Logo Logo
Help
Contact
Switch language to German
A search for functional connectivity rules in the visual thalamus and hippocampus
A search for functional connectivity rules in the visual thalamus and hippocampus
Computations in the brain arise from the functional connectivity of individual neurons, brain regions and circuits. Understanding the fundamental connectivity rules in the brain is an important step to better understand the brain itself. In this dissertation, a suite of methods and approaches were employed to investigate such rules at the synaptic and circuit level. The first part of this dissertation dealt with the conflicting results of recent structural (Rompani et al., 2017) and functional (Howarth, Walmsley, & Brown, 2014; Jaepel et al., 2017; Sommeijer et al., 2017) reports regarding the degree of functional binocular convergence in the dorsal lateral geniculate nucleus (dLGN). To address this, a novel dual-color optogenetic assay was developed to map functional connectivity between RGCs and individual dLGN cells in vitro. While structural convergence is large, with > 60 % of dLGN cells receiving binocular input, the dLGN is functionally monocular: not only did the dominant eye provide > 95 % of a dLGN cell’s retinogeniculate input, but the non-dominant eye was unable to elicit the firing of action potentials under resting conditions. Analysis of dLGN cell morphology in relation to the axonal input pattern of RGC afferents in the dLGN revealed axo-dendritic overlap could not explain the levels of functional monocularity observed in the in vitro assay. Instead, the dominant and non-dominant eye differed with regards to the expression of AMPA and NMDA receptors, seemingly favoring the dominant eye: fine-scale input selection and refinement were found to limit the functional convergence in the retinogeniculate pathway, resulting in a winner-takes-all wiring rule in this part of the visual circuit. In the second part of this dissertation a deep learning tool for the detection of dendrites and dendritic spines, termed DeepD3, was developed. DeepD3 directly addresses the current need for automated methods of spine detection. Unlike other areas of neuroscience, where data collection and analysis throughput has improved considerably in the last years, most studies to date still only investigate dozens of dendritic spines per neuron. The analysis - such as identification or segmentation of dendritic spines in image data - represents the main bottleneck in current analysis efforts. DeepD3 was tested against a number of in vitro and in vivo datasets with varying image properties to ensure that this method performs well in a large range of data qualities. DeepD3 performed as well as human in both in vivo and in vitro data. Importantly, DeepD3 fully processes large datasets within hours, a procedure which would take months if done via the current gold standard, human annotation. DeepD3 can be flexibly employed for counting dendritic spines or to measure 2D, 3D or time-series fluorescence values of spines and dendrites. The third part of this dissertation aims towards understanding the functional connectivity rules of LTP-induced dendritic spines. Dendritic spines grow in neurons undergoing LTP (Engert & Bonhoeffer, 1999; Toni et al., 1999) and rapidly form functional synapses (Nägerl et al., 2007). However, it remains unclear which presynaptic neurons are chosen to establish functional connectivity, and hence whether there is a fundamental wiring rule followed by the brain. To address this, improvements to an existing assay to map functional synaptogenesis following LTP in vitro (Coneva, 2015) were made. By modifying the timeline of the assay and introducing high-throughput volumetric calcium imaging methods, the throughput of the assay was improved several-fold. Moreover, a molecular approach was devised to combat a potential confounding variable, the lack of spine maturity in nascent spines, when determining functional connectivity rules of LTP-induced dendritic spines. Lastly, several pharmacological and computational means were tested in their ability to assess functional connectivity on a single-spine level despite the occurrence of dendritic calcium events, which otherwise prevent such assessments. Neither the pharmacological, nor the computational methods applied proved effective in this undertaking. As a consequence, determining functional connectivity rules of nascent, LTP-induced dendritic spines remains outstanding. This dissertation paved the way for attempts of this undertaking in the future.
Not available
Fernholz, Martin
2023
English
Universitätsbibliothek der Ludwig-Maximilians-Universität München
Fernholz, Martin (2023): A search for functional connectivity rules in the visual thalamus and hippocampus. Dissertation, LMU München: Faculty of Biology
[thumbnail of Fernholz_Martin.pdf] Licence: Creative Commons: Attribution-NonCommercial 4.0 (CC-BY-NC)
PDF
Fernholz_Martin.pdf

14MB

Abstract

Computations in the brain arise from the functional connectivity of individual neurons, brain regions and circuits. Understanding the fundamental connectivity rules in the brain is an important step to better understand the brain itself. In this dissertation, a suite of methods and approaches were employed to investigate such rules at the synaptic and circuit level. The first part of this dissertation dealt with the conflicting results of recent structural (Rompani et al., 2017) and functional (Howarth, Walmsley, & Brown, 2014; Jaepel et al., 2017; Sommeijer et al., 2017) reports regarding the degree of functional binocular convergence in the dorsal lateral geniculate nucleus (dLGN). To address this, a novel dual-color optogenetic assay was developed to map functional connectivity between RGCs and individual dLGN cells in vitro. While structural convergence is large, with > 60 % of dLGN cells receiving binocular input, the dLGN is functionally monocular: not only did the dominant eye provide > 95 % of a dLGN cell’s retinogeniculate input, but the non-dominant eye was unable to elicit the firing of action potentials under resting conditions. Analysis of dLGN cell morphology in relation to the axonal input pattern of RGC afferents in the dLGN revealed axo-dendritic overlap could not explain the levels of functional monocularity observed in the in vitro assay. Instead, the dominant and non-dominant eye differed with regards to the expression of AMPA and NMDA receptors, seemingly favoring the dominant eye: fine-scale input selection and refinement were found to limit the functional convergence in the retinogeniculate pathway, resulting in a winner-takes-all wiring rule in this part of the visual circuit. In the second part of this dissertation a deep learning tool for the detection of dendrites and dendritic spines, termed DeepD3, was developed. DeepD3 directly addresses the current need for automated methods of spine detection. Unlike other areas of neuroscience, where data collection and analysis throughput has improved considerably in the last years, most studies to date still only investigate dozens of dendritic spines per neuron. The analysis - such as identification or segmentation of dendritic spines in image data - represents the main bottleneck in current analysis efforts. DeepD3 was tested against a number of in vitro and in vivo datasets with varying image properties to ensure that this method performs well in a large range of data qualities. DeepD3 performed as well as human in both in vivo and in vitro data. Importantly, DeepD3 fully processes large datasets within hours, a procedure which would take months if done via the current gold standard, human annotation. DeepD3 can be flexibly employed for counting dendritic spines or to measure 2D, 3D or time-series fluorescence values of spines and dendrites. The third part of this dissertation aims towards understanding the functional connectivity rules of LTP-induced dendritic spines. Dendritic spines grow in neurons undergoing LTP (Engert & Bonhoeffer, 1999; Toni et al., 1999) and rapidly form functional synapses (Nägerl et al., 2007). However, it remains unclear which presynaptic neurons are chosen to establish functional connectivity, and hence whether there is a fundamental wiring rule followed by the brain. To address this, improvements to an existing assay to map functional synaptogenesis following LTP in vitro (Coneva, 2015) were made. By modifying the timeline of the assay and introducing high-throughput volumetric calcium imaging methods, the throughput of the assay was improved several-fold. Moreover, a molecular approach was devised to combat a potential confounding variable, the lack of spine maturity in nascent spines, when determining functional connectivity rules of LTP-induced dendritic spines. Lastly, several pharmacological and computational means were tested in their ability to assess functional connectivity on a single-spine level despite the occurrence of dendritic calcium events, which otherwise prevent such assessments. Neither the pharmacological, nor the computational methods applied proved effective in this undertaking. As a consequence, determining functional connectivity rules of nascent, LTP-induced dendritic spines remains outstanding. This dissertation paved the way for attempts of this undertaking in the future.