 |
 Previous Article | Next Article 
Volume 17, Number 11,
Issue of June 1, 1997
pp. 4461-4472
Copyright ©1997 Society for Neuroscience
A Slow Outward Current Activated by FMRFamide in Heart
Interneurons of the Medicinal Leech
Farzan Nadim and
Ronald L. Calabrese
Department of Biology, Emory University, Atlanta, Georgia 30322
 ABSTRACT
- INTRODUCTION
- MATERIALS AND METHODS
- RESULTS
- DISCUSSION
- FOOTNOTES
- REFERENCES
ABSTRACT 
The endogenous neuropeptide FMRFamide
(Phe-Met-Arg-Phe-NH2) can accelerate the oscillation of
reciprocally inhibitory pairs of interneurons that pace heartbeat in
the medicinal leech. A model based on all available biophysical data of
a two-cell heart interneuron oscillator provides a theoretical basis
for understanding this modulation. Previously observed modulation of
K+ currents by FMRFamide cannot account for this
acceleratory effect in the model. This observation prompted the present
reexamination of K+ currents in heart interneurons. We
devised better methods for separation of the various components of
K+ current and more accurately measured their activation
and deactivation kinetics. Moreover, we demonstrated that FMRFamide
activates a previously undetected K+ current
(IKF), which has very slow activation and
deactivation kinetics. Addition of physiologically measured amounts of
IKF to the model two-cell oscillator can
account for the acceleratory effect of FMRFamide.
Key words:
FMRFamide;
slow outward currents;
neural oscillator;
conductance-based model;
leech;
K+
INTRODUCTION
Rhythmic motor patterns, such as feeding and
chewing, locomotion, breathing, and heartbeat (in certain
invertebrates), are programmed in part by rhythmically active neural
networks called pattern generators (Delcomyn, 1980 ). Oscillation in
these networks derives from the interplay of the inherent membrane
properties of the component neurons and their synaptic interactions
(Getting, 1989; Jacklet, 1989; Arshavsky et al., 1993; Harris-Warrick,
1993; Rossignol and Dubuc, 1994; Dean and Cruse, 1995; Marder and
Calabrese, 1996). To produce adaptive behavior, these pattern
generators must respond to the changing internal and external needs of
the animal by producing motor outflow that is gated on or off or is altered in its frequency, strength, and activity phases. Numerous studies have indicated that both synaptic interactions and membrane properties can be modulated to effect these changes (for reviews, see
Harris-Warrick and Marder, 1991; Dickinson, 1995; Katz, 1996). Often
more than one neuron, synapse, or membrane property is altered by a
modulatory substance that produces an adaptive change in motor output
(e.g., Johnson et al., 1995), so that it can be difficult to determine
which modulated parameters(s) is critical for producing a given motor
outflow (Harris-Warrick et al., 1995). The use of computer modeling
studies and hybrid simulation tools such as the dynamic clamp (Sharp et
al., 1993) have been fruitful in identifying such critical parameters
(Golowasch et al., 1992; Sharp et al., 1993; Harris-Warrick et al.,
1995).
We have been studying the pattern generator for heartbeat in the
medicinal leech and its modulation by the endogenous (Evans et al.,
1991) neuropeptide FMRFamide (Phe-Met-Arg-Phe-NH2) (for a
recent review, see Calabrese et al., 1995). Activity in the FMRFamide
immunoreactive cell 204 or bath application of the peptide ( 5 × 10 8 M) increases the cycle rate of this
pattern generator (Simon et al., 1992). Bath application of higher
concentrations of FMRFamide leads to a disruption of rhythmic activity
(Simon et al., 1992). The "beat timing oscillator" that paces this
pattern generator consists of two bilateral pairs of reciprocally
inhibitory segmental interneurons that are linked together by two pairs
of segmental coordinating interneurons (Calabrese et al., 1995). Each
of these reciprocally inhibitory pairs can act as an independent
elemental oscillator, when the segmental ganglion in which they reside
is isolated from the rest of the ventral nerve cord. In such reduced preparations, bath-applied FMRFamide exerts both its acceleratory and
disruptive effects on the elemental oscillators (Simon et al.,
1992).
Here we report experimental studies that identify a new voltage-gated
outward current that is modulated by FMRFamide and can account for the
acceleratory effect of FMRFamide when incorporated into our
conductance-based model of a heart interneuron elemental oscillator.
MATERIALS AND METHODS
Leeches (Hirudo medicinalis) were obtained from
Leeches USA and Biopharm and maintained in artificial pond water at
15°C. After the animals were anesthetized in ice-cold saline,
individual ganglia were dissected and pinned in small petri dishes.
Ganglia were superfused continuously with normal leech saline
containing (in mM): 115 NaCl, 4 KCl, 1.8 CaCl2,
10 glucose, 10 HEPES buffer; adjusted to pH 7.4. Equimolar amounts of
N-methyl-D-glucamine and Co++
replaced Na+ and Ca2+, respectively, in 0 Na+, 0 Ca2+ solutions. FMRFamide (Bachem,
Torrance, CA) was dissolved in HPLC-grade water at a concentration of
103 M, stored frozen, and diluted to
106 M in physiological saline immediately
before use; it was applied at this concentration in all experiments
described herein.
Cells were penetrated with borosilicate microelectrodes (1 mm outer
diameter, 0.75 mm inner diameter) filled with 4 M potassium acetate with 20 mM KCl (20-35 M ). Electrodes were
coated with Sylgard 182 (Dow Corning) up to ~50 µm from the tip to
reduce capacitance. Currents were measured using switching
single-electrode voltage clamp (Axoclamp 2A, Axon Instruments, Foster
City, CA). Sample rates were between 2.9 and 3.5 kHz, and clamp gain
was from 8 to 50 nA/mV. The output bandwidth was set at 0.3 kHz. At the
end of the experiment, microelectrodes were withdrawn from the cell and
only those preparations in which the electrode was within ±5 mV of the
bath potential were accepted. Voltage steps were generated by a
computer (PC with 486 processor), and data were digitized and stored
using pCLAMP software (Axon Instruments). The measurement of time
constants was performed with CLAMPFIT 6.0, and leak subtractions were
performed on a Sun SPARCstation LX. Deactivation the opposite process
from activation, i.e., closing of the "activation gate"was
measured after an activating voltage step, after returning to a more
hyperpolarized potential. Because currents measured in FMRFamide
deactivated slowly, the automatic leak-subtraction protocol of pCLAMP
(using one or several hyperpolarizing voltage steps before each
depolarizing step and adding the resulting currents) was not used.
Instead, we applied 4-10 hyperpolarizing steps, of magnitude 10 mV and
duration 4 sec, both before and after each experimental protocol. We
used the average leak current from all these steps to leak-subtract the
currents measured with depolarizing steps.
The simulations were performed with Neurolab (Olsen, 1994) on a Sun
SPARCstation LX and on a PC with Pentium processor under Linux. The
simulations were performed using a variable time-step method (LSODES).
The ionic currents in the model are given by Hodgkin-Huxley type
equations (Hodgkin and Huxley, 1952) and are described in detail in
Nadim et al. (1995) and Olsen et al. (1995).
RESULTS
Rationale for the reexamination of outward currents in
heart interneurons
Several ionic currents have been identified in single-electrode
voltage-clamp studies that contribute to the activity of oscillator heart interneurons. These include, in addition to the fast
Na+ current that mediates spikes and the leak current
(Il, Erev =  52.5 mV),
two low-threshold Ca2+ currents (Angstadt and Calabrese,
1991) [one rapidly inactivating (ICaF) and one
slowly inactivating (ICaS)]; three outward
currents (Simon et al., 1992) [a fast transient K+ current
(IA) and a delayed rectifier-like K+
current (IK), consisting of an inactivating
(IK1), component and a persistent
(IK2) component]; a hyperpolarization-activated inward current (DiFrancesco and Noble, 1989)
(Ih) [mixed Na+/K+,
Erev = 20 mV] (Angstadt and Calabrese,,
1989); and a low-threshold persistent Na+ current
(IP) (Opdyke and Calabrese, 1994). The
inhibition between oscillator interneurons consists of a graded
component that is associated with the low-threshold Ca2+
currents (Angstadt and Calabrese, 1991) and a spike-mediated component
that seems to be mediated by high-threshold Ca2+ current
(Simon et al., 1994; Lu et al., 1997). Blockade of synaptic transmission with bicuculline leads to tonic activity in oscillator heart interneurons (Schmidt and Calabrese, 1992).
Much of this biophysical data was incorporated into a conductance-based
model of an elemental (two-cell) oscillator (Nadim et al., 1995; Olsen
et al., 1995), using standard Hodgkin-Huxley (Hodgkin and Huxley,
1952) representations of each voltage-gated current. The model also
contains explicit independent formulations for both spike-mediated and
graded synaptic transmission (Calabrese and De Schutter, 1992; De
Schutter et al., 1993; Nadim et al., 1995). The model generates
activity that closely approximates that observed for an elemental
oscillator under control and various experimental conditions. This
model has recently been tested experimentally by voltage clamping
oscillator interneurons with realistic waveforms and has proved to be
reliable (Olsen and Calabrese, 1996).
Voltage-clamp and current studies of oscillator interneurons have
revealed several modulatory effects of FMRFamide (bath applied), including (1) negative shifts in the steady-state activation and inactivation of IK (Simon et al., 1992), (2)
activation of an IP-like current (Schmidt et
al., 1995), and (3) an apparent reduction in spike-mediated synaptic
transmission (Simon et al., 1994). In preliminary studies using our
model, we have found that none of these effects satisfactorily account
for the acceleratory effect of FMRFamide on the oscillator
interneurons, although they can account for the disruptions observed at
higher concentrations. These observations have led us to reexamine
outward currents in oscillator heart interneurons and their modulation
by FMRFamide.
The outward current in heart interneurons comprises
three components
We measured leak-subtracted outward currents in heart interneurons
in 0 Na+, 0 Ca2+, 1.8 mM
Co++ saline. Our measurements confirmed those of Simon et
al. (1992). When the cells were held at 70mV and activated by
depolarizing pulses, two outward currents were present: a fast,
transient current IA and a slower current
IK, which only partially inactivated. When the
cells were voltage-clamped at 35 mV, depolarizing pulses produced
only IK, indicating that the transient current
IA was completely inactivated at 35 mV. Use of
1.5 sec or longer depolarizing pulses revealed that
IK itself comprised two components: a component (IK1) that inactivated with a time constant of
400-800 msec, and a component (IK2) that was
either noninactivating or inactivated with a time constant >3 sec.
Figure 1 shows the outward current elicited using a 4 sec pulse to 0 mV, from the holding potential of 70 mV, together with
fits to IA, IK1, and
IK2. We used data described in this manuscript
together with data from Simon et al. (1992) to obtain mathematical
fits. The fits were made using Hodgkin-Huxley kinetic models (Hodgkin
and Huxley, 1952) and were calculated using voltage-dependent time
constants and steady states. The ionic currents were represented
as:
|
(1)
|
where q = 1 for IA and
IK1, and q = 0 for
IK2. The activation variable m
(similarly for the inactivation variable h) was governed by
the differential equation:
|
(2)
|
The steady-state activation curve m
(similarly for h) is given by the
sigmoid:
|
(3)
|
The parameters are given in Tables 1 and
2.
Fig. 1.
The outward currents measured in the heart
interneurons. The outward currents comprise a rapidly inactivating
IA, and a delayed rectifier consisting of a
slowly inactivating component IK1 and a
persistent component IK2. A,
Outward currents measured in heart interneurons (ganglion 4), in
response to a 4 sec voltage step from 70 mV to 0 mV. Also shown are
the fit to the current using parameters given in Tables 1 and 2, and
the three components of the fit: IA,
IK1, and IK2.
B, Outward currents measured in the same cell, in
response to a 4 sec voltage step from 35 mV to 0 mV. Also shown are
the trace fit to the current using parameters given in Tables 1 and 2. The holding potential of 35 mV almost completely inactivates
IA, the component of the current that gives the fast transient peak.
[View Larger Version of this Image (11K GIF file)]
Oscillation in model heart interneurons is sensitive to the
activation kinetics of IK1 and
IK2
Using our model of oscillator heart interneurons (Nadim et al.,
1995), we have looked at the possible extent and range of activation of
the outward currents during oscillations. A sensitivity analysis of the
oscillations in the model cells had revealed that the period is
particularly sensitive to variations in the maximal conductance of
IK1 and IK2 (Olsen et
al., 1995). In our model we used equations derived by Simon et al.
(1992), in which IK1 activated and deactivated
rapidly ( = 1-12 msec), and IK2
activated and deactivated slowly ( = 60-100 msec). With
these equations, our canonical model produced oscillations with a
period of 7.5 sec (Fig. 2A). Varying
time constants of activation and deactivation of
IK1 and IK2 affected the
period and amplitude of the model oscillation. By making the activation
rate of IK2 as fast as that of
IK1 (but keeping the deactivation slow), the
period of oscillations was reduced to 3.6 sec (Fig.
2B). Measurements of IK2
during these model oscillations showed that the activation variable
(m) of IK2 integrated rapidly with
action potentials, without deactivating fast. Therefore, there was more
IK2 available during the burst phase of the
oscillations compared with the canonical case. The increase in
IK2 reduced the spike frequency and caused the
speed-up of the oscillation. [See Olsen et al. (1995) for a discussion of the effect of spike frequency on the period of oscillations.]
Fig. 2.
A 10 sec window of the oscillation in one heart
interneuron of a model elemental oscillator (two reciprocally
inhibitory cells). Dashed line denotes membrane
potential of 50 mV. The values of time constants are given in Table
2. A, Bursting oscillation of a model cell using
canonical parameters of the ionic and synaptic currents, as described
in Nadim et al. (1995). The period of the canonical oscillation is 7.5 sec. B, Model cell oscillation where the activation time
constant of IK2 is changed from its
canonical value to be as fast as the activation time constant of
IK1. The period has decreased to 3.6 sec.
C, Model cell oscillation where the deactivation time
constant of IK2 is changed from the
canonical value to be as fast as the deactivation time constant of
IK1. The period has increased to 10.6 sec.
D, Model cell oscillation where time constants of both
activation and deactivation of IK2 are fast.
In this case, the model cells cannot support action potentials on the
plateau.
[View Larger Version of this Image (33K GIF file)]
By making the deactivation rate of IK2 as fast
as that of IK1 (but keeping the activation
slow), the period of oscillations was increased to 10.6 sec (Fig.
2C). The activation variable (m) of
IK2 was set back to almost zero with the
repolarization after each action potential. There was little
accumulation of IK2 during the burst, and the
current became small compared with the canonical case. If both
activation and deactivation of IK2 were fast,
then the model cells could not support action potentials on the
bursting plateau (Fig. 2D). The failure to support
action potentials was caused by inactivation of the fast
Na+ current INa, which is
responsible for spiking. When deactivation of
IK2 is slow, there is sufficient removal of
inactivation from INa during the intervals
between action potentials. Therefore, action potentials do not fail
merely by making the activation of IK2 fast. If
the activation time constant of IK2 was
decreased gradually (both in the activation and in the deactivation
range), then initially the effect of fast deactivation was dominant and the period of oscillation increased; subsequently, as time constants were made faster, spiking ceased and model cells produced oscillations as shown in Figure 2D. These theoretical studies
pointed out the importance of the kinetics of outward currents in
determining the period of oscillation, and thus motivated a careful
experimental reevaluation of the activation and deactivation kinetics
of IK2 and the effects of FMRFamide on these
parameters.
Activation and deactivation kinetics of
IK2
Simon et al. (1992) measured activation and deactivation kinetics
of IK. They voltage-clamped cells at 35 mV to
inactivate IA, used 150 msec depolarizing pulses
to measure activation time constants, and used a 75 msec prepulse to 20 mV, followed by postpulses to a range of potentials, to measure
deactivation time constants. The protocol they used, therefore,
measured activation and deactivation time constants for the sum of both
IK1 and IK2. We were
interested in obtaining activation and deactivation time constants for
IK2 separately, to see any difference in the
activation kinetics of IK1 and
IK2. We could not separate the currents
pharmacologically; thus it was not possible to measure the activation
kinetics of IK1 and IK2
separately. We made use of the inactivation of
IK1 to isolate IK2 and
measure its deactivation kinetics in a range of membrane potentials
that was as wide as possible. Because deactivation is generally
believed to be the opposite process to activation, the kinetics
measured would provide information about the activation kinetics of
IK2 as well. Deactivation time constants of
IK were measured by holding the cell at 35 mV
and applying a prepulse to 0 mV to activate the current, followed by a
postpulse at various membrane potentials. Two sets of experiments were
performed: one used a 100 msec prepulse (Fig.
3A) and the other used a 4 sec prepulse (Fig.
3B). Time constants of deactivation were obtained from tail
currents measured at membrane potentials between 55 and 25 mV.
Because IK1 inactivates with time constant of
400-800 msec (Simon et al., 1992), it was completely inactivated by
the end of the 4 sec pulse. Therefore, the time constants measured after the long pulse were deactivation time constants of
IK2, and the time constants measured after the
short pulse were deactivation time constants of
IK1. Postpulses above 25 mV did not result in
clearly decaying currents, and postpulses below 55 mV were too close
to the reversal potential to produce a clear tail current. For
measuring the time constraints, a 5 msec initial period of the
postpulse was omitted to allow the electrode to settle and establish
voltage control. The cells were initially voltage-clamped at 35 mV so
that the transient current IA was
inactivated.
Fig. 3.
Deactivation time constants for
IK. Cells were voltage-clamped at 35 mV.
Outward currents were activated by a prepulse to 0 mV. The prepulse was
followed immediately by a postpulse from 60 mV to 25 mV, in steps
of 5 mV. A, Currents used to measure deactivation time
constants after a brief (100 msec) activation prepulse.
B, Currents used to measure deactivation time constants after a long (4 sec) activation prepulse. C, Plot of
average deactivation time constants against membrane potential after a
short (solid circles) or a long (open
circles) prepulse (mean ± SEM; n = 6). Asterisks indicate values that are significantly
different (t test; p < 0.01).
[View Larger Version of this Image (14K GIF file)]
At most membrane potentials measured, there was significant difference
between time constants of deactivation after a short prepulse versus a
long prepulse (Fig. 3C). The current, activated with a brief
prepulse, deactivated with a time constant between 20 and 50 msec. This
range is consistent with the deactivation rates reported in Simon et
al. (1992), in which deactivation time constants were measured after a
75 msec prepulse. After a 4 sec prepulse, however, the current
deactivated with a time constant between 100 and 250 msec.
Deactivation of IK in the presence and
absence of FMRFamide
Deactivation time constants of IK2 were
measured and compared in the absence and presence of FMRFamide. Cells
were voltage-clamped at 70 mV, and a 6 sec pulse to 0 mV was applied,
followed by a 4 sec postpulse to potentials ranging from 100 mV to
40 mV (Fig. 4A). Difference currents
were obtained by subtracting leak-subtracted current traces measured in
the absence of FMRFamide from current traces, corresponding to the same
voltage pulse, measured in the presence of FMRFamide. An increase in
the size of the currents measured during successive prepulses to 0 mV
was observed in FMRFamide. This increase indicated accumulation of an
outward current over several pulses (Fig. 4A).
Fig. 4.
Tail currents of IK in
the absence and presence of FMRFamide. A,
Leak-subtracted outward currents measured by applying a 6 sec
depolarizing pulse to 0 mV from a holding potential of 70 mV,
followed immediately by 4 sec pulses from 100 mV to 40 mV, in steps
of 10 mV. In the absence of FMRFamide, tail currents were small,
whereas large tail currents were observed in the presence of
106 M FMRFamide. Currents measured in the
absence of FMRFamide were subtracted from currents measured in the
presence of FMRFamide to reveal the difference current.
B, The I-V plot of tail currents. Currents were measured from traces of the difference current at the
beginning of postpulse. The current values indicate that the reversal
potential of the difference current is between 70 mV and 60 mV.
C, Plot of deactivation time constants against membrane potential in the presence (solid squares) and absence
(open circles) of FMRFamide. Tail currents in the
presence of FMRFamide were not only larger in magnitude, but also had a
slower decay rate at membrane potentials of 70 mV and below.
Asterisks indicate values that are significantly
different (t test; p < 0.05;
n = 7).
[View Larger Version of this Image (20K GIF file)]
The amplitude of the difference current at the beginning of the
postpulse was plotted against membrane potential (Fig.
4B). This I-V plot shows that the
reversal potential of the difference current is between 70 mV and
60 mV. This estimate for the reversal potential is consistent with
previous estimates of EK in leech heart
interneurons (Simon et al., 1992). The difference currents measured at
membrane potentials above the reversal potential were generally smaller
than those measured at membrane potentials below the reversal
potential. Compare the size of currents measured at 40 mV with those
measured at 90 mV in Figure 4B. This difference in
amplitude was independent of the order of the postpulse potentials applied, and it suggests that the channels involved may pass current preferentially in the inward direction.
Using the tail currents, deactivation time constants were measured in
the absence and presence of FMRFamide. At potentials above the reversal
potential of the difference current, there was no significant
difference between deactivation time constants in the absence and
presence of FMRFamide. At membrane potentials of 70 mV and below,
however, the decay of tail currents was significantly slower
(t test; p < 0.05; n = 7)
in the presence of FMRFamide than the decay in the absence of FMRFamide
(Fig. 4C). It is not possible to measure the decay of tail
currents near reversal and thus determine whether there is also
significant slowing in the important physiological range around 60
mV; however, the fact that IK deactivates very
slowly in the presence of FMRFamide at membrane potentials below 70
mV suggests that the deactivation may also be slow around the reversal
potential. Slow deactivation of IK in the
presence of FMRFamide at membrane potentials around 60 mV could help
activate the hyperpolarization-activated inward current
Ih, as it does in the model described in
Results.
FMRFamide activates a slowly activating outward current
IKF
To reveal any slowly activating outward current that might be
elicited by FMRFamide, we used a long depolarizing protocol and
compared the activation of IK in the absence and
presence of FMRFamide. From a holding potential of 70 mV, a 17.5 sec
pulse to 0 mV was applied and followed by an 8 sec pulse to 100 mV to
reveal any tail currents. This protocol was used in both the absence
and presence of FMRFamide, and leak-subtracted current traces were used
to obtain the difference current (Fig. 5). The difference current during the depolarizing pulse to 0 mV comprised an
initial transient inward current followed by a slowly activating outward current ( = 10-12 sec). The slowly activating
outward current was also slow in deactivation ( = 2-3
sec), resulting in a large tail current. The transient inward current
was possibly caused by the negative shift of inactivation steady-state
of IK1 in FMRFamide, as reported by Simon et al.
(1992) (see Discussion). We shall refer to this novel
FMRFamide-sensitive slowly activating outward current as
IKF.
Fig. 5.
Bath application of FMRFamide activates a slowly
activating outward current. From a holding potential of 70 mV,
outward currents were activated by a 17.5 sec depolarizing pulse to 0 mV, followed by an 8 sec pulse to 100 mV to reveal the tail current.
This protocol was used in both the absence and presence of FMRFamide, and a difference current was obtained by subtracting the traces. The
difference current (bottom trace) reveals
a slowly activating ( = 9 sec) outward current that deactivated
slowly ( = 3 sec) as well. The difference current was fit using the
parameters given by Tables 1 and 2 (bottom trace, dotted
lines). A maximal conductance of 23 nS for
IKF gave a good fit of the current in
response to the step pulse to 0 mV, but underestimated the tail current
in response to the step down to 100 mV. A good fit of the tail
current was obtained by increasing the maximal conductance to 70 nS.
[View Larger Version of this Image (12K GIF file)]
We fit the difference current using the parameters given by Tables 1 and 2 (Fig. 5, bottom trace, dotted lines). A maximal conductance of 23 nS for IKF gave a good fit of
the current in response to the step pulse to 0 mV, but underestimated
the tail current in response to the step down to 100 mV. A good fit
of the tail current was obtained by increasing the maximal conductance threefold to 70 nS. The increase in conductance to obtain a good fit of
the tail current confirms that IKF is larger as
an inward current than as an outward current.
Activation of IKF
To measure activation of IKF, we
voltage-clamped the cells at 70 mV and activated outward currents
using six depolarizing steps to membrane potentials from 50 mV to 0 mV, both in the absence and in the presence of FMRFamide (Fig.
6A). A time interval of at least 30 sec was allowed between pulses in FMRFamide to allow the current to
deactivate and therefore to prevent the accumulation of the current
over several pulses. We used the amplitude of the difference current at
the end of the pulse as a measurement of the activation
IKF. A sample plot of IKF
(measured at the end of the pulse) against membrane potential is shown
in Figure 6B. Assuming a reversal potential of 65
mV, the IKF conductance was calculated and
plotted against membrane potential (Fig. 6C). Because the
conductances did not saturate at 0 mV, a steady-state activation curve
was not plotted. Voltage pulses to potentials higher than 0 mV were not
used because good voltage clamp could not be obtained at such
potentials. Also, because of difficulty in keeping the cells at
depolarized potentials for long periods of time, the length of the
activating pulse was restricted to 12 sec, even though the current did
not completely activate in that period of time.
Fig. 6.
Activation of IKF.
A, Currents activated from a holding potential of 70
mV using a series of 12 sec depolarizing steps from 50 mV to 0 mV.
Currents obtained in the absence (left panel) and
presence (right panel) of FMRFamide.
B, Current-voltage relationship of the difference
current at the end of the 12 sec depolarizing step. C,
Conductance of the difference current IKF at
the end of the 12 sec depolarizing step shown against membrane
potential. To calculate conductance, the reversal potential of
IKF was assumed to be 65 mV.
[View Larger Version of this Image (18K GIF file)]
To observe the activation of IKF over longer
periods of time, and to observe how IKF might be
integrated during the oscillation of the heart interneurons, the
following protocol was used. The cells were voltage-clamped at 70 mV,
a sequence of four 6 sec pulses to 0 mV was applied, and the time
interval between these pulses was varied from 2 to 12 sec (Fig.
7A). Over the four pulses, the amplitude of
the outward current increased without saturating, provided the
interpulse interval was brief (2-6 sec). When the interpulse interval
was increased to 12 sec, there was an increase an the amplitude of
IK from the first to the second pulse, and a
smaller increase from the second to third pulse. In all measurements with a 12 sec interpulse interval, the current eventually saturated, and there was little or no increase in IK from
the third to the fourth pulse. Figure 7B shows the
difference between the amplitude of IK at the
end of the first pulse and at the end of each consecutive pulse for
traces plotted in Figure 7A. The difference between the
fourth-pulse and first-pulse amplitude of IK was
more than double when the interpulse interval was 2 sec
(black) as compared with 12 sec (white). The 6 sec interpulse period (gray) resulted in an increase
in IK that was intermediate between the 2 sec
interval and the 12 sec interval case. It should be noted that the
increased amplitude of the IK does not decay
during the 12 sec intervals between pulses. This fact, together with
our measurements of the deactivation of IK in
the presence of FMRFamide (Fig. 4C), implies that there is a
residual part of the current that remains active despite the long wait
intervals in our voltage-clamp protocols.
Fig. 7.
Cumulative activation of
IKF. A, A sequence of 6 sec
pulses to 0 mV were applied from a holding potential of 70 mV. The
interpulse interval was varied from 2 sec (top traces)
to 6 sec (middle traces) to 12 sec (bottom
traces). The accumulation of gKF was
reduced by increasing the interpulse interval. The calibration bar
refers to all three sets of traces. B, Increase in the
amplitude of IK at the end of the 6 sec
pulse. The difference between the amplitude of the current at the end
of the first pulse and at the end of each consecutive pulse is plotted.
For the 2 sec interpulse period (black), the increase
was large and did not saturate over the four pulses. For the 12 sec
interpulse period (white), the increase was small and
saturated. The 6 sec interpulse period (gray) was intermediate between the 2 sec interval and the 12 sec interval case in
amplitude increase.
[View Larger Version of this Image (18K GIF file)]
IKF speeds up the oscillation of model
heart interneurons
Simon et al. (1992) showed that bath application of FMRFamide at
low concentrations (5 × 108 M) causes
an increase in the oscillation rate of heart interneurons (Fig.
8A). The increase in the oscillation
rate is reversed by washing out the FMRFamide. We used our model of a
reciprocally inhibitory pair of heart interneurons (Nadim et al., 1995;
Olsen et al., 1995) to test the effect of IKF on
the oscillation of model cells. The current IKF
was modeled as a noninactivating, slowly activating, and slowly
deactivating K+ current:
The activation variable m was governed by
Equation 2, where the steady-state activation curve
m is given by Equation 3. The parameters
defining the model IKF are given in Tables 1 and
2. The maximal conductance of IKF was increased
from 0 to 40 nS over 40 sec, held at 40 nS for 20 sec, and then reduced back to 0 over 40 sec. Addition of IKF reduced
the period of oscillations in the model heart interneurons from 7.8 sec
to 5.6 sec. Analysis of the ionic currents in the model cells revealed
two factors that contributed to the acceleration of the rhythm:
additional activation of the hyperpolarization-activated inward current
Ih and decrease in spike frequency, and
therefore spike-mediated synaptic transmission.
Ih increased because the slow deactivation of
IKF caused the model membrane potential to
approach EK more closely and linger there.
Additional activation of Ih caused the model
cells to escape more quickly from the inhibited phase and hence reduced
the period of oscillations. This effect was similar to increasing the
maximal conductance of Ih, as described in Olsen et al. (1995). The decrease in spike frequency in the model cell resulted in less spike-mediated inhibition onto the contralateral cell,
in turn allowing the contralateral cell to escape inhibition more
easily, hence speeding up the oscillation. According to Nadim et al.
(1995) and Olsen et al. (1995), spike-mediated inhibition is the
dominant form of synaptic transmission in the oscillation of model
heart interneurons, and possibly in the oscillation of the heart
interneurons themselves. The effect of
IKF in reducing the spike frequency was
similar to the effect of increasing the maximal conductance of
IK2, as described in Olsen et al. (1995).
Fig. 8.
Acceleration of the heartbeat rhythm by FMRFamide.
A, The period of the activity rhythm in oscillator heart
interneurons (HN) is reduced from
7.5-8.5 sec to ~6 sec by bath application of 5 × 108 M FMRFamide. B,
The effect of FMRFamide is mimicked in the model heart interneurons by
introducing IKF and shifting the
steady-state inactivation curve of IK1 by
10 mV (i.e., to the left). Simon et al. (1992) demonstrated that
FMRFamide produced such a negative shift in steady-state inactivation
of IK1. HN cells are indexed by ganglion
number and body side.
[View Larger Version of this Image (32K GIF file)]
The heart interneuron model neurons were improved using the
new data
We used the new kinetic equations for IK1,
IK2, and IA in the model
of heart interneurons described in Nadim et al. (1995) and Olsen et al.
(1995). In Nadim et al. (1995), we had reported that the model cells
reproduce the behavior of the biological cells in producing
oscillations in reduced-Na+ saline (slow oscillations). To
produce these oscillations, however, the maximal conductance of the
graded synaptic current (gSynG) in the
cells had to be increased from the experimentally measured value of
20-30 nS to 300 nS. With the new kinetic equations for the outward
currents, the model cells reproduced the slow oscillations with
gSynG = 30 nS (Fig. 9). This
improvement in the model was caused by the slow deactivation of
IK2. During the depolarized phase of the
oscillations, IK2 was activated, counteracting
the depolarizing effect of the inward currents. During the inhibited phase, IK2, because of its slow deactivation
time constant, pulled the membrane potential toward the reversal
potential of K+ (75 mV) and caused a delay in the rise of
the membrane potential. This delay resulted in removal of inactivation
from the Ca2+ currents, which were responsible for
producing the graded synaptic transmission in the model cells (Nadim et
al., 1995). The increase in the Ca2+ currents resulted in
sufficient graded synaptic transmission in the opposite cell to keep
the oscillation stable.
Fig. 9.
Slow oscillations of oscillator heart interneurons
(HN) of the third ganglion in 10%
Na+ saline containing 5 mM Ca2+
(the normal [Ca2+] is 1.8 mM) triggered by a
hyperpolarizing pulse (not shown) in one cell. Similar oscillations are
produced in the model cells by reducing ENa
and Eh. The model cells are as described in
Nadim et al. (1995), with K+ currents given by Equations 1
and 2, and Tables 1 and 2. The maximal conductance
gSynG of the graded synaptic current is 30 nS, which is within the biological range of and an order of magnitude smaller than the value used in Nadim et al. (1995). HN cells are indexed by body side.
[View Larger Version of this Image (18K GIF file)]
DISCUSSION
From model to experiment: outward currents revisited
In Nadim et al. (1995), we described a realistic conductance-based
model of the elemental two-cell oscillator that gives rise to rhythmic
activity underlying the leech heartbeat. Analysis of the model cells
predicted activation levels of ionic currents that were confirmed in
experiments in which oscillator interneurons were voltage-clamped using
realistic waveforms (Olsen and Calabrese, 1996). Moreover, this model
has suggested potential mechanisms for modulation of cycle period in
the real heart interneurons.
Our previous studies of outward currents in leech heart
interneurons (Simon et al., 1992) showed that they comprise three components: a rapidly inactivating A-like current
IA, a slowly inactivating component
IK1, and a noninactivating component
IK2 [called IA,
IKF(ast), and IKS(low),
respectively, in Simon et al., (1992)]. Sensitivity analysis on the
model heart interneurons showed that the period of oscillation is
particularly sensitive to variations in IK1 and
IK2 (Olsen et al., 1995). Simon et al. (1992)
showed that FMRFamide affects the steady-state inactivation and to a
smaller extent the steady-state activation of
IK1 (and IK2) by shifting
these curves to more negative potentials. In particular, from a holding
potential of 70 mV, there would be more inactivation of
IK1 in the presence of FMRFamide. The additional inactivation in FMRFamide could explain why the difference current in
Figure 5 is initially negative. We observed the initial negative current in most difference currents measured. If, however, several depolarizing pulses were applied in a sequence in the presence of
FMRFamide, the cumulative activation of IKF
eventually became larger than the initial extra inactivation of
IK1, and the initial negative difference current
was obscured.
Because the measurements of activation of IK
performed by Simon et al. (1992) applied to the sum of both currents
IK1 and IK2, from their
data it was inconclusive whether IK1 and
IK2 have different activation and deactivation
time courses. We show that such a difference has large effects on
oscillations in the model heart interneurons (Fig. 2). Whether
FMRFamide shifts the steady-state activation curve of
IK2 could also not be determined from
experiments described in Simon et al. (1992). To examine this question,
we needed to find experimental protocols to measure the activation or
deactivation time constants of each current separately.
The slow current IKF is not caused by a
change in activation of IK2 in FMRFamide
We were interested to see whether the activation of
IK2 is sensitive to FMRFamide. A standard
activation voltage-clamp protocol in the absence and presence of
FMRFamide seemed to indicate extra activation of
IK2 by FMRFamide. A scrutiny of the currents
measured during automatic leak-subtraction steps, however, showed that the leak current after a depolarizing pulse had a time-dependent, decaying component. The nonlinearity observed in the measurement of the
leak current indicated either slow activation or slow deactivation of a
current that was inward at membrane potentials below 70 mV. We
hypothesized that in the presence of FMRFamide a new current IKF is activated, and we verified that the
nonlinearity in the leak-subtraction steps was caused by the
deactivation of IKF. We have not entirely
excluded the possibility that the extra current activated in the
presence of FMRFamide is attributable to a change in the kinetics of
IK2; however, when several long depolarizing voltage pulses are applied in a sequence, the extra current is activated cumulatively over these steps (Fig. 7), indicating very slow
activation kinetics, whereas even during the first depolarizing pulse,
IK2 is activated. Therefore it is implausible
that the new current is caused by a change in the activation kinetics
of IK2.
Because IKF is very slow in activation, it
activates cumulatively over a sequence of depolarizing pulses.
Therefore, in Figures 3B and 4A, there is
cumulative activation of IK over several pulses to the same potential. This cumulative activation, however, should not
change the time constant of activation and deactivation measured, but
just the amplitude.
Auxiliary mechanisms may play a role in activation and deactivation
of IKF
We have shown that the activation and deactivation of
IKF are voltage dependent, but the dependence is
slow ( values of the order of seconds). There is a possibility that
activation of IKF is not solely a
voltage-dependent process but that it is is also influenced by some
internal second messenger: for example, Ca2+ released from
internal stores. Such an "auxiliary" mechanism for activation is
supported by the evidence that in addition to the voltage-dependent
deactivation described in the results, there is a slower deactivation
of IKF in which after activation by a sequence
of pulses the cell remains hyperpolarized by a few millivolts for a
minute or two before returning to baseline.
From experiment to model: activation of IKF
may be one mechanism of modulation by FMRFamide of the heartbeat
rhythm
Heartbeat in the leech has a period of ~8-12 sec at room
temperature. Various sensory pathways and identified modulatory neurons can accelerate this rhythm of activity (Arbas and Calabrese, 1984, 1987). The target of this modulation must be the elemental two-cell oscillators that pace this centrally programmed rhythm. Among these
modulatory pathways are the swim-gating interneuron cell 204 (Weeks and
Kristan, 1978; Arbas and Calabrese, 1984). Activity in these neurons
gates on the swimming motor pattern in both semi-intact preparations
and in isolated nerve cords, and likewise accelerates the cycling of
the heartbeat pattern generator. These neurons are immunoreactive
(Kuhlman et al., 1985) for the endogenous neuropeptide FMRFamide (Evans
et al., 1991), and bath-applied FMRFamide at lower concentrations
(5 × 108 M) accelerates the cycling
of the heartbeat pattern generator (Simon et al., 1992). Although
several modulatory effects of FMRFamide have been documented in
heartbeat oscillator interneurons (Simon et al., 1992, 1994; Schmidt et
al., 1995), these changes cannot easily account for the observed
acceleratory effects of FMRFamide or cell 204. Our results here suggest
that in addition to the previously documented effects of FMRFamide,
this peptide elicits a novel voltage-gated outward current,
IKF, in oscillator heart interneurons. This
current is characterized by very slow activation and deactivation
kinetics. When this current is introduced into our model of an
elemental two-cell oscillator, the cycling rate of the model increases.
This result indicates that the primary effect of FMRFamide, which
accounts for its acceleratory action on the heartbeat motor pattern, is
its activation of IKF. The slow dynamics of
IKF give rise to a form of cellular memory
similar to that observed when an artificial conductance based on the
Kv1.3 (K+) channels, which have slow inactivation and
deinactivation kinetics, was introduced into lobster stomatogastric
neurons in culture (Turrigiano et al., 1996).
From experiment to model: slow kinetics of
IK2 contributed to slow oscillations
When we modified our canonical model of a heart interneuron
two-cell oscillator using the new measurements of
IK1 and IK2 activation
and deactivation kinetics, we discovered that slow oscillations in the
model could now be produced with a realistic value of graded synaptic
transmission (gSynG). Previously,
gSynG had to be increased to ~10 times the
value used in the canonical model (i.e., the realistic value) to
produce oscillations in the absence of action potentials (Nadim et al.,
1995). These oscillations are observed in heart interneurons when they
are recorded in reduced Na+/high Ca2+ saline
(Arbas and Calabrese, 1987; Nadim et al., 1995). During slow
oscillations in the model, the slow deactivation of
IK2 helps keep the cells in the hyperpolarized
phase long enough to remove inactivation from the low-threshold
Ca2+ currents that are required for producing the plateau
phase and graded inhibition during the slow oscillations. Therefore the large gSynG fudge factor that we used in Nadim
et al. (1995) is now unnecessary. The new, more realistic value of
gSynG does not affect the canonical model
described in Nadim et al. (1995) and Olsen et al. (1995), because
generation of oscillations in the canonical model depends primarily on
spike-mediated transmission, and only minimally on graded
transmission.
This reanalysis of outward currents in heart interneurons and their
modulation by FMRFamide has solidified our detailed model of a heart
interneuron two-cell oscillator and thus strengthened our understanding
of how this half-center oscillator works. Our experiments demonstrate
that a single neuromodulator affects multiple currents in a neuron, and
analysis of these currents within the context of our model indicates
that each makes a different functional contribution to the network
behavior. Moreover, our results indicate that currents with very slow
activation and deactivation kinetics can contribute substantially to
pattern generation.
FOOTNOTES
Received Nov. 27, 1996; revised March 17, 1997; accepted March 21, 1997.
This work was supported by National Institutes of Health Grants NS24072
and NS34975.
Correspondence should be addressed to Ronald L. Calabrese, Department
of Biology, Emory University, 1510 Clifton Road, Atlanta, GA 30322.
Dr. Nadim's present address: Volen Center, Brandeis University,
Waltham, MA 02254.
REFERENCES
-
Angstadt JD,
Calabrese RL
(1989)
A hyperpolarization-activated inward current in heart interneurons of the medicinal leech.
J Neurosci
9:2846-2857[Abstract].
-
Angstadt JD,
Calabrese RL
(1991)
Calcium currents and graded synaptic transmission between heart interneurons of the leech.
J Neurosci
11:746-759[Abstract].
-
Arbas EA,
Calabrese RL
(1984)
Rate modification in the heartbeat central pattern generator of the medicinal leech.
J Comp Physiol [A]
155:783-794.
-
Arbas EA,
Calabrese RL
(1987)
Slow oscillations is membrane potential in interneurons that control heartbeat in the medicinal leech.
J Neurosci
7:3953-3960[Abstract].
-
Arshavsky YI,
Orlovsky GN,
Panchin YV,
Roberts A,
Soffe SR
(1993)
Neuronal control of swimming locomotion: analysis of the pteropod mollusc Clione and embryos of the amphibian Xenopus.
Trends Neurosci
1993:16:227-233.
-
Calabrese RL,
De Schutter
(1992)
Motor-pattern-generating networks in invertebrates: modeling our way toward understanding.
Trends Neurosci
15:439-445[ISI][Medline].
-
Calabrese RL,
Nadim F,
Olsen ØH
(1995)
Heartbeat control in the medicinal leech: a model system for understanding the origin, coordination, and modulation of rhythmic motor patterns.
J Neurobiol
27:390-402[ISI][Medline].
-
Dean J,
Cruse H
(1995)
Motor pattern generation.
In: The handbook of brain theory and neural networks (Arbib MA,
ed), pp 600-605. Cambridge, MA: MIT.
-
Delcomyn F
(1980)
Neural basis of rhythmic behavior in animals.
Science
210:492-498[Abstract/Free Full Text].
-
De Schutter E,
Angstadt JD,
Calabrese RL
(1993)
A model of graded synaptic transmission for use in dynamic network simulations.
J Neurophysiol
69:1225-1235[Abstract/Free Full Text].
-
Dickinson PS
(1995)
Neuromodulation in invertebrate nervous systems.
In: The handbook of brain theory and neural networks (Arbib MA,
ed), pp 631-634. Cambridge, MA: MIT.
-
DiFrancesco D,
Noble D
(1989)
Current If and its contribution to cardiac pacemaking.
In: Neuronal and cellular oscillators (Jacklet JW,
ed), pp 31-58. New York: Marcel Dekker.
-
Evans BD,
Pohl J,
Kartsonis NA,
Calabrese RL
(1991)
Identification of RFamide neuropeptides in the medicinal leech.
Peptides
12:897-908[ISI][Medline].
-
Getting PA
(1989)
Emerging principles governing the operation of neural networks.
Annu Rev Neurosci
12:184-204.
-
Golowasch J,
Marder E
(1992)
Proctolin activates an inward current whose voltage dependence is modified by external Ca2+.
J Neurosci
12:810-817[Abstract].
-
Harris-Warrick RM
(1993)
Pattern generation.
Curr Opin Neurobiol
3:982-988[Medline].
-
Harris-Warrick RM,
Marder E
(1991)
Modulation of neural networks for behavior.
Annu Rev Neurosci
14:39-57[ISI][Medline].
-
Harris-Warrick RM,
Coniglio LM,
Levini RM,
Gueron S,
Guckenheimer J
(1995)
Dopamine modulation of two subthreshold currents produces phase shifts in activity of an identified motoneuron.
J Neurophysiol
74:1401-1420.
-
Hodgkin AL,
Huxley AF
(1952)
A quantitative description of membrane current and its application to conduction and excitation in nerve.
J Physiol (Lond)
117:500-544.
-
Jacklet JW
(1989)
In: Neuronal and Cellular Oscillators. New York: Marcel Dekker.
-
Johnson BR,
Peck JH,
Harris-Warrick RM
(1995)
Distributed amine modulation of graded chemical transmission in the pyloric network of the lobster stomatogastric ganglion.
J Neurophysiol
72:437-452.
-
Katz PS
(1996)
Neurons, networks, and motor behavior.
Neuron
16:245-253[ISI][Medline].
-
Kuhlman JR,
Li C,
Calabrese RL
(1985)
FMRF-amide-like substances in the leech. I. Immunocytochemical localizataion.
J Neurosci
5:2301-2309[Abstract].
-
Lu J, Dalton IV JF, Stokes DR, Calabrese RL (1997) Functional
role of Ca2+ currents in graded and spike-mediated synaptic
transmission between leech heart interneurons. J Neurophysiol, in
press.
-
Marder E,
Calabrese RL
(1996)
Principles of rhythmic motor pattern generation.
Physiol Rev
76:687-717[Abstract/Free Full Text].
-
Nadim F,
Olsen ØH,
De Schutter E,
Calabrese RL
(1995)
Modeling the leech heartbeat elemental oscillator: I. Interactions of intrinsic and synaptic currents.
J Computat Neurosci
2:215-235[ISI][Medline].
-
Olsen ØH (1994) Exploring temporal computation in neuronal
systems. PhD thesis, University of Glasgow.
-
Olsen ØH,
Calabrese RL
(1996)
Activation of intrinsic and synaptic currents in leech heart interneurons by realistic waveforms.
J Neurosci
16:4858-4970.
-
Olsen ØH,
Nadim F,
Calabrese RL
(1995)
Modeling the leech heartbeat elemental oscillator: II. Exploring the parameter space.
J Computat Neurosci
2:237-257[ISI][Medline].
-
Opdyke CA,
Calabrese RL
(1994)
A persistent sodium current contributes to oscillatory activity in heart interneurons of the medicinal leech.
J Comp Physiol [A]
175:781-789[Medline].
-
Rossignol S,
Dubuc R
(1994)
Spinal pattern generation.
Curr Opin Neurobiol
4:894-902[Medline].
-
Schmidt J,
Calabrese RL
(1992)
Evidence that acetylcholine is an inhibitory transmitter of heart interneurons in the leech.
J Exp Biol
171:339-347.
-
Schmidt J,
Gramoll S,
Calabrese RL
(1995)
Segment-specific effects of FMRFamide on membrane properties of heart interneurons in the leech.
J Neurophysiol
74:1485-1497[Abstract/Free Full Text].
-
Sharp AA,
O'Neil MB,
Abbott LF,
Marder E
(1993)
Dynamic clamp: artificial conductances in biological neurons.
Trends Neurosci
16:389-394[ISI][Medline].
-
Simon TW,
Opdyke CA,
Calabrese RL
(1992)
Modulatory effects of FMRFamide on outward currents and oscillatory activity in heart interneurons of the medicinal leech.
J Neurosci
12:525-537[Abstract].
-
Simon TW,
Schmidt J,
Calabrese RL
(1994)
Modulation of high-threshold transmission between heart interneurons of the medicinal leech by FMRF-NH2.
J Neurophysiol
71:454-466[Abstract/Free Full Text].
-
Turrigiano GG,
Marder E,
Abbott LF
(1996)
Cellular short-term memory from a slow potassium conductance.
J Neurophysiol
75:963-966[Abstract/Free Full Text].
This article has been cited by other articles:
|
|
|
|
 
A.-E. Tobin and R. L. Calabrese
Endogenous and Half-Center Bursting in Morphologically Inspired Models of Leech Heart Interneurons
J Neurophysiol,
October 1, 2006;
96(4):
2089 - 2106.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A.-E. Tobin and R. L. Calabrese
Myomodulin Increases Ih and Inhibits the Na/K Pump to Modulate Bursting in Leech Heart Interneurons
J Neurophysiol,
December 1, 2005;
94(6):
3938 - 3950.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A. L. Weaver and S. L. Hooper
Relating Network Synaptic Connectivity and Network Activity in the Lobster (Panulirus interruptus) Pyloric Network
J Neurophysiol,
October 1, 2003;
90(4):
2378 - 2386.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A. L. Weaver and S. L. Hooper
Follower Neurons in Lobster (Panulirus interruptus) Pyloric Network Regulate Pacemaker Period in Complementary Ways
J Neurophysiol,
March 1, 2003;
89(3):
1327 - 1338.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
G. S. Cymbalyuk, Q. Gaudry, M. A. Masino, and R. L. Calabrese
Bursting in Leech Heart Interneurons: Cell-Autonomous and Network-Based Mechanisms
J. Neurosci.,
December 15, 2002;
22(24):
10580 - 10592.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
M. A. Masino and R. L. Calabrese
A Functional Asymmetry in the Leech Heartbeat Timing Network Is Revealed by Driving the Network across Various Cycle Periods
J. Neurosci.,
June 1, 2002;
22(11):
4418 - 4427.
[Abstract]
| |
