Summary Synaptic connections undergo activity-dependent plasticity during development and learning, as well as homeostatic re-adjustment to ensure stability. Little is known about the relationship between these processes, particularly in vivo. We addressed this with novel quantal resolution imaging of transmission during locomotive behavior at glutamatergic synapses of the Drosophila larval neuromuscular junction. We find that two motor input types – Ib and Is – provide distinct forms of excitatory drive during crawling and differ in key transmission properties. While both inputs vary in transmission probability, active Is synapses are more reliable. High frequency firing “wakes up” silent Ib synapses and depresses Is synapses. Strikingly, homeostatic compensation in presynaptic strength only occurs at Ib synapses. This specialization is associated with distinct regulation of postsynaptic CaMKII. Thus, basal synaptic strength, short-term plasticity and homeostasis are determined input-specifically, generating a functional diversity that sculpts excitatory transmission and behavioral function.

Cerebellar circuits are patterned into an array of topographic parasagittal domains called zones. The proper connectivity of zones is critical for motor coordination and motor learning, and in several neurological diseases cerebellar circuits degenerate in zonal patterns. Despite recent advances in understanding zone function, we still have a limited understanding of how zones are formed. Here, we focused our attention on Purkinje cells to gain a better understanding of their specific role in establishing zonal circuits. We used conditional mouse genetics to test the hypothesis that Purkinje cell neurotransmission is essential for refining prefunctional developmental zones into sharp functional zones. Our results show that inhibitory synaptic transmission in Purkinje cells is necessary for the precise patterning of Purkinje cell zones and the topographic targeting of mossy fiber afferents. As expected, blocking Purkinje cell neurotransmission caused ataxia. Using in vivo electrophysiology, we demonstrate that loss of Purkinje cell communication altered the firing rate and pattern of their target cerebellar nuclear neurons. Analysis of Purkinje cell complex spike firing revealed that feedback in the cerebellar nuclei to inferior olive to Purkinje cell loop is obstructed. Loss of Purkinje neurotransmission also caused ectopic zonal expression of tyrosine hydroxylase, which is only expressed in adult Purkinje cells when calcium is dysregulated and if excitability is altered. Our results suggest that Purkinje cell inhibitory neurotransmission establishes the functional circuitry of the cerebellum by patterning the molecular zones, fine-tuning afferent circuitry, and shaping neuronal activity.

BackgroundAutism spectrum disorder (ASD) diagnosis can be delayed due in part to the time required for administration of standard exams, such as the Autism Diagnostic Observation Schedule (ADOS). Shorter and potentially mobilized approaches would help to alleviate bottlenecks in the healthcare system. Previous work using machine learning suggested that a subset of the behaviors measured by ADOS can achieve clinically acceptable levels of accuracy. Here we expand on this initial work to build sparse models that have higher potential to generalize to the clinical population.MethodsWe assembled a collection of score sheets for two ADOS modules, one for children with phrased speech (Module 2; 1319 ASD cases, 70 controls) and the other for children with verbal fluency (Module 3; 2870 ASD cases, 273 controls). We used sparsity/parsimony enforcing regularization techniques in a nested cross validation grid search to select features for 17 unique supervised learning models, encoding missing values as additional indicator features. We augmented our feature sets with gender and age to train minimal and interpretable classifiers capable of robust detection of ASD from non-ASD.ResultsBy applying 17 unique supervised learning methods across 5 classification families tuned for sparse use of features and to be within 1 standard error of the optimal model, we find reduced sets of 10 and 5 features used in a majority of models. We tested the performance of the most interpretable of these sparse models, including Logistic Regression with L2 regularization or Linear SVM with L1 regularization. We obtained an area under the ROC curve of 0.95 for ADOS Module 3 and 0.93 for ADOS Module 2 with less than or equal to 10 features.ConclusionsThe resulting models provide improved stability over previous machine learning efforts to minimize the time complexity of autism detection due to regularization and a small parameter space. These robustness techniques yield classifiers that are sparse, interpretable and that have potential to generalize to alternative modes of autism screening, diagnosis and monitoring, possibly including analysis of short home videos.