Researchers at the University of Houston and the University of Cincinnati are using machine learning to create a clearer picture of how heroin affects the brain, potentially opening new doors for addiction treatment.
In a groundbreaking study published April 30 in the journal “Science Advances,” Demetrio Labate, professor of mathematics at UH, his PhD students, Michela Marini and Heng Zhao, and his colleague Yabo Niu, an assistant professor in the Department of Mathematics and the Department of Health Systems and Population Health Sciences at UH, collaborated with Anna Kruyer, an assistant professor in UC’s Department of Pharmacy, to apply object recognition technology to track changes in brain cell structure and provide new insights into how the brain responds to heroin use, withdrawal and relapse.
“Essentially the Holy Grail in the study of substance use disorder is how to find treatments that prevent opioid users from relapsing,” said Marini, who was the lead author on the study. “There are treatments for alcohol intoxication, but not for drugs like heroin, so if we can find a way to block people from relapsing, that would have a huge impact.”
“Collaborating with the University of Cincinnati on this study has been incredibly rewarding. This kind of work is essential for developing better drug addiction treatments in the future.” - Michela Marini, PhD student at UH
Study Background
According to Kruyer, whose lab at UC is focused on relapses from heroin, many overdose deaths occur when people overestimate their capacity for drug use during a relapse. Her team developed an animal model to study interactions between brain cells and the reward center of the brain that orchestrate the relapse process.
“We want to understand the neurons that are involved and all of the different cells and molecules that can shape that activity,” Kruyer said. “The idea would be if you can interfere with relapse, you can help someone stay clean.”
While neurons are the more commonly studied brain cells, Kruyer’s team focused on another type of cell called an astrocyte. Astrocytes support neurons by supplying metabolic energy, providing the building blocks for neurotransmitters and regulating synaptic activity by shielding or exposing different receptors.
“Astrocytes are a kind of protective cell that can restore synaptic homeostasis,” Kruyer said. “They are super dynamic relative to the synapse, and they’re moving toward and away from the synapse in real time in a way that can impact drug seeking. So if you prevent this reassociation with synapses during relapse, you can increase and prolong relapse.”
A New Approach
But Kruyer’s animal model wasn’t translatable to human subjects, so the team focused on an astrocyte protein that essentially acts as the cell’s cytoskeleton.
“We thought if we could figure out a way to translate what we’re seeing at the synaptic level to changes in the cytoskeleton, maybe we could see if astrocytes are doing something critical during relapse in humans,” Kruyer said.
That’s when Kruyer called Labate and Marini for help.
“A central focus of my research is the development and application of mathematical techniques to uncover meaningful patterns in non-Euclidean data, such as the analysis of complex shapes,” Labate said. “The study of astrocytes provides an ideal setting for this type of investigation – these cells are highly heterogeneous, varying widely in size and shape, and are capable of dynamically remodeling their morphology in response to external stimuli.”
Labate and Marini used machine learning to train a computer to recognize astrocyte cells in images, similar to how recognition software identifies objects like cars or people in photographs. Once the computer found an astrocyte, it measured 15 structural features – including size, elongation and branching properties to better understand how they were different.
The team then applied the computer model to identify subpopulations of astrocytes in the nucleus accumbens, a region of the brain associated with drug relapse. The model could determine the cells’ origin with 80% accuracy, suggesting that astrocytes are not a homogenous group as once thought. Instead, their structure — specifically their shape and size — appears to vary by location and may be linked to differences in function.
Using the animal model and the knowledge gained from the computer models, they found these astrocyte subpopulations appear to shrink and become less malleable after exposure to heroin.
“These data suggest that heroin is doing something molecularly that makes astrocytes less able to respond to synaptic activity and maintain homeostasis,” Kruyer said.
Next Steps
As the team continues to uncover the specific mechanisms of astrocytes, their findings could inform the development of new treatments for addiction – aimed at restoring or replacing the normal functions of astrocytes that are disrupted by heroin exposure.
Additionally, the machine learning method Labate and Marini developed can be adapted and applied to other types of cells with intricate structures.
“By enabling precise quantification and comparison of single-cell morphological features, this approach opens the door to the development of novel techniques for identifying cellular or molecular biomarkers that reflect biological processes, disease states or responses to therapeutic interventions,” Labate said. “More broadly, our work introduces a new quantitative framework for uncovering and validating fundamental mechanistic models underlying complex brain conditions, such as drug addiction.”
For Marini, the work with the UC team emphasized the importance of interdisciplinary collaboration in advancing brain research.
“Collaborating with the University of Cincinnati on this study has been incredibly rewarding,” Marini said. “By uncovering how astrocytes are altered by heroin use, we’re opening new doors not just for addiction research, but for understanding the brain’s response to a wide range of drugs and neurological conditions. This kind of work is essential for developing better drug addiction treatments in the future.”
