A dynamical role for acetylcholine in synaptic renormalization

Abstract

Although sleep is a fundamental behavior observed in virtually all animal species, its functions remain unclear. One leading proposal, known as the synaptic renormalization hypothesis, suggests that sleep is necessary to counteract a global strengthening of synapses that occurs during wakefulness. Evidence for sleep-dependent synaptic downscaling (or synaptic renormalization) has been observed experimentally, but the physiological mechanisms which generate this phenomenon are unknown. In this study, we propose that changes in neuronal membrane excitability induced by acetylcholine may provide a dynamical mechanism for both wake-dependent synaptic upscaling and sleep-dependent downscaling. We show in silico that cholinergically-induced changes in network firing patterns alter overall network synaptic potentiation when synaptic strengths evolve through spike-timing dependent plasticity mechanisms. Specifically, network synaptic potentiation increases dramatically with high cholinergic concentration and decreases dramatically with low levels of acetylcholine. We demonstrate that this phenomenon is robust across variation of many different network parameters.

Figure 1. Effects of acetylcholine on phase response curves, network synchrony, and overall network synaptic potentiation in 1000-cell cortical neuronal network models.

(a,b) Phase response curves of individual neurons with (a) high simulated ACh concentration and (b) low simulated ACh concentration. (c, d) Raster plots of the activity of a model cortical network with (c) high and (d) low ACh concentration. Blue (Red) dots represent spikes of excitatory (inhibitory) neurons. Note the higher synchronization in the network with low cholinergic modulation compared to the network with high cholinergic modulation. (e) Average final distributions of synaptic strengths for a typical high-ACh network, with a network potentiation value of . (f) Average final distribution of synaptic strengths for a typical low-ACh network. This distribution constitutes a much lower network potentiation value () due to a greater proportion of synapses with zero synaptic strength values. In panels (c)–(f), the re-wiring probability was 0.60 and . Panels (e) and (f) represent histograms averaged over ten different network initializations.

Figure 2. Effects of acetylcholine on network potentiation and synchronization for varied network parameters with an additive STDP rule.

(a,b) Network potentiation as a function of re-wiring probability (controlling randomness of network connections, x-axis) and maximum synaptic strength (, y-axis) for model cortical networks both with (a) high and (b) low simulated cholinergic modulation. Note the much greater potentiation of high-ACh networks for virtually all network parameters, and especially for . (c,d) Network synchrony, as measured by mean phase coherence, as a function of re-wiring probability and for networks with (c) high and (d) low simulated cholinergic modulation. All results represent averages over ten randomly-initialized network simulations. Arrows indicate network parameters which gave rise to panels c, d, e, and f in Fig. 1.

Figure 3. Structure of neuronal firing of pre- and post-synaptic cell pairs in high-ACh and low-ACh cortical networks.

(a,b) Spike-timing histogram of phases of post-synaptic cell firing relative to pre-synaptic cell firing in the model cortical network both with (a) high and (b) low simulated cholinergic modulation. These plots were constructed by averaging the spike-timing histograms of all pre-post pairs throughout the entire network. (a) In high-ACh networks, post-synaptic cells were much more likely to fire shortly after (as opposed to shortly before) pre-synaptic spikes, as evidenced by the fact that the cumulative probability of firing within the interval (0.30) was substantially larger than the cumulative probability of firing within the interval (0.22). (b) In low-ACh networks, post-synaptic spike timings were more balanced between shortly preceding and shortly succeeding pre-synaptic spikes, leading to much lower network potentiation via the STDP rule. Both histograms were computed from simulations in which the re-wiring probability was 0.60 and . Note the different scales on the y-axes.

Figure 4. Effects of noise amplitude on the difference in network potentiation between networks with high and low cholinergic modulation.

(a,b) Network potentiation as a function of re-wiring probability (x-axis) and maximum synaptic strength (y-axis) for networks with (a) high and (b) low cholinergic modulation, with noise amplitude fixed at . Note that high-ACh networks exhibited much greater potentiation than low-ACh networks for . (c) Difference in network potentiation between high- and low-ACh networks as a function of noise amplitude for the network parameters indicated by arrows in panels (a) and (b).

Figure 5. Effects of varying the slow potassium conductance,

, upon network potentiation and PRC. (a,b) Examples of network potentiation as a function of and re-wiring probability for and . (c) Network potentiation and synchronization (as measured by mean phase coherence) as a function of for the network parameters indicated in (a) and (b). (d) Phase response curves corresponding to and .

Figure 6. Effects of connectivity density upon network potentiation.

(a,b) Network potentiation of high-ACh and low-ACh networks with 4.0% connectivity density. Network potentiation is displayed as a function of and re-wiring probability, as in Fig. 2. Note the difference in scale between these plots and Fig. 2. (c) Difference between high-ACh and low-ACh network potentiation values as a function of connectivity density for networks with parameters analogous to those indicated by arrows in panels (a) and (b). In order to investigate similar regimes of network excitability, we decreased in proportion to the increase in connectivity density.

Figure 7. Effects of the modulation of the STDP window,

, upon network potentiation. (a,b) Network potentiation of high-ACh and low-ACh networks as a function of and re-wiring probability for . (c,d) Network potentiation of high-ACh and low-ACh networks as a function of and re-wiring probability for . (e) Network potentiation of both high-ACh and low-ACh networks as a function of , with and a re-wiring probability of 0.60.

Figure 8. Effects of acetylcholine on network potentiation (a,b) and synchronization (c,d) for varied network parameters with an asymmetric STDP rule that favors LTD over LTP.

STDP parameters were , , , and .

Figure 9. Effects of acetylcholine on network potentiation (a,b) and synchronization (e,f) for varied network parameters with a multiplicate (weight-dependent) STDP rule.

As found in previous studies, the distribution of synaptic weights is not bimodal (c,d). Note the difference in scale between network potentation plots for the multiplicative STDP rule (a,b) versus the additive STDP rule (Fig. 2a,b). In both cases, high ACh concentration results in significantly greater network potentiation than low ACh concentration.

Figure 10. Effects of alternately switching between high and low levels of acetylcholine in a cortical network with an embedded cluster.

(a) Network potentiation and synchronization (as measured by mean phase coherence) of the cortical network as a function of time as the level of ACh was alternated between high and low levels (different intervals are demarcated by dashed lines). (b) Distribution of synaptic strength values at the end of the last high-ACh interval. (c) Representative raster plot of network activity during the last high-ACh interval. The first 50 neurons comprise the cluster. (d) Distribution of synaptic strength values at the end of the last low-ACh interval. Note how the number of connections whose synaptic strength went to 0 is greater than the number that went to . (e) Representative raster plot of network activity during the last low-ACh interval. Note how the tight bursting of the cluster drove activity in the rest of the network. (f) Network potentiation computed from distributions of synaptic weights for all synaptic connections (heavy blue curve, as shown in (a)), for synapses originating in the cluster and projecting outside the cluster (green curve), and for synaptic connections originating outside the cluster and projecting to the cluster (light blue curve). During the low-ACh intervals, the connections originating outside the cluster and projecting to the cluster showed extreme relative depotentiation due to the driving of the rest of the network by the cluster.