Neurohormonal Modulation of the Limulus Heart: Amine Actions on Cardiac Ganglion Neurones

In addition to classical synaptic interactions, some neurones are capable of more global, neurohormonal communication (Barker & Smith, 1979; Kupferman, 1979; Kandel, Krasne, Strumwasser & Truman, 1979; Kravitz & Treherne, 1980). This mode of communication imparts the ability to interact with many targets simultaneously and thus to modulate the activity of a collection of neurones. In this paper we study such a response.

The cardiac ganglion of Limulus generates the rhythmic beating of the tubular array of muscle cells which forms the heart. Biochemical, histological and pharmacological studies have identified a number of neurohormone candidates which modulate the activity of this system (see Watson & Augustine, 1982). Four of these endogenous amines, octopamine (OCT), dopamine (DA), epinephrine (EPI) and norepinephrine (NE), increase the rate and amplitude of the heartbeat at micromolar (or lower) concentrations (Augustine, Fetterer & Watson, 1982).

In this paper and the subsequent one (Watson, Hoshi, Colburne & Augustine, 1985) we have used electrophysiological techniques to identify the cellular sites of action of these amines. We demonstrate here that both types of neurone within the cardiac ganglion possess at least two pharmacologically distinct receptors sensitive to these amines. The combined responses of these neurones account for the ability of amines to increase heart rate. Actions of the amines upon cardiac neuromuscular transmission and cardiac muscle are responsible for increasing heartbeat amplitude (Watson et al. 1985). Thus amines simultaneously act upon multiple cellular targets to modulate the Limulus cardiac system.

Preliminary accounts of some of these results have appeared in abstracts (Fetterer & Augustine, 1977; Augustine, Watson & Fetterer, 1978; Augustine, Fetterer & Watson, 1979) and a review (Watson & Augustine, 1982).

Experiments were performed on isolated cardiac ganglia of Limulus polyphemus. The ganglia were isolated as described by Augustine et al. (1982). Ganglia were pinned to silicone resin (Sylgard 184, Dow Chemical Co.) in a 5 ml capacity Plexiglas chamber containing physiological saline. This chamber was attached to a perfusion system which permitted exchange of bathing solutions without disrupting recording electrodes. Superfusion of a dye solution indicated that exchange within the chamber was complete within 45 s of initiating flow.

Somata of cardiac ganglion neurones were impaled with a glass microelectrode filled with 3 mol 1 ¹ KC1 or 2·5 mol 1^-1 potassium-acetate (5-20 MΩ) resistance in sea water). Neurones could be identified in unstained ganglia by appropriate trans-or epi-illumination. Follower cells were much more readily located, and could be distinguished from pacemaker cells by their larger size (80-150 μm diameter; Bursey & Pax, 1970) and their characteristic ‘bursting’ electrical activity (Palese, Becker & Pax, 1970; Lang, 1971). Transmembrane potentials were recorded via a high impedence preamplifier with negative capacitance compensation and displayed on both a storage oscilloscope (Tektronix 5113) and a chart recorder (Brush 220, Gould Inst.). Burst frequency was measured manually or with a Grass 7P4 tachograph. In experiments where current was injected intracellularly, an amplifier (WPI Model no. 701) with a bridge circuit was used and a Grass S-44 stimulator was used as a current source. To verify results obtained in this way, a second current-passing microelectrode was occasionally inserted into follower cells. This method, although more reliable than the use of a bridge circuit, was not routinely used because the thick connective tissue sheath surrounding each neurone (Bursey & Pax, 1970; Palese et al. 1970) made multiple electrode insertion difficult.

In some experiments dopamine was applied on to individual neurones by iontophoresis or pressure ejection. lontophoretic application was accomplished by passing current from an isolated pulse generator (Grass S-44 with SIU-9 isolator in series with a 100MQ resistor) through a microelectrode containing lmoll^-1 dopamine-HCl (Sigma Chemical Co.). Braking current was applied, when necessary, to reduce dopamine leakage from the electrode tip. Pressure ejection was accomplished with the device described by McCaman, McKenna & Ono (1977).

Filtered natural sea water ordinarily was used as physiological saline (Augustine et al. 1982). Artificial sea water (Instant Ocean) or Limulus saline (Chao, 1933) were used occasionally and yielded similar results. Amines and other drugs were generally applied to cardiac ganglia via bath superfusion. Drug sources and methods used to prepare drug solutions were as described by Augustine et al. (1982). All solutions were applied to preparations at room temperature (18-24°C).

The Limulus cardiac ganglion consists of several hundred neurones (Bursey & Pax, 1970) which have been grouped into only two functional types (Watson & Augustine, 1982). Pacemaker cells are spontaneously active neurones responsible for generating the heartbeat rhythm, while follower cells are postsynaptic to pacemakers and are thought to be motor neurones (Palese et al. 1970; Lang, 1971). The synaptic organization of the Limulus ganglion thus appears to be simpler than that of crustacean ganglia, whose nine neurones interact in complex ways to produce the heartbeat rhythm (Maynard, 1966; Hartline, 1979; Benson & Cooke, 1984). We describe below experiments which indicate that the amines OCT, DA, EPI and NE affect the electrical properties of both types of neurones present within the Limulus cardiac ganglion. Like their effects on the Limulus heart and isolated cardiac ganglia (Augustine et al. 1982), the predominant effect of these amines was to increase the rate of neuronal activity.

It was difficult to insert microelectrodes into the small (30-60μm diameter) pacemaker cells, so that our observations on pacemaker cells were limited to three experiments where we recorded simultaneously the activity of single pacemaker and follower cells. Pacemaker neurones generate overshooting action potentials which occur once per heartbeat cycle (Lang, 1971). Bath application of DA increased the frequency of pacemaker action potentials (Fig. 1). Similar effects were produced by OCT and EPI in other experiments. DA also produced a small, transient inhibition of cardiac ganglion activity which was reflected in a slight decrease in pacemaker action potential frequency (Fig. 1). No such inhibition was caused by OCT or EPI. Between pacemaker action potentials there is a gradually depolarizing ‘pacemaker potential’ (Lang, 1971). DA, OCT and EPI increased the rate of rise of this pacemaker potential and also produced a small overall depolarization, as illustrated for DA in Fig. 1. Such effects may underlie the amine-induced increases in pacemaker cell firing rate.

Simultaneous recordings from pacemaker and follower neurones showed that application of the amines caused simultaneous increases in the frequency of both pacemaker spikes and follower bursts, so that the interval between successive events in these two neurone types always declined in parallel. Increases in the frequency of follower bursts were accompanied by decreases in their duration, as described in greater detail below.

The most prominent effect of the amines on Limulus follower neurones was an increase in the frequency of bursting activity (Fig. 2). At all concentrations OCT was most effective and NE was generally least effective (Fig. 3). Approximate ED50 values for the effects on burst frequency are: OCT, 6 × 10⁻⁸mol l⁻¹; DA, 10⁻⁷mol l⁻¹; EPI, 5 × 10⁻⁷mol¹; and NE, 10⁻⁶mol l¹.

The amine-induced increases in burst frequency were rather slow in onset,reaching a maximum over a period of 5-10 min (Fig. 4). These increases in burst frequency recovered even more slowly, requiring 20–60 min to return to control values after rinsing with amine-free saline. DA, and to a small extent NE, also produced small, transient decreases in burst rate which preceded their excitatory effects.

Coincident with the amine-induced increases in follower cell burst frequency, the duration of the slow plateau component of the burst decreased and the number of small antidromic spikes superimposed on this plateau also decreased (e.g. Fig. 2). Interburst interval, burst duration and number of spikes per burst generally decreased, in parallel, with increased amine concentrations (Fig. 5). Concentrations of OCT greater than 10⁻⁷moll⁻¹ produced a progressive increase in the number of spikes per burst. This effect may be due to the depolarizing response of follower cells to OCT which is described below.

One prominent feature of the amine responses of the intact Limulus heart is the differential sensitivity of the excitatory and transient inhibitory components to pharmacological agents (Augustine et al. 1982). In intact hearts the increase in heart rate produced by any of the four amines studied here is preferentially blocked by phentolamine, an a-adrenergic antagonist which blocks octopamine responses in arthropods (Evans, 1981), while the transient inhibition of heart rate produced by DA and NE is eliminated by metergoline, an antagonist of serotonin responses of vertebrate central neurones (Beretta, Ferrini & Glasser, 1965). These antagonists also selectively blocked the excitatory and inhibitory effects of amines on follower cell bursts (e.g. Fig. 6). These results suggest the presence of at least two distinct amine responses, one excitatory and the other inhibitory. More detailed pharmacological studies will be needed to determine whether two or more distinct receptors are involved in these responses.

Bath perfusion of the amines gradually depolarized the membrane potential of follower neurones (Fig. 7). To quantify this effect, measurements were made in four experiments (of the 49 performed) where 10⁻⁶moll⁻¹ concentrations of all four amines were sequentially applied to the same neurone. In these experiments the net depolarization was 3·3 + 2·lmV (S.E.M.) for OCT, 2·9 + l·8mV for DA, 4·4+ 2·4 mV for EPI and 2·0 + 1·6 mV for NE. DA, and occasionally NE, also produced transient increases in the hyperpolarizing phase of the follower cell burst (Fig. 7). For the same four experiments described above this transient hyperpolarization was – 3·8 + 1·1 mV for DA and – (M + OdmV for NE. These hyperpolarizations transiently reduced the number of axonal spikes appearing on the burst plateau.

The large, oscillatory changes in follower cell conductance and potential which are associated with bursting activity made it quite difficult to characterize the effects of amines upon follower cells. To eliminate bursting activity the calcium channel blocker C0CI2 (25-50 mmol 1⁻¹) was added to the bathing medium. In this condition the follower cells became roughly 15 mV less negative than they were at the most hyperpolarized potential of a burst cycle and had a somewhat higher (typically less than 50 % greater) input resistance. These changes are presumably due to blockade of the ionic conductance changes underlying bursting (Augustine, 1979). In cobalt-containing saline, cells were either quiescent or spontaneously produced the tetrodotoxin-sensitive, antidromic spikes normally found on the burst plateau (Augustine, 1979). Perfusion of the four amines in the presence of cobalt revealed at least two types of potential changes (Fig. 8). OCT slowly depolarized follower cells and increased the frequency of axonal spikes. DA produced a rapid hyperpolarization which decreased spike frequency. This response was due to a conductance increase (Fig. 9). EPI produced a depolarization substantially smaller than that generated by an equal OCT concentration, while NE had little effect on membrane potential except for an occasional small, transient hyperpolarization. These effects were quantified by measuring the potential changes produced in four cells (of the 21 examined) bathed in cobalt-containing saline to which 10⁻⁶moll⁻¹ of each amine had been added. In these experiments OCT depolarized 1·9± 0·4mV and EPI depolarized 1·0±0·5mV. DA hyperpolarized these cells by – 2·2± 0·3 mV and NE produced a slight ( – 0·7± 0·6mV) hyperpolarization. The slow, gradual depolarization which normally occurred after applying DA or NE was eliminated by cobalt. Similar effects were also observed in neurones in which bursting was suppressed by exposure to saline containing high (200mmoll⁻¹) Mg and low (2mmoll⁻¹) Ca.

We also examined some of the pharmacological properties of the hyperpolarizing DA response. This response was reduced by metergoline (Fig. 9), but was not affected by phentolamine (not shown). Ergonovine, an ergot which antagonizes DA responses in molluscan neurones (Ascher, 1972) and DA-induced increases in the Limulus heartbeat (Fetterer & Augustine, 1977), irreversibly blocked the follower cell DA response (Fig. 10). However, ergonovine also transiently hyperpolarized follower cells and is thus an agonist as well.

Because pacemaker neurones alone are thought to be responsible for generating the Limulus heart rhythm (Lang, 1971; Watson & Augustine, 1982), the amine-induced increases in heart rate and pacemaker firing frequency would be expected to be caused by direct action upon pacemaker neurones. However, because bath perfusion permits access of the amines to all ganglionic elements, it is possible that the effects are indirectly mediated. We examined this possibility (in eight experiments) by locally applying DA from micropipettes on to individual pacemaker and follower neurones, while monitoring ganglionic burst rate by recording follower cell activity.

Ejection of DA on to small cells, near their cell bodies, increased follower burst frequency (Fig. 11), while similar ejections upon various regions of follower cells never altered burst frequency or duration (not shown). Transient decreases in burst rate were never observed with local DA application on to either neurone. These results indicate that (1) DA can act directly upon pacemaker cells to increase their firing rate, and (2) this effect of DA upon pacemaker neurones can subsequently increase activity in the follower neurones. Possible causes of the transient inhibition of pacemaker firing rate are considered in the Discussion.

Local application of DA on to pacemaker neurones also decreased follower cell plateau duration and decreased the number of spikes appearing on the plateau (Fig. 11). All these effects are likely to be mediated via pacemaker neurones, because DA should not have access to follower neurones in these conditions.

The amine responses of cobalt-treated follower neurones are presumably due to direct effects upon these cells, because cobalt treatment blocks pacemaker synaptic input on to follower cells (G. Augustine & D. McCulloh, unpublished observations). This was verified (in six experiments) by local iontophoretic application of DA on to follower cells (Fig. 10, top). Hyperpolarizing responses could be consistently elicited by ejecting DA on to follower cell neurites, but applications on to follower cell somata were seldom effective. It therefore appears that amines also act directly upon follower cells and, at least for DA, sensitivity is greatly reduced at the cell body. Similar regional sensitivity to neurohormones has been reported for neurones in decapod cardiac ganglia (Cooke & Hartline, 1975; Miller, Benson & Berlind, 1984).

The amines OCT, DA, EPI and NE affect the physiological properties of Limulus cardiac ganglion pacemaker and follower neurones. Because local application of DA on to pacemaker cells (but not follower neurones) increases follower burst frequency (Fig. 11), it appears that the amine-induced increases in the frequency of pacemaker action potentials and follower cell bursts are due to amine excitation of pacemaker neurones. In addition to pacemaker-mediated effects, amines also directly elicit (at least) two distinct responses from follower neurones whose synaptic input and bursting activity have been blocked by cobalt. Thus these putative neurohormones affect the cardiac ganglion network at its two known cellular levels.

Although the excitatory effects of the amines on ganglionic activity are apparently mediated via pacemaker neurones, it is possible that the transient inhibition of the ganglion produced by DA and NE is due to an effect upon follower neurones. Experiments where DA was locally applied on to pacemaker neurones did not provide any evidence for a direct inhibition preceding the excitatory response. Such a response could have been small and undetectable, or inhibitory ‘receptors’ could be located on an area of the pacemaker neurone which was not exposed to the local applications. Alternatively these responses could be mediated via follower neurones, because (1) the same amines which transiently inhibit ganglionic activity (i.e. DA and NE) also transiently hyperpolarize follower neurones and (2) metergoline blocks both the inhibition of ganglionic activity (Fig. 6) and hyperpolarization of follower neurones (Fig. 9) which are produced by DA. DA and NE are thus either evoking pharmacologically similar responses on both pacemaker and follower cells or are inhibiting pacemaker cells via followers. Unlike follower neurones of other cardiac ganglia (Otani & Bullock, 1959; Tazaki & Cooke, 1979; Benson, 1980), hyperpolarization of single Limulus follower neurones does not decrease the burst frequency of the entire ganglion (Palese et al. 1970; G. Augustine & D. McCulloh, unpublished observations). Thus the transient inhibitory response to DA and NE is the first suggestion of a follower-pacemaker feedback pathway in Limulus.

The amine-induced decrease in the plateau component of the follower cell burst appears to be, at least in part, a secondary consequence of increased pacemaker activity, since plateau duration decreased when DA was locally applied on to pacemaker neurones (Fig. 11). This may reflect the reciprocal relationship between duration and frequency described in other cardiac ganglia (Brown, 1964; Tazaki & Cooke, 1979; Benson, 1980).

Cardiac ganglia of other arthropods have also served as model systems for the study of neurohormone action (Cooke & Sullivan, 1982). Recent studies have examined the sites of action of several putative neurohormones and have also found that neurohormones affect multiple neuronal targets within these cardiac ganglia (Miller et al. 1984; Benson, 1984; Sullivan & Miller, 1984). Our experiments complement these other studies by providing the first intracellular recordings of neurohormone actions upon pacemaker neurones (Fig. 1).

The four amines examined here have several effects on the isolated Limulus heart (Augustine et al. 1982). They increase the rate and amplitude of heart contractions, while decreasing the duration of individual contractions. In addition, DA and NE transiently decrease the rate and amplitude of the Limulus heartbeat. The results presented in this paper are sufficient to explain some of these effects in terms of cellular actions of amines.

It is likely that the amine-induced changes in heart rate are due to actions on cardiac ganglion neurones, because of the numerous parallels between the amine responses of the intact heart and the isolated cardiac ganglion. These parallels include: similarities in time course, presence of DA/NE inhibition, selective insensitivity to phentolamine (or metergoline) and relative sensitivity to amines (OCT > DA > EPI > NE). One apparent discrepancy is that EPI is less potent than DA on isolated ganglia (Fig. 3), although the two amines are approximately equipotent on the intact heart (Augustine et al. 1982). This could reflect a genuine difference between the two preparations, or could merely be a consequence of the wide variability of the Limulus EPI response (Augustine et al. 1982).

The effects of the amines on the amplitude and duration of heart contractions can be partially explained by the responses of the isolated cardiac ganglion. The decrease in contraction duration observed when amines increase heart rate is probably due to decreases in the duration of the follower cell plateau. This decrease in plateau duration is associated with a reduction in the number of spikes per burst (Figs 2, 5) which reduces the duration of cardiac neuromuscular transmission (Fig. 5 of Watson et al. 1985). The transient decreases in contraction amplitude produced by DA and NE may be due to their ability to transiently hyperpolarize follower cells; such a hyperpolarization reduces the frequency of axonal spikes (Fig. 8) and should decrease the number of junction potentials transmitted to the muscle. The increase in contraction amplitude is not simply explained by ganglionic effects: amines decrease the number of follower spikes per burst, which would probably decrease contraction amplitude. The following paper (Watson et al. 1985) demonstrates that increased contraction amplitude is due to an enhancement of cardiac neuromuscular transmission and heart muscle contractility.

In sum, we have characterized the actions of several putative neurohormones upon a simple neuronal network, and have ascribed the response of this network to actions upon its various cellular constituents. Our conclusions are compiled in Table 1.

We thank H. Levitan, S. Pierce, D. Prior and W. Watson for providing equipment and laboratory space, W. Crews and the late M. Dutterer for help in transferring equipment to Woods Hole, and J. Benson, L. Beres and W. Watson for reviewing drafts of this paper. Financial support was provided by an NSF Graduate Fellowship and University of Maryland Chesapeake Bay Research Funds to GJA and a WHO Fellowship to RHF. This is contribution no. 226 of the Tallahassee, Sopchoppy and Gulf Coast Marine Biological Association, Inc.