Synaptic and Hormonal Modulation of A Neuronal Oscillator: A Search for Molecular Mechanisms

In the last two decades the use of gastropod molluscs for neurobiological studies has become widespread. Such animals possess relatively simple central nervous systems, consisting of several ganglia communicating with each other and with the periphery by means of long connective nerves. Within each ganglion there is a complex central neuropile region containing neuronal processes, glial cells, and synaptic contacts, whereas the neuronal cell bodies are arranged around the periphery and are easily accessible to the experimenter. These cell bodies are often very large (in some cases as much as 0·5–1 mm in diameter) and many can be identified reliably from one animal to the next on the basis of their specific topographical, morphological, chemical, and electrical characteristics. In such widely studied gastropod species as the ophisto-branch marine snail Aplysia californica, and the pulmonate terrestrial snails Helix pomatia, Helix aspersa and Otala lactea, the characteristics of many individual neurones and of synaptic connections between them have been thoroughly documented (Frazier et al. 1967; Gainer, 1972; Kerkut et al. 1975). Thus, the gastropod nervous system is the tissue of choice for many electrophysiological, pharmacological, and biochemical studies of synaptic transmission and other aspects of neuronal function.

The central ganglia of Helix, Otala and Aplysia contain neurones which exhibit similar patterns of endogenous electrical activity, in which a regularly oscillating membrane potential gives rise to alternating periods of action potential bursts and interburst hyperpolarizations (Fig. 1). The depolarizing phase of the oscillation is due to the presence of a voltage and time-dependent inward current (see Meech, p. 93). which in Aplysia neurone R15 may be carried largely by Na⁺ ions (Smith, Barker & Gainer, 1975), and in Helix neurone F-1 largely by Ca²⁺ ions (Meech, p. 93; Eckert & Lux, 1976). The hyperpolarizing phase results from inactivation of the inward current, together with activation of an ion-, time-, and voltage-dependent outward current carried by K⁺ ions (Meech, p. 93; Heyer & Lux, 1976). This endogenous oscillatory behaviour can be altered by physiological and pharmacological treatments, including stimulation of appropriate presynaptic nerves, and application of hormones or neutrotransmitters. For example, stimulation of the right connective nerve or Aplysia at low stimulus intensities produces a fast excitatory synaptic potential (EPSP) in R15, leading to a net increase in the activity of the cell (Parnas & Strum-wasser, 1974; Schlapfer et al. 1974). Stimulation of the branchial nerve, or of the right connective at higher stimulus intensities, produces a complex postsynaptic response in R15, consisting of a fast EPSP followed by one or more inhibitory components (IPSP’s). The predominant phase is the inhibitory (hyperpolarizing) one, which eliminates oscillatory activity and can last seconds, minutes or even hours, depending on the stimulus parameters (Parnas & Strumwasser, 1974). A similar long-lasting IPSP can be evoked in Helix neurone F-i by stimulation of the right palliai nerve (Lambert, 1975).

The rhythmic activity of neurone F-1 (and the homologous neurone 11 in Otala) can also be modified by application of vertebrate neurohypophyseal peptides such as vasopressin and oxytocin (Barker & Gainer, 1974). At low concentrations these peptides increase the frequency and number of action potentials within each burst, and also cause a marked increase in the depth and duration of the interburst hyperpolarization. The net effect is a long-lasting stimulation of the activity of the cell, and an increase in oscillatory activity, in contrast to the effects of synaptic hyperpolarization. A similar response can be observed in these neurones following application of a crude peptide-containing extract obtained from molluscan ganglia (Ifshin, Gainer & Barker, 1975; Treistman & Levitan, 1976a; Levitan & Treistman, 1977b). That is, the molluscan nervous system itself contains a factor or factors which can alter the activity of these rhythmically oscillating neurones.

Because their endogenous oscillatory activity can be enhanced or inhibited for long periods (many seconds or minutes) by environmental manipulations, it seemed that these neurones might be particularly appropriate for studying molecular events involved in long-term modulation of oscillatory behaviour. Relatively long-lasting changes in neuronal function may be mediated by different intracellular events than the much shorter changes often associated with neurotransmitter action. In particular, rapidly reversible fluctuations in ionic conductances may reflect rapidly reversible conformational changes in membrane components which control ionic permeability. On the other hand, long-lasting conductance changes may require more stable metabolic modification of the appropriate membrane components.

One possible metabolic modification that could play such a role is cyclic nucleotide-mediated phosphorylation of membrane proteins which, directly or indirectly, control membrane permeability to various ions. It has been proposed that cyclic nucleotides may be involved in oscillatory behaviour in many systems (Rapp & Berridge, 1977), and Greengard (1976) has put forward a detailed scheme for cyclic nucleotide involvement in the nervous system in particular. Much of the early evidence in favour of this scheme is based on work with the mammalian superior cervical ganglion, in which McAfee & Greengard (1972) showed that dibutyryl cAMP could produce a long-lasting hyperpolarization similar to the IPSP resulting from pre-ganglionic stimulation. However, these results and their interpretation have been challenged by several authors (Gallagher & Schinnick-Gallagher, 1977; Dun & Karczmar, 1977), and the controversy remains unresolved. The other major evidence for a cyclic nucleotide involvement in mammalian nervous system is that of Bloom and his polleagues, who have concluded that cAMP may mediate the nor-adrenergic synaptic input from the locus coeruleus onto cerebellar Purkinje cells (for review see Bloom, 1975). In Aplysia, there are recent indications that cAMP may be involved in pre-synaptic facilitation of neurotransmitter release (Brunelli, Castellucci & Kandel, 1976; Shimahara & Tauc, 1977; Klein & Kandel, 1978). The present report describes some attempts to characterize both synaptic and hormonal modulation of the oscillatory activity in Aplysia neurone R15 and Helix neurone F-i, with respect to the ionic basis of the responses and the possible involvement of cyclic nucleotides. The methodology used for the biochemical measurements, and for intracellular recording of electrical activity, is standard and has been described in detail elsewhere (Levitan, Madsen & Barondes, 1974; Treistman & Levitan, 1976a, b; Levitan & Treistman, 1977a, b; Adams & Levitan, in preparation).

The slow membrane conductances that mediate the oscillatory activity observed in some molluscan neurones are described in detail by Meech (p. 93). We will review them briefly here since we wish to compare them with the synaptic conductances mediating the long-lasting inhibition of bursting, elicited in R15 and F-1 by stimulation of appropriate pre-synaptic nerves. Molluscan neurones appear to have slow conductance pathways for sodium (g_Na), potassium (g_K) and calcium (g_Ca). These pathways can be distinguished from the fast conductances that produce action potentials (1) by having kinetics that are two to three orders of magnitude slower, and (2) by pharmacological means (the slow sodium channel is not blocked by tetrodo-toxin; lithium can substitute for sodium in the fast channel but not the slow). During the depolarizing phase of oscillatory activity sodium and/or calcium conductance is high and potassium conductance is low, resulting in a net inward flow of current. During the hyperpolarizing phase the situation is reversed and current flows outward.

Long-lasting synaptic inhibition of oscillatory activity can be elicited in these cells by stimulation of appropriate pre-synaptic nerve trunks with single electrical pulses (Fig. 2). It appears that these responses are not monosynaptic, but the interneurones cannot be readily located (Frazier et al. 1967). The inhibitory response in Helix neurone F-i consists of a fast (< 1 s) and a slow (1-10 s) phase (Fig. 2). The outward currents which give rise to the voltage response can be readily studied under voltage clamp (Fig. 2). Ion replacement experiments indicate that the slow inhibitory phase in F-i is mediated in part by an increase ing_K and, in part, by a decrease ing_Na (Fig. 3). Since Na⁺ tends to flow down its electrochemical gradient into the cell, reducing g_Na leads to a reduction of inward current and hyperpolarization. In Helix it appears that the decrease in g_Na predominates: reversal potentials for the slow synaptic response are rarely found in normal medium (Fig. 3), but reduction of Na⁺ decreases the Na⁺ component and gives reversal potentials at or near K⁺ equilibrium (Fig. 3).

Stimulation of the pre-synaptic nerve with a train of pulses produces an inhibition in R15 that can last for periods of up to several hours (Parnas & Strumwasser, 1974). Ionic mechanisms similar to those in F-i appear to mediate this response, but in R15 it is possible to separate the Na⁺/Ca²⁺ and K⁺ components temporally (Fig. 4a). Shortly after the stimulus train a reversal potential close to the presumed potassium equilibrium potential is found (Fig. 4b) indicating that an increase in g_k predominates. With increasing time the reversal potential shifts in the negative direction (Fig. 4c) until finally no reversal potential is seen (Fig. 4d). At these later times g_K appears to be of minimal importance and the synaptic response is due to a long-lasting decrease in inward (Na⁺/Ca²⁺) current. We have not yet carried out ion replacement experiments to determine whether Ca²⁺ or Na⁺ ions carry the inward current in R15 which is turned off by synaptic stimulation.

The similarities between the conductances that mediate the inhibitory synaptic response and those that mediate the hyperpolarizing phase of oscillatory activity, namely the slow kinetics and the concomitant increase in g_K and decrease in g_Na (and/or g_Ca), led us to suspect that they might be related and might in fact be the same conductances. To test this idea we measured the synaptically induced current in R15 immediately preceding and immediately following a burst. The rationale was as follows: application of a voltage clamp just before the beginning of a burst detects net inward, current due to a high g_Na/_Ca and low g_K. Application of a clamp immediately following a burst, at the same potential as before, detects net outward current (Adams & Levitan, in preparation). If the individual conductances that are modulated by the synapse are the same as those which change during the oscillatory cycle, then activation of the synapse when g_Na/Ca is high and g_K is low (immediately before a burst) should elicit a larger net current flow than when the reverse is true (immediately after a burst). Fig. 5 shows the results of such an experiment. The net current elicited by synaptic stimuli delivered before the burst are indeed larger than those elicited after the burst and remain larger through a series of measurements (Fig. 5).

To enhance the differences in current flow before and after the burst, we augmented the bursts by injecting current and then applied synaptic stimuli before and after the augmented bursts (Fig. 6). The difference in the current flowing across the membrane pre-burst and post-burst is approximately 8 nA, almost exactly the same as the difference between the currents elicited by synaptic stimulation pre-burst and postburst (Fig. 6). That is, the difference in the synaptic response pre-burst and post-burst can be attributed entirely to the changing currents responsible for the oscillations. Furthermore, the currents at the peak of synaptic activation reach approximately the same value before and after the burst, suggesting that activation of the synapse eliminates the conductance changes that occur during the oscillatory cycle. These data suggest that oscillatory activity in these cells, and long-lasting inhibition elicited by synaptic stimulation, are indeed mediated by the same ionic conductances.

In the molluscan nervous system, physiological responses to conventional neurotransmitters, such as acetylcholine, dopamine and 5-hydroxytryptamine, are rarely observed at concentrations below about 10⁻⁶ M. In contrast, vasopressin and related neurohypophyseal hormones are remarkably potent in stimulating oscillatory activity in Helix neurone F-i (Fig. 7), and in neurone 11 in Otala (Barker & Gainer, 1974), threshold responses being observed at hormone concentrations as low as 10⁻⁹ M (Barker, Ifshin & Gainer, 1975). The high sensitivity of these neurones to vasopressin and related hormones suggested the presence in molluscan ganglia of receptors specifically sensitive to these peptides. This, together with the observation that crude acetic acid extracts of molluscan brain contain a factor that has effects qualitatively similar to those of vasopressin (Fig. 8, Ifshin, Gainer & Barker, 1975), led us to search for a vasopressin-like peptide in such extracts. At present, no conclusive proof for the existence in invertebrates of peptides resembling the neurohypophyseal hormones is available (Sawyer, 1977). Such proof would not only be of considerable evolutionary significance, but would allow us to study in detail the mechanism by which an endogenous molluscan hormone modulates oscillatory behaviour in individual neurones.

All of the known posterior pituitary peptides are stored in the pituitary in association with specific binding proteins (neurophysins) and may be isolated by precipitation as a complex with neurophysin of endogenous or exogenous origin (Acher, Light & Du Vigneaud, 1958). We have been able to purify a vasopressin-like factor from the brain of Helix pomatia, using the technique of affinity chromatography. The affinity support used was Sepharose, to which we linked bovine neurophysin. The neurophysin was isolated from bovine posterior pituitary powder (Hollenberg & Hope, 1968), and coupled to cyanogen bromide-activated Sepharose 4B (Flouret et al. 1977). A summary of the purification procedure is given in Table 1. Prior to application to a neurophysin-Sepharose affinity column, acetic acid extracts of snail brain were passed through two gel filtration columns - the first (Sephadex G-10) to remove amino acids and other small molecules (material smaller than 700 daltons was discarded), the second (Sephadex G-75) to remove material of higher molecular weight (greater than 10000 daltons). After gel filtration, the extract was applied to a 2 ml column of neurophysin-Sepharose under conditions optimal for the binding of the neuro-hypophyseal hormones (0·1 M ammonium acetate, pH 5-7). The column was extensively washed with the same buffer, which removed over 99% of the applied material (estimated by optical density at 280 nm, see Fig. 9). The column was then eluted with O’i M formic acid; a small peak of material which was fluorescamine-reactive and which possessed detectable optical density at 280 nm was eluted (Fig. 9).

This formic acid eluate has effects upon the oscillatory activity of Helix neuron F-i (Fig. 10), which are qualitatively similar to those of vasopressin or of crude snail brain extract (Figs. 7, 8). A number of controls indicate that the absorption of biologically active material on to neurophysin-Sepharose is specific. Firstly, no material which affects oscillatory activity is retained by a column of Sepharose substituted with glycine in place of neurophysin. Furthermore, the neurophysin affinity column appears to extract the biologically active material almost quantitatively from the crude extract, since, if material which was not retained by neurophysin-Sepharose is applied to the column a second time, no further biological activity is retained (Fig. 11). It could be argued that the biological activity eluted from neurophysin-Sepharose is due to leakage of vasopressin, oxytocin, or other material present as contaminants in the initial neurophysin preparation. However, no such activity is eluted from columns subjected to a ‘dummy’ elution cycle in which no sample is applied to the column (not shown).

Although the approach described above has provided strong evidence for the presence of a vasopressin-like peptide in the molluscan nervous system, conclusive proof will require the purification to homogeneity and chemical characterization of the active principle. The extremely small quantities of material available to us may render such characterization difficult, if not impossible. Estimates of the concentration of active material likely to be present in snail nervous tissue (obtained by paper electrophoresis of the purified material followed by staining with fluorescamine) indicate that less than 10 pmol are present in each animal. Several hundred micrograms of material (tens of thousands of snail ganglia) are likely to be required for protein sequence analysis by conventional techniques. However, amino acid analysis is now possible in the picomole range, and methods employing reversed-phase high-performance liquid chromatography have been reported to be capable of detecting as little as 15 pmol of vasopressin and oxytocin (Gruber et al. 1976). The growing availability of such techniques may render purification and sequence analysis feasible within the near future.

At any rate, the quantities of material available to us should be sufficient to establish whether vasopressin and the endogenous Helix peptide have a common mechanism of action in modulating oscillatory behaviour. It will be particularly interesting to ask which ionic conductances are affected during the peptide-induced stimulation of oscillatory activity and how this relates to synaptic modulation (see previous section). With respect to the biochemical mechanism of action of vasopressin and of the endogenous peptide, two possible modes of action might be envisaged by analogy with the known effects of neurohypophyseal hormones upon vertebrate tissues. The first would involve stimulation of neuronal adenylate cyclase in a manner analogous to that observed in many ion-transporting epithelia in vertebrates (Dousa, 1977; see below). A second possible mechanism of action would involve the gating of calcium influx by vasopressin, as appears to be the case in contractile tissues such as smooth muscle (Altura & Altura, 1977). Experiments to compare the effects of vasopressin and of the endogenous peptide on neuronal oscillatory activity are in progress.

It is interesting to speculate on the role that neuronal oscillations, and their modulation, may play in behaviour of the organism. The function of R15 in Aplysia and its homologues in the land snail is not clear, but it is known that these neurones share several other homologies in addition to oscillatory activity. Neurosecretory-like granules are present in their cytoplasm (Coggeshall, 1967), and the synthesis and processing of low-molecular-weight peptides, which may be neurosecretory products, has been thoroughly documented (Berry, 1975; Loh, Barker & Gainer, 1976; (Strumwasser & Wilson, 1976). In an investigation of the possible function of these neuro-secretory peptides, Kupfermann & Weiss (1976) injected crude extracts of R15 intqi Aplysia and found a small but rapid weight gain apparently due to water retention. Thus, R15 may synthesize and release an anti-diuretic peptide, and long-lasting synaptic hyperpolarization may represent feed-back inhibition of release of this peptide, from the target organ on which it acts. It is particularly significant in this regard that the branchial nerve, which contains axons of pre-synaptic neurones responsible for long-lasting hyperpolarization, also contains processes of sensory neurones from the osphradium, the organ which controls water balance in Aplysia (Jahan-Parwar, Smith & von Baumgarten, 1969). Furthermore, the oscillatory activity in R15 can be completely inhibited for long periods of time by appropriate chemical, osmotic or mechanical stimulation of the osphradium (Jahan-Parwar et al. 1969; Stinnakre & Tauc, 1966).

In view of our isolation from molluscan ganglia of a vasopressin-like peptide which alters neuronal oscillatory behaviour, it is tempting to compare this system with the hypothalamo-neurohypophyseal system in mammalian brain. The supra-optic nucleus area of the hypothalamus contains neurones which exhibit regular oscillatory activity (Gähwiler, Sandoz & Dreifuss, 1978), and which synthesize and release vasopressin, the antidiuretic hormone (Dreifuss, Harris & Tribollet, 1976). Furthermore, vasopressin itself can modulate the oscillatory behaviour of these neurones (Gähwiler, Sandoz & Dreifuss, 1978). Thus, although the osmotic stresses to which gastropods and mammals are subject clearly must differ in many respects, one may speculate that some aspects of control of water balance may be similar, and that the molluscan system may be a suitable model for studying control of release of neurohypophyseal-like peptides.

As a first approach to investigating the possibility that cAMP and/or cGMP might play a role in modulation of the endogenous oscillatory activity of neurones R15 and F-1, we examined the effects of phosphodiesterase (PDE) inhibitors on this activity. The PDE inhibitor isobutylmethylxanthine (IBMX), added to the bathing medium at concentrations sufficient to produce increases in both cAMP and cGMP (Treistman & Levitan, 1976a), markedly alters the activity of R15 (Fig. 12 A) and F-1 (not shown). Both the burst and interburst phases of the oscillatory cycle are affected, with the net effect being stimulation of the oscillatory activity. IBMX is also effective in the presence of tetrodotoxin (Levitan, 1979), which eliminates Na⁺-dependent action potentials in R15 (and other cells in the ganglion) but leaves the oscillations in membrane potential (Strumwasser, 1971). This indicates that the change in activity produced by IBMX may not require interneuronal communication, which is minimized in tetrodotoxin. The non-methylxanthine PDE inhibitor, papaverine, affects R15 and F-i in the same way as does IBMX (Treistman & Levitan, unpublished observations).

We examined the effect of cAMP and cGMP applied separately or together, extra-cellularly or intraneuronally, on the activity of R15 and F-1. Although in several experiments small transient changes were seen, cAMP and cGMP had no dramation effects on oscillatory activity resembling those produced by IBMX (Treistman & Levitan, unpublished results). One possible explanation is that the high phosphodiesterase activity in Aplysia and Helix ganglia (Levitan & Bergstrom, in preparation) destroys the cAMP and cGMP before they can reach their site of action. Accordingly, we tested the effects of several 8-position substituted derivatives of cAMP that are not broken down by phosphodiesterase (Meyer & Miller, 1974). A mixture of 8-parachlorophenylthio cAMP and 8-benzylthio cAMP, applied to the bathing medium, alters the activity of R15 (Fig. 12B) and F-i (Fig. 12C) in the same way as does IBMX (Fig. 12 A). These derivatives, like cAMP itself, are potent activators of mammalian (Meyer & Miller, 1974) and molluscan (Levitan & Bergstrom, in preparation) protein kinases. In addition, they can inhibit cAMP and cGMP phosphodiesterases, and at the concentrations used in these experiments they cause intraneuronal accumulation of both cAMP and cGMP (Levitan & Bergstrom, in preparation).

To determine whether the cAMP derivatives were acting directly on R15 and F-1, rather than on some pre-synaptic or even non-neuronal element, we injected them intraneuronally. Immediately following their injection into R15, the cell hyperpolarized and oscillatory activity was abolished for a number of minutes (Fig. 13). When oscillations resumed, the interburst hyperpolarizing phase was strongly enhanced in both amplitude and duration and the bursts themselves were similar to those observed following extracellular application of the derivatives or IBMX (Fig. 12). Certainly, then, 8-substituted cAMP derivatives can act directly on R15 to modulate its oscillatory behaviour. However, it is not clear whether the apparent difference seen following extracellular and intracellular application of the derivatives is simply a quantitative one, or whether the response is indeed qualitatively different following intraneuronal injection. The prolonged hyperpolarizations produced in the latter case (Fig. 13) were also observed in some extracellular application experiments (Levitan & Norman, in preparation).

As another approach to manipulation of intraneuronal cAMP levels we utilized the GTP derivative guanylylimidodiphosphate [Gpp(NH)p], which is not hydrolysis by GTPase. As in many other systems, Gpp(NH)p is a potent activator of Aplysia and Helix adenylate cyclase (Treistman & Levitan, 1976b; Levitan, Bergstrom & Simonet, 1978). However, in contrast to IBMX and the cAMP derivatives, it does not affect cAMP or cGMP phosphodiesterase (Treistman & Levitan, 1976b). We injected Gpp(NH)p into R15, at a concentration sufficient to fully activate adenylate cyclase. The cell rapidly hyperpolarized, and no further oscillatory activity was seen for the duration of the experiment (approximately 4 h) (Fig. 14). The cell was not damaged, since action potentials could be elicited by injection of depolarizing current, rather the neurone appeared to be ‘chemically clamped’ at about −75 mV (Treistman & Levitan, 1976b), which is close to its K+ equilibrium potential.

This complete and long-lasting abolition of oscillatory activity resembles that produced during long-lasting synaptic hyperpolarization (Fig. 2; Parnas & Strum-wasser, 1974). Furthermore, a number of putative amine neurotransmitters can activate adenylate cyclase (Levitan & Drummond, in preparation) and increase cAMP levels (Cedar & Schwartz, 1972; Levitan & Barondes, 1974; Levitan et al. 1974) in Helix and Aplysia. Unfortunately, however, the synaptic neurotransmitter mediating long-lasting hyperpolarization in R15 has not yet been identified, so it remains unclear whether these cAMP measurements are relevant to this particular synaptic response. We are currently attempting to determine whether Gpp(NH)p affects the same ionic conductances which give rise to oscillatory activity and which are altered during long-lasting synaptic hyperpolarization.

The similarity in the changes in oscillatory activity in R15 and F-1 produced on the one hand by peptides (Figs. 7, 8), and on the other hand by PDE inhibitors and cAMP derivatives (Figs. 12, 13), suggested to us that cAMP and/or cGMP might play a role in the neuronal response to peptides. This led us to examine the effect of vasopressin and oxytocin, as well as of the crude peptide-containing extract (PE) from molluscan ganglia, on ganglionic cyclic nucleotide levels. As shown in Fig. 15, the PE does indeed cause cAMP and cGMP concentrations to increase; the amounts of PE required are similar to those which modulate neuronal oscillatory activity (Fig. 8). Vasopressin and oxytocin have similar effects, although the duration and amplitude of the cAMP and cGMP increases are smaller than those produced by PE (Levitan, 1978a; Levitan & Treistman, 1977b). These cyclic nucleotide changes are not observed in neuronal cell bodies isolated following PE treatment, but rather appear to be restricted to the ganglionic neuropile (Treistman & Levitan, 1976a).

The PE also stimulates adenylate cyclase activity in a crude membrane fraction prepared from Helix and Aplysia ganglia (Fig. 16). Note that the stimulation is maintained for only a short time in this in vitro system, as in many others. Stimulation is also seen in membranes prepared from isolated Helix and Aplysia neuronal cell bodies (Table 2), indicating that these cell bodies contain receptors for the active component of the PE (Levitan, 1978b). We have not found conditions under which vasopressin (Levitan, Bergström & Simonet, 1978) or the neurophysin-purified component of the PE (Harmer & Levitan, unpublished results) stimulate Helix or Aplysia adenylate cyclase in vitro. Thus, the factor in the PE which stimulates adenyate cyclase is distinct from the vasopressin-like factor which we purify on neuro-physin-Sepharose.

One of the implications of our findings is that cyclic nucleotides are involved in both enhancement and inhibition of oscillatory activity, and it is appropriate to ask how two apparently opposing phenomena may both be mediated by cAMP. Our data suggest that the ionic conductances which are affected by synaptic stimulation are those directly involved in generation of oscillatory activity. This indicates that vasopressin (Barker & Smith, 1977) and the molluscan vasopressin-like peptide on the one hand, and the synaptic neurotransmitter on the other hand, both may modulate the same conductances. It then seems possible that cAMP alone may act on one or several of the relevant ionic conductances to cause long-lasting hyperpolarization, whereas it may act together with some other factor on the same or different conductances in such a way as to produce a net increase in oscillatory activity. Gpp(NH)p treatment, which appears to give a relatively ‘pure’ increase in cAMP, produces long-lasting hyperpolarization; in contrast, those agents which produce an increase in oscillatory activity (vertebrate peptides, crude PE, IBMX, cAMP derivatives) all cause an increase in both cAMP and cGMP. Although it is possible that cAMP and cGMP may act together, it must be emphasized that such a scheme is highly speculative at present. We cannot exclude the possibility that other factors, for example Ca²⁺, in addition to or in place of cyclic nucleotides, may play a role in long-term modulation of oscillatory behaviour. It is certainly clear that manipulation of cyclic nucleotide concentrations within individual neuronal oscillators can cause profound changes in their oscillatory activity. Definitive proof that such changes are relevant to physiological phenomena must await further investigation.