Lateral Connections Improve Generalizability of Learning in a Simple Neural Network.

To navigate the world around us, neural circuits rapidly adapt to their environment learning generalizable strategies to decode information. When modeling these learning strategies, network models find the optimal solution to satisfy one task condition but fail when introduced to a novel task or even a different stimulus in the same space. In the experiments described in this letter, I investigate the role of lateral gap junctions in learning generalizable strategies to process information. Lateral gap junctions are formed by connexin proteins creating an open pore that allows for direct electrical signaling between two neurons. During neural development, the rate of gap junctions is high, and daughter cells that share similar tuning properties are more likely to be connected by these junctions. Gap junctions are highly plastic and get heavily pruned throughout development. I hypothesize that they mediate generalized learning by imprinting the weighting structure within a layer to avoid overfitting to one task condition. To test this hypothesis, I implemented a feedforward probabilistic neural network mimicking a cortical fast spiking neuron circuit that is heavily involved in movement. Many of these cells are tuned to speeds that I used as the input stimulus for the network to estimate. When training this network using a delta learning rule, both a laterally connected network and an unconnected network can estimate a single speed. However, when asking the network to estimate two or more speeds, alternated in training, an unconnected network either cannot learn speed or optimizes to a singular speed, while the laterally connected network learns the generalizable strategy and can estimate both speeds. These results suggest that lateral gap junctions between neurons enable generalized learning, which may help explain learning differences across life span.

Region-Specific Phosphorylation Determines Neuroligin-3 Localization to Excitatory Versus Inhibitory Synapses.

BACKGROUND: Neuroligin-3 is a postsynaptic adhesion molecule involved in synapse development and function. It is implicated in rare, monogenic forms of autism, and its shedding is critical to the tumor microenvironment of gliomas. While other members of the neuroligin family exhibit synapse-type specificity in localization and function through distinct interactions with postsynaptic scaffold proteins, the specificity of neuroligin-3 synaptic localization remains largely unknown. METHODS: We investigated the synaptic localization of neuroligin-3 across regions in mouse and human brain samples after validating antibody specificity in knockout animals. We raised a phospho-specific neuroligin antibody and used phosphoproteomics, cell-based assays, and in utero CRISPR/Cas9 (clustered regularly interspaced short palindromic repeats/Cas9) knockout and gene replacement to identify mechanisms that regulate neuroligin-3 localization to distinct synapse types. RESULTS: Neuroligin-3 exhibits region-dependent synapse specificity, largely localizing to excitatory synapses in cortical regions and inhibitory synapses in subcortical regions of the brain in both mice and humans. We identified specific phosphorylation of cortical neuroligin-3 at a key binding site for recruitment to inhibitory synapses, while subcortical neuroligin-3 remained unphosphorylated. In vitro, phosphomimetic mutation of that site disrupted neuroligin-3 association with the inhibitory postsynaptic scaffolding protein gephyrin. In vivo, phosphomimetic mutants of neuroligin-3 localized to excitatory postsynapses, while phospho-null mutants localized to inhibitory postsynapses. CONCLUSIONS: These data reveal an unexpected region-specific pattern of neuroligin-3 synapse specificity, as well as a phosphorylation-dependent mechanism that regulates its recruitment to either excitatory or inhibitory synapses. These findings add to our understanding of how neuroligin-3 is involved in conditions that may affect the balance of excitation and inhibition.

Rapid modeling of an ultra-rare epilepsy variant in wild-type mice by in utero prime editing.

Generating animal models for individual patients within clinically-useful timeframes holds great potential toward enabling personalized medicine approaches for genetic epilepsies. The ability to rapidly incorporate patient-specific genomic variants into model animals recapitulating elements of the patient's clinical manifestations would enable applications ranging from validation and characterization of pathogenic variants to personalized models for tailoring pharmacotherapy to individual patients. Here, we demonstrate generation of an animal model of an individual epilepsy patient with an ultra-rare variant of the NMDA receptor subunit GRIN2A, without the need for germline transmission and breeding. Using in utero prime editing in the brain of wild-type mice, our approach yielded high in vivo editing precision and induced frequent, spontaneous seizures which mirrored specific elements of the patient's clinical presentation. Leveraging the speed and versatility of this approach, we introduce PegAssist, a generalizable workflow to generate bedside-to-bench animal models of individual patients within weeks. The capability to produce individualized animal models rapidly and cost-effectively will reduce barriers to access for precision medicine, and will accelerate drug development by offering versatile in vivo platforms to identify compounds with efficacy against rare neurological conditions.

Enhancing Precision and Efficiency of Cas9-Mediated Knockin Through Combinatorial Fusions of DNA Repair Proteins.

Cas9 targets genomic loci with high specificity. For knockin with double-strand break repair, however, Cas9 often leads to unintended on-target knockout rather than intended edits. This imprecision is a barrier for direct in vivo editing where clonal selection is not feasible. In this study, we demonstrate a high-throughput workflow to comparatively assess on-target efficiency and precision of editing outcomes. Using this workflow, we screened combinations of donor DNA and Cas9 variants, as well as fusions to DNA repair proteins. This yielded novel high-performance double-strand break repair editing agents and combinatorial optimizations, yielding increases in knockin efficiency and precision. Cas9-RC, a novel fusion Cas9 flanked by eRad18 and CtIP[HE], increased knockin performance in vitro and in vivo in the developing mouse brain. Continued comparative assessment of editing efficiency and precision with this framework will further the development of high-performance editing agents for in vivo knockin and future genome therapeutics.