CA1 pyramidal neuron: as a 2-layer NN and subthreshold synaptic summation (Poirazi et al 2003)

 Download zip file   Auto-launch 
Help downloading and running models
We developed a CA1 pyramidal cell model calibrated with a broad spectrum of in vitro data. Using simultaneous dendritic and somatic recordings, and combining results for two different response measures (peak vs. mean EPSP), two different stimulus formats (single shock vs. 50 Hz trains), and two different spatial integration conditions (within vs. between-branch summation), we found the cell's subthreshold responses to paired inputs are best described as a sum of nonlinear subunit responses, where the subunits correspond to different dendritic branches. In addition to suggesting a new type of experiment and providing testable predictions, our model shows how conclusions regarding synaptic arithmetic can be influenced by an array of seemingly innocuous experimental design choices.
1 . Poirazi P, Brannon T, Mel BW (2003) Arithmetic of subthreshold synaptic summation in a model CA1 pyramidal cell. Neuron 37:977-87 [PubMed]
2 . Poirazi P, Brannon T, Mel BW (2003) Pyramidal neuron as two-layer neural network. Neuron 37:989-99 [PubMed]
3 . Poirazi P, Brannon T, Mel BW (2003ab-sup) Online Supplement: About the Model Neuron 37 Online:1-20
4 . Polsky A, Mel BW, Schiller J (2004) Computational subunits in thin dendrites of pyramidal cells. Nat Neurosci 7:621-7 [PubMed]
Citations  Citation Browser
Model Information (Click on a link to find other models with that property)
Model Type: Neuron or other electrically excitable cell;
Brain Region(s)/Organism:
Cell Type(s): Hippocampus CA1 pyramidal GLU cell;
Channel(s): I Na,p; I Na,t; I L high threshold; I T low threshold; I A; I K; I M; I h; I K,Ca; I Calcium;
Gap Junctions:
Receptor(s): GabaA; GabaB; NMDA; Glutamate;
Simulation Environment: NEURON;
Model Concept(s): Action Potential Initiation; Activity Patterns; Dendritic Action Potentials; Active Dendrites; Influence of Dendritic Geometry; Detailed Neuronal Models; Action Potentials; Depression; Delay;
Implementer(s): Poirazi, Panayiota [poirazi at];
Search NeuronDB for information about:  Hippocampus CA1 pyramidal GLU cell; GabaA; GabaB; NMDA; Glutamate; I Na,p; I Na,t; I L high threshold; I T low threshold; I A; I K; I M; I h; I K,Ca; I Calcium;
cad.mod *
cagk.mod *
cal.mod *
calH.mod *
car.mod *
cat.mod *
d3.mod *
gabaa.mod *
gabab.mod *
glutamate.mod *
h.mod *
hha_old.mod *
hha2.mod *
kadist.mod *
kaprox.mod *
kca.mod *
km.mod *
nap.mod *
nmda.mod *
somacar.mod *
mosinit.hoc *
mosinit.hoc.old *
Simple synaptic mechanism derived for first order kinetics of
binding of transmitter to postsynaptic receptors.

A. Destexhe & Z. Mainen, The Salk Institute, March 12, 1993.

During the arrival of the presynaptic spike (detected by threshold 
crossing), it is assumed that there is a brief pulse (duration=Cdur)
of neurotransmitter C in the synaptic cleft (the maximal concentration
of C is Cmax).  Then, C is assumed to bind to a receptor Rc according 
to the following first-order kinetic scheme:

Rc + C ---(Alpha)--> Ro							(1)

where Rc and Ro are respectively the closed and open form of the 
postsynaptic receptor, Alpha and Beta are the forward and backward
rate constants.  If R represents the fraction of open gates Ro, 
then one can write the following kinetic equation:

dR/dt = Alpha * C * (1-R) - Beta * R					(2)

and the postsynaptic current is given by:

Isyn = gmax * R * (V-Erev)						(3)

where V is the postsynaptic potential, gmax is the maximal conductance 
of the synapse and Erev is the reversal potential.

If C is assumed to occur as a pulse in the synaptic cleft, such as

C     _____ . . . . . . Cmax
      |   |
 _____|   |______ . . . 0 
     t0   t1

then one can solve the kinetic equation exactly, instead of solving
one differential equation for the state variable and for each synapse, 
which would be greatly time consuming...  

Equation (2) can be solved as follows:

1. during the pulse (from t=t0 to t=t1), C = Cmax, which gives:

   R(t-t0) = Rinf + [ R(t0) - Rinf ] * exp (- (t-t0) / Rtau )		(4)

   Rinf = Alpha * Cmax / (Alpha * Cmax + Beta) 
   Rtau = 1 / (Alpha * Cmax + Beta)

2. after the pulse (t>t1), C = 0, and one can write:

   R(t-t1) = R(t1) * exp (- Beta * (t-t1) )				(5)

There is a pointer called "pre" which must be set to the variable which
is supposed to trigger synaptic release.  This variable is usually the
presynaptic voltage but it can be the presynaptic calcium concentration, 
or other.  Prethresh is the value of the threshold at which the release is

Once pre has crossed the threshold value given by Prethresh, a pulse
of C is generated for a duration of Cdur, and the synaptic conductances
are calculated accordingly to eqs (4-5).  Another event is not allowed to
occur for Deadtime milliseconds following after pre rises above threshold.

The user specifies the presynaptic location in hoc via the statement
	connect pre_GLU[i] , v.section(x)

where x is the arc length (0 - 1) along the presynaptic section (the currently
specified section), and i is the synapse number (Which is located at the
postsynaptic location in the usual way via
	postsynaptic_section {loc_GLU(i, x)}
Notice that loc_GLU() must be executed first since that function also
allocates space for the synapse.
  GLUTAMATE SYNAPSE (AMPA-Kainate receptors)

  Parameters estimated from whole cell recordings of synaptic currents on
  Cochlear neurons (Raman & Trussel, Neuron 9: 173-186, 1992) as well as
  from sharp electrode EPSPs recordings in thalamocortical neurons (LGN)
  (Crunelli et al. J. Physiol. 384: 603, 1987).



	RANGE C, R, R0, R1, g, gmax, lastrelease
	GLOBAL Cmax, Cdur, Alpha, Beta, Erev, Prethresh, Deadtime, Rinf, Rtau
	(nA) = (nanoamp)
	(mV) = (millivolt)
	(umho) = (micromho)
	(mM) = (milli/liter)

	Cmax	= 1	(mM)		: max transmitter concentration
	Cdur	= 1.1	(ms)		: transmitter duration (rising phase)
	Alpha	= 10	(/ms mM)	: forward (binding) rate
	Beta	= 0.5	(/ms)		: backward (unbinding) rate
	Erev	= 0	(mV)		: reversal potential
	Prethresh = 0 			: voltage level nec for release
	Deadtime = 0	(ms)		: mimimum time between release events
	gmax		(umho)		: maximum conductance

	v		(mV)		: postsynaptic voltage
	i 		(nA)		: current = g*(v - Erev)
	g 		(umho)		: conductance
	C		(mM)		: transmitter concentration
	R				: fraction of open channels
	R0				: open channels at start of release
	R1				: open channels at end of release
	Rinf				: steady state channels open
	Rtau		(ms)		: time constant of channel binding
	pre 				: pointer to presynaptic variable
	lastrelease	(ms)		: time of last spike

	R = 0
	C = 0
	Rinf = Cmax*Alpha / (Cmax*Alpha + Beta)
	Rtau = 1 / ((Alpha * Cmax) + Beta)
	lastrelease = -9e9

	SOLVE release
	g = gmax * R
	i = g*(v - Erev)

PROCEDURE release() { LOCAL q
	:will crash if user hasnt set pre with the connect statement 

	q = ((t - lastrelease) - Cdur)		: time since last release ended

						: ready for another release?
	if (q > Deadtime) {
		if (pre > Prethresh) {		: spike occured?
			C = Cmax			: start new release
			R0 = R
			lastrelease = t
	} else if (q < 0) {			: still releasing?
		: do nothing
	} else if (C == Cmax) {			: in dead time after release
		R1 = R
		C = 0.

	if (C > 0) {				: transmitter being released?
		    R = Rinf + (R0 - Rinf) * exptable (- (t - lastrelease) / Rtau)

: This synapse produces a negative R when it is first hit with
: neurotransmitter.
:		if (R < 0) {
:			printf("\t**negative R: %e\n", R)
:			}
	} else {				: no release occuring

  	   R = R1 * exptable (- Beta * (t - (lastrelease + Cdur)))

:	printf ("----------------------------------------\n")
:	printf ("q: %f Deadtime: %f pre: %f\n", q, Deadtime, pre )
:	printf ("C: %f R: %.12f t: %f lastrelease: %f\n", C, R, t, lastrelease)

	return 0;

FUNCTION exptable(x) { 
	TABLE  FROM -10 TO 10 WITH 2000

	if ((x > -10) && (x < 10)) {

		exptable = exp(x)
	} else {
		exptable = 0.