Multimethodological approaches to tackle glycogen: therapeutic and modelling opportunities in Lafora disease

Protein aggregation is a key pathological hallmark of many neurodegenerative diseases. However, such a paradigm does not universally prevail. Indeed, some neurological glycogen storage disorders manifest through the accumulation of aberrant glycogen into polyglucosan bodies (PGBs). These include Lafora disease (LD), a lethal myoclonus epilepsy, caused by loss-of-function mutations in either EPM2A or NHLRC1 genes encoding laforin or malin, respectively. In physiological conditions, malin ubiquitinates its substrates, among which protein targeting to glycogen (PTG), in a laforin-dependent manner. PTG emerged as a master regulator of brain glycogen synthesis. In LD, PTG escapes degradation due to the dysfunctional laforin-malin complex, leading to the accumulation of PGBs and neurodegeneration. Presently, the landscape for LD remains barren of efficacious treatments, cellular models, or clinically relevant targets. In this window of opportunities, two parallel strategies have been investigated. On one hand, a reverse chemogenomic approach has been employed with PTG as a therapeutic target. The protein structure of the PTG carbohydrate-binding motif 21 has been experimentally determined, the crystallisation condition has been optimised, and molecular docking was performed resulting in suboptimal druggability. Consequently, this crystallographic system emerged as a reliable platform for large-scale screenings. Thus, covalent ligands and fragments have been tested to identify PTG-interacting compounds. Different ligands resulted in the covalent binding of PTG, however with only minor efficiency (up to 26%) and selectivity for specific reactive cysteines. In addition, the positive non-covalent fragment hits lay on the protein surface or crystallisation interface, which is unideal for drug design purposes. Therefore, the test of mini-fragments and covalent mini-fragments libraries is ongoing to sharpen the design of PTG-specific inhibitors or PROTACs. On the other hand, a forward chemogenomic approach has been explored to set up a robust and unbiased HCS strategy. A phenotypic assay tailored to detect glycogen with GST-tagged PTG-CBM21 domain has been validated, showing the advantage of in-house and limitless production, since only two non-commercial anti-glycogen antibodies have been reported in the literature. SH-SY5Y neuroblastoma cells have been chosen for their homogenous and reproducible accumulation of glycogen and the possibility of neuron-like differentiation. A pilot screening of the KCGS library has been performed to validate the platform, resulting in five glycogen-reducing hits after DRC tests. These hits have been preliminarily evaluated in LD mouse primary astrocytes. Additionally, LD patient-derived iPSCs have been reprogrammed and characterised for being differentiated in brain organoids, astrocytes, and neurons. In particular, cortical organoids derived from one patient showed a statistically significant increase in glycogen accumulation compared to WT control, resulting in the first human in vitro model of LD. Taken together, these findings implemented a multimethodological approach for modelling LD and performing screenings to advance the design and development of LD therapeutics