Untangling the association of amyloid-β and tau with synaptic and axonal loss in Alzheimer's disease

Abstract

It is currently unclear how amyloid-β and tau deposition are linked to changes in synaptic function and axonal structure over the course of Alzheimer's disease. Here, we assessed these relationships by measuring presynaptic (synaptosomal-associated protein 25, SNAP25; growth-associated protein 43, GAP43), postsynaptic (neurogranin, NRGN) and axonal (neurofilament light chain) markers in the CSF of individuals with varying levels of amyloid-β and tau pathology based on 18F-flutemetamol PET and 18F-flortaucipir PET. In addition, we explored the relationships between synaptic and axonal markers with cognition as well as functional and anatomical brain connectivity markers derived from resting-state functional MRI and diffusion tensor imaging. We found that the presynaptic and postsynaptic markers SNAP25, GAP43 and NRGN are elevated in early Alzheimer's disease i.e. in amyloid-β-positive individuals without evidence of tau pathology. These markers were associated with greater amyloid-β pathology, worse memory and functional changes in the default mode network. In contrast, neurofilament light chain was abnormal in later disease stages, i.e. in individuals with both amyloid-β and tau pathology, and correlated with more tau and worse global cognition. Altogether, these findings support the hypothesis that amyloid-β and tau might have differential downstream effects on synaptic and axonal function in a stage-dependent manner, with amyloid-related synaptic changes occurring first, followed by tau-related axonal degeneration.

Keywords: MRI; PET; amyloid-β; neurofilament; neurogranin; tau PET.

Figure 1

Synaptic and axonal markers change in earlier and later Alzheimer’s disease stages. (A) Representation of the neuronal aspects measured by the postsynaptic [NRGN (Ng)], axonal (NfL) and presynaptic markers (SNAP25; GAP43). (B) In early Alzheimer’s disease stages, SNAP25, GAP43 and NRGN show significant increases in amyloid-β+ (Aβ+) compared to amyloid-β− (Aβ−) individuals without evidence of tau PET retention (Tau−). (C) In later Alzheimer’s disease stages, NfL shows significant increases in Tau+ compared to Tau− individuals who are all amyloid-β+. (D) The increases in NfL in the previous groups can still be observed when only cognitively impaired (CI) individuals are considered. These analyses were performed using t-tests, while controlling for covariates. The values presented in the plots are the residuals of SNAP25, GAP43, NRGN and NfL after regressing out the effects of age (all CSF markers) in addition to sex (NfL).

Figure 2

Synaptic degeneration and axonal damage correlate with high amyloid-β and tau. (A) Among the synaptic and axonal markers, increasing NRGN (Ng) correlates best with greater amyloid-β PET retention, i.e. in individuals who are amyloid− but do not show evidence of tau pathology. (B) In contrast, higher NfL correlates best with greater tau PET retention in later Alzheimer’s disease stages, i.e. in individuals who are Tau− and Tau+ with amyloid-β pathology. The analyses were carried out using linear regression models. The values presented in the plots are the residuals of NRGN and NfL after regressing out the effects of age (all markers) in addition to sex (NfL). AD = Alzheimer’s disease.

Figure 3

Memory and global cognition are associated with synaptic and axonal markers in a stage-dependent manner. (A) Memory measured by the word list recall test (ADAS-Cog) correlates with GAP43 and NRGN levels in early Alzheimer’s disease stages (amyloid-β+Tau− and amyloid-β−Tau−), whereas (B) global cognition measured by the Mini-Mental State Examination correlates with NfL in later Alzheimer’s disease stages (amyloid-β+Tau+ and amyloid-β+Tau−). The correlation analyses were performed using Spearman’s ρ. These results are in line with findings showing memory deficits in early Alzheimer’s disease and global cognitive impairment in later disease stages. The values presented in the plots are the residuals of GAP43, NRGN and NfL after regressing out the effects of age, sex and education. AD = Alzheimer’s disease.

Figure 4

Synaptic loss predicts functional connectivity changes in the default-mode networks. In earlier Alzheimer’s disease stages (amyloid-β+Tau− and amyloid-β−Tau−), our linear regression models show that NRGN (Ng) is the marker that correlates best with connectivity increases in the anterior default-mode network (A) and connectivity decreases in the posterior default-mode network (B). The values presented in the plots are the residuals of functional MRI (fMRI) signals and NRGN after regressing out the effects of age and sex. AD = Alzheimer’s disease.

Figure 5

Tau pathology correlates with structural connectivity changes in the parahippocampal tract. In later Alzheimer’s disease stages (amyloid-β+Tau+ and amyloid-β+Tau−), our linear regression models show that tau PET retention is the measure that correlates best with reduced fractional anisotropy (FA) in the parahippocampal tract (left). The values presented in the plots are the residuals of fractional anisotropy values and tau PET levels after regressing out the effects of age and sex.

Figure 6

Schematic hypothetical representation of the downstream effects of amyloid-β and tau on synaptic and axonal function in earlier and later Alzheimer’s disease stages. In earlier disease stages (amyloid-β+Tau− and amyloid-β−Tau−), the accumulation of amyloid-β into plaques reduces the number of postsynaptic spines and the number of presynaptic vesicles, altering local synaptic communication. In later disease stages (amyloid-β+Tau+ and amyloid-β+Tau−), as synaptic changes become more pronounced, tau becomes phosphorylated and aggregates into insoluble paired helical filaments (PHF), which form neuropil threads and neurofibrillary tangles. These changes are followed by the dissociation of neurofilaments from microtubules, axonal degeneration and ultimately global cognitive impairment. Aβ = amyloid-β; AD = Alzheimer’s disease.