Learning-guided automatic three dimensional synapse quantification for drosophila neurons

Synaptic aging disrupts synaptic morphology and function in cerebellar Purkinje cells

Synapses are key structures in neural networks, and are involved in learning and memory in the central nervous system. Investigating synaptogenesis and synaptic aging is important in understanding neural development and neural degeneration in diseases such as Alzheimer disease and Parkinson's disease. Our previous study found that synaptogenesis and synaptic maturation were harmonized with brain development and maturation. However, synaptic damage and loss in the aging cerebellum are not well understood. This study was designed to investigate the occurrence of synaptic aging in the cerebellum by observing the ultrastructural changes of dendritic spines and synapses in cerebellar Purkinje cells of aging mice. Immunocytochemistry, DiI diolistic assays, and transmission electron microscopy were used to visualize the morphological characteristics of synaptic buttons, dendritic spines and synapses of Purkinje cells in mice at various ages. With synaptic aging in the cerebellum, dendritic spines and synaptic buttons were lost, and the synaptic ultrastructure was altered, including a reduction in the number of synaptic vesicles and mitochondria in presynaptic termini and smaller thin specialized zones in pre- and post-synaptic membranes. These findings confirm that synaptic morphology and function is disrupted in aging synapses, which may be an important pathological cause of neurodegenerative diseases.

Analyzing Synaptic Modulation of Drosophila melanogaster Photoreceptors after Exposure to Prolonged Light

The nervous system has the remarkable ability to adapt and respond to various stimuli. This neural adjustment is largely achieved through plasticity at the synaptic level. The Active Zone (AZ) is the region at the presynaptic membrane that mediates neurotransmitter release and is composed of a dense collection of scaffold proteins. AZs of Drosophila melanogaster (Drosophila) photoreceptors undergo molecular remodeling after prolonged exposure to natural ambient light. Thus the level of neuronal activity can rearrange the molecular composition of the AZ and contribute to the regulation of the functional output. Starting from the light exposure set-up preparation to the immunohistochemistry, this protocol details how to quantify the number, the spatial distribution, and the delocalization level of synaptic molecules at AZs in Drosophila photoreceptors. Using image analysis software, clusters of the GFP-fused AZ component Bruchpilot were identified for each R8 photoreceptor (R8) axon terminal. Detected Bruchpilot spots were automatically assigned to individual R8 axons. To calculate the distribution of spot frequency along the axon, we implemented a customized software plugin. Each axon's start-point and end-point were manually defined and the position of each Bruchpilot spot was projected onto the connecting line between start and end-point. Besides the number of Bruchpilot clusters, we also quantified the delocalization level of Bruchpilot-GFP within the clusters. These measurements reflect in detail the spatially resolved synaptic dynamics in a single neuron under different environmental conditions to stimuli.

The parameter sensitivity of random forests

Background

The Random Forest (RF) algorithm for supervised machine learning is an ensemble learning method widely used in science and many other fields. Its popularity has been increasing, but relatively few studies address the parameter selection process: a critical step in model fitting. Due to numerous assertions regarding the performance reliability of the default parameters, many RF models are fit using these values. However there has not yet been a thorough examination of the parameter-sensitivity of RFs in computational genomic studies. We address this gap here.

Results

We examined the effects of parameter selection on classification performance using the RF machine learning algorithm on two biological datasets with distinct p/n ratios: sequencing summary statistics (low p/n) and microarray-derived data (high p/n). Here, p, refers to the number of variables and, n, the number of samples. Our findings demonstrate that parameterization is highly correlated with prediction accuracy and variable importance measures (VIMs). Further, we demonstrate that different parameters are critical in tuning different datasets, and that parameter-optimization significantly enhances upon the default parameters.

Conclusions

Parameter performance demonstrated wide variability on both low and high p/n data. Therefore, there is significant benefit to be gained by model tuning RFs away from their default parameter settings.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-016-1228-x) contains supplementary material, which is available to authorized users.

Mapping the connectivity of serotonin transporter immunoreactive axons to excitatory and inhibitory neurochemical synapses in the mouse limbic brain

Serotonin neurons arise from the brainstem raphe nuclei and send their projections throughout the brain to release 5-HT which acts as a modulator of several neuronal populations. Previous electron microscopy studies in rats have morphologically determined the distribution of 5-HT release sites (boutons) in certain brain regions and have shown that 5-HT containing boutons form synaptic contacts that are either symmetric or asymmetric. In addition, 5-HT boutons can form synaptic triads with the pre- and postsynaptic specializations of either symmetrical or asymmetrical synapses. However, due to the labor intensive processing of serial sections required by electron microscopy, little is known about the neurochemical properties or the quantitative distribution of 5-HT triads within whole brain or discrete subregions. Therefore, we used a semi-automated approach that combines immunohistochemistry and high-resolution confocal microscopy to label serotonin transporter (SERT) immunoreactive axons and reconstruct in 3D their distribution within limbic brain regions. We also used antibodies against key pre- (synaptophysin) and postsynaptic components of excitatory (PSD95) or inhibitory (gephyrin) synapses to (1) identify putative 5-HTergic boutons within SERT immunoreactive axons and, (2) quantify their close apposition to neurochemical excitatory or inhibitory synapses. We provide a 5-HTergic axon density map and have determined the ratio of synaptic triads consisting of a 5-HT bouton in close proximity to either neurochemical excitatory or inhibitory synapses within different limbic brain areas. The ability to model and map changes in 5-HTergic axonal density and the formation of triadic connectivity within whole brain regions using this rapid and quantitative approach offers new possibilities for studying neuroplastic changes in the 5-HTergic pathway.

Bioimage Informatics for Big Data

Bioimage informatics is a field wherein high-throughput image informatics methods are used to solve challenging scientific problems related to biology and medicine. When the image datasets become larger and more complicated, many conventional image analysis approaches are no longer applicable. Here, we discuss two critical challenges of large-scale bioimage informatics applications, namely, data accessibility and adaptive data analysis. We highlight case studies to show that these challenges can be tackled based on distributed image computing as well as machine learning of image examples in a multidimensional environment.