- Cable theory
Classical cable theory describes the development of mathematical models that can calculate the flow of electric current (and accompanying
voltage ) along passive ref|classical neuronal fibers (neurites) particularlydendrite s that receive synaptic (seesynapse ) inputs at different sites and times. Such estimates can be achieved by regarding dendrites and axons as cylinders composed of segments withcapacitance s c_m and resistances r_m combined in parallel (See Figure 1). The capacitance of a neuronal fiber comes about becauseelectrostatic forces are acting through the very thin phospholipid bilayer (See Figure 2). The resistances in series along the fiber r_l is due to thecytosol ’s significant resistance to movement ofelectric charge .History
Cable theory in computational neuroscience has roots leading back to the 1850s, when Professor William Thomson (later known as Lord Kelvin) began developing mathematical models of signal decay in submarine (underwater) telegraphic cables. The models resembled the partial differential equations used by
Fourier to describe heat conduction in a wire.The 1870s saw the first attempts by Hermann to model
axon alelectrotonus also by focusing on analogies with heat conduction. However it was Hoorweg who first discovered the analogies with Kelvin’s undersea cables in 1898 and then Hermann and Cremer who independently developed the cable theory for neuronal fibers in the early 20th century. Further mathematical theories of nerve fiber conduction based on cable theory were developed by Cole andHodgkin (1920s-1930s), Offner et.al. (1940), and Rushton (1951).Experimental evidence for the importance of cable theory in modeling real nerve
axon s began surfacing in the 1930s from work done by Cole, Curtis,Hodgkin ,Katz , Rushton, Tasaki and others. Two key papers from this era are those of Davis and Lorente de No (1947) andHodgkin and Rushton (1946).The 1950s saw improvements in techniques for measuring the electric activity of individual neurons. Thus cable theory became important for analyzing data collected from intracellular microelectrode recordings and for analyzing the electrical properties of neuronal dendrites. Scientists like Coombs, Eccles, Fatt, Frank, Fuortes and others now relied heavily on cable theory to obtain functional insights of neurons and for guiding them in the design of new experiments.
Later, cable theory with its mathematical derivatives allowed ever more sophisticated neuron models to be explored by workers such as Jack,
Christof Koch , Noble, Poggio, Rall, Redman, Rinzel, Idan Segev, Shepherd, Torre and Tsien.One important avenue of research became to analyze the effects of different synaptic input distributions over the dendritic (seedendrite ) surface of a neuron.Deriving the cable equation
r_m and c_m introduced above are measured per fiber-length unit (usually centimeter (cm)). Thus r_m is measured in
ohm s times centimeters (Omega imes cm ) and c_m in microfarads per centimeter (mu F/cm ). This is in contrast to R_m (Omega imes cm^2) and C_m (mu F/cm^2) , which represent the specific resistance and capacitance of the membrane measured within one unit area of membrane cm^2) . Thus if the radius "a" of the cable is known ref|fiberradius and hence its circumference 2 pi a , r_m and c_m can be calculated as follows:r_m = frac{R_m}{2 pi a } (1) c_m = C_m 2 pi a (2) This makes sense because the bigger the circumference the larger area for charge to escape through the membrane and the smaller resistance (we divide R_m by 2 pi a ); and the more membrane to store charge (we multiply C_m by 2 pi a .)In a similar vein, the specific resistance R_l of the cytoplasm enables the longitudinal intracellular resistance per unit length (Omega/cm ) r_l to be calculated as: r_l = frac{R_l}{pi a^2 } (3) Again a reasonable equation, because the larger the cross sectional area (pi a^2 ) the larger the number of paths for the current to flow through the cytoplasm and the less resistance.
To better understand how the cable equation is derived let's first simplify our fiber from above even further and pretend it has a perfectly sealed membrane (r_m is infinite) with no loss of current to the outside, and no capacitance (c_m = 0 .) A current injected into the fiber ref|linearity at position x = 0 would move along the inside of the fiber unchanged. Moving away from the point of injection and by using
ohms law (V=I R ) we can calculate the voltage change as: Delta V = -i_l r_l Delta x (4) If we let Delta x go towards zero and have infinitely small increments of x we can write (4) as: frac{partial V}{partial x} = -i_l r_l (5) or frac{1}{r_l} frac{partial V}{partial x} = -i_l (6) Bringing r_m back into the picture is like making holes in a garden hose. The more holes the more water will escape to the outside, and the less water will reach a certain point of the hose. Similarly in the neuronal fiber some of the current travelling longitudinally along the inside of the fiber will escape through the membrane.If i_m is the current escaping through the membrane per length unit (cm), then the total current escaping along y units must be y i_m . Thus the change of current in the cytoplasm Delta i_l at distance Delta x from position x=0 can be written as: Delta i_l = -i_m Delta x (7) or using continuous infinitesimally small increments: frac{partial i_l}{partial x}=-i_m (8) i_m can be expressed with yet another formula, by including the capacitance. The capacitance will cause a flow of charge (current) towards the membrane on the side of the cytoplasm. This current is usually referred to as displacement current (here denoted i_c .) The flow will only take place as long as the membrane's storage capacity has not been reached. i_c can then be expressed as: i_c = c_m frac{partial V}{partial t} (9) where c_m is the membrane's capacitance and frac{partial V}{partial t} is the change in voltage over time. The current that passes the membrane (i_r ) can be expressed as:i_r=frac{V}{r_m} (10) and because i_m = i_r + i_c the following equation for i_m can be derived if no additional current is added from an electrode:
frac{partial i_l}{partial x}=-i_m=-(frac{V}{r_m}+c_m frac{partial V}{partial t}) (11) where frac{partial i_l}{partial x} represents the change per unit length of the longitudinal current.
By combining equations (6) and (11) we get a first version of a cable equation:
frac{1}{r_l} frac{partial ^2 V}{partial x^2}=c_m frac{partial V}{partial t}+frac{V}{r_m} (12)
which is a second-order
partial differential equation (PDE.)By a simple rearrangement of equation (12) (see later) it is possible to make two important terms appear, namely the length constant (sometimes referred to as the space constant) denoted lambda and the time constant denoted au. The following sections focus on these terms.
The length constant
The length constant denoted with the symbol lambda (lambda) is a parameter that indicates how far a current will spread along the inside of a neurite and thereby influence the voltage along that distance. The larger lambda is, the farther the current will flow. The length constant can be expressed as:
lambda = sqrt frac{r_m}{r_l} (13)
This formula makes sense because the larger the membrane resistance (r_m ) (resulting in larger lambda ) the more current will remain inside the cytosol to travel longitudinally along the neurite. The higher the cytosol resistance (r_l ) (resulting in smaller lambda ) the harder it will be for current to travel through the cytosol and the shorter the current will be able to travel.It is possible (albeit not straightforward) to solve equation (12) and arrive at the following equation:
V_x = frac {V_0}{e^frac{x}{lambda (14)
Where V_0 is the depolarization at x=0 (point of current injection), e is the exponential constant (approximate value 2.71828) and V_x is the voltage at a given distance x from x=0 . When x=lambda then
frac{x}{lambda}=1 (15)
and
V_x = frac{V_0}{e} (16)
which means that when we measure V at distance lambda from x=0 we get
V_lambda = frac{V_0}{e} = 0.368 V_0 (17)
Thus V_lambda is always 36.8 percent of V_0 .
The time constant
Neuroscientists are often interested in knowing how fast the membrane potential V_m of a neurite is changing in response to changes in the current injected into the cytosol. The time constant au is an index that provides information about exactly that. au can be calculated as:
au = r_m c_m (18)
which seem reasonable because the larger the membrane capacitance (c_m ) the more current it takes to charge and discharge a patch of membrane and the longer this process will take. Thus membrane potential (voltage across the membrane) lags behind current injections. Response times vary from 1-2 milliseconds in neurons that are processing information that needs high temporal precision to 100 milliseconds or longer. A typical response time is around 20 milliseconds.
The cable equation with length and time constants
If we multiply equation (12) by r_m on both sides of the equal sign we get:
frac{r_m}{r_l} frac{partial ^2 V}{partial x^2}=c_m r_m frac{partial V}{partial t}+ V (19)
and recognize lambda^2 = frac{r_m}{r_l} on the left side and au = c_m r_m on the right side. The cable equation can now be written in its perhaps best known form:
lambda^2 frac{partial ^2 V}{partial x^2}= au frac{partial V}{partial t}+ V (20)
References
* "Methods in Neuronal Modeling, From synapses to networks", edited by
Christof Koch and Idan Segev. ISBN 0-262-61071-X
* "Biophysics of Computation, Information Processing in Single Neurons", byChristof Koch . ISBN 0-19-518199-9
* "Studies from the Rockefeller Institute for Medical Research", by Davis,L.,Jr. and Lorente de No. R (1947). 131: 442-496.
*"The electrical constants of a crustacean nerve fibre", byHodgkin , A.L. and Rushton, W.A.H. (1946). Proc. Roy. Soc. London. B 133: 444-479.Notes
:Notes::1 Passive here refers to the membrane resistance being voltage-independent. However recent experiments (Stuart and Sakmann 1994) with dendritic membranes shows that many of these are equipped with voltage gated
ion channel s thus making the resistance of the membrane voltage dependent. Consequently there has been a need to update the classical cable theory to accommodate for the fact that most dendritic membranes are not passive.:2 Classical cable theory assumes that the fiber has a constant radius along the distance being modeled.
:3 Classical cable theory assumes that the inputs (usually injections with a micro device) are currents which can be summed linearly. This linearity does not hold for changes in synaptic membrane conductance.
Wikimedia Foundation. 2010.