Proteomics for synaptic markers of cognitive decline in neurodegenerative diseases

This scientific commentary refers to ‘Synaptic markers of cognitive decline in neurodegenerative diseases: a proteomic approach’ by Bereczki et al. (doi:10.1093/brain/awx352). Loss of synapses and a decrease in synaptic proteins are constant features of neurodegenerative disorders associated with dementia, including Alzheimer’s disease, Parkinson’s disease dementia (PDD), dementia with Lewy bodies (DLB) and frontotemporal lobe degeneration. These signs of synaptic dysfunction have been shown to be better correlates of cognitive impairment than the protein aggregates that are considered pathological hallmarks of most proteinopathies. In Alzheimer’s disease, synapse loss was identified as a major correlate of cognitive impairment by the late Bob Terry (Terry et al., 1991). This relationship has since been confirmed in many other neurodegenerative diseases using (quantitative) ultrastructural and molecular biochemical methods. More recently, quantitative proteomic analyses of brain tissue, as well as blood and CSF, have provided insights into basic physiological processes underlying morphological changes in both normal ageing and disease. In this issue of Brain, Bereczki and co-workers expand upon these data by presenting the first in-depth quantitative proteomic study in a series of neurodegenerative diseases associated with dementia (Bereczki et al., 2018). In Alzheimer’s disease, PDD and DLB, pilot proteomic studies have correlated changes in synaptic proteins with neuropathological markers (Bereczki et al., 2016) and with exposure to soluble N-terminally truncated tau isoforms that cause alterations in synaptic activity (Florenzano et al., 2017). Neocortical synaptic dysfunction is probably caused at least in part by presynaptic deposition of pathological proteins, such as toxic α-synuclein oligomers (Rockenstein et al., 2014), leading to alterations in the concentrations of other synaptic proteins. These pathomechanisms result in synapse loss, which is followed by impaired axonal transport and finally neuronal demise and functional disconnection (Bellucci et al., 2016). Downregulation of the presynaptic protein Rab3A and the postsynaptic proteins synaptopodin and neurogranin—suggesting defective synaptic transmission—has been reported in DLB and PDD (Bereczki et al., 2016). The largely similar proteome of these conditions matched for amyloid burden suggests that variations in concomitant Alzheimer’s disease pathology and protein–protein or other molecular interactions may contribute to the observed clinical differences between the two (Datta et al., 2017). Synaptic dysfunction with a decrease in neurogranin may be an early pathological process in neurodegeneration and a sensitive marker of age-related cognitive impairment (Casaletto et al., 2017). Varying reductions in RPT6 protein—a component of the proteasome—and in peptidyl-glutamyl peptide-hydrolysing and chymotrypsin-like proteasome activity have also been described in the cerebellar cortex of patients with PDD and DLB (Alghamdi et al., 2017). These changes, all associated with cognitive impairment, may reflect differing involvement of the ubiquitin proteasome system dependent on disease severity, but the causative mechanisms are still unclear. Using RNA sequencing, protein expression analysis, and histological evaluation, we recently found decreased levels of proteins related to synapse efficiency, in post-mortem tissues of the pulvinar nuclei of the thalamus in DLB. Of note, these changes were observed in a region typically only mildly affected by Lewy body pathology; however, our findings suggest that they contribute to cognitive changes in DLB, further strengthening the evidence for synaptic dysfunction in this disorder (Erskine et al., unpublished results). Bereczki et al. used modern proteomic methods to assess prefrontal post-mortem cortical tissues (Brodmann area 9) from 92 cases of autopsy-confirmed PDD (n = 24), DLB (n = 26), Alzheimer’s disease (n = 18) and age-matched non-neurological controls (n = 24) (Bereczki et al., 2018). Cognitive data were available for the majority of cases (Mini-Mental State Examination scores 1–2 years prior to death), and the severity of amyloid-β, phospho-tau and α-synuclein pathology was assessed semi-quantitatively. Proteomic studies were performed using liquid chromatography-mass spectrometry (LS-MS), gene analogy and pathway enrichment analysis, western blotting, sandwich enzyme-like immunosorbent assays, immunoblotting and multivariant data analysis. A total of 1035 proteins were identified, of which 851 were synaptic proteins. In all three disorders, the synaptic protein with the largest reduction was LRFN2. In PDD, concentrations of SNAP47 and SYT2 were decreased compared to controls, while GAP43 was decreased in comparison to Alzheimer’s disease. In DLB, five of eight synaptic proteins were decreased (LRFN2, SNAP47, SV2C, SYBU, SYT2) compared to non-neurological controls, while in Alzheimer’s disease, only LRFN2 and GRIA3 concentrations were decreased. A clear discrimination between PDD and Alzheimer’s disease was observed (77.8% sensitivity, 80% specificity), as well as between DLB and controls, and Alzheimer’s disease versus controls. On the other hand, no significant changes were detected in Rab3 or SNAP25, possibly due to post-mortem changes. Significant negative correlations were seen between α-synuclein and SNAP47 and GRIA3 in PDD, and between α-synuclein and SV2C in DLB. GRIA4 was negatively correlated with amyloid-β in both PDD and DLB. The only significant negative correlation between a synaptic protein and tangle scores was that between GRIA3 and tangle scores in PDD. Significant associations between reduced levels of synaptic proteins and cognitive decline were found in PDD, DLB and Alzheimer’s disease, whereas no associations were found between motor symptoms and synaptic proteins in PDD. Cognitive impairment and the rate of cognitive decline significantly correlated with loss of a number of synaptic proteins (SNAP47, SYBU, LRFN2, SV2C, and GRIA3). While levels of single synaptic proteins alone could not discriminate between groups, multivariate analysis using a panel of synaptic proteins could distinguish between controls and different disease groups, as well as between Alzheimer’s disease and PDD. Of note, no such differences were seen between DLB and either Alzheimer’s disease or PDD. This supports the assumption that despite the obvious similarities between DLB and PDD (i.e. α-synuclein pathology and Alzheimer’s disease-related lesions) the two diseases do not share the exact same pathomechanisms. Furthermore, neuropathological studies report a significantly higher prevalence of Alzheimer pathology in DLB compared to PDD, as well as the absence of clinical parkinsonism in up to 40% of DLB cases (Ferman et al., 2011). Proteomics will certainly be helpful in identifying similarities and differences between PDD and DLB. This is of utmost interest in view of the renewed discussion about the nosological positions of the two conditions—whether they are one or two separate diseases or phenotypes of an α-synuclein-associated spectrum of Lewy body diseases sharing genetic risk factors (Fields, 2017; Jellinger and Korczyn, unpublished results). In conclusion, proteomic assessment of synaptic proteins allowed reliable discrimination of PDD from Alzheimer’s disease, and of PDD, DLB and Alzheimer’s disease from neurologically unimpaired controls, with high specificity and sensitivity. The work by Bereczki and colleagues presented in this issue of Brain indicates that proteomics may be ideally suited to addressing the role of synaptic dysfunction in the pathogenesis of neurodegeneration, as well as to identifying novel targets for early disease intervention.