edoc

Biophysical analysis of neuronal computations in the zebrafish olfactory forebrain

Rupprecht, Peter. Biophysical analysis of neuronal computations in the zebrafish olfactory forebrain. 2018, Doctoral Thesis, University of Basel, Faculty of Science.

[img]
Preview
PDF
116Mb

Official URL: http://edoc.unibas.ch/diss/DissB_12909

Downloads: Statistics Overview

Abstract

Higher brain function and cognition arise from associative computations that are most likely mediated by recurrent connections among distributed neurons. Due to its prominent recurrent connections and its proximity to the sensory periphery, the olfactory cortex is an ideal model to study these processes. The main focus of this work is the study of the biophysical properties of the olfactory cortex and how these properties affect computations like auto-associative processing.
The first part of this thesis describes methods for the acquisition and analysis of large-scale calcium imaging data. I developed a method for fast remote z-scanning in a two-photon microscope based on a voice coil motor, which allowed me record from more than 1500 neurons in the forebrain of adult zebrafish almost simultaneously. In addition, to infer spike rates from calcium imaging data, I developed a method based on deep artificial neuronal networks that surpassed the previous state of the art.
In the second major part of this work, I used electro- and optophysiological recordings in the zebrafish homolog of olfactory cortex, called Dp, to analyze its biophysical properties and how they affect the computations performed by the network. As a central result, I found that the posterior part of Dp (pDp) enters a transient state of precise balance upon odor stimulation. The classical "balanced network" is an established concept in theoretical neuroscience. Recent theoretical studies suggest that such a network could become more efficient with a more precise balance of excitatory and inhibitory inputs in time (tight balance) and coding space (detailed balance).
For the present study, I used whole-cell voltage-clamp recordings in the intact zebrafish brain to directly analyze synaptic inputs to neurons in Dp. Local silencing of activity by injection of the GABAA-agonist muscimol confirmed that Dp neurons receive strong inputs from recurrent connections within Dp. During an odor response, Dp neurons exhibited the hallmarks of a balanced state: strong and balanced excitatory and inhibitory inputs and a large synaptic conductance. Using the odor-evoked 20 Hz oscillations that originate in the olfactory bulb as a reference clock, I aligned excitatory and inhibitory inputs recorded in Dp and found a tight balance, with inhibition tracking excitation by a 3 ms delay. Finally, by studying the odor-specificity of excitatory and inhibitory inputs for a set of odor stimuli, I found inhibition and excitation to be co-tuned. These findings could not be explained by a purely random network, indicating that excitation and inhibition exhibit a detailed, high-dimensional balance in stimulus space.
Together, these experimental results show that Dp enters a balanced state during an odor response that is both tight and detailed, that is, precise. This suggests that this network could be the substrate for a pattern classification process that is fast, as in classical balanced networks, but also stable in many coding directions.
Advisors:Friedrich, Rainer and Rinaldi Barkat, Tania
Faculties and Departments:09 Associated Institutions > Friedrich Miescher Institut FMI
Item Type:Thesis
Thesis Subtype:Doctoral Thesis
Thesis no:12909
Thesis status:Complete
Number of Pages:1 Online-Ressource (99 Seiten)
Language:English
Identification Number:
edoc DOI:
Last Modified:14 Jan 2019 15:21
Deposited On:14 Jan 2019 15:21

Repository Staff Only: item control page