Abstract
Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease characterized by the progressive loss of motor neurons. While several pathogenic mutations have been identified, the vast majority of ALS cases have no family history of disease. Thus, for most ALS cases, the disease may be a product of multiple pathways contributing to varying degrees in each patient. Using machine learning algorithms, we stratify the transcriptomes of 148 ALS postmortem cortex samples into three distinct molecular subtypes. The largest cluster, identified in 61% of patient samples, displays hallmarks of oxidative and proteotoxic stress. Another 19% of the samples shows predominant signatures of glial activation. Finally, a third group (20%) exhibits high levels of retrotransposon expression and signatures of TARDBP/TDP-43 dysfunction. We further demonstrate that TDP-43 (1) directly binds a subset of retrotransposon transcripts and contributes to their silencing in vitro, and (2) pathological TDP-43 aggregation correlates with retrotransposon de-silencing in vivo.