Large-scale deep multi-layer analysis of Alzheimer’s disease brain reveals strong proteomic disease-related changes not observed at the RNA level

RNA-Seq offers an unbiased view of known and novel transcripts and has identified candidate hub genes within networks linked to lipid metabolism and innate immunity.

“RNA-Seq offers an unbiased view of known and novel transcripts and has identified candidate “hub” genes within networks linked to lipid metabolism and innate immunity122,130,159,160.”

Integrating genomics, epigenomics, and transcriptomics within the same biological samples can yield highly informative insights into the molecular mechanisms of AD.

“There is increasing recognition that integrating genomics, epigenomics, and transcriptomics within the same biological samples, a strategy commonly referred to as multi-omics integration, can yield highly informative insights into the molecular mechanisms of AD159,160.”

Multi-omics integration that combines genomic, transcriptomic, proteomic, and metabolomic layers is creating more comprehensive models of AD biology.

“Multi-omics integration that combines genomic, transcriptomic, proteomic, and metabolomic layers is creating more comprehensive models of AD biology156,159,160.”

Proteomics and metabolomics directly capture functional endpoints such as proteins and metabolites, which often act as early indicators or amplifiers of disease progression.

“Although genetic and transcriptomic studies have provided critical insights into AD, proteomics and metabolomics directly capture functional endpoints such as proteins and metabolites, which often act as early indicators or amplifiers of disease progression156,159.”

Recent advances in multi-omics technologies have enabled simultaneous profiling of genomics, transcriptomics, proteomics, metabolomics, and epigenomics within AD cohorts.

“Recent advances in multi-omics technologies have enabled simultaneous profiling of genomics, transcriptomics, proteomics, metabolomics, and epigenomics within AD cohorts.”

A study integrating metabolomics and proteomics in APOE ε4 carriers revealed disruptions in sphingolipid and glycerophospholipid pathways.

“A study integrating metabolomics and proteomics in APOE ε4 carriers revealed disruptions in sphingolipid and glycerophospholipid pathways.”

One study combining proteomic and metabolomic data identified dysregulated sphingolipid and ceramide metabolism in APOE ε4 carriers, findings supported by post-mortem analyses.

“One study combining proteomic and metabolomic data identified dysregulated sphingolipid and ceramide metabolism in APOE ε4 carriers, findings supported by post-mortem analyses.”

MDK, SMOC1, and NTN1 have been shown to colocalize with Aβ plaque and are promising biomarkers associated with AD pathology.

“These proteins have been shown to colocalize with Aβ plaque and are promising biomarkers associated with AD pathology.”

SMOC1, NTN1, and MDK are strongly associated with amyloid plaques and are considered hub proteins in both AsymAD and AD.

“These proteins are strongly associated with amyloid plaques and are considered hub proteins in both AsymAD and AD. 43 , 51 , 80”

The correlation between mRNA levels and their encoded protein levels is weak.

“Additionally, the correlation between levels of mRNA and their encoded proteins is weak8,9, while proteins are typically the molecule that most-directly act on disease.”

Studies correlating transcriptomic profiles with proteomic, metabolomic, and epigenomic data have revealed key discrepancies between RNA and protein expression in AD.

“Studies correlating transcriptomic profiles with proteomic, metabolomic, and epigenomic data have already revealed key discrepancies between RNA and protein expression in AD89,159,160.”

TMT-MS peptide-level data following extraction with urea or TBS revealed that peptides mapping to the phosphatase domain are reduced in AD while C-terminal peptides are elevated.

“Analyses of peptide-level data of INPP5D using TMT-MS34 following extraction with urea or TBS both revealed that peptides mapping to the phosphatase domain are reduced in AD while C-terminal peptides are elevated (Fig. 2l–o, Supplementary Data 12).”

Discordance or decoupling in mRNA and protein expression of the same genes has been reported in human aging and neurodegenerative disorders such as tauopathy and AD.

“The discordance or decoupling in mRNA and protein expression of the same genes has been reported in studies with human aging [67] and neurodegenerative disorders such as tauopathy [68] and AD [69]”

We found a 60% π1 replication rate with earlier studies (n = 516 DLPFC samples) for clinical diagnosis.

“We also found a 60% π1 replication rate with earlier studies (n = 516 DLPFC samples) for clinical diagnosis, 43 suggesting the replicability of prior results.”

The kernel-based model used in that study was non-transparent and performed poorly in non-European cohorts, highlighting the challenge of cross-population generalizability.

“However, the kernel-based model used was non-transparent and performed poorly in non-European cohorts, highlighting the challenge of cross-population generalizability159,293.”

Predictive accuracy declined in ethnically diverse cohorts, underscoring the importance of ancestry-inclusive validation.

“However, predictive accuracy declined in ethnically diverse cohorts, underscoring the importance of ancestry-inclusive validation41,159,284”

MDK was the most overexpressed protein in 5xFAD mice.

“MDK was the most overexpressed protein in 5xFAD mice and has recently been identified as also highly overexpressed in human AD brains [74].”