Relative Contributions of Specific Activity Histories and Spontaneous Processes to Size Remodeling of Glutamatergic Synapses

The idea that synaptic properties are defined by specific pre- and postsynaptic activity histories is one of the oldest and most influential tenets of contemporary neuroscience. Recent studies also indicate, however, that synaptic properties often change spontaneously, even in the absence of specific activity patterns or any activity whatsoever. What, then, are the relative contributions of activity history-dependent and activity history-independent processes to changes synapses undergo? To compare the relative contributions of these processes, we imaged, in spontaneously active networks of cortical neurons, glutamatergic synapses formed between the same axons and neurons or dendrites under the assumption that their similar activity histories should result in similar size changes over timescales of days. The size covariance of such commonly innervated (CI) synapses was then compared to that of synapses formed by different axons (non-CI synapses) that differed in their activity histories. We found that the size covariance of CI synapses was greater than that of non-CI synapses; yet overall size covariance of CI synapses was rather modest. Moreover, momentary and time-averaged sizes of CI synapses correlated rather poorly, in perfect agreement with published electron microscopy-based measurements of mouse cortex synapses. A conservative estimate suggested that ~40% of the observed size remodeling was attributable to specific activity histories, whereas ~10% and ~50% were attributable to cell-wide and spontaneous, synapse-autonomous processes, respectively. These findings demonstrate that histories of naturally occurring activity patterns can direct glutamatergic synapse remodeling but also suggest that the contributions of spontaneous, possibly stochastic, processes are at least as great.Contrary to expectations, specific activity histories account for less than half of the remodeling exhibited by individual glutamatergic synapses in spontaneously active networks of cortical neurons.Author Summary: The modification of synaptic connections by specific activity histories (a phenomenon known as synaptic plasticity) is widely believed to represent a major substrate of processes collectively referred to as learning and memory. Recent studies indicate, however, that synapses also change spontaneously, even in the absence of specific activity histories—or, for that matter, any activity whatsoever. This raises a fundamental question: how do changes directed by specific activity histories quantitatively compare to spontaneous changes in synaptic properties? Put differently—what is the “signal-to-noise ratio” of synaptic plasticity at individual synapses? To address this question we followed—over several days—pairs of synapses formed between the same neurons under the assumption that their common activity histories should drive similar changes in their sizes. Indeed, sizes of such synapses tended to change in a correlated manner; yet the extent of this correlation was surprisingly modest, accounting for less than half of the changes that such synapses exhibited. Moreover, sizes of synapses with apparently common activity histories tended to be quite different. Our findings thus indicate that the “signal-to-noise ratio” of synapse remodeling might be rather poor, on the order of 1:1 or less.