First published online March 21, 2005
Journal of Experimental Biology 208, 1347-1361 (2005)
Published by The Company of Biologists 2005
doi: 10.1242/jeb.01500
Central pattern generator for swimming in Melibe
Stuart Thompson1,* and
Winsor H. Watson, III2
1 Department of Biological Sciences, Hopkins Marine Station, Stanford
University, Pacific Grove, CA 93950, USA
2 Zoology Department, Center for Marine Biology, University of New
Hampshire, Durham, NH 03824, USA
View larger version (89K):
[in a new window]
|
Fig. 1. Axon distributions of sint1 and sint2. (Top) A diagram of
the dorsal aspect of the Melibe (rostral at the top), illustrating
the three major subdivisions of the CNS: the cerebral ganglia (C), pleural
ganglia (Pl) and pedal ganglia (Pd). The cell bodies and major projections of
sint1 and sint2 are shown diagrammatically, based on
interpretation of 24 successful Lucifer Yellow dye fills. Examples of Lucifer
Yellow stained interneurons are shown below. Sint1 branches near its
cell body located just caudal to the tentacular lobe (T) in the pleural
ganglion and sends a process to the ipsilateral pedal ganglion that arborizes
in the pedal ganglion neuropil. It does not project across the midline
via the central commissure or beyond the CNS. Sint2 branches
in the pedal ganglion and sends a process via the pedal-pedal
connective to the opposite pedal ganglion. No processes were seen to project
to the pleural ganglia or leave the CNS. The dye fills illustrated in this
figure were done in isolated ganglion preparations.
|
|
View larger version (25K):
[in a new window]
|
Fig. 2. Firing pattern in swim interneuron 1. (A) Lower trace: intracellular
recording from Rsint1 during a short episode of swimming. Swimming
was initiated by separating the animal's foot from a surface at the first
arrow and terminated by returning the surface at the second arrow. Upper
trace: a record of the animal's side-to-side swimming movements (upward
deflection indicates bending toward the right). (B) Lower trace: bursting
activity in Rsint1 at higher gain and on an expanded time base.
Insert: the trajectory of membrane voltage at the end of a burst on an
expanded scale. Action potentials were truncated by the recording device in
this example. Upper trace: a record of the animal's movement. (C) Simultaneous
recording of activity in Rsint1 (middle) and Lsint1 (bottom)
during a long swimming episode along with the record of swimming movements
(top). Note that the spike bursts in these two cells do not overlap.
|
|
View larger version (23K):
[in a new window]
|
Fig. 3. Graded interruption of swimming by increasing hyperpolarization of
sint1. (A–D) The upper traces in A–D show activity in
Rsint1 during a continuous episode of swimming. The lower traces show
the animal's swimming movements. Hyperpolarizing currents were applied
via the recording electrode using a bridge circuit. The timing of
current pulses is indicated by solid bars drawn under the behavioral record,
and the strength of the hyperpolarizing current is indicated under each bar.
Changes in absolute membrane voltage during stimulation are inaccurate because
of errors in bridge balance.
|
|
View larger version (23K):
[in a new window]
|
Fig. 4. Resetting the phase of swimming by stimulating sint1. The firing
pattern in Lsint1 was recorded along with a record of behavior (upper
trace in each panel) during a continuous episode of swimming. (A–C)
Hyperpolarizing current pulses sufficient to prevent soma and axon spikes in
Lsint1 were applied via the recording electrode. The timing
and duration of pulses were varied (pulse timing indicated by solid lines
below the voltage traces). In each experiment, the interval between the
centers of Lsint1 bursts was measured during the 10 cycles preceding
the onset of the stimulus (indicated by vertical strokes above the voltage
recording). These measurements were used to calculate the means ±
S.D. of the swim period, which were then used to predict the time
of occurrence of the Lsint1 burst projected forward in time beyond
the period of stimulation, based on the assumption that stimulation of
Lint1 has no effect on pattern generation. The predicted times of the
center of Lsint1 bursts are shown as vertical lines above the voltage
recording, with the S.D. represented by horizontal tics through the
lines. The results show that the assumption fails and that the phase of
swimming behavior is reset by the stimulus. The difference between the
predicted and actual time of occurrence of the burst provides a quantitative
measure of phase resetting (see details in the text). (D) The experiment was
repeated with an extra burst driven in Lsint1. The extra burst also
reset the phase of swimming in a quantitative fashion (see text). The bridge
circuit used to deliver stimulating currents was imperfectly balanced and,
therefore, the absolute membrane voltage is not accurately represented during
periods of stimulation. This experiment was conducted 28 times in four
different whole animal preparations with consistent results.
|
|
View larger version (20K):
[in a new window]
|
Fig. 5. Reciprocal inhibition between sint1 neurons. (A,B). Simultaneous
recordings from Rsint1 (upper trace) and Lsint1 (lower
trace) in a quiescent whole animal preparation. In A, Rsint1 was
driven to fire a burst of action potentials while the voltage in
Lsint1 was recorded at 4 x higher gain. The same procedure was
followed in B, where Lsint1 was stimulated. (C,D). Unitary ipsp in
Lsint1 during stimulation of Rsint1. Rsint1 was driven to
fire either a single action potential (C) or a series of seven action
potentials at a rate of one per second (D, sweeps superimposed). Asterisks
indicate the time of occurrence of the ipsp. This experiment was repeated in
six different whole animal preparations with consistent results.
|
|
View larger version (29K):
[in a new window]
|
Fig. 6. Excitatory synaptic output from sint1 to synergistic motoneurons.
The dorsal aspect of the right half of the CNS, including the right pleural
and pedal ganglia, is shown diagrammatically (rostral direction upward). The
diagram also shows the relative positions of the soma of Rsint1 in
the pleural ganglion and of eight individual motoneurons in the ipsilateral
pedal ganglion that are known to participate in swimming behavior. Single
spikes were driven in Rsint1 at a rate of one per second in a
quiescent whole animal preparation while recording membrane voltage at higher
gain in the motoneurons. The results from 5–10 repetitions are
superimposed. In each case the spike in Rsint1 elicits an epsp in the
motoneuron. The time calibration bar (near A) corresponds to 40 ms
(A–C,G) and 20 ms (D–F,H). For the postsynaptic potentials, the
vertical calibration corresponds to 2 mV (A,B), 4 mV (F–H), and 20 mV in
the other recordings. The epsps in A and B occur with latencies of 20–40
ms, while those in C–H occur with latencies of 2–5 ms. These
results are characteristic of those obtained in 30 separate whole animal
preparations involving >150 paired recordings.
|
|
View larger version (39K):
[in a new window]
|
Fig. 7. Inhibitory synaptic output from sint1 to an antagonistic
motoneuron recorded in a quiescent whole animal preparation. (A) A driven
burst of action potentials in Rsint1 (lower trace) causes
hyperpolarization of an identified antagonistic motoneuron located in the
opposite pedal ganglion. (B) A driven burst in the motoneuron has no effect on
Rsint1.
|
|
View larger version (34K):
[in a new window]
|
Fig. 8. Firing pattern in swim interneuron 2 (sint2). (A) Activity in
Lsint2 during a brief episode of behavior. Swimming was initiated by
removing a surface from the animal's foot at the first arrow and terminated by
replacing the surface at the second arrow. (B) Action potential bursts in
Lsint2 shown at higher gain and on an expanded time scale. (C)
Alternating bursts in the two antagonistic sint2 neurons recorded
simultaneously during a long episode of swimming.
|
|
View larger version (29K):
[in a new window]
|
Fig. 15. The timing of bursts in sint1 and sint2 during swimming.
(A,B) Simultaneous recording from Rsint1 (A) and Rsint2 (B)
during an episode of swimming in a whole animal preparation. After this
recording was taken, the electrode was removed from Rsint1 and
Lsint1 was impaled. (C,D) Simultaneous recording from Rsint2
(C) and Lsint1 (D) during the same swimming episode. The two sets of
records were aligned to the midpoint of the center burst in Rsint2 so
that the timing of bursts could be compared. All recordings are from the same
whole animal preparation. Similar results were obtained in three separate
whole animal experiments.
|
|
View larger version (19K):
[in a new window]
|
Fig. 9. Resetting the phase of swimming by stimulating sint2. (Top) An
intracellular recording from Lsint2 during swimming; (bottom) a
record of the animal's movements. The movement detector saturated and did not
report the full range of side-to-side movement in this example.
Hyperpolarizing current sufficient to silence the cell was applied
via the recording electrode during the time indicated by the bar
under the voltage recording. The time of maximal right flexion (shown by
vertical lines above the behavior record) was measured during the 10 cycles
preceding the onset of the stimulus in order to calculate the mean ±
S.D. of the swimming period. These values were projected forward in
time to predict the expected time of occurrence of maximal right flexion The
bars and vertical lines above the behavioral record show the predicted time of
occurrence of peak right flexion after the stimulus ends. Horizontal tics show
the S.D. The difference between the predicted time of occurrence
and the actual occurrence of peak flexion provides a measure of phase
resetting. This experiment was repeated 15 times in three different whole
animal preparations with consistent results.
|
|
View larger version (16K):
[in a new window]
|
Fig. 10. Electrical coupling between sint1 and the synergistic
sint2. (A,B) Simultaneous recordings from Rsint1 (upper
traces) and Rsint2 (lower traces) in a quiescent whole animal
preparation. Hyperpolarizing current pulses were applied to one of the cells
via the recording electrode while membrane voltage was recorded in
the other. Bridge balance was checked using short hyperpolarizing pulses of
the same amplitude before and after the recording. (Ai) A current pulse
applied to Rsint1 causes hyperpolarization of Rsint2 due to
current flow across the electrical junction. (Aii) The reciprocal connection.
(Bi) The electrically coupled epsp in Rsint2 (lower trace; recorded
at higher gain) in response to driven action potentials in Rsint1
(eight superimposed traces, stimulus rate one per second). (Bii) The strongly
attenuated electrically coupled epsp recorded in Rsint1 (upper trace;
recorded at higher gain) in response to single action potentials driven in
Rsint2 (seven superimposed traces, stimulus rate one per second). (C)
Bode plot showing conduction of sinusoidal currents across the electrical
junction in both directions. Rsint1 and Rsint2 were
stimulated, one at a time, with subthreshold currents of fixed amplitude but
varying frequencies while recording the coupled sine wave in the other cell.
The electrical junction passes current symmetrically in both directions and
has the characteristics of a low-pass filter with a corner frequency of 1.5 Hz
and final slope of 6 dB per octave in frequency. These results were confirmed
in each of six separate whole animal preparations.
|
|
View larger version (22K):
[in a new window]
|
Fig. 11. Mutual inhibition between antagonistic sint2 neurons. Simultaneous
recordings from Lsint2 and Rsint2 in a quiescent whole
animal preparation. (A) Lsint2 (upper trace) was driven to fire two
bursts of action potentials while recording membrane voltage in
Rsint2 (lower trace) at higher gain (time of stimulus marked by bar).
(B) Stimulation of Rsint2 to fire three bursts while recording from
Lsint2. Action potentials were truncated by the recording device in
the high gain records. The firing frequency during driven bursts was similar
to what is seen during normal swimming (see
Fig. 8). Similar results were
obtained in each of five whole animal experiments.
|
|
View larger version (14K):
[in a new window]
|
Fig. 12. Mutual inhibition between the antagonistic sint1 and
sint2. Simultaneous recordings from Lsint2 (upper traces)
and Rsint1 (lower traces) in a quiescent whole animal preparation.
(A) Single action potentials were driven in Lsint2 at a rate of one
per second while recording from Rsint1 at higher gain (7 superimposed
traces). (B) The reciprocal ipsp in Lsint2 during stimulation of
single action potentials in Rsint1 at 1 Hz (6 superimposed traces).
Similar results were obtained in each of 12 separate whole animal
experiments.
|
|
View larger version (30K):
[in a new window]
|
Fig. 13. Synaptic output from sint2 to motoneurons. (A) Excitatory output
from sint2 to a synergistic motoneuron recorded in a quiescent
isolated brain preparation. Sint2 (lower trace) was driven to fire a
burst of action potentials by direct stimulation while recording membrane
voltage in the motoneuron (upper trace). Individual spikes in sint2
correspond with unitary epsps in the motoneuron. (B) Ipsps in an antagonistic
motoneuron coincident with action potentials in sint2 during fictive
swimming in an isolated brain preparation.
|
|
View larger version (24K):
[in a new window]
|
Fig. 14. Inhibitory output from sint2 to an antagonistic motoneuron. (A)
Single action potentials were driven in sint2 at a rate of one per
second while recording from an antagonistic motoneuron in the opposite pedal
ganglion in a quiescent whole animal preparation. Two traces are superimposed.
The ipsp occurs with a delay of about 35 ms. (B) A driven burst in
sint2 causes sustained hyperpolarization in the motoneuron due to
summation of ipsps. These results are characteristic of 7 separate
experiments.
|
|
View larger version (33K):
[in a new window]
|
Fig. 16. Recordings from sint1 and sint2 during turning movements.
(A) A turning movement to the right was initiated by a tactile stimulus
applied to the left side of the body while recording from Rsint1
(upper trace) and Rsint2 (lower trace). (B) Activity in the same two
cells during a spontaneous turn to the right. A and B are from a whole animal
preparation that was allowed to crawl on a blade of seagrass during the
recording. Similar observations were made in each of 12 separate whole animal
experiments.
|
|
View larger version (35K):
[in a new window]
|
Fig. 17. Network model for the Melibe swim CPG. The mutually inhibitory
connections linking the left (L) and right (R) sint1 and
sint2 neurons are shown by lines terminating in circles. The
electrical synapses between the synergistic sint1 and sint2
neurons are shown by lines terminating in bars.
|
|
© The Company of Biologists Ltd 2005
