Multimodal Approaches to Investigate Neuronal Activity Using Optical Imaging Techniques and High-density Microelectrode Arrays

Abstract

This thesis presents the concepts, development, and applications of two novel multimodal approaches to studying neuronal activity that combine high-density microelectrode arrays (HD-MEAs) and optical imaging techniques. The aim of this work was to exploit the advantages of combinations of HD-MEAs, calcium imaging, and voltage imaging, versatile and powerful tools that facilitate a wide range of investigations in the field of neuroscience. HD-MEAs, based on complementary metal-oxide-semiconductor (CMOS) technology, have emerged as a promising approach for electrophysiological studies that facilitate the understanding of brain functions and the profiling of neurological disorders. Modern CMOS-based HD-MEAs feature thousands of electrodes enabling high-yield recordings at high spatiotemporal resolution. However, certain limitations of HD-MEAs, such as the inability to detect sub-threshold activity and challenges associated with assigning spikes to specific neurons (`spike sorting'), constrain the scope of investigations when employing HD-MEAs alone. Optical imaging techniques, such as calcium imaging and voltage imaging, offer unique features in assessing neural activity. Calcium imaging can be used to measure distinct neuronal events, including individual synaptic activation, at exceptional spatial resolution. Voltage imaging features fast kinetics and has great potential to simultaneously detect membrane potential changes in multiple neuronal compartments. In combination with large-scale HD-MEA recordings, optical methods provide powerful tools for capturing highly complex activity dynamics at single-neuron and neuronal network level. The first multimodal approach introduced in this thesis is based on the combination of large-scale HD-MEA recordings and dendritic spine calcium imaging. While subcellular-resolution calcium imaging gives access to evoked postsynaptic signals, HD-MEAs allow for recording precise spiking sequences of large-scale networks. We developed a novel and data-driven analysis pipeline to infer synaptic connectivity using data acquired through such combination recordings. Furthermore, the analysis toolkit and the data sets were made publicly available to promote a collaborative environment for further exploration by the research community. The second multimodal approach presented in this thesis combines HD-MEA recordings and voltage imaging of neuronal populations. We employed this method for high-throughput validation of spike sorting algorithms, which enabled a comprehensive evaluation and enhancement of spike sorting accuracy and robustness, crucial features of extracellular data analysis. A novel and scalable analysis pipeline was developed to extract ground-truth spiking data from multiple units using voltage imaging and to validate spike sorters using the acquired ground-truth data. The novel multimodal approaches developed in this thesis represent significant advancements for investigating synaptic connections and refining electrophysiological data analysis. In addition, these newly established methods facilitate the exploration of a wide range of other research topics in experimental and computational neuroscience, such as synaptic plasticity and the construction of multi-compartment models of individual neurons. Ultimately, these approaches will contribute to a more comprehensive and systematic understanding of fundamental neuroscientific questions, which paves the way to the development of targeted interventions and therapies to combat neurodegenerative diseases.