Proteomic landscape of Alzheimer's Disease: novel insights into pathogenesis and biomarker discovery

Abstract

Mass spectrometry-based proteomics empowers deep profiling of proteome and protein posttranslational modifications (PTMs) in Alzheimer's disease (AD). Here we review the advances and limitations in historic and recent AD proteomic research. Complementary to genetic mapping, proteomic studies not only validate canonical amyloid and tau pathways, but also uncover novel components in broad protein networks, such as RNA splicing, development, immunity, membrane transport, lipid metabolism, synaptic function, and mitochondrial activity. Meta-analysis of seven deep datasets reveals 2,698 differentially expressed (DE) proteins in the landscape of AD brain proteome (n = 12,017 proteins/genes), covering 35 reported AD genes and risk loci. The DE proteins contain cellular markers enriched in neurons, microglia, astrocytes, oligodendrocytes, and epithelial cells, supporting the involvement of diverse cell types in AD pathology. We discuss the hypothesized protective or detrimental roles of selected DE proteins, emphasizing top proteins in "amyloidome" (all biomolecules in amyloid plaques) and disease progression. Comprehensive PTM analysis represents another layer of molecular events in AD. In particular, tau PTMs are correlated with disease stages and indicate the heterogeneity of individual AD patients. Moreover, the unprecedented proteomic coverage of biofluids, such as cerebrospinal fluid and serum, procures novel putative AD biomarkers through meta-analysis. Thus, proteomics-driven systems biology presents a new frontier to link genotype, proteotype, and phenotype, accelerating the development of improved AD models and treatment strategies.

Keywords: Abeta; Alzheimer’s disease; Amyloidome; Biomarker; Mass spectrometry; PTM; Pathogenesis; Proteome; Proteomics; Tau.

Fig. 1

Major historical events in mass spectrometry and AD research. Edman degradation is also included. The AD proteomics studies are highlighted. The information is compiled from several online resources (https://www.nature.com/collections/aajfehieag, https://www.hupo.org/Proteomics-Timeline, https://masspec.scripps.edu/learn/ms, https://www.alzforum.org/timeline, and https://www.alzheimers.net/history-of-alzheimers) and references [, , , , , –102]. MS detection instruments include sector MS, time-of-flight (TOF), quadrupole, Fourier-transform ion cyclotron resonance (FTICR), triple quadrupole, and orbitrap. Common ionization methods include atmospheric pressure chemical ionization (APCI), electrospray ionization (ESI), matrix-assisted laser desorption/ionization (MALDI), desorption electrospray ionization (DESI), and direct analysis in real time (DART). Biomolecules may be fractionated by nanoscale liquid chromatography (LC) with sub-2 μm resin to improve resolution, or by multi-dimensional protein identification technology (MudPIT). MS precursor ions can be fragmented by collision-induced dissociation (CID), electron-capture dissociation (ECD), electron-transfer dissociation (ETD), or higher-energy collisional dissociation (HCD). Quantitative strategies include two dimensional polyacryl amide gel electrophoresis (2D PAGE), isotope-coded affinity tag (ICAT), stable isotope labeling by/with amino acids in cell culture (SILAC), tandem mass tag (TMT), isobaric tags for relative and absolute quantitation (iTRAQ), and the data-independent acquisition (DIA) methods. Selected/multiple reaction monitoring (SRM/MRM) is a MS technique for analyzing pre-defined molecules. Database search tools contain SEQUEST, MASCOT, and the target-decoy strategy. CAA: cerebral amyloid angiopathy; LCM: laser capture microdissection; GWAS: genome-wide association study; CSF: cerebrospinal fluid; and FDA: the United States Food and Drug Administration

Fig. 2

Protein samples are analyzed by pre-MS sample processing, MS data acquisition and post-MS bioinformatic data processing. Protein quantification can be achieved by label-free methods, such as spectral counting (SC), extracted ion current (XIC), and data-independent acquisition (DIA), or by isotope-labeling methods, such as SILAC and TMT. In addition, MS may also be operated to analyze targeted proteins/peptides by multiple reaction monitoring (MRM) or parallel reaction monitoring (PRM)

Fig. 3

A deep proteomic workflow of TMT-LC/LC-MS/MS. The 16-plex TMT reagents are represented by different colors, which react with amine groups at peptide N-termini and lysine residues. In addition to the amine-reactive group, the isobaric TMT reagents also contain a reporter ion group and a balance group. The mass difference in the reporter group is offset by the balance group, enabling isobaric labeling and pooling. The pooled peptides are fractionated by LC/LC, and identified as mixed, isobaric precursor ions. After fragmentation, the TMT tags are cleaved between the reporter and balance groups, generating reporter ions for quantification

Fig. 4

A meta-analysis integrating 7 deep AD datasets and identifying 12,017 proteins. In each dataset, the p value for control-AD comparison is derived by one-tailed t-test. The p values are combined by the Fisher’s method, followed by multiple-hypothesis correction using the Benjamini-Hochberg FDR procedure. A total of 2,698 DE proteins are accepted with the FDR cutoff of 1 %. Compared to 167 reported AD genes and risk loci, 35 are overlapped with the protein DE list. Based on cell type specific genes, the DE list contains 638 genes/proteins specific to all five major cell types in the brain

Fig. 5

The equilibrium model of deleterious and protective factors during AD disease progression. Among the biological processes and cellular pathways activated at the asymptomatic stage of AD, some may exert protective roles. However, with exacerbation of the harmful insults during disease progression, the protective effect is exhausted. The resulting imbalance leads to neuronal degeneration and the onset of clinical symptoms. The pathway information is extracted from current AD proteomic studies [74, 75, 78, 161]

Fig. 6

A meta-analysis integrating 6 deep CSF datasets and identifying 5,939 proteins. Using the same method in Fig. 4, individual and combined p values for control-AD comparison are computed, followed by the FDR analysis. With the FDR cutoff of 1 %, 476 DE proteins are accepted, 109 are overlapped with the list of DE proteins in AD brain tissues