INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The feeding motor system of Aplysia is a useful experimental preparation for examining how a neural network generates and controls a variety of behaviors. This system gives rise to consummatory feeding behaviors that are effected by the buccal muscles (for review, see Elliott and Susswein 2002). Contractions of these muscles produce a number of different feeding behaviors, such as biting, swallowing, and rejection (Jing and Weiss 2001, 2002; Kupfermann 1974; Morton and Chiel 1993a,b). All of these behaviors consist of two phases of radula movement, first a radula protraction and then a radula retraction. The protraction-retraction sequence is variably coupled with radula opening and closing movements. The activity of the buccal muscles is organized by a central pattern generator (CPG) contained in the buccal ganglia (Hurwitz and Susswein 1996; Hurwitz et al. 1997; Kirk 1989; Susswein and Byrne 1988; Teyke et al. 1993), which drive the motoneurons that innervate the buccal muscles (Church and Lloyd 1994; Church et al. 1993; Cohen et al. 1978; Hurwitz et al. 1994, 1996, 2000; Morton and Chiel 1993a,b).

The mechanisms of motor program generation have been studied predominantly in isolated buccal ganglia (Hurwitz and Susswein 1996; Hurwitz et al. 1997; Kirk 1989; Susswein and Byrne 1988; Teyke et al. 1993). However, motor programs are strongly influenced by connections from the cerebral ganglion, which senses the presence of the food stimuli that initiate feeding and also modulates feeding behaviors (Rosen et al. 1991). In addition, there is good evidence that complex motor programs can be elicited by a small group of cerebral-buccal interneurons (CBIs). CBI-2, the best studied of these neurons, is capable of driving robust feeding motor programs. Although some of the connections that CBI-2 makes in the buccal ganglion have been described (Hurwitz et al. 1999a; Jing and Weiss 2001, 2002; Sanchez and Kirk 2000), little is known about how the CBIs elicit and control feeding motor programs. To examine the nature of the connections made by different CBIs to various key buccal neurons that function in the initiation of buccal motor programs (BMPs), and the role of the CBIs in shaping the form of BMPs, we have investigated the connections of CBI-2, as well as those of CBI-1 and CBI-3, to key protraction phase neurons that are part of the CPG, the previously identified buccal ganglion neurons B31/B32, B34, and B63. Other studies have focused on the role of additional neurons that function in modifying motor programs so that they become more similar to those seen in specific behaviors, such as ingestion or rejection (Jing and Weiss 2001, 2002; Morgan et al. 2002; Nargeot et al. 1997).

We have found that CBI-2 drives motor programs via ipsilateral and contralateral monosynaptic excitatory postsynaptic potentials (EPSPs) to several protraction-phase interneurons that are capable of initiating feeding motor programs. CBI-2 was found to elicit both fast and slow EPSPs in its followers. The present report characterizes in detail the fast EPSPs elicited by CBI-2. A future report (I. Hurwitz, R. A. DiCaprio, and K. R. Weiss, unpublished data) will characterize in detail the slow EPSPs. The fast EPSPs are apparently cholinergic, and they display a prominent facilitation on repeated stimulation of CBI-2. The facilitation is an integral component of the ability of CBI-2 to recruit key protraction phase interneurons. Other CBIs differentially excite alternate combinations of protraction phase interneurons and may therefore act as modulators of the pattern elicited by the CPG. Thus the ability to initiate and determine the form of motor programs in part arises from the patterns of connectivity of the CBIs to different protraction-phase interneurons.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The experimental subjects for this study were Aplysia californica weighing 150-300 g. They were purchased from Marinus, Long Beach, CA, and from the National Resource for Aplysia at the University of Miami. They were maintained at 14-16°C in holding tanks containing aerated, filtered seawater. Before being dissected, animals were immobilized by injection with isotonic MgCl₂ (50% of body weight). In earlier experiments, the buccal and cerebral ganglia were removed with the cerebral-buccal connectives (CBCs) intact with the cerebral ganglia remaining attached via the pleural and pedal connectives to the pleural and pedal ganglia. This allowed us to characterize the synaptic connections from neuron C-PR to neurons CBI-2 and CBI-12, and thereby to identify these neurons. We then characterized the connections of CBI-2 and CBI-12 to the buccal ganglia. The connections from CBI-2 and CBI-12 to buccal ganglia neurons were very different (Sanchez and Kirk 2001). After the different patterns of connectivity were characterized, these patterns were used to identify CBI-2 or CBI-12. For this reason, in later experiment, the pleural and pedal ganglia were not included in the preparation, and stimuli from C-PR were not needed to identify separately CBI-2 and CBI-12. The data on connections from CBI-12 to the buccal ganglia will be presented in detail elsewhere. The cerebral ganglion was pinned to the floor of the recording chamber with the ventral side up, and the buccal ganglion was pinned with the caudal surface up. The sheath overlying the outermost surface of the ganglia was removed using ultrafine scissors.

Electrophysiology

For intracellular recording and stimulation, neurons were impaled with single-barreled microelectrodes that were made of thin-walled glass filled with 1.9 M potassium acetate and 0.1 M potassium chloride. The electrodes were pulled so that their impedances ranged from 10 to 15 M, and following beveling, they had a final resistance of 6-10 M. The activity of up to four neurons was monitored via intracellular recording using conventional electrometers. The extracellular activity in up to two nerves was also monitored. A Grass stimulator (S88) controlled the intracellular stimuli delivered to the neurons.

Recording apparatus and bathing solutions

The preparations were bathed in artificial seawater [ASW, containing (in mM) 460 NaCl, 10 KCl, 11 CaCl₂, 55 MgCl₂, and 5 NaHCO₃] at pH = 7.64. In some experiments, the buccal and cerebral ganglia were placed in a solution containing an increased concentration of divalent cations [HiDi, with a composition of (in mM) 311 NaCl, 9 KCl, 33 CaCl₂, 132 MgCl₂, and 5 NaHCO₃] to reduce polysynaptic activity of coupled neurons and follower neurons. In general, intracellular recordings were obtained from isolated ganglia preparations maintained at room temperature (18-22°C). Most experiments were performed under continuous fluid exchange using a peristaltic pump at a rate of 10% volume/min.

Intracellular recordings were made from three cerebral-buccal interneurons, CBI-1, -2, and -3, as well as from the buccal-cerebral interneurons (BCIs) B63 and B34 (Hurwitz et al. 1997) and the buccal protraction muscle motoneurons B31/B32 and B61/B62. Other neurons that were often recorded include neurons B4/B5, B64, and radula closure motoneuron B8. The neurons were identified by previously established morphological and/or physiological criteria (Hurwitz and Susswein 1996; Hurwitz et al. 1994, 1996, 1999b; Rosen et al. 1991). For example, B4/B5 was identified based on soma position and physiological criteria (Gardner and Kandel 1977), and the B8 neurons were identified by soma position and their large axon spikes in the radula nerve (RN) (Morton and Chiel 1993b). CBI-2 was differentiated from CBI-12 via the following criteria: CBI-12 but not CBI-2 is coupled to CBI-3, CBI-2 and CBI-12 receive different patterns of synaptic activity from neuron CPR, CBI-2 but not CBI-12 (or any of the other CBIs) elicits a profound facilitation in the protraction-phase neurons, and only CBI-2 elicits an EPSP with an amplitude that exceeds 2 mV.

To generate an action potential in a neuron, 20-ms depolarizing current pulses were injected, and the appearance of one-for-one action potentials was monitored. CBI-2 was fired at 7-12 Hz, for a period that did not exceed 3 min unless otherwise noted.

Data analysis

Comparisons between multiple groups were performed using one-way ANOVAs. When only two groups were compared, t-test were used.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Activity of protraction phase interneurons in CBI-2-elicited motor programs

A number of CBIs can drive BMPs. However, CBI-2 is the most effective (Rosen et al. 1991). Although a number of studies have examined the properties of CBI-2 and have contributed to characterizing this neuron (Hurwitz et al. 1999a,b; Morgan et al. 2002; Rosen et al. 1991; Sanchez and Kirk 2000; Teyke et al. 1993), the mechanisms that underlie the ability of CBI-2 to initiate BMPs have not been studied systematically.

Sustained depolarization of CBI-2 initiates rhythmic activity in the buccal ganglia in which protraction and retraction phases alternate. CBI-2 itself fires during the protraction phase of BMPs and is inhibited during the retraction. In the present study, CBI-2 was stimulated with short pulses that each initiated a single spike. This allowed a fine-grained control of the CBI-2 firing frequency. Using this stimulus, continuous stimulation of CBI-2 was found to initiate rhythmic activity similar to that seen with continuous depolarization even though CBI-2 continued to fire during the retraction. (Fig. 1A). Data to be presented elsewhere (Hurwitz and Weiss, unpublished results) will explain the apparent anomaly that firing in CBI-2 during retraction apparently does not affect BMPs. Because the I2 nerve contains the axons of motoneurons that innervate and excite a major protraction muscle (the I2 muscle), the duration of protraction phases can be monitored via extracellular recordings from this nerve (Hurwitz et al. 1994). The intracellularly recorded firing of B61/B62 and B31/B32 (Fig. 1, A and B) coincides with the extracellularly recorded activity in the I2 nerve, and previous data have shown that firing of B61/B62 and B31/B32 causes I2 muscle contractions (Hurwitz et al. 1994, 2000). Extracellular recordings can also be used to monitor the duration of the retraction phase via recordings from both the I2 nerve and the RN. The RN contains the axons of neurons (B8) that drive contractions of radula closer muscle (the I4 muscle). Thus the duration of the retraction phase can be monitored by the extracellularly recorded activity in RN that continued after I2 nerve activity ceases (see for details Morton and Chiel 1993a,b; Morgan et al. 2002). Intracellularly recorded activity in neuron B8 largely coincides with the extracellularly recorded activity in RN (Fig. 1A), and I4 muscle contractions.

View larger version (45K): [in this window] [in a new window]   Fig. 1. Intracellular depolarization causing firing in CBI-2 drives multiple cycles of buccal motor programs (BMPs) with a progressive decrease in the duration of the programs. CBI-2 was stimulated with 25-nA pulses of 20-ms duration at a frequency of 15 Hz. The stimulus generated cycles of the biphasic motor programs that were monitored via intracellular or extracellular recordings. The protraction phase was monitored by intracellular recordings from protraction phase interneurons B34 or B63 and from motoneurons B31 or B61. In addition, the protraction phase could be monitored via extracellular recording of the I2 nerve (I2N) that contains the axons of these motoneurons. The retraction phase was monitored via intracellular recordings from interneuron B64 or via the hyperpolarization that the firing of B64 elicited in the protraction neurons. In addition, the retraction phase could also be monitored via extracellular recording from the radula nerve (RN) or via intracellular recordings of the radula closer motoneuron, B8, which has an axon in the RN. Because B8 may also fire during the protraction phase, only the B8 firing that persists on termination of protraction phase was a monitor of the retraction phase. A: multiple cycles of biphasic biting motor program were generated by sustained firing of CBI-2. Firing of CBI-2 first drives a protraction phase that was monitored by the firing of B61 and B34 with latency of several seconds followed by a retraction phase that was monitored by the firing of B8 and the inhibition of the protractors. Note that there is a progressive decrease in the duration of the successive cycles of BMPs. B: a briefer stimulation of CBI-2 initiates a single cycle of a BMP, in which B31/B32 and B63 are active. C: multiple cycles of biphasic BMP are generated by stimulating CBI-2. In this recording, activity in B61/B62 is a monitor of the protraction phase, and activity in B64 monitors the retraction phase.

Intracellular and extracellular patterns of activity were investigated in the I2 and RN nerves and in several key buccal ganglia neurons during motor programs elicited by CBI-2 stimulation (n > 30). Typical recordings are shown in Fig. 1. Tonic firing of CBI-2 induced rhythmic cycles of BMPs with protraction phase interneuron B34 and the B61/B62 motor neurons firing together during the protraction phase (Fig. 1A). B34 and B61/B62 fired out of phase with the firing of B8. Firing of CBI-2 also drives activity in interneuron B63 and in the multi-functional B31/B32 neurons during the protraction phase (Fig. 1B). As in the isolated buccal ganglia, B31/B32 activity is characterized by a sustained plateau depolarization with small, fast depolarizations of a variety of sizes superimposed on the sustained plateau. The largest of these fast depolarizations represents axon spikes that fail to invade the soma (Hurwitz et al. 1994). In addition, B64 is recruited into CBI-2 elicited BMP during the retraction phase (Fig. 1C). Previous studies have shown that in an isolated buccal ganglia preparation, intracellular stimulation of B31/B32, B34, B63, and B64 can all initiate BMPs (Hurwitz and Susswein 1996; Hurwitz et al. 1994, 1996, 1997; Susswein and Byrne 1988). Thus in principle, the ability of CBI-2 to drive BMPs could arise via its ability to drive one or all of these neurons. However, a closer analysis of BMPs elicited by CBI-2 indicates that these are unlikely to be initiated via B64. In all BMPs elicited by CBI-2, the initial phase was protraction (n > 100) with firing in B64 in BMPs occurring only after the end of a protraction phase (Fig. 1C).

The transition from protraction to retraction was sharply demarcated by a rapid termination of firing in protraction phase neurons B31/B32, B34, B61/B62 and B63 (Fig. 1, A-C). In isolated buccal ganglia, the retraction phase interneuron B64 remained active throughout the retraction phase (Hurwitz and Susswein 1996), and termination of its activity coincides with termination of activity in other retraction phase neurons, such as B4 (Hurwitz and Susswein 1996). Consistent with observations made in isolated buccal ganglia (Hurwitz and Susswein 1996; Hurwitz et al. 1997), the protraction phase neurons B31/B32, B34, B61/B62, and B63 remained hyperpolarized below their resting membrane potential throughout the retraction phase. Thus similar to observations made in isolated buccal ganglia, the hyperpolarizations that occur rhythmically in the protraction phase neurons can serve as markers of the retraction phase. Coincident with the termination of the retraction phase (monitored by the onset of high-frequency firing of B8 or via firing in B64), protraction phase neurons showed a characteristic rapid return to their rest potentials. When CBI-2 was continuously stimulated with pulses, after a short delay, the return to rest was followed by a new cycle of a motor program (see also Church and Lloyd 1994; and Rosen et al. 1991). However, if firing in CBI-2 was terminated, BMPs were terminated (Fig. 1, B and C), indicating that activity in CBI-2 is required to activate repetitive BMPs. Buccal motor programs similar to those illustrated in Fig. 1 were elicited by stimulating CBI-2 in 30 preparations

Key buccal interneurons receive monosynaptic excitatory inputs from CBI-2

The preceding data show that CBI-2 can drive BMPs because it drives important elements of the buccal ganglia CPG. However, the CPG is a complex network of electrically and chemically connected neurons, many of which are able to drive BMPs when stimulated (e.g., B31/B32, B34, B63, and B64) (see Hurwitz and Susswein 1996; Hurwitz et al. 1994, 1997; Susswein and Byrne 1988). To characterize in greater detail the mechanisms by which CBI-2 drives BMPs, its synaptic connections to some of the CPG neurons were characterized. Previous data (Rosen et al. 1991; Sanchez and Kirk 2000) reported that B31/B32, B34, and B61/B62 are recruited by CBI-2, but there is no information on possible connections to B63 or to B64, interneurons that are considered central for generating buccal motor programs (Hurwitz and Susswein 1996; Hurwitz et al. 1997, 1999b). In addition, the connections to the other protraction phase neurons were incompletely characterized in the previous studies.

CBI-2 monosynaptically excited B63 in addition to also exciting B31/B32 and B34, (Figs. 2, 3, n = 4 for the data in Fig. 2, n = 5 for the data in Fig. 3). There were no direct connections from CBI-2 to B64 in either experiments performed in ASW or HiDi (not shown). Feeding is normally generated by activating bilaterally symmetrical motor neurons, and therefore it was not surprising to find that synaptic potentials elicited by CBI-2 were observed in both ipsilateral and contralateral buccal interneurons, although the synaptic potentials were 30-70% larger ipsilaterally than contralaterally (Fig. 2).

View larger version (22K): [in this window] [in a new window]   Fig. 2. CBI-2 elicits facilitating excitatory postsynaptic potentials (EPSPs) in the ipsilateral and the contralateral neurons. A: EPSPs elicited by firing CBI-2 are twice as large in the ipsilateral than in the contralateral B31 neuron. B: similarly, EPSPs elicited by firing CBI-2 are 3 times as large in the ipsilateral than in the contralateral B34 neuron.

View larger version (36K): [in this window] [in a new window]   Fig. 3. Stimulating CBI-2 elicits 1-for-1 facilitating EPSPs in two major protraction phase interneurons. A1: EPSPs recorded in B63. B1: EPSPs recorded in B34. The EPSPs persisted in a high-divalent cation saline (HiDi saline), indicating that the connections are monosynaptic in neuron B63 (A2) and B34 (B2).

CBI-2 elicited complex synaptic potentials in ipsilateral and contralateral buccal ganglion protraction phase neurons. Using neurons B31/B32 and B34 as examples, Fig. 2 illustrates the basic features of the synaptic potentials elicited by CBI-2. CBI-2 elicited both fast and slow excitatory synaptic potentials in B31/B32, B34, and B63. The fast EPSPs persisted in a solution containing a high concentration of divalent cations (HiDi), in interneurons B63 and B34 (Fig. 3, A and B) and in the motoneurons B31/B32 and B61/B62 (not shown), suggesting that the connections are monosynaptic.

In addition to this fast component, synaptic potentials elicited by CBI-2 also displayed a slow component (Fig. 3A), which also persisted in a HiDi saline (Figs. 3B and 4). The properties of the slow EPSPs will be explored in detail in a future communication (see Hurwitz et al. 1999a).

View larger version (16K): [in this window] [in a new window]   Fig. 4. In addition to producing within-train facilitation, stimulating CBI-2 also produced between-train enhancement of the fast EPSPs recorded in B63 and in B34. The recording was in presence of HiDi saline. Note that firing CBI-2 induces EPSPs with both fast and slow components, and the between-train enhancement affects both the fast and the slow components. Note that the lines below the B34 and B63 traces show the resting potential, so that the slow EPSPs can be seen by reference to this potential.

As was previously reported by Sanchez and Kirk (2000), the fast EPSPs showed pronounced facilitation. In some cases, the first few spikes within a train produced no discernible postsynaptic potentials (Figs. 3 and 4), whereas when the fast EPSP was fully facilitated, it displayed values of >5 mV. We have found that the fast EPSPs elicited by CBI-2 in B63 displayed within-train facilitation as well as between-train enhancement, even when the ganglia were bathed in HiDi saline (Fig. 4). There was also a between-train enhancement of the slow component of the EPSPs that CBI-2 elicits in B34 and B63 (Fig. 4).

Contribution of B31/B32, 34, and B63 to CBI-2-elicited motor programs

CBI-2 monosynaptically excites protraction phase interneurons B31/B32, B34, and B63, which are all capable of initiating BMPs when depolarized (Hurwitz and Susswein 1996; Hurwitz et al. 1994, 1997; Susswein and Byrne 1988), suggesting that the ability of CBI-2 to initiate BMPs is via its direct connections to these neurons. CBI-2 may initiate a BMP via effects on a specific neuron or via effects on all of the protractor-phase interneurons. To distinguish between these possibilities, we examined systematically the effects of CBI-2 firing on the different protraction phase interneurons. CBI-2 also excites the B61/B62 protraction motoneurons, but previous work demonstrated that B61/B62 are not involved in BMP generation (Hurwitz et al. 1994).

The synaptic potentials that CBI-2 elicited in B34 and in B61/B62 were generally twice as large as those in B63, and 10 times larger than the ones recorded in B31/B32. For example, when the amplitude of the facilitated EPSPs reached 10 mV in B34 and 8 mV in B61, it was only 5 mV in B63 and as low as 1 mV in B31 (Figs. 2 and 3; note differences in voltage calibrations). Furthermore, B31/B32 have a high threshold for triggering their typical activity pattern during a BMP, a plateau-like potential in the somata and spiking in the axon (Hurwitz et al. 1994; Susswein and Byrne 1988). Thus the direct excitation that B31/B32 receives from the firing of CBI-2 is not likely to depolarize B31/B32 sufficiently to initiate motor programs. Because B31/B32 receive strong excitatory inputs from B34 and B63 (Hurwitz et al. 1997), we hypothesized that CBI-2 may access the buccal CPG by first activating neurons B34 and B63 and then via a feed forward excitation of neurons B31/B32. To characterize the sequence of activation of B31/B32, B34 and B63 during motor programs elicited by CBI-2, we recorded simultaneously (n = 3) from these followers of CBI-2 while motor programs were elicited by CBI-2 stimulation (Fig. 5). Firing CBI-2 rapidly depolarized neurons B34 and B63, which on reaching threshold began to fire. By contrast, almost no depolarization of B31/B32 was observed until after B34 and B63 were firing. Thus the evidence supports the idea that the strong inputs from B34 and B63 are major contributors to the recruitment of B31/B32 into motor programs.

View larger version (21K): [in this window] [in a new window]   Fig. 5. Firing in CBI-2 at a frequency of 12 Hz initiates a BMP that begins within seconds. Spikes in B63 and B34 began ~5 s after activity in CBI-2, whereas spikes in B31 (arrow), which are restricted to the axon, and do not invade the soma, began much later, after ~10 s. Note the onset of activity in the I2 nerve that represent firing in motoneurons B61/B62. Firing later in the I2 nerve represents both B31/B32 and B61/B62 activity.

To determine whether CBI-2 induces BMPs predominantly via either its excitation of B34 or B63, these neurons were selectively hyperpolarized during CBI-2 firing, thereby preventing them from firing (n = 3 for each neuron). As was seen previously in Fig. 5, when the B34 or B63 neurons were not hyperpolarized, firing in CBI-2 first depolarized neurons B34 and B63, and B31/B32 became depolarized only after they began to fire (Fig. 6A). Hyperpolarizing a single B63 neuron was found to obstruct the expression of a motor program as monitored by activity in B31/B32 (Fig. 6B). Hyperpolarizing B63 also reduced spike activity in the radula nerve (Fig. 6B). These findings extend a previous finding (Hurwitz et al. 1999b) that showed that when BMPs are elicited by tonic stimulation of CBI-2, unilateral hyperpolarization of B63 leads to an interruption of the BMPs until the hyperpolarization is released. These experiments also monitored spike activity in the I2 nerve that innervates the I2 muscle, which has a major role in effecting protraction (Hurwitz et al. 1996). Hyperpolarizating B63 did not prevent the typical protraction phase activity in this nerve. In CBI-2-induced BMPs, activity in the I2 nerve is predominantly due to firing of protraction phase motor neuron B61/B62, which independently receives a direct, strong monosynaptic input from CBI-2 (Sanchez and Kirk 2000). This excitation is likely to drive B61/B62 firing even in the absence of inputs from the buccal ganglion CPG.

View larger version (14K): [in this window] [in a new window]   Fig. 6. CBI-2 initiates BMPs via B63. A: firing in CBI-2 drives protraction-phase neurons B31, B63, and B34. These neurons are inhibited at the start of retraction. Note that B8 activity as recorded from the RN persists during both protraction and retraction phases. B: B63 was hyperpolarized (arrows), preventing its firing in response to activity in CBI-2. This prevented CBI-2 from driving a BMP. Specifically, although B34 fired in response to CBI-2 stimulation, B31 did not fire, and the retraction phase, as monitored via RN activity and the onset of a rapid inhibition (presumably from firing B64), did not appear. Note that B31 and B34 did display an inhibition with a slower onset than usually seen when B64 fires, and presumably this is driven by neurons other than B64. C: when B34 was hyperpolarized and prevented from firing (arrows), activity in CBI-2 successfully drives a BMP, indicating that B34 is not necessary for driving a BMP via CBI-2 (compare C with A).

Unlike the unilateral hyperpolarization of B63, unilateral hyperpolarization of B34 (n = 5) did not obstruct the expression of CBI-2-elicited motor programs (Fig. 6C). Bilateral hyperpolarization of B34 (n = 3), however, affected buccal motor programs (Fig. 7). Specifically, the slope of depolarization of B31/B32 in response to CBI-2 stimulation was shallower when the two B34s were hyperpolarized (Fig. 7). Also, the duration during which the B31/B32 axon generated action potentials was shorter when the two B34s were hyperpolarized. Interestingly, the transition from protraction to retraction occurred at the same time relative to the onset of stimulation of CBI-2 stimulation, independent of whether the two B34s were allowed to fire or were hyperpolarized. However, the duration of the retraction phase was shortened when the B34 neurons were hyperpolarized, and in the absence of B34 activity, the retraction phase was always 2 s. These observations are consistent with a previous suggestion that B34 is not a critical element for triggering motor programs. Instead, B34 is likely to act in shaping the form of a BMP (Hurwitz et al. 1997; Jing and Weiss 2002).

View larger version (25K): [in this window] [in a new window]   Fig. 7. Hyperpolarizing both B34 neurons during firing of CBI-2 does not block the ability of CBI-2 to generate a BMP. A: firing CBI-2 drives protraction neuron B31 as well as both B34 neurons during the protraction phase. On initiation of the retraction phase, these neurons were inhibited by retraction-phase neurons. Note that B8 activity as recorded from the RN during the retraction phase. B: when both B34 neurons are hyperpolarized and prevented from firing, activity in CBI-2 still is able to drive a BMP. Specifically, B31 was active, and its acivity was terminated via a fast onset of inhibition. However, hyperpolarizing both B34 neurons decreased the time B31/B32 was maximally depolarized and delayed the onset its activity. Note that during B34 hyperpolarization, the duration of the retraction phase was also reduced (both firing of B8 and the postprotraction hyperpolarization of B31 were shortened). C: releasing B34 neurons from hyperpolarization restored the characteristic CBI-2-elicited BMP. The duration of B8 firing and the duration of the post protraction hyperpolarization of B31 were restored (compare C with B). Note that the B34 neurons were somewhat depolarized in this panel.

The preceding data indicate that in CBI-2-driven programs, the two protraction-phase interneurons apparently have different functions. B63 initiates a plateau depolarization in B31/B32 and, with a delay, initiates the retraction phase. B34 is responsible for producing an earlier onset in the activity of B31/B32 as well as an increase in the duration of the retraction phase.

Facilitation enhances the ability of CBI-2 to recruit protraction neurons

The preceding data, as well as a previous report (Sanchez and Kirk 2000), indicated that the fast EPSPs from CBI-2 to protraction interneurons undergo a prominent facilitation and the facilitated EPSPs summate (see Figs. 2-4). These findings suggested that aspects of the control of BMPs by CBI-2 could depend on the frequency of CBI-2 firing, and the concomitant build-up and the decay of facilitation that may depend on different firing frequencies. We examined this possibility, by exploring systematically the effect of different frequencies of CBI-2 firing on the protraction-phase interneurons.

Facilitation and summation of the EPSPs elicited by CBI-2 affected the ability of CBI-2 to initiate firing in the protraction-phase neurons. When the CBI-2 firing was set to duration of 1.5 s at frequency of 10 Hz, the amplitude of the facilitated EPSPs was below threshold for firing both B34 and B61 (Fig. 8A). Increasing the firing frequency to 15 Hz caused a corresponding large increase in the amplitude of individual EPSPs from 2 to 4 mV (tested on the 5th EPSP) but did not cause firing in either protraction phase interneuron (Fig. 8B). An increase in CBI-2 firing frequency to 20 Hz caused an increase of the EPSP to 6 mV and firing in B61 (Fig. 8C), and a further increase in CBI-2 firing frequency recruited both B61 and B34 (n > 15, Fig. 8D). Thus the ability of CBI-2 to drive protraction-phase neurons is dependent on the CBI-2 firing frequency, and the resultant facilitation of the fast EPSPs.

View larger version (23K): [in this window] [in a new window]   Fig. 8. The ability of CBI-2 to initiate activity in protraction-phase neurons depends on facilitation and summation of the EPSPs that CBI-2 elicits. Firing of CBI-2 for 1.5 s elicited facilitating EPSPs in B34 and B61. A and B: at firing frequencies <20 Hz, CBI-2 did not drive action potentials in B61 or B34. C: at a firing frequency of 20 Hz, CBI-2 generated only 1 action potential in B61. D: by contrast, when CBI-2 fired at 25 Hz, several action potentials in both B61 and B34 were elicited. Note that both B61 and B34 remained depolarized for several seconds after the termination of CBI-2 stimulation.

After the facilitation was initiated, it could be maintained for close to a minute, thereby affecting the ability of successive bursts of activity in CBI-2 to become progressively more effective in initiating activity in protraction phase neurons. The relationship was examined between the facilitation of the EPSPs that CBI-2 elicits in B63 and B34 and the latency of the subsequent initiation of firing in these neurons (n = 3). A conditioning train (1.5 s at 20 Hz) of action potentials was triggered in CBI-2 (Fig. 9, left). After a variable delay (3, 10, 40, and 60 s) a test train of lower frequency was triggered in CBI-2. The size of the synaptic potentials elicited by the test train was inversely related to the duration of the delay between the conditioning and the test train (see Fig. 9, A-D, right, and E, 1 and 2, which illustrate the 1st 3 EPSPs elicited by CBI-2 at higher gain and with an expanded time base). The size of the EPSPs elicited by CBI-2 declined as the delay between the conditioning and test trains increased. Importantly, the latency to initiate spiking also increased as the duration of delay was increased. These data indicate that the size of the facilitation of the EPSPs that CBI-2 elicits in a key protraction phase interneuron contributes to a reduction in the latency for the onset of spikes in its buccal followers and may thus advance the initiation and perhaps accelerate the rate of buccal motor programs.

View larger version (27K): [in this window] [in a new window]   Fig. 9. Facilitation of the EPSPs elicited by firing of CBI-2 can last for tens of seconds. The intervals between 2 trains of spikes elicited in CBI-2 affect both the amplitude of subsequent EPSPs as well as the latency for firing of protraction interneurons B34 and B63. The 1st stimulus train was delivered for 2 s at 15 Hz, and the 2nd train was set to 7 Hz and lasted until the CBI-2 follower neurons began to fire. A: when the interval between the 1st and 2nd train was set to 3 s, the onset of firing in the follower cells occurred at 2.5 s for B34, and at 3 s for B63. B: at a 10-s interval, the B34 latency was 4.2 s and the B63 latency was 4.8 s. C: at a 40-s interval between the 2 trains, the latency for firing in B34 was 4.9 s and the B63 latency was 5.7 s. D: at a 60-s interval, both B34 and B63 latencies were ~6.5 s. Longer intervals lead to a similar latency (not shown). E: a decrease in the EPSPs amplitudes was observed when interval-length was increased. E1: the 3rd EPSPs recorded in B34 were 10, 6, 5, and 4.5 mV in amplitude to intervals of 3, 10, 40, and 60 s, respectively. E2: similarly, the 3rd EPSPs recorded in B63 were 13, 8, 4, and 3 mV in amplitude to intervals of 3, 10, 40, and 60 s, respectively.

The background rate of firing in CBI-2 also contributed to the amplitude of the EPSPs elicited by CBI-2 during a burst. Tonic firing at a relatively high rate maintained the facilitation for longer periods than did lower firing frequencies (n = 4). To explore this effect systematically, CBI-2 was excited with a high-frequency burst under different conditions of tonic background firing (Fig. 10). Decreasing the background firing rate of CBI-2 from nearly 3 (Fig. 10A) to 2 Hz (Fig. 10B) decreased the decay time of facilitation from >8 s to <4 s. Furthermore, interrupting the tonic firing in CBI-2 immediately after the burst for ~8 s decreased the EPSPs amplitude below the base line value (Fig. 10C). These data indicate that following a burst, the firing frequency of CBI-2 is critical for maintaining the facilitation. The data also indicate that under the conditions of this experiment, in which CBI-2 was firing individual spikes, the facilitation decays within several seconds (see Fig. 10), whereas in other conditions, when CBI-2 is firing in bursts, the facilitation lasts for almost a minute (see Fig. 9).

View larger version (34K): [in this window] [in a new window]   Fig. 10. The decay of facilitation elicited by firing of CBI-2 is dependent on the background firing rate of CBI-2. A: CBI-2 was stimulated to produce a background firing frequency of 3 Hz. On this background, a 15-Hz train of firing was elicited, and this elicited a profound facilitation. The EPSP amplitude declined to baseline values during the subsequent 9 s. B: the facilitated EPSPs declined to baseline values within ~4 s when the background firing frequency was decreased to 2 Hz. , the point at which PSPs return to baseline values. C: cessation of stimulation of CBI-2 immediately after the train of spikes led to a decrease in the EPSP amplitude even below the baseline value.

We also examined systematically the effect of triggering CBI-2 with different firing frequencies on the relative length of the protraction and retraction phases of a BMP (Fig. 11). To generate a BMP, CBI-2 firing frequencies were 7 Hz because lower frequencies did not initiate even a single motor pattern within a minute. CBI-2 was stimulated to fire at frequencies of 7, 10, 15, 20, 25, and 30 Hz, until the completion of a single BMP. The experiment was performed with intervals of 1 min between stimulus trains. Data were discarded from the first four BMP cycles elicited because data from these trials were quite variable. Data that were collected for subsequent analysis were from the fourth and later BMP cycles, when the BMPs became highly reproducible and regular. Each of the frequencies was tested three times in each animal in a random order to ensure a stable baseline, and the three tests were averaged. The buccal ganglia were not desheathed to reduce the potential for damage. The protraction phase was monitored via extracellular recording of the I2 nerve (I2N), and the time interval between cessation of I2N activity and the cessation of RN activity was used as a monitor of the duration of the retraction phase.

View larger version (46K): [in this window] [in a new window]   Fig. 11. The characteristics of a BMP depend on the CBI-2 firing frequency. Increasing the firing frequency of CBI-2 results in a decreased duration of a BMP. The duration of the protraction phase was monitored via activity recorded extracellularly in the I2N. The duration of the retraction phase was monitored via activity in the RN that outlasted the protraction phase activity. A1: firing of CBI-2 for the full duration of a BMP at a frequency of 7 Hz generates a protraction phase with a latency of 10 s and a retraction phase that ceased almost 20 s later. A, 1-6: both the latency to onset of a BMP and the duration of the BMP display progressive decreases in duration as the firing frequency is increased. Note that at a firing frequency of 30 Hz (A6) the latency for the start of the protraction phase was less than a second, and the duration of the BMP was <10 s. Note that the figures given in the text for latencies and durations of programs refer to the these data and not to the mean data. B: the means ± SE of 4 parameters of the BMPs at each of the 6 frequencies of stimulation are presented before normalization. C: statistical tests revealed significant differences in the effect of CBI-2 firing frequency on the protraction latency, protraction duration and the BMP duration. To perform the statistics, for each of the preparations data were averaged for each of the 6 tested frequencies (each frequency was tested 3 times) and expressed as a percentage of the data obtained at a 7-Hz stimulation frequency. Data from all 7 preparations were then combined and tested for statistical significance ( = 0.05) by a 1-way ANOVA. These tests revealed: a significant difference for the latency to the start of protraction [C1; F(5,36) =17.8; P < 0.0001]; a significant difference in the duration of the protraction phase [C2; F(5,36) = 4; P < 0.01]. a significant difference in the duration of the BMP [C3; F(5,36) = 3.7, P < 0.01]. Note the particularly large effect of high firing frequencies on the latency for the onset of protraction. *, P < 0.05; **, P < 0.01; ***, P < 0.005 in a Dunnet's post hoc test.

Several parameters of the BMP were sensitive to increasing the firing frequency of CBI-2 (Fig. 11A). Specifically, at a frequency of 7 Hz, the latency to the onset of protraction and the duration of the protraction phase were both relatively long. In the example shown in Fig. 11A, the latency for the onset of the protraction phase was >10 s, and the duration of the protraction phase reached 15 s (Fig. 11A1). As the stimulation frequencies were progressively increased, the latencies for the onset of the protraction phase decreased and reached values of <1 s (Fig. 11A, 2-6). The duration of the protraction phase also decreased to <7 s. The duration of a BMP also decreased (from ~20 to ~10 s), even though the duration of the retraction phase appeared resistant to changes in frequency (see the table in Fig. 11B for the calculated averages). The mean values of four parameters at the six frequencies of stimulation are presented in Fig. 11B.

In view of the variability between different preparations, data for statistical analyses were expressed as percentages of the results obtained in response to a 7-Hz stimulation. Data from all seven preparations were then combined. The latency for the onset of activity in the I2N showed a significant difference [F(5,36) = 17.8; P < 0.0001] among the six stimulus frequencies (Fig. 11C1). In addition, there was a significant difference in the duration of protraction phase [F(5,36) = 4; P < 0.01] among the six stimulus frequencies (Fig. 11C2). There was also a significant difference in the duration of BMP cycles [F(5,36) = 3.7, P < 0.01; Fig. 11C3]. There was no difference in the length of the retraction phase in the different CBI-2 firing frequencies [F(5,36) = 0.65, P > 0.05; all tests are 1-way ANOVAs; data not shown].

Hexamethonium blocks fast synaptic potentials elicited by CBI-2 and delays the onset of firing of its followers

To examine further the properties and functions of the fast synaptic potentials from CBI-2 to the buccal ganglia CPG, we sought to identify an antagonist of a transmitter that elicits these EPSPs. Hexamethonium, a cholinergic antagonist that has been extensively used to block fast cholinergic EPSPs and end junction potentials (EJPs) in Aplysia (Cohen et al. 1978; Gardner and Kandel 1977; Hurwitz et al. 2000), was found to be an effective blocker of the fast EPSPs that CBI-2 elicited in its followers, suggesting that CBI-2 is a cholinergic neuron. Figures 12 and 13 demonstrate the block of the fast EPSPs that CBI-2 elicited in various neurons, including B31/B32, B63, B61/62, and B34. In the example illustrated in Fig. 12, the follower neurons of CBI-2 (B31/B32, B63, and B61/62) were slightly hyperpolarized to preventing spiking of CBI-2 followers and to increase the amplitude of the EPSPs. In this condition, a train of action potentials that was triggered in CBI-2 elicited one for one EPSPs in the CBI-2 followers (Fig. 12A). A complete elimination of the fast EPSPs was observed during bath application of 10³ M hexamethonium (Fig. 12B). The action of hexamethonium was reversible if the washout began shortly after the fast component of the EPSPs was blocked (Fig. 12C). If, however, hexamethonium was left in the bath for >20 min, only a partial recovery occurred even after a 1 h of washout (not shown).

View larger version (19K): [in this window] [in a new window]   Fig. 12. EPSPs evoked in B61, B63, and B31 by firing of CBI-2 are blocked by the cholinergic antagonist hexamethonium. A: firing of CBI-2 for 1.5 s at a frequency of 11 Hz evoked a train of facilitated EPSPs that was observed in B61, B63, and B31. B: hexamethonium (103 M) caused a decrease in the amplitude or completely blocked the EPSPs. C: the effects of hexamethonium were partially reversed after 10 min of washing in artificial seawater (ASW).

View larger version (13K): [in this window] [in a new window]   Fig. 13. Fast EPSPs evoked in B34 by firing of CBI-2 are blocked by the cholinergic antagonst hexamethonium. Note that the recording is in HiDi. A: firing of CBI-2 for 1.5 s at a frequency of 14 Hz evokes a train of facilitated EPSPs in B34. B: the amplitude of the fast EPSPs was decreased in 105 M hexamethonium. C: a higher concentration of hexamethonium (104 M) reduced the summated EPSPs amplitude still further. D: in 103 M hexamethonium, the fast EPSP was completely blocked. E: the effects of hexamethonium were partially reversed after 10 min of washout in HiDi (post Hex).

The EPSPs elicited by CBI-2 in B34 were examined, to test the response to different concentrations of hexamethonium. In this experiment, ASW was replaced with HiDi to prevent polysynaptic activity. This allowed us to record from neuron B34 at -45 mV. Under these conditions, the slow component of the EPSP is readily observed. A train of action potentials triggered in CBI-2 elicited one-for-one EPSPs in B34 (Fig. 13A). Application of 10⁵ M hexamethonium caused a reduction in the amplitude of the fast EPSPs (Fig. 13B) by 25%. Increasing the concentration of hexamethonium to 10⁴ M reduced the EPSPs by 70% (Fig. 13C), and the fast EPSPs were abolished at a concentration of 10³ M hexamethonium (Fig. 13D). The action of hexamethonium was reversible (Fig. 13E).

Figures 12 and 13 also demonstrate that hexamethonium preferentially blocks the fast component of the EPSP while leaving essentially intact the slow component that CBI-2 elicited in its followers.

CBI-2 elicits both fast and slow EPSPs in its followers, and in principle, the ability of CBI-2 to initiate BMPs could depend on either or both types of EPSP. Because hexamethonium preferentially blocks the fast EPSPs, we attempted to use hexamethonium to characterize the contribution of fast synaptic potentials to the initiation of BMPs. In every case (n = 3 for B61/B62 and B63, n = 4 for B31/B32, and n = 7 for B34), a partial block of the fast EPSPs by a brief (<1 min) perfusion with 10⁴ M hexamethonium delayed the onset of activity in the protraction phase interneurons (Fig. 14, A-D). However, in the presence of hexamethonium, the slow build-up of the depolarization elicited by CBI-2 stimulation in its followers was also reduced. In part, this reduction could be explained by the lack of summated EPSPs and in part because the slow EPSP may be partially voltage-dependent and the lack of the fast EPSPs reduces its depolarization-dependent amplification.

View larger version (18K): [in this window] [in a new window]   Fig. 14. The latency for the onset of the protraction phase is sensitive to hexamethonium. A: CBI-2 was fired at a frequency that drives the protraction phase neurons. In ASW, the latency for B34 firing was <2 s (see ). After hexamethonium (104 M) application, the latency increased. The effects of hexamethonium were partially reversible. B-D: similar effects of hexamethonium application were observed on the latency for firing B31, B61, and B63. Note that the time bar for A and B represents 2 s, but for C and D, it represents 5 s. E: summary of the data from experiments similar to that in A-D in a number of preparations. For each preparation, several runs were performed, and the data were averaged and normalized to the results obtained before the hexamethonium was applied. A total of 3 B63, 7 B34, and 4 B31 preparations were tested. Data were combined and tested for statistical significance, which revealed that hexamethonium application significantly affected the latency for firing in each of the protraction neurons tested [for B63, P < 0.05; t(2) = 4.91; for B34, P < 0.05; t(6) = 2.29, and for B31, P < 0.02; t(3) = 5.7; 2-tailed paired t-test].

To quantify the data on the delay of activity in CPG elements as a result of treatment with hexamethonium, the data were expressed as a percentage of the result obtained when stimulation of CBI-2 was performed in ASW (Fig. 14E). An analysis of these data showed that there were significant increases in the latency to trigger the first action potential in B31/B32 [P < 0.05; t(3) = 5.7], B34 [P < 0.05; t(6) = 2.29], and B63 [P < 0.05; t(2) = 4.91; 2-tailed paired t-test; Fig. 14]. For B63, the onset was delayed from 5.2 to 7.7 s, for B34, the onset was delayed from 8.6 to 13.5 s, and for B31, the onset was delayed from 4.4 to 5.8 s.

Hexamethonium attenuates fast synaptic potentials elicited by B34 and B63 and differentially delays the onset of firing in their followers

The preceding data indicate that CBI-2 may be a cholinergic neuron because the fast EPSPs it elicits in the protraction phase neurons of the buccal ganglia were blocked by hexamethonium. We found that the buccal ganglion protraction-phase neurons are also likely to be cholinergic because hexamethonium was also found to block the interconnections between protraction phase interneurons (Fig. 15). Hexamethonium (10⁴ M) attenuated the connection from B63 to B31/B32 by 70% and blocked the connection from B34 to B31/B32. It is important to note that the EPSP from B63 to B31/B32 was previously shown to be a mixed electrical/chemical synapse, and the residual EPSP displayed in the presence of hexamethonium is likely to be the electrical component of this connection. The effects of hexamethonium were fully reversed when it was washed out.

View larger version (34K): [in this window] [in a new window]   Fig. 15. EPSPs elicited in B31 by firing protraction-phase interneurons B63 and B34 are hexamethonium sensitive. Note that this experiment was performed in HiDi. A, left: firing of B63 for one second at a frequency of 10 Hz evoked a train of EPSPs in B31. A, middle: the EPSPs amplitude was attenuated in hexamethonium (5 × 104 M). Note that the residual component of the EPSP is a previously described electrical PSP. A, right: the block of the chemical EPSP caused by hexamethonium was reversed when the preparation was washed in HiDi. B: similar results were obtained when EPSPs in B31 were elicited by firing B34. B, left: firing B34 for 1 s at frequency of 10 Hz evoked a train of facilitated EPSPs in B31. B, middle: the EPSPs were fully abolished in hexamethonium (5 × 104 M). B, right: the effect of hexamethonium was reversed following a washout in HiDi.

Previous data showed that B63 elicits monosynaptic EPSPs in B31/B32 and also initiates a BMP, which consists of a protraction and, after a delay, a retraction. The finding that B63 is a cholinergic neuron prompted us to examine whether B63 can drive either protraction or retraction when cholinergic transmission is blocked with hexamethonium.

Firing B63 at a low rate elicited a BMP in which firing in the B31/B32 axon was initiated after ~6 s, and retraction was initiated after ~21 s (Fig. 16A). In the presence of hexamethonium (10⁴ M), firing in the B31/B32 axon was delayed and began after ~16 s, but the time of onset of retraction was minimally affected (Fig. 16B). The action of hexamethonium was partially reversible (Fig. 16C). This experiment demonstrates that hexamethonium slows but does not block the protraction phase while having a minimal effect on retraction. Thus B63 activity is likely to activate receptors that are not sensitive to hexamethonium, which set into play a delayed protraction and a normally timed retraction. Experiments similar to those in Fig. 16 were performed in seven preparations.

View larger version (16K): [in this window] [in a new window]   Fig. 16. Hexamethonium delays the onset of B31 firing but leaves intact the onset of the retraction phase during B63-elicited BMPs. A: in this experiment, B63 was stimulated to fire at a low frequency. This drives a cycle of BMP with ~30 s duration. The B31 axon fires ~6 s after the start of the stimulation. In this and in all recordings from B31, spikes in the B31 axon fail to invade the soma and are recorded as fast depolarizations of ~10 mV that are superimposed on a sustained depolarization of B3. The axon spikes occur on a background of EPSPs from B63 and from other neurons. Activity in B63 and in B31 is terminated and B4 is depolarized and fires, at the start of the retraction phase. B: hexamethonium (104 M) delayed the onset of firing in B31 () in response to B63 stimulation to ~16 s. However, the latency for onset of retraction phase was unchanged. C: the hexamethonium effect was reversible. Note that in hexamethonium, B63 still elicited electrical EPSPs in B31 that are ~60% smaller than the mixed chemical-electrical EPSPs elicited in the absence of hexamethonium.

CBI-1 and CBI-3 modulate BMPs

The frequency of CBI-2 firing was shown to affect the length of the protraction phase, and the onset of retraction (see Fig. 11). In part, this is likely to arise from the facilitation of the EPSPs elicited by CBI-2 in B63 and an increase in B63 firing. However, an increased firing frequency in CBI-2 can also have additional effects. Previous data (Rosen et al. 1991) have shown that activity in CBI-2 causes firing in additional CBIs, which in turn could affect the buccal ganglion CPG. These CBIs are also activated by food stimuli in parallel to their activation of CBI-2. The ability of CBI-2 to recruit other CBIs is likely to be dependent on the firing frequency of CBI-2. We examined the effects of CBI-2 on the protraction-phase neurons with and without the additional activity of two other CBIs, CBI-1 and CBI-3.

To test whether firing CBI-1 along with CBI-2 could affect the length of the protraction phase, and the transition to retraction, we initiated BMPs by firing CBI-2 at a relatively low frequency (12 Hz until the end of a single BMP). Superimposed on this activity, CBI-1 was stimulated at different frequencies, beginning at 5 s after the start of the CBI-2 stimulus. Increasing the firing frequency of CBI-1 systematically decreased the BMP duration (Fig. 17A).

View larger version (36K): [in this window] [in a new window]   Fig. 17. CBI-1 was fired at different frequencies 5 s after the start of CBI-2 stimulation at 12 Hz. A: when CBI-1 was inactive, the BMP was terminated after 27 s. Progressively increasing CBI-1 firing frequency in 5-Hz steps led to a graded decrease of the BMP duration to <10 s. Note that CBI-2 was fired at a constant frequency of 15 Hz with individual pulses, but the pulses were superimposed on a DC depolarization. B: the means ± SE of 2 parameters of the BMPs at each of the 5 frequencies of stimulation are presented before normalization. C, 1 and 2: statistical tests revealed significant differences in the duration of BMP and in the duration of the protraction phase as a function of CBI-1 firing frequency. For each preparation, data were averaged for each of the 5 tested frequencies (each frequency was tested several times) and normalized to the result obtain when CBI-1 was not stimulated. *, P < 0.05; **, P < 0.001 in a Dunnet's post hoc test.

To quantify the effects of CBI-1 activity on BMPs elicited by CBI-2, In five preparations, CBI-1 was stimulated at five different frequencies (0, 5, 10, 20, and 30 Hz), during programs initiated by CBI-2. The different frequencies were applied in a random order with a rest of 60 s between trials. Stimulation of CBI-1 began 5 s after the start of CBI-2 stimulation and was continued until the end of the elicited motor program. The data were averaged and then expressed as a percentage of the results in the absence of CBI-1 activity. There were significant differences in the overall duration of the BMP [F(4,20) = 19.8; P < 0.0001] as well as in the duration of the protraction phase [F(4,20) = 39.8; P < 0.0001; Fig. 17, B and C]. However, there was no significant difference in the duration of the retraction phase [F(4,20) = 0.4; P > 0.5; 1-way analyses of variance; not shown]. These data indicate that an increase in CBI-2 frequency may affect the pattern of a BMP in part via effects on CBI-1 as well as via the increased facilitation.

The activity of CBI-1 affects the pattern of a BMP via its direct effects on the protraction phase neurons in the buccal ganglia. We found that when both the cerebral and buccal ganglia were bathed in HiDi (n = 5), firing CBI-1 for 2 s evoked a 2-mV depolarization in B63, but not in B34, when the membrane potentials of the cells were held at -65 mV (Fig. 18A1). However, when B63 and B34 were held at -50 mV, CBI-1 firing led to a stronger depolarization and firing in B63 (Fig. 18A2) with little or no effect on B34 (Fig. 18A, 1 and 2). In a higher gain recording (Fig. 18B), a small, summated inhibitory postsynaptic potential (IPSP) of <1 mV was detected in B34 and a small, summated EPSP of <0.5 mV was detected in B31. These PSPs lasted for ~1 s. To investigate the differential effects of CBI-1 activity on B63 and B34, B34 was depolarized with brief suprathreshold pulses with and without stimulating CBI-1, and the effect of this stimulus on EPSPs in B31/B32 was examined. CBI-1 activity caused a suppression of B34 firing (Fig. 18C) and a cessation of the B34-elicited EPSPs in B31/B32. In place of the excitation from B34, a barrage of EPSPs was seen in the B31/B32 neuron from another source, (presumably B63 because the EPSPs are similar to those elicited by firing of B63). These data indicate that recruiting CBI-1 leads to a modulation of BMPs by amplifying the activity of B63 while inhibiting B34. The differential targeting of protraction phase neurons is likely to affect the motor output because neurons such as B34 excite some of the motorneurons (see Hurwitz et al. 1997 for its excitatory effects on B61/B62 and B8). Other CBIs may also modulate the form of a BMP, as shown by a previous study that demonstrated the role of CBI-3 in shaping CBI-2 programs (Morgan et al. 2002). Thus these data are consistent with the hypothesis that recruiting CBI-1 affects the protraction phase via the direct effects of CBI-1 in the buccal ganglia.

View larger version (21K): [in this window] [in a new window]   Fig. 18. CBI-1 differentially targets central pattern generating (CPG) elements. A1: firing CBI-1 elicited EPSPs in B63 but not in B34 when B63 and B34 were held at -65 mV. A2: when B63 and B34 were depolarized, stimulation of CBI-1 elicited firing in B63 with no detectable effect in B34. B: replacing the ASW with HiDi uncovered a slow inhibitory postsynaptic potential (IPSP) in B34 elicited by CBI-1. Note that B31 was depolarized but the summated EPSP was <2 mV. Also note that a previous report by Sanchez and Kirk (2001) did not find this slow IPSP presumably because of its small size, but they did report finding a fast EPSP that decremented with repeated stimulation of CBI-1. C: brief depolarizing current pulses were injected into B34, eliciting firing in B34 and EPSPs in B31. Firing CBI-1 blocked the B34 spikes. Note that the EPSPs in B31 were similar in amplitude and shape to those elicited by firing of B63.

We also examined the effects of stimulating CBI-3, a second CBI neuron excited by CBI-2. These experiments (n = 4) were done in HiDi. In contrast to the effects of firing CBI-1, firing of CBI-3 elicited fast excitatory EPSPs of low amplitude (0.5 mV) in B34 with no comparable fast EPSPs in B31/B32 or B63 (not shown). Firing CBI-3 amplified the effects caused by weak depolarizing currents injection into B34. Pulses that were below threshold for inducing firing in B34 in the absence of CBI-3 activity induced firing during and somewhat after CBI-3 stimulation (Fig. 19A). A similar amplification of subthreshold pulses was not observed in B63 (not shown). However, a further investigation of the effects of CBI-3 on B63 showed that suprathreshold stimuli to B63 were weakly amplified in that CBI-3 caused a twofold increase in the firing frequency (Fig. 19B).

View larger version (23K): [in this window] [in a new window]   Fig. 19. CBI-3 differentially targets CPG elements. A: depolarizing current pulses injected into B34 were subthreshold for firing in the absence of CBI-3 stimulation. However, when CBI-3 was fired, the same current pulses elicited B34 firing. Note that ~12 Hz firing rate of CBI-3 was sufficient for generating ~15-Hz firing rate in B34. B: by contrast, ~30 Hz firing in CBI-3 that was superimposed on suprathreshold depolarizing pulses to B63 increased the B63 firing only moderately, from ~8 to 12 Hz. C: CBI-3 drives B31/B32 via its effects on B34. Firing CBI-3 elicited firing in both B34 and B31/B32 (left). When B34 was hyperpolarized, firing CBI-3 did not elicit depolarization and firing in B31/B32 (middle). This effect was reversed when B34 was released from the hyperpolarization (right).

To examine in greater detail the differential effects of CBI-3 activity on B63 and B34, CBI-3 was depolarized sufficiently, so that firing was seen in both the B31 axon and in B34 (Fig. 19C, left). Hyperpolarizing B34, and thereby preventing it from firing, led to a large decrease in the depolarization, from a value of 40 to <20 mV, recorded in B31, and a cessation of its axon firing (Fig. 19C, middle). The effect of B34 hyperpolarization was easily reversed (Fig. 19C, right). B34 cannot alone drive a depolarization of B31 as large as that seen in Fig. 19C (Hurwitz et al. 1997), and this depolarization is likely to be via the effects of B34 on B63, which is able to elicit strong excitation of B31. These data indicate that both CBI-1 and -3 act on the protraction phase neurons that are followers of CBI-2. This action presumably shapes the form of the BMPs elicited by CBI-2. These effects are likely to be components of the modulation of BMPs elicited by CBI-2.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We found that the buccal motor programs driven by the CBI neurons that were examined are similar to those observed when CPG neurons in the buccal ganglia are directly stimulated (Figs. 1, 5, 6, 7, 16, and 17). This is because the ability of the CBIs to drive motor programs stems in part from monosynaptic excitatory connections onto key protraction-phase interneurons (Figs. 2, 3, 4, 8, 18, and 19), which can themselves initiate programs. One of the cells examined, CBI-2, causes firing of the protraction-phase neurons only after the direct connections undergo a strong facilitation (Fig. 8). Thus the facilitation is an integral feature contributing to the ability of CBI-2 to recruit the CPG neurons and to initiate buccal motor programs.

The CBIs were found to excite monosynaptically only protraction phase interneurons and motorneurons, such as B31/B32, B34, B61/B62, and B63, and not retraction-phase interneurons, such as B64. Additional studies (Jing and Weiss 2001, 2002) have shown that the CBIs also excite other protraction phase neurons, such as B20 and B40. Previous studies have shown that BMPs initiated by stimulating peripheral nerves or identified neurons, as well as feeding responses to food, always begin with a protraction phase, and this is followed by retraction (Kupfermann 1974; Susswein and Byrne 1988). The sequencing of motor programs seems to arise in part from the direct recruitment of protraction phase neurons by the CBIs, with the subsequent recruitment of retraction-phase neurons via circuitry within the buccal ganglia. In Aplysia, all of the consummatory behaviors that have been identified, as well as all of the BMPs that are generally observed, are characterized by the protraction phase preceding the retraction. However, in Helisoma, there are programs that can begin with retraction (Quinlan and Murphy 1996)

We also found that the different CBIs examined differentially connect with the various protraction-phase neurons (Fig. 20). The three protraction-phase neurons examined in detail, B31/B32, B34, and B63, are all able to drive motor programs when depolarized (Hurwitz et al. 1994, 1997; Susswein and Byrne 1988) but are not equal targets for inputs from the CBIs. Some CBIs cause depolarization and firing of all of these neurons, whereas others differentially excite and inhibit the different interneurons. The differential activation of protraction phase neurons changes the form of a BMP (Figs. 11 and 17). Previous studies have shown that the differential activity of different CBIs can act as a mechanism for selecting specific types of BMPs (Jing and Weiss 2001, 2002; Morgan et al. 2002). In addition, firing CBIs at different frequencies also changes the form of a BMP, suggesting that modulating the firing frequency may also allow the CBIs to play a role in shaping a specific consummatory behavior.

View larger version (18K): [in this window] [in a new window]   Fig. 20. Summary of the synaptic connections discussed in this report. A: connections from CBI-2 to elements of the buccal ganglia central pattern generator (bold lines) as well as connections between the CPG elements. B: differential connection patterns from CBI-1, -2, and -3 to protraction phase interneurons B63 and B34.

It is important to note that there is little information on the firing frequencies of the CBIs in behaving animals. However, previous studies (Rosen et al. 1991) have characterized the firing frequency in response to natural stimuli such as food on the lips as well as characterizing the frequencies during buccal motor programs. The stimulus frequencies used in our experiments on the CBI cells are well within those that have been found previously to drive BMPs or to be characteristic of the activity of these neurons during BMPs. According to Rosen et al. (1991), the firing rate of CBI-2 in response to inner lips stimulation by seaweed is >10 Hz. Because such stimuli activate both the left and the right CBI-2, the effective input to buccal ganglia may be stronger than that represented by the rates of stimulation that we used with a unilateral stimulation of CBI-2. In addition, close examination of Fig. 8 in Rosen et al. (1991) shows that CBI-1 fires at frequencies in excess of 25 Hz. Thus the various effects of the CBIs that are described in this paper are likely to occur in behaving animals.

COMMAND-LIKE AND MODULATORY FUNCTIONS OF THE CBIS. CBI-2 has some of the properties of a command neuron. Command neurons are defined by being both necessary and sufficient to induce a specific behavior (Kupfermann and Weiss 1978). To be necessary and sufficient, the command neuron must be recruited by stimuli that cause the behavior, firing the command neuron should induce the behavior, and removing the command neuron should block the behavior (Kupfermann and Weiss 1978). CBI-2 is excited by food stimulating the lips (Rosen et al. 1991), the natural stimulus that initiates biting, one of the consummatory behaviors that are generated by the buccal ganglia CPG. Stimulating CBI-2 initiates motor programs, and previous data using chronic recording techniques have shown that the motor programs elicited by the protraction-phase interneurons that are recruited by CBI-2 are correlates of consummatory behaviors, such as biting, swallowing, and rejection (Hurwitz et al. 1996). Thus firing CBI-2 is sufficient for initiating a consummatory behavior, perhaps biting (see Morgan et al. 2002). There are no data directly testing whether CBI-2 is necessary for behavior. However, our data as well as that of others (Hurwitz et al. 1999b; Morgan et al. 2002; Rosen et al. 1991), suggest that CBI-2 is unlikely to act alone in initiating a consummatory response and is therefore unlikely to fulfill the criterion of necessity. In addition to exciting CBI-2, food stimuli also excite other CBIs as does the firing of CBI-2 (Rosen et al. 1991). Furthermore, the specific combination of CBIs that are activated biases the CPG to produce ingestion- or egestion-like behaviors (Jing and Weiss 2001, 2002; Morgan et al. 2002). Thus it is unlikely that CBI-2 acts as a command neuron eliciting a unique motor program, but rather CBI-2 is likely to act as an important element in a system that both initiates programs and also biases the CPG toward one of a number of different motor patterns.

DIFFERENTIAL SELECTION OF PROTRACTION PHASE NEURONS. CBI-2 initiates BMPs in part via its monosynaptic excitation of B31/B32, B34, and B63. However, it is possible that motor programs elicited by stimulation of other neurons, or via natural stimuli initiating consummatory behaviors, differentially act on other CPG elements.

FUNCTION OF FACILITATION. We have shown that the fast EPSPs from CBI-2 to the protraction-phase interneurons undergo a prominent facilitation (Figs. 2, 3, 4, 9, and 10). This is consistent with previous findings (Hurwitz et al. 1999a), which showed that the output of CBI-2 to B63, as well as the output to protraction phase motor neurons B61/B62 and to interneuron B34 (Sanchez and Kirk 2000), undergo facilitation. Our data extend the previous findings by showing that facilitation affects both the fast and slow components of the EPSP, by showing that facilitation in part underlies the ability of CBI-2 to initiate motor programs, and by examining systematically the effects of different firing frequencies as well as by examining a number of properties of the EPSPs. We confirmed that the facilitation elicited by a brief high-frequency burst of activity in CBI-2 is maintained for 1 min as was demonstrated in a previous study (Sanchez and Kirk 2000). However, the previous work did not distinguish between the effects of the fast and slow components of the EPSP. In addition, tonic firing preceding a burst of activity was now shown to produce a less profound facilitation lasting only several seconds, an issue that was not previously explored. There are functional differences in the effects of the quickly and slowly decaying components of facilitation. The quickly decaying component of facilitation is likely to affect only a single cycle of a program (Fig. 10) because it decays during the retraction phase. By contrast, the slowly decaying component of facilitation will affect multiple cycles of BMPs (Fig. 9). Our data also indicate that the ability of CBI-2 to command a BMP is functionally related to the facilitation because firing in the protraction phase neurons is elicited only after the EPSPs are facilitated.

CHOLINERGIC TRANSMISSION BY CBIS AND BY PROTRACTION INTERNEURONS. Our data suggests that acetylcholine (ACh) is the transmitter underlying the fast EPSPs from CBI-2 to the buccal ganglion protraction interneurons as well as underlying the fast EPSPs caused by the firing of the protraction-phase interneurons. Previous studies (Hurwitz et al. 2000) have shown that ACh is also the transmitter released by protraction-phase motor neurons B31/B32 and B61/B62 onto muscle I2, the major muscle that effects protraction. Additional data supporting the notion that these neurons are cholinergic stem from the finding that muscarinic antagonists block the slow components of the EPSPs. However, additional experiments will be needed to demonstrate conclusively that these neurons are cholinergic. By contrast, B20 and B65, two additional interneurons that are active during the protraction phase, utilize dopamine as their transmitter (Kabotyanski et al. 1998; Teyke et al. 1993). Thus the protraction phase neurons are heterogeneous with respect to the transmitter used.