Large-scale deep multi-layer analysis of Alzheimer's disease brain reveals strong proteomic disease-related changes not observed at the RNA level.
Johnson ECB, Carter EK, Dammer EB, Duong DM, Gerasimov ES, Liu Y, Liu J, Betarbet R, Ping L, Yin L, Serrano GE, Beach TG, Peng J, De Jager PL, Haroutunian V, Zhang B, Gaiteri C, Bennett DA, Gearing M, Wingo TS, Wingo AP, Lah JJ, Levey AI, Seyfried NT. · 2022
Read Large-scale deep multi-layer analysis of Alzheimer's disease brain reveals strong proteomic disease-related changes not observed at the RNA level. by Johnson ECB, Carter EK, Dammer EB, Duong DM, Gerasimov ES, Liu Y, Liu J, Betarbet R, Ping L, Yin L, Serrano GE, Beach TG, Peng J, De Jager PL, Haroutunian V, Zhang B, Gaiteri C, Bennett DA, Gearing M, Wingo TS, Wingo AP, Lah JJ, Levey AI, Seyfried NT. (2022) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc8825285.txt.
Introduction
Johnson and co-authors map protein changes in the Alzheimer's disease brain at large scale. They analyzed proteins from more than 1,000 brain tissues. The goal was to find groups of proteins that rise and fall together. Such groups can reveal which processes break down in the disease. The team compared protein groups with RNA groups from the same cohorts.
Nearly half of the protein groups had no match at the RNA level. Two disease-linked groups stood out as protein-only finds. One tied to MAPK signaling and metabolism. The other tied to the matrisome, the tissue scaffold. The MAPK group tracked closely with the pace of mental decline. The authors offer both groups as leads for drugs and biomarkers. The study argues that Alzheimer's is, at heart, a protein disease.
Catalog record
| CREATOR | Johnson ECB, Carter EK, Dammer EB, Duong DM, Gerasimov ES, Liu Y, Liu J, Betarbet R, Ping L, Yin L, Serrano GE, Beach TG, Peng J, De Jager PL, Haroutunian V, Zhang B, Gaiteri C, Bennett DA, Gearing M, Wingo TS, Wingo AP, Lah JJ, Levey AI, Seyfried NT. |
|---|---|
| YEAR | 2022 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 100 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Large-Scale Deep Multi-Layer Analysis of Alzheimer’s Disease Brain Reveals Strong Proteomic Disease-Related Changes Not Observed at the RNA Level By Erik C.B. Johnson, E. Kathleen Carter, Eric B. Dammer, Duc M. Duong, Ekaterina S. Gerasimov, Yue Liu, Jiaqi Liu, Ranjita Betarbet, Lingyan Ping, Luming Yin, Geidy E. Serrano, Thomas G. Beach Published in: Nature neuroscience DOI: 10.1038/s41593-021-00999-y ABSTRACT The biological processes that are disrupted in the Alzheimer’s disease (AD) brain remain incompletely understood. Here, we analyzed the proteomes of >1000 brain tissues to reveal new AD-related protein co-expression modules that were highly preserved across cohorts and brain regions. 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. 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. INTRODUCTION Alzheimer’s disease (AD) remains an immense personal and public health burden without effective disease-modifying therapies. As part of a national effort to develop therapeutics and biomarkers for AD, the Accelerated Medicines Partnership—Alzheimer’s Disease (AMP-AD) Consortium has been leveraging unbiased molecular profiling data at the genomic, transcriptomic, proteomic, and metabolomic levels to further our understanding of AD pathogenesis. Genetics has significantly advanced our understanding of AD heritable risk, yet how genetic risk factors affect biological pathways that influence AD pathophysiology is not always clear1. Understanding the biological effects of AD risk factor polymorphisms in human brain often requires additional levels of analysis using other –omics approaches. To this end, transcriptomics has been widely used to measure mRNA transcripts in AD brain, and the resulting transcriptomic data have been integrated with AD genetic risk2. However, the ultimate biological effectors of AD genetic and environmental risk are often the proteins and the metabolic pathways they modulate. Compared to genomics and transcriptomics, proteomics approaches have to date provided comparatively less in-depth coverage of the target analyte due to increased complexity and technical demands of analyzing amino acid polymers versus nucleic acid polymers. In this study, we used a tandem mass tag mass spectrometry (TMT-MS) approach3–6 and the AMP-AD consortium of postmortem brain tissues to generate a deep TMT AD protein network that significantly expanded our previous label-free quantitation mass spectrometry (LFQ-MS) network7 and revealed new AD-related protein co-expression modules. We leveraged brain tissues from cohorts that also have been profiled using other –omics modalities, including genomics and transcriptomics, to perform a multi-layer genomic, transcriptomic, and proteomic analysis of the TMT AD protein network in order to better understand the relationships among these different data types in the context of AD. RESULTS TMT Consensus AD Protein Co-expression Network For this study, we analyzed a total of 516 dorsolateral prefrontal cortex (DLPFC) tissues from control, asymptomatic AD (AsymAD), and AD brains from the Religious Orders Study and Memory and Aging Project (ROSMAP, n=84 control, 148 AsymAD, 108 AD)8–10 and Banner Sun Health Research Institute (Banner, n=26 control, 58 AsymAD, 92 AD)11 by tandem mass tag mass spectrometry (TMT-MS)-based quantitative proteomics (Figure 1A, Supplementary Table 1). Cases were defined based on a unified classification scheme using semi-quantitative histopathological measures of Aβ and tau neurofibrillary tangle deposition12–15, as well as cognitive function near time of death, as previously described7. AsymAD cases were those with neuropathological burden of Aβ plaques and tau tangles similar to AD cases, but without significant cognitive impairment near time of death, which is considered to be an early preclinical stage of AD16. After data processing and outlier removal, a total of 8619 proteins were used to build a protein co-expression network using the weighted co-expression network (WGCNA) algorithm17 (Figure 1B, Extended Data Figure 1, Supplementary Tables 2-4). This network consisted of 44 modules, or communities of proteins related to one another by their co-expression across control and disease tissues. Compared to our previous AD consensus network constructed using label-free quantitative (LFQ) proteomic data, the TMT consensus network contained over five times as many proteins that could be assigned to a module (6337 versus 1205), as well as a larger fraction of quantified proteins that could be assigned to a module (73% versus 36%), highlighting the improved depth and coherence of the TMT data compared to the LFQ consensus data. Of the 13 modules previously identified in the LFQ consensus network7, every module except the smallest module (module 13 consisting of 20 proteins) was preserved in the TMT network (Extended Data Figure 2A), also highlighting the consistency of the LFQ and TMT proteomic data. Because different network clustering algorithms can produce disparate networks, we tested the robustness of the TMT consensus network generated by the WGCNA algorithm by also generating a co-expression network using an independent algorithm—the MONET M1 algorithm. MONET M1 was identified as one of the top performers in the Disease Module Identification DREAM challenge, and is based on a modularity optimization algorithm rather than the hierarchical clustering approach used in WGCNA18, 19. We found that all 44 WGCNA modules were highly preserved in the MONET M1 network (Extended Data Figure 2B), demonstrating the robustness of the TMT consensus network to clustering algorithm. The biology represented by each TMT consensus network module was determined using gene ontology (GO) analysis of its constituent proteins (Figure 1B, Supplementary Data). Most modules could be assigned a primary ontology, and those that were ambiguous in their ontology were left unannotated or assigned as “ambiguous.” Module 42 was assigned the term “matrisome,” which refers to the collection of extracellular matrix-associated proteins20, 21, due to its strong enrichment in extracellular matrix and glycosaminoglycan binding proteins. To assess whether a module was related to features of AD, we correlated each module eigenprotein, or the first principal component of module protein expression, to neuropathologic or cognitive traits present in the ROSMAP and Banner cohorts (Figure 1B, Supplementary Table 5; individual protein trait correlations are provided in Supplementary Table 6). We also assessed the cell type nature of each module by determining whether it was enriched in cell type-specific protein markers (Figure 1B, Supplementary Tables 7 and 8). Because the network was highly powered, with the ability to observe a significant correlation of 0.1 at p=0.05 for most pathological and cognitive traits, we were able to observe a large fraction of the 44 modules that significantly correlated with at least one pathological or cognitive trait. Twelve modules or module families were noted to correlate more strongly to AD traits than the others. These included post-synaptic density, glycosylation/endoplasmic reticulum (ER), oligodendrocyte/myelination, RNA splicing, matrisome, cell-extracellular matrix (ECM) interaction, synapse/neuron, ubiquitination, MAPK signaling and metabolism, mitochondria, sugar metabolism, and protein transport modules (Figure 1B). Four of these modules—M5 post-synaptic density (global pathology r=–0.32, p=2.7e−9; global cognitive function r=0.35, p=7.4e−11), M7 MAPK signaling and metabolism (global pathology r=0.37, p=4.9e−12; global cognitive function r=–0.42, p=1.2e−15), M11 cell-ECM interaction (global pathology r=0.34, p=4.1e−10; global cognitive function r=–0.33,