Olfactory bulb mitral and granule cell column formation (Migliore et al. 2007)

 Download zip file   Auto-launch 
Help downloading and running models
Accession:114665
In the olfactory bulb, the processing units for odor discrimination are believed to involve dendrodendritic synaptic interactions between mitral and granule cells. There is increasing anatomical evidence that these cells are organized in columns, and that the columns processing a given odor are arranged in widely distributed arrays. Experimental evidence is lacking on the underlying learning mechanisms for how these columns and arrays are formed. We have used a simplified realistic circuit model to test the hypothesis that distributed connectivity can self-organize through an activity-dependent dendrodendritic synaptic mechanism. The results point to action potentials propagating in the mitral cell lateral dendrites as playing a critical role in this mechanism, and suggest a novel and robust learning mechanism for the development of distributed processing units in a cortical structure.
Reference:
1 . Migliore M, Inzirillo C, Shepherd GM (2007) Learning mechanism for column formation in the olfactory bulb. Front Integr Neurosci 1:12 [PubMed]
Model Information (Click on a link to find other models with that property)
Model Type: Realistic Network;
Brain Region(s)/Organism: Olfactory bulb;
Cell Type(s): Olfactory bulb main mitral cell; Olfactory bulb main interneuron granule MC cell;
Channel(s): I Na,t; I A; I K;
Gap Junctions:
Receptor(s): AMPA; NMDA; Gaba;
Gene(s):
Transmitter(s): Gaba; Glutamate;
Simulation Environment: NEURON;
Model Concept(s): Activity Patterns; Dendritic Action Potentials; Active Dendrites; Detailed Neuronal Models; Synaptic Plasticity; Long-term Synaptic Plasticity; Action Potentials; Learning; Olfaction;
Implementer(s): Migliore, Michele [Michele.Migliore at Yale.edu];
Search NeuronDB for information about:  Olfactory bulb main mitral cell; Olfactory bulb main interneuron granule MC cell; AMPA; NMDA; Gaba; I Na,t; I A; I K; Gaba; Glutamate;
/
plast
readme.html
kamt.mod *
kdrmt.mod *
naxn.mod *
nmdanetOB.mod *
2mc-w05-w00-e2i3-int220.hoc
2mt-s1-w05-w00-e2i3-int220.txt
2mt-s2-w05-w00-e2i3-int220.txt
2mt-s4-w05-w00-e2i3-int220.txt
gc-plast.hoc
mitral-plast-2.hoc
mosinit.hoc
plasticity-disp.hoc
screenshot.jpg
trace-gc0dend0-w05-w00-e2i3-int220.txt
trace-gc33dend0-w05-w00-e2i3-int220.txt
trace-mt0dend066-w05-w00-e2i3-int220.txt
trace-mt0soma05-w05-w00-e2i3-int220.txt
trace-time-w05-w00-e2i3-int220.txt
                            
load_file("nrngui.hoc")
load_file("mitral-plast-2.hoc")
load_file("gc-plast.hoc")
use_mcell_ran4(1)
mcell_ran4_init(125)
cvode.active(1)

Vrest = -65
dt = 1
celsius=35
tstop=10000
diff=0

objref nconp[2], net, netp[2], g, b, synp, nil, stim, apc, time,mt0soma05, gc0dend0, gc33dend0, mt0dend066
objref mt[2], gc[50], train[50], trainmt[2][50], outfile1, outfile2, outfile4
objref outtracetime,outtracemt0,outtracegc0 ,outtracegc33,outtracemt0dend066
objref nc[300], apc[50], apcmt[2][50], distrx, distry, distry2, sw
double prev[50], sigmoid[50], prevmt[2][50], sigmoidmt[2][50]


sw = new Random()
sw.uniform(0, 25)

strdef filename
outfile1 = new File()
outfile2 = new File()
outfile4 = new File()
outtracetime = new File()
outtracemt0= new File()

outtracegc0= new File()


outtracegc33= new File()

outtracemt0dend066= new File()

distrx = new Vector()
distry = new Vector()
distry2 = new Vector()
time = new Vector()
mt0soma05 = new Vector()

gc0dend0 = new Vector()


gc33dend0= new Vector() 

mt0dend066= new Vector() 

outfile1.wopen("2mt-s1-w05-w00-e2i3-int220.txt")
outfile2.wopen("2mt-s2-w05-w00-e2i3-int220.txt")
outfile4.wopen("2mt-s4-w05-w00-e2i3-int220.txt")
outtracetime.wopen("trace-time-w05-w00-e2i3-int220.txt")
outtracemt0.wopen("trace-mt0soma05-w05-w00-e2i3-int220.txt")

outtracegc0.wopen("trace-gc0dend0-w05-w00-e2i3-int220.txt")


outtracegc33.wopen("trace-gc33dend0-w05-w00-e2i3-int220.txt")

outtracemt0dend066.wopen("trace-mt0dend066-w05-w00-e2i3-int220.txt")




for i=0, 1 {mt[i] = new Mitral()}

for i=0, 49 {
	gc[i] = new GC()

	train[i] = new Vector()
	gc[i].dend[0] apc[i] = new APCount(i*0.02)
	apc[i].thresh=-40
	apc[i].record(train[i])
	prev[i]=0
	sigmoid[i]=0//sw.repick()

	trainmt[0][i] = new Vector()
	mt[0].secden[1] apcmt[0][i] = new APCount(i*0.02)
	apcmt[0][i].thresh=-40
	apcmt[0][i].record(trainmt[0][i])
	prevmt[0][i]=0
	sigmoidmt[0][i]=0//sw.repick()

	trainmt[1][i] = new Vector()
	mt[1].secden[0] apcmt[1][i] = new APCount((49-i)*0.02)
	apcmt[1][i].thresh=-40
	apcmt[1][i].record(trainmt[1][i])
	prevmt[1][i]=0
	sigmoidmt[1][i]=0//sw.repick()

	distrx.append(i)
	//print i, sigmoid[i], sigmoidmt[0][i], sigmoidmt[1][i]
}



weight=0.5
amp=.03
rel=0.2
inh=3
synstr=2
nmdafactor=0.0035



for i=0, 1 {
access mt[i].soma
	distance()
	netp[i] = new NetStim(0)
	netp[i].number=200
	netp[i].interval=220
	netp[i].noise=0
	netp[i].start=2
}
	nconp[0]= new NetCon(netp[0],mt[0].synodor,0.5,0,0.5*1.e-3) 
	nconp[1]= new NetCon(netp[1],mt[1].synodor,0.5,0,0*1.e-3) 

////////////////// circuit definition  

///// gc <-> mt

for z=0, 49 {

mt[0].secden[1]	 nc[0+z*6]= new NetCon(&v(z*0.02),gc[z].synmt[0],-40,1,0*nmdafactor) 
mt[0].secden[1]	 nc[1+z*6]= new NetCon(&v(z*0.02),gc[z].sampa[0],-40,1,0*1e-3) 
gc[z].dend[0]	 nc[2+z*6]= new NetCon(&v(1),mt[0].igp[1][z],-40,1,0*inh*1e-3) 

mt[1].secden[0]  nc[3+z*6]= new NetCon(&v((49-z)*0.02),gc[z].synmt[0],-40,1,0*nmdafactor) 
mt[1].secden[0]	 nc[4+z*6]= new NetCon(&v((49-z)*0.02),gc[z].sampa[0],-40,1,0*1e-3) 
gc[z].dend[0]    nc[5+z*6]= new NetCon(&v(1),mt[1].igp[0][49-z],-40,1,0*inh*1e-3) 

}
//stampa pesi nconp
print nconp[0].weight,nconp[1].weight, inh, synstr
////////////////// end circuit definition


proc init() {
	t=0
	disp=50
	finitialize(Vrest)
        fcurrent()
        forall {
		v=Vrest
		if (ismembrane("nax")) {e_pas=v+(ina+ik)/g_pas
		} else {
		e_pas=v+ik/g_pas
		}
	}
	for z=0, 49 {
	nc[2+z*6].weight=inh*1.e-3*plast(sigmoid[z])
	nc[5+z*6].weight=inh*1.e-3*plast(sigmoid[z])
	nc[0+z*6].weight=synstr*nmdafactor*plast(sigmoidmt[0][z])
	nc[1+z*6].weight=synstr*1.e-3*plast(sigmoidmt[0][z])
	nc[3+z*6].weight=synstr*nmdafactor*plast(sigmoidmt[1][z])
	nc[4+z*6].weight=synstr*1.e-3*plast(sigmoidmt[1][z])
	}
	cvode.re_init()
	
	status()
}

proc advance() {
	fadvance()
	
	for (z=0; z<=49; z=z+1) {
	if (train[z].size()>=2 && train[z].size()>prev[z]) { 
		prev[z]=train[z].size()
		diff= train[z].x[train[z].size()-1]-train[z].x[train[z].size()-2]
		if (diff>1000/4) {fact=0}
		if (diff<1000/4 && diff>1000/30) {fact=-1}
		if (diff<1000/30) {fact=1}
		sigmoid[z]=sigmoid[z]+fact
		if (sigmoid[z]<=0) {sigmoid[z]=0}
		if (sigmoid[z]>=50) {sigmoid[z]=50}
		nc[2+z*6].weight=inh*1.e-3*plast(sigmoid[z])
		nc[5+z*6].weight=inh*1.e-3*plast(sigmoid[z])
	outfile2.printf("%d %g %g %d %g\n", z, t, diff, sigmoid[z], nc[2+z*6].weight)
//	print " GC ", t, z, prev[z], diff, fact, sigmoid[z], nc[2+z*6].weight
	}

	if (trainmt[0][z].size()>=2 && trainmt[0][z].size()>prevmt[0][z]) { 
		prevmt[0][z]=trainmt[0][z].size()
		diff= trainmt[0][z].x[trainmt[0][z].size()-1]-trainmt[0][z].x[trainmt[0][z].size()-2]
		if (diff>1000/4) {fact=0}
		if (diff<1000/4 && diff>1000/30) {fact=-1}
		if (diff<1000/30) {fact=1}
		sigmoidmt[0][z]=sigmoidmt[0][z]+fact
		if (sigmoidmt[0][z]<=-10) {sigmoidmt[0][z]=-10}
		if (sigmoidmt[0][z]>=60) {sigmoidmt[0][z]=60}
		nc[0+z*6].weight=synstr*nmdafactor*plast(sigmoidmt[0][z])
		nc[1+z*6].weight=synstr*1.e-3*plast(sigmoidmt[0][z])
	outfile1.printf("%d %g %g %d %g\n", z, t, diff, sigmoidmt[0][z], nc[1+z*6].weight)
//	print " MT0 ",t, z, 0, " ", prevmt[0][z], diff, fact, sigmoidmt[0][z], nc[1+z*6].weight
	}

	if (trainmt[1][z].size()>=2 && trainmt[1][z].size()>prevmt[1][z]) { 
		prevmt[1][z]=trainmt[1][z].size()
		diff= trainmt[1][z].x[trainmt[1][z].size()-1]-trainmt[1][z].x[trainmt[1][z].size()-2]
		if (diff>1000/4) {fact=0}
		if (diff<1000/4 && diff>1000/30) {fact=-1}
		if (diff<1000/30) {fact=1}
		sigmoidmt[1][z]=sigmoidmt[1][z]+fact
		if (sigmoidmt[1][z]<=0) {sigmoidmt[1][z]=0}
		if (sigmoidmt[1][z]>=50) {sigmoidmt[1][z]=50}
		nc[3+z*6].weight=synstr*nmdafactor*plast(sigmoidmt[1][z])
		nc[4+z*6].weight=synstr*1.e-3*plast(sigmoidmt[1][z])
	outfile4.printf("%d %g %g %d %g\n", z, t, diff, sigmoidmt[1][z], nc[4+z*6].weight)
//	print "MT1 ",t, z, 1, " ", prevmt[1][z], diff, fact, sigmoidmt[1][z], nc[4+z*6].weight
	}
	}

	

}

proc run() {

	
	time.record(&t)
	mt0soma05.record(&mt[0].soma.v(0.5))
	
	gc0dend0.record(&gc[0].dend[0].v(0.5))
	
	
	gc33dend0.record(&gc[33].dend[0].v(0.5))
	
	mt0dend066.record(&mt[0].secden[1].v(0.66))
	 

	
	stdinit()
	
	continuerun(tstop)
	for(i=0; i<mt0soma05.size(); i=i+1){	
		outtracetime.printf("%g \n", time.x[i])
		outtracemt0.printf("%g \n",mt0soma05.x[i])
		
		outtracegc0.printf("%g \n",gc0dend0.x[i])
		
	
		outtracegc33.printf("%g \n", gc33dend0.x[i])
		
		outtracemt0dend066.printf("%g \n", mt0dend066.x[i])
		
		
	}
	outfile1.close()
	outfile2.close()
	outfile4.close()
	
	outtracetime.close()
	outtracemt0.close()
	
	outtracegc0.close()
	

	outtracegc33.close()
	
	outtracemt0dend066.close()

	

}




proc status() {

	
	cvode.event(t+250, "status()")
}


func plast() {
	return (1-(1/(1+exp(($1-25)/3))))
}
run()

Loading data, please wait...