Study uncovers and predicts widespread molecular changes in ALS brain cells

Researchers at the University of Toronto’s Temerty Faculty of Medicine have found widespread genetic and protein changes in the brain cells of people with amyotrophic lateral sclerosis (ALS), which may help shed light on mechanisms of the disease.

The study, recently published in Cell Genomics, also describes development of a deep-learning tool to help predict whether the changes that contribute to ALS and other neurodegenerative diseases are present in cells.

Paul McKeever, a co-lead author of the study, says the researchers developed an ‘atlas’ of RNA dysregulation across the brain cells in the orbito-frontal cortex, through a broad analysis of genetic changes. That brain region is associated with behavioural changes in ALS and frontotemporal lobar degeneration (FTLD).

“Many researchers study the dysregulation of specific RNA binding proteins in ALS brain cells and how that impacts cell function. We took a step back and asked whether we can identify those proteins in the cells, how they might impact RNA processing, and therefore how genes are expressed in these cells,” says McKeever, a postdoctoral fellow with Janice Robertson, an investigator at the Tanz Centre for Research in Neurodegenerative Diseases who holds the James Hunter Family Chair in ALS Research. 

“Our approach was broad in scope and deeply comparative, allowing us to examine many cell types and disease subtypes side by side and uncover patterns of change that hadn’t been seen before.”

The team used a technique called single nucleus RNA sequencing, which analyzes RNA from individual cell nuclei rather than whole cells, and is ideal for sequencing previously frozen patient samples. Understanding RNA sequences gives insight into the cell pathways affected in disease and the mechanisms involved, and can lead to identification of new therapeutic targets.

McKeever, working with researchers at the Tanz Centre and the department of laboratory medicine and pathobiology, sequenced RNA from pea-sized samples of brain tissue containing a variety of cell types — from people with genetic and sporadic forms of ALS, as well as from people without ALS.

Their results show some differences in RNA transcription between cell types, brain regions and ALS subtypes, but also some consistent patterns of changes across cell types, especially in neurons — suggesting those changes are fundamental mechanisms that drive the advancement of disease.

“We found there are shared and distinct changes both between cell types and disease cell types, and this extended to independent studies that really validated our findings,” says McKeever. “This provides a more granular understanding of the differences between brain cells in ALS and FTLD, and illuminates the disease mechanisms specific to cell types, offering a valuable resource for advancing therapeutic research.”

McKeever says that a better understanding of changes to RNA and the cell’s protein networks can help identify signatures, or biomarkers of disease, that can aid diagnosis. It could also help researchers learn more about the cell pathways that lead to disease.

One of the most widespread changes the researchers saw in many ALS brain cells, especially neurons and glia, was dysregulation of alternative polyadenylation (APA) — a process that takes place after RNA is transcribed and affects the process of translation into proteins.

Building on this finding, Gary Bader, professor of molecular genetics at the Donnelly Centre for Cellular and Biomolecular Research, and his PhD student at the time, Aiden Sababi, used computational biology to develop a predictive deep-learning tool.

They used data from McKeever’s RNA sequencing analysis to predict whether a cell would have dysregulated APA based on its RNA sequences and networks of RNA binding proteins.

Sababi, now a postdoctoral fellow at McGill University, says the deep-learning tool is valuable for predicting which cells might have dysregulated APAs, but also for generating ideas on future research about proteins most likely to drive disease development and progression.

“We try to focus on biological function, rather than prediction alone. We don’t just want to see where APA changes but learn why it changes in ALS, so the tool is valuable for generating hypotheses and avenues of research,” says Sababi.

“The advantage of this deep-learning tool is that once the model makes reliable predictions, you can start to look at why the model makes the prediction it does. The model confirmed the evidence from the Tanz Centre team that APA dysregulation is happening, and now the question is why.”

Graham Collingridge, director of the Tanz Centre, says the research represents an important step forward in understanding how the disease develops.

“ALS is a devastating disease, and understanding of the causes and developing effective treatments is a high priority for Tanz researchers and others around the world,” he says. “This study, and the associated deep learning tool that has been developed, constitute important steps in this direction.”

Sababi used a relatively small data set to train the deep learning model, but based on the promising results in this study, the research team wants to expand its capability.

Robertson, McKeever, Sababi and Bader were recently awarded a McLaughlin Accelerator Grant in Genomic Medicine from U of T to advance the model, expand its capabilities and potentially use it to analyze similar data for other neurodegenerative diseases, and cancers.

“This new grant will allow us to advance the model, train it on larger data sets, and hopefully validate what we see in this current data,” says Sababi.

“As we get more confident with this model, we can generate better hypotheses for mechanisms of disease and better validate the approaches we have. I think we are getting closer to having a good impact from this avenue of biology.”

This research was funded by the ALS Society of Canada, Canadian Institutes for Health Research, the ALS Association, ALS Double Play, Canadian Consortium on Neurodegeneration in Aging, and the U.S. National Institutes of Health.