Preview

PDF, English Download (19MB) | Terms of use

Abstract

The goal of this work was two-fold: 1) To apply serial block-face electron microscopy (SBEM) to the spinal cord of a larval zebrafish, in order to gain a mechanistic understanding of motoneuron (MN) recruitment, based on a reconstruction of the wiring between spinal interneurons and MNs and 2) to implement technological improvements to SBEM that would allow datasets to be acquired at much higher speed, in order to acquire a dataset of a complete larval zebrafish brain. The spinal cord of vertebrates contains a neural circuit known as a central pattern generator (CPG), which can generate the rhythmic muscle contractions underlying locomotion independently of the brain. In fish, the rhythm consists of muscle contractions that alternate between the left and right side of the tail and that travel down the length of the fish, from head to tail. When swimming fast, such as during escapes, the rhythm has a high frequency and muscles contract vigorously. During slow, routine swimming, the rhythm has a low frequency and muscles contract with less strength. The MNs in the spinal cord, which elicit the contractions of the tail musculature, are recruited to different degrees during these different behaviors. With increasing contraction strength, more and larger MNs are activated. This phenomenon is called orderly recruitment. The rhythmic excitation that recruits MNs is provided by Circumferential Descending (CiD) interneurons located in the spinal cord. These interneurons also follow a specific recruitment pattern: During weak swimming, ventral cells are active exclusively and dorsal cells are silent. As swims increase in vigor, the activity in these cells shifts towards more dorsal cells, with more ventral cells becoming inactive. The aim of the first part of this thesis was to reconstruct the MNs along with the CiDs that excite them, using a high resolution SBEM dataset of the spinal cord, to identify the pattern of connectivity between these types of neurons and distinguish between competing hypotheses of orderly MN recruitment. Conceptually, orderly recruitment could either be implemented with unspecific connectivity, in which case it would be a consequence of the interplay of size-dependent biophysical properties (in particular the input resistance) with the strengths of the synapses driving them. Alternatively, the wiring pattern could be specific and the CiDs could select the subset of MNs to activate by making synapses with just those cells. MNs in the larval zebrafish spinal cord clustered into distinct subtypes, depending on their size: Small, intermediate and large. The small MNs received almost no synaptic inputs and appeared to be immature. CiDs differentially innervated the intermediate and large MNs: Ventrally located CiDs did not differentiate between the two subtypes, but the dorsal CiDs made synapses onto large MNs with high specificity. Since dorsal CiDs are active only during the fastest swims, this finding can be interpreted as a labeled line specifically recruiting the strongest MNs during the most vigorous behaviors. During weaker behaviors, when the dorsal CiDs are inactive and the more ventral ones are active exclusively, differences in MN excitability due to size would encode the recruitment order. The second objective was to improve SBEM technology to acquire a whole larval zebrafish brain in a relatively short period of time. Due to the very high resolution required to trace small neurites and to identify synapses, even very small brains, such as the brain of a larval zebrafish, would take many months to acquire using a typical SBEM setup. Two main techniques were used to increase net speed. First, line-scanning of individual image tiles was implemented, where the electron beam scans the image in one axis only and the other axis is scanned by moving the stage. This allows larger individual images to be taken, greatly reducing the number of motor moves between images. Second, dynamic adaptation of the image tile mosaic to the shape of the sample was used to avoid scanning the blank plastic regions surrounding an irregularly shaped sample. These improvements allowed the complete brain of a 5 day old larval zebrafish to be imaged in less than 30% of the time than would have been required previously. In a collaborative project with Dr. Fumi Kubo, two-photon calcium imaging was performed prior to EM imaging, revealing pretectal cells active during optokinetic stimulation. The two-photon dataset was successfully registered to the EM data and a functionally identified pretectal cell could be traced. This dataset will be used to reconstruct the complete neural networks that compute the optokinetic response.