Glutamatergic N2v Cells Are Central Pattern Generator Interneurons of
the Lymnaea Feeding System: New Model for Rhythm Generation
M. J. BRIERLEY, M. S. YEOMAN, AND P. R. BENJAMIN
Sussex Centre for Neuroscience, School of Biological Sciences, University of Sussex, Brighton, East Sussex BN1 9QG,
United Kingdom
INTRODUCTION
The N2v cells are rhythmically active plateauing interneurons forming part of the network controlling feeding in the
snail, Lymnaea. They have complex synaptic connections
with feeding motor neurons (Brierley et al. 1997a) that appear to be mediated by glutamate (Brierley et al. 1997b).
Three types of evidence suggests that they may be part of
the feeding central pattern generator (CPG) (Brierley et al.
1997a): they are driven by the modulatory interneuron called
3396
the slow oscillator (SO) and fire during the N2/rasp phase
of the feeding rhythm; they are coupled electrotonically to
other retraction phases CPG interneurons the N2d (dorsal)
cells; and they have a pattern of synaptic connections with
motor neurons, suggesting that, with the N2ds, they control
the rasp-swallow phases of the feeding cycle involved in
food ingestion movements. The work discussed in this paper
will provide further and more conclusive evidence for the
role of the N2vs in the Lymnaea feeding CPG by describing
their synaptic connections with other interneurons of the
feeding network.
Lymnaea feeding interneurons divide into two types, the
N cells of the CPG itself and modulatory neurons like the
slow oscillator (SO) and the serotonergic cerebral giant cells
(CGCs), equivalent to the metacerebral giant cells of other
mollusks (e.g., Weiss et al. 1978). So far, five types of CPG
interneurons have been described, the N1Ms (N1 medials)
and N1Ls (N1 laterals), the N2d (dorsals), and the N3t
(tonics) and N3p (phasics). We will argue that the N2vs
are a sixth member of the CPG network and that they are
the principal element active during the rasp phase of the
feeding cycle. A new model of the Lymnaea CPG network
will be presented to account for activity in all six cell types
of CPG interneuron. It will reveal a pattern of synaptic connectivity of remarkable richness with the duration and
strength of synaptic connections playing an important role
in determining the timing of different phases of the feeding
pattern.
METHODS
Previously described methods were used to carry out multiple
simultaneous intracellular recordings from interneurons and motor
neurons of the Lymnaea feeding network (Brierley et al. 1997a;
Yeoman et al. 1994a). The twisted buccal ganglion preparation
was used to simultaneously record from N2v on the ventral surface
and different combinations of the SO, N1M, N1L, N3t, N3p, and
N2d interneurons on the dorsal surface of the buccal ganglia (Fig.
1). The only nonbuccal CNS recorded were the CGCs. These are
a symmetrical pair of uniquely identifiable giant neurons (90- to
120-mm diam) located on the medial surfaces of the anterior lobes
of the cerebral ganglia (McCrohan and Benjamin 1980). They can
be identified easily by visual observation. N-2-hydroxyethylpiperazine-N *-2-ethanesulfonic acid–buffered normal saline (Benjamin
and Winlow 1981) or, alternatively, high Mg 2/ /high Ca 2/ (HiDi)
saline to raise the threshold for polysynaptic connections (Elliott
and Benjamin 1989), were perfused continuously over the preparation. Acetylcholine was applied focally to the surface of the N2v
or N2d cells by pressure pipette while continuously bath perfusing
the preparation with normal saline.
0022-3077/97 $5.00 Copyright q 1997 The American Physiological Society
/ 9k22$$de26 J862-6
11-26-97 13:07:34
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 18, 2017
Brierley, M. J., M. S. Yeoman, and P. R. Benjamin. Glutamatergic N2v cells are central pattern generator interneurons of
the Lymnaea feeding system: new model for rhythm generation.
J. Neurophysiol. 78: 3396–3407, 1997. We aimed to show that
the paired N2v (N2 ventral) plateauing cells of the buccal ganglia
are important central pattern generator (CPG) interneurons of the
Lymnaea feeding system. N2v plateauing is phase-locked to the
rest of the CPG network in a slow oscillator (SO)-driven fictive
feeding rhythm. The phase of the rhythm is reset by artificially
evoked N2v bursts, a characteristic of CPG neurons. N2v cells
have extensive input and output synaptic connections with the rest
of the CPG network and the modulatory SO cell and cerebral
giant cells (CGCs). Synaptic input from the protraction phase
interneurons N1M (excitatory), N1L (inhibitory), and SO (inhibitory-excitatory) are likely to contribute to a ramp-shaped prepotential that triggers the N2v plateau. The prepotential has a highly
complex waveform due to progressive changes in the amplitude
of the component synaptic potentials. Most significant is the facilitation of the excitatory component of the SO r N2v monosynaptic
connection. None of the other CPG interneurons has the appropriate
input synaptic connections to terminate the N2v plateaus. The modulatory function of acetylcholine (ACh), the transmitter of the SO
and N1M/N1Ls, was examined. Focal application of ACh (50-ms
pulses) onto the N2v cells reproduced the SO r N2v biphasic
synaptic response but also induced long-term plateauing (20–60
s). N2d cells show no endogenous ability to plateau, but this can
be induced by focal applications of ACh. The N2v cells inhibit the
N3 tonic (N3t) but not the N3 phasic (N3p) CPG interneurons.
The N2v r N3t inhibitory synaptic connection is important in
timing N3t activity. The N3t cells recover from this inhibition and
fire during the swallow phase of the feeding pattern. Feedback N2v
inhibition to the SO, N1L protraction phase interneurons prevents
them firing during the retraction phase of the feeding cycle. The
N2v r N1M synaptic connection was weak and only found in
50% of preparations. A weak N2v r CGC inhibitory connection
prevents the CGCs firing during the rasp (N2) phase of the feeding
cycle. These data allowed a new model for the Lymnaea feeding
CPG to be proposed. This emphasizes that each of the six types
of CPG interneuron has a unique set of synaptic connections, all
of which contribute to the generation of a full CPG pattern.
N2V CELLS ARE PART OF THE FEEDING CPG
RESULTS
Resetting of the feeding CPG by the N2v cells
One type of evidence that would indicate that the N2v
cells were part of the CPG would be if artificial activation
could reset the rhythm in the rest of the CPG network. An
example of five such experiments is shown in Fig. 2. Here
fictive feeding was generated by steady depolarization of the
SO. This produced bursting in the SO itself and phase-locked
activity in the N2v cell. Bursts of N2v spikes coincided with
large hyperpolarizing inputs on the SO, indicating, on the
basis of previous analysis (Rose and Benjamin 1979) that
bursts of spikes on the N2v occur during the N2/rasp phase
of the feeding cycle. Once the regular feeding pattern was
generated, an extra burst of activity was evoked by current
injection in the N2v during the interburst interval of the ongoing rhythm. This delayed the onset of the next SO-driven
N2v burst and subsequent bursts in both cells (Fig. 2). This
sudden delay was greater than would have been predicted
by the progressive slowing of the rhythm that also was occurring in Fig. 2 (a common feature of SO-driven rhythms).
It was obvious that the ‘‘extra’’ evoked burst of N2v cell
spikes caused a hyperpolarization in the SO (upward sloping
arrow) due to an inhibitory synaptic connection from the
N2v cell to the SO (see further text). This would have been
partly responsible for the resetting ability of the N2v cell,
but further synaptic connections that the N2v makes with
other members of the CPG network are also important. These
could include the electrotonic coupling with the N2d cells
(Brierley et al. 1997a).
Synaptic inputs to the N2vs
Based on a previous analysis of the feeding system (Elliott
and Benjamin 1985a,b; Yeoman et al. 1993, 1995), examination of the waveforms and timing of synaptic inputs to
/ 9k22$$de26 J862-6
the N2v cells enabled predictions to be made about the likely
interneurons providing these inputs. During a SO-driven fictive feeding rhythm, the N2v plateaus were triggered by a
gradual depolarizing prepotential (Fig. 3). This prepotential
was highly complex, consisting of depolarizing and hyperpolarizing components, likely to be originating from several
types of presynaptic interneurons.
At the start of SO spike activity (Fig. 3) the N2v received
small unitary inhibitory postsynaptic potentials (IPSPs) that
appeared to be 1:1 with SO action potentials (confirmed in
Fig. 4). At the beginning of N1/P (protraction) phase (shallow smooth hyperpolarizing input on B3) of the cycle, the
N2v began to slowly depolarize, and the plateau was triggered at the end of this phase. However, superimposed on
the underlying depolarizing waveform (ramp shaped) were
unitary IPSPs that were again 1:1 with the SO spikes. Despite
their large amplitude, these IPSPs did not prevent the triggering of the plateau in the N2v. The plateau appeared to
be generated endogenously as it persisted when chemical
synaptic connections were blocked (Brierley et al. 1997a).
During the N3 phase of the cycle (N3 IPSPs on the SO and
excitatory postsynaptic potentials (EPSPs) on the B3), there
were no clear PSPs on the N2v that corresponded with the
N3t (N3 tonic) input to the B3 cell. This suggested that the
falling phase of the plateau was due to an intrinsic property
of the N2v. This was supported by evidence reported in
Brierley et al. (1997a) showing that the N2v plateau, including the falling phase, still occurred when all (chemical)
synaptic input was blocked in high Mg 2/ /zero Ca 2/ saline.
These data suggested that the N2vs were hyperpolarized
by the SO during the late N3 phase and N1/protraction phase
of the fictive feeding cycle. They also were depolarized during the N1 phase, which could be due to N1 and/or SO
inputs, but appeared to receive no synaptic input from the
N3 swallow phase CPG interneurons. These predicted patterns of synaptic connections were tested by direct paired
recordings between the N2vs and the other interneuronal
types.
FIG . 2. Current injection into the slow oscillator (SO) drives a fictive
feeding rhythm in the N2v cells. Bursts of N2v spikes occur at the same
time as large hyperpolarizing waves in the SO. This is defined as the
N2/rasp phase of the feeding rhythm based on previous behavioral analysis.
A short depolarizing pulse evokes an ‘‘extra’’ burst of spikes in the N2v
cell (electrode unbalanced) and a corresponding hyperpolarizing wave in
the SO (upward sloping arrows). This delays the occurrence of subsequence
bursts in the both the recorded cells. This delay is greater than would be
expected by the gradual slowing of the rhythm that also was occurring.
11-26-97 13:07:34
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 18, 2017
FIG . 1. Schematic diagram of neurons in the paired buccal ganglia of
Lymnaea in the ‘‘twisted’’ ganglion preparation. Ventral surface of the right
buccal ganglion shows the N2v (N2 ventral) interneuron and three types
of motoneuron (B4Cl, B8, and B9 cells). Dorsal surface of the left buccal
ganglion shows N1L (N1 lateral), N1M (N1 medial), N2d (N2 dorsal),
N3p (N3 phasic), and N3t (N3 tonic) central pattern generator interneurons
as well as the B1, B2, B3, B4, B4Cl, B5, B7, B8, B9, and B10 types of
motoneurons. This type of preparation allows simultaneous intracellular
recordings from cells on the ventral and dorsal surfaces.
3397
3398
M. J. BRIERLEY, M. S. YEOMAN, AND P. R. BENJAMIN
This SO r N2v biphasic synaptic connection appeared to
contribute significantly to the N2v prepotential during the
SO-driven rhythm shown in Fig. 3. It is responsible for the
unitary IPSPs seen during the late N3 phase (SO), the large
IPSPs, and the gradually depolarizing waveform occurring
during the P phase (N1 / SO) of the N2v feeding cycle.
The facilitating depolarizing component of the SO r N2v
response is likely to be particularly important in generating
the ramp-shaped depolarization of the N2v prepotential. The
progressive increase in this depolarization due to a train of
SO spikes is clear in Fig. 4A.
N1M r N2v
SO r N2v
Experiments of the type shown in Fig. 4 (n Å 6) showed
that the SO had striking but variable amplitude biphasic
synaptic effects on the N2v cells. The preparation had been
bathed in HiDi saline for 20 min to block any polysynaptic
pathway. Stimulation of the SO therefore revealed its postsynaptic effects in isolation from any other pathway and
produced a unitary, presumably monosynaptic, potential. Examination of the response to a train of SO spikes (Fig. 4A)
showed that the biphasic PSP consisted of an initial faster
hyperpolarizing component, followed by a slower depolarizing waveform. This second depolarizing component only
became prominent after five or more SO spikes (Fig. 4A).
The PSPs in the N2v followed 1:1 with a constant latency
(10 ms), reinforcing the evidence for a monosynaptic connection (Fig. 4B, i and ii). Both the hyperpolarizing and
depolarizing components of the SO r N2v PSP facilitated
with repetitive firing of the SO (Fig. 4A). During the first
five SO spikes, the amplitude of the hyperpolarizing component increased most rapidly (Fig. 4Bi) with little increase
in the second depolarizing component obvious until the fifth
spike of the SO train. After the fifth or sixth spike, the
depolarization became increasingly important, causing an
overall depolarization of the N2v. After Ç12 SO spikes, the
depolarizing component reached a maximum amplitude and
remained at this level (Fig. 4Bii).
/ 9k22$$de26 J862-6
FIG . 4. SO r N2v biphasic synaptic connection. A: current-driven
spikes in the SO generates 1:1 synaptic potentials in the N2v. These are
predominantly hyperpolarizing at the beginning of the SO spike train (1–
5 spikes), but a delayed depolarizing potential becomes obvious later and
increases in amplitude as the SO spike train proceeds. Both components
of the biphasic response reach a maximum amplitude after Ç12 spikes.
Experiment was carried out in high Mg 2/ /high Ca 2/ (HiDi) saline suggesting a SO r N2v monosynaptic connection. Bi: progressive changes in
amplitude of the inhibitory (i) component of the response is shown more
clearly when the SO spike triggers the oscilloscope trace with 5 superimposed traces corresponding to SO spikes 1–5 in A. Bii: by 12–14 spikes
the amplitude of both inhibitory and excitatory (e) components remain
constant. In A and B, the responses follow SO spikes 1:1 with a constant
latency of Ç10 ms. This again provides evidence for a monosynaptic connection.
11-26-97 13:07:34
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 18, 2017
FIG . 3. Synaptic inputs received by the N2v cells. One cycle of fictive
feeding activity recorded on the SO, N2v, and B3 motoneuron in a
SO-driven fictive feeding rhythm. On this faster time base, the waveforms
of N2 synaptic inputs during the N1/P (protraction), N2/R (rasp), and
N3/S (swallow) phases of the feeding cycle can be examined. Small unitary
inhibitatory postsynaptic potentials (IPSPs) occur during the late N3 phase
that are 1:1 with SO spikes. Beginning of the N1 phase is indicated by the
shallow hyperpolarizing potential on the B3 motoneuron. During the
N1/P phase of the fictive feeding cycle, the N2v gradually depolarizes with
an overall ramp-shaped waveform. However, large unitary IPSPs also occur
that are 1:1 with SO spikes. This input during N1 is likely to be due to SO
input but also N1 CPG interneurons active at the same time (Figs. 4–6).
At the end of N1, a N2v plateaus triggered with a sudden depolarization
of Ç30 mV. N2v spikes continue throughout the N2/R phase. Plateau falls
in the first part of the N3 phase of the feeding cycle. This is probably
mainly due to the endogenous properties of the N2v, but small inflections
on the falling phase of the plateau potential may indicate a small contribution from synaptic input. These inputs do not appear to correspond to the
N3t inputs recorded on the B3 cell.
We tested the hypothesis that the N1M cells provide a
depolarizing component of the prepotential that triggers N2v
plateaux by recording N1Ms with N2v cells in the twisted
preparation (Fig. 5B, n Å 6). This was most instructive in
preparations where slow spontaneous fictive feeding rhythms
occur in the absence of SO activity. In the example shown
in Fig. 5A, a slow depolarizing wave ( f ) preceded in the
N2v plateaus in the absence of inhibitory synaptic potentials,
indicating that the SO was inactive. This was important because it suggested that the N1Ms may be generating the
depolarizing potential in the N2v alone. The record in Fig.
5Bi showed that the depolarization of the N2v started exactly
at the moment when the N1M began to fire. In these spontaneous patterns, the duration of N1M firing is much longer
N2V CELLS ARE PART OF THE FEEDING CPG
FIG . 5. N1M r N2v excitatory synaptic connection. SO was inactive
in the 2 preparations shown here, allowing the N1M contribution to the N1
phase prepotential in the N2v cell to be assessed conveniently. A: in this
spontaneous pattern, spike activity in the N1M is accompanied by a slowly
rising depolarizing potential in the N2v ( f ). Bi: details of this are shown
on this faster time base display. Fluctuations in firing rate in the N1M
( x ) produce corresponding changes in the N2v depolarizing potential.
Eventually this triggers a N2v plateau. Bii: steady current injection in the
N1M drives spikes and a depolarizing postsynaptic potential in the N2v.
Again fluctuations in spike rate in the N1M are marked ( z ). Unitary
postsynaptic potentials are never seen to follow N1M spikes (see text for
discussion).
/ 9k22$$de26 J862-6
synaptic connection shown in Fig. 4. The N1M r N2v synaptic connection may be monosynaptic but this was not certain on the basis of the data described here.
N1L r N2v
Recently a new type of N1 CPG cell type, the N1L, was
described (Yeoman et al. 1995). These were a single pair
of symmetrical cells, one in each buccal ganglion, that fired
during the N1/protraction phase of a SO-driven fictive feeding pattern (Fig. 6 A) because of their electrotonic coupling
with the SO (Yeoman et al. 1995). Although they only fired
relatively few spikes during the N1 phase compared with
N1Ms, they still could be contributing to the complex synaptic input the N2v receives during this protraction phase of
the feeding cycle (Fig. 3B). Simultaneous recording of the
N1Ls with the N2v cells (n Å 4) in the twisted preparation
show that they inhibited the N2v cells, the opposite to the
N1Ms. Evoking a burst of spikes in the N1L generated an
inhibitory wave on the N2v (Fig. 6B). This was followed
by a delayed depolarization and bursts of spikes in the N2v.
We have no evidence that this excitation is due to direct
N1L r N2v effects because detailed examination of the
PSPs after each N1L spike never showed a depolarizing
component to the response (Fig. 6, C and D). It is more
likely to be due to the strong N1L r N1M excitatory synaptic
connection described by Yeoman et al. (1995) and therefore
an indirect effect due to the N1L r N1M r N2v pathway,
where the N1Ms rather than the N1L cells excited the N2v
(Fig. 5). Examining the specific N1L r N2v response, when
the N1L was made to fire by current injection at more physiological rates (Fig. 6C), showed that each N1L spike was
followed by a unitary IPSP in the N2v. These were of constant latency (5 ms in Fig. 6D) and appeared to be monosynaptically produced. The interesting feature of the IPSPs
evoked by a train of NIL spikes was that, after an initial
facilitation, they gradually reduced in amplitude as the train
progressed (Fig. 6, C and D), indicating synaptic depression.
No change in membrane potential (e.g., long-term hyperpolarization) could account for this reduction in IPSP amplitude (dashed line in Fig. 6C, N2v trace). This gradual decrease in IPSP amplitude was the opposite to the facilitatory
nature of the SO r N2v biphasic synaptic connection.
It has proved technically impossible to record N2v, N1L,
and SO (all three small interneurons) at the same time to
produce a SO-driven rhythm, so the likely effects of the
N1L r N2v synaptic input only could be assessed indirectly
on the basis of their firing pattern and synaptic potentials
recorded on the N2v. The N1Ls typically fired three to five
spikes in a SO-driven rhythm (Fig. 6A) and were likely to
contribute, with the SO, to the IPSPs occurring during the
P (protraction) phase of complex synaptic input received by
the N2v and shown in Fig. 3.
It seemed possible that the N1Ls were providing a second
inhibitory component to the N2v cells during the N1/protraction phase so that the complex waveform of the prepotential (ramp shaped), which triggered the N2v plateau (Fig.
3), was presumably due to a combination of excitatory
(N1M), inhibitory (N1L), and inhibitory/excitatory (SO)
synaptic inputs. Despite the inhibitory component of the
prepotential, the combined excitatory effects of the SO and
11-26-97 13:07:34
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 18, 2017
than in SO-driven rhythms (Elliott and Benjamin 1985b),
increasing the period of each feeding cycle. Under these
circumstances, fluctuations in firing rate of the N1M often
occurred and these were accompanied by corresponding
peaks in the depolarizing waveform of the N2v (Fig. 5Bi,
x ). Eventually the plateau was triggered followed by a
cessation of N1M firing. More direct evidence that the N1Ms
are connected synaptically to the N2v cells comes from the
experiment shown in Fig. 5Bii where driving the N1M by
current injection depolarized the N2v. Again fluctuations in
firing rate (indicated by a reduction in spike amplitude)
caused corresponding changes in the compound PSP waveform. We were never able to see unitary EPSPs in the N2v
after N1M stimulation. This is partly because the N1M always fire high-frequency sustained bursts when depolarized
leading to summated postsynaptic potentials but also because
they are an electrotonically coupled network of several cells
in each buccal ganglia (Rose and Benjamin 1981a). They
fire together when one N1M is depolarized, although rarely
1:1. Therefore injecting current into one N1M causes similar
activity in the other N1Ms that are also likely to be synaptically connected to the N2vs. The result is a summating EPSP
in the N2v, probably due to activity in several presynaptic
N1Ms. This was similar to the result obtained with the other
type of N2 cell, the N2d, that also was depolarized by summated potentials from the N1M cells (Elliott and Benjamin
1985a).
We concluded that the depolarizing prepotential on the
N2v cells seen in a SO-driven rhythm (Fig. 3B) was partly
due to N1M r N2v excitatory synaptic connection as well
as the depolarizing component of the SO r N2v biphasic
3399
3400
M. J. BRIERLEY, M. S. YEOMAN, AND P. R. BENJAMIN
N1M finally triggers a N2v plateau leading to the behavioral
transition between protraction and retraction (rasp).
Mimicking the SO/N1M synaptic inputs to the N2v by
acetylcholine application produced long-lasting
modulatory effects
Considerable evidence indicated that acetylcholine (ACh)
was the main transmitter of both the SO (Yeoman et al.
1993) and the N1 cells (N1Ms: Elliott and Kemenes 1992;
N1Ls: Vehovszky and Elliott 1995). These data predict that
application of ACh onto the N2vs should evoke biphasic
(hyperpolarizing/depolarizing) responses to mimic the postsynaptic effects of both the SO and the N1M/N1L cells on
the N2vs. This was shown to be true in all five experiments
where it was tried. In the example shown in Fig. 7A, a
50-ms focal application of 10 04 M ACh onto the surface of
a N2v evoked an initial hyperpolarization followed by a
depolarization. However, the depolarization lasted for longer
than the SO r N2v synaptic response, 20 s compared with
Ç50 ms, despite the continuous perfusion of the preparation
with normal saline. At higher pipette concentrations (10 02
M) (Fig. 7B), the long-lasting depolarizing component of
the ACh response evoked plateau in the N2v and EPSPs on
the B3 motoneuron due to the N2v r B3 synaptic connection
(Brierley et al. 1997a). ACh (10 02 M) also could evoke
long periods of lower amplitude plateaus without spike activity (Fig. 7C) in another preparation. Additional activation
/ 9k22$$de26 J862-6
FIG . 7. Modulatory effect of acetylcholine (ACh) on the N2v and N2d
cells. A: a 50-ms focal application of 10 04 M ACh onto the surface of
the N2v causes an initial hyperpolarization (i) followed by a long-lasting
depolarizing wave. Initial fast transient is a pressure artifact. B: at the
higher concentration of 10 02 M, the delayed depolarizing component of the
response triggered N2v plateaus. These resulted in corresponding burst
activity in the B3, presumably due to the previously described N2 r B3
excitatory synaptic connection. C: in this preparation, a 50-ms application
of 10 02 M ACh produced long periods of plateauing activity. Later in the
record spikes were superimposed on the plateau potentials. Initial part of
the response to ACh is unclear in this record because the N2v cell is at the
equilibrium potential for the hyperpolarization. Depolarizing the SO recorded at the same time showed the expected N2v r SO inhibitory synaptic
potentials. D: similar biphasic responses were seen with ACh in the N2d
cell. Plateau potentials accompanied the delayed depolarizing response.
11-26-97 13:07:34
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 18, 2017
FIG . 6. N1L r N2v inhibitory synaptic connection. A: firing pattern of
the N1L in a SO-driven fictive feeding rhythm. B: a strong burst of currentevoked N1L spikes induced inhibition (i) followed by excitation in the
N2v cell (see text for detailed interpretation). C: driving the N1L at more
physiological rates generates unitary IPSPs in the N2v that follow 1:1 with
N1L spikes. IPSPs first facilitate (1–4 spikes) then gradually reduce in
amplitude up to the 18th spike. D: superimposed N1L spikes triggered the
trace, and constant latency IPSPs follow in the N2v. Numbered IPSPs
correspond to those in C.
of the SO in this example increased the amplitude of the
plateau with associated spiking.
The lack of activity of other components of the CPG in
the preparation shown in Fig. 7, A and B, suggested that
ACh was capable of producing these long-term modulatory
effects on the N2v without effects on other elements of the
CPG. All other CPG neurons or modulatory elements such
as the SO are on the opposite dorsal surfaces of the buccal
ganglia and could not have been affected by focal application
of the ACh onto the ventrally located N2vs. The pipette
concentration of ACh used was high (10 02 M) and so was
probably producing an exaggerated effect, but the actual
concentration of ACh reaching the receptors was thought to
be lower (probably 1–2 orders of magnitude less) under
the constant perfusion conditions used here (Walden et al.
1988).
It was interesting that focal application of 10 04 M ACh
produced similar biphasic (hyperpolarizing followed by depolarizing) responses on the other type of N2 cell, the N2d
cells (Fig. 7D, n Å 3). Again prolonged depolarizing responses were seen (Fig. 7D) with plateau potentials repeatedly triggered, like the N2vs (Fig. 7C). As these experiments were carried out in high Mg 2/ /zero Ca 2/ saline to
block chemical synapses, the effects of ACh must be a direct
effect on the N2d cells. It appeared that ACh is capable of
producing modulatory effects on both types of N2 cells and
N2V CELLS ARE PART OF THE FEEDING CPG
3401
that this could contribute to the long-term oscillatory activity
in the feeding CPG network.
Lack of N3t and N3p r N2v synaptic connections
Synaptic connections from the N2v cells to other feeding
interneurons
The above experiments showed that the N2v cells received
the appropriate input synaptic connections to be an integral
part of the CPG network. This entrained their endogenous
plateauing activity to the rest of the CPG network and ensured that they fired during the correct rasp/N2 phase of the
feeding cycle. The following results will indicate that the
N2v cells also possess important output synaptic connections
with other known members of the CPG network and the
modulatory SO and CGC interneurons.
N2v r N1M, N1L
Both types of N1 cells were hyperpolarized strongly during the N2/rasp phase of the feeding cycle (Figs. 5A and
6A), and the N2vs fired appropriately to provide this inhibitory feedback. Paired recordings of the N1 cells and the
N2v (n Å 10) were carried out to test the hypothesis that
the N2vs provide the main inhibitory synaptic input to the
N1M/N1L cells.
The N2vs inhibited both types of N1 cells (Fig. 9), but
the response was much stronger for the N1Ls compared with
the N1Ms. In the record shown in Fig. 9A, spontaneous
bursts of spikes in the N2v were accompanied by inhibition
of the N1L as expected. Evoking bursts of N2v spikes by
current injection generated a similar strong hyperpolarization
in the N1L at firing levels of membrane potential providing
more direct evidence for the N2v r N1L synaptic connection. This experiment was carried out in HiDi saline suggesting a monosynaptic connection. However, episodes of
weaker hyperpolarization in the N1Ls were seen when the
recorded N2vs were not active and only showed subthreshold
plateau potentials (i in Fig. 9A). IPSPs were reduced in
amplitude compared with inputs where the N2v was active
/ 9k22$$de26 J862-6
FIG . 8. Absence of N3p or N3t r N2v synaptic connections. A: evoked
N3p spikes produced no response on the N2v cell. Bi: N3t cell produced
1:1 EPSPs on the B3 motoneuron but nothing on the N2v cell. Bii: this is
confirmed on the faster time base record where N3t triggered the oscilloscope trace.
but the residual hyperpolarizations still have to be explained.
It was possible that the unrecorded contralateral N2v was
active, but this seems unlikely as left and right N2vs normally fired together due to strong electrotonic coupling (Brierley et al. 1997a). More likely was that it was due to
N2d activity through the monosynaptic N2d r N1L synaptic
connection reported by Yeoman et al. (1995). An interesting
difference was seen on the B3 motoneuron compared with
the N1L in Fig. 9A. Lack of N2v spikes meant a complete
loss of EPSP synaptic input to the B3, suggesting that the
N2v r B3 connection was predominant in providing N2
input.
The N2v r N1M inhibitory connection was much weaker
than that for the N1Ls and only occurred in 50% of the cells
tested. Evoked bursts of spikes only moderately slowed N1M
activity, although a clear (reversed) IPSP could be seen
when the N1Ms were inactive (arrowed in Fig. 9B). The
other type of N2, the N2d also is known to inhibit the N1Ms
(Elliott and Benjamin 1985a). These can stop N1M firing
and appear to be more important than the N2vs.
N2v r N3t
N3t (tonics) are inhibited throughout the N2/rasp phase
of the feeding cycle (Elliott and Benjamin 1985a) (summa-
11-26-97 13:07:34
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 18, 2017
Recording of the N2vs in a SO-driven rhythm suggested
that the termination of the N2v plateau was not due to the
activity of the N3 interneurons (Fig. 3). These fire immediately after the N2vs (see Fig. 14) and could have been
terminating the plateau by feedback due to inhibitory synaptic connections. This lack of N3 r N2v connectivity was
tested directly by the experiments shown in Fig. 8. No synaptic response was recorded on the N2v cell as a result of the
evoked spikes in the N3p cell recorded at the same time
(Fig. 8A). The membrane potential of the N2vs was varied
but still no synaptic response was seen. Similar recordings
with the N3t cells showed no synaptic responses on the N2v
after evoked action potentials in the N3t cells (Fig. 8). The
N3t r B3 synaptic connection produced the expected unitary
EPSPs on the B3 recorded at the same time (Fig. 8B, i and
ii). This suggested that the N2v plateaus were terminated
by an endogenous mechanism, confirming the data from
Brierley et al. (1997a) where the falling phase of the N2v
plateau potentials remained unchanged when all chemical
synapses had been blocked in high Mg 2/ /low Ca 2/ saline.
3402
M. J. BRIERLEY, M. S. YEOMAN, AND P. R. BENJAMIN
whereas the N3ts are inhibited for much longer due to the
long-duration N2v burst and recover and fire at the beginning
of the N3/swallow. The N2v/N2d cells both had similar
differential effects on retraction phase motoneurons (Brierley et al. 1997a), and this was crucial for producing the
sequence of motoneuron firing required for the rasp/swallow
phases of the behavioral feeding cycle.
N2v r SO
Like the N1 cells, the SO was hyperpolarized strongly
throughout the N2/rasp phase of the feeding cycle (e.g.,
Figs. 2 and 3), and the N2v cells are strong candidates for
providing this inhibition. Evidence for this is shown in Fig.
11, where evoked bursts of spikes in the N2v strongly hyperpolarized the SO both in normal (Fig. 11A) and HiDi saline
(Fig. 11B). The latter suggests a monosynaptic connection.
FIG . 9. N2v r N1L, N1M inhibitory synaptic connections. A: spontaneous and evoked ( r ) bursts of spikes hyperpolarized the N1L cell, suggesting a N2v r N1L inhibitory synapse. Excitation of the B3 motoneuron
occurred at the same time. Absence of spikes in the N2v, during spontaneous
plateauing event, still left a residual hyperpolarization on the N1L (i) but
completely removed the synaptic input to the B3 cell. Occurrence of N2v r
N1L synaptic response in the HiDi saline suggests that it is a monosynaptic
connection. Bi: N2v cell only generates a weak inhibition of the N1M cell.
This is more clearly seen as a reversed IPSP (i) when the N1M was in a
more hyperpolarized state.
rized in Fig. 14), and experiments like that shown in Fig.
10 showed that spontaneous N2v bursts were accompanied
by IPSP on the N3t recorded at the same time, suggesting
that a N2v r B3 synaptic connection exists. Although the
N3t was not patterning in this preparation, each spike was
accompanied by 1:1 EPSPs on the B3 motoneurons. This is
the only CPG interneuron that produces these unitary EPSPs
on the B3 cell (Elliott and Benjamin 1985a). Evoked activity
in the N2v also generated a hyperpolarizing response on the
N3t providing more direct evidence for a synaptic connection. Whether the N2v r N3t connection is monosynaptic
has yet to be determined.
It recently has been shown that the N3p (phasics) receive
only a brief N2 inhibitory input and thus fire earlier in the
feeding cycle than the N3ts at the end of N2 and beginning
of N3 phase of the feeding cycle (Yeoman et al. 1995) (Fig.
14). It thus seems unlikely that they would be inhibited by
N2v cells as these fire throughout the N2 phase of the feeding
cycle. This was confirmed by direct recording in two experiments where no N2v r N3p connection could be found (not
shown). The other type of N2 cell, the N2d, fire more briefly
and are more likely to provide the inhibitory input to the
N3p cells (Elliott and Benjamin 1985a).
It appeared that the two types of N2 CPG interneurons
have differential effects on the N3t and N3p cells and can
thus account for their firing pattern in the SO-driven rhythm
(summarized in Fig. 14). The N3ps receive brief inhibition
from the N2d and recover and fire at the end of N2/rasp,
/ 9k22$$de26 J862-6
The CGCs fired in a slow pattern in both intact (Yeoman
et al. 1994a) and isolated ganglia (McCrohan and Benjamin
1980) but their activity was modulated weakly by synaptic
input to make them fire mainly in the protraction phase of
the feeding cycle rather than during the retraction phase.
McCrohan and Benjamin (1980) showed that they appeared
to be excited during the N1 protraction phase and weakly
inhibited during N2 rasp phase to account for this pattern
of activity. Figure 12 showed that the inhibition occurred at
the same time as spontaneous and evoked bursts of N2v
spikes, suggesting that these cells were responsible for the
CGC inhibitory input during the rasp phase of the feeding
cycle. There was no statistical difference (P õ 0.05, Student’s t-test) in the numbers of spontaneous [15.9 { 3.1
(mean { SE)] spikes in N2v burst compared with current
driven (14.4 { 2.1; n Å 14). The N2v r CGC connection
may be monosynaptic as the experiment was carried out in
HiDi saline. Because the axonal and neuritic processes of
the N2vs are restricted to the buccal ganglia and postbuccal
nerve (Brierley et al. 1997a), the CGC r N2v synapse must
be located in the same structures.
DISCUSSION
N2v cells are part of the Lymnaea feeding CPG
A variety of evidence strongly supported the hypothesis
that the N2v cells are important retraction phase interneurons
of the Lymnaea feeding CPG.
FIG . 10. N2v r N3t inhibitory synaptic connection. Spontaneous or
current-evoked (electrode unbalanced) bursts of N2 spikes cause inhibitory
responses on the N3t cell. N3t generates 1:1 EPSPs on the B3 motoneuron
confirming its identity.
11-26-97 13:07:34
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 18, 2017
N2 r CGCs
N2V CELLS ARE PART OF THE FEEDING CPG
3403
N2v input synaptic connections
These mainly occurred in the N1/protraction phase of the
feeding cycle and trigger the N2v plateau. The effects of
CGC excitation also contributes to excitation during protraction, but the long-term effects of CGC firing may occur
across the whole feeding cycle (Yeoman et al. 1996). Once
the plateau is triggered, no further significant synaptic inputs
appear to be involved in terminating the N2v plateaus. Any
inhibitory input might have been expected to come from the
N3 CPG interneurons. However, direct recordings showed
FIG . 11. N2v r SO inhibitory synaptic connection. A: a burst of N2v
spikes inhibited the SO. B: this persisted in HiDi saline, suggesting a
monosynaptic connection. Reduction in amplitude of hyperpolarizing wave
in B compared with A is presumably due to the SO being in a more
hyperpolarized state.
FIG . 12. N2v r CGC inhibitory synaptic connection. Both currentevoked and spontaneous N2v bursts generate weak hyperpolarizing responses on the CGC. This occurs in HiDi saline, suggesting a monosynaptic
connection.
/ 9k22$$de26 J862-6
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 18, 2017
1)Their activity is phase locked to the rest of the CPG
network in a SO-driven fictive feeding rhythm (Fig. 2).
2)The phase of the CPG rhythm was reset by artificially
inducing extra bursts of N2v spikes, a characteristic of all
CPG neurons (Fig. 2).
3)N2vs have inhibitory and excitatory synaptic connections with retraction phase motoneurons, and these are necessary (with the N2ds) to produce the rasp-swallow phases of
the motor neuron firing pattern (Brierley et al. 1997a).
4)N2vs have extensive input and output synaptic connections with the rest of the CPG network (summarized in Fig.
13A) and provide the main component of N2/rasp phase
inhibitory input to several of these (N1L, Fig. 9; N3t, Fig.
10) as well as driving the N2d cells via a strong electrotonic
synapse (Brierley et al. 1997a).
5)N2vs have reciprocal synaptic connections with the
modulatory SO (Figs. 4 and 11) and CGC (Fig. 12) (Yeoman et al. 1996) interneurons. These are important in the
cycle-by-cycle control of the feeding oscillator, but also
long-term modulatory effects on the N2vs are present (Yeoman et al. 1996).
We believe that the N2vs are the most significant type of
retraction phase interneuron cell so far discovered because
they fire throughout the N2/rasp of the feeding cycle (Brierley et al. 1997a), unlike the N2ds, which only fire briefly
at the beginning of this phase (Elliott and Benjamin 1985a).
Their ability to generate plateau potentials by an endogenous
mechanism (Brierley et al. 1997a) is also likely to be significant in providing oscillatory ‘‘drive’’ to the whole CPG
rhythm.
FIG . 13. Summary of N2v synaptic connections with central pattern
generator CPG and modulatory neurons of the Lymnaea feeding network.
A: apart from the electrotonic coupling of the N2v and the N2d cells, all
the output connections of the N2v cells are inhibitory ( ● ). In contrast, the
input connections to the N2v are a mixture of excitatory ( j ) and inhibitory/
excitatory (closed cycle-bar) biphasic connections. B: interpretation of synaptic potentials triggering plateau potentials during late N3 are due to the
inhibitory (i) component of the SO r N2v biphasic response. During the
protraction/N1 phase of the feeding cycle, depolarization comes from the
facilitatory depolarizing (e) component of the SO r N2v connection as
well as the N1M r N2v excitatory synaptic potential. CGCs provide an
additional weak excitatory component to the N2vs as recently described by
Yeoman et al. (1996). Inhibition comes from the N1Ls and the SO. Plateau
is due to an endogenous mechanism, and this produces sustained firing of
the N2v cell throughout the rasp/N2 phase of the feeding cycle. Falling
phase is also probably due to an endogenous mechanism, but several weak
synaptic potentials can be seen in this falling phase and could be contributory. These do not appear to be due to activity in either the N3t and N3p
neurons.
11-26-97 13:07:34
neupa
LP-Neurophys
3404
M. J. BRIERLEY, M. S. YEOMAN, AND P. R. BENJAMIN
/ 9k22$$de26 J862-6
Again this reduced the duration of the N1/protraction phase
of the feeding cycle (Yeoman et al. 1996).
The final type of input synaptic connection to be considered is the reciprocal electrotonic connection between the
N2vs and the N2ds (Fig. 13A) (Brierley et al. 1997a). This
mainly was considered to play a role in the N2vs driving
the N2ds (see further) but during the brief firing of the N2ds
there could be a minor N2d r N2v excitatory effect.
Modulatory effects of ACh
ACh was applied focally to the surface of the N2v to see
if it could mimic the biphasic (i–e) synaptic effect of SO
neuronal stimulation. Previously published data (Yeoman et
al. 1993) showed that ACh was the likely transmitter of the
SO. This was successful in that an initial hyperpolarizing
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 18, 2017
that no feedback connection occurred from the N3t or N3p
cells to the N2v cells.
The SO, N1M, and N1L are the three main types of interneurons that provided the N1/protraction phase synaptic input to the N2v (Fig. 13B). The final waveform of the potential changes were highly complex and not only determined
by the number of presynaptic cells involved but also by
activity-dependent changes in the amplitude of particular
synaptic responses. The SO is active during the late N3
phase, and this provided a purely inhibitory synaptic input.
However, during N1/protraction, the delayed depolarizing
component of the SO r N2v biphasic PSP (Fig. 13A) became progressively larger and thus increasingly important in
determining the overall depolarizing ramp-shaped waveform
occurring during this phase of the cycle. However, the IPSP
component of the SO r N2v response still persisted, and
the large IPSPs obvious during N1/protraction were also
due to the SO. The other main depolarizing input to the N2v
was due to the N1M r N2v excitatory synaptic connection
(Fig. 13A). The N1Ms accelerate their spike activity toward
the end of protraction, and their excitatory effects must be
particularly important in the steeper part of ramp. The N1Ms
form an electrotonically coupled subnetwork ( Ç4 cells on
each side) (Rose and Benjamin 1981a) of N1M and were
capable of driving N2v plateau alone (Fig. 5Bi). However,
the N1M rhythm was weak and irregular, and a fast fictive
feeding rhythm required the SO (depolarizing) input to the
N2v cells. The final cell type providing input to the N2v
cells are the N1L cells (Fig. 13A). These cells fired four or
five spikes during the N1/protraction phase of the cycle, and
we were surprised to find that these cells inhibited the N2v.
This was the opposite to the N1Ms. However, the inhibitory
effect of the N1Ls became weaker with successive spikes,
and so we presume it was less important in the steeper parts
of ramp just before plateau triggering.
In summary (Fig. 13B), the shape of ramp-like N2v prepotential occurring during N1/protraction initially was determined by SO/N1L inhibitory inputs plus some excitatory
component from the SO and the N1Ms. The latter became
dominant toward the end of the ramp, finally leading to the
triggering of the N2v plateau.
This study emphasized the complexity of the synaptic
control of the protraction phase of the feeding cycle. Changing the slope of the N1/protraction phase ramp of the N2v
cells would determine the duration of protraction phase of
the CPG network and the whole feeding cycle. The steeper
the slope the more rapid the N2v plateau would be triggered,
resulting in a reversal of the cycle from protraction to retraction (see the overall CPG pattern in Fig. 14). Previous authors also have emphasized that changes in the duration of
the protraction phase was a major target for control of frequency of the fictive feeding rhythm (e.g., Elliott and Benjamin 1985a). The SO r N1M excitatory synaptic connection
was considered to be the main site of control, but recently
it was found that the CGCs also could control the frequency
of the feeding oscillator by influencing the duration of the
N1/protraction phase of the cycle. The CGCs excited both
the SO and N1M cells (Yeoman et al. 1996) and so added
to the excitatory drive on these cells that normally would
come from food in the intact animal (Kemenes et al. 1986).
FIG . 14. Summary of firing patterns and synaptic connectivity of all 6
types of CPG interneurons in a SO-driven rhythm. SO was depolarized for
the duration of the traces (horizontal bar). Four cycles of feeding interneurons is shown with vertical bars (solid lines) dividing feeding cycles and
vertical bars (dashed lines) separating out the protraction (P), rasp (R),
and swallow (S) phases within each cycle. Synaptic connections, most of
which are probably monosynaptic, are indicated (arrows) with e indicating
an excitatory postsynaptic potential and i indicating an IPSP. Synaptic
connection of the SO and N2v (black boxes) are shown in the 1st cycle
from the left, the N2d in the 2nd cycle, the N1L in the 3rd cycle, and the
N1M, N3p, and N3t in the 4th. See text for the detailed description of the
figure.
11-26-97 13:07:34
neupa
LP-Neurophys
N2V CELLS ARE PART OF THE FEEDING CPG
N2v output synaptic connections
N2v cells had significant output synaptic connections with
other CPG interneurons and the SO and CGC modulatory
cells. These were all inhibitory (Fig. 13A) apart from the
N2d where electrotonic coupling, N2v r N2d, provided a
strong excitatory drive (Brierley et al. 1997a).
The N2vs provided strong feedback inhibition to the
N1Ls, SO, and CGC cells, preventing them firing during the
N2/retraction phase of the feeding cycle (Fig. 14). N2vs
only had weak inhibitory feedback connections to the N1M
cells, so the strong hyperpolarizing synaptic input that terminates N1M firing must be provided by the N2ds or some
unknown type of N2 cell. The only other important connection the N2v made was with the N3t cells. The N3t were
strongly inhibited by the N2v throughout the N2/rasp phase
of the feeding cycle (Fig. 14) and recovered and fired by
postinhibitory rebound (PIR) (detailed evidence in Elliott
and Benjamin 1985a). There was no connection between
the N2v cells and the N3p CPG interneurons. These are
inhibited briefly during the N2/rasp phase of the feeding
cycle and so presumably must receive inhibitory input from
the N2ds alone.
The discovery of the differential inhibitory effects of the
N2 cells, N2v r N3t N2d r N3p, was important because it
/ 9k22$$de26 J862-6
allowed the N3ps to recover and fire earlier in the cycle than
the N3t cells (Fig. 14). This is crucial in the feedback role
of the N3 cells on the N1M/N1L and SO cells. The N3p
inhibited these three cell types immediately after the end
of N2 inhibition (Fig. 14), and this was continued by the
overlapping N3t induced inhibition until the start of the next
feeding cycle. Thus three phases of sequential feedback inhibition, N2v r N3p r N3t, prevented activity in the protraction phase CPG interneurons N1M/N1L and SO during the
rasp and swallow phases of the feeding cycle.
We conclude that the output synaptic connections of the
N2v are essential for producing the pattern of activity in the
rest of the Lymnaea interneuronal network.
A new model for the Lymnaea feeding CPG
A full summary of the Lymnaea CPG is given in Fig. 14.
It includes the firing patterns and synaptic connections of
the SO and six CPG interneuronal types. It assumes that the
SO is driving the rhythm although recent evidence showed
that the natural food stimulus in the semi-intact preparation
produced a similar fictive feeding rhythm (Yeoman et al.
1995). Direct multicell recordings have been made from all
these cell types and evidence provided in most cases for
monosynaptic connections (Elliott and Benjamin 1985 a,b;
Rose and Benjamin 1981a,b; Yeoman et al. 1995). Further
data are known for the CV1s (McCrohan and Kyriakides
1989) and the CGCs (Yeoman et al. 1996), but these are
not included in Fig. 14 for simplicity. It is now known that
full activation of the feeding rhythm by food (sucrose) requires coactivation of the SO, CGCs, and CPG interneurons
(Yeoman et al. 1995). In the SO-driven rhythm of Fig. 14,
the gating input from CGCs is assumed to be present. In the
isolated preparation, they fired spontaneously at tonic levels
sufficient to allow the SO to drive the pattern. In the intact
animal, food is required to activate the CGCs, SO, and CPG
interneurons (Kemenes et al. 1986; Yeoman et al. 1995).
The transmitters for some of these neurons are known.
This is acetylcholine for the SO, N1M, and N1L cells (Elliott
and Kemenes 1992; Vehovszky and Elliott 1995; Yeoman
et al. 1993) and provisionally glutamate for the N2vs (Brierley et al. 1997b). Neuropeptides also are known to be
colocalized in some of the cells but their function is unknown. The CGCs and SO contain peptides related to myomodulin and the N2vs myomodulin and buccalin-like peptides. Buccalin also may be present in one of the N1M cells
(Santama et al. 1994).
The sequence of activity summarized in Fig. 14 begins
with the SO. This drives the N1L and N1M by electrotonic
coupling (N1Ls) or a strongly facilitating excitatory (e)
synaptic connection (N1Ms). A strong chemical synaptic
connection N1L r N1M also helps to excite the N1Ms during protraction. The activity in the SO is reinforced by reciprocal N1L r SO electrotonic connection and by a weaker
N1M r SO excitatory chemical synaptic connection. All
three protraction phase interneurons provide synaptic inputs
to the N2v cells during N1/protraction. The N2vs initially
are inhibited (i) by the SO but the SO r N2v biphasic (i–e)
synaptic response facilitates so that the delayed e component
gradually becomes more important toward the end of N1/
protraction. A second component of excitation comes from
11-26-97 13:07:34
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 18, 2017
followed by a longer lasting depolarizing response was seen.
However a short (50 ms) application of ACh, in a preparation continuously perfused with normal saline, produced a
long-lasting depolarization lasting for °1 min (10 02 M
ACh). Superimposed on the depolarizing waveform were
sequences of plateau potentials. This suggested that ACh, as
well as being involved in short-term, cycle-by-cycle synaptic
transmission, could cause long-term potentiation of plateauing in the N2v cells. This may well be involved in maintaining the feeding pattern and could account for the observation (Rose and Benjamin 1981a) that the CPG continued to
oscillate for a number of cycles after the end of SO activation. For comparison, a similar experiment was carried out
on the other type of N2 cell type, N2d, and again long-term
activation was seen (Fig. 7D). It should be noted that plateau in the N2vs can be generated by current injection in
quiescent preparations where the SO was inactive, showing
that ACh was not necessary for the basic plateauing. In
contrast, current injection in quiescent preparations cannot
induce plateaus in the N2d cells (Brierley et al. 1997a), and
so ACh may play an important role in inducing long-term
plateauing in this cell type. This would modify the conclusion in Brierley et al. (1997a) that the plateaus in the N2ds
were entirely due to electrotonic potentials arising from the
N2vs. We cannot completely rule out the possibility that
focal application of ACh to the N2d cells also may have
activated adjacent cells (N1Ms or N1Ls) occurring on the
same surface of the buccal ganglia, so more detailed evidence is required. This is unlikely to be the case for ACh
effects on the N2v cells as these are on the opposite ventral
surface away from the main CPG population. The ability of
modulatory substances to initiate plateauing in CPG cells
was seen in the other preparations (Ramirez and Pearson
1991; Sigvardt et al. 1985) and appears to be a common
feature of oscillatory systems.
3405
3406
M. J. BRIERLEY, M. S. YEOMAN, AND P. R. BENJAMIN
Feeding CPGs of other mollusks
The feeding systems of other mollusks have similar multiple types of interneurons forming part of the CPG. The
system from the related snail Planorbis (Arshavsky et al.
1988a–c) has cells of the N1 and N2 subtypes with similar
synaptic connectivity to Lymnaea but apparently no N3s.
Cells equivalent to N1–N3 may occur in Helisoma (Quinlan
and Murphy 1991), but how closely they resemble the Lymnaea cells is not yet clear except that the B2 CPG interneuron
(Quinlan et al. 1995), being glutamatergic, may be similar
to N2d cells of Lymnaea (Brierley et al. 1997a). The B31
cells of Aplysia show plateauing properties similar to the
N2v cells of Lymnaea (Susswein and Byrne 1988), but
whether they play a similar role needs to be further investigated, given that Aplysia has a more flexible feeding system
than Lymnaea (Benjamin 1983).
We thank A. Bacon for help in typing the manuscript, G. Kemenes for
reading the manuscript, and S. Dunn for producing Fig. 4.
This work was supported by the Biotechnology and Biological Sciences
Research Council of the United Kingdom.
Present address of M. J. Brierley: Dept. of Life Sciences, University of
Nottingham, University Park, Nottingham NG7 2RD, UK.
/ 9k22$$de26 J862-6
Address for reprint requests: P. R. Benjamin, Sussex Centre for Neuroscience, School of Biological Sciences, University of Sussex, Falmer, Brighton,
East Sussex BN1 9QG, UK.
Received 20 October 1996; accepted in final form 8 August 1997.
REFERENCES
ARSHAVSKY, Y. I., DELIAGINA, T. G., MEIZEROV, E. S., ORLOVSKY, G. N.,
AND PANCHIN, Y. V. Control of feeding movements in the freshwater
snail Planorbis corneus. I. Rhythmical neurons of buccal ganglia. Exp.
Brain Res. 70: 310–322, 1988a.
ARSHAVSKY, Y. I., DELIAGINA, T. G., ORLOVSKY, G. N., AND PANCHIN,
Y. V. Control of feeding movements in the freshwater snail Planorbis
corneus. II. Activity of isolated neurons of buccal ganglia. Exp. Brain
Res. 70: 323–331, 1988b.
ARSHAVSKY, Y. I., DELIAGINA, T. G., ORLOVSKY, G. N., AND PANCHIN,
Y. V. Control of feeding movements in the freshwater snail Planorbis
corneus. III. Organisation of the feeding rhythm generator. Exp. Brain
Res. 70: 332–341, 1988c.
BENJAMIN, P. R. Gastropod feeding: behavioural and neural analysis of a
complex multicomponent system. In: Symposia of the Society for Experimental Biology: Neural Origins of Rhythmic Movements, edited by A.
Roberts and B. Roberts. Cambridge, UK: Cambridge, 1983, p. 159–193.
BENJAMIN, P. R. AND ELLIOTT, C.J.H. Snail feeding oscillator: the central
pattern generator and its control by modulatory interneurons. In: Neuronal
and Cellular Oscillators, edited by J. W. Jacklet. New York: Dekker,
1989, p. 173–214.
BENJAMIN, P. R. AND WINLOW, W. The distribution of three wide acting
synaptic inputs to identified neurons in the isolated brain of Lymnaea
stagnalis (L). Comp. Biochem. Physiol. A Physiol. 70: 293–307, 1981.
BRIERLEY, M. J., STARAS, K., AND BENJAMIN, P. R. Behavioral function of
glutamatergic interneurons in the feeding system of Lymnaea: plateauing
properties and synaptic connections with motor neurons. J. Neurophysiol.
78: 3386–3395, 1997a.
BRIERLEY, M. J., YEOMAN, M. S., AND BENJAMIN, P. R. Glutamate is the
transmitter for the N2v retraction phase interneurons of Lymnaea feeding
system. J. Neurophysiol. 78: 3408–3414, 1997b.
ELLIOTT, C.J.H. AND BENJAMIN, P. R. Interactions of pattern generating
interneurons controlling feeding in Lymnaea stagnalis. J. Neurophysiol.
54: 1396–1411, 1985a.
ELLIOTT, C.J.H. AND BENJAMIN, P. R. Interaction of the slow oscillator
interneuron with feeding pattern generating interneurons in Lymnaea
stagnalis. J. Neurophysiol. 54: 1412–1421, 1985b.
ELLIOTT, C.J.H. AND BENJAMIN, P. R. Esophageal mechanoreceptors in the
feeding system of the pond snail, Lymnaea stagnalis. J. Neurophysiol.
61: 727–736, 1989.
ELLIOTT, C.J.H. AND KEMENES, G. Cholinergic interneurons in the feeding
system of the freshwater pond snail, Lymnaea stagnalis. II. N1 interneurons make cholinergic synapses with feeding motoneurons. Philos. Trans.
R. Soc. Lond. B. Biol. Sci. 336: 167–180, 1992.
KEMENES, G., ELLIOTT, C.J.H., AND BENJAMIN, P. R. Chemical and tactile
inputs to the Lymnaea feeding system: effects on behaviour and neural
circuits. J. Exp. Biol. 122: 113–138, 1986.
MCCROHAN, C. R. AND BENJAMIN, P. R. Patterns of activity and axonal
projections of the cerebral giant cells of the cerebral giant cells of the
snail Lymnaea stagnalis. J. Exp. Biol. 85: 149–168, 1980.
MCCROHAN, C. R. AND KYRIAKEDES, M. Cerebral interneurons controlling
feeding motor output in the snail Lymnaea stagnalis. J. Exp. Biol. 147:
361–374, 1989.
QUINLAN, E. M., GREGORY, E., AND MURPHY, A. D. An identified glutamatergic interneuron patterns feeding motor activity via both excitation
and inhibition. J. Neurophysiol. 73: 945–956, 1995.
QUINLAN, E. M. AND MURPHY, A. D. Glutamate as a putative transmitter
in the buccal central pattern generator of Helisoma trivolvis. J. Neurophysiol. 66: 1264–1271, 1991.
RAMIREZ, J.-M. AND PEARSON, K. G. Octopamine induces bursting and
plateau potentials in insect neurons. Brain Res. 549: 332–337, 1991.
ROSE, R. M. AND BENJAMIN, P. R. The relationship of the central motor
pattern of the feeding cycle of Lymnaea stagnalis. J. Exp. Biol. 80: 137–
163, 1979.
ROSE, R. M. AND BENJAMIN, P. R. Interneuronal control of feeding in Lymnaea stagnalis. I. Initiation of feeding by a single buccal interneuron. J.
Exp. Biol. 92: 187–201, 1981a.
11-26-97 13:07:34
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 18, 2017
the N1M r N2v. The N1Ls inhibit the N2v. This is likely
to be important at the beginning of N1/protraction, but as
the N1L r N2v inhibitory response progressively weakens,
it is not significant by the end of the protraction phase. The
N2d cells have similar inputs from the SO and N1M cells,
but unlike the N2vs, the N1L input is excitatory, adding to
the input from the other two protraction phase interneurons.
At the end of the N1/protraction phase, a plateau is triggered
in the N2v and this simultaneously activates the N2d via the
strong N2v r N2d electrotonic synapse. ACh released from
the SO and N1M also may contribute to the N2d plateauing.
The N2v fires throughout the N2/rasp phase of the cycle,
whereas the N2d only fires briefly at the beginning of N2.
Both cells provide inhibitory feedback to the N1L/N1Ms.
This is strong for the N2v r N1L connection but weak for
N2v r N1M. The N2d cells inhibit both the N1Ls and N1Ms
equally but only can be important at the beginning of the
N2/rasp phase of the cycle, as they only fire one or two
spikes. This suggests that another type of N2 may exist to
produce the prolonged inhibition of the N1Ms during the
N2/rasp phase of the cycle. Alternatively the N3ps may be
most important. They fire, earlier than previously was
thought (Yeoman et al. 1995), toward the end of N2 phase
of the feeding cycle. The N3t and N3p cells are inhibited
by both the N1M and N1L cells during the N1/protraction
phase of the cycle, and the inhibition is maintained by a
second phase of inhibition from the N2 cells during N2/
rasp. This is differential, with the N3t being inhibited solely
by the N2v cells and the N3p by the N2d cells. Both cells
recover from this inhibition to fire by PIR, N3ps earlier than
N3ts. Both types of N3 cells provide inhibitory feedback to
the N1 cells (both N1Ms and N1Ls) and weaker input to
the SO. Unitary IPSPs can be seen first, 1:1 with N3p spikes
followed by a smoother hyperpolarizing input from the N3t
cells. This inhibition is superimposed on a basically depolarizing waveform driven by the SO, leading to the protraction
phase of the next cycle of activity.
N2V CELLS ARE PART OF THE FEEDING CPG
/ 9k22$$de26 J862-6
buccal musculature by a serotonergic neuron (metacerebral cell) in
Aplysia. J. Neurophysiol. 41: 181–203, 1978.
YEOMAN, M. S., BRIERLEY, M. J., AND BENJAMIN, P. R. Central pattern
generator interneurons are targets for the modulatory serotonergic cerebral giant cells in the feeding system of. Lymnaea. J. Neurophysiol. 75:
11–25, 1996.
YEOMAN, M. S., KEMENES, G., BENJAMIN, P. R., AND ELLIOTT, C.J.H. Modulatory role for the serotonergic cerebral giant cells in the feeding system
of the snail, Lymnaea. II. Photoinactivation. J. Neurophysiol. 72: 1372–
1382, 1994a.
YEOMAN, M. S., PARISH, D. C., AND BENJAMIN, P. R. A cholinergic modulatory interneuron in the feeding system of the pond snail. Lymnaea. J.
Neurophysiol. 70: 37–50, 1993.
YEOMAN, M. S., PIENEMAN, A. W., FERGUSON, G. P., TER MAAT, A., AND
BENJAMIN, P. R. Modulatory role for the serotonergic cerebral giant cells
in the feeding system of the snail Lymnaea. I. Fine wire recording in
the intact animal and pharmacology. J. Neurophysiol. 72: 1357–1371,
1994b.
YEOMAN, M. S., VEHOVSZKY, Á., KEMENES, G., ELLIOTT, C.J.H., AND BENJAMIN, P. R. Novel interneuron having hybrid modulatory central pattern
generator properties in the feeding system of the snail Lymnaea stagnalis.
J. Neurophysiol. 73: 112–124, 1995.
11-26-97 13:07:34
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 18, 2017
ROSE, R. M. AND BENJAMIN, P. R. Interneuronal control of feeding in Lymnaea stagnalis. II. The interneuronal mechanism generating feeding cycles. J. Exp. Biol. 92: 203–228, 1981b.
SANTAMA, N., BRIERLEY, M. J., BURKE, J. F., AND BENJAMIN, P. R. Neural
network controlling feeding in Lymnaea stagnalis: immunocytochemical
localization of myomodulin, small cardioactive peptide, buccalin and
FMRFamide-related peptides. J. Comp. Neurol. 342: 352–365, 1994.
SIGVARDT, K. A., GRILLNER, P., WALLÉN, P., AND VAN DONGEN, P.A.M.
Activation of NMDA receptors elicits fictive locomotion and bistable
membrane properties in the lamprey spinal cord. Brain Res. 336: 390–
395, 1985.
SUSSWEIN, A. J. AND BYRNE, J. H. Identification and characterization of
neurons initiating patterned neural activity in the buccal ganglia of.
Aplysia. J. Neurosci. 8: 2049–2061, 1988.
VEHOVSZKY, Á. AND ELLIOTT, C.J.H. The hybrid modulatory/pattern generating N1L interneuron in the buccal feeding system of Lymnaea is cholinergic. Invert. Neurosci. 1: 67–74, 1995.
WALDEN, J. W., SPECKMANN, E. J., AND WITTE, O. Membrane currents
induced by pentylenetelrazol in identified neurons of Helix panatia. Brain
Res. 473: 294–304, 1988.
WEISS, K. R., COHEN, J. L., AND KUPFERMANN, I. Modulatory control of
3407