Microtubules are tubular filaments that form part of the eukaryotic cytoskeleton. They are important for cell division, shape and motor-driven transport. Axons and dendrites of neurons are filled with longitudinal arrays of microtubules. They serve as tracks for long-range cargo transport driven by the microtubule-based motors kinesin and dynein. This transport can be visualized in light microscopy but the structure of motile motors has not yet been observed inside cells. Similarly, the organization and ultrastructural features of neuronal microtubules remain unclear. Cryogenic electron tomography allows for visualization and structure determination of purified or cellular specimen in their native-like, vitrified state. I used this method in combination with subtomogram averaging to study intracellular microtubules and attempted to study dynein-driven transport. Starting with tomographic data of mouse axons and Drosophila melanogaster (Drosophila) neurites collected by my colleague Helen Foster, I implemented subtomogram classification workflows to determine the polarity and protofilament number of microtubules. This revealed that 13-protofilament microtubules are maintained within mouse axons, even close to sites of lattice damage. In contrast, Drosophila neurites contained mixed arrays of 12- and 13-protofilament microtubules that could transition between each other. Subtomogram averaging of globular microtubule inner proteins found within mouse axons showed that they form ring-like structures.

I next aimed to label dynein cargoes, with a focus on RNA granules, to target them by cryogenic correlative light and electron microscopy (cryo-CLEM). I describe targeting strategies and attempts at cryo-CLEM, most of which were unsuccessful. As an alternative, I established the sample preparation of human induced neurons and Drosophila S2 cells on electron microscopy grids to capture moving motors without specific targeting. I tested the use of microfluidics devices and optimized data processing workflows. I surveyed intracellular compartments from induced neurons and induced protrusions of S2 cells in the acquired tomograms, which revealed interesting features of unknown identity. An example are membrane-bound protein complexes resembling the outer membrane dome protein found in lamellar bodies of human cells.

While surveying compartments inside induced protrusions of S2 cells, I noticed rare occurrences of microtubule luminal filaments, which have similar appearance to luminal filaments found inside human induced protrusions. The luminal filaments accumulated upon inhibition of the sarco/endoplasmic reticulum Ca²⁺ ATPase (SERCA) with a small-molecule inhibitor. Subtomogram averaging suggests that they are formed of cofilin-bound actin (cofilactin). To confirm these results, I reduced cofilin levels using RNA interference and this led to a change in luminal filament morphology, detectable by visual inspection and subtomogram classification. Together, these results suggest that cofilin is involved in microtubule luminal filament formation.