Topographic maps, in which neighbour relations between neurons are conserved over a projection between brain areas, are a recurrent feature of visual systems. By examining the retinal ganglion cell (RGC) synapses to the zebrafish tectum at the population level, I have explored hitherto inaccessible questions regarding development, refinement and alignment of distinct maps within the same target field. Given that functional types of RGC stratify into laminae within the tectum, what are the developmental dynamics and parameters governing these maps and alignment between them? Do they encode visual space with the same precision? The parameters governing axonal arbour refinement vary across species, likely related to differences in nurture and environmental pressures. In the zebrafish, the situation is complicated by mismatched growth of the tectum with respect to the retina, necessitating constant remodelling of retinotectal connectivity. Although topography is maintained during this process, it is not known how precise this topography is. How does map precision change in response to experience during development? This question is particularly interesting given recent theoretical results suggesting that perfect topography may not be optimal for decoding the visual world. These, and other salient issues, are explored in greater detail within my introduction (Chapter 1). Quantifying the precision of topographic mappings is a non-trivial problem, so metrics require empirical validation. Chapter 2 is a comparative study of various metrics for topographic precision using in-silico modelled data, and particularly focusing on quantification of maps derived from multiple experimental subjects. Metrics were compared on their ability to discriminate different levels of order, their resistance to global shape distortion, and the amount of data they required to perform optimally. Having selected one metric for further development, a statistical framework for testing differences in order between multiple-subject datasets was derived. Finally, a method for interpreting sources of topographic disorder in biologically relevant distance units was developed. I constructed a novel visual presentation system, maximising visual coverage and resolution, the details of which are described in Appendix A. Using simultaneous visual presentation and confocal imaging of a calcium indicator in RGC axon terminals allowed characterisation of RGC functional selectivity. In Chapter 3 I explored the development of two feature-selective topographic maps formed by retinal ganglion cells (RGCs) in the tectum. I focused on 3 orientation-selective (OS) and direction-selective (DS) RGCs, pooling data from fish of different ages (3, 7 and 10 days post fertilisation, dpf) into standardised anatomical spaces. Such experiments revealed nonuniform, nonmatching coverage of the tectum by OS and DS RGCs, suggestive of regional specialisation. DS and OS maps also exhibit differing levels of topographic order, with the DS map more ordered than the OS at all ages. For both functional types, order changes nonlinearly during development, with maps at 10dpf less ordered than those at 7dpf, providing empirical evidence to the notion that a topographic encoding of visual space may be non-optimal. Finally, in Chapter 4, fish were reared in altered visual environments in order to specifically explore the role of visual experience in the development of topographic maps formed by DS and OS RGCs. Three different conditions were used: complete darkness, an enriched/naturalistic environment, and strobe lights. Functional imaging of topographic maps in 10dpf animals reared in these conditions suggests complex relationships between visual experience and map properties, including the representation of the cardinal axes of visual space, and overall map precision. Visual experience affects feature-selective maps differentially, with a far greater impact on OS maps than DS. My thesis offers insight into how feature-selective topographic maps in the tectum develop. It suggests that, despite a globally uniform density of RGCs in the retina, regions could be specialised for the detection of specific features, although specialisation is dependent on the visual environment of the animal as it develops. This is of great interest to the community, raising questions about how animals might behaviourally maximise the matching of retinal biases with the statistics of their environment. Finding that topographic order does not necessarily increase with age adds evidence for the hypothesis that topographic maps are not directly used in decoding the visual scene.
|Date of Award
|1 May 2019
|Jon Clarke (Supervisor), Gareth Barker (Supervisor) & Andrew Lowe (Supervisor)