The bag cells of Aplysia as a multitransmitter system: identification of alpha bag cell peptide as a second neurotransmitter

The bag cell neurons of the marine mollusk, Aplysia, are a putative multitransmitter system that utilizes two or more peptide transmitters derived from a common precursor protein. Two putative transmitters are egg-laying hormone (ELH), a 36 amino acid peptide that induces egg laying and mediates bag cell-induced excitatory effects on certain abdominal ganglion neurons, and alpha-bag cell peptide (alpha BCP), which mimics bag cell-induced inhibition of the left upper quadrant (LUQ) neurons and the depolarization of the bag cells that occurs during the bag cell burst discharge. Alpha BCP was previously purified from bag cell extracts in three neuroactive forms: alpha BCP(1–9), a nine amino acid peptide encoded on the ELH/BCP precursor protein, and two NH2-terminal fragments, alpha BCP(1–8) and alpha BCP(1–7). Analyzing bag cell-induced inhibition of LUQ neurons, we report here that alpha BCP fulfills the main criteria for transmitter identification: stimulation of individual bag cells produces inhibition of the neurons; inhibitory activity is present in releasate collected following an elicited bag cell burst discharge in the presence of protease inhibitors; alpha BCP(1–9) and alpha BCP(1–8) are detected in the releasate in the presence of protease inhibitors; alpha BCP is rapidly inactivated after release, as indicated by the lack of detectable alpha BCP or inhibitory activity in the releasate in the absence of protease inhibitors, and by the increase in potency of the arterially perfused peptide in the presence of protease inhibitors; alpha BCP and the endogenously released transmitter produce apparently identical changes in membrane conductance; bag cell-induced inhibition is reduced or abolished following desensitization of the inhibitory response by long-term application of high concentrations of alpha BCP. The results provide additional evidence that the bag cells are a multitransmitter system and also suggest that many of the physiological properties of alpha BCP-mediated neurotransmission differ from those of ELH. First, unlike ELH, alpha BCP is rapidly inactivated after release. Second, alpha BCP(1–9) may be activated by carboxypeptidase cleavage since alpha BCP(1–8) and alpha BCP(1–7) are 30 and 10X as potent, respectively, as alpha BCP(1–9). Third, the inhibitory action of alpha BCP on its targets has a more rapid onset and a shorter time course than the excitatory actions of ELH. Thus, alpha BCP may diffuse to less distant targets than ELH and serve to regulate the more rapidly occurring neural events underlying egg-laying behavior.

The bag cell neurons of the marine mollusk, Aplysiu, are a putative multitransmitter system that utilizes two or more peptide transmitters derived from a common precursor protein. Two putative transmitters are egg-laying hormone (ELH), a 36 amino acid peptide that induces egg laying and mediates bag cellinduced excitatory effects on certain abdominal ganglion neurons, and a-bag cell peptide (aBCP), which mimics bag cell-induced inhibition of the left upper quadrant (LUQ) neurons and the depolarization of the bag cells that occurs during the bag cell burst discharge. (YBCP was previously purified from bag cell extracts in three neuroactive forms: aBCP(l-9), a nine amino acid peptide encoded on the ELH/BCP precursor protein, and two NH,-terminal fragments, aBCP( l-8) and aBCP(l -7). Analyzing bag cell-induced inhibition of LUQ neurons, we report here that (uBCP fulfills the main criteria for transmitter identification: (1) stimulation of individual bag cells produces inhibition of the neurons; (2) inhibitory activity is present in releasate collected following an elicited bag cell burst discharge in the presence of protease inhibitors; (3) aBCP(l-9) and aBCP(l-8) are detected in the releasate in the presence of protease inhibitors; (4) (rBCP is rapidly inactivated after release, as indicated by the lack of detectable (uBCP or inhibitory activity in the releasate in the absence of protease inhibitors, and by the increase in potency of the arterially perfused peptide in the presence of protease inhibitors; (5) aBCP and the endogenously released transmitter produce apparently identical changes in membrane conductance; (6) bag cell-induced inhibition is reduced or abolished following desensitization of the inhibitory response by long-term application of high concentrations of (uBCP. The results provide additional evidence that the bag cells are a multitransmitter system and also suggest that many of the physiological properties of crBCP-mediated neurotransmission differ from those of ELH. First, unlike ELH, (xBCP is rapidly inactivated after release. Second, aBCP(l-9) may be activated by carboxypeptidase cleavage since aBCP(l-8) and aBCP(l-7) are 30 and 10 x as potent, respectively, as aBCP(l-9).
Third, the inhibitory action of aBCP on its targets has a more rapid onset and a shorter time course than the excitatory actions of ELH. Thus, (uBCP may diffuse to less distant targets than ELH and serve to regulate the more rapidly occurring neural events underlying egg-laying behavior.
Studies in both vertebrates and invertebrates suggest that many types of neurons contain more than one biologically active peptide that are thought to function as neurohormones or neurotransmitters (Erichsen et al., 1982;Hunt et al., 198 1;Lundberg et al., 1982;Rothman et al., 1983a;Schultzberg et al., 1980;Watson et al., 1978). The bag cells of Apfysia are a convenient system for investigating the role of peptidergic neurons in the processing of information in the CNS. A bag cell-specific gene encodes a precursor protein that contains the sequences for egglaying hormone (ELH), cu-bag cell peptide ((rBCP), acidic peptide (AP), p-, y-, and GBCP, and several other peptides (Scheller et al., 1983a). There is strong evidence that ELH, a 4400 Da peptide, mediates two types of bag cell-induced excitatory responses in abdominal ganglion neurons (Branton et al., 1978;Mayeri and Rothman, 1982a;Mayeri et al., 1985), but does not mediate other bag cell-induced responses. (uBCP is a candidate as a second bag cell transmitter for mediating the inhibitory responses . There are neurons in other parts of the CNS that are immunoreactive for ELH (Chiu and Strumwasser,198 l), and ELH may mediate effects on other central neurons, most notably identified buccal ganglion neurons . Bag cell-induced responses on a subpopulation of abdominal ganglion neurons provide a convenient preparation for understanding how various bag cell peptides act. Bag cellinduced inhibition occurs in the left upper quadrant (LUQ) neurons, the right upper quadrant white cells R3-R14, and in L14A-C, LlO, and other identified and unidentified neurons (Mayeri et al., 1979a, b). The LUQ neurons, which were used previously as an assay system in the isolation and sequencing of olBCP, were used in the present study to further elucidate the role of olBCP. The suggestion that aBCP is a neurotransmitter is based on the following evidence: (1) Three neuroactive forms of (uBCP have been isolated from bag cell extracts: aBCP( l-9), (l-8), and (l-7) . The sequence of olBCP( l-9) is Ala-Pro-Arg-Leu-Arg-Phe-Tyr-Ser-Leu; aBCP( l-8) and olBCP(l-7) are NH,-terminal fragments lacking the COOH-terminal leucine and serylleucine residues, respectively.
(2) The three forms of aBCP mimic inhibition of LUQ neurons and depolarization of the bag cells. Although they differ in their relative potencies, the three forms otherwise produce apparently identical effects on nerve cells. (uBCP(I-7), purified from bag cell extracts, and synthetic (uBCP( l-7) have identical inhibitory actions on LUQ neurons at a given concentration . (3) L~BCP acts directly on LUQ neurons . (4) aBCP(l-9) is encoded on the bag cell gene between basic residues that represent cleavage sites (Scheller et al., 1983a). (5) Immunohistochemical studies using affinity-purified antibodies to (YBCP( l-9) show that the somata of the left and right upper quadrant neurons are surrounded by an extensive network of cuBCP( 1-9)Gmmunoreactive bag cell processes that terminate in the ganglionic sheath and are coextensive with bag cell processes immunoreactive for ELH (Pulst et al., 1985).
We present here further evidence that aBCP fulfills the major criteria for identification as a transmitter and provide additional documentation of the physiological characteristics of aBCP. We find that the physiological characteristics of (uBCP differ in important ways from those of ELH and provide a rationale for why it is functionally useful for a group of neurons to use more than one transmitter derived from a common precursor.

Materials and Methods
Electrophysiology and arterial perfusion of peptides Experiments were carried out on over 100 preparations of Aplysiu californicu obtained from Sea Life Supply (Sand City, CA). The results of each of the physiological experiments described below were obtained on at least three different preparations. Electrophysiology and arterial perfusion of peptides were performed as described previously (Mayeri and Rothman, 1982b;Mayeri et al., 1979aMayeri et al., , b, 1985. In brief, an abdominal ganglion was dissected from animals weighing 400-l 500 gm and pinned in a 1 ml capacity chamber. The preparation was steadily superfused with locally obtained seawater containing 10 mM HEPES (pH 7.6) and 1 gm/liter dextrose at a rate of 30 ml/hr. In experiments using low Ca*+ high Mg*+ bathing medium, the seawater was replaced with a solution containing 220 mM MgCl,, 1 .O mM CaCl,, 228 mM NaCl and 10 mM KCl. Experiments were performed at room temperature (19-23"C), which usually did not vary more than 1" in a single experiment. Intracellular recordings were made from up to four neurons simultaneously, using conventional methods. The bag cells were continuously monitored during all experiments by placing an extracellular recording electrode on one or both bag cell clusters.
Solutions of the various peptides to be tested were perfused through the caudal artery of the abdominal ganglion at a rate of 3 pl/min by means ofa micrometer syringe. The peptides to be tested were dissolved in filtered (0.22 urn oore size), buffered seawater. Unless noted, the perfusion medium also contained a mixture of protease inhibitors (see below). In all experiments, an aliquot of the perfusion medium without the peptide was applied before the test solution both as a control and to preincubate the ganglion in protease inhibitors. In experiments in which protease inhibitors were not used, BSA (250 &ml) was added to the perfusion medium as a carrier. Neither the protease inhibitors alone nor the BSA alone affected target neuron activity. otBCP( l-7), (l-8), and (l-9), and /3-, y-, and 8BCP were synthesized by Peninsula Laboratories (Belmont, CA). The serial perfusion experiments were performed as described previously .
In the experiments in which releasate was collected for high-pressure liquid chromatography (HPLC) analysis, the perfusion medium consisted of bathing medium containing 3 mM NaHCO, and 62 &ml each of leupeptin, antipain, diprotin A (Peninsula Laboratories), serylleucine, leucylarginine, and phenylalanylalanine (Vega Biochemicals, Tucson, M. The two groups of protease inhibitors, which were chosen to protect aBCP from degradation, are known to be effective against several of the major classes of proteases, including aminopeptidases (bacitracin, diprotin A), endopeptidases (ovoinhibitor, ovomucoid inhibitor, leupeptin, antipain), and carboxypeptidases (angiotensin-converting enzyme inhibitor and the dipeptides).

HPLC analysis of releasate
An abdominal ganglion was placed in a small sealed chamber of 200 ~1 dead volume. The caudal artery was cannulated and bag cell activity was monitored with two extracellular electrodes, one on each cluster.
The two electrodes also served as stimulating electrodes . The ganglion was perfused through the cannula and superfused by means of a second tube entering the chamber. The perfusion and superfusion rates were each 250 rllhr. Releasate (combined perfusate/ superfusate) was led out of the chamber and collected at 0°C. Control releasate was collected for 30 min prior to initiation of a bag cell burst discharge; bag cell burst releasate was collected for 30 min following burst initiation. Releasates were stored at -20°C. For analysis, control releasates from six ganglia were pooled to yield approximately 2.1 ml; bag cell burst releasates (eight bag cell bursts in six ganglia) were also pooled to yield a total volume of 2.75 ml.

Direct bag cell stimulation produces inhibition
During a bag cell burst discharge, the approximately 400 cells within each cluster fire nearly synchronously for about 20 min (Kupfermann and Kandel, 1970), producing prolonged inhibition of the LUQs (Mayeri et al., 1979a). The usual method for triggering this burst is by focal stimulation of the bag cell cluster with a short (l-2 set) train ofelectrical pulses via an extracellular electrode placed directly on the surface of either cluster (Mayeri et al., 1979a). The stimulus intensity is carefully adjusted to a level that produces a spike or depolarizing response in the bag cells without producing synaptic potentials or other responses in abdominal ganglion neurons recorded intracellularly. The presence of postsynaptic potentials one-for-one with the stimulus in other ganglion neurons indicates that the stimulus has excited other axons in the connective in addition to bag cell neurons. Although it is probable that this procedure activates only bag cells, it has not been directly demonstrated that the inhibition of the LUQs is due to bag cell activation and not to other neurons that might be activated by the extracellular stimulating electrode.
To test directly for bag cell-induced inhibition, a bag cell and an LUQ neuron were recorded from simultaneously. As shown in Figure lA, direct depolarization of the bag cell by a train of intracellular depolarizing current pulses (100 msec pulses at S/set) resulted in repetitive spike activity in the bag cell and inhibition of the LUQ neuron. The repetitive spike activity was most likely restricted to one or a few bag cells of the entire cluster because no spikes were recorded by an extracellular electrode placed on the cluster (not shown). The inhibition was characterized by hyperpolarization of the cell membrane potential, slowing of the cell's bursting pacemaker activity, and a decrease in the spike amplitude. When depolarizing current was again injected into the bag cell, as shown in Figure 1B (arrow), it was then sufficient to initiate the bag cell burst discharge in cells of both bag cell clusters; this resulted in a large and prolonged inhibition of L6. The larger amplitude and longer time course of this response presumably reflects a larger quantity of transmitter released during the burst of the entire bag cell population. Intracellular stimulation of a single bag cell after the end of a long bag cell burst (Fig. lB, right) did not result in further inhibition, apparently because transmitter is no longer released from the bag cells. It did not result from desensitization, since, in other experiments, arterially perfused (YBCP applied following the end of a bag cell burst discharge still resulted in inhibition of the LUQs. The ability to trigger a bag cell burst discharge by intracellular current injection into a single bag cell neuron is most likely the result of two processes acting together: electrical coupling between bag cells, allowing injected current to spread throughout the population (Kupfermann and Kandel, 1970), and autoex-  . Activation of a bag cell burst by this method produced the same effects in the LUQs (and other recorded ganglion neurons) as that produced by carefully controlled extracellular focal stimulation (for example, see Fig. 3, Mayeri et al., 1979a), thus confirming the results obtained by the latter method. Not every bag cell, when activated individually, produced an inhibitory effect; bag cells located closer to target cells seemed to be more effective.
Only cvBCP produces inhibition of LUQs (uBCP is one of several peptides encoded on the ELH precursor gene (Scheller et al., 1983a). Five of these peptides (ELH, AP, a-, p-, and -/BCP) have been isolated from bag cell extracts and identified (ELH, Chiu et al., 1979;AP, Scheller et al., 1983a; (Y-, @-, and ?BCP, Rothman et al., 1983aRothman et al., , 1985b. The amino acid sequences of p-and yBCP are Arg-Leu-Arg-Phe-His and Arg-Leu-Arg-Phe-Asp, respectively. cy-, /3-, and rBCP have a common sequence of four residues. Because of their structural similarity, it is important to know whether PBCP, yBCP, or the other peptides have an effect on the LUQ neurons. When a solution containing 1 PM each of ELH, AP, p-, y-, and &bag cell peptide was arterially perfused into the abdominal ganglion, there was no effect on the LUQ neurons ( Fig. 24. Furthermore, concentrations of these peptides as high as 100 PM, applied individually, had no effect on LUQs (not shown). When czBCP( l-7) was added to the solution containing the five other bag cell peptides and perfused, the LUQs were inhibited (Fig. 2B). arBCP( l-7), applied alone, had the same effect (for example, see Fig. 3B). Therefore, the inhibition of the LUQs produced by the bag cell discharge is mimicked only by ar-bag cell peptide. It was previously shown that the effects of arterially perfused (uBCP on all four LUQ cells ended within a few minutes after the end of application . Thus, the peptide Without protease inhibitors (no Z's), a concentration between 1.0 and 3.0 PM aBCP(l-7) was necessary to produce the same effect. B, Intracellular record from L3, the activity of which is plotted in A, shows that arterial perfusion of 1 PM (YBCP( l-7) without protease inhibitors did not affect the neuron's activity, whereas perfusion of the same concentration with protease inhibitors added caused a large inhibition. Arterial perfusion of a solution containing only the protease inhibitors had no effect on the cell's activity (not shown). mimics all aspects of bag cell-induced inhibition in L2 and L4 and all but the duration of inhibition in cells L3 and L6 (see Discussion). In the present study, we found that the long duration of inhibition was not mimicked by any of the three forms of cllBCP applied singly or in combination with other bag cell peptides @-, y-, or GBCP, ELH, or AP). It may be that the longer 20 set Figure 4. olBCP( l-7) is about 10 x more potent than cuBCP( l-9). A, Percentage decrease in mean spike rate from control rate of LUQ neuron L6 versus concentration of peptide in the perfusate. The spike rate at each concentration was the average of 4 min before and after application of the peptide. Solid line, czBCP( l-7); dashed line, (YBCP( l-9). The effect of tuBCP( l-9) at 0.3 PM was about the same as the effect of aBCP( l-7) at 0.03 PM, and 1 .O PM (YBCP( l-9) had less effect than 0.1 PM aBCP( l-7). B, Intracellular record from L6, whose activity is plotted in A, shows that arterial perfusion (indicated by bar) of 0.1 PM (YBCP( l-7) produced complete inhibition, whereas the same concentration of olBCP( l-9) produced only a slight decrease in spike activity.
duration response in L3 and L6 is produced by a second, as yet unidentified, transmitter released by the bag cells that acts together with (uBCP to prolong the inhibitory response. In other experiments, with pressure application of cvBCP onto L3 from a micropipette placed directly over the soma, we confirmed that the action of (uBCP is direct . This indicates that there are receptors for the transmitter on the neuron's soma. Bag Cells ASSAY LC L6 5 min Figure 5. Inhibition of LUQ neurons produced by releasate from stimulated bag cells, provided that protease inhibitors are present. An abdominal ganglion was placed in a small "source" chamber with extracellular electrodes for stimulating and recording from the bag cell cluster. A second abdominal ganglion was placed in an "assay" chamber for simultaneous intracellular recording from three LUQ neurons (L2, L6, and ~53) and a left lower quadrant neuron, LC. The source ganglion was simultaneously perfused and super-fused with a solution containing protease inhibitors, and the combined perfusate/superfusate was fed directly into the caudal artery of the assay ganglion. An electrically triggered bag cell burst in the source ganglion (arrow) resulted in prolonged inhibition of the LUQs and prolonged excitation of the LC neuron in the assay ganglion.
cxBCP is rapidly inactivated in the abdominal ganglion Inactivation of (uBCP can be demonstrated by comparing the potency of the peptide in the presence or absence of protease inhibitors in the perfusate (Fig. 3). The graph in Figure 3A plots the percentage decrease in spike rate of L3 from control levels as a function of (uBCP concentration. With protease inhibitors in the perfusate, a concentration of 0.3 PM cuBCP( l-7) produced strong inhibition of L3. Without protease inhibitors, an approximately lo-fold higher concentration of peptide was necessary to produce the same effect. Intracellular recordings from the same experiment (Fig. 3B) show the effects on L3 of arterially perfusing 1 PM concentrations of (YBCP( l-7) with and without inhibitors. In the presence of protease inhibitors, the cell was strongly inhibited, whereas in their absence, the cell was unaffected, indicating that the peptide had been rapidly inactivated. A similar result was obtained for (YBCP(I-9) (not shown); (wBCP(l-8) was not tested. These results suggest that there are membrane-bound proteases in the vascular and interstitial spaces of the ganglion that are highly effective in inactivating (YBCP. olBCP(I-7) is about IO times more potent than aBCP(l-9) Of the three forms of (uBCP, (rBCP( l-8) has been shown to be approximately 3 x more potent than czBCP( l-7) in producing inhibitory effects on the LUQs, and cuBCP(l-9) has been reported to be the least potent . We confirm here that the largest of the peptides, (YBCP( l-9), is the least potent form; it is approximately one-tenth as potent as (uBCP( l-7) (Fig. 4). The graph in Figure 4A plots the percentage decrease from control levels in spike activity in L6 versus the concentration of peptide in the perfusate. The peptides were applied in the presence of protease inhibitors. The effects of aBCP(l-9) at 0.3 and 1 .O PM were equal to or smaller than the effects of cuBCP( l-7) at lo-fold lower concentrations. At 0.1 PM (Fig. 4B, from the same experiment), L6 was only slightly inhibited by (YBCP( l-9), while at the same concentration of aBCP( l-7), the cell was nearly maximally inhibited. All three forms of (uBCP produce apparently identical inhibition.

Inhibitory activity is released during the bag cell burst
If aBCP is released from the bag cells, one might expect to detect it and its activity in releasate (i.e., perfusate plus superfusate) collected from an abdominal ganglion following a bag cell burst discharge. To test for release, an abdominal ganglion was placed in a small, sealed chamber and simultaneously perfused and superfused, and bag cell activity was monitored with an extracellular electrode. The medium exiting the chamber was fed directly into the artery of a second abdominal ganglion, located in a normal recording chamber, and the activity of neurons was monitored intracellularly. In an earlier study, following a bag cell burst in the first ganglion, releasate produced excitatory responses mediated by ELH in LB and LC cells and R 15 in the second ganglion. However, no effects were seen on target neurons that are normally inhibited (LUQ neurons) or transiently excited (Ll, Rl) by the bag cells (Mayeri et al., 1985, Fig. 6b). We reasoned that, if the inhibition and transient excitation were indeed mediated by bag cell peptides, these peptides may have been inactivated by proteolysis before leaving the ganglion.
To test this possibility, we repeated the experiment just described, but added eight protease inhibitors to the perfusate. (See Materials and Methods; either BSA or three inhibitorsbacitracin, ovomucoid, and lima bean trypsin inhibitor-were used by Mayeri et al., 1985.) During a 60 min control period before the bag cell discharge, the neurons fired at a steady rate. Six minutes after a discharge was triggered in the first ganglion (Fig. 5), cells L2, L3, and L6 in the assay ganglion were inhibited, the releasate was greater than the reponse to a bag cell burst normally seen in a single ganglion in the presence of the protease inhibitors. We attribute the greater magnitude of the responses to two factors. First, in the release experiments, the bag cell transmitters were collected and fed directly into the vascular spaces of the second ganglion. In contrast, in the single ganglion experiments, the ganglion was superfused in such a way that the released transmitters were carried by bulk flow rostrally away from the caudally located target cells, which would tend to reduce the amount of bag cell transmitter that reaches the target cells. Second, in the release experiments, the bag cell transmitters were released into space with a much smaller volume than in single-ganglion experiments. As a consequence of both factors, there may have been a higher concentration of transmitter reaching the target neurons in the serial perfusion experiments, particularly in the first few minutes, although the precise concentrations cannot be determined. The initial increase in LC firing (mediated by ELH; see Mayeri et al., 1985) was followed by a decrease in firing not normally seen in response to a bag cell burst in a single ganglion. The decrease in firing rate is attributable to an inhibitory effect of (rBCP on these cells due to a high concentration of (rBCP (see below). In the experiment shown in Figure 5, an extracellular recording showed that no bag cell activity occurred in the second ganglion. In three of six experiments, however, a bag cell burst discharge was initiated in the second ganglion within 1-3 min after onset of inhibition evoked by releasate from the first ganglion. This result is consistent with the proposal that (uBCP and/ or other bag cell peptides are excitatory autotransmitters involved in the initiation and maintenance of the burst discharge within each cluster . points for aBCP and BCB were each taken from the three-step current pulse that occurred during the peak of the inhibitory response. The data points fir the control are averages from three current pulses just prior to the inhibitory response. (The data points for the control membrane potential at 0 nA were taken at the inflection point on the upswing of the membrane potential between two successive action potentials.) aBCP and the BCB produced apparently identical changes in the currentvoltage relationship, indicating the ionic mechanisms underlying the two responses are identical. B, Intracellular record of L6, from which data in A are taken. Upper record, Inhibitory response after application of (YBCP(I-7) (bar). Lower record, Inhibitory response following an electrically triggered bag cell response (arrow; tops of spikes are not shown). gradient of n-propanol in pyridine acetate at pH 4.0. A 94 min segment of a 270 min elution profile of four test solutions is presented in Figure 6. Analysis of 320 ~1 of pooled control releasate collected before the bag cell burst discharge showed no detectable material (Fig. 6A). (The protease inhibitors in the releasate eluted prior to the segment shown.) The limit of detection was 10 ng. Analysis of pooled releasate collected following the initiation of the bag cell burst discharge showed at least 13 peaks of material (Fig. 6B). Two peaks eluted at times close to those of the orBCP( l-9) and ~YBCP( l-8) standards (Fig. 60), suggesting that these two peptides were present in the releasate. To confirm the identity of these two peaks, 128 ng of each (uBCP standard was added to the pooled releasate and analyzed on HPLC. As seen Right t / 5407 mv 2min Figure 9. Cross-desensitization of inhibitory responses produced by CYBCP and bag cell activity. When LUQ cells are desensitized by prolonged application of aBCP at high concentration, the inhibition normally produced by a bag cell burst is reduced or abolished. Simultaneous recordings from right bag cell cluster (extracellular recording) and LUQ neurons, L2 and L6 (intracellular recordings). Left, Onset of continuous arterial perfusion of 1 mM aBCP(l-7) for 30 min (beginning at lower arrow), which initially caused inhibition of the LUQ cells. Right, After 25 min, with orBCP still present in the perfusate, activity had returned to baseline, indicating that the LUQ cells had become desensitized. An electrically triggered bag cell burst (beginning at upper arrow) resulted in a lack of inhibitory response in L2 and a greatly reduced inhibitory response in L6.
in Figure 6C, the endogenous peptide and exogenous (rBCP( l-9) standard eluted as a single peak at 148 min. [The trBCP(l-9) peak occurred 2.5 min earlier in the run shown in Fig. 6B than the corresponding peaks in Fig. 6, Band D because of small variations in the elution profile over such long-lasting gradients. cuBCP(l-7) elutes 1 min earlier in Fig. 6C than in Fig. 60, and the peak eluting at 127 min in Fig. 6B elutes 2 min earlier in Fig. 6C.l The small peak of material thought to be crBCP(l-8) apparently eluted with the peak of the cuBCP( l-8) standard, but because the amount of standard was over 8-fold greater than endogenous peptide, precise co-elution could not be unequivocally demonstrated in this analysis. No peak in the pooled releasate ( Fig. 6B) eluted close to the time of the aBCP(l-7) standard, indicating that it was not present in the releasate in detectable amounts. Subsequent peaks (data not shown) were identified as AP and ELH on the basis of comigration with AP and ELH, which were purified and identified from bag cell extracts Rothman et al., 1985b). ELH and AP, but not aBCP(l-9, 1-8, nor l-7), were detected in releasate without protease inhibitors (unpublished observations). An additional comigration experiment was done to confirm the identification of (rBCP( l-8) and (l-9) in the releasate. Material corresponding to the elution times of (uBCP( l-9) and (l-8) was purified from 2.2 ml of pooled releasate from the same experiment by preparative HPLC under the same conditions as the analysis above. The purified material was again compared to standards on HPLC, but at pH 2.4 instead of pH 4.0, and under isocratic conditions instead of a gradient (Fig. 7). The material corresponding to tiBCP(l-9) eluted at the same time as an aBCP(l-9) standard, and the standard and endogenous peptides eluted together as a single peak. Similar results were obtained for aBCP(l-8) at a lower isocratic concentration of n-propanol (Fig. 7, right traces).
The amount of aBCP present in releasate was estimated by comparisons of peak heights in releasate and standards. There was 96 ng of (rBCP(l-9) and 20 ng of (uBCP(I-8) in 320 ~1 of pooled releasate (Fig. 6B), which corresponds to 103 and 21 ng, respectively, per bag cell burst. These amounts are in good agreement with the amounts recovered from purification of releasate.
These results indicate that some, if not all, (rBCP is originally cleaved from the ELH/BCP precursor and released from bag cells as a 9 amino acid peptide. (uBCP(I-8) may be released simultaneously with cuBCP( l-9). Alternatively, only (YBCP( l-9) is released and subsequently converted to (YBCP( l-8) and perhaps aBCP( l-7) by carboxypeptidase cleavage after release. Our data do not distinguish between these two possibilities (see Discussion). aBCP and the endogenous inhibitory transmitter produce the same conductance change in target neurons When arterially perfused at an appropriate concentration, (uBCP mimics the time course of onset and the magnitude of hyperpolarization that occurs in LUQ cells in response to a bag cell burst discharge . To test whether the effects of the exogenously applied peptide and the endogenously released transmitter on membrane conductance are identical, the membrane potential reponse to three levels of imposed hyperpolarizing current was measured following perfusion of aBCP(l-7) and compared to the response following a subsequent bag cell burst discharge (Fig. 8). The data from a typical (n = 7) experiment (shown in Fig. 8B) are graphed in Figure 8A. Each Z-V plot in Figure 8A is constructed from a single, threestep current pulse and represents the steady-state current-voltage characteristic of the target cell membrane at the time each current pulse was applied. The Z-V plots for aBCP application and bag cell burst were constructed from the three-step current pulse that was applied during the peak of the hyperpolarizing response. The shift in the I-Vplot away from the control levels was similar following application of cvBCP and discharge of the bag cells, suggesting that the change in the ionic conductances underlying the hyperpolarizing response is the same in both cases. As shown in Figure 8B, under control conditions, the first three current pulses produced stable changes in the membrane potential. Application of (YBCP (upper trace) and stimulation of the bag cells (lower trace) produced similar hyperpolarizing re-sponses. In each case, the current-voltage relationship plotted from the current pulse applied during the peak of the hyperpolarizing response changed from control in an identical manner, and these changes were maintained during the two subsequent current pulses. The changes from control produced by the first (-4 nA) step are the most prominent, though there are also changes for the other two steps. The change in the slope of the Z-Y curves in each case (Fig. 8A) indicates that there is an increase in membrane conductance with a reversal potential that extrapolates to near -80 mV, the equilibrium potential for potassium. The results are consistent with an earlier study indicating that bag cell-induced inhibition of LUQ cells is caused by a conductance increase to potassium ions plus a second hyperpolarizing mechanism that is relatively independent of membrane conductance (Brownell and Mayeri, 1979).
Desensitization of the bag cell-induced response by prior application of aBCP At present, an antagonist to the binding of aBCP to cell receptors has not been developed. In lieu of an antagonist, a cross-desensitization experiment was performed to test whether the inhibitory responses to the bag cell transmitter and to applied (YBCP are mediated via the same receptor or postreceptor mechanism. If the same receptor mediates both responses, then the bag cell discharge should produce no inhibition when the inhibitory response to aBCP has been desensitized by prior application of (*BCP at high concentration. CXBCP was therefore applied to the ganglion by continuous arterial perfusion at high concentration (I mM) for 30 min, initially producing inhibition of the two LUQ neurons (Fig. 9, left traces beginning at lower arrow). After 25 min of application, the spontaneous pacemaker activity of the LUQ neurons was at or near their level prior to application of the peptide, indicating that desensitization had occurred (Fig.  9, right traces). Under these conditions, an electrically triggered bag cell discharge (upper arrow) produced little or no inhibitory response, suggesting that cross-desensitization of the responses had occurred and that the applied (rBCP binds to the same receptors or acts through the same postreceptor mechanism as the transmitter released by the bag cells. It should be noted that there was a slight response to the bag cell burst that remained after desensitization (Fig. 9, right traces). In L6 (and L3), the bag cell burst produced a slight inhibition, whereas in L2 (and L4) there was actually net excitation. One interpretation of this result is that a second transmitter is involved in mediating bag cell effects on L2 and L4 and that the effects of the second transmitter become apparent only after the response to the first has become desensitized (see Discussion).

Inhibition of other neurons by (uBCP
The LUQ neurons were used in the present study as convenient cells in which to study the role of (uBCP in mediating bag cellinduced inhibition. However, a bag cell discharge inhibits other identified and unidentified neurons in the abdominal ganglion, including R2, the right upper quadrant white cells (R3-R14), the ink gland motor neurons (L14A, B, C), and LlO (Mayeri et al., 1979a, b). Unlike the other cells, LlO is transiently inhibited for 5-10 min and then undergoes prolonged excitation (Mayeri et al., 1979b), which is thought to be mediated by ELH (unpublished observations). Arterial perfusion of the abdominal ganglion with aBCP inhibited LlO (Fig. 1 OA), R2, and R3-R 14 (not shown). Recording from both identified and unidentified neurons, we found that all cells that were inhibited by the bag cells were also inhibited by application of arBCP. This is consistent with the hypothesis that CVBCP is the mediator of all bag cell-induced inhibition in the ganglion. In addition, we found that cells that were excited by the bag cells were inhibited by CXBCP. This included left lower quadrant (LLQ) neurons and R 15 (Fig. IOB). Although we did not examine the concentration CUBCP, A Second Neurotransmitter in Aplysia Bag Cells 811 dependence of the response of the neurons to cvBCP in detail, it generally paralleled that of (rBCP on LUQ neurons, and like the LUQ cells, the effects ended within a few minutes after the end of application. The inhibitory actions of L~BCP on cells not normally inhibited by a bag cell burst discharge were reduced or abolished when (uBCP was arterially perfused with ELH. As in an experiment described previously (Scheller et al., 1983b, Fig. 5) we confirmed that arterial perfusion of 1 PM each of ELH, (YBCP, and acidic peptide (AP) mimicked prolonged excitation of LLQ cells and slow inhibition of L6 and other LUQ cells. We also found that perfusion of this set of peptides mimicked burst augmentation of R15 and depolarization of the bag cells. Thus, for LLQ cells, the hyperpolarizing effect of aBCP is largely offset by the depolarizing effect of ELH in the first 2.5 min after the start of arterial perfusion, and after cvBCP is washed away, prolonged excitation by ELH continues. For L6 and other LUQ cells, ELH by itself has no effect except at very high concentrations (2 100 PM), and the response is entirely due to (uBCP. None of the three peptides, when applied singly or together, has an effect on L 1 or R 1 even at very high concentrations (I 100 PM).

Discussion
otBCP is a bag cell neurotransmitter The results presented here, together with those previously reported , provide strong evidence that aBCP is a bag cell neurotransmitter, as follows: (1) Synthesis of the transmitter by the bag cells is supported by purification of large amounts of cvBCP from bag cell extract  and by the molecular genetic data indicating that a bag cell-specific gene codes for a precursor molecule containing the amino acid sequence of aBCP(l-9) (Scheller et al., 1983a). (2) arBCP( l-9) immunoreactive terminals are in close association with LUQ cell bodies (Pulst et al., 1985). (3) aBCP mimics inhibition of L2 and L4 and all but the duration of inhibition of L3 and L6 (see below). (4) aBCP acts directly on the LUQs to produce the inhibition. (5) The changes in the current-voltage relationship in the LUQs in response to the bag cell burst, and to application of aBCP, are apparently identical, suggesting an identical ionic mechanism of action. The precise underlying ionic mechanisms need to be elucidated by further studies. (6) Inhibitory activity characteristic of aBCP is released during a bag cell discharge, and an amount of aBCP consistent with this activity is present in releasate. The amount of L~BCP present in the releasate appears to be sufficient to cause inhibition. More specifically, neurally evoked inhibition can be mimicked by a 10 min arterial perfusion of 15% of the amount of (rBCP(l-9) or 2% of the amount of aBCP( l-8) detected in the releasate per bag cell burst discharge. (7) Desensitization of the response of the LUQ neurons to applied (YBCP results in reduction or abolition of the response to a subsequently elicited bag cell burst discharge. (8) A mechanism for rapid inactivation (proteolysis) for cvBCP exists.
Thus, the key criteria for identification of a substance as a neurotransmitter have been established for cuBCP. Although blockage of the postsynaptic response to endogenous and exogenous transmitter was not attempted because of the lack of a specific antagonist, the results of the cross-desensitization experiment suggest that cvBCP may bind to the same receptor as the endogenous transmitter.
(rBCP and ELH do not mediate all aspects of bag cell-induced responses aBCP and ELH mimic all but two aspects of known bag cellinduced responses: transient excitation of Ll and R 1, for which @BCP is a candidate transmitter (Rothman et al., 1985a), and the long-lasting aspect of inhibition of L3 and L6. Arterially perfused L~BCP mimics the onset and amplitude of the inhibitory response produced in all the LUQ neurons by a bag cell burst discharge and mimics the duration of the response in L2 and L4; it does not, however, mimic the longer duration of the bag cell-induced response seen in L3 and L6 and this study). After the initiation of a bag cell burst, all four neurons show the initial hyperpolarizing response that inhibits all spike activity. L2 and L4 return to the baseline membrane potential and firing rate within l-2 min after the cessation of the bag cell burst, but L3 and L6 exhibit a prolonged inhibitory phase characterized by a sustained hyperpolarization and decrease in the rate of spontaneous spike activity of these neurons for more than 2 hr.
The simplest hypothesis for the prolonged inhibition of L3 and L6 is that another transmitter, in addition to aBCP, is released by the bag cells and that the combined effects of the two transmitters produce the prolonged inhibition. This second transmitter might act in either of two ways: (1) to potentiate the inhibitory effect of aBCP but have no effect by itself or (2) to act similarly to (YBCP but with a more prolonged action. One class of candidates for the second transmitter includes modified forms of olBCP (e.g., amidated, phosphorylated, or cleaved from the precursor in an extended form). Whatever the nature of the second transmitter, the principal transmitter on LUQs and other inhibited cells is very likely to be aBCP, because it is released in such large amounts.
Physiological characteristics of aBCP d$er from those of ELH Compared to ELH, (rBCP has a more rapid onset and a shorter duration of action and is inactivated more rapidly. After the start of the bag cell burst, the onset of inhibition in LUQ cells begins within a few seconds and reaches a maximum in about 30 set (Mayeri et al., 1979a). In contrast, bag cell-induced responses mediated by ELH are much slower, requiring l-2 min to result in spike activity in LLQs and 10-15 min to reach a maximum in R15 (Mayeri et al., 1979a, b). The rates of onset of the responses are mimicked by arterial perfusion of arBCP and ELH, applied separately or together (Branton et al., 1978;Mayeri et al., 1985;Rothman et al., 1983a;Schelleret al., 1983b). This indicates that, to a considerable degree, the more rapid action of olBCP results from a faster cellular response once the peptide has reached the cell.
The effects of arterially perfused aBCP end l-2 min after perfusion is stopped, whereas the effects of ELH persist for more than an hour (Branton et al., 1978;Mayeri et al., 1985). This suggests that the continued presence of aBCP near target cells is needed for inhibition to persist for as long as the duration of the bag cell burst, as is the case for inhibition of cells L2 and L4.
L~BCP is rapidly inactivated after release, as indicated by the lack of detectable (uBCP or inhibitory activity in releasate containing no protease inhibitors, and by the increase in potency of cvBCP when arterially perfused in the presence of protease inhibitors. It seems unlikely that a rapid-uptake system for the peptides could explain these results, since such systems are not known to be inhibited by protease inhibitors. Although direct evidence of proteolytic cleavage is needed before definitive conclusions can be made, the results suggest that there are extracellular proteases bound to connective tissue and/or cell membranes lining the interstitial and vascular spaces of the ganglion and overlying sheath. According to this hypothesis, these proteases totally inactivate crBCP before the peptide is able to diffuse into the medium surrounding the ganglion, which, in intact animals, corresponds to the general circulation. It is therefore unlikely that aBCP released from bag cells has any action on tissues outside the abdominal ganglion. This is in sharp contrast to ELH, which is relatively resistant to proteolysis, is easily detected in the bathing medium  and is Vol. 6, No. 3, Mar. 1986 released into the general circulation in apparently sufficient amounts to act on other neural tissue  and on gonadal tissue . Therefore, the relative susceptibilities of olBCP and ELH to degradation by extracellular proteolytic enzymes may be important determinants of the distance over which the two peptides can act. After release, aBCP may dljiie longer distances than conventional synaptic transmitters Although aBCP may diffuse a shorter distance than ELH before being inactivated, there are indications that it diffuses a longer distance from release sites to its targets than conventional synaptic transmitters such as ACh, at the vertebrate skeletal neuromuscular junction. First, in contrast to the discrete IPSPs produced in LUQ cells by the choline& neuron, LlO, there are no discrete PSPs occurring with each bag cell spike; instead, the inhibitory response produced by the bag cells is slow and smoothly graded, as expected should aBCP be released from distant sites. Second, the onset of neurally evoked inhibition can be mimicked by diffuse application, as is accomplished by arterial perfusion, whereas focal application of the peptide, which more nearly mimics synaptic release, produces a response that is too rapid to mimic the neurally evoked response. Third, there are large numbers of profusely branched bag cell axons within tens of microns of the nearest neurons, including those inhibited by a bag cell burst. All of these axons are almost certainly release sites for aBCP since (1) they are immunoreactive for (YBCP( l-9), (2) large amounts of otBCP are recovered from bag cell extracts , and (3) large amounts of cvBCP can be detected in the releasate when protease inhibitors are present. Thus, even with rapid proteolysis of the peptide after release, there appears to be enough (YBCP released to reach targets at long distances (as compared to conventional synaptic transmitters) before being completely inactivated.
Selective action of cvBCP Target neurons located in the rostra1 (LUQ and RUQ cells) and rostroventral (cells L12-L 14) regions of the ganglion are inhibited by bag cell activity or by application of cuBCP, but are unaffected by ELH at apparently physiological concentrations. On the other hand, the target neurons located in the caudal ganglion (LLQs and R15) are excited by bag cell activity and by application of ELH, but are inhibited when (uBCP is applied. What, then, determines the selectivity of action of (uBCP on neurons within the ganglion; i.e., which cells are ultimately inhibited by it during a bag cell discharge? There are likely to be at least two contributing factors. First, the relatively rapid inactivation of (uBCP following its release may play a role. If one assumes that ELH and (YBCP are released from bag cell axons in equal amounts, then the inhibition of rostrally located neurons and the lack of inhibition of caudally located target neurons can be accounted for by the higher density of bag cell axonal processes in the rostra1 and rostroventral portions of the ganglion (Chiu and Strumwasser, 1981;Pulst et al., 1985); cells located closest to densely packed bag cell release sites (e.g., LUQs) will therefore receive higher concentrations of aBCP than those located near less densely packed sites (e.g., LLQs and R 15) because (YBCP is more rapidly inactivated than ELH when diffusing from the release sites to the target cells. Consequently, if most or all cells have the same sensitivity to CXBCP, as seems to be the case, then cells located closest to densely packed release sites will be strongly inhibited and cells located farther away will be less inhibited or unaffected by CXBCP. A major determinant of whether a cell is inhibited by (uBCP may therefore be not the presence or absence of aBCP receptors on the target cells, but the degree to which aBCP is degraded before reaching the target cell. In contrast, the concentrations of ELH are likely to be higher than (uBCP throughout the gan-glion, and selectivity of the response to ELH is thought to be determined by the presence or absence of ELH receptors on neurons (Mayeri and Rothman, 1982a;Mayeri et al., 1985).
A second factor for determining the selectivity of the various neurons to inhibition by aBCP may be the presence or absence of ELH receptors on individual neurons and the consequent interaction between aBCP and ELH on individual target cells where both receptors are present. Cells without high-affinity ELH receptors (e.g., LUQ cells) are affected only by aBCP, and therefore inhibited. In cells with both ELH and (wBCP receptors, the effects of arterially perfused ELH and aBCP interact to produce the observed effect. For example, in LLQs, the hyperpolarizing effect of aBCP appears to sum with or be occluded by the depolarizing effect of ELH to produce net excitation. Summation of effects may also occur in R15, although more data are needed to determine whether additional interactions, such as occlusion or potentiation of the effects of one peptide by the other, also play a role.
If one assumes that bag cell peptides are uniformly processed from a single precursor in the bag cells and released in equimolar amounts to diffuse long distances to their targets , then it may eventually be possible to perfuse equimolar amounts of bag cell neurotransmitters arterially and totally reconstruct the bag cell-induced responses.
Possible activation of aBCP after release Of the three neuroactive forms of (uBCP, aBCP(l-9) was the prevalent form in the releasate (Fig. 6). aBCP(l-8) was also detected, but not (l-7). This strongly suggests that the (l-9) form is an end product of the processing of the ELH/BCP precursor protein within the bag cells before release. This conclusion is consistent with the finding that the nine-residue peptide is the prevalent form in bag cell extracts when proteolysis is carefully controlled (B. S. Rothman et al., 1985b). It is also consistent with molecular genetic data, which show that the cuBCP( l-9) amino acid sequence that is encoded on the precursor is flanked by presumed cleavage sites consisting of a single arginine residue on the amino terminus and three arginine residues on the carboxy terminus (Scheller et al., 1983a).
Although the releasate contained only one-fifth as much aBCP( l-8) as (l-9), aBCP( l-8) was 30 x as potent as (l-9) and therefore represented the major portion of inhibitory activity in the releasate. The presence of aBCP(l-8) in the releasate presumably results from carboxypeptidase A-like cleavage of the nine-residue form. There are two sites at which this cleavage might occur: either within the secretory granules before release, or in extracellular space after release. Although more data are needed to resolve the issue, we favor the latter possibility, since intragranular cleavage would require a processing step (cleavage of a carboxy-terminal leucine residue) that is not known to occur in secretory granules (Gainer et al., 1985). The proposed activation of aBCP after release is similar to the activation of the polypeptide hormone, angiotensin II, by converting enzyme, a dipeptidyl carboxypeptidase present in the circulation. The present data provide evidence for similar processing of a neuropeptide in the CNS.
If it does occur after release, the process of activation must occur simultaneously with inactivation. Detection of released aBCP appeared to require the presence of carboxypeptidase inhibitors in the perfusion medium. This is consistent with the hypothesis that aBCP( l-9) is activated to (l-8) and even (l-7) by carboxypeptidase activity and then inactivated by continued carboxypeptidase activity to (l-6) and smaller forms. Inactivation is, in addition, likely to involve proteolysis by aminoand/or endo-peptidases, since inhibitors of these two classes of proteases were required to detect released (YBCP. Since (uBCP( l-7) was not detected in the releasate, its contribution as a neurotransmitter remains to be established. However, under the experimental conditions used, the lack of aBCP(l-7) in the releasate may have occurred because the rate of its production from (l-9) and (l-8) was much slower than the rate of inactivation of (l-9) and (1-8) by endo-and/or aminopeptidases.