This is a summary of the paper:

Regulation of the Segmental Swim Generating System by a Pair of Identified Interneurons in the Leech Head Ganglion


by Peter D. Brodfuehrer, Hannah J. Parker, Andrew Burns and Melissa Berg
The aim of this study was to identify neurons that modulate activity of segmental swim gating interneurons. We found a pair of bilaterally symmetrical interneurons, cells SE1, whose activity level directly influences three groups of segmental neurons associated with generating swimming in the medicinal leech.

Introduction

In the medicinal leech, Hirudo medicinalis, the initiation of swimming following body wall stimulation can be traced, neuron to neuron, from the mechanosensory neurons that perceive the stimulus to motor neurons that produce the swimming movements (Brodfuehrer and Friesen 1986a&b; Kristan et al. 1988; Friesen 1989). However, in intact leeches presentation of a constant mechanosensory input or stimulus can elicit one of several well-defined motor outputs or behaviors, including swimming (Kristan et al. 1982). Two studies have demonstrated that the firing frequency of swim gating interneurons, cells 204, is a critical factor in determining whether swimming is initiated in the isolated leech nervous system. First, Brodfuehrer and Friesen (1986c) showed that the effectiveness with which cell Tr1 triggers swimming is highly correlated with the level of activity in cells 204. Second, Kristan et al. (1988) demonstrated that in the isolated segmental nerve cord increasing cell 204's firing frequency switches a sub-threshold mechanosensory stimulus for the initiation of swimming into a supra-threshold stimulus.

Although it is clear from both of these studies that an elevated level of excitation in cells 204 is necessary for initiating swimming, it is not known how the activity level of cells 204 is regulated in the leech. In both of the aforementioned studies, the firing frequency of cell 204 was manipulated by intracellular current injection. The only known direct excitatory input to cell 204 is from cells Tr1 located in the subesophageal ganglion (Brodfuehrer and Friesen 1986a&b). However, in isolated nerve cord preparations cells Tr1 are not spontaneously active (Brodfuehrer and Friesen 1986b). Unless, cells Tr1 act differently in intact leeches, their activity pattern precludes them from being involved in modulating the firing frequency of cells 204. Thus, other as yet unidentified neurons in the leech CNS must be able to modulate the activity level of cell 204.

In this paper we characterize a pair of interneurons, cells SE1, whose cell bodies are located in the subesophageal ganglion, that strongly excites several members of the segmental swim-generating system including segmental swim gating interneurons (cells 204), swim oscillatory interneurons (cells 115, 28 and 208) and segmental motor neurons (cells 3, 5, and 7). Moreover, we demonstrate that the level of excitation in cells 204, 115 and 3 is positively correlated with the activity level in cells SE1. Based upon the potent and widespread output connections of cells SE1, we hypothesize that cells SE1 function as swim exciters or gain control interneurons for the initiation of swimming by setting the level of excitation in the segmental swim-generating system.

Part of these results have appeared in abstract form, and in these abstracts cell SE1 was referred to as cell Tr4 (Brodfuehrer et al. 1993a; Brodfuehrer and Burns 1993).

Methods

Animal preparation: Medicinal leeches, Hirudo medicinalis, were purchased from Leeches USA and maintained in aquaria at room temperature. The CNS of the medicinal leech consists of head and tail ganglia connected by the ventral nerve cord, a chain of 21 segmental ganglia and their intersegmental connectives. The head ganglion is composed of the subesophageal ganglion, consisting of four neuromeres, and the supraesophageal ganglion. The four neuromeres of the subesophageal ganglion are referred to as R1 - R4, anterior to posterior. On the dorsal side of the subesophageal ganglion each neuromere is subdivided into an anterior and a posterior packet, except for R1 (Fig. 1A). The segmental ganglia are numbered sequentially, from M1 - M21, beginning with the ganglion posterior to the subesophageal ganglion (Muller et al. 1981; Brodfuehrer and Cohen 1992). The segmental ganglia are joined by connectives that are composed of two large lateral bundles of nerve fibers and a thin median bundle called Faivre's nerve. A pair of nerve roots arises from each segmental ganglion to innervate the body wall surrounding that segment (Muller et al. 1981). The ipsilateral and contralateral side of the leech CNS is referenced to the location of the soma of cell SE1.

All experiments were performed on isolated leech nerve cords extending from the head ganglion to the tail ganglion. During dissection and recording, nerve cords were bathed in leech physiological saline containing 115 mM NaCl; 4 mM KCl; 1.8 mM MgCl2; 2 mM CaCl2 and 10 mM Hepes buffer. To test if two neurons were connected monosynaptically the concentration of CaCl2 and MgCl2 in the leech physiological saline was elevated to 10 mM, with an equivalent reduction in the Na+ concentration (Berry and Pentreath 1976; Nicholls and Purves 1972). This physiological saline is referred to as high Mg++ and Ca++ saline. To determine if a synaptic connection was chemically mediated, the concentration of Ca++ and Mg++ was reduced to zero and 10 mM, respectively, in leech physiological saline.

Electrophysiological and morphological techniques: In isolated nerve cords, swimming activity was monitored by recording extracellularly, with suction electrodes, from a branch of a segmental peripheral nerve root, the dorsal posterior (DP) nerve. Rhythmic bursts of spikes from cell 3, a motor neuron whose axon runs in the DP nerve, are indicative of swimming (Kristan et al. 1974). Intracellular membrane potentials were measured with glass microelectrodes filled with 4 M KAc. To facilitate intracellular penetration of neurons in the subesophageal ganglion and in segmental ganglia, the connective sheath surrounding these ganglia was removed with fine scissors. This procedure did not adversely affect the functioning of the nervous system and the preparation remained viable for several hours (Friesen 1985). All electrical signals from the intracellular and extracellular recordings were recorded on VCR tape and analyzed off-line on a Macintosh IIci computer equipped with a GW Instruments A/D board (sampling rate = 2000 samples/s per channel) and software (SuperScope II).

The level of excitation induced in cells 204 and 115 by stimulation of cell SE1 was quantified in the following manner. One intracellular microelectrode was used to inject 1 s depolarizing current pulses, of varying intensity, into cell SE1, while a second intracellular microelectrode was used to record the resulting voltage change produced in either cell 204 or cell 115. Using SuperScope II, the average spike frequency of cell SE1 during the current pulse was calculated along with the area (volt-sec) under the membrane depolarization induced in either cell 204 or cell 115. To ensure that the baseline voltage of cell 204 or cell 115 used in the depolarization area calculation was approximately equal between trials, and to eliminate negative voltage values in the depolarization area calculation, activity records of cell 204 or cell 115 were first offset so that their lowest voltage value equaled zero before calculating the depolarization area. The beginning and end of the induced depolarization area were set manually. Membrane depolarization area was then graphed as a function of the average firing frequency in cell SE1 and a linear regression line drawn through the points using Cricket Graph. A similar comparison was performed on the connection from cell SE1 to cell 3, except that the average firing frequency of cell 3 was used as a measure of the postsynaptic response. With this data set, a value for the synaptic transfer coefficient from cell SE1 to cell 3 was calculated from the slope of the linear regression line (post- and presynaptic neurons) drawn through the data points (Granzow et al. 1985). Cell 3 spikes were recorded extracellularly from DP nerves along the ventral nerve cord. We were not concerned that our measurement of cell 3's firing frequency would be significantly skewed by the inclusive of L cell spikes, which also can be recorded extracellularly in the DP nerve (Ort et al. 1974), because our results indicated that intense stimulation of cell SE1 rarely led to the production of L cell action potential (see results).

The morphology of cell SE1 was determined by intracellular injection of Lucifer Yellow (Stewart 1978). Once cell SE1 was iontophoretically stained with Lucifer Yellow, nerve cords consisting of the head ganglion through M2 were pinned out in a sylgard-coated petri dish, fixed in 4% paraformaldehyde for 30 min, washed in 0.1 M phosphate-buffered saline for 30 min, dehydrated though an alcohol series and cleared in xylene. A line drawing of cell SE1 was produced using a drawing tube attached to an Olympus BH-2 microscope, which was then scanned into a Macintosh IIci computer and printed.

Results

Morphology cell SE1. Cells SE1 (Swim Exciter 1) occur as a bilaterally symmetrical pair of interneurons whose somata are located on the dorsal aspect of the subesophageal ganglion near the midline of the anterior packet of R2 (Fig. 1A). Generally, the soma of cell SE1 was located below and partially obstructed by one or two large, midline somata. The primary neurite of cell SE1 projects rearward and exits the subesophageal ganglion in the ipsilateral connective. To determine how far the axon of cell SE1 extended in the ventral nerve cord, we recorded extracellularly from the ipsilateral connective between M14 and M15, while recording intracellularly from the soma of cell SE1. As is shown in Fig. 1B, a computer-averaged cell SE1 impulse was recorded between M14 and M15. Thus, the axon of cell SE1 extends most, it not the entire length, of the ventral nerve cord. Processes of cell SE1 branch primarily along the midline of the subesophageal ganglion, with a few processes also projecting into the circumesophageal connective and supraesophageal ganglion. No processes of cell SE1 were observed in any peripheral nerves of the subesophageal ganglion.

Physiological properties of cells SE1. The physiological characterization of cells SE1 grew out of an initial observation that a short duration (0.5 - 1 s), high frequency burst of action potentials in cell SE1 could initiate swimming (Fig. 2A). In this respect, cells SE1 behave exactly like two previously described trigger neurons in the subesophageal ganglion of the leech, cells Tr1 and Tr2 (Brodfuehrer and Friesen 1986a). However, because the cell bodies of cells SE1 and cells Tr1 and Tr2, and hence their intracellular recording sites, are located on opposite sides of the subesophageal ganglion a direct comparison between the swim-initiating ability of cells SE1 with cells Tr1 and Tr2 was not performed. Nevertheless, characteristics of cell SE1's swim-initiating ability, along with several of its physiological properties, strongly suggested that cells SE1 were not functionally identical to cells Tr1 and Tr2.

Three significant differences were observed between the manner in which cell SE1 initiated swimming compared to cells Tr1 and Tr2. First, swimming often occurred within 1 s of cell SE1 stimulation. In fact, cell SE1 stimulation sometimes led to the nearly immediate activation of the swim oscillator (Fig. 2A). In these instances the 'preparatory phase' of swimming, where excitation gradually builds in swim gating and oscillatory interneurons, was missing. In contrast, swimming usually followed stimulation of cells Tr1 and Tr2 by several seconds. Surprisingly, a preparatory phase was observed in cell SE1 before the onset of spontaneous swim episodes (swim episodes initiated without obvious input from the experimenter)(Fig. 2B). Second, stimulation of cell SE1 occasionally elicited abbreviated swim bouts consisting of only one or two swim cycles (Fig. 2C). Abbreviated swim episodes are not observed following stimulation of cells Tr1 and Tr2. Lastly, the average firing frequency of cell SE1 necessary to trigger swimming ranged from 50 Hz to 70 Hz, while firing frequencies as low as 30 Hz could trigger swimming in cells Tr1 and Tr2.

The physiological response of cells SE1 during swimming was also different than cells Tr1 and Tr2 . During swimming the membrane potential of cell SE1 depolarized 2 - 5 mV (Fig. 2A), and in some preparations, modulated rhythmically (Figs. 2B and 6A1). Depending on the resting membrane potential of cell SE1, a 2 - 5 mV depolarization caused the tonic firing frequency of cell SE1 to increase slightly. For example, in Fig. 2A the firing frequency of cell SE1 increased from a tonic rate of approximately 2 Hz before cell SE1 was stimulated to an average firing frequency of 7 Hz during the swim episode. The average firing frequency of cell SE1 during swimming was generally between 5 and 10 Hz, but could be as high as 20 - 25 Hz. Excitatory feedback from the swim oscillator is obviously responsible for depolarizing cells SE1 during swimming, but the specific neurons have yet to be identified. Neither cells Tr1 nor Tr2 are active during swimming in the isolated nerve cord.

Tonic activity in cells SE1 was also markedly different than in cells Tr1 and Tr2. In quiescent preparations (when swimming was not occurring), the tonic firing frequency of cells SE1 ranged from 0 Hz to 15 Hz during the time course of an experiment. In addition, cells SE1 received a lot of synaptic input, particularly large inhibitory postsynaptic potentials (see Fig. 4). Cells Tr1 and Tr2 are usually silent throughout an experiment. Like cells Tr1 and Tr2, the SE1 cell pair was weakly, electrically coupled as demonstrated by their physiological responses in leech saline containing 10 mM Mg++ and zero Ca++ (not shown). Since this coupling was weak, a high firing frequency in one cell SE1 did not substantially increase the firing frequency of the other cell SE1 (Fig. 6A1).

Interaction between cell SE1 and cell 204. Since stimulation of cell SE1 can elicit swimming, we examined the connection from cell SE1 to swim gating interneurons, cells 204. Using pairwise intracellular recordings we found that a high frequency burst of action potentials in cell SE1 rapidly depolarized cell 204 by several millivolts and usually led to an increase in cell 204's firing frequency (Figs. 3A and 5A). The response elicited in cell 204 by stimulation of cell SE1 was unlike the gradual increase in cell 204's activity observed following stimulation of cell Tr1. Further examination revealed that cell SE1 was directly connected to cell 204, as was evident by the presence of constant-latency excitatory postsynaptic potentials (EPSPs) in normal saline and their persistence in high Mg++ and Ca++ saline in all segmental ganglia (M10 - M14) examined (Fig. 3B). In fact, both cells SE1 connected directly to cell 204 (not shown). In three preparations we measured the amplitude and duration, at half-maximum amplitude, of the EPSP in cell 204 from stretches of data in which cell SE1 was firing at a less than 10 Hz. In high Mg++ and Ca++ saline the average amplitude of the cell SE1-induced EPSP in cell 204 was 1.7 mV, while the average EPSP duration was 32 ms. The homolog of cell 204 in M9 (Weeks 1982b), cell 205, also received direct excitatory synaptic input from cell SE1 (not shown). Stimulation of neither cell 204 nor of cell 205 affected cell SE1.

Two experiments were performed to determine whether the level of excitation in cell 204 was modulated by the firing frequency of cell SE1. First, we examined the effect that elimination of spiking activity in cell SE1 had on activity in cell 204. As is shown in Fig. 4, hyperpolarization of one cell SE1 with approximately 1 nA, which abolished all spiking activity in cell SE1, completely eliminated all excitatory events in cell 204. Both action potentials and EPSPs occurring in cell 204 ceased during the hyperpolarizing current pulse. The most likely reason that eliminating spiking activity in just one cell SE1 suppressed all EPSPs and action potentials in cell 204 is that the other cell SE1 was silent, either spontaneously or induced through electrical coupling with its cell pair. In fact, we rarely observed an EPSP in cell 204 that did correlate with a cell SE1 spike in quiescent preparations. Spontaneously occurring IPSPs in cell 204 were not affected by hyperpolarizing cell SE1. Second, we injected 1 s depolarizing current pulses, of varying intensity, into cell SE1 and recorded the response, the amount of membrane depolarization, produced in cell 204 (Fig. 5A). In three preparations, the amount of membrane depolarization elicited in cell 204 increased linearly over the cell SE1 frequency range tested (Fig. 5B). Thus our results suggested that in quiescent preparations almost all tonic EPSPs observed in cell 204 are due primarily to synaptic input from cell SE1, and that the level of excitation in cell 204 is positively correlated with the firing frequency of cell SE1.

Cell SE1 and the swimming rhythm. The firing frequency of cell 204 is inversely related to swim period (Debski and Friesen 1985; Weeks and Kristan 1978). Since the level of excitation in cell 204 is directly related to the firing frequency of cell SE1, we examined the effect that stimulation of cell SE1 had on the swimming rhythm. We found that the effect cell SE1 stimulation had on swimming depended primarily on the duration and intensity level of cell SE1 activity. Short duration (0.5 sec) pulses in cell SE1 effectively shifted the phase of the swimming rhythm (Fig. 6A). Stimulating cell SE1 for a slightly longer duration (2 -4 s) often caused both the swim period and burst duration to dramatically increase (Fig. 6B), while continuous stimulation of cell SE1 disrupted the swimming rhythm and often prematurely halted swimming (not shown). The intensity at which the above effects were observed varied considerably. In most preparations, phase shifts required the highest cell SE1 firing frequencies, 40 - 50 Hz, while changes in burst duration occurred at approximately 25 Hz. No consistent changes were observed in the swimming rhythm at cell SE1 frequencies less than 25 Hz.

Interaction between cell SE1 and swim oscillatory interneurons. The ability of cell SE1 to shift the phase of the swimming rhythm suggested that cell SE1 was strongly connected to the swim oscillatory network. Pairwise intracellular recordings revealed that cell SE1 was connected directly to three oscillatory interneurons, cells 115, 28 and 208, as evident by the presence of constant latency EPSPs following cell SE1 impulses in normal saline and in high Mg++ and Ca++ saline (Figs. 7A, 8 and 14). Stimulation of cell SE1 did not consistently excite two other oscillatory interneurons examined, cells 27 and 33 (not shown). In addition, no reciprocal connections were found between these oscillatory interneurons and cell SE1.

Although cell SE1 was connected directly to cells 28, 208 and 115, two major differences were observed between their respective synaptic interactions with cell SE1. First, in high Mg++ and Ca++ saline, cell 28 received direct excitatory input from only the ipsilateral cell SE1 (Fig. 8), while cells 115 and cell 208, an unpaired interneuron, were directly excited by both cells SE1 (not shown). Second, cell SE1 had a stronger excitatory effect on cell 115 than on cells 28 and 208. This observation was based upon the fact that stimulation of cell SE1 always strongly excited cell 115. In contrast, stimulation of cell SE1 sometimes inhibited cell 28 (Fig. 9A), and often had little or no effect on a tonically active cell 208 (Fig. 9B). Furthermore, hyperpolarizing a tonically active cell SE1 generally eliminated all EPSPs in cell 115 and drastically reduced its tonic spike frequency (Fig. 7B). In contrast, hyperpolarizing cell SE1 did not consistently affect the tonic firing frequency of cells 28 and 208 (not shown).

Because the interactions from cell SE1 to cell 115 mimicked the effect of cell SE1 on cell 204, we determined whether the level of excitation in cell 115 was also positively correlated with the firing frequency of cell SE1. As is shown in Fig. 10A, the amount of membrane depolarization area in cell 115 increased linearly with increasing firing frequency of cell SE1. In one preparation, we were able to compare the level of excitation produced in the left and right cell 115 by stimulation of one cell SE1. In this preparation, the amounts of membrane depolarization area produced in each cell 115 were approximately equal at all cell SE1 firing frequencies (Fig. 10B). Thus the activity level in both cells 204 and cells 115 seems to be regulated in a similar manner by cell SE1.

Interaction between cell SE1 and swim motor neurons. Besides affecting swim gating and oscillatory interneurons, cell SE1 had a strong and rapid excitatory effect on the excitatory motor neurons, cells 3 (Fig. 11), 5 (Fig. 12) and 7 (not shown) that innervate dorsal longitudinal muscle. In fact, in quiescent preparations a defining physiological characteristic of cell SE1 was its ability to produce a burst of cell 3 action potentials in DP nerves all along the segmental nerve cord (Fig. 11A). Neither ventral excitatory nor dorsal and ventral inhibitory motor neurons to the longitudinal muscles were consistently excited by stimulation of cell SE1 (not shown). However, stimulation of cell SE1 excited L cells, segmental motor neurons that innervate both dorsal and ventral longitudinal muscles. High frequency stimulation of cell SE1 usually depolarized L cells by 2 -5 mV, but this level of depolarization only elicited a few L cell action potentials (Fig. 13).

To determine the strength of the connection from cell SE1 to cell 3 we injected 1 s depolarizing current pulses into cell SE1 while recording extracellularly from DP nerves at several locations along the nerve cord. The average firing frequency of cell SE1 during the current pulse was then compared with the average firing frequency of cell 3 recorded in each DP nerve. In four preparations analyzed we found that the firing frequency of cell 3 recorded in each DP nerve increased linearly with the firing frequency of cell SE1 (Fig. 11B). The average synaptic transfer coefficient (average firing frequency of cell 3 / average firing frequency of cell SE1) for the connection from cell SE1 to cell 3 was 0.53 + 0.07 (n = 12 DP nerves). In one preparation, we were able to compare the synaptic transfer coefficient from both cells SE1 to cell 3. In this preparation, we found that the synaptic transfer coefficient from the left and right cell SE1 to the same cell 3 was approximately equal, 0.54 and 0.49, respectively.

The strong effect of cell SE1 stimulation on cell 3 suggested that the connection between these two neurons was direct. However, in high Mg++ and Ca++ saline we were unable to identify a cell SE1 induced constant-latency EPSP in cell 3 (10 preparations). It was surprising that a one-to one relationship was not observed from cell SE1 to cell 3 for two reasons. First, constant-latency EPSPs were readily observed in cell 5 following action potentials in cell SE1 (Fig. 12B). Second, in high Mg++ and Ca++ saline, stimulation of cell SE1 still evoked a barrage of cell 3 action potentials (Fig. 14). Thus, although two physiological tests for monosynapticity failed to confirm the presence of a direct connection from SE1 to cell 3, the value of the synaptic transfer coefficient suggested that this connection was very strong. In fact, the relative strength of the connection from cell SE1 to cell 3 could be made by comparing it to other known direct excitatory inputs to cell 3. One such direct input is from cell 208 which connects directly to cell 3 in segmental ganglia posterior to its own location (Weeks, 1985). As is shown in Fig. 14, stimulation of cell SE1 produced a much greater cell 3 response recorded extracellularly in the DP nerve from M13 than stimulation of cell 208 in M10, even though the average firing frequency of cell 208 was twice as high as the firing frequency of cell SE1 (Fig. 14).

Discussion

The primary goal of this study was the identification of neurons that modulate the activity level of segmental swim-gating interneurons, cells 204. Cells SE1, which are located in the subesophageal ganglion, are clearly such interneurons. We've shown that: 1) cell SE1 directly excites cell 204, 2) the level of excitation evoked in cell 204 is positively correlated with the firing frequency of cell SE1, and 3) when the swim motor program is not active, elimination of spiking activity in cell SE1 abolishes almost all EPSPs in cell 204. The connection from cell SE1 to cell 204, however, represents only part of cell SE1's output connections (Fig. 15). We've also shown that cell SE1 directly excites three oscillatory interneuron, cells 115, 28 and 208, and that the connection from cell SE1 to cell 115 mimics that from cell SE1 to cell 204. Furthermore, cell SE1 strongly excites three motor neurons, cells 3, 5 and 7, that innervate dorsal longitudinal muscles, and that a highly positive correlation exists between with the firing frequency of cell 3 and cell SE1.

Are cells SE1 trigger neurons? We initially classified cells SE1 as swim trigger neurons because one of their physiological properties mimicked a defining property of previously described trigger neurons in the leech; namely, brief activation of cell SE1 could lead to the initiation of a prolonged swim episode. However, our results suggest that the ability of cells SE1 to trigger swimming may simply be a by-product of its strong and direct output connections to cells 204 and several oscillatory interneurons, and unrelated to its normal function in the leech. This assertion is based primarily upon differences in how cells SE1 and previously described trigger neurons, cells Tr1 and Tr2, initiate swimming.

The salient difference between the swim initiating ability of cells Tr1 and Tr2 versus that of cells SE1 pertains to swim latency. In a freely behaving leech, swimming usually occurs several seconds following body wall stimulation (P.D. Brodfuehrer personal observation). This latent period is referred to as the 'preparatory phase'. Hydrodynamically, this preparatory phase is necessary, as a leech must prepare itself for swimming by extending and flattening its body (Sawyer 1986; Stent et al. 1978). At the cellular level, several physiological changes take place during the preparatory phase including a gradual build up of excitation in cells 204 and the swim oscillator (Ort et al. 1974; Weeks 1982a&b; Kristan and Weeks 1983; Nusbaum et al. 1987). Swim episodes initiated by stimulation of cells Tr1 and Tr2, and by mechanosensory stimulation are preceded by this preparatory phase (Brodfuehrer and Friesen 1986c; Debski and Friesen 1986). In addition, in the isolated leech nerve cord a preparatory phase precedes spontaneous swim episodes. On the other hand, intracellular stimulation of cell SE1 often initiated swimming in less than 1 s with little or no preparatory phase occurring in cells 204 and oscillatory interneurons. Instead, we observed that oscillatory interneurons began oscillating almost immediately after cell SE1 was stimulated, while the excitation level in cells 204 rapidly jumped to a higher, sustained level. In fact, the response induced in cell 204 by cell SE1 stimulation resembles the response produced in cell 204 by direct, intracellular current injection; the firing frequency of cell 204 increases rapidly to a sustained level and swimming ensues with a very short latency. However, the initiation of swimming by intracellular depolarization of cell 204 is itself 'artificial' since intracellular depolarization of cell 204 only reliably elicits swimming in reduced preparations (Brodfuehrer and Friesen, 1986e; Brodfuehrer and Burns, 1994). Hence the very short swim latencies observed following cell SE1 stimulation do not resemble the 'natural' sequence of events that normally precede leech swimming when swimming is initiated by stimulation of trigger neurons and mechanosensory neurons, or occurs spontaneously. Furthermore, the cell SE1 firing frequencies (50 - 70 Hz) necessary to elicit swimming appear unphysiologically high. Stimulation of mechanosensory neurons, the only know sensory input to cells SE1 (Brodfuehrer et al. unpublished observations), do not excite cells SE1 to the extent required to initiate swimming by current injection.

In many respects, cells SE1 resemble higher order homologs of cells 204 rather than trigger neurons. Specifically, both cells 204 and cells SE1 provide, excitatory drive to swim oscillatory interneurons, cells 115, 28 and 208 (Fig. 15), receive excitatory feedback from the swim oscillatory network during swimming and, under certain circumstances, evoke swimming when stimulated intracellularly. However, not all the physiological properties of cells SE1 mimic those of cells 204. One key difference is that intense stimulation of cell SE1 for several seconds during swimming increases both swim period and burst duration, while prolonged stimulation often stops swimming. In contrast, stimulation of cell 204 decreases swim period (Debski and Friesen 1986; Weeks and Kristan 1978). In addition, unlike cells 204, cells SE1 directly excite longitudinal motor neurons. Thus cells SE1 do not appear to be functional homologs of segmental swim gating neurons. Instead, our results strongly suggest that cells SE1 are a functionally distinct class of higher order neurons that are involved in regulating leech swimming.

Functional role of cells SE1. An alternative function of cells SE1, suggested by the physiological properties of their output connections, is that of a swim exciter or gain control interneuron involved in regulating the initiation of swimming in the leech CNS. As swim exciters, cells SE1 would help regulate whether a given sensory input initiates swimming by modulating the level of excitation in the segmental swim-generating system. Hypothetically, as the tonic firing frequency in cells SE1 increases, the level of excitation in the segmental swim-generating system would move closer to threshold for the initiation of swimming. Thus when the firing frequency of cells SE1 is high, the probability of initiating swimming by a given sensory input would also be high.

At a first glance, it should be possible to test whether cells SE1 act as swim exciters by intracellularly manipulating their firing frequency. Increasing the firing frequency of cells SE1 should effectively change a stimulus that is sub-threshold for the initiation of swimming into a supra-threshold stimulus. However, control of the leech swim-generating system is not that simple. The above experiment assumes that only excitatory inputs control leech swimming. In fact, both excitatory and inhibitory systems or pathways emanating from the head ganglion influence the segmental swim-generating system (Brodfuehrer and Burns 1994, Brodfuehrer and Friesen, 1986e). Cells SE1, given their extensive output connections to the segmental swim-generating system, are most likely major components of the descending, excitatory system. However, the descending, inhibitory control system appears stronger than the descending, excitatory system (Brodfuehrer and Friesen 1986e; Brodfuehrer et al. 1993b). Thus, simultaneous modulation of both the excitatory and inhibitory systems is most likely required to affect the probability of initiating swimming (Brodfuehrer and Burns 1994). This in fact appears to be the case since preliminary attempts at switching a sub-threshold sensory input to a supra-threshold one by just increasing the firing frequency of cell SE1 have had only limited success (Brodfuehrer et al. unpublished observations). Clearly, testing the functional role of cells SE1 will require a reliable means of manipulating both the excitatory and inhibitory pathways that affect the segmental swim-generating system.

If cells SE1 function as swim exciters, a mechanism must also exist in the leech CNS for controlling their activity level. We presently know very little about synaptic inputs to cells SE1. A possible function of the descending, inhibitory system mentioned above may be that of controlling the activity level of cells SE1. To date, only one member of the descending inhibitory system, cell SIN1, has been identified. Stimulation of cell SIN1, however, does not affect cell SE1, although stimulation of cell SIN1 indirectly inhibits cell SIN1 (Brodfuehrer and Burns 1994).

Control of CNS activity in the leech Besides cells SE1, widespread control of neuronal activity in the leech CNS has also been shown to be regulated by an identified pair of segmental neurons, cells 151 (Wadepuhl 1989). However, several physiological properties of cells SE1 are different than those of cells 151 which suggests that these interneurons most likely do not function in the same capacity within the leech CNS. First, cell SE1 primarily affects dorsal, excitatory motor neurons whereas cell 151 regulates the excitability of several classes of motor neurons. Second, injection of hyperpolarizing and depolarizing current pulses into cell SE1 decreases and increases dorsal motor neuron activity, respectively. In contrast, hyperpolarization of cell 151 decreases motor neuron activity, while depolarization has no effect. Finally, cell SE1 selectively influences several different functional classes of neurons all of which are associated with swimming, while only motor neurons are affected by cell 151. Thus, cells SE1 and cells 151 are not functional homologs. The function of cells SE1 is clearly more specific, being directly related to swimming, than cells 151.

References

Berry, M. S., and Pentreath, V. W. Criteria for distinguishing between monosynaptic and polysynaptic transmission. Brain Res. 105:1-20, 1976.

Brodfuehrer, P. D., and Burns, A. Neuronal factors influencing the decision to swim in the medicinal leech. Behav. Neural Biol. (submitted).

Brodfuehrer, P. D., and Burns, A. To swim or not to swim: control of swimming by neurons in the head ganglion of the medicinal leech. Physiologist 36:A-24, 1993.

Brodfuehrer, P. D., Burns, A., and Berg, M. Regulation of segmental swim-initiating interneurons by a pair of identified interneurons in the leech head ganglion. Neurosci. Abstr. 19:1600, 1993a.

Brodfuehrer, P. D., Kogelnik, A. M., Friesen W. O., and Cohen, A. H. Effect of the tail ganglion on swimming activity in the leech. Behav. Neural Biol. 59:162-166, 1993b.

Brodfuehrer, P. D., and Cohen, A. H. Glutamate-like immunoreactivity in the leech central nervous system. Histochemistry 97:511-516, 1992.

Brodfuehrer, P. D., and Friesen, W. O. From stimulation to undulation: A neuronal pathway for the control of swimming in the leech. Science 234:1002-1004, 1986a.

Brodfuehrer, P. D., and Friesen, W. O. Initiation of swimming activity by trigger neurons in the leech subesophageal ganglion. I. Output connections of Tr1 and Tr2. J. Comp. Physiol. A 159:489-502, 1986b.

Brodfuehrer, P. D., and Friesen, W. O. Initiation of swimming activity by trigger neurons in the leech subesophageal ganglion. II. Role of segmental swim-initiating interneurons. J. Comp. Physiol. A 159:503-510, 1986c

Brodfuehrer, P. D., and Friesen, W. O. Initiation of swimming activity by trigger neurons in the leech subesophageal ganglion. III. Sensory input to Tr1 and Tr2. J. Comp. Physiol. A 159:511-519, 1986d.

Brodfuehrer, P. D., and Friesen, W. O. Control of leech swimming activity by cephalic ganglia. J. Neurobiol. 17:697-705, 1986e.

Debski, E. A., and Friesen, W. O. Role of central interneurons in habituation of swimming activity in the medicinal leech. J. Neurophysiol. 55:977-993, 1986.

Friesen, W. O. Neuronal control of leech swimming movements In: Jacklet JW (ed) Neuronal and Cellular Oscillators. New York, Basel: Marcel Dekker Inc, p. 269-316, 1989.

Friesen, W. O. Neuronal control of leech swimming movements: interactions between cell 60 and previously described oscillator neurons. J. Comp. Physiol. A 156:231-242, 1985.

Granzow, B., Friesen, W. O., and Kristan, W. B. Jr. Physiological and morphological analysis of synaptic transmission between leech motor neurons. J. Neurosci. 5:2035-2050, 1985.

Kristan, W. B. Jr., Wittenberg, G., Nusbaum, M. P., and Stern-Tomlinson, W. Multifunctional interneurons in behavioral circuits of the medicinal leech. Experientia. 44:383-389, 1988.

Kristan, W. B. Jr., McGirr, S. J. and Simpson, G. V. Behavioral and mechanosensory neurone responses to skin stimulation in leeches. J. exp. Biol. 96:143-160, 1982.

Kristan, W. B. Jr., Stent, G. S., and Ort, C. A. Neuronal control of swimming in the medicinal leech. I. Dynamics of the swimming rhythm. J. Comp. Physiol. 94:97-119, 1974.

Muller K. J., Nicholls J. G., and Stent, G. S. Neurobiology of the Leech. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory, 1981.

Nicholls, J. G., and Purves, D. A comparison of chemical and electrical synaptic transmission between single sensory cells and a motorneurone in the central nervous system of the leech. J. Physiol. Lond. 225:637-656, 1972.

Nusbaum, M. P., Friesen, W. O., Kristan, W. B. Jr., and Pearce, R. A. Neural mechanisms generating the leech swimming rhythm: Swim-initiator neurons excite the network of swim oscillator neurons. J. Comp. Physiol. A 161:355-366, 1987.

Ort, C. A., Kristan, W. B. Jr., and Stent, G. S. Neuronal control of swimming in the medicinal leech. II. Identification and connections of motor neurons. J. Comp. Physiol. 94:121-154, 1974.

Sawyer, R. T. Leech Biology and Behavior. Vol 1. Anatomy, Physiology and Behaviour. Oxford, Clarendon Press, p. 271, 1986.

Stent, G. S., Kristan, W. B. Jr., Friesen, W. O., Ort, C. A., Poon, M., and Calabrese, R. L. Neuronal generation of the leech swimming movement. Science 200:1348-1357, 1978.

Stewart, W. W. Intracellular marking of neurons with a highly fluorescent naphtalimide dye. Cell 14:741-759, 1978.

Wadepuhl, M. Depression of excitatory motoneurones by a single neurone in the leech central nervous system. J. exp. Biol. 143:509-527, 1989.

Weeks, J. C. Synaptic basis of swim initition in the leech. I. Connections of a swim-initiating neuron (cell 204) with motor neurons and pattern-generating 'oscillator' neurons. J. Comp. Physiol. A 148:253-263, 1982a.

Weeks, J. C. Segmental specialization of leech swim-initiating interneuron, cell 205. J. Neurosci. 7:972-985, 1982b.

Weeks, J. C., and Kristan, W. B. Jr. Initiation, maintenance and modulation of swimming in the medicinal leech by the activation of a single neurone. J. exp. Biol. 77:71-88, 1978.

Figure Legends

1. Morphology of cell SE1. A. Camera lucida drawing of a Lucifer Yellow fill of cell SE1. The soma of cell SE1 lies on the dorsal aspect of the anterior division of R2 in the subesophageal ganglion. The axon of cell SE1 projects posteriorly and exits the subesophageal ganglion in the ipsilateral connective (ipsilateral with respect to the position of cell SE1's cell body). Branches of cell SE1 extend primarily along the midline of the subesophageal ganglion, with a few branches projecting anteriorly into the circumesophageal connective and supraesophageal ganglion (SupraEG; partially drawn). Dashed lines indicate the approximate borders of the cell packets (labeled R1-R4) in the subesophageal ganglion. B. Cell SE1 connective spike. Computer-averaged connective spike (30 iterations) of cell SE1 recorded extracellularly in the ipsilateral connective between M14 and M15 (bottom trace; C(14,15)). The averaging program was triggered from a spike recorded intracellularly from the soma of cell SE1 (top trace). The slight rise in the middle of the top trace was due to the periodic occurrence of a second spike in cell SE1 that was subsequently averaged out. In this and all succeeding figures, the letter in parentheses indicates the location of the neuronal cell body, either the right (R) or left (L) side of the subesophageal ganglion or segmental ganglion, while the number indicates the segmental ganglia from which the recording is made.

2. Physiological properties of cell SE1. A. Depolarization of cell SE1 (middle trace) with a short duration (1.2 s) current pulse produces a high frequency burst of action potentials (average frequency = 62 Hz) that initiates swimming. Note that membrane potential oscillations in cell 208, a swim oscillatory interneuron (bottom trace), and bursts of action potentials in the dorsal posterior (DP) of M13 begin almost immediately after intracellularly depolarizing cell SE1. Swimming in this and all subsequent figures is indicated by rhythmic bursts of actions potential in DP nerves (Kristan et al. 1974). During swimming the membrane potential of cell SE1 depolarizes 2 - 3 mV from rest (dashed line indicates the resting membrane potential of cell SE1) and its firing frequency increases from that before the onset of swimming. Membrane potential of cell SE1 returns to its resting level when swimming ends. Vertical scale bar pertains to both the middle and bottom traces. B.Spontaneous swim episode. Preceding the onset of a spontaneous swim episode, the membrane potential and spike frequency of cell SE1 increase slowly. A general increase in motor activity in DP nerves coincides with the changes in cell SE1's activity. C. An abbreviated swim episode. Depolarization of cell SE1 (middle trace) rapidly excites cell 115, a swim oscillatory interneuron (bottom trace), and triggers one well-defined swim cycle in both cell 115 and DP(8) (top trace).

3. Stimulation of cell SE1 excites cell 204. A. Stimulation of cell SE1 (bottom trace) rapidly depolarizes and increases the firing frequency of cell 204 (middle trace). Note that swimming starts within approximately 0.5 s following stimulation of cell SE1. Vertical scale bar pertains to the bottom two traces. B. In high Mg++ and Ca++ saline, each cell SE1 spike (bottom trace) is followed by a constant-latency EPSP in cell 204 of M11 and M12 (top two traces). The 2 mV scale bar pertains to the top and middle traces.

4. A. Hyperpolarizing cell SE1 (middle trace) with approximately 1 nA of current, which abolishes all spiking activity in cell SE1, eliminates all EPSPs and spiking activity in cell 204 (bottom trace), along with all large motor unit activity in DP(8) (top trace). Upon release of cell SE1 from hyperpolarization, excitatory activity immediately returns in both cell 204 and DP(8). B. Expanded section (indicated by the dashed lines) of A. In both A and B the asterisk (*) indicates the occurrence of a stimulus artifact caused by the injection of a hyperpolarizing current pulse into cell SE1. Vertical scale bar corresponds to both cell SE1 and cell 204.

5. Relationship between the level of activity in cell SE1 and cell 204. A. Example demonstrating the response evoked in cell 204 (top trace) and in DP(12) by a 1 s current pulse injected into cell SE1 (middle trace). Diagonal lines indicate the parameter measured (depolarization area) to estimate the level of excitation in cell 204. B. Graph comparing the amount of depolarization area evoked in cell 204 as a function of cell SE1's firing frequency. In three preparations, the amount of depolarization area induced in cell 204 increases linearly over the range of cell SE1 firing frequencies employed; r = 0.92 (), 0.93 () and 0.88 (O).

6. Effect of cell SE1 on the swimming rhythm. A. Phase shift. A1. Depolarization of cell SE1(R) with a 1 s current pulse during the 5 th swim cycle increases substantially the burst duration of the 5 th swim cycle recorded extracellularly in DP(10) and the period of the next swim cycle (top trace). Note that intense stimulation of cell SE1(R) produces only a slight increase in the firing frequency of its cell pair, SE1(L). Average firing frequency of cell SE1(R) during the first and second current pulses is 66 Hz and 69 Hz, respectively. A2.. Graph of swim period as a function of swim cycle number. Data were calculated from swim episode in A1. Solid bar indicates the approximate timing of the second current pulse injected into cell SE1(R). B1. Period and burst duration of the swim cycles increase while cells SE1 are simultaneously depolarized. B2. Graph of swim period and burst duration as a function of swim cycle number. Data were calculated from swim episode in B1. Solid bar indicates when cells SE1 were depolarized. In A1 and B1 the vertical scale bar pertains to both the middle and bottom traces.

7. Properties of the connection from cell SE1 to cell 115. A. In high Mg++ and Ca++ saline, each cell SE1 spike (top trace) is followed by a constant-latency EPSPs in cell 115 (bottom trace). Vertical scale bar pertains to both traces. B. Hyperpolarizing cell SE1 (middle trace) with approximately 1 nA of injected current , which abolishes spiking activity in cell SE1, hyperpolarizes cell 115 and eliminates all EPSPs and a majority of the tonic spiking activity in cell 115 (bottom trace). In addition, large motor unit activity in DP(10) (top trace) ceases while cell SE1 is hyperpolarized. C. Expanded section (indicated by the dashed lines) of B. In both A and B the asterisk (*) indicates a stimulus artifact caused by the injection of a hyperpolarizing current pulse into cell SE1. In B and C the vertical scale bar corresponds to both cell SE1 and cell 115.

8. Monosynaptic connection from cell SE1 to cell 28, an oscillatory interneuron. A. In high Mg++ and Ca++ saline each cell SE(L) impulse (top trace) is followed by a constant-latency EPSP in the ipsilateral cell 28(L) (bottom trace). B. However, impulses in cell SE1(L) do not evoke a postsynaptic response in the contralateral cell 28(R). The records in A and B are from the same preparation and the experiments were performed sequentially. In B, the asterisk (*) denotes an impulse in cell 28, which is slightly distorted because the membrane potential of cell 28 has spontaneously hyperpolarized during the experiment.

9. Response of cells 28 and 208 to stimulation of cell SE1. Stimulation of cell SE1 inhibits cell 28 (A.), but has no observable effect on cell 208 (A.). Note that in both A and B stimulation of cell SE1 strongly activated large motor unit activity in the DP nerve.

10. Relationship between the level of activity in cell SE1 and cell 115. A. In three different preparations the amount of depolarization area evoked in cell 115 increases linearly with the firing frequency of cell SE1; r = 0.86 (), 0.92 () and 0.76 (O). B. Comparison of the effect that stimulation of one cell SE1 has on the level of excitation in the left and right cell 115. The amount of depolarization area induced in cell 115(R) and cell 115(L) in M10 by stimulation of cell SE1(R) are approximately equal at all cell SE1 firing frequencies tested; r = 0.76 () and 0.76 (O).

11. Relationship between the level of activity in cell SE1 and cell 3. A. Depolarization of cell SE1 (second trace) with increasing amounts of current (top trace) produces a corresponding increase in motor unit activity in all three DP nerves (bottom traces). The largest unit recorded extracellularly in the DP nerve is predominantly cell 3, an excitatory motor neuron that innervates the dorsal longitudinal muscles (see Methods). B. Graph showing that the firing frequency of cell 3 recorded extracellular from DP nerves at M2, M9 and M15 increases linearly with the firing frequency of cell SE1; r= 0.84 (), 0.89 () and 0.92 (O), respectively. A and B are from different preparations.

12. Cell SE1 directly excites cell 5. A. Stimulation of cell SE1(middle trace) rapidly depolarizes and increases the firing frequency of cell 5, an excitatory motor neuron to the dorsal longitudinal muscles (bottom trace), and initiates swimming (top trace). b. In high Mg++ and Ca++ saline a computer-averaged EPSP (10 iterations) is observed in cell 5 (bottom trace). The averaging program was triggered from a spike recorded intracellularly from the soma of cell SE1 (top trace). In both A and B the vertical scale bar pertains to both cell SE1 and cell 5.

13. Properties of the connection from SE1 to the L-cell. A high frequency burst of action potentials in cell SE1 (bottom trace) depolarizes L-cell (middle trace) by approximately 3 mV, but elicits only 3 action potentials in the L-cell. In contrast, 39 large, motor unit action potentials occur in DP(13) in response to cell SE1 stimulation (top trace). The large, motor units are primarily action potentials from cell 3.

14. Comparison of the effects that stimulation of cell 208 and cell SE1 have on DP nerve activity, respectively. In high Mg++ and Ca++ saline, intense stimulation of cell 208 (average firing frequency = 45.5 Hz; bottom trace) elicits 7 cell 3 action potentials in DP(13) (top trace), while stimulation of cell SE1 (average firing frequency = 19.3 Hz; third trace) elicits 28 such action potentials. In addition, stimulation of cell SE1 produces summated EPSP activity in cell 208, while cell 208 stimulation has no effect on cell SE1.

15. Summary diagram illustrating the output connections from cell SE1 to intrasegmental swim-generating neurons. Bold lines () are used to highlight the direct, excitatory connections of cell SE1, while thin lines () indicate previously identified direct, excitatory connections from cell 204 to swim oscillatory interneurons. The dashed lines denotes the fact that cell SE1 strongly excites all three swim motor neurons, but only the connection from cell SE1 to cell 5 has physiologically been shown to be direct. Note that the output connections of cell SE1 duplicate those of cell 204.