Inferring protein expression changes from mRNA in Alzheimer’s dementia using deep neural networks

Abstract

Identifying the molecular systems and proteins that modify the progression of Alzheimer’s disease and related dementias (ADRD) is central to drug target selection. However, discordance between mRNA and protein abundance, and the scarcity of proteomic data, has limited our ability to advance candidate targets that are mainly based on gene expression. Therefore, by using a deep neural network that predicts protein abundance from mRNA expression, here we attempt to track the early protein drivers of ADRD. Specifically, by applying the clei2block deep learning model to 1192 brain RNA-seq samples, we identify protein modules and disease-associated expression changes that were not directly observed at the mRNA level. Moreover, pseudo-temporal trajectory inference based on the predicted proteome became more closely correlated with cognitive decline and hippocampal atrophy compared to RNA-based trajectories. This suggests that the predicted changes in protein expression could provide a better molecular representation of ADRD progression. Furthermore, overlaying clinical traits on protein pseudotime trajectory identifies protein modules altered before cognitive impairment. These results demonstrate how our method can be used to identify potential early protein drivers and possible drug targets for treating and/or preventing ADRD.

Synaptic proteomics reveal distinct molecular signatures of cognitive change and C9ORF72 repeat expansion in the human ALS cortex

“…However, protein alterations in small structures such as the synapse can be diluted in whole cell/tissue analyses. Furthermore, deep proteomic profiling of human brain can reveal changes in protein expression that are not highlighted by RNA expression analysis [ 19 , 20 ]. This highlights the importance of using well-characterised human samples for the development of comprehensive proteomic datasets.…”

PICALM and Alzheimer’s Disease: An Update and Perspectives

“…Cautions need to be taken to interpret the data about PICALM expression in human autopsy tissues. The results of PICALM expression are often based on the bulk tissue of human postmortem brains [ 32 , 33 , 124 ]. Since the cell type composition of brain tissue undergoes alterations during the progression of neurodegenerative disease, such bulk data may confound the interpretation of differences in gene expression between AD patients and controls.…”

“…During the progression of AD, PICALM mRNA expression levels are increased in the brain [ 32 , 124 ] and in the blood [ 125 ] of AD patients. In contrast, the PICALM protein level in the total bulk tissue is decreased in AD brains [ 33 ].…”

“…In line with this discordant observation, Tasaki et al, have recently reported that there is a general divergence between the levels of transcripts (mRNA) and protein expression in postmortem AD brain tissue. By mass-spectrometry, they observed a significant reduction of PICALM protein in a correlation with 14.3% faster cognitive decline [ 124 ].…”

Unbiased Classification of the Human Brain Proteome Resolves Distinct Clinical and Pathophysiological Subtypes of Cognitive Impairment

“…Yet, marked spatial, temporal, and quantitative differences between mRNA and protein expression make proteomic subtyping a potential source of unique biological insights [25,26]. Furthermore, compared to RNA differences, protein changes associate more strongly with AD clinical and pathological phenotypes, consistent with their being more proximate mediators of disease manifestations [27][28][29]. Using unbiased co-expression network analysis, we have demonstrated that the cortical brain regions of those with pathologically defined early-and late-stage LOAD feature a wide range of altered protein systems not observed in the transcriptome [30][31][32][33].…”