Yang, H.-W. | bioRxiv Science

bioRxiv2025

Axon collateral pattern of a sparse locus coeruleus norepinephrine neuron in mouse cerebral cortex.

The locus coeruleus (LC) contains predominantly norepinephrine (NE) neurons that project widely throughout the brain. The LC plays a critical role in controlling behavior, particularly arousal. Historically, it was thought that the LC-NE system performed its behavioral control function by uniformly releasing NE throughout most brain regions. However, recent evidence suggests that the LCs cortical projections are organized into modules, which allows for the coordination of diverse, and sometimes opposing, functions such as fear memory formation and extinction. Nevertheless, many details remain unclear and require data from the axon collaterals of sparse neurons. We modified a viral tracing protocol using a dual-recombinase system to trace the axonal collaterals of sparse LC neurons projecting to the cingulate cortex (CgC). Our results show that even a small number of LC neurons have broad cortical projections, though the pattern is not uniform. Centered-log ratio transformation of NE fiber distribution across the cortex and hippocampus reveals a few preferential target areas (PTAs) of the labeled LC-NE neurons axonal projections. The summed NE fiber length in these defined PTs is enriched relative to the geometric mean of all other cortical and hippocampal regions where NE fibers were detected. Notably, the defined PTAs--including the rostral splenial cortex, dorsal hippocampus, somatosensory cortex, and CgC (the retrograde viral labeling injection site)--are functionally related to navigation. These results demonstrate that LC-NE neurons are organized into distinct projection modules, each comprising a small number of neurons with functionally correlated major cortical targets.

neuroscience↗

bioRxiv2023

Pre-trained Inspired MocFormer: Efficient and Predictive Models of Drug-target Interactions

Drug-target interactions (DTIs) is essential for advancing pharmaceuticals. Traditional drug-target interaction studies rely on labor-intensive laboratory techniques. Still, recent advancements in computing power have elevated the importance of deep learning methods, offering faster, more precise, and cost-effective screening and prediction. Nonetheless, general deep learning methods often yield low-confidence results due to the complex nature of drugs and proteins, bias, limited labeled data, and feature extraction challenges. To address these challenges, a novel two-stage pre-trained framework is proposed for DTIs prediction. In the first stage, pre-trained molecule and protein models develop a comprehensive feature representation, enhancing the frameworks ability to handle drug and protein diversity. This also reduces bias, improving prediction accuracy. In the second stage, a transformer with bilinear pooling and a fully connected layer (FCN) enables predictions based on feature vectors. Comprehensive experiments were conducted using DrugBank dataset and Epigenetic-regulators dataset to evaluate the frameworks effectiveness. The results demonstrate that the proposed framework outperforms the state-of-the-art methods regarding accuracy, area under the ROC curve (AUC), recall, and the area under the precision-recall curve (AUPRC). The code will be available after being accepted: https://github.com/rickwang28574/MocFormer

bioinformatics↗