Altered Synapse Stability in the Early Stages of Tauopathy

Abstract

Synapse loss is a key feature of dementia, but it is unclear whether synaptic dysfunction precedes degenerative phases of the disease. Here, we show that even before any decrease in synapse density, there is abnormal turnover of cortical axonal boutons and dendritic spines in a mouse model of tauopathy-associated dementia. Strikingly, tauopathy drives a mismatch in synapse turnover; postsynaptic spines turn over more rapidly, whereas presynaptic boutons are stabilized. This imbalance between pre- and post-synaptic stability coincides with reduced synaptically driven neuronal activity in pre-degenerative stages of the disease.

Keywords: 2-photon microscopy; axon; bouton; cortex; dementia; dendritic spine; neurodegeneration.

Graphical abstract

Figure 1

Progressive Simultaneous Loss of Pre- and Post-synaptic Components in rTg4510 Mice (A and B) Dendrites (A) and axons (B) bearing TBs were repeatedly imaged at weekly intervals in WT and rTg4510 animals. Scale bar, 5μm. In some image panels, for clarity of display, fluorescence not associated with main neurite was removed. (C) There is a progressive decrease in spine density with age in rTg4510 dendrites relative to WT (mean ± SEM; early, F_(1,54) = 0.31, p = 0.580, WT: n = 39 dendrites/5 animals, rTg4510 n = 16/4; mid, F_(1,64) = 2.07, p = 0.169, WT: n = 8/3, rTg4510 n = 10/3; late, F_(1,50) = 44.09, p < 0.001, WT n = 15/4, rTg4510 n = 12/3). (D) Quantification of TB loss in the same three batches of animals shows that loss of axonal TBs follows a similar age-dependent decrease relative to WT (mean ± SEM; early, F_(1,29) = 0.12, p = 0.728, WT:n = 16/3, rTg4510: n = 7/3; mid, F_(1,54) = 1.92; p = 0.183; WT: n = 13/3, rTg4510: n = 14/3; late, F_(1,57) = 10.76; p = 0.004; WT: n = 13/4, rTg4510 n = 8/4). Two-way repeated-measures ANOVA. Error bars represent mean ± SEM.

Figure 2

Dissociated Turnover and Altered Stability of Dendritic Spines and Axonal Boutons (A) rTg4510 dendrites had a significantly increased TOR of dendritic spines at mid and late time points driven by a significant increase in lost spines and gained spines at the mid and late time points respectively (mean ± SEM; p = 0.003; mid, WT: n = 16/4, rTg4510 n = 17/4; late, WT: n = 17/4, rTg4510 n = 8/3). (B) rTg4510 axonal TBs showed a significantly reduced turnover at both mid and late time points due to decreases in both gains and losses of TB (p = 0.003; mid, WT: n = 17/3, rTg4510 n = 9/3; late, WT: n = 13/4, rTg4510 n = 9/4). (C) TBs and spines from the same WT individuals showed a balanced turnover (p = 0.786; n = 7 animals). This association was lost in rTg4510s where dendritic spines had a significantly increased TOR compared to axonal boutons (p = 0.007; n = 4 animals). (D) Survival of dendritic spines from the first to last imaging session is decreased in rTg4510s in the late group (F_(1,41) = 14.87; p < 0.001; WT: n = 17/4, rTg4510: n = 8/3), but not at the mid time points (F_(1,46) = 5.90; p = 0.053; WT: n = 16/4, rTg4510: n = 17/4). (E) The survival fraction of rTg4510 axonal TBs was increased in the mid group (F_(1,46) = 5.90; p = 0.023; WT: n = 16/3, rTg4510: n = 8/3) and in the late group (F_(1,36) = 6.60; p = 0.019; WT: n = 14/4, rTg4510: n = 6/3). (F) The relative proportions of persistent TBs and spines is shifted in rTg4510 animals (p = 0.054; n = 7 animals/group) compared to WT (p = 0.04; n = 4 animals/group). Error bars represent mean ± SEM. (G) Boxplots showing decreased spine head diameter for rTg4510 compared to WT for each imaging week (box, 25^th/75^th percentile; line, median; whiskers, full range; WT n = 4,805 spines, 4 animals, TG n = 3,516 spines, 4 animals). (H) Boxplots showing no change in TB head diameter for rTg4510 compared to WT (box, 25^th/75^th percentile; line, median; whiskers, full range; WT n = 1,608 TBs, 4 animals, TG n = 410 TBs, 4 animals). Two-way repeated-measures ANOVA (A, B, D, and E), Student’s unpaired t test (C and F), and linear mixed model (G and H) were used; ^∗p < 0.05, ^∗∗p < 0.01.

Figure 3

Decreased Cortical Neuronal Activity and Aberrant Stimulus Encoding in rTg4510 Mice (A) Image shows representative field of view showing GCaMP6m fluorescence in layer 2 neurons of barrel cortex (R, rostral; M, medial). Right: example GCaMP6m fluorescence traces (ΔF/F) for five randomly selected neurons. Red stars denote the detection of spontaneous GCaMP6m transients. (B) Percentages of WT (n = 268, 3 mice, 89 ± 14 per mouse [mean ± SEM]) and rTg4510 (n = 233, 3 mice, 78 ± 6 per mouse [mean ± SEM]) neurons with isolated GCaMP6m transients (p = 1e-5, χ²₍₁₎ = 19.32; WT = 62/268 neurons, rTg4510 = 20/233 neurons). (C) Cumulative frequency and inset bar graph (mean ± SEM) illustrating the frequency of spontaneous GCaMP6m transients in active WT (n = 62) and rTg4510 (n = 20) neurons (p = 0.005, rank-sum test; WT = 0.39 ± 0.05 events per min, rTg4510 = 0.16 ± 0.02 events per min). (D) Cumulative frequency and inset bar graphs (mean ± SEM) illustrating the amplitudes of spontaneous GCaMP6m transients in WT (n = 208 transients, 62 cells) and rTg4510 (n = 27 transients, 20 cells) neurons (p = 0.001, rank-sum test; WT = 1.32 ± 0.06 ΔF/F, rTg4510 = 0.88 ± 0.12 ΔF/F). (E) Representative GCaMP6m transients showing single trial responses to contralateral principal whisker stimulation (10 Hz, 1 s; 25 trials; red dashes). (F) Percentages of WT (n = 268 neurons, three mice) and rTg4510 (n = 233 neurons, three mice) neurons responding to whisker stimulation (p = 0.03, χ²₍₁₎ = 4.62; WT = 8/268 neurons, rTg4510 = 1/233 neurons).