Rate, Timing, and Cooperativity Jointly Determine Cortical Synaptic Plasticity

Abstract

Cortical long-term plasticity depends on firing rate, spike timing, and cooperativity among inputs, but how these factors interact during realistic patterns of activity is unknown. Here we monitored plasticity while systematically varying the rate, spike timing, and number of coincident afferents. These experiments demonstrate a novel form of cooperativity operating even when postsynaptic firing is evoked by current injection, and reveal a complex dependence of LTP and LTD on rate and timing. Based on these data, we constructed and tested three quantitative models of cortical plasticity. One of these models, in which spike-timing relationships causing LTP "win" out over those favoring LTD, closely fits the data and accurately predicts the build-up of plasticity during random firing. This provides a quantitative framework for predicting the impact of in vivo firing patterns on synaptic strength.

Complexity of calcium signaling in synaptic spines

“…2). This finding is robust and has been found in neurons from the hippocampus using acute, (23) organotypic slices, (31) or dissociated cell culture, (32) and in layers V (33,34) and II/III (35) of the cerebral cortex, as well as in vivo preparations using cat (36,37) and Xenopus (38) visual systems. The temporal window permissive for LTP induction is 10-20 ms (milliseconds) whereas the window in which the presynaptic spike can follow the postsynaptic spike in order to induce LTD ranges from 20 to 100 ms in different preparations.…”

“…For example, LTP induction by prebefore-post correlated pairing depends on the frequency of pairing. (33,34) Similarly, the increase in [Ca 2þ ] i in cerebellar Purkinje neurons following a train of action potentials is highly non-linear, with small increases in [Ca 2þ ] i resulting from initial spikes but very large increases in [Ca 2þ ] i towards the end of the train. (85) The authors show that the amount of Ca 2þ influx following each spike is constant, and that the difference in [Ca 2þ ] i is due to saturation of the buffering capacity.…”

“…Interestingly, 10 Hz is the lowest pairing frequency at which a small number of positively correlated pairings will reliably induce LTP. (33,34) If the frequency at which pairing becomes cooperative is due to saturation of the buffering capacity of the spine, then changing total buffering capacity should modulate this frequency dependence. Indeed, halving and doubling the CBP concentration resulted in increased and decreased frequency-dependence and CaM-4 levels, respectively.…”

“…Note that Sjö strö m and colleagues were able to induce LTP with low frequency pairing when the soma was mildly depolarized. (34) Although the chemical consequences at the synapse remain unknown, it is tempting to imagine that this depolarization leads to calcium spikes in dendritic branchlets, (87,88) such that the frequency-dependence of CaM activation is dramatically altered by an additional mode of Ca 2þ entry. Future direction Clearly, this model does yet not explain the differences between the intracellular Ca 2þ dynamics required for the induction of LTP and LTD, or the critical time-window of STDP.…”

Time scales of memory, learning, and plasticity

“…However, the generality of STDP is heavily debated (Lisman andSpruston 2005, 2010;Shouval et al 2010). Experiments show that the temporal order is only important in a small regime of presynaptic activation (Sjöström et al 2001) and, furthermore, that synaptic modifications seem to be independent of the spiking of the postsynaptic cell (Golding et al 2002). Thus, alternative models have been developed, for instance, calcium-based plasticity (Lisman 1989;Shouval et al 2002;Yeung et al 2004;Graupner and Brunel 2012).…”