Synaptic plasticity in hippocampal slices

Questions to investigate

1. Does one train of high-frequency stimulation produce lasting potentiation in control slices, and how much of it decays during the hour?
2. Does AP5 prevent the induction of potentiation while leaving baseline transmission and the paired-pulse ratio essentially unchanged?
3. Raise or lower the release probability. How does the paired-pulse ratio change, and why does it change with release probability?
4. Run low-frequency stimulation, and a time control without conditioning. Which changes exceed slow drift?
5. How many animals are needed for a clear answer, and how misleading is it to count slices as independent replicates?
6. Change the exclusion criteria. How many slices are lost, and could unequal exclusion bias the comparison?

In a real laboratory

Transverse hippocampal slices, 350–400 µm thick, are cut on a vibratome into ice-cold oxygenated solution and recover for at least an hour in artificial cerebrospinal fluid (ACSF) bubbled with 95% O2 and 5% CO2. In a chamber perfused at 30–32 °C, a stimulating electrode in stratum radiatum activates the Schaffer collaterals, and a glass electrode records the field potential in stratum radiatum of CA1. Each slice is characterised by an input–output curve and paired pulses, the test intensity is set to a fraction of the maximal response, and a stable baseline of at least 20 minutes precedes induction. Exclusion criteria are written down in advance. Drug and control experiments are interleaved in random order, solutions are coded where possible, and both the number of slices and the number of animals are reported, with the animal as the unit of analysis. The work requires approval by an Institutional Animal Ethics Committee (in India, under the CCSEA and the Prevention of Cruelty to Animals Act 1960; elsewhere, the equivalent national body); taking several slices from each animal reduces the number of animals.

How the simulation works

Each sweep is a waveform sampled at 20 kHz: a stimulus artefact, a fibre volley that grows as a sigmoid function of stimulus intensity, and an fEPSP, with filtered recording noise and sweep-to-sweep variability. The slope is measured as in a laboratory, by a straight-line fit between the 10% and 90% points of the initial negative-going phase, and is reported as a positive number in mV/ms. For the time course the window is set on the average of the baseline sweeps and applied to every sweep; single sweeps measured with their own window underestimate small, noisy slopes. At 20 ms the decaying first response slightly reduces the second slope, as in real recordings. Paired-pulse ratios follow an extension of the Tsodyks–Markram model that adds facilitation. Long-term change follows a calcium rule: calcium enters through NMDA receptors, which AP5 blocks by 95%; a weak signal gives no change, an intermediate one depression and a strong one potentiation, which partly decays over tens of minutes. A short post-tetanic potentiation, independent of NMDA receptors, follows high-frequency trains. Slices and animals differ in response size, recruitment, potentiation and drift, and some slices are unhealthy. All values are approximate; the AP5 concentration (50 µM), the ACSF recipe and the stimulation parameters are typical laboratory values, not taken from the cited studies. Population spikes, inhibition, slice temperature, age and the late, protein-synthesis-dependent phase of LTP are not simulated.