The researchers set out to answer a question that has shadowed addiction neuroscience for decades: what allows drugs to drive dopamine release at intensities that ordinary rewards never reach? Their hypothesis centered on energy. Dopamine terminals in the nucleus accumbens are extraordinarily metabolically demanding structures, packed with mitochondria positioned near active zones where vesicles of dopamine await release signals. When dopaminergic neurons fire at high frequency, the energy demands of packaging, mobilizing, and releasing vesicles can outstrip local ATP supplies. Previous work had established that presynaptic mitochondria buffer calcium and synthesize ATP on demand, but whether this mitochondrial machinery was specifically recruited during drug reward—rather than reward in general—remained unknown.

To watch mitochondrial calcium in real time, the team used a genetically encoded fluorescent indicator called 4mt-jGCaMP8s, targeted to the mitochondrial matrix, expressed selectively in dopaminergic neurons of mice. Using fiber photometry, they monitored calcium signals from dopaminergic terminals in the nucleus accumbens while animals received heroin, morphine, methamphetamine, or cocaine. The results were unambiguous: heroin and methamphetamine produced robust, rapid mitochondrial calcium transients in accumbal dopamine terminals, while cocaine did not. This selectivity held up under two-photon imaging with a head-mounted microscope in freely moving animals, and the signal was strongest in the core of the nucleus accumbens, a subregion closely tied to drug seeking. Critically, when animals consumed natural rewards such as sucrose, the same terminals showed no comparable mitochondrial calcium influx.

The gateway for this calcium was identified as the mitochondrial calcium uniporter, or MCU, a channel in the inner mitochondrial membrane that was molecularly identified more than a decade ago. MCU allows calcium to flow down its steep concentration gradient into the mitochondrial matrix, where it activates calcium-sensitive enzymes of the tricarboxylic acid cycle and stimulates oxidative phosphorylation. In essence, mitochondrial calcium influx is how a mitochondrion learns that its host cell is working hard and needs more fuel. The Beijing team confirmed that the drug-evoked signals depended on MCU: in dopaminergic neurons lacking the uniporter, or in neurons treated with the MCU inhibitors Ru360 and berberine, the drug-induced mitochondrial calcium rise was abolished, while cytoplasmic calcium responses to the drugs remained intact.

Why would drugs, but not natural rewards, engage this pathway? The answer emerged from optogenetic experiments in which the researchers artificially drove dopaminergic terminals at defined frequencies. At low stimulation frequencies of around 4 hertz, mimicking the modest firing rates associated with physiological rewards, mitochondrial calcium barely moved, and deleting MCU had no effect on evoked dopamine release. But at 20 and 40 hertz—intensities characteristic of drug-evoked activation—mitochondrial calcium influx became prominent, and its absence sharply reduced dopamine release measured with the GRAB_DA2m fluorescent sensor. The conclusion is elegant: MCU-mediated mitochondrial calcium is recruited only when dopaminergic neurons are pushed beyond their normal metabolic comfort zone, which is precisely the regime that addictive drugs occupy and that natural rewards do not.

The mechanism linking calcium to release proved to be energy. Using the genetically encoded ATP sensor ATeam1.03 targeted to mitochondria, the researchers showed that drug-like high-intensity stimulation drove a rapid rise in mitochondrial ATP production that required MCU. When they blocked ATP synthesis with oligomycin, or supplied exogenous ATP to neurons lacking MCU, the picture sharpened further: ATP replenishment rescued the vesicle release deficit in MCU-deficient neurons, while ATP depletion mimicked it. Probing the vesicle machinery directly with a pHluorin-tagged vesicular monoamine transporter, the team found that the readily releasable pool of dopamine vesicles was depleted in MCU knockout neurons, and that this deficit was restored by ATP supplementation. In other words, the mitochondrial calcium influx does not directly trigger release; it manufactures the ATP that keeps the release machinery stocked during metabolically punishing bursts of activity.

The behavioral consequences were decisive. Mice engineered to lack MCU specifically in dopaminergic neurons showed dramatically reduced locomotor responses to heroin and morphine, blunted conditioned place preference for opioids, and diminished cue-induced reinstatement of cocaine seeking after extinction—a laboratory model of relapse. Circuit-specific deletion confirmed the anatomy: removing MCU only from dopaminergic neurons projecting from the ventral tegmental area to the nucleus accumbens impaired heroin reward, whereas deletion in the projection to the prefrontal cortex did not. Pharmacological validation followed. Berberine, a natural compound recently characterized as an MCU inhibitor that disrupts the channel’s assembly with its regulatory partner EMRE, reduced mitochondrial calcium uptake in cultured dopaminergic neurons, attenuated heroin conditioned place preference when microinjected into the nucleus accumbens, and suppressed cue-induced reinstatement of cocaine seeking when given systemically.

Perhaps most striking is what did not happen. Mice lacking MCU in dopaminergic neurons performed indistinguishably from controls across a battery of tests probing dopamine-dependent physiology: spontaneous locomotion, working memory in the Y-maze, anxiety-like behavior, despair-related immobility, sucrose preference, motor coordination on the rotarod, and spatial learning in the Morris water maze. Food-conditioned place preference, a model of natural reward learning, was preserved. Whole-brain MCU knockout mice, generated with Nestin-Cre, developed normally with expected Mendelian ratios and showed no deficits in dopamine-related behaviors. The therapeutic window implied by these results is unusually wide: the bioenergetic pathway appears to be dispensable for everyday reward processing yet essential for the supraphysiological dopamine output that drugs commandeer.

The study reframes addiction as, in part, a metabolic disease of synapses. Drugs of abuse differ in their molecular targets—opioids bind receptors, methamphetamine reverses transporters, cocaine blocks reuptake—but they converge on a common output: high-intensity dopaminergic activation in the accumbens. This work shows that convergence extends to the mitochondria, which are forced into emergency ATP production to sustain the pathological signal. By targeting MCU, or the calcium-driven bioenergetics it enables, it may be possible to starve the addictive circuitry of its energy supply while leaving the mitochondria of ordinary reward signaling untouched. Considerable distance remains between mouse models and human therapy, and berberine’s properties as a drug candidate will need rigorous evaluation. But the identification of a drug-selective mitochondrial mechanism offers something addiction research has rarely possessed: a molecular distinction between the dopamine that fuels compulsion and the dopamine that fuels life.

Subject of Research: Mitochondrial calcium uniporter-mediated bioenergetics in dopaminergic terminals that selectively enable drug-induced dopamine release and addictive behavior

Article Title: Mitochondrial calcium influx-driven bioenergetics selectively enable drug addiction

Article References: Gao, J., Zhao, H., Han, X., Zeng, L., Pan, J., Liu, G., Wei, X., Liu, C., Wu, W., Chen, S., Chen, J., Li, T., Yin, J., Zhou, T., Zhang, X.-M., Li, A.-L., Li, T., & Pan, X. (2026). Mitochondrial calcium influx-driven bioenergetics selectively enable drug addiction. Nature Neuroscience. https://doi.org/10.1038/s41593-026-02421-x

Image Credits: AI Generated

DOI: 10.1038/s41593-026-02421-x

Keywords: addiction, mitochondria, mitochondrial calcium uniporter, dopamine, nucleus accumbens, reward, MCU, bioenergetics, opioids, methamphetamine, ATP, berberine