Temporal and spatial patterns of secondary motor cortex calcium activity in cocaine self-administration: A study using miniScope imaging and machine learning

Abstract

Addiction is a chronic mental disorder caused by disruptions in brain function. While most research has focused on the medial prefrontal cortex, our recent findings highlight the secondary motor cortex (M2) as a key region modulating cocaine-seeking behaviors during relapse. Mechanisms underlying the role of M2 in addiction remain unclear. We hypothesize that initial drug-taking behaviors directly reshape M2 neuronal activity. This study investigated the effects of five 1-hr daily intravenous self-administration (IVSA) sessions on M2 neuronal activity in male C57BL/6J mice using in vivo Ca²⁺ imaging via miniScopes. Temporal and spatial patterns of Ca²⁺ transients were analyzed across three IVSA factors: IV substance (i.e., saline vs. cocaine), IVSA days, and within-session stages. Machine learning models, including Recurrent Neural Networks for temporal patterns, and Neural Networks, Support Vector Machines, and Extreme Gradient Boosting for spatial patterns, were employed.

Results

Cocaine-treated mice displayed consistent drug-taking behaviors within sessions and increased intake on Day 5 compared to Day 1, unlike saline-treated mice, which showed reduced operant behaviors within each daily session. IV substance was the most sensitive factor influencing both temporal and spatial patterns of M2 Ca²⁺ transients, characterized by frequency and frequency-amplitude interactions, but not amplitude alone. A 15-s bin size optimized differentiation between saline and cocaine groups.

Conclusion

Both temporal and spatial alterations in M2 neuronal activity were detected during early cocaine exposure, revealing early changes that may contribute to later drug relapse. These findings underscore the importance of advanced imaging and machine learning techniques in advancing addiction research.