Internal Brain States of Opioid Addiction

Abstract

Fentanyl withdrawal induces a complex internal state characterized by adverse physiological, emotional, and cognitive symptoms that reinforce drug taking behaviors. Although studies have identified many local transcriptional, cellular, and circuit adaptations that contribute to, or are a result of opioid withdrawal, the widespread expression of opioid receptors across the brain suggests that repeated opioid exposure and withdrawal may reorganize brain-wide neural communication. However, whether such large-scale adaptations emerge, and how they relate to a withdrawal state, remains unclear. Here, we combine multisite local field potential (LFP) recordings with interpretable machine learning tools to uncover brain network dynamics engaged by repeated opioid exposure and withdrawal. We identify a sensitization-related network primarily characterized by monotonic increases in widespread high-beta activity in response to repeated fentanyl exposure. Then, following naloxone-precipitated withdrawal in fentanyl-dependent mice, we additionally uncovered a large-scale, highly reproducible, and interpretable brain network, termed EN-Withdrawal. Activity within this network coupled with broad cellular unit firing and was prominently defined by rapid, region-specific gamma oscillations, paired with slower interregional delta and theta coherence. While dense connectivity patterns pointed towards the dorsolateral striatum and amygdala as relevant network hubs, orbitofrontal cortex gamma power emerged as the most dominant network feature. And although the network was solely trained to classify non-withdrawal from naloxone-precipitated withdrawal states in fentanyl-dependent mice, EN-Withdrawal generalized across two independent cohorts of morphine-dependent mice, accurately predicting both spontaneous and precipitated withdrawal in these cohorts, while explaining over 70% of neurophysiological variance across conditions. After deriving a data-driven behavioral severity index from somatic and exploratory behaviors, we found that EN-Withdrawal activity scaled with behavioral severity and persisted across contexts, timepoints, and behavioral manifestations, suggesting it encodes a sustained internal state rather than transient motor output. Lastly, leveraging an intravenous fentanyl self-administration model, we discovered that escalation in EN-Withdrawal activity patterns across days significantly correlated with escalation in fentanyl use on a mouse-by-mouse basis, profoundly shaping motivated drug seeking behaviors. These findings demonstrate that opioid withdrawal engages a distributed, multiscale brain network and provides a biologically grounded, generalizable framework for tracking internal states and identifying novel brain-wide networks relevant to addiction.