Search

The biological processes that are disrupted in the Alzheimer’s disease (AD) brain remain incompletely understood. We recently performed a proteomic analysis of >2000 brains to better understand these changes, which highlighted alterations in astrocytes and microglia as likely key drivers of disease. Here, we extend this analysis by analyzing >1000 brain tissues using a tandem mass tag mass spectrometry (TMT-MS) pipeline, which allowed us to nearly triple the number of quantified proteins across cases. A consensus protein co-expression network analysis of this deeper dataset revealed new co-expression modules that were highly preserved across cohorts and brain regions, and strongly altered in AD. Nearly half of the protein co-expression modules, including modules significantly altered in AD, were not observed in RNA networks from the same cohorts and brain regions, highlighting the proteopathic nature of AD. Two such AD-associated modules unique to the proteomic network included a module related to MAPK signaling and metabolism, and a module related to the matrisome. Analysis of paired genomic and proteomic data within subjects showed that expression level of the matrisome module was influenced by the APOE ε4 allele, but was not related to the rate of cognitive decline after adjustment for neuropathology. In contrast, the MAPK/metabolism module was strongly associated with the rate of cognitive decline. Disease-associated modules unique to the proteome are sources of promising therapeutic targets and biomarkers for AD.

Background There is an association between repetitive head injury (RHI) and a pathologic diagnosis of chronic traumatic encephalopathy (CTE) characterized by the aggregation of proteins including tau. The underlying molecular events that cause these abnormal protein accumulations remain unclear. Here, we hypothesized that identifying the human brain proteome from serial CTE stages (CTE I-IV) would provide critical new insights into CTE pathogenesis. Brain samples from frontotemporal lobar degeneration due to microtubule associated protein tau (FTLD-MAPT) mutations were also included as a distinct tauopathy phenotype for comparison. Methods Isobaric tandem mass tagged labeling and mass spectrometry (TMT-MS) followed by integrated differential and co-expression analysis (i.e., weighted gene co-expression network analysis (WGCNA)) was used to define modules of highly correlated proteins associated with clinical and pathological phenotypes in control (n = 23), CTE (n = 43), and FTLD-MAPT (n = 12) post-mortem cortical tissues. We also compared these findings to network analysis of AD brain. Results We identified over 6000 unique proteins across all four CTE stages which sorted into 28 WGCNA modules. Consistent with Alzheimer’s disease, specific modules demonstrated reduced neuronal protein levels, suggesting a neurodegeneration phenotype, while other modules were increased, including proteins associated with inflammation and glial cell proliferation. Notably, unique CTE-specific modules demonstrated prominent enrichment of immunoglobulins, including IGHM and IGLL5, and extracellular matrix (ECM) proteins as well as progressive protein changes with increasing CTE pathologic stage. Finally, aggregate cell subtype (i.e., neurons, microglia, astrocytes) protein abundance levels in CTE cases were similar in expression to AD, but at intermediate levels between controls and the more exaggerated phenotype of FTLD-MAPT, especially in astrocytes. Conclusions Overall, we identified thousands of protein changes in CTE postmortem brain and demonstrated that CTE has a pattern of neurodegeneration in neuronal-synaptic and inflammation modules similar to AD. We also identified unique CTE progressive changes, including the enrichment of immunoglobulins and ECM proteins even in early CTE stages. Early and sustained changes in astrocyte modules were also observed. Overall, the prominent overlap with FTLD-MAPT cases confirmed that CTE is on the tauopathy continuum and identified CTE stage specific molecular phenotypes that provide novel insights into disease pathogenesis.

The hallmark amyloid-β and tau deposition of Alzheimer’s disease (AD) represents only a fraction of its diverse pathophysiology. Molecular subtyping using large-scale -omic strategies can help resolve this biological heterogeneity. Using quantitative mass spectrometry, we measured ~8,000 proteins across >600 dorsolateral prefrontal cortex tissues from Religious Orders Study and Rush Memory and Aging Project participants with clinical diagnoses of no cognitive impairment, mild cognitive impairment (MCI), and AD dementia. Unbiased classification of MCI and AD cases based on individual proteomic profiles resolved three classes with expression differences across numerous cell types and biological ontologies. Two classes displayed molecular signatures atypical of those previously observed in AD neurodegeneration, such as elevated synaptic and decreased inflammatory markers. In one class, these atypical proteomic features were associated with clinical and pathological hallmarks of cognitive resilience. These results promise to better define disease heterogeneity within AD and meaningfully impact its diagnostic and therapeutic precision.

Post-translational modifications (PTMs) within splicing factor RNA-binding proteins (RBPs), such as phosphorylation, regulate several critical steps in RNA metabolism including spliceosome assembly, alternative splicing and mRNA export. Notably, the arginine-/serine-rich (RS) domains in SR proteins are densely modified by phosphorylation compared with the remainder of the proteome. Previously, we showed that dephosphorylation of SRSF2 regulated increased interactions with similar arginine-rich RBPs U1-70K and LUC7L3. In this work, we dephosphorylated nuclear extracts using phosphatase in vitro and analyzed equal amounts of detergent-soluble and -insoluble fractions by mass spectrometry-based proteomics. Correlation network analysis resolved 27 distinct modules of differentially soluble nucleoplasm proteins. We found classes of arginine-rich RBPs that decrease in solubility following dephosphorylation and enrich to the insoluble pelleted fraction, including the SR protein family and the SR-like LUC7L RBP family. Importantly, increased insolubility was not observed across broad classes of RBPs. Phosphorylation regulated SRSF2 structure, as dephosphorylated SRSF2 formed high molecular weight oligomeric species in vitro. Reciprocally, phosphorylation of SRSF2 by serine-/arginine protein kinase 2 (SRPK2) in vitro prevented high molecular weight SRSF2 species formation. Furthermore, we pharmacologically inhibited SRPKs in mammalian cells and observed increased cytoplasmic granules as well as the formation of cytoplasmic SRSF2 tubular structures that associate with microtubules by immunocytochemical staining. Collectively, these findings demonstrate that phosphorylation may be a critical modification that prevents arginine-rich RBP insolubility and oligomerization.

Alzheimer’s disease (AD) is characterized by an early, asymptomatic phase (AsymAD) in which individuals exhibit amyloid-beta (Aβ) plaque accumulation in the absence of clinically detectable cognitive decline. Here we report an unbiased multiplex quantitative proteomic and phosphoproteomic analysis using tandem mass tag (TMT) isobaric labeling of human post-mortem cortex (n = 27) across pathology-free controls, AsymAD and symptomatic AD individuals. With off-line high-pH fractionation and liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) on an Orbitrap Lumos mass spectrometer, we identified 11,378 protein groups across three TMT 11-plex batches. Immobilized metal affinity chromatography (IMAC) was used to enrich for phosphopeptides from the same TMT-labeled cases and 51,736 phosphopeptides were identified. Of these, 48,992 were quantified by TMT reporter ions representing 33,652 unique phosphosites. Two reference standards in each TMT 11-plex were included to assess intra- and inter-batch variance at the protein and peptide level. This comprehensive human brain proteome and phosphoproteome dataset will serve as a valuable resource for the identification of biochemical, cellular and signaling pathways altered during AD progression., Measurement(s) Proteomic Profile • Phosphoprotein Profile Technology Type(s) liquid chromatography-tandem mass spectrometry • immobilized metal affinity chromatography Factor Type(s) pathology-free control; asymptomatic AD; symptomatic AD • intra-batch variance • inter-batch variance Sample Characteristic - Organism Homo sapiens Machine-accessible metadata file describing the reported data: 10.6084/m9.figshare.12901817

Posttranslational modifications (PTMs) such as phosphorylation of RNA-binding proteins (RBPs) regulate several critical steps in RNA metabolism, including spliceosome assembly, alternative splicing, and mRNA export. Notably, serine-/arginine- (SR)-rich RBPs are densely phosphorylated compared with the remainder of the proteome. Previously, we showed that dephosphorylation of the splicing factor SRSF2 regulated increased interactions with similar arginine-rich RBPs U1-70K and LUC7L3. However, the large-scale functional and structural impact of these modifications on RBPs remains unclear. In this work, we dephosphorylated nuclear extracts using phosphatase in vitro and analyzed equal amounts of detergent-soluble and -insoluble fractions by mass-spectrometry-based proteomics. Correlation network analysis resolved 27 distinct modules of differentially soluble nucleoplasm proteins. We found classes of arginine-rich RBPs that decrease in solubility following dephosphorylation and enrich the insoluble pelleted fraction, including the SR protein family and the SR-like LUC7L RBP family. Importantly, increased insolubility was not observed across broad classes of RBPs. We determined that phosphorylation regulated SRSF2 structure, as dephosphorylated SRSF2 formed high-molecular-weight oligomeric species in vitro. Reciprocally, phosphorylation of SRSF2 by serine/arginine protein kinase 2 (SRPK2) in vitro decreased high-molecular-weight SRSF2 species formation. Furthermore, upon pharmacological inhibition of SRPKs in mammalian cells, we observed SRSF2 cytoplasmic mislocalization and increased formation of cytoplasmic granules as well as cytoplasmic tubular structures that associated with microtubules by immunocytochemical staining. Collectively, these findings demonstrate that phosphorylation may be a critical modification that prevents arginine-rich RBP insolubility and oligomerization.

Our understanding of the biological changes in the brain associated with Alzheimer’s disease (AD) pathology and cognitive impairment remains incomplete. To increase our understanding of these changes, we analyzed dorsolateral prefrontal cortex of control, asymptomatic AD, and AD brains from four different centers by label-free quantitative mass spectrometry and weighted protein co-expression analysis to obtain a consensus protein co-expression network of AD brain. This network consisted of 13 protein co-expression modules. Six of these modules correlated with amyloid-β plaque burden, tau neurofibrillary tangle burden, cognitive function, and clinical functional status, and were altered in asymptomatic AD, AD, or in both disease states. These six modules reflected synaptic, mitochondrial, sugar metabolism, extracellular matrix, cytoskeletal, and RNA binding/splicing biological functions. The identified protein network modules were preserved in a community-based cohort analyzed by a different quantitative mass spectrometry approach. They were also preserved in temporal lobe and precuneus brain regions. Some of the modules were influenced by aging, and showed changes in other neurodegenerative diseases such as frontotemporal dementia and corticobasal degeneration. The module most strongly associated with AD pathology and cognitive impairment was the sugar metabolism module. This module was enriched in AD genetic risk factors, and was also highly enriched in microglia and astrocyte protein markers associated with an anti-inflammatory state, suggesting that the biological functions it represents serve a protective role in AD. Proteins from the sugar metabolism module were increased in cerebrospinal fluid from asymptomatic AD and AD cases, highlighting their potential as biomarkers of the altered brain network. In this study of >2000 brains and nearly 400 cerebrospinal fluid samples by quantitative proteomics, we identify proteins and biological processes in AD brain that may serve as therapeutic targets and fluid biomarkers for the disease.

Our understanding of Alzheimer's disease (AD) pathophysiology remains incomplete. Here we used quantitative mass spectrometry and coexpression network analysis to conduct the largest proteomic study thus far on AD. A protein network module linked to sugar metabolism emerged as one of the modules most significantly associated with AD pathology and cognitive impairment. This module was enriched in AD genetic risk factors and in microglia and astrocyte protein markers associated with an anti-inflammatory state, suggesting that the biological functions it represents serve a protective role in AD. Proteins from this module were elevated in cerebrospinal fluid in early stages of the disease. In this study of >2,000 brains and nearly 400 cerebrospinal fluid samples by quantitative proteomics, we identify proteins and biological processes in AD brains that may serve as therapeutic targets and fluid biomarkers for the disease.

Patients with Alzheimer’s disease (AD) and Parkinson’s disease (PD) often have overlap in clinical presentation and brain neuropathology suggesting that these two diseases share common underlying mechanisms. Currently, the molecular pathways linking AD and PD are incompletely understood. Utilizing Tandem Mass Tag (TMT) isobaric labeling and synchronous precursor selection-based MS3 (SPS-MS3) mass spectrometry, we performed an unbiased quantitative proteomic analysis of post-mortem human brain tissues (n=80) from four different groups defined as controls, AD, PD, and co-morbid AD/PD cases across two brain regions (frontal cortex and anterior cingulate gyrus). In total, we identified 11 840 protein groups representing 10 230 gene symbols, which map to ~65% of the protein coding genes in brain. The utility of including two reference standards in each TMT 10-plex assay to assess intra- and inter-batch variance is also described. Ultimately, this comprehensive human brain proteomic dataset serves as a valuable resource for various research endeavors including, but not limited to, the identification of disease-specific protein signatures and molecular pathways that are common in AD and PD. Machine-accessible metadata file describing the reported data (ISA-Tab format)

Figure S1. SDS-PAGE of Brain Homogenates. Dorsolateral prefrontal cortex (DLPFC) brain tissue homogenates from cases shown in Table S1 were analyzed by SDS-PAGE to assess sample integrity prior to TMT labeling and mass spectrometry analysis. Gels were stained with Coomassie Blue to visualize protein. AD, Alzheimer’s disease; AS, asymptomatic Alzheimer’s disease; CT, control; MCI, mild cognitive impairment. (PDF 22000 kb)

Figure S14. Differential Abundance of Alternative Exon-Exon Junction Peptides by TMT Network Module. For case group comparisons AD vs. AsymAD (top), AD vs. control (middle), and AsymAD vs. control (bottom), the fraction of alternative exon-exon junction (alt-EEjxn) peptides within each network module that were significantly different between the two case groups was plotted by bar graph, with each bar color coded according to the average log2 difference of the alt-EEjxn peptides in each direction (increased or decreased). The arrows in the AD vs. AsymAD comparison highlight modules that showed an increase in the fraction of alt-EEjxns or an increase in the magnitude of differential abundance, or both, compared to AD vs. control. (PDF 718Â kb)

Table S7. AD Risk Factor Protein Alternative Exon-Exon Junction Peptides Significantly Altered by Case Status. Of the five AD risk factor proteins that had quantifiable alt-EEjxn peptides in the BLSA-TMT analysis, three had alt-EEjxn peptides that were significantly, or nearly significantly, different in abundance by case status. For PTK2B 20038, module eigenprotein correlation was not performed due to the number of missing values for quantitation (≥25). A complete description of each alt-EEjxn peptide is provided in Supplementary Data. ME bicor, module eigenprotein bicorrelation; kME, correlation value to the module eigenprotein; AD, Alzheimer’s disease; AsymAD, asymptomatic Alzheimer’s disease. (DOCX 29 kb)

Figure S7. Differential Protein Abundance between AD and AsymAD. Proteins that were significantly increased (160) or decreased (190) in AD compared to AsymAD, color-coded by TMT network module membership, are shown. The horizontal dotted line represents p = 0.05. Interactive plots for AD vs. AsymAD, AD vs. control, and AsymAD vs. control differential protein abundance are provided in Supplementary Data. (PDF 10400 kb)

Figure S5. TMT Protein Network Modules Enriched for AD Risk Factors. Graphical representation of the correlation relationships among TMT network module proteins for the four modules identified to contain enrichment of AD risk factors from GWAS, along with the relationship of each module to case status, neuritic amyloid plaque load (CERAD score), and tau tangle burden (Braak stage). Proteins identified by GWAS as AD risk factors are highlighted in red. Only the top 100 proteins by kME value are shown for the M4 yellow (257 total proteins) and M7 black (162 total proteins) modules. (PDF 462Â kb)

Figure S2. Protein Quantitation in TMT-LysC and LFQ-trypsin Analyses. The relationship between the number of quantifiable proteins at a given threshold of missing values in the 47 brain samples from the BLSA cohort for TMT-LysC and LFQ-trypsin analyses is shown. The point at 23 samples and 6533 proteins represents the threshold used for the TMT-LysC analysis pipeline in this study. This point falls slightly below the TMT curve because 11 MCI samples were included in the TMT analysis workflow, for a total of 58 samples, but were later dropped from the analysis (see Methods). The increased number of samples when including the 11 MCI cases slightly reduced the number of quantifiable proteins at the ~ 50% missing value threshold. (PDF 79 kb)

Table S3. TMT Experimental Design. MCI cases (n = 11) were removed after batch correction. The final cohort used for quantification was n = 47 DLPFC non-MCI samples. TMT, tandem mass tag; GIS, global internal standard; BLSA, Baltimore Longitudinal Study of Aging; AD, Alzheimer’s disease; AsymAD, asymptomatic Alzheimer’s disease; MCI, mild cognitive impairment; DLPFC, dorsolateral prefrontal cortex. (PDF 888 kb)

Table S4. List of Biological Terms for GO Network in Figure S8. GO, gene ontology; UP, UniProt; KEGG, Kyoto Encyclopedia of Genes and Genomes; SMART, Simple Modular Architecture Research Tool; FDR, false discovery rate. (DOCX 35Â kb)

Table S2. Case Characteristics. Values shown are means ± SD. AD, Alzheimer’s disease; AsymAD, asymptomatic Alzheimer’s disease; MCI, mild cognitive impairment; CERAD, Consortium to Establish a Registry for Alzheimer’s Disease amyloid-β plaque load score; Braak, Braak stage for tau tangle burden; PMI, post-mortem interval; ApoE, apolipoprotein E isoform genotype. (DOCX 30 kb)

Table S6. Number of Alternative Exon-Exon Junctions in AD Risk Factor Proteins. From the twenty proteins identified as risk factors for AD by GWAS at genome-wide significance [12], five had alt-EEjxn peptides that were observed and quantifiable in the BLSA-TMT analysis (observed). The number of observed and quantifiable alt-EEjxn peptides for each of these five proteins was a subset of the total number of alt-EEjxn peptides predicted to exist after LysC digestion (peptide database). This number was a further subset of the total number of alt-EEjxns observed for each of the five proteins from RNAseq data (transcript level). For details on generation of the peptide database and transcript level numbers, see Methods. (DOCX 28Â kb)

Table S5. Number of Alternative Exon-Exon Junction Peptides Identified by TMT-LysC and LFQ-trypsin Approaches. The number of alt-EEjxn peptides identified by matching to the listed databases (Swiss-Prot, Trembl, or RNAseq) is shown, along with the number of quantifiable alt-EEjxn peptides. A peptide was considered quantifiable in this analysis if it had a minimum of 2 measurements in at least 2 different case groups. RNAseq data from control and AD patient brains (n = 6) were used to generate the RNAseq alt-EEjxn peptide database, as described in Methods. (DOCX 28 kb)

Here, we report proteomic analyses of 129 human cortical tissues to define changes associated with the asymptomatic and symptomatic stages of Alzheimer's disease (AD). Network analysis revealed 16 modules of co-expressed proteins, 10 of which correlated with AD phenotypes. A subset of modulesoverlapped with RNA co-expression networks, including those associated with neurons and astroglial cell types, showing altered expression in AD, even in the asymptomatic stages. Overlap of RNA and protein networks was otherwise modest, with many modules specific to the proteome, including those linked to microtubule function and inflammation. Proteomic modules were validated in an independent cohort, demonstrating some module expression changes unique to AD and several observed in other neurodegenerative diseases. AD genetic risk loci were concentrated in glial-related modules in the proteome and transcriptome, consistent with their causal role in AD. This multi-network analysis reveals protein- and disease-specific pathways involved in the etiology, initiation, and progression of AD.