Title: A stochastic model for retinocollicular map development

Abstract: We present a theoretical model for retinocollicular map development, which can account for intriguing behaviors observed in gain-of-function experiments in knock-in mice by Brown et al., including bifurcation in heterozygous Isl2/EphA3 knock-ins. The model is based on known chemical labels, axonal repulsion/competition, stochasticity and uses Markov chain description. Our model suggests that the map in heterozygotes is single-valued in the temporal region of retina due to reduced gradient of ephrin in the corresponding region of SC. The remaining map is double-valued since the gradient of ephrin is high there. We predict therefore that if gradient of ephrin is reduced by a genetic manipulation, the single-valued region of the map should occupy a larger portion of temporal retina, i.e. the point of transition between single- and doulble-valued maps should move to a more nasal position in Isl2-EphA3 heterozygotes. We also discuss the importance of inhomogeneous EphA gradient and mapping in Isl2/EphB knock-ins.