The developmental and epileptic encephalopathies (DEEs) are heterogeneous disorders with a strong genetic contribution, but the underlying genetic etiology remains unknown in a significant proportion of individuals. To explore whether statistical support for genetic etiologies can be generated on the basis of phenotypic features, we analyzed whole-exome sequencing data and phenotypic similarities by using Human Phenotype Ontology (HPO) in 314 individuals with DEEs. We identified a de novo c.508C>T (p.Arg170Trp) variant in AP2M1 in two individuals with a phenotypic similarity that was higher than expected by chance (p ¼ 0.003) and a phenotype related to epilepsy with myoclonic-atonic seizures. We subsequently found the same de novo variant in two individuals with neurodevelopmental disorders and generalized epilepsy in a cohort of 2,310 individuals who underwent diagnostic whole-exome sequencing. AP2M1 encodes the m-subunit of the adaptor protein complex 2 (AP-2), which is involved in clathrin-mediated endocytosis (CME) and synaptic vesicle recycling. Modeling of protein dynamics indicated that the p.Arg170Trp variant impairs the conformational activation and thermodynamic entropy of the AP-2 complex. Functional complementation of both the m-subunit carrying the p.Arg170Trp variant in human cells and astrocytes derived from AP-2m conditional knockout mice revealed a significant impairment of CME of transferrin. In contrast, stability, expression levels, membrane recruitment, and localization were not impaired, suggesting a functional alteration of the AP-2 complex as the underlying disease mechanism. We establish a recurrent pathogenic variant in AP2M1 as a cause of DEEs with distinct phenotypic features, and we implicate dysfunction of the early steps of endocytosis as a disease mechanism in epilepsy.

Expansion microscopy (ExM) is a recently developed technique that allows for the resolution of structures below the diffraction limit by physically enlarging a hydrogel-embedded facsimile of the biological sample. The target structure is labeled and this label must be retained in a relative position true to the original, smaller state before expansion by linking it into the gel. However, gel formation and digestion lead to a significant loss in target-delivered label, resulting in weak signal. To overcome this problem, we have here developed an agent combining targeting, fluorescent labeling and gel linkage in a single small molecule. Similar approaches in the past have still suffered from significant loss of label. Here we show that this loss is due to insufficient surface grafting of fluorophores into the hydrogel and develop a solution by increasing the amount of target-bound monomers. Overall, we obtain a significant improvement in fluorescence signal retention and our new dye allows the resolution of nuclear pores as ring-like structures, similar to STED microscopy. We furthermore provide mechanistic insight into dye retention in ExM.

No abstract

The combination of image analysis and fluorescence superresolution microscopy methods allows for unprecedented insight into the organization of macromolecular assemblies in cells. Advances in deep learning-based object recognition enables the automated processing of large amounts of data, resulting in high accuracy through averaging. However, while the analysis of highly symmetric structures of constant size allows for a resolution approaching the dimensions of structural biology, deep learning methods are prone to different forms of bias. A biased recognition of structures may prohibit the development of readouts for processes that involve significant changes in size or shape of amorphous macromolecular complexes. What is required to overcome this problem is a detailed investigation of potential sources of bias and the rigorous testing of trained models using real or simulated data covering a wide dynamic range of possible results. Here we combine single molecule localization-based superresolution microscopy of septin ring structures with the training of several different deep learning models for a quantitative investigation of bias resulting from different training approaches and finally quantitative changes in septin ring structures. We find that trade-off exists between measurement accuracy and the dynamic range of recognized phenotypes. Using our trained models, we furthermore find that septin ring size can be explained by the number of subunits they are assembled from alone. Our work provides a new experimental system for the investigation of septin polymerization.