Synapses are the central information processors in the brain. Their function, efficacy and plasticity are key determinants of all brain functions, and of the corresponding behavioral output. Conversely, aberrant synapse function is the cause of many neurological and psychiatric disorders. Our ultimate objective is to generate a functional virtual synapse, in silico, that covers both the presynaptic and postsynaptic compartments.

Our overall strategy, closely following our initial proposal from 2017, has been the following. Over the first funding period, we obtained a large set of molecular, structural and functional data on a prototypic, averaged model synapse, while only starting to engage in computational work. During the second funding period, we refined our datasets by further wet-lab work, while at the same time strongly expanding the computational aspects of our work, integrating several additional computational neuroscience projects into the CRC. This generated the foundational work for the third funding period, in which we will rely even more heavily on in silico modelling to combine the CRC data, along with data from the literature, into a structural and functional model of a prototypic, average synapse.

During the third funding period, we will build on all of these elements to generate our ‘Göttingen’ synapse model, which will enable the field to address open questions on synaptic function and dysfunction. This model will rely on defined experimental and computational work from the previous funding periods, along with results that will be obtained during the third funding period, and will include a concept that the CRC published recently, namely that the synaptic vesicle cluster (SVC) has a key role in regulating the presynapse, while also influencing postsynaptic function. By combining our basic science approaches with pathology-based analyses and an ageing- and turnover-based focus, we expect to obtain a model that can provide substantial guidance on how pharmacological intervention should be used to ameliorate or prevent synapse dysfunction, by indicating which processes should best be targeted and how to target them. We also foresee that our work will be extremely important in the context of connectomics efforts, since our model will ultimately help to predict functional features of synapses in a given structural connectome, thus fundamentally increasing the connectome’s functional information content. Finally, we expect these identified functional features to give new impulses in the development of neuro-inspired algorithms and energy-efficient hardware architectures advancing artificial intelligence.