Ionic Channels and Hormone Release From Peptidergic Nerve Terminals

The pioneering work of Douglas and coworkers, on the mechanism of catecholamine release from the adrenal medulla (Douglas & Rubin, 1963) and neurohypophysial peptide release (Douglas & Poisner, 1964a,b) has established the basic hypothesis for the mechanism by which a molecule, enclosed in a granule (or vesicle), is released into the external medium. These authors showed that calcium ions play a major role in the process of stimulus-secretion coupling, and that neurohypophysial hormone release is triggered by depolarization of the nerve terminals. Accumulating evidence over the last 15 years suggests that the events leading to the release of neurohormones are essentially the same as those observed at synapses (Llinas, Steinberg & Walton, 1981).

The steps which link the increased calcium concentration to the release of the neurosecretory granule (NSG) contents are, as yet, unknown (for a review see Nordmann, 1983). However, biochemical and morphological studies have clearly demonstrated that release occurs by exocytosis and that endocytosis is tightly coupled to this release mechanism. It is of interest that, as early as 25 years ago, Douglas chose the neural lobe to demonstrate his stimulus-secretion coupling hypothesis. This is because, as is also true for the crustacean sinus gland, the nerve endings (i.e. the release ‘site’) can be easily separated from the transport ‘site’ (the axons) and from the site of synthesis (the neuronal cell bodies). Unfortunately, the relatively small size of neurosecretory nerve endings has not, until now, allowed direct measurement of their electrophysiological properties.

Recently, patch-clamp methods have been applied to a number of secretory systems, such as chromaffin cells (Fenwick, Marty & Neher, 1982a, b; Kidokoro, 1985), acinar cells (Maruyama & Petersen, 1982) and pituitary tumour cells (Hagiwara & Byerly, 1983), for both macroscopic current and single-channel recordings. We have reported that it is possible to apply these techniques to terminals obtained from a crustacean neurohaemal organ, the sinus gland (Lemos, Nordmann, Cooke & Stuenkel, 1986) and now extend this analysis to isolated peptidergic nerve terminals from the rat posterior neurohypophysis.

The crustacean sinus gland has provided direct evidence that propagated action potentials invading the neurosecretory endings depolarize the terminal membrane and cause release of hormones (Cooke & Stuenkel, 1985). The isolated X-organ— sinus gland system is uncomplicated by the presence of non-neurosecretory or non-peptidergic neurones or of non-neuronal cellular elements other than connective tissue and glia. A clump of somata, the X-organ, sends its axons to form a profusion of terminals abutting blood sinuses in a discrete neurohaemal organ, the sinus gland (SG). In contrast to other neurosecretory systems, where targets have not been identified and release sites are inaccessible, the SG has been well characterized and the terminals can be recorded from in situ or easily dissociated.

Secretion can be monitored while continuously recording extracellularly from the axon tract and intracellularly from soma and terminal and has been shown to be Ca dependent (Stuenkel, 1985). There is no evidence for other than peptidergic secretion (Cooke, 1981). Furthermore, the activity, function and release mechanisms of this secretory structure are analogous to those of the vertebrate neurohypophysis (Cooke & Sullivan, 1982); thus findings in one system may be comparable or applicable to the other.

It is the size of the terminals (up to 30 μm in diameter in Cardisoma) which has provided the possibility of employing intracellular recording techniques to study individual peptidergic terminal electrical responses (Cooke, 1967). SG nerve endings have special electrical properties (Fig. 2) which may have significance for secretion.

1. Terminals continue to fire impulses without accommodation in response to a maintained depolarization (Cooke, 1977).

2. Certain terminals burst spontaneously (Fig. 2; Stuenkel, 1985) or can be induced to burst by brief axonal stimulation (Cooke, 1981; Nagano & Cooke, 1983). Action potentials occurring in bursts are of particular interest because in many secretory systems, including the neurohypophysis, they have been shown to dramatically enhance secretion (Cazalis, Dayanithi & Nordmann, 1985).

3. During repetitive firing and/or bursting in terminals there is an increase in action potential duration (Fig. 2C). A similar effect is induced by tetraethylammonium (TEA) application (Nagano & Cooke, 1981).

4. There is an augmentation in terminals, as compared to axons, of the proportion of voltage-dependent calcium channels (Cooke, 1981).

Sinus gland terminals from the land crab, Cardisoma carnifex, are easily isolated (Lemos et al. 1986). In electron micrographs the isolated nerve terminals appear as circular profiles densely packed with neurosecretory granules (Fig. 1C). Dissociated neuronal terminals are morphologically very similar to non-dissociated nerve endings in situ (compare with Fig. 1A). The same morphologically distinguishable terminal types as those observed in situ (Weatherby, 1981) can be recognized among the isolated terminals (Fig. 1C). A single sinus gland yields hundreds of isolated terminals, many greater than 10μm in diameter (Fig. 1B). Dissociated nerve terminals exhibit resting potentials between –30 and –50mV (intact nerve endings have values of –50 to –60mV; Stuenkel, 1985) and sometimes have spontaneous overshooting action potentials. These and other observations reported below show that the nerve terminals, after dissociation, are viable and comparable to nerve endings in situ.

Following brief treatment with collagenase, terminals readily form >10GΩ seals with fire-polished electrodes (Lemos, Stuenkel, Nordmann & Cooke, 1985), making feasible the application of patch-clamp techniques (Hamill et al. 1981) to characterize the ionic currents and channels of the terminal membrane. ‘Terminal attached’ patches can even be studied in situ. It has been more practical, however, to prepare isolated nerve endings (Lemos et al. 1986) of a size suitable for ‘whole terminal’ recording (Fig. 1B). In the whole-cell recording, the membrane within the pipette is ruptured, giving access with a low electrical resistance to the cell (terminal) interior. This makes possible not only the study, under excellent voltage-and spaceclamp conditions, of macroscopic transmembrane currents, but also, because there is rapid equilibration between the cell interior and the electrode solution, control of the internal milieu.

The currents, in response to depolarizing voltage-clamp commands, include initial inward followed by maintained, outward current (Fig. 3, top). In some nerve terminals, the inward current has the expected properties of a Ca²⁺ current (Hagiwara & Byerly, 1981): it is not blocked by tetrodotoxin (TTX) (Fig. 3), is reduced by external application of Cd²⁺ (Fig. 3) and begins to ‘run down’ after 30–40 min. The Ca²⁺ current relaxes more slowly in certain terminals and this could explain observed differences in release pattern. Release of red-pigment-concentrating hormone (Fernlund & Joseffson, 1972) is brief and is associated with a voltage-dependent Ca²⁺ entry which inactivates (Cooke & Haylett, 1984); other peptides are released in a more prolonged manner and are probably associated with non-inactivating Ca²⁺ entry (Stuenkel, 1985). In some terminals a Cd²⁺-resistant component (Fig. 3, bottom) could be blocked by TTX. Thus there are two inward currents: one carried by Na⁺ and the other by Ca²⁺. This is expected from previous studies showing that most crab nerve endings exhibit overshooting action potentials having both Na⁺ and Ca²⁺ components (Cooke, 1977; Nagano & Cooke, 1981). Outward currents were partially reduced by the application of TEA (Fig. 3, top), consistent with the increase by TEA of action potential duration in intact terminals, and were totally blocked if the terminal was internally perfused with Cs⁺ (Fig. 3, bottom). These results indicate that the outward currents are carried by K⁺. The magnitude of these currents is directly related to the size of the terminal (compare top and bottom of Fig. 3), presumably due to differences in their membrane surface area.

Single-channel currents were recorded from isolated terminals in cell-attached and inside-out patches. Two cation channels, not previously described, have been observed (Lemos et al. 1986).

One type of channel (‘f’) shows brief (milliseconds) transitions to the open state, sometimes occurring in bursts, with long (seconds) intervals between openings (Fig. 4A). This channel, in symmetrical K⁺, has a mean slope conductance of 69 ± 3·6 pS (Fig. 4C). The single channel l/V curve (Fig. 4C) is not changed by substitution of K₂SO₄ or Na₂SO₄ for KCl on the internal face of the patch, but shows rectification when CsCl is substituted, indicating failure of the channel to pass Cs⁺ or Cl⁻. Channel currents in the presence of a salt gradient show nearly perfect selectivity for cations versus anions (Lemos et al. 1986). These observations argue against a significant anion permeability of the channels. Na⁺ goes through the channel just as easily as K⁺ since the reversal potential with equal concentrations (310 mmol l⁻¹) of KCl outside and NaCl inside remains 0 mV. Furthermore, Na⁺ concentrations between 78 and 310 mmol l⁻¹ on the inside of the patch cause this type of channel to remain open for longer periods (Fig. 4A, bottom) upon depolarization. This cation channel is observable in solutions having [Ca²⁺] buffered to 10⁻⁸mol l⁻¹ with EGTA, on the inside face.

The other type of channel (‘s’) exhibits much longer (seconds) openings (Fig. 4B) and has a mean conductance of 213 ± 6·1 pS in symmetrical K⁺ (Fig. 4D). The openings appear to occur in bursts with rapid flickering back to the closed state. It is rarely observed in solutions having low [Ca²⁺]₁, except during large voltage steps (Fig. 5), but is activated by increasing the internal free Ca²⁺ concentration above 1 μmol l⁻¹ (Lemos & Stuenkel, 1986). Higher concentrations of Ca²⁺ cause only a transient activation of channel activity. Ion substitution experiments indicate that this channel also has nearly equal permeabilities for Na⁺ and K⁺, but, unlike the f channel, allows Cs⁺ to pass through (Fig. 4D). The s channel, therefore, has the characteristics of a Ca²⁺-activated cation channel (Colquhoun, Neher, Reuter & Stevens, 1981), but exhibits a much larger slope conductance than has been observed in other cells (Yellen, 1982; Maruyama & Petersen, 1982). It could be termed the ‘maxi’ Ca²⁺-activated cation channel by analogy with Ca²⁺-activated K⁺ channels (Latorre & Miller, 1983).

The probability of opening for both the f and s channels appears not to be dependent on voltage (Lemos & Stuenkel, 1986). The P_o for the /and s channels is increased, however, by intracellular Na⁺ and Ca²⁺, respectively (Fig. 5). A scheme can be imagined in which Na⁺ and/or Ca²⁺ entry activates the channel(s) which then become inactivated by increased [Ca²⁺] after a period of seconds. These channels might be responsible for the long-lasting plateau potentials which underlie bursting (see Fig. 2) in these terminals (Cooke, 1981; Cooke & Stuenkel, 1985). Certain terminals can be induced to burst by brief axonal stimulation, and it could be the entry of Ca²⁺ and/or Na⁺ during the subsequent terminal spikes or depolarization that activates the cation channels. Repolarization may be mediated by both Ca²⁺ inactivation of the channels and a Ca-dependent K⁺ current, which appears to exist in these terminals (Nagano & Cooke, 1983). Inward current events having the characteristics of the s channel are observable in whole-terminal recordings, under voltage-clamp, with crab saline (Pantin, 1948) in the bath.

It is important to characterize the entry of Ca²⁺ into nerve terminals to understand how release is regulated. The study of Ca²⁺ channels, which in other material have generally proved to have unitary currents of <1 pA, and to show rapid transitions and bursting behaviour, is greatly facilitated by the use of ‘tip-dip’ methodology (Coronado & Latorre, 1983) and of Ba²⁺ as the charge carrier in the electrode, since it increases single Ca channel current amplitudes (Lux & Brown, 1984) and prevents inactivation (Tsien, 1983).

Three distinct types of unitary currents can be resolved in recordings (Fig. 6) from crab sinus gland nerve terminal membranes reconstituted into an exogenous lipid bilayer. The approximate single-channel conductances for the three types of channels, in symmetrical 200 mmol l⁻¹ BaCl₂, are (a) 14 pS, (b) 27 pS and (c) 43 pS, respectively. The different types of Ca channels can also be distinguished by their voltage activation and sensitivity to nifedipine derivatives. Types b and c are activated by Bay K 8644, which greatly prolongs Ca²⁺ channel open times (Nowycky, Fox & Tsien, 1985), and blocked by NS-202 (J. R. Lemos, in preparation).

The posterior pituitary gland (neural lobe) is a convenient model for the study of the release of peptide neurohormones since it contains the distal parts of oxytocin-and vasopressin-containing neurones. The hormones are synthesized, as precursors, in the cell bodies, packaged into neurosecretory granules and transported down their axons to the posterior pituitary where they are released. The neural lobe (NL) contains, on average, 3·4×10⁷ nerve terminals. Each cell body gives rise to an average of 1·8× 10³ nerve terminals, whose mean diameter is about 2μm although much larger (approx. 8–10μm) endings can be observed. Arginine-vasopressin (AVP) and oxytocin (OT) are the two major peptides found in the neurohypophysial secretory granules (molar ratio >1000 compared with other molecules).

Biochemical and morphological knowledge of the hypothalamo-neurohypophysial complex provides insight into the mechanisms of synthesis, transport, release and storage of the neuropeptides synthesized by the supraoptic and paraventricular nuclei. In addition, much is known about the electrophysiology of these magno-cellular neurones (for a review, see Poulain & Wakerley, 1982). Oxytocin neurones are characterized by their synchronous high frequency discharge during suckling, which leads to the pulsatile release of OT and subsequent milk ejection. Vasopressin neurones are characterized by their asynchronous phasic activity (bursting) during prolonged, ‘trickle’ AVP release and regulation of water balance. In both cases it is the clustering of action potentials which facilitates hormone release (Cazalis et al. 1985), albeit with different time courses.

To analyse the mechanisms by which membrane depolarization is linked to the release of AVP and OT, we have developed a technique for isolating neural lobe terminals (Nordmann, Desmazes & Georgescault, 1982; D. Brethes, G. Dayanithi, L. Letellier & J. J. Nordmann, in preparation). These neurosecretosomes consist almost entirely of isolated nerve endings as judged by electron microscopy (Fig. 7) or by immunocytochemistry of the neuropeptides and neurophysins (J. J. Nordmann, unpublished observation).

The following steps in stimulus—secretion coupling, in the neural lobe, have been indirectly demonstrated (for a review see Nordmann, 1983). The arrival of action potentials induces the depolarization of the nerve terminals which have been shown to have in their plasma membrane both Na (Nordmann & Dyball, 1978) and Ca channels (Dreifuss, Grau & Nordmann, 1973; Russell & Thorn, 1974). The electrically induced depolarization promotes the entry of calcium into the nerve terminals which then triggers, by an unknown mechanism, the release of the NSG contents.

Using a fluorescent probe to measure changes of membrane potential, we have shown that the neurosecretosomes can be depolarized with increasing external potassium concentration or with agents such as veratridine (Nordmann et al. 1982). The depolarization of the isolated nerve terminals is correlated with the release of AVP, OT and neurophysins. Depolarization-induced hormone release requires external calcium and is abolished by agents known to block Ca²⁺ channels (Co²⁺, Mn²⁺, D600, Cd²⁺, Gd²⁺, nitrendipine and nicardipine; M. Cazalis, G. Dayanithi & J. J. Nordmann, in preparation).

The entry of calcium and its homeostasis in the nerve terminals has also been studied (Douglas & Poisner, 1964b; Nordmann, 1976; Russell & Thorn, 1974). Only recently has it been possible to show, using vasopressinergic activity as the stimulus, that depolarization of the nerve endings is associated with an increased ionized calcium concentration in their cytoplasm (Fig. 8). Similarly, we have shown that this increase, measured with Fura-2, is abolished by Ca²⁺ channel blockers (D. Brethes, G. Dayanithi, L. Letellier & J. J. Nordmann, in preparation). As in the intact neural lobe, calcium can be replaced by strontium as a trigger for hormone release.

Using detergent-permeabilized isolated nerve endings, hormone release can be observed (Fig. 9) with calcium concentrations in the micromolar range (Bicknell, Cazalis, Dayanithi & Nordmann, 1985; M. Cazalis, G. Dayanithi & J. J. Nordmann, in preparation). This is in contrast with ‘normal’ isolated preparations which release neuropeptides only when depolarized in the presence of millimolar concentrations of external calcium. Furthermore, the release mechanism is greatly potentiated by the presence of ATP. Other nucleotides have little or no effect on the secretion of AVP and OT. The permeabilized preparation is extremely useful for studying the steps which are hypothesized to link the entry of calcium to the exocytosis of the NSGs. We have found that trifluoroperazine, at high concentrations, only partially inhibits the calcium-dependent hormone release in this preparation (Fig. 10). A phorbol ester (TPA), on the other hand, stimulates hormone release from the permeabilized neurosecretosomes at low concentrations (M. Cazalis, G. Dayanithi & J. J. Nordmann, in preparation). These preliminary results suggest that C-kinase might be involved in the secretory response.

The rat neurosecretosomes also readily form giga-ohm seals with fire-polished electrodes, making feasible the application of patch-clamp techniques to characterize the ionic channels of the terminal membrane. Single-channel currents from isolated rat NL nerve terminal inside-out patches show both fast (J. R. Lemos & J. J. Nordmann, in preparation) and slow channel types (Fig. 11). The slow channel has a slope conductance of 221 pS. Both Na⁺ and K⁺ permeate through the channel, but anions do not, indicating that this is a cation channel. Openings are only seen with internal Ca²⁺ concentrations above lμmol l⁻¹. Thus this neural lobe Ca²⁺-activated cation channel seems comparable to the 5 channel found in SG nerve terminals (see Fig. 4). Since the f and s channels have not been found in other neurones or neuronal structures (such as X-organ somata; J. R. Lemos & B. Haylett, unpublished observations), they may be unique to peptidergic nerve terminals. More experiments, however, are necessary to establish this conclusion.

It is possible to separate and isolate neurosecretory granules on iso-osmotic gradients (Nordmann, Louis & Morris, 1979). The purity of such preparations makes possible the analysis of neurosecretory granule ion fluxes. We have been able to reconstitute neurosecretory granule membrane proteins and study their activity using tip-dip methods. Preliminary evidence indicates that the NSG membranes do not contain Ca channels but do exhibit at least two other ionic channels (J. R. Lemos & J. J. Nordmann, in preparation). One channel type is permeable to K⁺ and activated by internal Ca²⁺. It appears to be similar to a channel recently reported in secretory granule membrane from pituitary glands studied in planar bilayers (Stanley, Ehrenstein & Russell, 1986). It has been suggested that entry of Ca²⁺ into pituitary nerve endings activates this channel and that elevation of [K⁺] in the NSG causes subsequent entry of anions through hypothesized anion channel(s). We have observed such an anion channel (Fig. 12) in membranes prepared from isolated neural lobe NSGs. In asymmetrical salt gradients the channel appears to be permeable only to anions, such as Cl⁻, and not to cations. This NSG anion channel, in symmetrical 200 mmol l⁻¹ KCl, has a slope conductance of about 280 pS and opens even in the presence of only 10⁻⁸mol l⁻¹ free internal [Ca²⁺]. The existence of these two channels in NSG membrane lends support to the theory (Cohen, Akabas & Finkelstein, 1982; Ehrenstein & Stanley, 1986) that Ca²⁺ entry could lead to swelling of the NSG and thus promote fusion with the plasma membrane and release of the NSG contents.

Although there is considerable evidence that depolarization of nerve cell terminals leads to the entry of Ca²⁺ and to the secretion of neurohormones and neurotransmitters, the details of how ionic currents control the release of neuroactive substances from nerve terminals remain undetermined. This study presents two preparations in which stimulus-secretion coupling can be directly analysed. Much is already known about neurohormone release from both the neural lobe and the sinus gland, and the electrical activity of individual terminals has been well characterized. Patch-clamping of the isolated neurosecretosomes has now allowed the elucidation of some of the macroscopic currents underlying nerve terminal voltage responses. The finding of two previously undescribed cation channels, possibly unique to nerve terminals, may have particular importance for bursting activity. Such patterns have been shown to have a facilitatory effect on calcium entry and hence on hormone release. We do not know exactly at which step of the stimulus-secretion coupling mechanism facilitation occurs, but recent data (Cazalis et al. 1985) suggest that the phasic pattern of discharge has some effects on Ca²⁺ channels. Plausible explanations are that it increases either (1) the number of channels activated at a given time or (2) the time for a channel to inactivate. The ability to reconstitute Ca²⁺ and NSG channels from nerve endings should allow us to answer such questions. Furthermore, the abundance of granules, known to contain peptide hormones, together with histological and biochemical evidence for release by exocytosis, suggest that membrane capacitance measurements (Neher & Marty, 1982) could be utilized in these preparations to study directly the events coupling depolarization of the nerve terminal membrane with the release of peptide hormones from neurosecretory granules.

We wish to thank D. Brethes, M. Cazalis, I. Cooke, D. Dagan, G. Dayanithi, L. Letellier and E. Stuenkel who have greatly contributed to this work. Supported by NS15433 and BNS81-07289 grants to I. M. Cooke, NATO and Fondation de la Recherche Médicale grants to JJN, National Service Award NS07072to JRL, and Biomedical Research Support grant RR05528.