Download presentation

Presentation is loading. Please wait.

Published byLena Farman Modified over 2 years ago

1
Synapses and Multi Compartmental Models Computational Neuroscience 03 Lecture 2

2
Synapses: The synapse is remarkably complex and involves many simultaneous processes such as the production and degredation of neurotransmitter. The neurotransmitters directly (A) or indirectly (B) binds to a synaptic channel and activates it.

3
Synaptic conductances: Synaptic transmission begins when an action potential invades the presynaptic terminal and activates voltage dependent Ca 2+ channels. This causes transmitter molecules to enter the cleft and bind to receptors on the postsynaptic neuron. As a result ion channels open, which modifies the conductance of the postsynaptic neuron Synaptic conductance: P: open channel probability P rel : probability of transmitter release P s : probability that postsyn. channel opens Both stochastic processes

4
Postsynaptic conductance: closing rate of the channel opening rate dependent on transmitter conc Spike ignore during the opening process

5
for if there is no synaptic release immediately before the release at t=0 using a simple manipulation we can write in the general case Update rule after spikes

6
Fast synapse: For a fast synapse the rise of the conductance following a presynaptic action potential can be approximated as instantaneous. For a single presynaptic action potential occurring at t=0 we can write with A sequence of action potentials at arbitrary times can be modeled with an exponential decay and by updating the probability after each action potential with:

7
Slow synapse (e.g. GABA A and NMDA): For an isolated presynaptic action potential occurring at t=0 we can use the same model or a difference of two exponentials or the alpha function with a peak value at B is a normalization factor and ensures that the peak value is equal to P max

8
Examples of time-dependent open probabilities: Single exp. decay Diff of two exp. exponential decay Instantaneous rise

9
NMDA: Slow (20ms rise) Physiological correlate of the Hebb learning rule since both, the presynaptic and postsynaptic cell have to be active. The voltage dependence is mediated by magnesium ions which normally block NDMA receptors. The postsynaptic cell Must be sufficiently depolarized to knock out the blocking ions. Dependence of the NMDA conductance on the membrane potential V and the extracellular Mg 2+ concentration.

10
Probability of transmitter release and short-term plasticity: Depression (D)andfacilitation (F) of excitory intercortical synapses with the release probability after a long period of silence Threshold

11
Steady-state release probability for a presynaptic Poisson spike-train: average steady state release probability The facilitation after each spike is cancelled out by the average exponential decrement between presynaptic spikes. Consider two action potentials separated by an interval and the release probability at the time of the first spike is Immediately after the spike the release probability is set to By the time of the second spike it is decayed to

12
Since we are interested in the average release probability, we have to determine the average exponential decay with an interspike interval Probability density of a Poisson spike train with interspike intervall .

13
Steady-state release probability for a presynaptic Poisson spike-train: Facilitating synapse Depressing synapse : Synaptic transmission

14
Transmission for a depressing synapse: Due to the 1/r release probability at high rates, the synaptic transmission becomes independent of the firing rate. Thus, depressing synapses do not convey the value of high presynaptic firing rates. They emphasize changes in the firing rate. Prior to a change: After a change: for high rates r

15
AMPA: fast Glutamate activates two different kinds of receptors: AMPA and NMDA. Both receptors lead to an excitation of the membrane. Examples of some synapses:

16
GABA ( aminobutyric acid) is the principal inhibitory neurotransmitter. There are two main receptors for GABA, GABA A and GABA B. GABA A GABA A is responsible for fast inhibtion and require only brief stimuli to produce a response.

17
GABA B GABA B is a much more complex receptor. It involves so-called second messengers. GABA B responses occur when the GABA binds to another compound (G-potein) which in turn binds to a Potassim channel and opens it up. It takes 4 activated G-proteins to open the channel.

18
Gap junctions are not chemical synapses but electrical in nature. The produce a current proportional to the difference between pre-and postsynaptic potential. No transmitter or action potential is involved. Many non-neural cells, e.g. muscle, glia, are coupled in this manner.

19
Synaptic in integrate and fire Can also add synapses onto an integrate and fire using: P s is probability of firingand changes whenever the presynaptic neuron fires using one of the schemes mentioned previously. For simplicity, assume P rel is 1

20
Used to look at eg dynamics of large numbers of inputs and effects of inhibitory/excitatory synapses Have excitatory or inhibitory synapses depending on whether E S is above/below membrane potential interestingly inhibitory synapse produces more synchronous firing

21
Cable model Have been assuming no spatial variation in membrane potential: Not true especially for neurons with long narrow processes Attenuation and degradation of signal is most severe when current passes down the long cable-like dendritic or axonal branches => Cable theory: assumption is that cables are radially homogeneous Longtitudinal resistance over cable of length x: R L = r L x/( a 2 ) Where r L is intracellular reistivity and a is radius Simultaneous recordings from different parts of neurons

22
From Ohm’s law, we get a pde relating voltage difference to current i m (see abbott and Dayan, pp ) i m generally complex and must be integrated numerically. However for linear current, no synaptic current and infinite cable can solve analytically to get Where electrotonic length And voltage drops by a factor of exp(-n) at a distance of

23
This model made more realistic by Rall: equivalent cable model

24
Multi Compartment models Dendrites only locally uniform Make different compartments with different properties Each compartment reduces to the cable model Need some way of them linking at the ends: Where compartments are indexed by

25
More compartments means better approximation

26
However, splitting axons into locally unifrom sections OK since we have to do this in numerical integration The simplest method of solving ODE's is Euler's method: Rule of the thumb:

27
Piecewise approximation: For constant I: update

28
The value of the conductance between compartments can be calculated from Ohm’s law. If compartments have equal length L and radius a If not: Can solve the equations and so see how APs propagate along axons.

29
AP Propagation AP propagates since point 0 is depolarised to V 0 it causes pt 1 to depolarise to V 1. Before reaching V1 however it crosses threshold causing spike, which causes pt 2 which had been moving towards V2 to move towrds V1 etc

30
APs can move either way, but don’t go backwards because of refractory period. Speed of propagation can be shown to be proportional to sqrt(radius) for unmyelinated axons. However, for myelinated axons, at the optimal thickness of myelin inner=0.6outer which is actual thickness of axons (thickness effects membrane capacitance etc) with respect to speed, speed is proportional to radius

Similar presentations

© 2017 SlidePlayer.com Inc.

All rights reserved.

Ads by Google