Computational Biology, Part 20

Neuronal Modeling
Robert F. Murphy
Copyright 1996, 1999-2006.
All rights reserved.

Basic Neurophysiology
An imbalance of charge across a membrane
is called a membrane potential
The major contribution to membrane
potential in animal cells comes from
imbalances in small ions (e.g., Na, K)
The maintenance of this imbalance is an
active process carried out by ion pumps

Basic Neurophysiology
The cytoplasm of most cells (including
neurons) has an excess of negative ions over
positive ions (due to active pumping of
sodium ions out of the cell)
By convention this is referred to as a
negative membrane potential (inside
minus outside)
Typical resting potential is -50 mV

Basic Neurophysiology
Ion pumps require energy (ATP) to carry
ions across a membrane up a concentration
gradient (they generate a potential)
Ion channels allow ions to flow across a
membrane down a concentration gradient
(they dissipate a potential)

Basic Neurophysiology
A cell is said to be electrically polarized
when it has a non-zero membrane potential
A dissipation (partial or total) of the
membrane potential is referred to as a
depolarization, while restoration of the
resting potential is termed repolarization

Basic Neurophysiology
Ion channels can switch between open and
closed states
If an ion channel can switch its state due to
changes in membrane potential, it is said to
be voltage-sensitive
A membrane containing voltage-sensitive
ion channels and/or ion pumps is said to be
an excitable membrane

Basic Neurophysiology

An idealized neuron consists of

soma

or cell body

contains

nucleus and performs metabolic functions

dendrites
receive

signals from other neurons through synapses

axon
propagates

terminal
form

signal away from soma

branches

synapses with other neurons

Basic Neurophysiology

vv.carleton.ca/~neil/ neural/neuron-a.html

Basic Neurophysiology
Synapses pass signals from one neuron to
another
When synaptic vesicles fuse with axon
membrane, neurotransmitters are released
into space between axon and dendrite
Binding of neurotransmitters to dendrite
causes influx of sodium ions that diffuse
into soma

Basic Neurophysiology
The junction between the soma and the axon
is called the axon hillock
The soma sums (integrates) currents
(inputs) from the dendrites
When the received currents result in a
sufficient change in the membrane potential,
a rapid depolarization is initiated in the axon
hillock

Basic Neurophysiology
The depolarization is caused by opening of
voltage-sensitive sodium channels that
allow sodium ions to flow into the cell
The sodium channels only open in response
to a partial depolarization, such that a
threshold voltage is exceeded

Basic Neurophysiology
As sodium floods in, the membrane
potential reverses, such that the interior is
now positive relative to the outside
This positive potential causes voltagesensitive potassium channels to open,
allowing K+ ions to flow out
The potential overshoots (becomes more
negative than) the resting potential

Basic Neurophysiology
The fall in potential triggers the sodium
channels to close, setting the stage for
restoration of the resting potential by
sodium pumps
This sequential depolarization, polarity
reversal, potential overshoot and
repolarization is called an action potential

Action Potential
Stimulus
(uA)

15 0
10 0
50
0
60

Voltage (mV)

40
20
0
-20
-40
-60
-80

Conduc ta nce
(mS/c m2)

40
G(Na)

30

G(K)

20
10
0
0

4 Time (ms)

10

Basic Neurophysiology
The depolarization in the axon hillock
causes a depolarization in the region of the
axon immediately adjacent to the hillock
Depolarization (and repolarization)
proceeds down the axon until it reaches the
terminal branches
The depolarization causes synaptic vesicles
to fuse with the membrane, releasing
neurotransmitters to stimulate neurons with
which they form synapses

Basic Neurophysiology
These sequential depolarizations form a
traveling wave passing down the axon
Note that while a signal is passed down the
axon, it is not comparable to an electrical
signal traveling down a cable

Basic Neurophysiology

Current flows in an electrical cable

are

in the direction that the signal is propagating

consist of electrons

Current flows in a neuron

are

transverse to the signal propagation

consist of positively-charged ions

The Hodgkin-Huxley Model

Based on electrophysiological
measurements of giant squid axon
Empirical model that predicts experimental
data with very high degree of accuracy
Provides insight into mechanism of action
potential

The Hodgkin-Huxley Model

Define
voltage across the membrane at time t
q(t) net charge inside the neuron at t
I(t) current of positive ions into neuron at t
g(v) conductance of membrane at voltage v
C capacitance of the membrane
Subscripts Na, K and L used to denote specific
currents or conductances (L=other)
v(t)

The Hodgkin-Huxley Model

Start with equation for capacitor

v(t )

q(t )
C

The Hodgkin-Huxley Model

Consider each ion separately and sum

currents to get rate of change in charge and
hence voltage
dq

I Na I K I L

dt
I Na g Na (v v Na )
I K g K (v v K )
I L gL ( v vL )
dv
dt

1
C

g Na ( v)(v v Na ) g K ( v)( v v K ) g L (v v L )

The Hodgkin-Huxley Model

Central concept of model: Define three state

variables that represent (or control) the
opening and closing of ion channels
m

controls Na channel opening

h controls Na channel closing
n controls K channel opening

The Hodgkin-Huxley Model

Define relationship of state variables to

conductances of Na and K
g Na g Na m h
3

gK gK n 4
0 m, n, h 1

The Hodgkin-Huxley Model

Define empirical differential equations to

model behavior of each gate

dn
n (v)(1 n) n (v)n
dt
0.01(v
10)
n (v)
(v 10)/10
(e
1)

n (v) 0.125e

v / 80

The Hodgkin-Huxley Model

Define empirical differential equations to

model behavior of each gate

dm
m (v)(1 m) m (v)m
dt
0.1(v

25)
m (v)
(v 25)/10
(e
1)

m (v) 4e

v /18

The Hodgkin-Huxley Model

Define empirical differential equations to

model behavior of each gate

dh
h (v)(1 h) h (v)h
dt
v / 20
h (v) 0.07e
h (v) 1 (v 30)/10
(e
1)

The Hodgkin-Huxley Model

Gives set of four coupled, non-linear,
ordinary differential equations
Must be integrated numerically
Constants (g in mmho/cm2 and v in mV)

g Na 120

v Na 115

g K 36

v K 12
v L 10.5989

gL 0.3

Hodgkin-Huxley Gates
Stimulus
(uA)

15 0
10 0
50
0
60

Voltage (mV)

40
20
0
-20
-40
-60
-80

Gate pa ram
va lue

1.0
0.8

m gate (Na)

0.6

h ga te (Na)

0.4

n ga te (K)

0.2
0.0
0

4 Time (ms)

10

Interactive demonstration

(Integration of Hodgkin-Huxley equations

using Maple)

Interactive demonstration
> Ena:=55: Ek:=-82: El:= -59: gkbar:=24.34: gnabar:=70.7:
> gl:=0.3: vrest:=-69: cm:=0.001:
> alphan:=v-> 0.01*(10-(v-vrest))/(exp(0.1*(10-(v-vrest)))-1):
> betan:=v-> 0.125*exp(-(v-vrest)/80):
> alpham:=v-> 0.1*(25-(v-vrest))/(exp(0.1*(25-(v-vrest)))-1):
> betam:=v-> 4*exp(-(v-vrest)/18):
> alphah:=v->0.07*exp(-0.05*(v-vrest)):
> betah:=v->1/(exp(0.1*(30-(v-vrest)))+1):
> pulse:=t->-20*(Heaviside(t-.001)-Heaviside(t-.002)):
> rhsV:=(t,V,n,m,h)->-(gnabar*m^3*h*(V-Ena) +
>
gkbar*n^4*(V-Ek) + gl*(V-El)+
pulse(t))/cm:
> rhsn:=(t,V,n,m,h)-> 1000*(alphan(V)*(1-n) - betan(V)*n):
> rhsm:=(t,V,n,m,h)-> 1000*(alpham(V)*(1-m) - betam(V)*m):
> rhsh:=(t,V,n,m,h)-> 1000*(alphah(V)*(1-h) - betah(V)*h):

Interactive demonstration
> inits:=V(0)=vrest,n(0)=0.315,m(0)=0.042, h(0)=0.608;
> sol:=dsolve({diff(V(t),t)=rhsV(t,V(t),n(t),m(t),h(t)),
diff(n(t),t)=rhsn(t,V(t),n(t),m(t),h(t)),
diff(m(t),t)=rhsm(t,V(t),n(t),m(t),h(t)),
diff(h(t),t)=rhsh(t,V(t),n(t),m(t),h(t)),inits},
{V(t),n(t),m(t),h(t)},type=numeric, output=listprocedure);
> Vs:=subs(sol,V(t));
> plot(Vs,0..0.02);
> sol20:=dsolve({diff(V(t),t)=rhsV(t,V(t),n(t),m(t),h(t)),
diff(n(t),t)=rhsn(t,V(t),n(t),m(t),h(t)),
diff(m(t),t)=rhsm(t,V(t),n(t),m(t),h(t)),
diff(h(t),t)=rhsh(t,V(t),n(t),m(t),h(t)),inits},
{V(t),n(t),m(t),h(t)},type=numeric);
> with(plots):

Interactive demonstration
> J:=odeplot(sol20,[V(t),n(t)],0..0.02):

> display({J});
> pulse:=t->-2*(Heaviside(t-.001)-Heaviside(t-.002)):
> rhsV:=(t,V,n,m,h)->-(gnabar*m^3*h*(V-Ena) +
gkbar*n^4*(V-Ek) + gl*(V-El)+ pulse(t))/cm:
> sol2:=dsolve({diff(V(t),t)=rhsV(t,V(t),n(t),m(t),h(t)),
diff(n(t),t)=rhsn(t,V(t),n(t),m(t),h(t)),
diff(m(t),t)=rhsm(t,V(t),n(t),m(t),h(t)),
diff(h(t),t)=rhsh(t,V(t),n(t),m(t),h(t)),inits},
{V(t),n(t),m(t),h(t)},type=numeric);
> K:=odeplot(sol2,[V(t),n(t)],0..0.02,color=green):
> display({J,K});

Interactive demonstration
> L:=odeplot(sol20,[V(t),n(t)],0..0.02,numpoints=400,
color=blue):
> display({J,L});
> odeplot(sol20,[V(t),m(t)],0..0.02,numpoints=400);
> odeplot(sol20,[V(t),h(t)],0..0.02,numpoints=400);
> odeplot(sol20,[m(t),h(t)],0..0.02,numpoints=400);
> a:=0.7; b:=0.8; c:=0.08;
> rhsx:=(t,x,y)->x-x^3/3-y;
> rhsy:=(t,x,y)->c*(x+a-b*y);
> sol2:=dsolve({diff(x(t),t)=rhsx(t,x(t),y(t)),
diff(y(t),t)=rhsy(t,x(t),y(t)),x(0)=0,y(0)=-1},
{x(t),y(t)},type=numeric, output=listprocedure);
> xs:=subs(sol2,x(t)); ys:=subs(sol2,y(t));
> K:=plot([xs,ys,0..200],x=-3..3,y=-2..2,color=blue):
> J:=plot({[V,(V+a)/b,V=-2.5..1.5],[V,V-V^3/3,V=-2.5..2.2]}):
> plots[display]({J,K});

Interactive demonstration

(Fitzhugh-Nagamo simplification)