Amyloidosis and tauopathy progressively disrupt near-critical dynamics in selectively vulnerable circuits

Abstract

Mechanistically distinct neurodegenerative diseases result in similar cognitive deficits, and pathology is often concentrated in a set of circuits known to be selectively vulnerable. Selective vulnerability is an intrinsic property of a circuit: in APP/PS1 mice, where plaques are global, functional disruptions are confined to circuits within the default mode network. What characterizes the dynamics of vulnerable circuits from those of spared circuits is unknown. We asked whether the distinguishing property is the proximity of population dynamics to criticality, a homeostatic set-point of network activity that determines computational capacity. We recorded single units simultaneously in three selectively vulnerable regions (retrosplenial cortex, anterior cingulate cortex, and CA1 hippocampus) and in primary visual cortex of freely behaving young and old APP/PS1 and wild-type mice. Firing rate and spike-time variability differed by genotype inconsistently across regions and ages. In contrast, three measures from criticality theory each revealed deficits that emerged only with age and only in the vulnerable regions. No measure detected a progressive change in V1. Avalanche scaling is not fixed in neural activity, but varies along a scaling line as the network state changes. As a result, the lines generated by each genotype revealed a constitutive effect of the transgene in young animals, which inverted with age in amyloid-bearing animals, but only in vulnerable circuits. If the destabilization of near-critical dynamics directly reflects a circuit's functional capacity, the same signature should identify a mechanistically distinct degenerative disease. In a model of tauopathy (P301S/E4), CA1 dynamics were displaced from the control scaling line with the same age-dependent reversal observed in APP/PS1. Our results suggest a network-level principle of neurodegenerative diseases: molecularly unrelated pathologies converge on the loss of near-critical dynamics in selectively vulnerable circuits.