TY - JOUR
AU - Aarsland, Dag
AB - Abstract See Attems and Jellinger (doi:10.1093/brain/awx360) for a scientific commentary on this article. Cognitive changes occurring throughout the pathogenesis of neurodegenerative diseases are directly linked to synaptic loss. We used in-depth proteomics to compare 32 post-mortem human brains in the prefrontal cortex of prospectively followed patients with Alzheimer’s disease, Parkinson’s disease with dementia, dementia with Lewy bodies and older adults without dementia. In total, we identified 10 325 proteins, 851 of which were synaptic proteins. Levels of 25 synaptic proteins were significantly altered in the various dementia groups. Significant loss of SNAP47, GAP43, SYBU (syntabulin), LRFN2, SV2C, SYT2 (synaptotagmin 2), GRIA3 and GRIA4 were further validated on a larger cohort comprised of 92 brain samples using ELISA or western blot. Cognitive impairment before death and rate of cognitive decline significantly correlated with loss of SNAP47, SYBU, LRFN2, SV2C and GRIA3 proteins. Besides differentiating Parkinson’s disease dementia, dementia with Lewy bodies, and Alzheimer’s disease from controls with high sensitivity and specificity, synaptic proteins also reliably discriminated Parkinson’s disease dementia from Alzheimer’s disease patients. Our results suggest that these particular synaptic proteins have an important predictive and discriminative molecular fingerprint in neurodegenerative diseases and could be a potential target for early disease intervention. synaptic proteins, cognitive impairment, Lewy body dementias, Alzheimer’s disease, mass spectrometry Introduction The pandemic increase of dementia, hampering daily living of many millions, carries serious implications for society (Wimo et al., 2017). Alzheimer’s disease and the Lewy body dementias, i.e. dementia with Lewy bodies and Parkinson’s disease dementia, are the most common forms of neurodegenerative dementias (Campbell et al., 2001; McKeith et al., 2005). Cognition gradually declines in Alzheimer’s disease, dementia with Lewy bodies and Parkinson’s disease dementia leading to loss of function in everyday life, reduced quality of life, and increased mortality (Aarsland et al., 2003; McKeith et al., 2005; Maalouf et al., 2011). Synapse and synaptic protein loss seems to be a universal element in the pathologic changes associated with dementia (DeKosky and Scheff, 1990) as it is directly linked to cognitive deficits from early stages of dementia and it is believed that synaptic changes precede neuronal degeneration (DeKosky and Scheff, 1990). It has been shown that synaptic loss is a better correlate of cognitive impairment in Alzheimer’s disease than the hallmark tau and amyloid-β pathologies (Blennow et al., 1996; Masliah et al., 2001). Several studies have shown that changes in synaptic function are associated with alterations in the concentration of synaptic proteins (Gottschall et al., 2010), a characteristic feature in Alzheimer’s disease (Terry et al., 1991; Honer, 2003) and increasing attention is now being devoted to their role in synucleinopathies (Aarsland et al., 2005; Compta et al., 2011; Howlett et al., 2015; Bereczki et al., 2016). In a recent study, we reported changes in the concentration of presynaptic proteins SNAP25 and Rab3A as well as a postsynaptic protein, neurogranin, in post-mortem neocortical regions in Parkinson’s disease, dementia with Lewy bodies and Alzheimer’s disease patients. These changes correlated with the rate of cognitive decline in dementia with Lewy bodies and Alzheimer’s disease as well as with neuropathological markers (Bereczki et al., 2016). The development of biomarkers aiding early differential diagnosis and predicting disease progression from its earliest stage is of major importance both for research and therapeutic development. The complex structural and functional organization of the brain regarding its morphology, connectivity and function warrants the application of systematic approaches. Recent advances in mass spectrometry-based proteomics offer a reliable molecular phenotype comparison between diseased and control cases allowing in-depth coverage of quantitative changes (Kim et al., 2014). These methods permit the identification of alterations in the cellular proteome and provide insight into disease aetiology and mechanisms. In addition, they aid the discovery of biomarkers for monitoring disease progression as well as assessment of drug effects (Portelius et al., 2015; Moya-Alvarado et al., 2016). Whereas some explorative proteomic studies have already been performed in Alzheimer’s disease and Parkinson’s disease (Abdi et al., 2006; Blennow and Zetterberg, 2013; Brinkmalm et al., 2014; Halbgebauer et al., 2016), only few studies have been conducted in dementia with Lewy bodies (Abdi et al., 2006; Barthelemy et al., 2016; Biemans et al., 2016). Our study is among the first in-depth quantitative proteome studies on prefrontal post-mortem tissues where, beside the whole proteome comparison, we also aimed to profile the entire synaptic proteome of patients with Alzheimer’s disease, Parkinson’s disease dementia and dementia with Lewy bodies and compared them to non-demented control cases. Our in-depth analysis of the synaptic proteome identified key synaptic proteins underlying synaptic dysfunction in Alzheimer’s disease, Parkinson’s disease dementia and dementia with Lewy bodies suggesting shared mechanisms, with major implications for prognostic and diagnostic marker development as well as advancing future therapeutic interventions for improving the disease course. Materials and methods Brain tissue Post-mortem human brain tissues from prefrontal cortex, Brodmann area 9 (from 92 cases in total) were provided by the Brains for Dementia Research network. The prefrontal cortex was selected due to its role in cognition and executive functions involved across the three diseases (Fuster, 2001). The cohort included cases from the Newcastle Brain Tissue Resource (three cases), the Thomas Willis Oxford Brain Collections (seven cases), the London Neurodegenerative Diseases Brain Bank (56 cases) and the University Hospital Stavanger (26 cases). Autopsy protocols and sample collection was harmonized between centres. Detailed description of the diagnostic criteria has been previously published (Howlett et al., 2015). Final diagnoses for patients are clinic-pathological consensus diagnoses. In total, 24 Parkinson’s disease dementia patients (age 72–89 years), 26 dementia with Lewy bodies patients (age 65–91 years), 18 patients with Alzheimer’s disease (age 72–103 years) and 24 elderly non-neurological controls (age 65–96 years) were included. Controls did not have significant neurological or psychiatric diseases and presented only mild age-associated neuropathological changes (e.g. neurofibrillary tangle Braak stage II). Semi-quantitative assessments of senile amyloid-β plaques, phosphor-tau and α-synuclein pathology were conducted by experienced neuropathologists blind to clinical diagnosis, using a four-tiered scale of 0 (none), 1 (sparse), 2 (moderate) and 3 (severe/frequent) to score sections from each brain area, as described previously (Howlett et al., 2015). Hoehn and Yahr scale was available for 23 of 24 Parkinson’s disease with dementia patients, and assessment from the last OFF phase was used. Alzheimer’s disease patients with low α-synuclein pathology were chosen to ensure distinction between Alzheimer’s disease and dementia with Lewy bodies patients. Lewy body dementia cases selected were of pathologically ‘diffuse neocortical’ stage, with a cortical Lewy body score of 13.2 (±3.6), incorporating the 1-year rule to differentiate between dementia with Lewy bodies and Parkinson’s disease with dementia (McKeith et al., 2005). Neuropathological assessment was performed according to standardized neuropathological scoring/grading systems; assessment and diagnostic criteria have been previously described (Howlett et al., 2015). Cognitive data were available for most patients and consisted of the last Mini-Mental State Examination (MMSE) scores, assessed in most cases within 1–2 years before death (Folstein et al., 1975) and MMSE decline calculated as average decline over a period of clinical observation of 8–10 years. All participants gave informed consent for their tissue to be used in research and the study was approved by the UK National Research Ethics Service (08/H1010/4), the Norwegian committee for medical and health research ethics (2010/633) and the Regional Ethical Review Board of Stockholm (2012/920-31/4). Sample preparation for HiRIEF LC-MS proteomics The tissues were lysed in SDS-lysis buffer [4% (w/v) SDS, 25 mM HEPES pH 7.6, 1 mM DTT]. Lysates were then heated at 95°C for 5 min in a thermomixer, and were sonicated with a sonicator probe to shear DNA. Samples were centrifuged at 14 000g to remove cell debris, the supernatant was collected and protein concentration estimated by the DC-protein assay (BioRad). From each sample, 250 µg of total protein were taken and processed according to the FASP (Filter Aided Sample Preparation) protocol (Wisniewski et al., 2009) with one modification, i.e. the samples were digested on the filter with Lys-C for 3 h prior to trypsin digestion (16 h). Peptide concentration was estimated by the DC-protein assay (Bio-Rad), and 100 µg of peptides from each sample were labelled with the respective TMT10plex™ reagent (Thermo Fisher Scientific) according to the manufacturer’s instructions. HiRIEF separation Peptide pre-fractionation was done using HiRIEF (high resolution isoelectric focusing) (Branca et al., 2014). Briefly, after pooling the samples that belong together in each TMT™ (Tandem Mass Tag™) set, each TMT set was cleaned by strong cation exchange solid phase extraction (SCX-SPE, Phenomenex Strata-X-C, P/N 8B-S029-TAK). After drying in a SpeedVac™ (Thermo SPD111V with refrigerated vapor trap RVT400), the equivalent to 400 µg of peptides of each sample were dissolved in 250 µl of 8 M urea, 1% pharmalyte (broad range pH 3–10, GE Healthcare, P/N 17-0456-01), and this solution was used to rehydrate the IPG drystrip (pH 3–10, 24 cm, GE Healthcare, P/N 17-6002-44) overnight. Focusing was done on an Ettan IPGphor 3 system (GE Healthcare), ramping up the voltage to 500 V in 1 h, then to 2000 V in two more hours, and finally to 8000 V in six more hours, after which voltage was held at 8000 V for additional 20 h or until 150 kVh were reached. After focusing was complete, a well-former with 72 wells was applied onto each strip, and liquid-handling robotics (GE Healthcare prototype modified from a Gilson liquid handler 215), using three rounds of different solvents: (i) milliQ water; (ii) 35% acetonitrile; and (iii) 35% acetonitrile, 0.1% formic acid, added 50 µl of solvent to each well, waited 30 min incubating, and finally transferred the 72 fractions into a microtitre plate (96 wells, polypropylene, V-bottom, Greiner P/N 651201), which was then dried in a SpeedVac. LC-MS analysis For each liquid chromatography–mass spectrometry (LC-MS) run of a HiRIEF fraction, the autosampler (Ultimate 3000 RSLC nanoUPLC system, Thermo Scientific Dionex) dispensed 15 µl of mobile phase A [95% water, 5% dimethylsulphoxide (DMSO), 0.1% formic acid] into the corresponding well of the microtitre plate, mixed by aspirating/dispensing 10 µl 10 times, and finally injected 7 µl into a C18 guard desalting column (Acclaim pepmap 100, 75 µm × 2 cm, nanoViper, Thermo). After 5 min of flow at 5 µl/min with the loading pump, the 10-port valve switched to analysis mode in which the NC pump provided a flow of 250 nl/min through the guard column. The curved gradient (curve 6 in the Chromeleon software) then proceeded from 3% mobile phase B (90% acetonitrile, 5% DMSO, 5% water, 0.1% formic acid) to 45% B in 50 min followed by wash at 99% B and re-equilibration. Total LC-MS run time was 74 min. We used a nano EASY-Spray™ column (pepmap RSLC, C18, 2 µm bead size, 100 Å, 75 µm internal diameter, 50 cm long, Thermo) on the nano-electrospray ionization (NSI) EASY-Spray source (Thermo) at 60°C. Online LC-MS was performed using a hybrid Q-Exactive mass spectrometer (Thermo Scientific). FTMS master scans with 70 000 resolution (and mass range 300–1600 m/z) were followed by data-dependent MS/MS (35 000 resolution) on the top five ions using higher energy collision dissociation (HCD) at 30% normalized collision energy. Precursors were isolated with a 2 m/z window. Automatic gain control (AGC) targets were 1 × 106 for MS1 and 1 × 105 for MS2. Maximum injection times were 100 ms for MS1 and 150 ms for MS2. The entire duty cycle lasted ∼1.5 s. Dynamic exclusion was used with 60 s duration. Precursors with unassigned charge state or charge state 1 were excluded. An underfill ratio of 1% was used. Proteomics database search All MS/MS spectra were searched by MSGF+/Percolator using a target-decoy strategy. Raw MS/MS files were converted to mzML format using msconvert from the ProteoWizard tool suite (Kessner et al., 2008). Spectra were then searched using MSGF+ (Kim and Pevzner, 2014) (v10072) and Percolator (Kall et al., 2007) (v2.08), where eight subsequent search results were grouped for Percolator target/decoy analysis. The reference database that was used was the human subset of the Swiss-Prot database (version 2015_08, with 42 122 canonical and isoform protein entries, downloaded from uniprot.org). MSGF+ settings included precursor mass tolerance of 10 ppm, fully-tryptic peptides, maximum peptide length of 50 amino acids and a maximum charge of 6. Fixed modifications were TMT10plex on lysine residues and N -termini, and carbamidomethylation on cysteine residues; a variable modification was used for oxidation on methionine residues. Peptide and peptide-spectrum match (PSM) false discovery rates (FDRs) were recalculated after merging the percolator groups of eight search results into one result per TMT set. Quantification of TMT10plex reporter ions was done using OpenMS project’s IsobaricAnalyzer (Rost et al., 2016) (v2.0). PSMs found at 1% PSM- and peptide-level FDR were used to infer gene identities, whose respective protein products were quantified using the medians of PSM quantification ratios, which were subsequently normalized to the median protein value of each TMT channel ratio. Only one unique peptide was required to identify a protein, but a protein level FDR cut-off of 1% (calculated using the picked-FDR method) (Savitski et al., 2015) was applied to the list of gene-centred proteins. Thus, all PSMs, peptides and proteins included in the final results were filtered through both a 1% peptide level FDR and a 1% protein level FDR. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD006122. Gene ontology and pathway enrichment analyses Gene ontology (GO) terms were retrieved from uniprot.org for all proteins identified. Proteins with GO terms (in all ontologies: biological processes, molecular function, cellular component) containing the word ‘synapse’ or ‘synaptic’ were considered synaptic proteins and used for further enrichment analysis. T-tests comparing the sample groups using log2-transformed ratios were used to determine whether proteins were differentially accumulated (requirements: P < 0.05 and fold change <0.83 or >1.20, which is based on the 95% confidence interval of the variance between the two replicate internal pooled standard TMT channels). The proteins deemed significant were then assigned to Kyoto Encyclopaedia of Genes and Genomes (KEGG) pathway enrichment analysis using DAVID Bioinformatics Resources version 6.8 (Huang da et al., 2009). An EASE score (a modified Fisher’s exact test) <0.1 and P-value <0.05 were the criteria for significantly enriched biological pathways. The GOrilla (gene ontology enrichment analysis and visualization) tool (Eden et al., 2009) was used for detailed data analyses with two unranked lists of genes, a target list and a background list (composed by all genes identified at protein level in the MS experiment), with a GO database from 2017-01-21. Results were categorized into the functional groups of cellular component, biological process or molecular function. FDR q-values of <0.5, representing the correction of P-values for multiple testing, were considered significant. Preparation of tissue samples for western blotting and ELISA Preparation of tissue for western blotting and ELISA analyses was performed as previously described (Kirvell et al., 2006). Briefly, 500 mg of frozen tissue was homogenized in ice-cold buffer containing 50 mM Tris-HCl, 5 mM EGTA, 10 mM EDTA, protease inhibitor cocktail tablets (Roche, one tablet per 50 ml of buffer), and 2 mg/ml pepstatin A dissolved in ethanol:DMSO 2:1 (Sigma). The buffer was used at a ratio of 2 ml to every 100 mg of tissue, and homogenization was performed using an IKA Ultra-Turrax mechanical probe (IKA Werke) until the liquid appeared homogenous. Protein concentration of each sample was measured by using BCA Protein Assay Kit (Pierce, Thermo Fisher Scientific). Samples for ELISA measurements were further diluted to 0.5 μg/μl total protein in PBS (phosphate-buffered saline). Parts of Figs 1 and 4 were crafted using the Mind the Graph platform. Figure 1 View largeDownload slide Proteomic data analyses. Thirty-two post-mortem brain samples underwent proteome profile comparison. (A) Samples were labelled at peptide level with four sets of isobaric tags (TMT10plex), each containing eight channels with randomized samples and two channels with the internal reference sample (Ref), followed by fractionation into 72 fractions by HiRIEF with the broad range IPG 3-10 strip prior to LC-MS analysis. (B) Schematic representation of the number of differentially regulated proteins across disease groups. Differentially regulated proteins in dementia with Lewy bodies were further analysed for KEGG pathways (C) and gene ontology (GO) terms (D). From the significantly altered synaptic proteins GRIA3, SNAP47, LRFN2, SYBU, SYT2, GAP43, GRIA4 and SV2C were chosen for further validation with ELISA or western blot analyses in a larger cohort (E) (two synaptic proteins neurogranin, and CAMK2 were previously found to be altered by us within the same cohort). AD = Alzheimer’s disease; C = non-demented controls; DLB = dementia with Lewy bodies; LC-MS = liquid chromatography-mass spectrometry; PDD = Parkinson’s disease with dementia. Figure 1 View largeDownload slide Proteomic data analyses. Thirty-two post-mortem brain samples underwent proteome profile comparison. (A) Samples were labelled at peptide level with four sets of isobaric tags (TMT10plex), each containing eight channels with randomized samples and two channels with the internal reference sample (Ref), followed by fractionation into 72 fractions by HiRIEF with the broad range IPG 3-10 strip prior to LC-MS analysis. (B) Schematic representation of the number of differentially regulated proteins across disease groups. Differentially regulated proteins in dementia with Lewy bodies were further analysed for KEGG pathways (C) and gene ontology (GO) terms (D). From the significantly altered synaptic proteins GRIA3, SNAP47, LRFN2, SYBU, SYT2, GAP43, GRIA4 and SV2C were chosen for further validation with ELISA or western blot analyses in a larger cohort (E) (two synaptic proteins neurogranin, and CAMK2 were previously found to be altered by us within the same cohort). AD = Alzheimer’s disease; C = non-demented controls; DLB = dementia with Lewy bodies; LC-MS = liquid chromatography-mass spectrometry; PDD = Parkinson’s disease with dementia. Sandwich enzyme-linked immunosorbent assays Commercial sandwich ELISA kits for each of the selected synaptic proteins were purchased from Mybiosource. Assay procedures were followed according to the manufacturer’s protocol. Standard samples for SV2C (synaptic vesicle 2C) and SNAP47 (synaptosomal associated protein 47, 50 µl) and 100 µl for glutamate receptor (GRIA)3, GRIA4 and GAP43 (neuromodulin) were incubated with the corresponding HRP-conjugate reagent for 1 h at 37°C, followed by thorough washing steps. Chromogen solution was then applied and after stopping the reaction, absorbance measured immediately at 450 nm on a SpektraMax® Plus 384 microplate reader (Molecular Devices). The sigmoidal standard was evaluated with non-linear four-parameter fit using SoftMax® Pro 5.2 software and sample amounts were obtained using the fitted standard curve. Standards and samples were measured in duplicates. Samples of human brain were added in dilutions of 0.5 μg/μl of total protein and standards were diluted so that the sample absorbance values would fall near 50% binding (the linear range) of the standard curve. Concentrations were calculated after the mean blank value had been subtracted. Immunoblotting To minimize inter-blot variability, 20 μg of total protein per sample was loaded in each lane on 7.5–10% SDS-polyacrylamide gels (Criterion) for protein separation and then transferred to nitrocellulose membrane (Immobilon-P, Millipore). Each gel contained a control lane of pooled brain homogenates used as an internal standard. After blocking non-specific binding, membranes were incubated with primary antibodies (Supplementary Table 1) followed by HRP conjugated secondary antibody. GAPDH was used as a reference protein assessing equal loading. Bands were visualized using Chemiluminescent substrate (Millipore) in a LAS-3000 luminescent image reader (Fujifilm). Western blot data were evaluated and quantified using Multi Gauge Image Analyzer (version 3.0). Statistical analysis To compare synaptic protein levels between groups, Student’s t-tests were applied on log2-transformed data using SAM (Significance Analysis of Microarrays) under R (version 3.2.2, The R Foundation for Statistical Computing). SAM performs t-tests using permutation-based corrections for multiple comparisons. Although originally designed for array data, SAM has been shown to be valid also for LC-MS/MS data (Roxas and Li, 2008; Sandberg et al., 2012). Additional univariate analyses were carried out using non-parametric statistical tests due to the irregular non-Gaussian distribution of the samples in SPSS (IBM Statistics 22). To assess the relationship between synaptic proteins, and neuropathological and MMSE scores, Spearman correlations were performed. To compare protein levels between controls and the different patient groups we used Kruskal-Wallis tests, followed by Dunn’s post hoc test. In all cases, differences were considered statistically significant at P ≤ 0.05. Chi-squared tests (with Yates Continuity Correction) were used to explore differences in gender across diagnostic groups. A linear regression for the correlation studies was applied to synaptic proteins to regress out the effects of age. Prior to linear regression, logarithmic normalization was applied to synaptic proteins to achieve normality. Multivariate data analyses were performed to discriminate controls from the different patient groups using orthogonal partial least square analyses (OPLS in the program SIMCA, version 13.0; Umetrics AB, Sweden). Detailed description of the multivariate statistical analysis can be found in the Supplementary material. We calculated the sensitivity, specificity, positive predictive values and negative predictive values of the group separations from the Q2(Y) values obtained in each model. Results Demographic characteristics of the samples Key cohort characteristics are shown in Table 1. In the mass spectrometry studies, we included 32 (eight non-demented, eight Parkinson’s disease dementia, seven dementia with Lewy bodies, nine Alzheimer’s disease) patients, while in the validation studies 92 (24 non-demented, 24 Parkinson’s disease dementia, 26 dementia with Lewy bodies, 18 Alzheimer’s disease) patients were included. There were no significant differences in the pH (χ2 = 6.147, df = 3, P = 0.105) or in the post-mortem delay between the groups (χ2 = 5.037, df = 3, P = 0.169). In the larger cohort, Alzheimer’s disease patients were significantly older than all the other three groups (P = 0.001 versus controls; P = 0.019 versus Parkinson’s disease dementia; P = 0.005 versus dementia with Lewy bodies) while patients with Parkinson’s disease dementia and dementia with Lewy bodies did not differ significantly in age. Dementia with Lewy body and Alzheimer’s disease had longer dementia duration than Parkinson’s disease dementia (P = 0.006 Parkinson’s disease dementia versus dementia with Lewy bodies; P ≤ 0.001 Parkinson’s disease dementia versus Alzheimer’s disease; P = 0.002 versus Alzheimer’s disease). Correlations between age and MMSE decline scores were observed in Parkinson’s disease dementia (rho = 0.553, P = 0.008, n = 22). The last MMSE scores before death were lower and the rate of MMSE decline was higher in Alzheimer’s disease compared to Parkinson’s disease dementia or dementia with Lewy bodies. No significant association was observed between diagnosis and gender. Table 1 Demographics and clinical characteristics of the subjects included in this study   Controls  PDD  DLB  AD  MS (n = 8)  ELISA (n = 24)  MS (n = 8)  ELISA (n = 24)  MS (n = 7)  ELISA (n = 26)  MS (n = 9)  ELISA (n = 18)  Age (mean ± SD)  83 ± 3.8  80.2 ± 7.5  82.5 ± 6  81.8 ± 4.8  83.38 ± 3.5  81.1 ± 6.5  85.67 ± 2.7  88.1 ± 7.3  Gender  4 M/4 F  14 M/10 F  3 M/5 F  10 M/14 F  3 M/4 F  17 M/9 F  3 M/6 F  6 M/12 F  PMD (h) (mean ± SD)  35.2 ± 18.4  38.8 ± 23.4  25.3 ± 7.8  33.9 ± 15.8  21.7 ± 12.1  27.6 ± 21.3  23.7 ± 10  35.0 ± 22.8  pH (mean ± SD)  6.45 ± 0.3  6.46 ± 0.3  6.61 ± 0.3  6.53 ± 0.3  6.45 ± 0.5  6.43 ± 0.4  6.28 ± 0.3  6.30 ± 0.3  Years of dementia  -  -  3.8 ± 2.5  2.78 ± 2  4 ± 1.5  5.9 ± 3  9.1 ± 2.1  9.7 ± 2.8  Hoehn and Yahr scale  -  -  4.75 ± 0.46  4.57 ± 1.08  NA  NA  -  -  Last MMSE  NA  NA  12.6 ± 7.7  14.1 ± 8.0  22 ± 9.1  14.4 ± 9.8  7.67 ± 7.8  8.5 ± 7.6  MMSE decline (y)  NA  NA  2.1 ± 1.2  1.8 ± 1.2  2.5 ± 3.3  2.9 ± 2.8  4.2 ± 4.5  3.5 ± 3.5  Amyloid-β plaque  0.37 ± 0.7  0.36 ± 0.65  1.85 ± 1.3  1.3 ± 1.2  1.7 ± 1.4  1.96 ± 1.0  2.88 ± 0.3  2.72 ± 0.7  Tangle  0  0.18 ± 0.4  0.33 ± 0.5  0.43 ± 0.5  0.7 ± 0.5  0.85 ± 0.7  2.67 ± 0.5  2.5 ± 0.6  α-synuclein  -  -  0.7 ± 1.1  0.6 ± 0.8  2.3 ± 0.9  1.8 ± 1.1  0.22 ± 0.4  0.18 ± 0.4    Controls  PDD  DLB  AD  MS (n = 8)  ELISA (n = 24)  MS (n = 8)  ELISA (n = 24)  MS (n = 7)  ELISA (n = 26)  MS (n = 9)  ELISA (n = 18)  Age (mean ± SD)  83 ± 3.8  80.2 ± 7.5  82.5 ± 6  81.8 ± 4.8  83.38 ± 3.5  81.1 ± 6.5  85.67 ± 2.7  88.1 ± 7.3  Gender  4 M/4 F  14 M/10 F  3 M/5 F  10 M/14 F  3 M/4 F  17 M/9 F  3 M/6 F  6 M/12 F  PMD (h) (mean ± SD)  35.2 ± 18.4  38.8 ± 23.4  25.3 ± 7.8  33.9 ± 15.8  21.7 ± 12.1  27.6 ± 21.3  23.7 ± 10  35.0 ± 22.8  pH (mean ± SD)  6.45 ± 0.3  6.46 ± 0.3  6.61 ± 0.3  6.53 ± 0.3  6.45 ± 0.5  6.43 ± 0.4  6.28 ± 0.3  6.30 ± 0.3  Years of dementia  -  -  3.8 ± 2.5  2.78 ± 2  4 ± 1.5  5.9 ± 3  9.1 ± 2.1  9.7 ± 2.8  Hoehn and Yahr scale  -  -  4.75 ± 0.46  4.57 ± 1.08  NA  NA  -  -  Last MMSE  NA  NA  12.6 ± 7.7  14.1 ± 8.0  22 ± 9.1  14.4 ± 9.8  7.67 ± 7.8  8.5 ± 7.6  MMSE decline (y)  NA  NA  2.1 ± 1.2  1.8 ± 1.2  2.5 ± 3.3  2.9 ± 2.8  4.2 ± 4.5  3.5 ± 3.5  Amyloid-β plaque  0.37 ± 0.7  0.36 ± 0.65  1.85 ± 1.3  1.3 ± 1.2  1.7 ± 1.4  1.96 ± 1.0  2.88 ± 0.3  2.72 ± 0.7  Tangle  0  0.18 ± 0.4  0.33 ± 0.5  0.43 ± 0.5  0.7 ± 0.5  0.85 ± 0.7  2.67 ± 0.5  2.5 ± 0.6  α-synuclein  -  -  0.7 ± 1.1  0.6 ± 0.8  2.3 ± 0.9  1.8 ± 1.1  0.22 ± 0.4  0.18 ± 0.4  One-way ANOVA followed by Bonferroni post hoc tests showed Alzheimer’s disease patients were older compared to the other diagnostic groups [ANOVA, F(3,91) = 5.791, P = 0.001 in controls; P = 0.019 in Parkinson’s disease dementia; P = 0.05 in dementia with Lewy bodies]. Dementia with Lewy bodies and Alzheimer’s disease had longer dementia duration than Parkinson’s disease dementia [ANOVA, F(2,44) = 26.738 P = 0.006 Parkinson’s disease dementia versus dementia with Lewy bodies; P ≤ 0.001 Parkinson’s disease dementia versus Alzheimer’s disease; P = 0.002 dementia with Lewy body versus Alzheimer’s disease]. There were no significant differences between diagnostic groups in other variables except cognition and pathology, which is further discussed in Fig. 3 and Supplementary Tables 4 and 5. AD = Alzheimer's disease; DLB = dementia with Lewy bodies; PMD = post-mortem delay; MS = mass spectrometry; NA = not available; PDD = Parkinson's disease with dementia; SD = standard deviation. Proteome analyses To avoid large interindividual variation, a confounding factor in previous comparative proteomic studies on clinical material, we included a large number of cases per disease type (n = 7–9). The 32 prefrontal cortex samples were processed by Filter Aided Sample Prep (FASP) (Wisniewski et al., 2009) and cases were individually labelled at peptide level with four sets of isobaric tags (TMT10plex). Each TMT set contained eight channels with randomized samples (Supplementary Table 2) and two channels with the internal reference sample (made by pooling aliquots from all 32 samples). Each TMT set was fractionated into 72 fractions by HiRIEF (Branca et al., 2014) with the broad range IPG 3-10 strip prior to LC-MS analysis (Fig. 1A). A total of 10 325 proteins (gene-centric) were identified (1% FDR, protein level FDR) (of which 7033 were common to all 32 samples) as a result of the proteomic database search (Fig. 1A). The DAVID (The Database for Annotation, Visualization and Integrated Discovery) platform as well as the GOrilla tool were used for detailed data analyses. Across the disease groups, 102 proteins were commonly differentially regulated (Fig. 1B). In the dementia with Lewy bodies group, 1010 differentially expressed proteins (of which 448 were upregulated and 562 were downregulated in dementia with Lewy bodies compared to non-demented controls) were introduced in the DAVID platform. Of these, 392 were assigned to 22 predicted KEGG pathways, with the identified differentially accumulated proteins found to be enriched in pathways related to human diseases (40%), organismal systems (30.3%), cellular processes (10.6%), genetic information processing (9.2%), metabolism (8.3%) and environmental information processing (1.9%) (Fig. 1C and Supplementary Table 3). Interestingly, pathways such as Parkinson’s disease (n = 22), Alzheimer’s disease (n = 24) and Huntington’s disease (n = 27), dopaminergic synapses (n = 17), protein processing in endoplasmic reticulum (n = 27) and oxidative phosphorylation (n = 23) were significantly enriched with the highest number of alterations (Fig. 1C and Supplementary Table 3). In Parkinson’s disease dementia, 485 proteins (286 upregulated and 199 downregulated compared to non-demented controls) were introduced in the DAVID platform, of which 182 were assigned to eight KEGG pathways. In Alzheimer’s disease, of the 593 (255 upregulated and 338 downregulated compared to non-demented controls) proteins introduced, 241 were assigned to 12 KEGG pathways. Since there were generally fewer than 10 hits for the pathways, neither Parkinson’s disease dementia nor Alzheimer’s disease KEGG pathway enrichment was further scrutinized (data not shown). For the GO analyses, in the case of dementia with Lewy bodies, 1003 proteins with GO terms were assigned to the annotated 1010 proteins using the GOrilla tool and classified into three groups (biological process, molecular function and cellular component) (Fig. 1D). Within the three main categories, only significant classifications were found in the cellular compartment (FDR, q < 0.05), and these were related to organelles and mitochondria (Fig. 1D and Supplementary Table 4). In Alzheimer’s disease, the only significant hits were found for the molecular function category related to translation initiation and RNA binding, while in Parkinson’s disease dementia no significant classification was found (Supplementary Table 4). Synaptic dysfunction in Parkinson’s disease dementia, dementia with Lewy bodies and Alzheimer’s disease Using GO terminology, we identified 851 proteins related to synaptic transmission (Supplementary Table 5), of which 25 synaptic proteins were significantly altered in the various dementias as shown by the SAM analyses with low FDR (q < 3.5%) (Table 2). As the levels of CAMK2 and neurogranin have already been assessed in this cohort (Vallortigara et al., 2014; Bereczki et al., 2016), we selected eight additional differentially regulated synaptic proteins based on their function, fold change, and antibody availability, for further validation on a larger cohort (containing the mass spectrometry cohort) using ELISA or western blot analyses. The synaptic protein with the most conspicuous drop in concentration (29–33%) was LRFN2 (leucine-rich repeat and fibronectin type-III domain-containing protein 2) in all three dementias (Table 3 and Fig. 2A). In Parkinson’s disease dementia, SNAP47 and SYT2 concentrations (Supplementary Fig. 1) also decreased compared to controls while GAP43 concentration decreased in comparison to the Alzheimer’s disease group. Five of eight measured synaptic proteins were decreased in dementia with Lewy bodies compared to non-demented controls. In addition to concentrations of LRFN2, SNAP47 and SV2C, levels of SYBU and SYT2 were also decreased (Table 3, Fig. 2A and Supplementary Fig. 1). In Alzheimer’s disease, apart from LRFN2, only GRIA3 concentration was significantly decreased (Table 3 and Fig. 2A). Proteomic profiling revealed no significant differences in α-synuclein levels (Supplementary Table 5). Table 2 List of synaptic proteins differentially expressed between dementia cases and controls based on mass spectrometry data analysis   Gene ID  Uniprot ID  Protein name  Fold change (min-max)  q-value (%)  Min. peptides  Min. quant. PSMs  PDD  SV2C  Q496J9  Synaptic vesicle glycoprotein 2C  0.6 (1.7−0.3)  12  3  5  NRGNa,b  Q92686  Neurogranin  0.42 (1.5−0.2)  0  1  5  CBLN4a  Q9NTU7  Cerebellin-4  0.7 (0.9−0.5)  0  1  1  BDNFa  P23560  Brain-derived neurotrophic factor  0.73 (1.0−0.6)  0  1  1  GAP43a  P17677  Neuromodulin  0.74 (1.1−0.6)  0  57  714  DLB  GRIA3  P42263  Glutamate receptor 3  0.56 (1.3−0.3)  3.5  7  9  CAMK2Ab  Q9UQM7  Calcium/calmodulin-dependent protein kinase type II subunit alpha  0.6 (1.4−0.4)  3.5  24  100  SYBU  Q9NX95  Syntabulin  0.61 (1.1−0.4)  3.5  2  2  VDAC2  P45880  Voltage-dependent anion-selective channel protein 2  0.62 (1.6−0.4)  3.5  14  90  ARC  Q7LC44  Activity-regulated cytoskeleton-associated protein  0.62 (1.0−0.4)  3.5  1  1  RAB11A  P62491  Ras-related protein Rab-11A  0.63 (1.0−0.4)  3.5  1  1  PDYN  P01213  Proenkephalin-B  0.64 (1.3−0.4)  3.5  1  1  GRIA4  P48058  Glutamate receptor 4  0.64 (1.3−0.3)  3.5  2  2  SYT2  Q8N9I0  Synaptotagmin-2  0.64 (1.4−0.4)  3.5  7  9  CAMK2G  Q13555  Calcium/calmodulin-dependent protein kinase type II subunit gamma  0.64 (1.4−0.5)  3.5  19  32  CNIH2  Q6PI25  Protein cornichon homolog 2  0.65 (1.3−0.4)  3.5  1  1  KCNIP2  Q9NS61  Kv channel-interacting protein 2  0.65 (1.4−0.4)  3.5  1  1  SNAP47  Q5SQN1  Synaptosomal-associated protein 47  0.66 (0.9−0.4)  3.5  2  2  TECR  Q9NZ01  Very-long-chain enoyl-CoA reductase  0.67 (1.2−0.4)  3.5  5  8  CACNG2  Q9Y698  Voltage-dependent calcium channel gamma-2 subunit  0.67 (1.0−0.5)  3.5  2  2  PVRL3  Q9NQS3  Nectin-3  0.68 (1.1−0.4)  3.5  1  1  LRFN2  Q9ULH4  Leucine-rich repeat and fibronectin type-III domain-containing protein 2  0.68 (1.1−0.5)  3.5  3  3  GRIK2  Q13002  Glutamate receptor ionotropic, kainate 2  0.68 (1.2−0.5)  3.5  2  2  CACNG3  O60359  Voltage-dependent calcium channel gamma-3 subunit  0.7 (1.0−0.5)  3.5  1  1  TNK2  Q07912  Activated CDC42 kinase 1  0.71 (1.1−0.5)  3.5  2  2  CAMKK1  Q8N5S9  Calcium/calmodulin-dependent protein kinase kinase 1  0.73 (1.1−0.6)  3.5  12  18    Gene ID  Uniprot ID  Protein name  Fold change (min-max)  q-value (%)  Min. peptides  Min. quant. PSMs  PDD  SV2C  Q496J9  Synaptic vesicle glycoprotein 2C  0.6 (1.7−0.3)  12  3  5  NRGNa,b  Q92686  Neurogranin  0.42 (1.5−0.2)  0  1  5  CBLN4a  Q9NTU7  Cerebellin-4  0.7 (0.9−0.5)  0  1  1  BDNFa  P23560  Brain-derived neurotrophic factor  0.73 (1.0−0.6)  0  1  1  GAP43a  P17677  Neuromodulin  0.74 (1.1−0.6)  0  57  714  DLB  GRIA3  P42263  Glutamate receptor 3  0.56 (1.3−0.3)  3.5  7  9  CAMK2Ab  Q9UQM7  Calcium/calmodulin-dependent protein kinase type II subunit alpha  0.6 (1.4−0.4)  3.5  24  100  SYBU  Q9NX95  Syntabulin  0.61 (1.1−0.4)  3.5  2  2  VDAC2  P45880  Voltage-dependent anion-selective channel protein 2  0.62 (1.6−0.4)  3.5  14  90  ARC  Q7LC44  Activity-regulated cytoskeleton-associated protein  0.62 (1.0−0.4)  3.5  1  1  RAB11A  P62491  Ras-related protein Rab-11A  0.63 (1.0−0.4)  3.5  1  1  PDYN  P01213  Proenkephalin-B  0.64 (1.3−0.4)  3.5  1  1  GRIA4  P48058  Glutamate receptor 4  0.64 (1.3−0.3)  3.5  2  2  SYT2  Q8N9I0  Synaptotagmin-2  0.64 (1.4−0.4)  3.5  7  9  CAMK2G  Q13555  Calcium/calmodulin-dependent protein kinase type II subunit gamma  0.64 (1.4−0.5)  3.5  19  32  CNIH2  Q6PI25  Protein cornichon homolog 2  0.65 (1.3−0.4)  3.5  1  1  KCNIP2  Q9NS61  Kv channel-interacting protein 2  0.65 (1.4−0.4)  3.5  1  1  SNAP47  Q5SQN1  Synaptosomal-associated protein 47  0.66 (0.9−0.4)  3.5  2  2  TECR  Q9NZ01  Very-long-chain enoyl-CoA reductase  0.67 (1.2−0.4)  3.5  5  8  CACNG2  Q9Y698  Voltage-dependent calcium channel gamma-2 subunit  0.67 (1.0−0.5)  3.5  2  2  PVRL3  Q9NQS3  Nectin-3  0.68 (1.1−0.4)  3.5  1  1  LRFN2  Q9ULH4  Leucine-rich repeat and fibronectin type-III domain-containing protein 2  0.68 (1.1−0.5)  3.5  3  3  GRIK2  Q13002  Glutamate receptor ionotropic, kainate 2  0.68 (1.2−0.5)  3.5  2  2  CACNG3  O60359  Voltage-dependent calcium channel gamma-3 subunit  0.7 (1.0−0.5)  3.5  1  1  TNK2  Q07912  Activated CDC42 kinase 1  0.71 (1.1−0.5)  3.5  2  2  CAMKK1  Q8N5S9  Calcium/calmodulin-dependent protein kinase kinase 1  0.73 (1.1−0.6)  3.5  12  18  Differences were assessed with respect to the control group. aDifferences between Parkinson’s disease dementia (PDD) and Alzheimer’s disease. bProteins previously measured in this cohort, which are discussed in the ‘Results’ and ‘Discussion’ sections. Proteins in bold were chosen for further analyses with ELISA or western blot. In addition to fold change, including minimum and maximum fold-change values in parenthesis, q-value, the minimal number of unique peptides and the minimal number of quantified peptide-spectrum matches (PSMs) per tandem mass tag (TMT) set is shown. Table 3 Differences in synaptic protein levels between control and dementia groups using western blotting and ELISA data   Control (n = 24)  PDD (n = 24)  DLB (n = 25)  AD (n = 18)  SNAP47(pg/ml)  156.9 ± 48.6  118.8 ± 29.2  P = 0.005  121.9 ± 39.1  P = 0.022  139.5 ± 26.5  P = 0.848  SV2C (ng/ml)  9.9 ± 2.7  8.3 ± 1.9  P = 0.068  7.5 ± 1.9  P = 0.01  8.3 ± 1.5  P = 0.085  GRIA3 (ng/ml)  6.68 ± 2.1  6.16 ± 2.1  P = 1.00  5.3 ± 1.3  P = 0.109  4.9 ± 1.6  P = 0.036  GRIA4 (ng/ml)  14.1 ± 2.1  13.1 ± 2.0  P = 1.00  12.8 ± 2.9  P = 0.721  12.5 ± 2.8  P = 0.358  LRNF2 (ng/ml)  6.1 ± 2  4.3 ± 1.8  P = 0.032  4.1 ± 1.9  P = 0.01  4.3 ± 2.2  P = 0.05  GAP43 (pg/ml)  931.5 ± 282  747.4 ± 202  P = 0.004*  842.8 ± 148  P = 1.00  972.8 ± 156  P = 1.00  SYBU  1.88 ± 0.75  1.71 ± 0.67  P = 1.00  1.18 ± 0.54  P = 0.002, P = 0.033**  1.4 ± 0.61  P = 0.799  SYT2  2.03 ± 0.43  1.64 ± 0.36  P = 0.037  1.55 ± 0.48  P = 0.006  1.74 ± 0.70  P = 0.330    Control (n = 24)  PDD (n = 24)  DLB (n = 25)  AD (n = 18)  SNAP47(pg/ml)  156.9 ± 48.6  118.8 ± 29.2  P = 0.005  121.9 ± 39.1  P = 0.022  139.5 ± 26.5  P = 0.848  SV2C (ng/ml)  9.9 ± 2.7  8.3 ± 1.9  P = 0.068  7.5 ± 1.9  P = 0.01  8.3 ± 1.5  P = 0.085  GRIA3 (ng/ml)  6.68 ± 2.1  6.16 ± 2.1  P = 1.00  5.3 ± 1.3  P = 0.109  4.9 ± 1.6  P = 0.036  GRIA4 (ng/ml)  14.1 ± 2.1  13.1 ± 2.0  P = 1.00  12.8 ± 2.9  P = 0.721  12.5 ± 2.8  P = 0.358  LRNF2 (ng/ml)  6.1 ± 2  4.3 ± 1.8  P = 0.032  4.1 ± 1.9  P = 0.01  4.3 ± 2.2  P = 0.05  GAP43 (pg/ml)  931.5 ± 282  747.4 ± 202  P = 0.004*  842.8 ± 148  P = 1.00  972.8 ± 156  P = 1.00  SYBU  1.88 ± 0.75  1.71 ± 0.67  P = 1.00  1.18 ± 0.54  P = 0.002, P = 0.033**  1.4 ± 0.61  P = 0.799  SYT2  2.03 ± 0.43  1.64 ± 0.36  P = 0.037  1.55 ± 0.48  P = 0.006  1.74 ± 0.70  P = 0.330  Differences in protein levels between disease groups and controls were determined using Kruskal-Wallis test followed by Dunn’s post hoc test. ELISA values are expressed in pg/ml or ng/ml (means ± SD). Western blot changes are expressed in relative units. P-values represent statistically significant differences between dementia and non-demented control groups. *Significant differences between Parkinson’s disease dementia (PDD) and Alzheimer’s disease (AD) groups. **Significant differences between dementia with Lewy bodies (DLB) and Parkinson’s disease dementia groups. Figure 2 View largeDownload slide Changes in synaptic protein levels and their contribution to discriminating patient groups. (A) Synaptic protein levels differed between the dementia groups. (B–E) Univariate statistical analyses were performed using Kruskal-Wallis test followed by post hoc Dunn’s multiple comparison test. Multivariate analyses show the contribution of synaptic proteins to discriminate controls from the different patient groups. Plots showing the variables of importance and their corresponding jack-knifed confidence intervals for the separation between controls (C) and Parkinson’s disease dementia patients (PDD, B), controls and dementia with Lewy bodies patients (DLB, C), controls and Alzheimer’s disease patients (AD, D) and Parkinson’s disease dementia patients and Alzheimer’s disease patients (E). A measure with high covariance is more likely to have an impact on group separation than a variable with low covariance. Measures with confidence intervals that include zero have low reliability. Figure 2 View largeDownload slide Changes in synaptic protein levels and their contribution to discriminating patient groups. (A) Synaptic protein levels differed between the dementia groups. (B–E) Univariate statistical analyses were performed using Kruskal-Wallis test followed by post hoc Dunn’s multiple comparison test. Multivariate analyses show the contribution of synaptic proteins to discriminate controls from the different patient groups. Plots showing the variables of importance and their corresponding jack-knifed confidence intervals for the separation between controls (C) and Parkinson’s disease dementia patients (PDD, B), controls and dementia with Lewy bodies patients (DLB, C), controls and Alzheimer’s disease patients (AD, D) and Parkinson’s disease dementia patients and Alzheimer’s disease patients (E). A measure with high covariance is more likely to have an impact on group separation than a variable with low covariance. Measures with confidence intervals that include zero have low reliability. Panel of synaptic proteins discriminate between control and dementia diagnoses Multivariate analyses showed that synaptic protein levels were able to provide a clear separation between controls and the different patient groups; however, no synaptic protein alone was able to achieve clear discrimination between groups. In addition, a clear discrimination between patients with Parkinson’s disease dementia and Alzheimer’s disease was also observed (77.8% sensitivity with 80% specificity, Table 4). All models were statistically significant and showed sensitivity, specificity, and positive and negative predictive values that were above 80% in the case of dementia with Lewy bodies versus control and Alzheimer’s disease versus control groups, and were close to 75% in patients with Parkinson’s disease dementia versus controls (Table 4). Table 4 Sensitivity, specificity, positive and negative predictive values for each model Models  Sensitivity (95% CI)  Specificity (95% CI)  PPV (95% CI)  NPV (95% CI)  Controls versus PDD  73.7 (48.8–90.9)  73.9 (51.6–89.8)  70.0 (45.7–88.1)  77.3 (54.6–92.2)  Controls versus DLB  83.3 (62.6–95.3)  80.0 (59.3–93.2)  80 (59.3–93.2)  83.3 (62.6–95.3)  Controls versus AD  81.3 (54.4–95.6)  80.8 (60.7–93.5)  72.2 (46.5–90.3)  87.5 (67.6–97.3)  PDD versus AD  77.8 (52.4–93.6)  80.0 (56.3–94–3)  77.8 (52.4–93.6)  80.0 (56.3–94.3)  Models  Sensitivity (95% CI)  Specificity (95% CI)  PPV (95% CI)  NPV (95% CI)  Controls versus PDD  73.7 (48.8–90.9)  73.9 (51.6–89.8)  70.0 (45.7–88.1)  77.3 (54.6–92.2)  Controls versus DLB  83.3 (62.6–95.3)  80.0 (59.3–93.2)  80 (59.3–93.2)  83.3 (62.6–95.3)  Controls versus AD  81.3 (54.4–95.6)  80.8 (60.7–93.5)  72.2 (46.5–90.3)  87.5 (67.6–97.3)  PDD versus AD  77.8 (52.4–93.6)  80.0 (56.3–94–3)  77.8 (52.4–93.6)  80.0 (56.3–94.3)  AD = Alzheimer's disease; CI = confidence interval; DLB = dementia with Lewy bodies; NPV = negative predictive value; PDD = Parkinson's disease dementia; PPV = positive predictive value. The model comparing control with Parkinson’s disease dementia cases showed a modest predictive power of Q2(Y) = 0.173 in discriminating controls from patients with Parkinson’s disease dementia. With the exception of GRIA4 and SYBU, all variables contributed to the separation between these groups (Fig. 2B). The dementia with Lewy bodies model showed a good predictive power of Q2(Y) = 0.471 in discriminating controls from patients with dementia with Lewy bodies. All synaptic proteins significantly contributed to the separation between groups, with the exception of GRIA4 and GAP43 (Fig. 2C). The Alzheimer’s disease model showed a good predictive power of Q2(Y) = 0.427 in the discrimination of controls from Alzheimer’s disease patients. The synaptic proteins that significantly contributed to the separation were LRFN2, GRIA3, SV2C and SYT2 (Fig. 2D). A good predictive power of Q2(Y) = 0.438 reflected the capacity to distinguish Parkinson’s disease dementia and Alzheimer’s disease pathology based on the contribution of both GAP43 and SNAP47 (Fig. 2E). These results were still significant after correcting for the effects of age. Patients with Parkinson’s disease dementia and dementia with Lewy bodies could not be reliably discriminated from one another, supporting that they are part of the same disease spectrum. No significant differences were found between dementia with Lewy bodies and Alzheimer’s disease, most likely attributed to the common amyloid-related pathology. Associations between synaptic proteins and neuropathological scores Correlations between the eight synaptic proteins validated on the larger cohort and Alzheimer’s disease and dementia with Lewy bodies regional pathologies were analysed (Supplementary Table 6). In Parkinson’s disease dementia there were significant correlations between α-synuclein and SNAP47 (rho = −0.539, P = 0.008) and GRIA3 (rho = −0.449, P = 0.047) whereas in dementia with Lewy bodies, α-synuclein correlated with SV2C (rho = −0.441, P = 0.035). Amyloid-β scores correlated significantly with GRIA4 both in Parkinson’s disease dementia (rho = −0.471, P = 0.031) and in dementia with Lewy bodies (rho = −0.444, P = 0.05). The only significant association we found between synaptic proteins and tangle scores was in the case of GRIA3 in Parkinson’s disease dementia (rho = −0.460, P = 0.041). No neuropathological associations were found in Alzheimer’s disease. Correlations between synaptic proteins and cognitive impairment We explored whether synaptic protein changes were associated with cognitive impairment. Due to the exploratory nature of these analyses and small number of patients per group, these correlations are presented without adjusting for multiple comparisons. Our calculations revealed that only results at P < 0.0087 would be considered statistically significant with FDR corrections, which is quite a stringent threshold. Significant associations between synaptic proteins and cognitive decline were found in Parkinson’s disease dementia, dementia with Lewy bodies and in Alzheimer’s disease (Fig. 3 and Supplementary Table 7). Synaptic vesicle protein SV2C was strongly associated with the rate of cognitive decline, i.e. reduced levels correlated with faster cognitive decline in Parkinson’s disease dementia (rho = −0.486, P = 0.022) and dementia with Lewy bodies (rho = −0.889, P = 0.0001) and low last MMSE score in dementia with Lewy bodies (rho = 0.759, P = 0.0001) (Fig. 3A and B). Decrease in SNAP47 concentration was associated with worsening cognition reflected both by last MMSE scores (rho = 0.480, P = 0.027) and the rate of MMSE decline (rho = −0.559, P = 0.008) in Parkinson’s disease dementia. GRIA3 also presented associations both with cognitive decline (rho = −0.726, P = 0.0001) and last MMSE scores (rho = 0.479, P = 0.033) in dementia with Lewy bodies. Furthermore, in dementia with Lewy bodies lowered SYBU levels correlated with lower MMSE scores (rho = 0.493, P = 0.023). In Alzheimer’s disease only LRFN2 presented strong associations with worsening cognition (rho = −0.613, P = 0.012) and the last MMSE scores (rho = 0.730, P = 0.001) (Fig. 3C). No significant associations were found between MMSE scores and GRIA4, SYT2 or GAP43 proteins, however GAP43 was found to be associated to the total years of dementia in Alzheimer’s disease (rho = 0.499, P = 0.035). No associations were found between motor symptoms and synaptic proteins in patients with Parkinson’s disease with dementia (data not shown). The results remained significant after controlling for the effects of age (data not shown). Figure 3 View largeDownload slide Correlations between synaptic proteins and cognitive impairment in Parkinson’s disease dementia (A), dementia with Lewy bodies (B) and Alzheimer’s disease (C). Decreased SNAP47, SV2C and GRIA3 concentrations (A) correlated with cognitive impairment in Parkinson’s disease dementia (PDD). SV2C and GRIA3 concentrations are negatively correlated with the rate of MMSE decline in dementia with Lewy bodies (DLB, B) showing, along with SYBU levels, positive correlations with the last MMSE scores (B). Negative correlations between LRFN2 concentrations and the rate of MMSE decline as well as positive correlations with the last MMSE scores were observed in Alzheimer’s disease (AD, C). Associations were analysed using Spearman correlations. Figure 3 View largeDownload slide Correlations between synaptic proteins and cognitive impairment in Parkinson’s disease dementia (A), dementia with Lewy bodies (B) and Alzheimer’s disease (C). Decreased SNAP47, SV2C and GRIA3 concentrations (A) correlated with cognitive impairment in Parkinson’s disease dementia (PDD). SV2C and GRIA3 concentrations are negatively correlated with the rate of MMSE decline in dementia with Lewy bodies (DLB, B) showing, along with SYBU levels, positive correlations with the last MMSE scores (B). Negative correlations between LRFN2 concentrations and the rate of MMSE decline as well as positive correlations with the last MMSE scores were observed in Alzheimer’s disease (AD, C). Associations were analysed using Spearman correlations. Discussion Both Alzheimer’s disease and Lewy body diseases are characterized by substantial synaptic loss, which to date serves as the best correlate with cognitive impairment (Terry et al., 1991; Scheff et al., 2007; Pienaar et al., 2012). More than 1000 proteins participate in the finely tuned process of synaptic transmission, a process that comprises interactions between synaptic vesicle membrane proteins as well as presynaptic and postsynaptic membrane proteins (Sudhof and Rothman, 2009). While general synaptic loss is a common feature of dementia, specific pre- and postsynaptic proteins such as Rab3A, SNAP25, synaptophysin crucial for vesicle trafficking, exo- and endocytosis have been found specifically altered in neurodegenerative diseases (Whitfield et al., 2014; Bereczki et al., 2016) along with NMDA and AMPA receptors, PSD95 and neurogranin (Lipton and Rosenberg, 1994; Lee et al., 2008; Whitfield et al., 2014; Bereczki et al., 2016) playing a role in long-term potentiation (Fig. 4). Of note, in our current proteomics study we have not identified significant changes neither for SNAP25 nor for Rab3a, which could be due to the relative vulnerability of Rab3a to post-mortem delay time (Ferrer et al., 2007). The potential use of key synaptic proteins as biomarkers has recently come in the spotlight of discussions in various dementias (Bereczki et al., 2016; Wellington et al., 2016). Note that the regional specific post-mortem synaptic protein profile might differ from the synaptic protein profile of CSF (Bereczki et al., 2016, 2017; Remnestal et al., 2016). Although CSF biomarkers are most informative in portraying the biochemical picture of the brain, blood-based biomarkers are more desired for large-scale screening (Mattsson et al., 2015). Of note, even if a biomarker has shown high specificity and sensitivity, its utility as a theragnostic biomarker is not guaranteed (Mattsson et al., 2015). Figure 4 View largeDownload slide Schematic overview of synaptic proteins with altered levels in dementia. The diagram depicts the proteins involved in the synaptic vesicle cycle focusing on the docking and priming proteins (VAMP2, Syntaxin-1, SNAP, Munc18a), along with proteins involved in the recycling of synaptic vesicles (Rab3A, SV2C) as well as postsynaptic proteins (NRGN, PSD95, LRFN2) and receptor proteins (GRIA3, 4) found to be differentially regulated in the various dementias. AMPAR = AMPA receptor/GRIA; GAP43 = neuromodulin; NMDAR = N-methyl-d-aspartic acid receptor; NRGN = neurogranin; VDCC = voltage-dependent calcium channel. Figure 4 View largeDownload slide Schematic overview of synaptic proteins with altered levels in dementia. The diagram depicts the proteins involved in the synaptic vesicle cycle focusing on the docking and priming proteins (VAMP2, Syntaxin-1, SNAP, Munc18a), along with proteins involved in the recycling of synaptic vesicles (Rab3A, SV2C) as well as postsynaptic proteins (NRGN, PSD95, LRFN2) and receptor proteins (GRIA3, 4) found to be differentially regulated in the various dementias. AMPAR = AMPA receptor/GRIA; GAP43 = neuromodulin; NMDAR = N-methyl-d-aspartic acid receptor; NRGN = neurogranin; VDCC = voltage-dependent calcium channel. The technological advance in proteomics analyses has provided high-throughput screening methods in the quest for biomarkers of neurodegenerative disorders in post-mortem as well as in CSF or blood-based samples. To our knowledge, we are the first to provide a systematic proteome profile comparison on post-mortem human brains (n = 8 controls, n = 8 Parkinson’s disease dementia, n = 7 dementia with Lewy bodies, n = 9 Alzheimer’s disease, i.e. 32 samples in total) from the prefrontal cortex (Brodmann area 9), revealing a pattern of synaptic protein loss across different neurodegenerative diseases. We have adopted a proteomics-driven discovery approach and after identifying roughly half of the human proteome, we validated lead synaptic candidates in a larger post-mortem brain cohort of 92 cases. Comparative proteomics highlighted significant loss of several synaptic proteins across dementias including presynaptic proteins (GRIK2, CAMK2A, BDNF, PDYN), synaptic vesicle priming proteins (SNAP47) synaptic vesicle proteins (SV2C, SYT2), proteins found in both pre- and postsynaptic terminus (GAP43, LRFN2) and postsynaptic proteins (GRIA3, GRIA4, ARC, CNIH2, PVRL3, NRGN). Among these proteins, SNAP47, SV2C, GRIA3, SYBU and LRFN2 in the prefrontal cortex correlated with cognitive decline in demented cases. The levels of apocalmodulin-binding proteins, NRGN and GAP43 diminished, which in turn might further contribute to the altered CAMK2 and AMPA receptor (GRIA3, GRIA4) mediated synaptic transmission. Their reduction reflects a selective alteration in a subset of synaptic proteins, suggesting that a decline in synaptic function rather than synaptic loss plays a more relevant role in contributing to dementia progression. However, the mechanisms leading to synapse destabilization and neuronal death remain elusive. There is evidence showing that synaptic plasticity underlying learning and memory often involves activity-dependent recruitment of synaptic AMPA receptors (GRIA) (Nicoll and Malenka, 1999; Kandel, 2001). During long-term potentiation, GRIA exocytosis is mediated by Q-SNARE proteins syntaxin-3 and SNAP47 (Jurado et al., 2013). Dysregulation of AMPA receptors has also been implicated in numerous neurodegenerative and psychiatric disorders (Lipton and Rosenberg, 1994). Likewise, deletion of LRFN2 localized both to the presynaptic and postsynaptic membrane has been linked to selective working memory and executive deficits, impaired intellectual functioning and auditory-verbal problems (Thevenon et al., 2016). Additional proteins with a potential role in cognitive impairment such as the members of the SNARE family, synaptotagmins (SYT), PSD95 and synaptic vesicle 2 (SV2) proteins have also been incriminated in this captivatingly complicated process of synaptic plasticity (Bajjalieh et al., 1994; Xu et al., 2007; Dun et al., 2010; Whitfield et al., 2014) (Fig. 4). In line with this, SYT2 levels have also been shown to be decreased in plasma neuronal-derived exosomes (Goetzl et al., 2016). Likewise, CSF levels of GAP43 were found to be altered in separate studies in Parkinson’s disease (Sjogren et al., 2000) and in Alzheimer’s disease (Sjogren et al., 2001; Goetzl et al., 2016). Our observation of GRIA3 and its correlation with cognitive impairment supports previous observations of reductions of AMPA receptors trafficking, or anchoring into dendritic spines with synaptic and cognitive disturbances (Henley and Wilkinson, 2013). Interestingly, significant cognitive associations with GRIA3 were present only in dementia with Lewy bodies and Parkinson’s disease dementia with no apparent association found in Alzheimer’s disease. The marked reductions of synaptic proteins in patients with dementia with Lewy bodies and Parkinson’s disease dementia could reflect a greater frontal degeneration in Lewy body dementia in comparison with Alzheimer’s disease, which usually affects the prefrontal cortex less than other medial and lateral temporal areas (Burton et al., 2012). Together with previous findings showing alterations in levels of strategic synaptic proteins such as Rab3A, PSD95 and SNARE proteins, and their correlation to cognitive domains (Gottschall et al., 2010; Mukaetova-Ladinska et al., 2013; Howlett et al., 2015; Vallortigara et al., 2014; Whitfield et al., 2014), our results provide support to the link between cognitive performance and synaptic protein loss in Lewy body dementia. In line with our previous study, the current work confirms the power of synaptic proteins (Bereczki et al., 2016) in discriminating patients with neurodegenerative diseases from controls with good sensitivity and specificity (>80%). In addition, we also found that GAP43, together with SNAP47, contributed to a clear separation between patients with Parkinson’s disease dementia and Alzheimer’s disease, highlighting the potential role of these proteins in disease discrimination. α-Synuclein is deeply involved in the synaptic vesicle trafficking required for a proper neurotransmitter release (Sidhu et al., 2004). Although we selected Alzheimer’s disease cases with low α-synuclein pathology, we did not observe any difference in the overall levels of monomeric α-synuclein protein, neither between Parkinson’s disease dementia and Alzheimer’s disease nor between dementia with Lewy bodies and Alzheimer’s disease, which could be partly due to the relative vulnerability of α-synuclein to post-mortem delay and storage temperature (Ferrer et al., 2007). This finding is in agreement with a previous proteomics study (Shi et al., 2009) carried out in Parkinson’s disease patients while another proteomics study in patients with Parkinsonism-dementia complex of Guam reported accumulation of α-synuclein levels (Yang et al., 2007). The occasional correlations observed between synaptic proteins and α-synuclein scores in Parkinson’s disease dementia and dementia with Lewy bodies indicate that there is a potential association but more evidence is needed. There are some limitations related to the current study. Although we were able to validate our findings from the proteomic comparison study, it is possible that some important synaptic protein level changes may have been missed due to the relatively small number of patients in the LC-MS analyses, or due to post-mortem delay times. Due to the exploratory nature and the small sample size of the study, correlations with cognitive impairment are presented without adjustments for multiple comparisons; thus these findings should be interpreted with caution and require confirmation in larger samples. The Alzheimer’s disease group presented more severe cognitive impairment, with longer dementia duration time compared to dementia with Lewy bodies and Parkinson’s disease dementia, which may have influenced the findings. Another caveat to consider is that most patients had advanced disease and the levels of synaptic proteins may differ in the earlier stages of disease progression however, these changes are likely to start early on, which is supported by the association with cognitive impairment. In spite of these limitations, our results suggest that synaptic proteins have an important predictive and discriminative value in neurodegenerative disorders, which needs to be explored further. Moreover, the independent validation by antibodies of the level alterations of several synaptic proteins revealed by proteomics highlights the robustness of this method. In vivo studies using imaging and CSF are needed to explore synaptic protein changes at early disease stages. We believe that pinpointing overall alterations of synaptic proteins occurring in dementia patients brings us one step closer to a disease-specific biological target for prevention and therapeutic strategies. We anticipate their importance as a treatment target and potential as a future biomarker of disease progression for clinical trials as the therapeutic intervention window based on synaptic repair and regeneration is considerably longer than the currently used toxin-clearance approaches. Acknowledgements Human brain tissue was supplied by the Brains for Dementia Research Network comprising the MRC London Neurodegenerative Diseases Brain Bank, the Thomas Willis Oxford Brain Collection, the Newcastle Brain Tissue Resource and the University Hospital Stavanger. We would like to express our gratitude to all the donors for the tissue used in this study. Funding We would like to thank the NIHR Biomedical Research Centre for Mental Health and the NIHR Biomedical Research Unit for Dementia at King’s College London for supporting the involvement of Clive Ballard, Paul Francis and Tibor Hortobágyi in the study. The Newcastle Brain Tissue Resource is partially funded by a grant from the UK Medical Research Council (G0400074) and by Brains for Dementia research, a joint venture between Alzheimer’s Society and Alzheimer’s Research UK. We would like to gratefully acknowledge the funding that supported this research namely the KI-Astra Zeneca collaborative grant, Alzheimerfonden, Parkinsonfonden, Demensfonden, Gamla Tjanarinnor, Stohnes Stiftelse, Loo och Hans Osterman Foundation and Lindhes Advokatbyra foundation. Support from the Swedish National Infrastructure for Biological Mass Spectrometry (BioMS), funded by the Swedish Research Council, is gratefully acknowledged. Supplementary material Supplementary material is available at Brain online. Abbreviations Abbreviations GRIA glutamate receptor HiRIEF high resolution isoelectric focusing MMSE Mini-Mental State Examination PSM peptide-spectrum match TMT Tandem Mass Tag References Aarsland D , Andersen K, Larsen JP, Lolk A, Kragh-Sorensen P. Prevalence and characteristics of dementia in Parkinson disease: an 8-year prospective study. Arch Neurol  2003; 60: 387– 92. Google Scholar CrossRef Search ADS PubMed  Aarsland D , Perry R, Brown A, Larsen JP, Ballard C. Neuropathology of dementia in Parkinson's disease: a prospective, community-based study. Ann Neurol  2005; 58: 773– 6. Google Scholar CrossRef Search ADS PubMed  Abdi F , Quinn JF, Jankovic J, McIntosh M, Leverenz JB, Peskind E, et al.   Detection of biomarkers with a multiplex quantitative proteomic platform in cerebrospinal fluid of patients with neurodegenerative disorders. J Alzheimers Dis  2006; 9: 293– 348. Google Scholar CrossRef Search ADS PubMed  Bajjalieh SM , Frantz GD, Weimann JM, McConnell SK, Scheller RH. Differential expression of synaptic vesicle protein 2 (SV2) isoforms. J Neurosci  1994; 14: 5223– 35. Google Scholar PubMed  Barthelemy NR , Gabelle A, Hirtz C, Fenaille F, Sergeant N, Schraen-Maschke S, et al.   Differential mass spectrometry profiles of Tau protein in the cerebrospinal fluid of patients with Alzheimer's disease, progressive supranuclear palsy, and dementia with Lewy bodies. J Alzheimers Dis  2016; 51: 1033– 43. Google Scholar CrossRef Search ADS PubMed  Bereczki E , Bogstedt A, Höglund K, Tsitsi P, Brodin L, Ballard C, et al.   Synaptic proteins in CSF relate to Parkinson’s disease stage markers. NPJ Parkinsons Dis  2017; 3: 7. Google Scholar CrossRef Search ADS PubMed  Bereczki E , Francis PT, Howlett D, Pereira JB, Hoglund K, Bogstedt A, et al.   Synaptic proteins predict cognitive decline in Alzheimer's disease and Lewy body dementia. Alzheimers Dement  2016; 12: 1149– 58. Google Scholar CrossRef Search ADS PubMed  Biemans EA , Verhoeven-Duif NM, Gerrits J, Claassen JA, Kuiperij HB, Verbeek MM. CSF d-serine concentrations are similar in Alzheimer's disease, other dementias, and elderly controls. Neurobiol Aging  2016; 42: 213– 16. Google Scholar CrossRef Search ADS PubMed  Blennow K , Bogdanovic N, Alafuzoff I, Ekman R, Davidsson P. Synaptic pathology in Alzheimer's disease: relation to severity of dementia, but not to senile plaques, neurofibrillary tangles, or the ApoE4 allele. J Neural Transm  1996; 103: 603– 18. Google Scholar CrossRef Search ADS PubMed  Blennow K , Zetterberg H. The application of cerebrospinal fluid biomarkers in early diagnosis of Alzheimer disease. Med Clin North Am  2013; 97: 369– 76. Google Scholar CrossRef Search ADS PubMed  Branca RM , Orre LM, Johansson HJ, Granholm V, Huss M, Perez-Bercoff A, et al.   HiRIEF LC-MS enables deep proteome coverage and unbiased proteogenomics. Nat Methods  2014; 11: 59– 62. Google Scholar CrossRef Search ADS PubMed  Brinkmalm A , Brinkmalm G, Honer WG, Frolich L, Hausner L, Minthon L, et al.   SNAP-25 is a promising novel cerebrospinal fluid biomarker for synapse degeneration in Alzheimer's disease. Mol Neurodegener  2014; 9: 53. Google Scholar CrossRef Search ADS PubMed  Burton EJ , Mukaetova-Ladinska EB, Perry RH, Jaros E, Barber R, O'Brien JT. Neuropathological correlates of volumetric MRI in autopsy-confirmed Lewy body dementia. Neurobiol Aging  2012; 33: 1228– 36. Google Scholar CrossRef Search ADS PubMed  Campbell S , Stephens S, Ballard C. Dementia with Lewy bodies: clinical features and treatment. Drugs Aging  2001; 18: 397– 407. Google Scholar CrossRef Search ADS PubMed  Compta Y , Parkkinen L, O'Sullivan SS, Vandrovcova J, Holton JL, Collins C, et al.   Lewy- and Alzheimer-type pathologies in Parkinson's disease dementia: which is more important? Brain  2011; 134 (Pt 5): 1493– 505. Google Scholar CrossRef Search ADS   DeKosky ST , Scheff SW. Synapse loss in frontal cortex biopsies in Alzheimer's disease: correlation with cognitive severity. Ann Neurol  1990; 27: 457– 64. Google Scholar CrossRef Search ADS PubMed  Dun AR , Rickman C, Duncan RR. The t-SNARE complex: a close up. Cell Mol Neurobiol  2010; 30: 1321– 6. Google Scholar CrossRef Search ADS PubMed  Eden E , Navon R, Steinfeld I, Lipson D, Yakhini Z. GOrilla: a tool for discovery and visualization of enriched GO terms in ranked gene lists. BMC Bioinformatics  2009; 10: 48. Google Scholar CrossRef Search ADS PubMed  Ferrer I , Santpere G, Arzberger T, Bell J, Blanco R, Boluda S, et al.   Brain protein preservation largely depends on the postmortem storage temperature: implications for study of proteins in human neurologic diseases and management of brain banks: a BrainNet Europe Study. J Neuropathol Exp Neurol  2007; 66: 35– 46. Google Scholar CrossRef Search ADS PubMed  Folstein MF , Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res  1975; 12: 189– 98. Google Scholar CrossRef Search ADS PubMed  Fuster JM . The prefrontal cortex–an update: time is of the essence. Neuron  2001; 30: 319– 33. Google Scholar CrossRef Search ADS PubMed  Goetzl EJ , Kapogiannis D, Schwartz JB, Lobach IV, Goetzl L, Abner EL, et al.   Decreased synaptic proteins in neuronal exosomes of frontotemporal dementia and Alzheimer's disease. FASEB J  2016; 30: 4141– 8. Google Scholar CrossRef Search ADS PubMed  Gottschall PE , Ajmo JM, Eakin AK, Howell MD, Mehta H, Bailey LA. Panel of synaptic protein ELISAs for evaluating neurological phenotype. Exp Brain Res  2010; 201: 885– 93. Google Scholar CrossRef Search ADS PubMed  Halbgebauer S , Ockl P, Wirth K, Steinacker P, Otto M. Protein biomarkers in Parkinson's disease: focus on cerebrospinal fluid markers and synaptic proteins. Mov Disord  2016; 31: 848– 60. Google Scholar CrossRef Search ADS PubMed  Henley JM , Wilkinson KA. AMPA receptor trafficking and the mechanisms underlying synaptic plasticity and cognitive aging. Dialogues Clin Neurosci  2013; 15: 11– 27. Google Scholar PubMed  Honer WG . Pathology of presynaptic proteins in Alzheimer's disease: more than simple loss of terminals. Neurobiol Aging  2003; 24: 1047– 62. Google Scholar CrossRef Search ADS PubMed  Howlett DR , Whitfield D, Johnson M, Attems J, O'Brien JT, Aarsland D, et al.   Regional multiple pathology scores are associated with cognitive decline in Lewy body dementias. Brain Pathol  2015; 25: 401– 8. Google Scholar CrossRef Search ADS PubMed  Huang da W , Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc  2009; 4: 44– 57. Google Scholar CrossRef Search ADS PubMed  Jurado S , Goswami D, Zhang Y, Molina AJ, Sudhof TC, Malenka RC. LTP requires a unique postsynaptic SNARE fusion machinery. Neuron  2013; 77: 542– 58. Google Scholar CrossRef Search ADS PubMed  Kall L , Canterbury JD, Weston J, Noble WS, MacCoss MJ. Semi-supervised learning for peptide identification from shotgun proteomics datasets. Nat Methods  2007; 4: 923– 5. Google Scholar CrossRef Search ADS PubMed  Kandel ER . The molecular biology of memory storage: a dialogue between genes and synapses. Science  2001; 294: 1030– 8. Google Scholar CrossRef Search ADS PubMed  Kessner D , Chambers M, Burke R, Agus D, Mallick P. ProteoWizard: open source software for rapid proteomics tools development. Bioinformatics  2008; 24: 2534– 6. Google Scholar CrossRef Search ADS PubMed  Kim MS , Pinto SM, Getnet D, Nirujogi RS, Manda SS, Chaerkady R, et al.   A draft map of the human proteome. Nature  2014; 509: 575– 81. Google Scholar CrossRef Search ADS PubMed  Kim S , Pevzner PA. MS-GF+ makes progress towards a universal database search tool for proteomics. Nat Commun  2014; 5: 5277. Google Scholar CrossRef Search ADS PubMed  Kirvell SL , Esiri M, Francis PT. Down-regulation of vesicular glutamate transporters precedes cell loss and pathology in Alzheimer's disease. J Neurochem  2006; 98: 939– 50. Google Scholar CrossRef Search ADS PubMed  Lee CY , Lee CH, Shih CC, Liou HH. Paraquat inhibits postsynaptic AMPA receptors on dopaminergic neurons in the substantia nigra pars compacta. Biochem Pharmacol  2008; 76: 1155– 64. Google Scholar CrossRef Search ADS PubMed  Lipton SA , Rosenberg PA. Excitatory amino acids as a final common pathway for neurologic disorders. N Engl J Med  1994; 330: 613– 22. Google Scholar CrossRef Search ADS PubMed  Maalouf M , Ringman JM, Shi J. An update on the diagnosis and management of dementing conditions. Rev Neurol Dis  2011; 8: e68– 87. Google Scholar PubMed  Masliah E , Mallory M, Alford M, DeTeresa R, Hansen LA, McKeel DWJr, et al.   Altered expression of synaptic proteins occurs early during progression of Alzheimer's disease. Neurology  2001; 56: 127– 9. Google Scholar CrossRef Search ADS PubMed  Mattsson N , Carrillo MC, Dean RA, Devous MDSr, Nikolcheva T, Pesini P, et al.   Revolutionizing Alzheimer's disease and clinical trials through biomarkers. Alzheimers Dement  2015; 1: 412– 19. McKeith IG , Dickson DW, Lowe J, Emre M, O'Brien JT, Feldman H, et al.   Diagnosis and management of dementia with Lewy bodies: third report of the DLB Consortium. Neurology  2005; 65: 1863– 72. Google Scholar CrossRef Search ADS PubMed  Moya-Alvarado G , Gershoni-Emek N, Perlson E, Bronfman FC. Neurodegeneration and Alzheimer's disease (AD). What can proteomics tell us about the Alzheimer's brain? Mol Cell Proteomics  2016; 15: 409– 25. Google Scholar CrossRef Search ADS PubMed  Mukaetova-Ladinska EB , Andras A, Milne J, Abdel-All Z, Borr I, Jaros E, et al.   Synaptic proteins and choline acetyltransferase loss in visual cortex in dementia with Lewy bodies. J Neuropathol Exp Neurol  2013; 72: 53– 60. Google Scholar CrossRef Search ADS PubMed  Nicoll RA , Malenka RC. Expression mechanisms underlying NMDA receptor-dependent long-term potentiation. Ann N Y Acad Sci  1999; 868: 515– 25. Google Scholar CrossRef Search ADS PubMed  Pienaar IS , Burn D, Morris C, Dexter D. Synaptic protein alterations in Parkinson's disease. Mol Neurobiol  2012; 45: 126– 43. Google Scholar CrossRef Search ADS PubMed  Portelius E , Zetterberg H, Skillback T, Tornqvist U, Andreasson U, Trojanowski JQ, et al.   Cerebrospinal fluid neurogranin: relation to cognition and neurodegeneration in Alzheimer's disease. Brain  2015; 138 (Pt 11): 3373– 85. Google Scholar CrossRef Search ADS   Remnestal J , Just D, Mitsios N, Fredolini C, Mulder J, Schwenk JM, et al.   CSF profiling of the human brain enriched proteome reveals associations of neuromodulin and neurogranin to Alzheimer's disease. Proteomics Clin Appl  2016; 10: 1242– 53. Google Scholar CrossRef Search ADS PubMed  Rost HL , Sachsenberg T, Aiche S, Bielow C, Weisser H, Aicheler F, et al.   OpenMS: a flexible open-source software platform for mass spectrometry data analysis. Nat Methods  2016; 13: 741– 8. Google Scholar CrossRef Search ADS PubMed  Roxas BA , Li Q. Significance analysis of microarray for relative quantitation of LC/MS data in proteomics. BMC Bioinformatics  2008; 9: 187. Google Scholar CrossRef Search ADS PubMed  Sandberg A , Lindell G, Kallstrom BN, Branca RM, Danielsson KG, Dahlberg M, et al.   Tumor proteomics by multivariate analysis on individual pathway data for characterization of vulvar cancer phenotypes. Mol Cell Proteomics  2012; 11: M112.016998. Google Scholar CrossRef Search ADS PubMed  Savitski MM , Wilhelm M, Hahne H, Kuster B, Bantscheff M. A scalable approach for protein false discovery rate estimation in large proteomic data sets. Mol Cell Proteomics  2015; 14: 2394– 404. Google Scholar CrossRef Search ADS PubMed  Scheff SW , Price DA, Schmitt FA, DeKosky ST, Mufson EJ. Synaptic alterations in CA1 in mild Alzheimer disease and mild cognitive impairment. Neurology  2007; 68: 1501– 8. Google Scholar CrossRef Search ADS PubMed  Shi M , Bradner J, Bammler TK, Eaton DL, Zhang J, Ye Z, et al.   Identification of glutathione S-transferase pi as a protein involved in Parkinson disease progression. Am J Pathol  2009; 175: 54– 65. Google Scholar CrossRef Search ADS PubMed  Sidhu A , Wersinger C, Vernier P. Does alpha-synuclein modulate dopaminergic synaptic content and tone at the synapse? FASEB J  2004; 18: 637– 47. Google Scholar CrossRef Search ADS PubMed  Sjogren M , Davidsson P, Gottfries J, Vanderstichele H, Edman A, Vanmechelen E, et al.   The cerebrospinal fluid levels of tau, growth-associated protein-43 and soluble amyloid precursor protein correlate in Alzheimer's disease, reflecting a common pathophysiological process. Dement Geriatr Cogn Disord  2001; 12: 257– 64. Google Scholar CrossRef Search ADS PubMed  Sjogren M , Minthon L, Davidsson P, Granerus AK, Clarberg A, Vanderstichele H, et al.   CSF levels of tau, beta-amyloid(1-42) and GAP-43 in frontotemporal dementia, other types of dementia and normal aging. J Neural Transm  2000; 107: 563– 79. Google Scholar CrossRef Search ADS PubMed  Sudhof TC , Rothman JE. Membrane fusion: grappling with SNARE and SM proteins. Science  2009; 323: 474– 7. Google Scholar CrossRef Search ADS PubMed  Terry RD , Masliah E, Salmon DP, Butters N, DeTeresa R, Hill R, et al.   Physical basis of cognitive alterations in Alzheimer's disease: synapse loss is the major correlate of cognitive impairment. Ann Neurol  1991; 30: 572– 80. Google Scholar CrossRef Search ADS PubMed  Thevenon J , Souchay C, Seabold GK, Dygai-Cochet I, Callier P, Gay S, et al.   Heterozygous deletion of the LRFN2 gene is associated with working memory deficits. Eur J Hum Genet  2016; 24: 911– 18. Google Scholar CrossRef Search ADS PubMed  Vallortigara J , Rangarajan S, Whitfield D, Alghamdi A, Howlett D, Hortobagyi T, et al.   Dynamin1 concentration in the prefrontal cortex is associated with cognitive impairment in Lewy body dementia. F1000Res  2014; 3: 108. Google Scholar PubMed  Wellington H , Paterson RW, Portelius E, Tornqvist U, Magdalinou N, Fox NC, et al.   Increased CSF neurogranin concentration is specific to Alzheimer disease. Neurology  2016; 86: 829– 35. Google Scholar CrossRef Search ADS PubMed  Whitfield DR , Vallortigara J, Alghamdi A, Howlett D, Hortobagyi T, Johnson M, et al.   Assessment of ZnT3 and PSD95 protein levels in Lewy body dementias and Alzheimer's disease: association with cognitive impairment. Neurobiol Aging  2014; 35: 2836– 44. Google Scholar CrossRef Search ADS PubMed  Wimo A , Guerchet M, Ali GC, Wu YT, Prina AM, Winblad B, et al.   The worldwide costs of dementia 2015 and comparisons with 2010. Alzheimers Dement  2017; 13: 1– 7. Google Scholar CrossRef Search ADS PubMed  Wisniewski JR , Zougman A, Nagaraj N, Mann M. Universal sample preparation method for proteome analysis. Nat Methods  2009; 6: 359– 62. Google Scholar CrossRef Search ADS PubMed  Xu J , Mashimo T, Sudhof TC. Synaptotagmin-1, -2, and -9: Ca(2+) sensors for fast release that specify distinct presynaptic properties in subsets of neurons. Neuron  2007; 54: 567– 81. Google Scholar CrossRef Search ADS PubMed  Yang W , Woltjer RL, Sokal I, Pan C, Wang Y, Brodey M, et al.   Quantitative proteomics identifies surfactant-resistant alpha-synuclein in cerebral cortex of Parkinsonism-dementia complex of Guam but not Alzheimer's disease or progressive supranuclear palsy. Am J Pathol  2007; 171: 993– 1002. Google Scholar CrossRef Search ADS PubMed  © The Author(s) (2018). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: journals.permissions@oup.com
TI - Synaptic markers of cognitive decline in neurodegenerative diseases: a proteomic approach
JF - Brain
DO - 10.1093/brain/awx352
DA - 2018-02-01
UR - https://www.deepdyve.com/lp/oxford-university-press/synaptic-markers-of-cognitive-decline-in-neurodegenerative-diseases-a-jZzTpCmKJn
SP - 582
EP - 595
VL - 141
IS - 2
DP - DeepDyve
ER -