Morphine does two things at once. It dulls pain, and it teaches the brain to want the experience again. For decades, researchers have assumed those two effects share a single neural currency: dopamine. The opioid reward, the theory went, is essentially a dopamine signal in the nucleus accumbens, the brain’s motivation hub. A new study from Duke University shows that assumption is incomplete and that the two effects can be separated with surgical precision.
Writing in Nature, a team led by Michael Tadross built a molecular tool called naloxone DART that targets a specific population of cells, the cholinergic interneurons of the nucleus accumbens, and renders them insensitive to morphine, while leaving every other opioid response in the brain untouched. The result: mice learned nothing from the drug. No conditioned place preference developed. Yet morphine’s pain relief, its stimulation of movement, and its sensitization all remained fully intact, and dopamine still rose in the accumbens as before.
A tether for a drug
The technology is the key. DART, which stands for Drug Acutely Restricted by Tethering, was originally described in 2017 and refined in 2024 with a claimed thousandfold cellular specificity. Naloxone DART is a membrane-tethered version of the clinical opioid antagonist naloxone, delivered selectively to genetically defined cholinergic interneurons in the accumbens. It blocks the mu-opioid receptors on those cells and only those cells.
Getting it there required solving a viral-delivery puzzle. The team tested six AAV serotypes and three titers before settling on an optimized two-step strategy called fDIO, which produced roughly sixfold higher expression in the target cells than the conventional approach. With the tool in place, the experiments could ask a question that had been out of reach: what happens when opioid signaling is blocked in one small cell type, while dopamine signaling proceeds normally?
The answer was striking. Microdialysis showed morphine’s characteristic reduction of acetylcholine in the accumbens was prevented, while the dopamine increase remained. Behaviorally, the morphine-conditioned place preference, the standard assay for reward learning, was abolished, with no signs of anxiety, no impairment of context learning, and no disruption of locomotion. The dissociation was clean: no reward learning, but full analgesia.
A cholinergic gate for reward learning
The finding supports an emerging theory that dopamine and acetylcholine work as partners in reward learning, with the cholinergic interneurons acting as a gate. Block the cholinergic arm, and the dopamine signal alone is not enough to stamp in the association between drug and place. The authors describe their result in terms of a cholinergic gate for associative opioid-reward learning.
The regional analysis pinpointed the medial shell of the nucleus accumbens as the dominant site for the effect, with a smaller contribution from the medial septum. The precision is notable: even within the accumbens, the relevant cells are a specific subtype in a specific subregion.
Why it matters
The ultimate goal is a strategy that preserves morphine’s clinical utility while blunting its addictive pull. Opioid prescriptions remain essential for severe pain, and the addiction crisis has made the trade-off between pain relief and dependence a matter of life and death. If reward learning can be blocked at the cholinergic gate without sacrificing analgesia, it suggests a path toward opioid-based therapies that are harder to become addicted to, or adjunct treatments that reduce the learned component of addiction.
The study is in mice, and the tool is a research reagent, not a drug. The clinical distance is considerable: naloxone DART is a cell-type-specific probe for dissecting circuits, and translating a tethering approach into a therapy for humans is a different order of problem. But the conceptual advance is clear. Reward learning is not simply dopamine; it has a cholinergic gate, and that gate can be operated independently of pain relief.
Sources
- Yousefzadeh, S.A., Yan, H., Kwak, S.-H., Oh, Y., Jeong, P., Pogorelov, V., Ravenel, J.R., Lim, S.S.X., Roach, J.M., Shields, B.C., Rodriguiz, R.M., Wetsel, W.C., Hong, J., Tadross, M.R. “A cholinergic hub in the nucleus accumbens gates opioid-reward learning.” Nature (2026). DOI: 10.1038/s41586-026-10887-9
- Shields et al., Science (2017); DART.2, Nature Methods (2024)