Leveraging Neural Networks in Preclinical Alcohol Research

Abstract

Alcohol use disorder is a pervasive healthcare issue with significant socioeconomic consequences. There is a plethora of neural imaging techniques available at the clinical and preclinical level, including magnetic resonance imaging and three-dimensional (3D) tissue imaging techniques. Network-based approaches can be applied to imaging data to create neural networks that model the functional and structural connectivity of the brain. These networks can be used to changes to brain-wide neural signaling caused by brain states associated with alcohol use. Neural networks can be further used to identify key brain regions or neural "hubs" involved in alcohol drinking. Here, we briefly review the current imaging and neurocircuit manipulation methods. Then, we discuss clinical and preclinical studies using network-based approaches related to substance use disorders and alcohol drinking. Finally, we discuss how preclinical 3D imaging in combination with network approaches can be applied alone and in combination with other approaches to better understand alcohol drinking.

Keywords: addiction; alcohol use disorder; animal model; binge drinking; dependence; fMRI; graph theory; iDISCO; modularity; substance use disorder.

Figure 1

Illustration of imaging techniques used for neural network analysis of functional and structural brain connectivity. (A) Single-cell whole-brain imaging techniques, such as iDISCO, allow for analysis of the whole brain at the mesoscale (i.e., with region and cell-specific resolution), with results representing neural activity across minutes to hours. (B) Magnetic resonance imaging (MRI) techniques allow for analysis of the brain in more generalized resolution (i.e., at the macroscale) in anesthetized or immobilized animals, with results representing neural activity across seconds. Data from both of these methods can be interpreted using graph theory approaches to identify key brain regions, hubs, and global changes in neural networks during a given state.

Figure 2

Graph theory representation of the neural network associated with alcohol abstinence in alcohol-dependent mice. Analysis identified several underexplored regions within the extended amygdala module that may be critical for withdrawal-associated behavior and excessive alcohol drinking. The zoomed-in panel highlights several of the underexplored regions, which include the posterior cortical amygdala (COAp), intercalated amygdala (IA), parasubthalamic nucleus (PSTN), and tuberal nucleus (TU). Figure reproduced and modified from Kimbrough et al., 2020 PNAS.