Development of adaptive-optics and multi-photon microscopy techniques for high-resolution deep-tissue imaging in the in-vivo mouse

Two-photon microscopy is currently the technique of choice for deep imaging in scattering, opaque specimen such as the in-vivo mouse, due to inherent optical sectioning and longer wavelength excitation light, which is generally less effected by scattering. However, the maximum penetration depth of two-photon microscopes is fundamentally limited by the on-set of out-of-focus fluorescence near the surface with increasing excitation power, which for the mammalian brain prevents imaging beyond ~1mm. Three-photon excitation fundamentally improves the depth limit due to a significantly increased signal-to-background ratio at depth and longer wavelength excitation. Unfortunately, optical aberrations stemming from the optical system and inhomogeneities within the sample lead to a degradation of resolution and contrast and loss of signal intensity at depth. However, aberrations can be corrected and near diffraction limited resolution recovered with so-called adaptive optics strategies. While two-photon microscopy has been combined with adaptive optics to correct for aberrations, very fine, sub- micron structures such as spines in the mouse brain are difficult to resolve with current methods in deep cortical or even sub-cortical brain regions. To tackle this challenge, in my PhD work I developed a custom three-photon microscope with integrated adaptive optics to increase the practical imaging depth and resolution for non-invasive in-vivo imaging of mouse tissue with the main focus on neuroscience application. In particular, I have designed and build a custom multi-photon microscope based on 1300nm excitation and shown its capability to image GFP-labeled neuron somata and even small structures such as dendritic branches up to a depth of 1.2mm in the intact mouse brain, which is among the best achievements demonstrated so far in the literature. At such large tissue depths, however, heart pulsation leads to brain motion and thus to intra-frame artefacts which prevent frame averaging to improve signal-to-noise ratio (SNR) of small structures. Therefore, we developed dedicated software and hardware to actively synchronize our image acquisition in real time to the cardiac cycle of the mouse. This improves SNR of small structure at depth without the need for sophisticated image registration techniques in postprocessing. Another main achievement of my work has been the development and integration of adaptive optics and its control software into our multi-photon microscope. Here I chose an indirect wavefront sensing approach which is more suitable for ultra-deep imaging. Together with our active motion-correction, our adaptive optics three-photon microscope enabled high, synaptic resolution imaging throughout an entire cortical column in the in-vivo mouse. In particular, we were able to improve (axial) resolution by ~3-fold and thus to visualize fine structures in the hippocampus, a sub-cortical brain region, at over 1mm depth. To further highlight potential applications of our method in the field of neuroscience, I have also performed proof-of-principle experiments in which I imaged the calcium dynamics of astrocytes, a cell type of the glia family, in the white matter of the intact mouse brain. These so-called fibrous astrocytes which are prevalent among myelinated nerve fibers in the white matter were so far, to best of our knowledge, not accessible for other non-invasive imaging methods. In summary, I have developed a motion-corrected adaptive-optics multi-photon microscope which enables intravital imaging at unprecedented depths and with near diffraction limited resolution. While most of my demonstrations were related to mouse neurobiology, I expect our new methods to find further applications in other fields such as mouse cancer and developmental biology