mitted via parallel pathways to the visual centers in the brain. FLRT3 has been shown to act as a controlling factor of retinal vascular development. In order to distinguish vascular versus neuronal functions of FLRT3, we aimed to compare retinas totally de- pleted of FLRT3 by using SOX2-Cre with retinal ganglion cell (RGC)-specific Brn3b-Cre knock-out. Our results show that full depletion of Flrt3 results in cataract formation, eye malformation, blindness and in one case specific loss of RGCs. Interestingly, all these phenotypes are not present when FLRT3 is removed specifically from RGCs, suggesting that its e↵ects on the vascular system have a crucial role during retina and eye formation. Analysis using genetic markers showed that all FLRTs (1-3) are also expressed in the retina postnatally and mainly in a subset of retinal ganglion cells (RGCs). Therefore, we asked whether FLRTs could represent potential markers for a functional subpopulation of that cell type. Functional analysis has identified more than 30 di↵erent RGC sub- types in the GCL so far. However, many of the functionally identified subpopulations lack a genetic marker to target these cells for full characterisation of each subpopulation. Histochemical studies showed expression of FLRT3 in a specific cell population within the GCL. Quantification revealed that 23% of the FLRT3+ cells are RGCs while the remaining 67% are displaced amacrine cells. The FLRT3+ RGC cell population rep- resents only 6% of all RGCs. However, non-mosaic like FLRT3 RGC distribution and varying stratification depths within the inner plexiform layer revealed that the FLRT3+ RGC population consists of at least 6 morphologically defined subpopulations. The two biggest groups of FLRT3+ RGCs, which represent 78% of all reconstructed cells stratified within the ON layers 7-10, suggesting that those cells might react to increases in light intensity. Notably, 2 groups of ON-OFF bistratifying RGCs in layers 3/4 and 7/8 and a small population stratifying in the OFF layer 2 were identified. ON-OFF RGC spe- cific CART immunostaining confirmed that 23% of all FLRT3-RGCs are indeed ON-OFF RGCs. Finally, 9 out of 83 RGCs showed a very di↵use stratification pattern. Retinofugal projection analysis of the whole FLRT3-RGC population showed no subregion specific targeting in the superior colliculus or ventral/dorsal lateral geniculate nucleus (LGN), which is usually found for ON, OFF or ON-OFF direction selective ganglion cells. This finding confirms the result that FLRT3+ RGCs consist of more than one functional sub- population. Moreover FLRT3-RGCs innervate several nuclei of the accessory optic tract, which is important to control retinal image stabilisation. Interestingly FLRT3+ RGC projections were found in the medial terminal nucleus (MTN), which is the main target of ON direction selective ganglion cells (ON-DSGCs), thereby suggesting that at least a fraction of the FLRT3+ RGCs are ON-DSGCs. FLRT3+ RGCs completely avoided the intergeniculate leaf (IGL) and the suprachiasmatic nucleus (SCN), which excludes FLRT3-RGCs from being involved in circadian entrainment. Overall data supports the idea that FLRT3-RGCs are a mixtue of mostly ON but also some OFF and ON-OFF RGCs, which might be important for image stabilization.

Abstract