Researchers headed by a team at the University of California, Irvine, Joe C. Wen School of Population & Public Health have built what they suggest is the first cell type-specific gene regulatory network (GRN) map for Alzheimer’s disease (AD), which shows how genes causally regulate one another across different types of brain cells affected by AD.
The researchers developed a machine learning framework, SIGNET (Statistical Inference on Gene Regulatory Networks), which reveals cause-and-effect relationships rather than simple genetic correlations, and applied this to uncover key biological pathways that may drive memory loss and brain degeneration. Their results pointed to numerous influential “hub genes” that offer promising potential new targets for early detection and therapeutic intervention. The investigators say their methodology is also applicable to other complex diseases, including cancer.
Research leads Min Zhang, MD, PhD, and Dabao Zhang, PhD, and colleagues reported on the development and application of SIGNET in Alzheimer’s & Dementia: The Journal of the Alzheimer’s Association, in a paper titled “From correlation to causation: cell-type-specific gene regulatory networks in Alzheimer’s disease.” In their paper, the researchers concluded, “By identifying novel AD-associated hubs and key pathways as potential biomarkers, this study advances our understanding of the molecular mechanisms driving AD and offers significant potential for developing targeted diagnoses and treatments.”
Alzheimer’s disease is the leading cause of dementia and is projected to affect nearly 14 million Americans over the age of 65 by 2060. Scientists have already linked many genes, such as apolipoprotein E (APOE) and amyloid precursor protein (APP), to the disease, but still don’t fully understand how these genes disrupt healthy brain function.
The authors further noted, “… a major bottleneck for understanding AD pathogenesis lies in its biological complexity, involving intra- and intercellular interactions, neuronal loss, gliosis, and accumulation of pathological proteins, which are all modulated under complex gene regulatory networks (GRNs) … understanding interactions between genes and transcription factors (TFs) within specific cell types is crucial for uncovering the cellular processes underlying neurodegeneration and AD progression, ultimately guiding the development of targeted treatments and preventive care for AD.”
“Different types of brain cells play distinct roles in Alzheimer’s disease, but how they interact at the molecular level has remained unclear,” said Min Zhang, co-corresponding author and professor of epidemiology and biostatistics. “Our work provides cell type-specific maps of gene regulation in the Alzheimer’s brain, shifting the field from observing correlations to uncovering the causal mechanisms that actively drive disease progression.”
The researchers developed their scalable, high-performance computing method, SIGNET, which they used to integrate and analyze single-nucleus RNA sequencing (snRNAseq) and whole-genome sequencing (WGS) data from 272 AD patients in the Religious Orders Study and the Rush Memory and Aging Project (ROSMAP). The analyses enabled them to identify and construct cell-type-specific causal gene regulatory networks for six major types of brain cells. This then allowed the team to determine which genes are likely controlling other genes, something traditional correlation-based tools cannot reliably do. “The constructed GRNs show upstream non-TF genes regulating TFs and interconnected TF regulatory modules, highlighting the complexity of AD regulatory mechanisms beyond TF-centric assumptions,” they stated.
“Most gene-mapping tools can show which genes move together, but they can’t tell which genes are actually driving the changes,” said Dabao Zhang, co-corresponding author and professor of epidemiology and biostatistics. “Some methods also make unrealistic assumptions, such as ignoring feedback loops between genes. Our approach takes advantage of information encoded in DNA to enable the identification of true cause-and-effect relationships between genes in the brain.”
The scientists found that the most dramatic gene disruptions in Alzheimer’s disease occur in excitatory neurons—the nerve cells that send activating signals—with analyses of nearly 6,000 cause-and-effect interactions indicating that these cells undergo extensive rewiring as the disease progresses.
They also pinpointed hundreds of “hub genes” that act as major control centers, influencing many other genes and likely playing key roles in driving harmful changes. “Our comprehensive analysis of cell-type-specific causal GRNs revealed excitatory neurons as exhibiting the most extensive regulatory network and greatest diversity in regulatory effects, with regulatory hubs predominantly represented among hub genes,” they noted. These could serve as new targets for early detection and therapeutic intervention. In addition, the team discovered new regulatory roles for well-known genes such as APP, which strongly controlled other genes in inhibitory neurons.
Importantly, the researchers confirmed these findings using an independent set of human brain samples, strengthening confidence that these gene-to-gene relationships reflect real biological mechanisms involved in Alzheimer’s disease. “Moving forward, we will dive deeper into the current results to investigate networks involved in AD-specific pathologies across different cell types,” they stated. “We plan to perform differential gene regulatory analysis between AD and healthy samples in these pathways to identify AD-specific regulatory patterns. This comparison will allow us to distinguish the regulatory changes involved in neurodegeneration from normal cell activities during aging.
The team noted that SIGNET can also be used to study many other complex diseases, including cancer, autoimmune disorders, and mental health conditions. “… this analytical pipeline is broadly applicable to other complex diseases, including cancers, enabling the integration of multi-omics data for constructing cell-type-specific causal GRNs across diverse biological contexts,” they concluded.
