Modelling input selection at synaptic clusters on dendrites

一些发现支持这样的观点，即树突分支上的突触簇与被动和主动的树突过程一起，允许计算操作，如过滤、选通、路由和多路传输传入的信息。然而，剩下的一个核心问题是，树突小室如何“知道”哪些突触--因此它的哪些输入--对所需的操作“有用”，而哪些不“有用”？有用的突触应该通过结构可塑性来选择，通过突触可塑性来加强，而所有其他的突触应该被削弱和移除，导致一个支持所需操作的输入选择过程。在这项提案中，我们正在系统地解决这个问题，方法是利用一个用于大规模参数分析的抽象模型和一个生物学详细的模型，将我们的发现与特定的生物物理过程联系起来，从而实现可通过实验验证的预测。在我们的第一个目标中，我们假设突触后处理模拟某个操作符，并研究一个树突分支上的突触簇如何根据特定操作符、局部树突信号(如树突棘波)和可塑性执行输入选择。在第二个目标中，我们将研究全局树突信号，如反向传播的棘波和树突交界处的塑料突触对操作员执行和输入选择的影响。我们认为，这些不同过程之间的相互作用通过梳理和适应各个算子来形成复杂的算子，每个算子都在一个单独的树枝上实现。综上所述，该项目的结果将是一个在不同操作符限制下的树突输入选择的生物物理现实模型，并伴随着一个更抽象的模型，允许对参数依赖关系有更一般的理解。这是新的，因为到目前为止，树突计算的模型还没有考虑到树突操作与突触和结构可塑性的结合之间的相互作用。

Several findings support the notion that synaptic clusters at a dendritic branch together with passive and active dendritic processes allow calculating operations such as filtering, gating, routing, and multiplexing of the incoming information. A central remaining question, however, is how a dendritic sub-compartment “knows” which synapses – hence which of its inputs – would be “useful” for the desired operation and which not? Useful synapses should be selected by structural plasticity and strengthened by synaptic plasticity, while all others should be weakened and removed, leading to a process of input selection supporting the desired operation. We are systematically addressing this question in this proposal by utilizing an abstract model for large-scale parameter analyses and a biologically detailed model to link our findings with specific biophysical processes allowing experimentally verifiable predictions. In our first aim, we assume that post-synaptic processing emulates a certain operator and investigate how synaptic clusters at one dendritic branch perform input selection dependent on the specific operator, local dendritic signals such as dendritic spikes, and plasticity. In the second aim, we will investigate the influence of global dendritic signals, such as back-propagating spikes, and plastic synapses at dendritic junctions on operator execution and input selection. We suggest that the interplay between these different processes form complex operators by combing and adapting individual operators, each being realized on an individual dendritic branch. Taken together, the outcome of this project will be a biophysically realistic model of dendritic input selection under the constraints of different operators, accompanied by a more abstract model that allows a more general understanding of parameter dependencies. This is novel, because so far models of dendritic computation have not considered the interplay between dendritic operation and the combination of synaptic and structural plasticity.