A new study led by researchers at the Icahn School of Medicine at Mount Sinai offers what they describe as one of the most detailed molecular maps to date of how brain cells fail to communicate in Alzheimer’s disease (AD), identifying a protein network that provides an important new target for future treatments. The research, published in Cell, found that in AD glial cells become overactive and neurons exhibit decreased functionality leading to increased inflammation and the accumulation of toxic proteins in the brain.
“Alzheimer’s is not just about plaque buildup or dying neurons; it’s about how the entire brain ecosystem breaks down,” said senior author Bin Zhang, PhD, a professor of neurogenetics and director of the Center for Transformative Disease Modeling at the Icahn School of Medicine.
For this study, the team used large-scale proteomic profiling and computational modeling to analyze nearly 200 postmortem brain samples, primarily focusing on the para-hippocampal gyrus (PHG), a brain region vital for memory and spatial processing and known to be vulnerable to AD. The investigators analyzed more than 12,000 proteins from brains of people with AD and those without AD. They used an unsupervised systems biology approach to identify molecular disruptions within the brain.
“This study took a broader view, examining how more than 12,000 proteins interact inside the brain,” said co-senior author Junmin Peng, PhD, professor of structural biology and developmental neurobiology at St. Jude Children’s Research Hospital. “Using state-of-the-art proteomics profiling technology, we quantified protein expression across the brain, enabling a comprehensive view of proteomic alterations and interactions in Alzheimer’s.”
The research found that a core dysfunction in AD involves glia-neuron communication. In the healthy brain, glial cells—including astrocytes and microglia—support and regulate neurons. “The multiscale network analysis revealed an AD-associated subnetwork that captures the interactions among neurons, microglia, and astrocytes,” the researchers wrote. “Such interactions were validated by the so-far largest single-nucleus RNA sequencing (snRNA-seq) dataset in AD.”
A key protein identified in this network was AHNAK, a protein that is highly expressed in astrocytes. The team found that AHNAK levels increased as AD progressed and these increases were correlated with higher levels of amyloid beta and tau, hallmark proteins of AD pathology.
The researchers validated AHNAK’s role via testing with induced pluripotent stem cell (iPSC)-based models of AD. They found that when AHNAK was suppressed, it not only improved neuron function but also led to proteomic changes that matched the AD network disruptions observed in the human data.
“By lowering [AHNAK’s] activity, we saw both less toxicity and more neuronal activity, two encouraging signs that we may be able to restore healthier brain function,” said co-senior author Dongming Cai, MD, PhD, professor of neurology and director of the Grossman Center for Memory Research and Care at the University of Minnesota.
In all, the study identified 580 key driver proteins (KDPs), with 108 of them located in a core protein co-expression module that includes AHNAK. Of these, nearly 20% had not been studied previously in the context of AD. This includes proteins such as ERBB2IP, which is upregulated in AD and has known roles in immune modulation and autophagy, and OLFM3, previously identified as a potential cerebrospinal fluid biomarker for AD.
Earlier transcriptomic and multiomics studies had pointed to glial involvement in AD, but these new findings provide a more detailed, protein-level view of activity withing the brain.
“This study opens up a new way of thinking about Alzheimer’s, not just as a buildup of toxic proteins, but as a breakdown in how brain cells talk to each other,” Zhang said. “By understanding those conversations and where they go wrong, we can start to develop treatments that bring the system back into balance.”
The team is hopeful that their identification of these protein networks will provide a new basis for identifying therapeutic targets that may be more effective than treatments focused only on amyloid or tau. The study also found that biological variables such as sex and genetic background, such as the the presence of the APOE4 allele, influence how these protein networks behave, which could point to a precision approach.
“A critical step in developing AD therapeutics is target identification and prioritization,” the researchers wrote. “As protein networks developed from this study are more relevant to AD than transcriptomic networks and generic protein interaction networks, they offer a foundation for developing next-generation therapeutics for treating AD.”
The full proteomic data and models from the study are publicly available, for use in further investigations by other research teams.
