Fig 1.
The effects of carbachol and noradrenaline on feed-forward excitatory and inhibitory transmission in the mossy fiber pathway.
A) Left: Experimental setup indicating the location of stimulation and recording electrodes within a hippocampal slice. Right: Schema of feed-forward mossy fiber circuit. Granule cells (GC) in the dentate gyrus send mossy fiber axons to synapse onto feed-forward interneurons (IN) and CA3 pyramidal cells (PC). B) EPSCs and IPSCs evoked by granule cells stimulation were blocked by 1 μM DCG-IV, confirming responses were driven by mossy fiber activation. Top: Example traces of DCG-IV block (black) of EPSCs (red) and IPSCs (blue). Bottom: Time course of DCG-IV block of 4th EPSC (red, n = 9) and IPSC (blue, n = 6). C) 5 μM CCh mildly suppresses mossy fiber EPSCs. Top left: Example traces before and after bath application of 5 μM CCh. Bottom Left: Time course of CCh effect, and washout (n = 7). Top Right: Effect of CCh on response amplitudes for each pulse. Bottom Right: Effect of CCh on nth/1st Pulse Ratio. D) 5 μM CCh substantially reduces disynaptic mossy fiber driven IPSC amplitudes. Top left: Example traces before and after bath application of 5 μM CCh. Bottom Left: Time course of CCh effect, and washout (n = 5). Top Right: Effect of CCh on response amplitudes for each pulse. Bottom Right: Effect of CCh on nth/1st Pulse Ratio. E) 20 μM NA has no effect on mossy fiber EPSCs. Top left: Example traces before and after bath application of 20 μM NA. Bottom Left: Time course of NA effect (n = 7). Top Right: Effect of NA on response amplitudes for each pulse. Bottom Right: Effect of NA on nth/1st Pulse Ratio. F) 20 μM NA reduces disynaptic mossy fiber driven IPSC amplitudes. Top left: Example traces before and after bath application of 20 μM NA. Bottom Left: Time course of NA effect (n = 5). Top Right: Effect of NA on response amplitudes for each pulse. Bottom Right: Effect of NA on nth/1st Pulse Ratio.
Fig 2.
The mechanism of carbachol and noradrenaline action on excitatory and inhibitory feed-forward mossy fiber transmission determined by short-term plasticity models.
A) Irregular stimulation protocol modelled on naturalistic granule cell spike patterns. GC spike patterns recorded in vivo during a spatial memory task, with a bimodal inter-spike interval (ISI) distribution (top right). Bimodal ISI distribution modelled as a doubly stochastic Cox process (middle right), with irregular stimulation protocol a sample drawn from this process (bottom right). B-C) Experimentally recorded EPSCs and IPSCs evoked by irregular stimulation protocol in hippocampal slices under control or presence of 5 μM CCH or 20 μM NA. Evoked peaks highlighted by white dots. The same example burst is shown on expanded timescales. D) Tsodyks-Markram short-term plasticity model schematic illustrating facilitating (f) and depressing (d) presynaptic components with time constants (τf, τd) and postsynaptic scaling factor (g). E-F) Model selection and fitting for EPSCs (E) and IPSCs (F). Left: AIC and BIC weights for each fitted model. Model selection by highest AIC and BIC weights and evidence ratios. Right: Modulation by CCh or NA assessed by effect on parameter fits normalized by time-matched control. Error bars are standard deviations, * denotes significant parameter change.
Fig 3.
Carbachol and noradrenaline alter the Excitatory-Inhibitory ratio within the feed-forward mossy fiber pathway in a frequency-dependent manner.
A) Simplification of bursting spike trains into two parameter spaces: between burst interval (BI) describing interval between bursts in a spike train, and within BI describing interval between spikes within a burst. B) Example synaptic waveforms of expected mossy fiber EPSCs (red) and IPSCs (blue) generated by the short-term plasticity models under control, CCh and NA conditions. The 3 rows illustrate short-term plasticity dynamics at three pairs of within and between burst intervals (20ms and 2s, 20ms and 50s, 500ms and 50s respectively). C) Expected short-term plasticity of EPSCs across a wide range of within and between BIs. Light blue, dark blue and black crosses shown in C denote within and between BIs used in the examples shown in B. Data in the presence of CCh or NA not shown since CCh does not change the facilitation of EPSCs and NA has no effect on EPSCs. D) Expected short-term plasticity of IPSCs across a wide range of within and between BIs and in control, CCh and NA conditions. Pulse ratios for 2nd, 6th and 10th pulses compared to the 1st are shown to illustrate change in facilitation across a 10 pulse burst. E) Progression of Excitatory-Inhibitory ratio across the range of within and between BIs and in control, CCh and NA conditions. Pulse ratios for 2nd, 6th and 10th pulses compared to the 1st are shown to illustrate change in E-I ratio across a 10 pulse burst.
Table 1.
Best fit Tsodyks-Markram model parameter sets for EPSC and IPSC short-term plasticity.
Values that are changed by acetylcholine or noradrenaline are shown.
Fig 4.
Acetylcholine- and noradrenaline-mediated disinhibition facilitates back-propagation of EPSPs and action potentials into the dendrites of CA3 pyramidal cells.
A) Sketch of CA3 pyramidal cell and positioning within the layers of the hippocampus. Synaptic inputs are shown with location of contact (red—recurrent CA3-CA3 synapse, blue—feed-forward inhibitory synapse, gray—mossy fiber synapse). SLM–stratum lacunosum moleculare, SR–stratum radiatum, SL–stratum lucidum, SP–stratum pyramidale, SO–stratum oriens. B) Example traces produced by the biophysical CA3 neuron model of action potentials generated at the soma from summated mossy fiber EPSPs presented at 20 Hz (top), back-propagation into the radial oblique dendrites (middle), and dendritic calcium influx (bottom), in control and with acetylcholine (ACh) or noradrenaline (NA)-mediated disinhibition of feed-forward inhibition. C) Back-propagating action potential amplitude before (left) and after (middle) cholinergic or noradrenergic modulation, and the difference in amplitude (right) distributed across an example CA3 pyramidal cell. D) The number of stimuli required to generate a back-propagating action potential across all cell morphologies. Only dendrites that had back-propagating action potentials in control conditions are shown. E) The proportion of oblique dendrites in stratum radiatum reached by a back-propagating action potential per stimulus for all cell morphologies. F) Histogram of differences in back-propagating action potential amplitudes with and without acetylcholine disinhibition in stratum radiatum oblique dendritic compartments (< 1 μm diameter) from 15 cells. G) Distribution of back-propagating action potential amplitudes in stratum radiatum oblique dendrites for a range of excitation-inhibition ratios in 15 cells. In our simulations the effect of acetylcholine was modelled as a change in this ratio (the absence of acetylcholine, gI/gE = 3; in the presence of acetylcholine, gI/gE = 1). H) The probability of successful action potential back-propagation (bAP amplitude > 40 mV) in oblique dendrites in 15 cells as a function of excitation-inhibition ratio.
Fig 5.
Facilitating mossy fiber inputs generate rapid and stable ensemble formation.
A) Schematic showing the network properties of the spiking network model and long-term plasticity rules. Left: A population of excitatory (red) and inhibitory (blue) cells with all-to-all connectivity and mossy fiber input (gray). Right: Recurrent excitatory CA3-CA3 spike timing-dependent plasticity rule with a symmetric window that shifted from potentiation for correlated spiking at low rates (Ρpost = 0), to depression for uncorrelated spiking near the maximum postsynaptic firing rate (Ρpost = Ρmax). B) Comparison of mossy fiber to irregular ‘perforant path’ input. Synaptic weight evolution with time for CA3-CA3 recurrent connections (red) and inhibitory to excitatory connections (blue) for a two cell excitatory population (top), a 10 cell excitatory population (middle) and a population including 10 excitatory and 5 inhibitory cells (bottom).
Fig 6.
Carbachol but not noradrenaline enhances CA3 cellular excitability and depresses associational/commissural synaptic connections.
A) 5 μM CCh depresses associational/commissural evoked EPSCs. Top: Example traces before and after bath application of 5 μM CCh. Bottom: Time course of CCh effect, and washout (n = 4). B) 20 μM NA has no effect on associational/commissural evoked EPSCs. Top: Example traces before and after bath application of 20 μM NA. Bottom: Time course of NA effect, and washout (n = 5). C) 5 μM CCh but not 20 μM NA depresses associational/commissural evoked EPSCs and increases paired pulse ratio (PPR). D) 5 μM CCh but not 20 μM NA depolarizes CA3 pyramidal cell membrane potential and increases input resistance.
Fig 7.
Acetylcholine speeds up ensemble formation and lowers input frequency requirement by increasing cellular excitability.
A) Network setup: A population of excitatory and inhibitory cells connected in all-to-all fashion. Subpopulations of excitatory cells receive independent feed-forward input that drives ensemble formation. B) Example weight matrix driven by input with 30 Hz bursts in the presence of NA, 20 Hz bursts in the presence of NA or 20 Hz bursts in the presence of acetylcholine. Stronger weights indicate robust ensemble formation. C) Evolution of ensemble formation illustrated by the weight matrix error reduction over time for different input burst frequencies. Triangles denote which effects of acetylcholine on the CA3 network were included in each set of simulations.
Table 2.
CA3 network parameter changes to model cholinergic modulation (ACh) of CA3.
Note that noradrenaline (NA) did not cause any changes to CA3 network properties and therefore parameters for noradrenaline are the same as control conditions.
Fig 8.
Acetylcholine enables a CA3 network to form stable overlapping ensembles by reducing the strength of recurrent excitatory CA3-CA3 synapses.
A) Network setup: A population of excitatory and inhibitory cells connected in all-to-all fashion. Subpopulations of excitatory cells receive overlapping feed-forward input that drives ensemble formation. B) Example weight matrix driven by input with increasing degrees of overlap between ensembles [0, 2, 4 cells] in the presence of NA or ACh. C) Evolution of ensemble formation illustrated by the weight matrix error over time for different degrees of overlap between ensembles. Triangles denote which effects of acetylcholine were included in the simulation. D) Effect of increasing overlap between ensembles on the ability to discriminate between ensembles defined as the difference in ensemble population spiking rates.