Digital staining based on machine learning models can provide cellular specificity to label-free optical imaging. This concept is particularly interesting for in vivo applications in fundamental research of auto-immune diseases as well as for future clinical translations. In this project MICS Multiphoton imaging with computational specificity, I will develop and implement computational specificity for label-free multiphoton microscopy (MPM) using artificial intelligence (AI). The direct outcome of this project will be two AI modules to perform (i) automated classification of mucosal inflammation based on 3D images from colon tissue and (ii) digital staining of un-stained immune cells. This integration of computational specificity to label-free multiphoton microscopy will allow direct investigation of global tissue alteration as well as specific immune cell localization during inflammatory tissue remodelling. Digital staining is an emerging concept in the field of computational microscopy but has not yet been implemented for immune cells based on label-free MPM images. Building on my previous expertise in label-free in vivo imaging via endomicroscopy, future implementations of multiphoton endomicroscopy would profit from tools for computational specificity, developed during this project.

NITRO-EARTH is aiming to investigate the hardly explored nitrogen chemistry of the alkaline-earth metals (Ae) in order to disclose new reactivity and catalysis. In the biogeochemical nitrogen cycle oxidized (NOx), neutral (N2) and reduced (NH3) forms of N are interconverted by a complicated network of processes. In contrast, manipulation of N in industry is challenging and often needs brute-force methods. The Haber-Bosch process for N2-to-NH3 conversion is, despite being metal-catalysed, one of the most energy consuming industrial processes. This proposal focusses on the organometallic chemistry of imido [RN(2)] and nitrido [N(3)] complexes of the alkaline-earth metals, in particular Mg and Ca. While the amide (R2N) chemistry of the Ae metals is well-established, Ae-imido complexes are rare and Ae-nitrido compounds solely exist as insoluble salts, e.g. Mg3N2. Given the importance of imido and nitrido ligands in transition metal chemistry, access to soluble Ae=NR and AeN complexes promises a rich reactivity and is the prelude of new catalytic processes based on abundant, generally biocompatible, alkaline-earth metals. The various pathways to reach the target include utilization of recently introduced, highly reducing Mg(0) complexes by HARDER and nitreones which have been investigated by PATEL. Also HARDERs recently discovered N2 fixation with Ca will play a role in the synthetic approach. Owing to the highly ionic character and negative charge on N in Ae=NR or AeN complexes, these novel complexes will be extremely potent nucleophiles or deprotonating reagents. This will be strongly dependent on nuclearity and aggregation which will be controlled by a library of bulky ligands currently available in the HARDER group. The work will be heavily supported by ab initio calculations. The project ultimately leads to the generation of a new class of alkaline-earth metal catalysts which may provide sustainable alternative for transition metal based catalysts.

The emergence of a new era in neuromodulation is led by the intriguing potential of functional materials to replace or control neural activity. The ability to simultaneously analyse neural activity offers the potential to translate signals into a feedback loop for intuitive therapy or even to replace lost neurological functions. However, neuromodulation and recording in the deep brain commonly relies on chronic implantation of macroscale hardware with numerous safety concerns and often suffers from poor spatiotemporal resolution. BRAINMASTER will demonstrate scalable, wireless, minimally invasive neuromodulation relying on forces transformed to mechanosensory neurons by magnetic nanodiscs (MNDs) coupled to external magnetic fields (MFs). Neuromodulation will run concurrently with magnetic resonance imaging (MRI) of Ca2+ transients. BRAINMASTERs ambitious objectives will permit cell-type specific interrogation (write) and simultaneous imaging (read) of deep brain in untethered subjects without implanted hardware, overcoming major challenges present in existing approaches. MNDs will be engineered to selectively target neural mechanosensitive ion channels by release of viral vectors for exogenous channel expression or by recognition motifs for endogenous stimulation. MND surface with Ca2+ binding moieties will allow dynamic MRI imaging via formation of ferromagnetic clusters translated as MRI contrast variations. The bidirectional BRAINMASTER interface will include MRI Ca2+ imaging simultaneous with stimulus from large gradient forces pulling MNDs on mechanosensory cells and torques mediated by low frequency MFs from miniaturized, MRI compatible coils. Ultimately, I will develop the first-of-its-kind intuitive interface between the deep brain and an engineered system to facilitate cognitive training and therapies for developmental, neurodegenerative and mental disorders and demonstrate the technological breakthrough in the rodent model of early Alzheimers disease.