COMMENT
Updated Exp2Syn synapse with Mg-blocked nmda channel.
Defaul values of parameters (time constants etc) set to match synaptic channels in
striatal medium spiny neurons (Du et al., 2017; Chapman et al., 2003; Ding et al., 2008).
Robert . Lindroos @ ki . se
original comment:
________________
Two state kinetic scheme synapse described by rise time tau1,
and decay time constant tau2. The normalized peak condunductance is 1.
Decay time MUST be greater than rise time.
The solution of A->G->bath with rate constants 1/tau1 and 1/tau2 is
A = a*exp(-t/tau1) and
G = a*tau2/(tau2-tau1)*(-exp(-t/tau1) + exp(-t/tau2))
where tau1 < tau2
If tau2-tau1 -> 0 then we have a alphasynapse.
and if tau1 -> 0 then we have just single exponential decay.
The factor is evaluated in the
initial block such that an event of weight 1 generates a
peak conductance of 1.
Because the solution is a sum of exponentials, the
coupled equations can be solved as a pair of independent equations
by the more efficient cnexp method.
ENDCOMMENT
NEURON {
POINT_PROCESS adaptive_sglutamate
RANGE tau1_ampa, tau2_ampa, tau1_nmda, tau2_nmda
RANGE erev_ampa, erev_nmda, g, i
NONSPECIFIC_CURRENT i
RANGE i_ampa, i_nmda, g_ampa, g_nmda, I, G, mg, q, alpha, eta, nmda_ca_fraction
RANGE w0, NMDA_AMPA_ratio
RANGE weight, last_dopamine, KD1, KD2, n1, n2, KD_LTD, n_LTD
RANGE hthresh_LTP_const, hthresh_LTP_0, hthresh_max, n, delta, lthresh_LTD, lthresh_LTP, hthresh_LTP
RANGE learning_rate_w_LTP, learning_rate_w_LTD, thresh_LTP_0, learning_rate_thresh_LTP, thresh_LTD_0, learning_rate_thresh_LTD, hthresh_LTP_0
RANGE ca_nmdai_max, cali_max, active_syn_flag
POINTER dopamine, stimulus_flag
USEION ca_nmda READ ca_nmdai WRITE ica_nmda VALENCE 2
USEION cal READ cali VALENCE 2
}
UNITS {
(nA) = (nanoamp)
(mV) = (millivolt)
(uS) = (microsiemens)
}
PARAMETER {
erev_ampa = 0.0 (mV)
erev_nmda = 15.0 (mV)
tau1_ampa = 1.9 (ms)
tau2_ampa = 4.8 (ms) : tau2 > tau1
tau1_nmda = 5.52 (ms) : old value was 5.63
tau2_nmda = 231 (ms) : tau2 > tau1
mg = 1 (mM)
alpha = 0.062
q = 2
eta = 18
NMDA_AMPA_ratio = 1
w0 = 0.01
learning_rate_w_LTP = 0.01
learning_rate_w_LTD = 0.01
delta = 0.65
hthresh_LTP_const = 0.05
learning_rate_thresh_LTP = 0.005
learning_rate_thresh_LTD = 0.005
n1 = 4 : Hill coefficient
n2 = 4 : Hill coefficient
KD1 = 1
KD2 = 1
KD_LTD = 1
n_LTD = 1
nmda_ca_fraction = 0.15
ca_nmdai_max = 0
cali_max = 0
active_syn_flag = 1e-6
}
ASSIGNED {
v (mV)
i (nA)
g (uS)
factor_nmda
factor_ampa
i_ampa
i_nmda
g_ampa
g_nmda
block
I
G
stimulus_flag
dopamine
last_dopamine
ica_nmda (nA)
ca_nmdai (mM)
cali (mM)
weight
lthresh_LTP
lthresh_LTD
hthresh_LTP
}
STATE {
A (uS)
B (uS)
C (uS)
D (uS)
}
INITIAL {
LOCAL tp
if (tau1_nmda/tau2_nmda > .9999) {
tau1_nmda = .9999*tau2_nmda
}
if (tau1_ampa/tau2_ampa > .9999) {
tau1_ampa = .9999*tau2_ampa
}
: NMDA
A = 0
B = 0
tp = (tau1_nmda*tau2_nmda)/(tau2_nmda - tau1_nmda) * log(tau2_nmda/tau1_nmda)
factor_nmda = -exp(-tp/tau1_nmda) + exp(-tp/tau2_nmda)
factor_nmda = 1/factor_nmda
: AMPA
C = 0
D = 0
tp = (tau1_ampa*tau2_ampa)/(tau2_ampa - tau1_ampa) * log(tau2_ampa/tau1_ampa)
factor_ampa = -exp(-tp/tau1_ampa) + exp(-tp/tau2_ampa)
factor_ampa = 1/factor_ampa
weight = w0
lthresh_LTP = 0
lthresh_LTD = 0
hthresh_LTP = 0
active_syn_flag = 0
last_dopamine = 0
}
BREAKPOINT {
SOLVE state METHOD cnexp
: NMDA
g_nmda = (B - A)*weight*NMDA_AMPA_ratio
block = MgBlock()
i_nmda = g_nmda * (v - erev_nmda) * block
ica_nmda = nmda_ca_fraction*i_nmda
i_nmda = (1 - nmda_ca_fraction)*i_nmda
: AMPA
g_ampa = (D - C)*weight
i_ampa = g_ampa * (v - erev_ampa)
: total current
G = g_ampa + g_nmda
I = i_ampa
i = I
if (stimulus_flag == 1) {
ca_nmdai_max = max(ca_nmdai, ca_nmdai_max)
cali_max = max(cali, cali_max)
last_dopamine = dopamine
} else {
if (last_dopamine == 1 && active_syn_flag == 1) {
weight = weight + learning_rate_w_LTP * lthresh(ca_nmdai_max, KD1, n1) * hthresh(ca_nmdai_max, KD2, n2)
lthresh_LTP = lthresh(ca_nmdai_max, KD1, n1)
hthresh_LTP = hthresh(ca_nmdai_max, KD2, n2)
} else if (last_dopamine == -1 && active_syn_flag == 1) {
weight = weight - learning_rate_w_LTD * lthresh(cali_max, KD_LTD, n_LTD) * weight
lthresh_LTD = lthresh(cali_max, KD_LTD, n_LTD)
}
last_dopamine = dopamine
reset_max()
}
}
DERIVATIVE state {
A' = -A/tau1_nmda*q
B' = -B/tau2_nmda*q
C' = -C/tau1_ampa
D' = -D/tau2_ampa
}
NET_RECEIVE(dummy (uS)) {
active_syn_flag = 1
A = A + factor_nmda
B = B + factor_nmda
C = C + factor_ampa
D = D + factor_ampa
}
FUNCTION MgBlock() {
MgBlock = 1 / (1 + mg * eta * exp(-alpha * v) )
}
FUNCTION lthresh(conc, KD, n) {
lthresh = conc^n/(KD^n + conc^n)
}
FUNCTION hthresh(conc, KD, n) {
hthresh = KD^n/(KD^n + conc^n)
}
FUNCTION reset_max() {
ca_nmdai_max = 0
cali_max = 0
active_syn_flag = 0
}
FUNCTION max(current, maximum) {
if (current>maximum) {
max = current
} else {
max = maximum
}
}
FUNCTION min(current, minimum) {
if (current<minimum) {
min = current
} else {
min = minimum
}
}