 |
INTRODUCTION |
By changing the waveform of action potentials and
excitability, serotonin (5-HT)-induced modulation of membrane currents
in the sensory neurons that mediate the gill- and tail-withdrawal reflexes of Aplysia is believed to be a key mechanism
underlying short-term heterosynaptic facilitation (for recent review,
see Byrne and Kandel 1996
). The first-discovered
"serotonin-sensitive" current was a novel K+
current that was termed the S current
(IK,S) (Klein et al.
1982). Acting via elevated levels of intracellular adenosine
cyclic monophosphate (cAMP) and the subsequent activation of protein
kinase A (PKA), application of 5-HT decreased the magnitude of
IK,S (Fig.
1A) (Baxter and Byrne
1990a; Bernier et al. 1982; Jarrard et
al. 1993; Ocorr and Byrne 1985; Pollock
and Camardo 1987; Pollock et al. 1985;
Shuster and Siegelbaum 1987; Shuster et al.
1985; Siegelbaum et al. 1982; Sugita et
al. 1997a; Walsh and Byrne 1989). Because 5-HT
produced a broadening of the action potential and enhanced the
excitability of sensory neurons, both of these changes originally were
attributed to the reduction of IK,S
(e.g., Klein et al. 1986). It has become clear, however,
that the mechanisms underlying the 5-HT-induced changes in the
biophysical properties of sensory neurons are more complicated than the
activation of a single second-messenger/protein kinase system and the
modulation of a single K+ current.
View larger version (25K):
[in this window]
[in a new window]
|
Fig. 1.
Serotonergic modulation of sensory neurons. A: early
view of serotonergic modulation. In this scheme, adenosine cyclic
monophosphate (cAMP) and protein kinase A (PKA) are the only second
messenger/kinase system activated by serotonin (5-HT). PKA-mediated
reduction of a S-type K+ channel conductivity,
gK,S, is considered to underlie the
modulation of 2 physiological properties of the sensory neuron: spike
broadening and enhanced excitability. B: current view of
serotonergic modulation. Serotonin binds to  2 distinct receptors,
which in turn lead to the activation of 2 kinases, PKA and protein
kinase C (PKC). These kinases modulate a host of membrane conductances.
There is overlap between the 2 kinases and their targets as well as
cross-talk between the 2 second-messenger/kinase systems. +,
enhancement or positive regulation;  , decrease or negative
regulation. See text for additional details.
|
|
Several additional components contribute to 5-HT-induced modulation of
action potentials and excitability of sensory neurons (Fig.
1B). First, elevated levels of cAMP modulate at least three currents in addition to IK,S.
Walsh and Byrne (1989) described a slow component of the
Ca2+-activated K+ current
(IK,Ca-S) that was active near the
resting potential of the cell and that was decreased by intracellular
injection of cAMP (or application of 5-HT). Braha et al.
(1993; see also Edmonds et al. 1990;
Eliot et al. 1993) reported that intracellular injection
of cAMP (or application of 5-HT) enhanced a dihydropyridine-sensitive and slowly inactivating component of the Ca2+
current similar to the L-type Ca2+ current
(ICa-L). Baxter and Byrne
(1989; see also White et al. 1994) reported that
application of 5-HT decreased the conductance and slowed the kinetics
of a large, steeply voltage-dependent K+ current
(IK-V). Goldsmith and
Abrams (1992) reported that application of analogues of cAMP
partially mimicked the 5-HT-induced slowing of the activation kinetics
of IK-V. Similarly, Hochner and
Kandel (1992) reported that specific blockers of PKA partially
blocked the 5-HT-induced slowing of the activation kinetics of
IK-V. These results indicate that the
5-HT-induced modulation of IK-V is
mediated, at least in part, by the cAMP/PKA system. Moreover, studies
of Goldsmith and Abrams (1992; see also Shuster
et al. 1991) suggested that the originally described
IK,S consisted of two components, a
moderately voltage-dependent and slowly-activating component (IK,S-V), and an instantaneous (i.e.,
time-independent) "steady-state" component that was activated at
the resting potential (IK,S-I). Thus
5-HT-induced increases in the levels of cAMP can lead to the modulation
of a complex array of membrane currents with diverse biophysical properties.
Second, in addition to the cAMP/PKA system, application of 5-HT
activates protein kinase C (PKC) (Sossin 1997;
Sossin and Schwartz 1992; Sossin et al.
1994; see also Sacktor and Schwartz 1990).1
Moreover, pharmacological activation of PKC [i.e., application of
active phorbol esters such as 4
12-deoxyphorbol 13-isobutyrate (DPB), 4-phorbol 12,13-diacetate (PDAc), phorbol dibutyrate (PDBu), phorbol myristate (PMA)] mimics some aspects of 5-HT-induced
modulation of membrane currents. Braha et al. (1993)
reported that activation of PKC mimicked the 5-HT-induced increase of
ICa-L and that blockers of PKC blocked
5-HT-induced modulation of ICa-L.
Sugita et al. (1994a) found that activation of PKC
mimicked and partially occluded the 5-HT-induced modulation of
IK-V.2
Thus 5-HT-induced modulation of membrane current appears to involve at
least two kinase systems (i.e., PKA and PKC) that act on an array of
membrane conductances.
Third, recent studies indicate that there is cross-talk between the PKC
and PKA cascades. Sugita et al. (1997a) reported that activation of PKC induced an increase in the level of cAMP in sensory
neurons. It is likely that the PKC-induced increase in cAMP leads to
activation of PKA and subsequent PKA-mediated modulation of membrane
currents. For example, activators of PKC induced a modest increase in
the excitability of sensory neurons, thereby partially mimicking a well
know cAMP effect (Sugita et al. 1997a; see also
Baxter and Byrne 1990a; Manseau et al.
1998). In contrast, biochemical evidence indicates that
translocation of PKC was not induced by analogues of cAMP
(Sacktor and Schwartz 1990). These results suggest that
some of the biophysical effects that have been attributed directly to
the PKC cascade may be indirect effects that result from cross-talk
between the PKC and PKA cascades.
Because of overlapping responses to electrical and pharmacological
stimulation and because of cross-talk between second messenger/protein kinase cascades, it is difficult to accurately assess the how the
modulation of specific membrane currents (i.e.,
IK,S,
IK-V, IK,Ca-S,
ICa-L) or how the PKA- versus
PKC-mediated modulation of membrane currents contribute to 5-HT-induced
spike broadening and excitability enhancement. The present study
addresses these issues by developing and analyzing a
Hodgkin-Huxley-type mathematical model of the sensory neuron. First,
previously published voltage-clamp data were used to develop
mathematical descriptions of the ionic conductances in the somata of
sensory neurons. Second, simulations investigated whether the known
modulatory actions of 5-HT on membrane currents are sufficient to
account for the empirically observed increases in spike duration and
excitability. Third, simulations investigated the relative
contributions of individual currents to the overall effects of 5-HT.
Fourth, simulations investigated the consequences of cross-talk between
the PKC and PKA cascades. Finally, an empirical study was conducted to
test the predicted contribution of
IK,Ca-S to accommodation. The results
indicated that the model was sufficient to simulate the basic features
of action potential and excitability data from sensory neurons;
concurrent modulation of IK,S and
IK,Ca-S contributed significantly to
5-HT-induced increases in excitability; and modulation of
IK-V contributed significantly to
5-HT-induced spike broadening. The simulations also provided several
predictions that can help guide future experimental analysis, and the
results illustrate that the actions of modulatory agents and second
messengers cannot be understood on the basis of their direct effects
alone. It is also necessary to consider indirect effects that occur
through cross-talk between second-messenger systems and
Ca2+-dependent processes.
| |
METHODS |
Model development
GENERAL FEATURES.
The simulations were performed with SNNAP (Simulator for Neural
Networks and Action Potentials) (Ziv et al. 1994).
Version 5 of SNNAP was used and the software was run under the Windows 95/NT operating systems on PC-type microcomputers (Baxter and Byrne 1999). The Euler method with a fixed time step of 30-µs was used for numerical integration. When simulations were begun, there
typically was a small (<500 µV), brief (~1 s) transient before the
resting membrane potential settled to its steady-state value (50 mV).
To avoid analysis during this or any other transient, 10 s of
simulated time was allowed to elapse before data were taken.
The soma of a sensory neuron was modeled as a sphere 20 µm in
diameter that had a surface area of 1.2 × 104 µm2. This surface
area included a factor of 10 to account for membrane invagination
(Gorman and Mirolli 1972; Mirolli and Talbott
1972). In the present study, we assumed the axon did not make a
significantly contribution to the biophysical properties of the sensory
neuron or to their responses to 5-HT. This assumption was based, in
part, on previously published studies of action potentials,
excitability, and their modulation by 5-HT in isolated somata and in
ganglia preparations (cf. Sugita et al. 1992, 1997).
Apparently identical results have obtained in both types of
preparation. The model consisted of a membrane model represented by an
equivalent circuit (Fig. 2A)
coupled to equations that describe the regulation of intracellular
Ca2+ (Fig. 2B). The membrane model was
composed of a membrane capacitance (CM), which was assumed to be 1 µF/cm2 (Almers 1978), in
parallel with three inward currents (1 Na+ and 2 Ca2+ currents), six K+
currents, and a leak current. The ionic currents in the model were
described by Hodgkin-Huxley-type equations in which generalized Boltzman-type equations defined the voltage- and time-dependent activation and inactivation of conductances. In addition, the descriptions of four conductances (the 2 Ca2+
conductances and 2 of the K+ conductances) were
expanded to include Ca2+-dependent regulation
(e.g., Ca2+-dependent inactivation and
activation, respectively; Fig. 2B). The details of the
equations and parameters are given in the APPENDIX and
Table 1.
View larger version (39K):
[in this window]
[in a new window]
|
Fig. 2.
Model of a sensory neuron. A: equivalent electrical
circuit of the cell membrane. Linear conductances are indicated by
resistors and nonlinear (i.e., voltage-gated) conductances are
indicated by variable resistors. Each conductance is associated with a
equilibrium potential (E). In parallel with the membrane
capacitance (CM) are 10 ionic conductances:
a leakage conductance (gL); a fast
Na+ conductance (gNa); a slowly
inactivating L-type Ca2+ conductance
(gCa-L); a rapidly inactivating N-type
Ca2+ conductance (gCa-N); a fast
transient A-type K+ conductance
(gK-A); a fast voltage- and
Ca2+-activated K+ conductance
(gK,Ca-F); a slow voltage-independent
Ca2+-activated K+ conductance
(gK,Ca-S); an instantaneous, modestly
voltage-dependent component of the S ("serotonin")-type
K+ conductance (gK,S-I); a
voltage- and time-dependent component of the S-type K+
conductance (gK,S-V); and a delayed, steeply
voltage-dependent K+ conductance
(gK-V). The application of extrinsic
stimulating and/or bias currents is represented by
IStim. B: intracellular
regulatory pathways of the model. Model also incorporated a description
of an intracellular pool of Ca2+. Calcium influx via
gCa-L and gCa-N
contributed to the Ca2+ pool. Dynamics of the
Ca2+ pool were modeled as a first-order process. The
intracellular pool of Ca2+, in turn, regulated several
membrane conductances. Both gCa-L and
gCa-N were regulated negatively by
[Ca2+] (i.e., Ca2+-dependent inactivation),
and both gK,Ca-S and gK,Ca-F were activated
(i.e., positively regulated) by [Ca2+]. Although not
explicitly modeled as biochemical components, the regulatory
relationships of PKC and PKA are illustrated. See text for details. +,
enhancement or positive regulation; , decrease or negative
regulation.
|
|
Whenever possible, previously published voltage-clamp data from sensory
neurons were used to formulate the descriptions of the currents. These
data have been collected from sensory neurons in both abdominal and
pleural ganglia. In the present study, these two sets of sensory
neurons were assumed to have identical biophysical properties and
responses to 5-HT. This assumption was based, in part, on the work of
Wright and Kirschman (1995), who made direct comparisons
of the firing properties and effects of 5-HT on siphon versus tail
sensory neurons. They concluded that the properties and responses to
5-HT of these two classes of sensory neurons were indistinguishable.
Thus we (and others) generally have considered these two sets of
sensory neurons to be experimentally interchangeable. Voltage-clamp
data that illustrate the voltage- and time-dependent characteristics
are available for many of the membrane currents, including the L- and
N-type Ca2+ currents
(ICa-L,
ICa-N, respectively) (Braha et
al. 1993; Edmonds et al. 1990; Eliot et
al. 1993), a transient A-type K+ current
(IK-A) (Baxter and Byrne 1989,
1990b), a steeply voltage-dependent K+
current (IK-V) (Baxter and
Byrne 1989, 1990b; Goldsmith and Abrams 1992;
Hochner and Kandel 1992; White et al.
1994), a fast component of the
Ca2+-activated K+ current
(IK,Ca-F) (Baxter and Byrne
1989; Critz and Byrne 1992; Walsh and
Byrne 1989; see also Shuster et al. 1991), and a
time- and voltage-dependent component of the S-type
K+ current
(IK,S-V) (Baxter and Byrne
1989, 1990a,b; Braha et al. 1993; Hochner
and Kandel 1992; Klein et al. 1982;
Pollock et al. 1985; Sugita et al. 1994a;
Walsh and Byrne 1989). In general, the methods used to
develop mathematical descriptions of these current followed those of
Byrne (1980a,b) and White et al. (1994). Briefly, published voltage-clamp records were digitized with a commercially available software packages (SigmaScanPro; SPSS, Chicago,
IL), nonlinear parameter estimation (SigmaPlot) was performed on the
digitized data at each potential, and a Hodgkin-Huxley type description
of each current was formulated that was consistent with the parameters
derived from the voltage-clamp traces. The model was developed to
simulate the properties of sensory neurons at 15°C. Thus when
necessary, conductances and time constants that were derived from
empirical data were scaled to 15°C using a
Q10 of 2 (Adams and Gage
1979b; Andresen and Brown 1979; Gorman and Marmor 1970, Johnston 1980; Joyner
1981; Partidge and Connor 1978; Romey et
al. 1980; Thompson et al. 1986; see also
Sah et al. 1988; Thompson et al. 1985).
In the present study, the parameters for
IK-V were adjusted to duplicate a
previous empirical-based model of IK-V
(White et al. 1994). Results from patch-clamp studies indicated that the channels mediating
IK,S-I have a modest voltage dependency, a low probability of being open (P < 0.1)
and a single-channel conductance of ~50 pS (Brezina et al.
1987; Pollock and Camardo 1987; Shuster
et al. 1991; Siegelbaum et al. 1982, 1986; see
also Goldsmith and Abrams 1992). The conductance for
IK,S-I was calculated assuming an
average of 0.02 pores were open per square µm [i.e., gK,S-I = (0.02 × 50 pS/µm2) × 1.2 × 104 µm2 = 0.012 µS; see
Table 1] and the voltage dependency of
IK,S-I was matched to that of
single-channel currents. Finally, the leakage conductance was set to
produce a input resistance of 27 M
, which was based on empirical
data (see RESULTS). In the present model, 61% of the
outward current that flowed at the resting membrane potential was
carried by gK,S-I, 25% was carried by
gK,S-V and 10% was carried by carried
by
gK,Ca-S.3
Other aspects of the model were less constrained by available data.
First, the Na+ current
(INa) in sensory neurons has not been
characterized. Na+ currents have been
characterized in other neurons of Aplysia, however
(Adams and Gage 1979a; Byrne 1980a,b;
Farquharson and Jahan-Parwar 1984; Fieber
1995; Gilly et al. 1997; see also
Canavier et al. 1991). Thus a Hodgkin-Huxley-type
description of INa for the present
sensory neuron model was formulated that was consistent with
INa in these other cells. The
parameters for INa in the present sensory neuron model were adjusted, within limits set by the published examples of INa in Aplysia
neurons, to produce a current typical of sodium currents observed in
Aplysia neurons and to match the waveform of empirically
observed action potentials in sensory neurons (e.g., Baxter and
Byrne 1990a; Braha et al. 1993; Critz et
al. 1991; Eliot et al. 1994; Ghirardi et
al. 1992; Goldsmith and Abrams 1992;
Hochner and Kandel 1992; Mercer et al.
1991; Stark et al. 1996; Sugita et al.
1992, 1994b; Wright and Kirschman 1995).
Second, the kinetics and voltage dependency of the slow component of
the Ca2+-activated K+
current (IK,Ca-S) have not been
characterized extensively. This current appears to be active at
potentials near the resting potential of sensory neurons and to have
slow kinetics, however. For example, in voltage-clamp studies of
sensory neurons, low concentrations (e.g., 2-5 mM) of
tetraethylammonium (TEA), which selectively blocks
IK,Ca, blocked a component of membrane
current that was active at membrane potentials as hyperpolarized as
40 mV and that activated with an apparent time constant >2 s (Baxter
and Byrne, unpublished observations; Walsh and Byrne
1989). In the present model, the model of
IK,Ca-S was developed without explicit voltage and time dependency. Rather, the activation of
IK,Ca-S and its dynamics were
regulated solely by the concentration and dynamics of the intracellular
pool of Ca2+ (see the APPENDIX). The
value for the maximum conductance of
IK,Ca-S (i.e.,
gK,Ca-S in Table 1) was estimated from
the data of Walsh and Byrne (1989), who found that
manipulations blocking IK,CA (e.g.,
substituting Ba2+ for Ca2+;
blocking Ca2+ currents; applying low
concentrations of TEA) also induced an inward shift in the steady-state
holding current and a decrease in the membrane conductance. Their data
suggested that at membrane potentials near 30 mV, the magnitude of
IK,CA-S was ~3 nA. In the present
model, voltage clamping the membrane potential at 30 mV induced a
steady-state IK,CA-S of 2.7 nA.
Third, the Ca2+ dependency of
ICa-L and
ICa-N has not been investigated in
sensory neurons of Aplysia. In other systems where it has
been examined, however, L- and N-type Ca2+
currents are regulated both by voltage and intracellular concentrations of Ca2+ (for reviews, see Eckert and Chad
1984; Hille 1992; Tsien et al.
1988). Thus in present the sensory neuron model, the
descriptions of ICa-L and
ICa-N were extended to include
functions describing an inverse relationship between the
Ca2+ conductances
(gCa-L and
gCa-N) the concentration of
Ca2+ in the pool. This feature of the model made
an important contribution to the firing properties of the sensory
neuron under conditions of enhanced excitability, which suggests that
this mechanism warrants additional experimental investigation (see
following text).
SIMULATING 5-HT-INDUCED MODULATION OF MEMBRANE CURRENTS.
To simulate 5-HT-induced modulation of membrane currents, selected
parameters of the model were set to the values indicated in the column
labeled "5-HT-induced modulation" in Table
2. The actions of 5-HT were simulated by
decreasing the conductances of IK,S
(both IK,S-V and
IK,S-I) to ~50% of their control
values and decreasing the conductance of
IK,Ca-S to ~20% of its control value. The magnitudes of these changes were estimated from published data. Cell-attached patch-clamp studies of the channels mediating IK,S indicated that application of
5-HT or injection of cAMP closed between 46 and 53% of the channels in
any given patch (Shuster et al. 1985; Siegelbaum
et al. 1986). Thus PKA-mediated and 5-HT-induced modulation of
IK,S were simulated by decreasing the
conductances of IK,S-I and
IK,S-V by ~50%. Similarly, the
voltage-clamp studies of Walsh and Byrne (1989)
indicated that application of 5-HT or injection of cAMP blocked between
30 and 100% of the total IK,Ca-S (the
average was ~77 ± 8%). Thus PKA-mediated and 5-HT-induced modulation of IK,Ca-S were simulated
by decreasing the conductances of
gK,Ca-S to 23% of its control value.
In addition, the actions of 5-HT were simulated by increasing the
conductance of ICa-L to 250% of its
control value, which was based on published data indicating that
application of 5-HT induced an average increase in
ICa of 220 ± 36% (Eliot
et al. 1993). Finally, the actions of 5-HT were simulated by
modifying the properties of IK-V. This modification was more complex than simply decreasing
gK-V, however. White et al.
(1994) reported that in addition to decreasing
gK-V, 5-HT slowed the kinetics for its
activation and inactivation. Thus the mathematical description of
5-HT-induced modulation of IK-V
included increases in the time constants of activation and inactivation
(
A and B,
respectively). The magnitude of these changes in the present study were
adjusted so as to reproduce the data of White et al.
(1994). This ensemble of modifications to the model was assumed
to represent the maximal effects of 5-HT. This assumption was based on
previously published dose-response curves for 5-HT-induced modulation
of sensory neurons (Jarrard et al. 1993; Ocorr
and Byrne 1985; Stark et al. 1996; see also Bacskai et al. 1993). The reported
EC50 values for the actions of 5-HT ranged from
0.8 to 14 µM and average EC50 was 8 ± 3 µM. The previously published empirical studies, which provided the data for the present model, used an average concentration of 30 ± 5 µM 5-HT. Thus we assumed that a maximal effect was achieved in the
majority of previous experimental studies, and the results of these
studies were combined. Finally, this ensemble of modifications to the
model that reflect 5-HT-induced modulation represented the steady-state
actions of 5-HT. Thus the present study did not simulate the time
dependency of 5-HT modulation (for review, see Byrne and Kandel
1996).
SIMULATING THE ACTIVATION OF PKA.
Although it is not clear from the available empirical results that all
of the modulatory changes that are induced by elevated levels of cAMP
are mediated via activation of PKA (e.g., Braha et al.
1993), in the present study, these modulatory changes were referred to collectively as "PKA-mediated modulation." To simulate the modulatory actions of PKA, selected parameters of the model were
set to the values indicted in the column labeled "PKA-mediated modulation" in Table 2. Because PKA is believed to mediate many of
the actions of 5-HT, many of the PKA-mediated parameter changes were
identical to those described above for 5-HT-induced modulation. For
example, the conductances for IK,S
(both IK,S-I and
IK,S-V) and
IK,Ca-S were reduced to ~50 and
~20% of their control values, respectively, and the conductance for
ICa-L was increased to 250% of its
control value. The modulation of IK-V
was different, however. In the presence of 5-HT, the conductance as
well as the activation and inactivation time constants
(A and B)
were modulated. The available empiric data suggest that PKA only
modulates A, and this modulation is
equivalent to ~64% of that produced by 5-HT (Goldsmith and
Abrams 1992; Hochner and Kandel 1992). Thus the actions of PKA on IK-V were simulated
by slowing its activation kinetics to a level 64% of that used to
simulate the actions of 5-HT.
SIMULATING ACTIVATION OF PKC.
To simulate the modulatory actions of PKC, selected parameters of the
model were set to the values indicted in the column labeled
"PKC-mediated modulation" in Table 2. Activation of PKC has been
found to partially mimic and occlude the modulatory actions of 5-HT on
some membrane current [e.g., ICa-L
(Braha et al. 1993) and
IK-V (Sugita et al.
1994a)]. Thus some of the PKC-mediated parameter changes were
similar or identical to those described earlier here for 5-HT-induced
modulation. Specifically, gCa-L was
increased to 250% of its control value,
gK-V was reduced, and the kinetics of
its inactivation were slowed to match the data of White et al.
(1994). As suggest by the data of Sugita et al.
(1994a), the actions of PKC on
IK-V were simulated by slowing its
activation kinetics to a level 75% of that used to simulate the
actions of 5-HT and by modifying the conductance and inactivation kinetics of IK-V to levels identical
to those used to simulated the actions of 5-HT. In addition, to its
direct effects on membrane conductances, activation of PKC stimulated
an increase in the intracellular levels of cAMP equivalent to ~60%
of the increase in cAMP that was induced by 5-HT (Sugita et al.
1997a). Thus the modulatory effects of PKC also included
changes to conductances that were modulated by elevated levels of cAMP,
such as gK,S-I, gK,S-V, and
gK,Ca-S. The simulated actions of PKC
decreased these conductances to a level equivalent to ~60% the
PKA-mediated modulation (see Table 2).
In vitro preparation
Experimental procedures to measure the excitability of
sensory neurons have been described in detail previously (Baxter
and Byrne 1990a). Briefly, all experiments were performed on
clusters of somata of sensory neurons that were surgically isolated
from the ventrocaudal cluster of pleural ganglia in A. californica. Dissections were performed after anesthetizing the
animals by injecting a volume of isotonic MgCl2
equal to about one-half of the volume of the animal. An isolated
cluster was pinned to the floor of a recording chamber, which was
coated with a silicon elastomer and had a volume of ~300 µl. The
static bathing solution of artificial sea water (ASW; Instant Ocean,
Aquarium Systems, Mentor, OH) was buffered to pH 7.6 with 10 mM Trizma
(Sigma Chemical, St. Louis, MO) and was maintained at 15°C.
Conventional two-electrode current-clamp techniques were used. Sensory
neurons were impaled with two glass capillary microelectrodes that were
filled with 3 M potassium acetate and that had resistances of 2-6
M. The membrane potential of the sensory neuron was monitored and
was maintained at 45 mV by manually adjusting the constant DC current output of the current passing electrode. Excitability was measured by
counting the number of action potentials elicited during a 1-s, 2-nA
constant-current pulse. These stimulating current pulses were separated
by 60 s. To ensure that the responses to the stimulating current
were stable, at least three examples of excitability were recorded
before and after bath application of TEA (Eastman Kodak, Rochester,
NY). Small, concentrated aliquots TEA were added to the bath such that
the final bath concentration of TEA was 2 mM. Data were collected from
the last stimulus in ASW and from the first stable response after
bath application of TEA.
| |
RESULTS |
Simulating membrane currents, action potentials, and excitability
in control conditions
The first test of the model was to examine how well it simulated
the biophysical properties of sensory neurons under control conditions.
To simulate control conditions, the parameters of the model were set
the values indicated in Table 1. These values produced a model sensory
neuron with a resting membrane potential of 50 mV and an input
resistance of 27 M. (The input resistance was measured by injecting
a hyperpolarizing current pulse from a resting potential of 50 mV.)
These values are within the range of previously reported empirically
measured values. A survey of the published literature indicated that in
vitro preparations of sensory neurons have resting membrane potentials
ranging from 38 to 55 mV and input resistances ranging from 10 to
50 M. The available published data suggested that sensory neurons
have an average resting potential of about 48 mV and an average input resistance of ~27 M (Baxter and Byrne 1989,
1990a,b; Cleary et al. 1998; Pollock et
al. 1985; Walsh and Byrne 1989; White et al. 1994; Wright and Kirschman
1995).4
Figure 3 illustrates simulated membrane
currents that were elicited by voltage-clamp protocols similar to those
used in previous empiric studies. The current responses of the model
were in general agreement, both in time course and magnitude, with
published examples of isolated ionic currents in sensory neurons (cf.
Baxter and Byrne 1989, 1990a,b; Braha et al.
1993; Edmonds et al. 1990; Eliot et al.
1993; Goldsmith and Abrams 1992; Hochner
and Kandel 1992; Klein et al. 1980, 1982;
Pollock et al. 1985; Scholz and Byrne 1987; Sugita et al. 1994a,b; Walsh and
Byrne 1989; White et al. 1994; see also
Adams and Gage 1979a; Byrne 1980a;
Farquharson and Jahan-Parwar 1984, Fieber
1995; Gilly et al. 1997).
View larger version (23K):
[in this window]
[in a new window]
|
Fig. 3.
Voltage-clamp simulations of membrane currents in control conditions.
Membrane currents were described by voltage-gating equations of the
Hodgkin-Huxley type and were modified when necessary to include
Ca2+ dependence. With the exception of
INa, the model currents were based on
experimental data from Aplysia sensory neurons and
corresponded well to experimentally measured currents both in time
course and magnitude. Model of INa was
derived from experimental data for Na+ current in other
identified neurons of Aplysia, and the parameters were
adjusted to match the waveform of the action potential of the sensory
neuron. Traces illustrate simulated current responses elicited by
200-ms voltage-clamp steps from a holding potential of 70 mV. To
ensure that steady-states conditions existed in the model, 10 s of
simulated time was allowed to pass before any experimental manipulation
and data collection in this and all subsequent figures.
A: simulation of INa elicited
by voltage-clamp steps to 0 mV (a), 10 mV (b), and 20 mV (c). Note, to
illustrate the fast kinetics of INa, only
the 1st 15 ms of the voltage-clamp step are illustrated.
B: simulation of ICa-N
elicited by voltage-clamp steps to 20 mV (a), 0 mV (b), and 20 mV
(c). C: simulation of IK-V
elicited by voltage-clamp steps to 0 mV (a), 10 mV (b), and 20 mV (c).
D: simulation of IK,S
elicited by voltage-clamp steps to 0 mV (a), 10 mV (b), and 20 mV (c).
These currents are a combination of both the instantaneous
(IK,S-I) and the voltage-dependent
(IK,S-V) components of
IK,S. E: simulation of
ICa-L elicited by voltage-clamp steps to
20 mV (a), 0 mV (b), and 20 mV (c). F: simulation of
IK-A elicited by voltage-clamp steps to 20
mV (a), 10 mV (b), and 0 mV (c). G: simulation of
IK,Ca elicited by voltage-clamp steps to 0 mV (a), 10 mV (b), and 20 mV (c). These currents are a combination of
both the fast (IK,Ca-F) and the slow
(IK,Ca-S) components of
IK,Ca.
|
|
The simulated voltage-clamp experiments illustrated that the model
accurately reproduced the data from which it was derived. The more
complex biophysical properties of sensory neurons (e.g., the waveform
of the action potential and its excitability), however, emerge from
interactions among this ensemble of membrane currents and from
interactions between the membrane conductances and the intracellular
concentration of Ca2+. To examine how well the
present model simulated these emergent properties, single actions
potentials were elicited with a brief (3 ms) depolarizing current pulse
(15 nA) (Fig. 4A) and the
excitability of the cell was measured as the number of spikes elicited
by a series of 1-s depolarizing current pulses of increasing amplitude (Fig. 4B). These techniques closely mimicked protocols used
in previous experimental studies (e.g., Baxter and Byrne
1990a; Braha et al. 1993; Hochner and
Kandel 1992; Stark et al. 1996; Sugita et
al. 1992; Wright and Kirschman 1995) and allowed
for analysis of the waveform of the action potential without
contamination from the stimulating current.
View larger version (12K):
[in this window]
[in a new window]
|
Fig. 4.
Simulations of an action potential and excitability in control
conditions. Responses of the model sensory neuron to stimulation
corresponded well to experimentally measured action potentials and
excitability both in time course and magnitude. A:
resting membrane potential of the model sensory neuron was 50 mV.
Single spike was elicited by injecting a 3-ms, 15-nA depolarizing
current pulse (bar). Action potential had a peak amplitude of ~41 mV
and a duration of 4.9 ms. Duration of the spike was measured as the
time between the peak of the spike and the point of the repolarizing
phase at which the membrane potential was 10% of the peak amplitude.
B: excitability of the model sensory neuron was measured
as the number of action potentials elicited by a series of 1-s
depolarizing current pulses (bar) of increasing magnitude (1 nA,
B1; 2 nA, B2; and 3 nA,
B3). In all 3 examples, the resting membrane potential
of the model sensory neurons was 50 mV. Response of the model cell
accommodated during the sustained depolarization, and thus only a brief
burst of spikes was elicited at the beginning of the stimulus.
|
|
From the resting potential of 50 mV, the model produced an action
potential that reached a voltage of ~41 mV at its peak (i.e., the
spike had a total amplitude of 91 mV; Fig. 4A). The duration
of the simulated spike, which was measured as the time between the peak
voltage and the point on the falling phase at which the membrane
potential was 10% of the peak, was 4.9 ms. There are many examples of
sensory neuron action potentials in the published literature with which
to compare the results of the present simulation (e.g., Baxter
and Byrne 1990a; Braha et al. 1993; Critz
et al. 1991; Eliot et al. 1994; Ghirardi
et al. 1992; Goldsmith and Abrams 1992;
Hochner and Kandel 1992; Jarrard et al.
1993; Klein 1993; Mercer et al.
1991; Stark et al. 1996; Sugita et al.
1992, 1994b, 1997b; Wright and Kirschman 1995). Results from these empirical studies indicated that in vitro
preparations of sensory neurons generally have an action potential with
a total amplitude of ~90 ± 14 mV and a duration of ~5.1 ± 2.7 ms (means ± SE). In response to 1-s depolarizing current
pulses, the model sensory neuron exhibited accommodation similar to
that observed empirically (Fig. 4B). The simulated responses
to 1-, 2-, and 3-nA depolarizing current pulses were 1, 3, and 6 spikes, respectively. A survey of the published literature indicated
that in vitro preparations of sensory neurons generally produced an
average of ~1.8 ± 1.4 spikes in response to a 1-nA pulse;
~3.6 ± 0.9 spikes in response to a 2-nA pulse and ~5.4 ± 1.3 spikes in response to a 3-nA pulse (cf. Baxter and Byrne
1990a; Braha et al. 1993; Cleary et al. 1998; Critz et al. 1991; Dale et al.
1987; Eliot et al. 1994; Ghirardi et al.
1992; Goldsmith and Abrams 1992; Hochner
and Kandel 1992; Jarrard et al. 1993;
Klein et al. 1986; Manseau et al. 1998; Mercer et al. 1991; Stark and Carew 1999;
Stark et al. 1996; Sugita et al. 1992;
Wright and Kirschman 1995; Wright et al. 1996). The close agreement between the simulated responses of the model (i.e., membrane currents, spike waveform, and excitability) and empiric results indicated the mathematical descriptions of the available empirical data were sufficient to reproduce several key biophysical properties of sensory neurons in control conditions and that additional simulations of the model may provide insights into the mechanisms underlying serotonergic modulation of spike duration and excitability of sensory neurons.
Simulating serotonergic modulation of membrane currents, action
potentials, and excitability
A second test of the model was to examine how well it simulated
the 5-HT-induced modulation of the biophysical properties of sensory
neurons. The simulated actions of 5-HT induced a steady-state depolarization of the resting membrane potential of ~4.1 mV and an
increase in the input resistance of the model sensory neuron to ~34
M (i.e., an increase to ~126% of the control value). A survey of
previously published empirical results indicated that in sensory
neurons, 5-HT induces depolarizations ranging from 2.9 to 5.7 mV (the
average depolarization was ~4.5 mV) and increases in input resistance
ranging from 110 to 140% of control values (the average increase was
~130% of control) (cf. Braha et al. 1993;
Stark et al. 1996; Walsh and Byrne 1989;
Wright and Kirschman 1995). The simulated responses were
in general agreement with empirical observations in that 5-HT induced a
decrease in resting membrane conductance and a depolarizing of the
resting membrane potential.
SEROTONERGIC MODULATION OF SPIKE DURATION.
To allow for direct comparisons between action potentials (and
measurements of excitability) simulated in control conditions and in
the simulated presence of 5-HT, a constant bias current (0.11 nA) was
applied to the model to maintain the resting membrane potential at 50
mV during 5-HT-induced modulation. In the simulated presence of 5-HT
and from a resting potential of 50 mV, the model produced an action
potential with a total amplitude of ~96 mV and a duration of 6.8 ms
(i.e., the duration was increased to ~139% of control; Fig.
5A). Although 5-HT-induced
increases in spike amplitude are not a parameter generally investigated
in empirical studies, a review of the published literature indicated that on average 5-HT induces an increase of ~3 mV in the amplitude of
spikes. Thus the simulated increase in spike amplitude was consistent
with empirical studies. Similarly, the simulated increase in spike
duration was in general agreement with empirical studies. A survey of
the published literature indicated that on average 5-HT induced an
increase in spike duration to ~140% of control (cf. Baxter
and Byrne 1990a; Braha et al. 1993; Critz
et al. 1991; Eliot et al. 1994; Ghirardi
et al. 1992; Goldsmith and Abrams 1992;
Hochner and Kandel 1992; Hochner et al.
1986a,b; Jarrard et al. 1993; Mercer et
al. 1991; Pollock et al. 1985; Stark and Carew 1999; Stark et al. 1996; Sugita et
al. 1992, 1994a; Wright and Kirschman 1995;
Wright et al. 1996). The close agreement between the
empirical and simulated results suggest that our current understanding of the 5-HT-induced modulation of membrane currents is sufficient to
account for 5-HT-induced spike broadening. The relative contribution of
the various modulatory actions of 5-HT to spike broadening will be
considered in the following text.
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 5.
Simulating the effects of 5-HT on spike duration and excitability. In
the simulated presence of 5-HT, the resting membrane potential of the
model sensory neurons was maintained at 50 mV by applying constant
bias current (see Table 2). Stimuli used to elicit a single action
potential and to measure excitability are indicated by the bars and
were identical to those described in Fig. 4. Simulated effects of 5-HT
corresponded well to experimentally measured increases in spike
duration and excitability both in time course and magnitude.
A: 2 simulated action potentials are illustrated. Spike
labeled "control" (- - -) was identical to the action potential
illustrated in Fig. 4A. Spike labeled "5-HT" ( )
was elicited in the simulated presence of 5-HT (see Table 2). In the
simulated presence of 5-HT, the peak amplitude of the action potential
was increased slightly and its duration was increased to 6.8 ms.
B: simulated effects of 5-HT greatly enhanced
excitability (1 nA, B1; 2 nA, B2; 3 nA,
B3) as compared with the control excitability of the
model sensory neuron (see Fig. 4B).
|
|
SEROTONERGIC MODULATION OF EXCITABILITY.
As described previously (see Fig. 4B), the excitability of
the model cell was measured as the number of spikes elicited by a
series of 1-s depolarizing current pulses of increasing amplitude. In
the simulated presence of 5-HT, the model no longer exhibited accommodation (Fig. 5B). Rather, the model fired spikes
throughout the 1-s depolarizing current pulses. The simulated responses
to 1-, 2-, and 3-nA depolarizing current pulses were 4, 8, and 11 spikes, respectively. A survey of the published literature indicated that in the presence of 5-HT, sensory neurons fired an average of ~7,
~9, and ~10 spikes during 1-, 2-, and 3-nA depolarizing current
pulses, respectively (cf. Baxter and Byrne 1990a;
Braha et al. 1993; Critz et al. 1991;
Eliot et al. 1994; Ghirardi et al. 1992;
Goldsmith and Abrams 1992; Hochner and Kandel
1992; Jarrard et al. 1993; Klein et al.
1986; Mercer et al. 1991; Stark and Carew
1999; Stark et al. 1996; Sugita et al.
1997b; Wright and Kirschman 1995; Wright
et al. 1996). Although there was a small difference between the
simulated response to a 1-nA current pulse and the average empirical
response (see DISCUSSION), there was a general agreement
between the model and the empirical data in that sensory neurons did
not exhibit accommodation in the presence of 5-HT. The specific
membrane currents that mediated the 5-HT-induced anti-accommodation
will be considered in the following text.
In the present study, a constant bias current was used to maintain the
membrane potential at 50 mV during measurements of 5-HT-induced
increases in spike duration and excitability. This procedure also was
used in some, but not all, empirical studies. To examine whether the
5-HT-induced depolarization of the membrane potential influenced
measurements of spike duration and excitability, a separate set of
simulations was performed without the bias current. The results were
essentially identical to those described in the preceding text, which
suggested the 5-HT-induced depolarization does not play an important
role in increasing excitability or spike duration.
SEROTONERGIC MODULATION OF MEMBRANE CURRENTS.
As a first step toward gaining an understanding of which 5-HT-modulated
currents mediated changes in excitability and spike duration,
simulations investigated the relative contributions of
IK,S,
IK-V,
IK,Ca-S, and
ICa-L to 5-HT difference currents. In
previous voltage-clamp studies (cf. Baxter and Byrne 1989, 1990b; Braha et al. 1993; Critz et al.
1992; Hochner and Kandel 1992; Klein et
al. 1982; Sugita et al. 1994a,b; White et
al. 1994), 5-HT difference currents were generated by
subtracting currents in the presence of 5-HT from control currents.
This subtraction yields the net total current modulated by 5-HT. Figure
6 illustrates a simulated voltage-clamp
experiment from which 5-HT difference currents were generated. Membrane
currents in the model were elicited by voltage-clamp pulses from a
holding potential of 70 to 20 mV (Fig. 6A) and to 20 mV
(Fig. 6B). Currents were elicited first while the parameters
of the model were set to their control values (see Table 1) and again
after the parameters had been adjusted to reflect the simulated
presence of 5-HT (see Table 2). The 5-HT difference currents were
generated by subtracting the 5-HT responses from the control responses.
Thus 5-HT-induced decreases in net outward membrane currents were
represented as upward deflections in the difference currents, and
conversely, downward deflections represented 5-HT-induced increases in
net outward current.
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 6.
Simulating the effects of 5-HT on total membrane current at 2 different
membrane potentials. Simulated effects of 5-HT on total membrane
currents corresponded to similar experimental effects both in time
course and magnitude. Total membrane current represented the sum of
individual ionic currents in the model sensory with the exception of
INa and IL.
Simulated voltage-clamp protocol was similar to that used in Fig. 1 of
Baxter and Byrne (1989). A1: total
membrane currents in the model sensory neuron were elicited by 200-ms
voltage-clamp steps from 70 to 20 mV in simulated control
conditions (a) and in the simulated presence of 5-HT (b).
A2: 5-HT difference current was isolated by subtracting
the current response elicited in the simulated presence of 5-HT from
the control response (i.e., a b). The dashed line labeled
IK,S illustrates the contribution of the
decrease in IK,S (both
IK,S-I and
IK,S-V) to the 5-HT difference current.
B1: total membrane currents elicited by 200-ms
voltage-clamp steps from 70 to 20 mV, 1st in control conditions (a)
and in 5-HT conditions (b). B2: 5-HT difference current
was isolated by subtracting the 5-HT trace from the control trace
(i.e., a b). The dashed line labeled IK-V
illustrates the contribution of the modulation of
IK-V to the 5-HT difference current.
|
|
Throughout the voltage-clamp step to 20 mV, 5-HT-induced modulation
produced a decrease in the net outward membrane current (Fig.
6A1), as indicated by the upward deflection in the 5-HT difference current (Fig. 6A2). This 5-HT difference current
had several components, including modulation of
IK,S,
ICa-L,
IKCa-S, and
IK-V. These various components did not
contribute equally to the 5-HT difference current, however. By
examining the 5-HT-induced modulation of these currents individually,
it was possible to determine the relative contribution that each
current made to the 5-HT difference current. The primary component of
the 5-HT difference current resulted from the decrease in the outward
current IK,S (both
IK,S-I and
IK,S-V). The contribution of the
modulatory changes in IK,S (both
IK,S-I and
IK,S-V) was indicated by the dashed
line in Fig. 6A2, which was generated by subtracting the simulated IK,S in the presence of 5-HT
from the control IK,S. At the end of
the 200-ms voltage-clamp pulse, modulation of
IK,S accounted for 70% of the 5-HT
difference current, whereas, modulation of
IK,Ca-S and
IK-V accounted for 15 and 3%,
respectively. An additional component resulted from the enhancement of
the inward current ICa-L. An enhanced
inward current would decrease the net outward current and thus would be
represented as an upward deflection in the 5-HT difference current.
Modulation of ICa-L accounted for 12%
of the 5-HT difference current at the end of the voltage-clamp pulse.
The result that modulation of IK,Ca-S
made a relatively minor contribution to a brief voltage-clamp pulse to
20 mV agreed with the previous empirical observations of Walsh
and Byrne (1989). They found that manipulations that block
IK,Ca (e.g., applying low concentrations of TEA;
substituting Ba2+ for
Ca2+; or intracellular injection of
Ca2+ chelators) blocked only ~10% of the net
total current modulated by 5-HT during brief voltage-clamp pulses. In
contrast, during long voltage-clamp steps, blocking
IK,Ca reduced the response to 5-HT by
51-68%. This empirical observation suggested that the relative
contributions of individual currents to the 5-HT difference current
varied dramatically over time. To examine this possibility, the
duration of the simulated voltage-clamp step was extended to 1 s,
and a 5-HT difference current was generated (Fig.
7). The contributions to the 5-HT
difference current that were made by modulation of
IK,S (both
IK,S-I and
IK,S-V) and
IK,Ca-S were indicated by the dashed
lines labeled IK,S and
IK,Ca-S, respectively. During the
first half of the 1-s voltage-clamp step, modulation of
IK,S was the predominant component of
the 5-HT difference current, and because
IK,S did not inactivate, the amplitude
of the IK,S component was constant
throughout the later portions of the voltage-clamp step. In contrast,
modulation of IK,Ca-S made very little
contribution early in the voltage-clamp step. Because of its slow
kinetics, however, the contribution of
IK,Ca-S continued to grow throughout the voltage-clamp step. The relative contribution that modulation of
IK,Ca-S made to the 5-HT difference
current eventually exceeded that of
IK,S. At the end of the 1-s
voltage-clamp step, modulation of
IK,Ca-S accounted for 58% of the 5-HT
difference current, whereas IK,S
accounted for 37%. The observation that modulation of
IK,S and
IK,CA-S could account for the majority
of the 5-HT difference current at membrane potentials near the resting
potential suggested that modulation of these currents may play a key
role in mediating 5-HT-induced increases in excitability (see following
text).
View larger version (9K):
[in this window]
[in a new window]
|
Fig. 7.
Relative contribution of IK,S and
IK,Ca-S to long-duration voltage-clamp
pulses. Total membrane currents in the model sensory neuron were
elicited by a 1-s voltage-clamp pulses from 70 to 20 mV, 1st in
control conditions and again in the simulated presence of 5-HT (not
shown). 5-HT difference current (solid line) was generated by
subtracting the current response in 5-HT from the control response and
thus represented the total net current modulated by 5-HT. Short-dashed
line, contribution of 5-HT-induced modulation of
IK,S (both IK,S-I
and IK,S-V) to the 5-HT difference current
and was generated by subtracting the 5-HT-modulated
IK,S response from the control
IK,S response. Long-dashed line, contribution of
5-HT-induced modulation of IK,Ca-S to the
5-HT difference current and was generated by subtracting the
5-HT-modulated IK,Ca-S response from the
control IK,Ca-S response.
|
|
In contrast to voltage-clamp steps to membrane potentials <0 mV,
voltage-clamp steps to 20 mV revealed a complex 5-HT difference current
that represented both decreases and increases in the net outward
current during modulation by 5-HT. The rate of rise of the total
membrane current was slowed in the simulated presence of 5-HT and thus
decreased the net outward current early during the voltage-clamp pulse
(Fig. 6B1). In addition, the simulated actions of 5-HT
slowed the inactivation of the outward current and thus increased the
outward current later during the voltage-clamp pulse. The early upward
defection in the 5-HT difference current reflected the initial decrease
in the outward current, whereas, the downward defection represented the
late increase in outward current (Fig. 6B2).
The complexity of the 5-HT difference current at depolarized membrane
potentials resulted primarily from the modulatory changes in
IK-V, which was the predominant
outward current at membrane potentials more depolarized than ~0 mV
(see Fig. 3) (see also White et al. 1994). The
contribution of modulatory changes in IK-V was indicated by the dashed line,
which was generated by subtracting the simulated
IK-V in the presence of 5-HT from the control IK-V. Examination of Fig.
6B2 revealed that modulation of
IK-V accounted for the majority of the
5-HT-induced decrease in the outward current during the first ~10 ms
of the voltage clamp. The observation that the early component of the
5-HT difference current was due mainly to changes in
IK-V suggested that serotonergic modulation of IK-V may be the key
contributor to 5-HT-induce spike broadening (see following text).
Simulating the effects of elevated levels of cAMP
To gain additional insights into how the modulation of specific
currents and the activation of different second-messenger/protein kinase cascades contribute to 5-HT-induced changes in spike duration and excitability, simulations examined how the currents modulated as a
consequence of elevated levels of cAMP effected the biophysical properties of the model sensory neuron. A survey of the published literature indicated that in the presence of elevated levels of cAMP,
sensory neurons fired an average of ~7, ~11, and ~12 spikes during prolonged depolarizing current pulses of 1, 2, and 3 nA, respectively (cf. Baxter and Byrne 1990a; Braha
et al. 1993; Goldsmith and Abrams 1992;
Hochner and Kandel 1992; Jarrard et al.
1993; Klein et al. 1986; see also Sugita
et al. 1997a). Thus elevated levels of cAMP fully mimicked, and
to some degree exceeded, the actions of 5-HT on accommodation in
sensory neurons. In contrast, elevated levels of cAMP did not appear to
fully mimic the actions of 5-HT on spike broadening. A survey of the
published literature indicated that on average elevated levels of cAMP
induced an increase in spike duration to ~119% of control (cf.
Baxter and Byrne 1990; Goldsmith and Abrams
1992; Hochner and Kandel 1992; Klein
1993; Sugita et al. 1992, 1994b; see also
Abrams et al. 1984; Jarrard et al. 1993).
These empirical observations suggested that currents modulated by
elevated levels of cAMP preferentially modulate accommodation and to a
lesser degree spike duration.
PKA-MEDIATED MODULATION OF SPIKE DURATION AND EXCITABILITY.
As described previously, a constant bias current (see Table 2) was
applied to the model to maintain the resting membrane potential at 50
mV during PKA-mediated modulation. In the simulated presence of
elevated levels of cAMP and from a resting potential of 50 mV, the
model produced an action potential with a total amplitude of ~93 mV
and a duration of 5.7 ms (i.e., the duration was increased to 116% of
control; Fig. 8A). Although
very little empirical data is available regarding PKA-mediated
increases in spike amplitude, that which are available suggest that
PKA-mediated modulation induces a slight increase in spike amplitude
(Baxter and Byrne 1990a; Goldsmith and Abrams
1992; Sugita et al. 1994b). The magnitude of the
simulated PKA-mediated spike broadening was ~40% of the simulated
effect of 5-HT on spike duration (see preceding text). This
intermediate response to PKA by the model was similar to the available
empirical data, which suggested that magnitude of the spike broadening
induced by elevated cAMP was ~47% of the average response to 5-HT
(see preceding text).
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 8.
Simulating the PKA-dependent modulation of spike duration and
excitability. During the simulated activation of PKA, the resting
membrane potential was maintained at 50 mV by applying a constant
bias current (see Table 2). Stimuli used to elicit a single action
potential and to measure excitability are indicated by the bars were
identical to those described in Figs. 4 and 5. Simulated activation of
PKA corresponded well to experimentally measured increases in spike
duration and excitability both in time course and magnitude.
A: 2 simulated action potentials are illustrated. Spike
labeled control (- - -) was identical to the action potential
illustrated in Fig. 4A. Spike labeled PKA () is an
action potential elicited during the simulated activation of PKA (see
Table 2). Activation of PKA slightly increased the peak amplitude of
the action potential and increased its duration to 5.7 ms.
B: simulated effects of activating PKA greatly enhanced
excitability (1 nA, B1; 2 nA, B2; 3 nA,
B3) as compared with the control excitability of the
model sensory neuron (see Fig. 4B).
|
|
In the simulated presence of elevated cAMP, the model produced 4, 9, and 11 spikes in response to 1-s depolarizing current pulses of 1, 2, and 3 nA, respectively. Thus the changes in the excitability of the
model that were induced by PKA were similar to those produced by the
simulated actions of 5-HT. Indeed, in response to the 2-nA pulse, the
model produced an additional spike in the simulated condition of PKA
activation, which suggested that some aspect of 5-HT-induced modulation
may slightly decrease excitability.
RELATIVE CONTRIBUTION OF INDIVIDUAL CURRENTS TO PKA-MEDIATED SPIKE
BROADENING.
To evaluate which currents mediated PKA-induced changes in spike
duration (and excitability, see following text) the modulation of
individual currents was removed selectively from the ensemble of
PKA-mediated actions, and simulation tested the effects of these
manipulations on PKA-mediated spike broadening (and excitability enhancement). For example, previous qualitative models attributed 5-HT-induced spike broadening to PKA-mediated decreases in
IK,S (e.g., Kandel and Schwartz
1982; for review, see Byrne and Kandel 1986). A
prediction of such a model would be that spike broadening would be
blocked if the modulation of IK,S was
removed from the ensemble of PKA-mediate actions (see Table 2). A
simulation to test this predication found that removing only the
modulation of IK,S (both
IK,S-I and
IK,S-V) had no effect on PKA-mediated spike broadening (not shown). Alternatively, enhancement of an inward
current (e.g., ICa) has been suggested
to mediate spike broadening (e.g., Klein and Kandel
1978). A simulation to te
TOP
ABSTRACT
INTRODUCTION
