What are the first changes in the brain as Alzheimer’s pathology takes hold? According to a postmortem proteomics study published in Alzheimer’s and Dementia on April 19, it is proteins involved in vesicular functions, including endocytosis, secretion, and synaptic vesicle transport. Researchers led by Dietmar Thal at KU Leuven in Belgium identified 53 proteins whose abundance changed at different stages of AD, including seven proteins that had not been tied to AD before. The scientists reported that, as Aβ plaques and tau tangles overtake the brain and cognitive symptoms surface, other biological pathways show signs of change, and some proteins become harder to solubilize. In all, the findings implicate endosomal and synaptic malfunctions as foundational to the disease process.

  • Proteins changed in abundance in response to amyloid plaques.
  • Endocytosis and synaptic vesicle function is hit in preclinical stage.
  • Some proteins become harder to solubilize as disease progresses.

“This proteomic analysis provides one more piece of the puzzle that indicates that endosomal/secretory defects are an essential mechanism early in AD,” commented Claudia Almeida of the University of Lisbon, Portugal. “Moreover, it supports that synaptic dysfunction, namely of the synaptic vesicle cycle, is an early target in AD.”

Numerous proteomic studies have graced the field in recent years, with some taking stock of proteins in thousands of postmortem brain samples to peg proteomic signatures of AD (Jul 2017 conference news; Mendonça et al., 2019; May 2020 news). Thal and colleagues opted to perform a smaller, more in-depth study with strict standards for distinguishing controls from people in different stages of AD. They also analyzed proteins based on their solubility using an ad hoc extraction method.

Stage Fractions. Neocortical homogenates prepared from postmortem brain samples of controls, people with preclinical AD (p-pre-AD), and people with symptomatic AD (left) were fractionated based on solubility (middle). Proteomic analysis of each fraction compared abundance and solubility of proteins between disease groups (right). [Courtesy of Li et al., Alzheimer’s & Dementia, 2021.]

First author Xiaohang Li and colleagues tracked not only the abundance of proteins, but how their solubility shifted from one disease stage to the next. They analyzed postmortem neocortical samples from, coincidentally, 53 people, including 18 non-AD controls who had died without any signs of AD pathology in the brain; 18 people who had been cognitively normal but harbored Aβ pathology; and 17 people who had had symptomatic AD. They also sampled the occipital cortex, a region that accumulates amyloid plaques early in disease, but stays free of tau tangles or neurodegeneration until the latest stages. For two non-AD cases, only frontal cortex samples were available, but the researchers later established that these frontal proteomes did not differ markedly from occipital proteomes of other controls.

Using a fractionation method they had previously developed, the researchers split each brain sample into four: a soluble fraction; a dispersible fraction containing insoluble protein oligomers and small vesicles such as synaptosomes and endosomes; a detergent-soluble fraction containing larger organelles and membrane–associated proteins; and a formic acid fraction containing highly insoluble, aggregated proteins such as those associated with plaques (Upadhaya et al., 2014). Then they measured the proteins in each fraction via mass spectrometry.

All told, the researchers detected 1,449 unique proteins across all fractions and samples. Of those, 53 differed in abundance between control, preclinical, and symptomatic AD groups. Levels of four proteins differed between controls and preclinical AD; the scientists called them “early responding proteins.” Another 26, called “late responding proteins,” changed between the preclinical and symptomatic stages, while another 20 proteins changed gradually in abundance from control through symptomatic stages.

Five proteins shifted fractions—moving from the dispersible fraction in controls or preclinical samples to the detergent fraction in symptomatic AD. The researchers reported that levels of early responding proteins correlated strongly with the presence of Aβ plaques. By contrast, the gradually changing and late-responding proteins correlated more tightly with measures of disease progression, including Aβ load, Braak neurofibrillary tangle stage, and neuritic plaque score.

In the earliest stage of disease, when Aβ plaques were present but cognitive symptoms had not yet surfaced, levels of calmodulin-1, endophilin-A1, adaptor-related protein complex 1-b (AP1B1), and ribosomal protein L31 had changed relative to controls. Calmodulin-1 increased, but the other three—found within the vesicle-laden, dispersible fraction—all decreased in preclinical AD. Both endophilin-A1 and AP1B1 facilitate clathrin-mediated endocytosis. Endophilin-A is crucial for the endocytosis of synaptic vesicles and the maintenance of dendritic spines. L31 is part of the large ribosomal subunit.

Endocytosis, secretory pathways, and/or synaptic vesicles also featured prominently among the gradually changing and late-responding proteins. In addition, pathway analysis placed the gradually and late-responding proteins in caspase-mediated cleavage and MHC Class II antigen presentation pathways. Notably, APP/Aβ (mass spec does not distinguish between the two) ramped up in the formic acid fraction, while tau increased in the dispersible, i.e., oligomeric, fraction as disease progressed.

Two of the fraction-shifting proteins—plectin and UCH-L1—not only were less soluble in samples from more advanced disease, but were also less abundant in people with symptomatic AD.

Big Picture
Using a protein-protein interaction network database called STRING, the researchers next asked how the 53 AD-related proteins fit into a wider picture of protein function. The analysis identified two functional networks—APP metabolism and endocytosis. When the researchers wove in AD risk genes identified in genome-wide association studies, three functional networks emerged: APP/Aβ, tau-related, and synaptic vesicle/endocytosis/secretory pathways.  

In all, the findings suggest that endocytic and synaptic vesicle functions become impaired in the earliest stage of AD, at a time when a person has Aβ plaques but tau tangles have not yet spread extensively in his or her cortex. This suggests that these early neocortical changes are caused by Aβ-related processes, rather than by tau accumulation. This is especially true in the sampled region—the occipital lobe—which is inundated by Aβ pathology early in disease but does not bear tau tangles until the latest stages. An alternative explanation, Thal said, is that tau pathology in subcortical regions such as the nucleus basalis of Meynert, the raphe nuclei, and the locus coeruleus, which arises early in disease, may affect neurons in connected regions of the brain, including the occipital lobe.

Thal hesitated to speculate about the role of any one protein in AD, noting that proteomic studies are best suited for assessing the entire forest, rather than individual trees. Future studies will focus on how specific proteins and cellular processes influence disease progression in detail, he said.—Jessica Shugart


  1. The study, led by Dietmar Thal, is fascinating since it does an excellent pathological characterization of the postmortem tissue, focusing on the earliest changes and disease progression.

    I am pleased to see that proteins functioning in the secretory and endocytic pathway changed in the preclinical phase. This proteomic analysis provides one more piece of the puzzle that indicates that endosomal/secretory defects are an essential mechanism early in AD. Moreover, it supports that synaptic dysfunction, namely of the synaptic vesicle cycle, is an early target in AD.

    I was a bit surprised that only four proteins changed from control to pathological but not yet demented frontal cortex. I am also left wondering about the patient's genotypes. Maybe for futures studies, the presence of genes related to AD could be correlated with proteome changes.

  2. Lee et al. describe a creative approach to the proteomic analysis of AD brain tissue by tracking not only changing protein levels but also shifts in solubility characteristics of individual proteins and their compartmentalization in cells. Although the brain sample is cell heterogeneous, there is reason to believe that some of these shifts may be taking place in multiple cell types given the conserved functions of vesicular trafficking processes implicated by the authors. The endocytic pathway highlighted by the authors is one such pathway, which is strongly linked to genetic risk for AD.

    Another particularly interesting group of differentially expressed proteins (DEPs) not highlighted in the report is related to autophagy. Protein changes implying autophagy induction and autophagosome biogenesis include RAB1a and CLUS elevations and decreasing UCHL1 and PLEC levels. Even some of the proteins characterized as a “secretory pathway” cluster encompass Golgi functions that are relevant to lysosome functioning in the autophagy pathway (Lie et al., 2021). The autophagy response suggested here is consistent with our evidence in single population analyses of CA1 hippocampal neurons that autophagy is progressively induced starting at early AD stages, even though flux through the pathway ultimately fails as lysosomal function declines (Bordi et al., 2016).

    It is interesting to speculate that increased levels of certain DEPs in the “dispersible” brain fraction, which comprises vesicular compartments of the endosomal-lysosomal-autophagy network, could reflect the upregulated sequestration of substrates into autophagic vacuoles as well as into endocytic compartments, which swell due to upregulated endocytosis in AD.

    Curiously, the rises in some abundant cytoskeleton-related proteins that are autophagy substrates (e.g. NFH, tau, vimentin) along with others in the “dispersible” vesicle and fraction-shifting groups may reflect their gradual accumulation as substrates due to the progressive declines of lysosomal efficiency and autophagy flux as AD advances (Bordi et al., 2016). 

    Future studies involving detection of a larger population of brain proteins and higher numbers of DEPs will yield improved definition of the disease-related biological pathways in order to confirm the intriguing leads reported here.


    . Post-Golgi carriers, not lysosomes, confer lysosomal properties to pre-degradative organelles in normal and dystrophic axons. Cell Rep. 2021 Apr 27;35(4):109034. PubMed.

    . Autophagy flux in CA1 neurons of Alzheimer hippocampus: Increased induction overburdens failing lysosomes to propel neuritic dystrophy. Autophagy. 2016 Dec;12(12):2467-2483. Epub 2016 Nov 4 PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Alzheimer’s Proteomics Treasure Trove?
  2. Massive Proteomics Studies Peg Glial Metabolism, Myelination, to AD

Paper Citations

  1. . Proteomic signatures of brain regions affected by tau pathology in early and late stages of Alzheimer's disease. Neurobiol Dis. 2019 Oct;130:104509. Epub 2019 Jun 15 PubMed.
  2. . Biochemical stages of amyloid-β peptide aggregation and accumulation in the human brain and their association with symptomatic and pathologically preclinical Alzheimer's disease. Brain. 2014 Mar;137(Pt 3):887-903. Epub 2014 Feb 10 PubMed.

Further Reading


  1. . Analysis of a membrane-enriched proteome from postmortem human brain tissue in Alzheimer's disease. Proteomics Clin Appl. 2012 Apr;6(3-4):201-11. PubMed.
  2. . Changes in the detergent-insoluble brain proteome linked to amyloid and tau in Alzheimer's Disease progression. Proteomics. 2016 Dec;16(23):3042-3053. PubMed.

Primary Papers

  1. . Sequence of proteome profiles in preclinical and symptomatic Alzheimer's disease. Alzheimers Dement. 2021 Apr 19; PubMed.