Control of Secretion in Anterior Pituitary Cells - Linking ion Channels, Messengers and Exocytosis

Cells of the anterior pituitary gland are the source of important physiological effector hormones. This endocrine tissue secretes growth hormone (GH), prolactin (PRL), luteinizing hormone (LH), follicle-stimulating hormone (FSH), thyroid-stimulating hormone (TSH), adrenocorticotrophic hormone (ACTH), beta-lipotropin, beta-endorphin and melanocyte-stimulating hormone (MSH). This wide spectrum of hormones in turn controls such important bodily functions as growth, reproduction and fertility, milk production, thyroid hormone output, and maternal behaviour. In most cases, the ‘one cell -one hormone’ concept seems to apply.

It is clear that most pituitary cells have receptors for small-peptide-releasing factors which are synthesized by and secreted from discrete groups of cells in the hypothalamus, ultimately finding their way to the anterior pituitary gland through a specialized portal circulation. Many of these cells also possess receptors for neurotransmitter substances such as γ-aminobutyric acid (GABA), acetylcholine, noradrenaline and dopamine, which appear to modulate hormone secretion. Given the heterogeneity of the anterior pituitary gland, how should one set about studying the secretion of a particular hormone from a particular cell type?

One approach was provided by Tashjian’s group working, not on normal cells, but on clonal cells derived from a rat pituitary cell tumour (Tashjian et al. 1968). This work produced the most widely studied group of pituitary tumour cell lines available, the GH3 and GH4 clonal lines and their various sub-clones. Work on these has provided much detail about basic aspects of endocrine cells but they are poor models for studying specific pituitary cell types. Compared with normal anterior pituitary cells secreting either prolactin or growth hormone, most GH3/GH4 lines secrete both hormones and at abnormally high rates; the cells are virtually totally degranulated; they lack receptors for dopamine (the tonic inhibitory transmitter implicated in prolactin secretion) and GRF (growth-hormone-releasing factor –the peptide promoting growth hormone secretion from somatotrophs). There was clearly a strong case for wishing to study the properties of single, identified cells associated with secretion of a particular hormone.

Most groups, including our own, have relied on separation and enrichment of cell populations on the basis of cell density. For prolactin and growth hormone cells, we have used the density centrifugation technique employing isotonic Percoll gradients (Mason & Ingram, 1986). Another approach has been to identify the cells using immunological techniques such as the reverse haemolytic plaque assay (DeRiemer & Sakmann, 1986; Lingle, Sombati & Freeman, 1986), which can detect hormone secretion from a single cell, or by specific fluorescence labelling of cells (Marchetti, Childs & Brown, 1987). More recently, anterior pituitary cells have also been separated using fluorescence-activated cell sorting (FACS) with antibodies or cholera toxin as a probe (Lewis, Goodman, St. John & Barker, 1988). In the case of the gonadotrophin-secreting cells, we have employed a different approach altogether, taking advantage of the observation that the ovine pars tuberalis contains gonadotrophs as the only hormone-secreting cell type (Mason & Waring, 1985). However, our approaches provide enrichment, not purification, and we have further employed post-recording staining with highly specific antibodies to ensure certainty of identification for most single-cell work.

Our interest lies broadly in the mechanisms that link the stimulus of a peptide or a neurotransmitter to eventual secretion of hormone from anterior pituitary cells. The aim of this article is to explore the current knowledge about this complex pathway, and how it might be modulated within the cell. At present, we believe this to involve a sequence which may be approximated by: (1) binding of a peptide or a neurotransmitter (a ligand) to its specific plasma membrane receptor on the cell; (2) activation of either ligand-ion channel complexes or intracellular metabolic cascades, the latter producing new sets of molecules (intracellular messengers); (3) interaction of these molecules with membrane ionic channels to cause currents to flow across the plasma membrane and possibly also mobilize ionic current flow within the cell; (4) release of hormone when secretory granules fuse with the plasma membrane in an exocytotic event.

Within this framework, the specific questions we would like to address in this article concern some of the fundamental properties of membrane events responsible for stimulus-secretion coupling. Is hormone secretion a voltage-dependent process, or is it linked to membrane potential in any way? Are electrical events, such as subthreshold fluctuations or action potential firing, important for hormone secretion? How is electrical activity modulated by peptide-releasing factors? Can evidence be found for activation of intracellular messengers that may influence ionic current flow? How are these changes linked ultimately to hormone secretion? Is Ca²⁺ vital for hormone secretion from these cells, or might there be other substances essential for exocytosis? Hence do changes in intracellular Ca²⁺ concentration occur when cells are stimulated with specific releasing factors? What is the source of changes in cell calcium concentration? Do all types of cell behave in the same way?

Ion channels are thought to be integral protein units within the lipid bilayer of a membrane which can in simple terms be regarded as a gate that is either open or closed. The mechanism by which the channel gate is opened or closed depends upon the type of channel. For convenience the channels that have been found in anterior pituitary cells can be classified into a number of different subtypes. These are listed below, and discussed in more detail later in the text.

channels that are activated by a change in membrane potential; e.g. the tetrodotoxin TTX-sensitive Na⁺ channels and TTX-insensitive Ca²⁺ and K⁺ channels seen in many anterior pituitary cells.

channels that are directly coupled to a receptor; e.g. the GABA-activated Cl⁻ channel seen in bovine lactotrophs (Inenaga & Mason, 1987) and melanotrophs (Kehl, Hughes & McBurney, 1987; Taleb et al. 1987).

channels that are gated by a reaction with a molecule whose availability depends on another cellular event such as receptor occupancy or changes in the cellular metabolism; e.g. the gonadotrophin-releasing hormone (GnRH)-activated Ca²⁺ channel in gonadotrophs (Mason & Waring, 1986). Meldolesi & Pozzan (1987) have referred to these as second-messenger-operated calcium channels (SMOCs).

channels that are activated by changes in membrane potential, but are also modified by chemical messengers; e.g. the calcium-dependent K⁺ channel which is opened by calcium, and closed by cyclic AMP.

Work from a number of laboratories has recently shown that anterior pituitary cells contain a variety of voltage-activated ion channels, many bearing similarities to those found in neurones. Kidokoro (1975) showed that GH3 cells exhibited spontaneous action potentials, and that these were generated partially by Ca²⁺ flux through voltage-activated channels. Thyrotrophin-releasing hormone (TRH), at concentrations known to stimulate PRL release, increased the firing rate in these cells, thus allowing (by implication) more Ca²⁺ into the cell. These data suggest an analogous situation to that found in neurones where firing frequency governs secretion of neurotransmitter. This therefore represented a possible mechanism for the control of hormone release. Work with normal anterior pituitary cells has shown that they differ to some extent from GH3 cells, and this work will be discussed below.

Using tight-seal, whole-cell recording techniques, normal anterior pituitary cells have been shown to have resting membrane potentials more negative than –60mV and a high input resistance (usually 2 –20 G Ω). Normal lactotrophs fire action potentials spontaneously, but at a frequency far lower than that seen in neurones (Taraskevich & Douglas, 1979; Cobbett, Ingram & Mason, 1987a), usually about 2 –3 spikes s⁻¹. In both somatotrophs (Mason & Rawlings, 1988) and gonadotrophs (Mason & Sikdar, 1988), we in common with others (Israel, Denef & Vincent, 1986) have not observed any significant level of spontaneous action potentials, although low-frequency, current-evoked action potentials may be observed (Fig. 1B). In all pituitary cells examined, this is because the Na⁺ currents (see below) require 1 –2 s to recover from inactivation following activation. In neoplastic (GH3) rat pituitary cells, action potential firing is also low-frequency -for similar reasons -and has a significant Ca²⁺ component as well as being unaffected by the voltage-activated sodium channel blocker tetrodotoxin (TTX). Normal pituitary cells fire action potentials which are TTX-sensitive, but which cannot be supported by Ca²⁺ alone (Cobbett et al. 1987a; Mason & Sikdar, 1988). Ironically, despite the presence of these action potentials, release of GH, LH and PRL -either basal or evoked by releasing factors -is not affected by TTX, suggesting the Na⁺ action potentials may not be important.

Voltage-clamp studies show that anterior pituitary cells possess voltage-activated Na⁺, Ca²⁺ and K⁺ channels, and that the currents carried by these channels underlie the action potential. In lactotrophs, for example, a rapidly activating and inactivating, TTX-sensitive Na⁺ current can be reliably evoked, similar in some respects to that recorded in neurones (Fig. 2A). A much slower inactivating, TTX-insensitive inward current can also be recorded when the external medium contains Ca²⁺ or Ba²⁺ (Fig. 2B). Both inward currents are maximal at about –10 to 0mV (Fig. 2C).

Under current-clamp, spontaneous depolarizing and hyperpolarizing potential fluctuations of 2 –20mV are often recorded. Since these cells have no synaptic input in vitro or in vivo, these are proposed to be due to brief, random, spontaneous openings of single ion channels. Because of the cell’s high input resistance, spontaneous openings of single ion channels can produce large changes in membrane potential, and depolarizing fluctuations could activate voltage-activated Ca²⁺ channels. Indeed depolarizing voltage fluctuations could be the result of spontaneous opening of voltage-activated Ca²⁺ channels themselves, or some other channel. For instance, a single channel inward current of 1pA (10 ⁻¹² A) flowing across a resistance typical of whole-cell recorded pituitary cells, say 10 G Ω, would result in a depolarizing voltage of about 10 mV.

In anterior pituitary cells there are two voltage-activated Ca²⁺ channels which may be differentiated by their kinetic properties, and voltage sensitivities (Fig. 3) (DeRiemer & Sakmann, 1986; Lingle et al. 1986; Marchetti et al. 1987; Cobbett, Ingram & Mason, 1987b; Kehl, 1987). One has a low threshold (usually about –30mV), activates and inactivates rapidly, and can only be reactivated by returning the membrane potential to the original holding potential (–60 mV or more negative). The other is activated by voltage pulses to more positive potentials (above –10 mV) and is mainly non-inactivating. The ion channels carrying these currents could be opened during an action potential and thus could mediate Ca²⁺ influx required to support secretion. Because the low-threshold channels are probably substantially inactivated at the resting potential (Cobbett et al. 1987b) it may be that only the higher-threshold current could contribute to the action potential.

The effects of peptide action and of channel blockers in lactotrophs appear to bear some correlation with membrane potential. Application of Mn²⁺ or Co²⁺ to single lactotrophs recorded intracellularly hyperpolarizes the membrane potential (Fig. 4A,B), and TRH application leads to a complex chain of events: initial hyperpolarization followed by a small to negligible depolarization (2 –5 mV) and low-frequency firing which is also blocked by Co²⁺ (Fig. 4B). Dopamine, the tonic inhibitory transmitter involved in PRL secretion, strongly depolarizes the membrane potential (Fig. 4C) and suppresses PRL release. Unfortunately, Co²⁺, Mn²⁺ and most divalent cation channel blockers also suppress voltage-activated Na⁺ currents, as they do in other tissues. Nevertheless, modulation of membrane potential is clearly correlated with hormone secretion.

In GH3 cells, at least, recent work with calcium-sensitive dyes has confirmed that intracellular Ca²⁺ concentration may rise during an evoked action potential or an evoked train of action potentials. But given the failure of a number of groups to observe action potentials in normal cell types, other than lactotrophs, it must be questioned whether the action potential is a ubiquitous mechanism for rises in intracellular calcium level. This point will be discussed in more detail below. In particular, can release be stimulated by a mechanism independent of action potentials, such as subthreshold channel openings and fluctuations in membrane potential close to the resting state?

The evidence for the involvement of hypophysiotrophic-peptide-activated plasma membrane ion channels comes from different studies. Altering membrane ionic flux either by using classical ion channel blockers -such as those acting upon the Ca²⁺ channel -or by changing the composition of the external medium can alter the hormone release response to releasing factors (Vale, Burgus & Guillemin, 1967 ). Also, the use of electrophysiological techniques has enabled us to study directly the ionic flow through individual ion channels, and these techniques have given irrefutable evidence that the binding of a number of releasing (or release-inhibiting) factors can change channel activity (Mason & Waring, 1985). Whether the altered state of these channels, and the subsequent changes in ionic flux, are critical for secretion is still not fully resolved.

An important role for extracellular Ca²⁺ in the regulation of anterior pituitary hormone secretion was first proposed by Vale and colleagues (Vale et al. 1967). They showed that the action of TRH on TSH release from the rat anterior pituitary was dependent upon extracellular calcium, and that high extracellular K⁺ concentration enhanced TSH release in the presence of calcium. Depolarizing concentrations of K⁺ were later shown to stimulate hormone release from other anterior pituitary cells (Kraicer et al. 1969; MacLeod & Fontham, 1970; Parsons, 1970; Wakabayashi, Kamberi & McCann, 1969). These findings suggested that voltage-dependent influx of Ca²⁺ is one method of linking stimulus to secretion in anterior pituitary cells. Recently Schlegel and colleagues have shown a clear rise in [Ca²⁺]_i following action potential firing in GH3/B6 cells (Schlegel et al. 1987). A 10 ms action potential raised the intracellular calcium to a level sufficient to stimulate hormone secretion. However, there is some evidence to suggest that in other cell types certain factors may cause hormone release in the absence of changes in membrane potential.

GRF is a potent stimulator of GH release from somatotrophs (Guillemin et al. 1982). However, intracellular recording has shown that although GRF applied to somatotrophs causes an increase in membrane voltage noise, no absolute change in membrane potential is observed. This noise is blocked by 4 mmol l ⁻¹ Co²⁺ (Mason et al. 1986) suggesting that it is caused directly or indirectly by an increase in the membrane Ca²⁺ permeability. A similar action is seen when ovine gonadotrophs respond to GnRH (Mason & Waring, 1985). Release studies on bovine and ovine cells have shown that GRF-, TRH-and GnRH-stimulated release of GH, PRL and LH, respectively, persists in the presence of doses of TTX capable of blocking action potential firing in these cells (see Fig. 1A) (Mason & Rawlings, 1988; Cobbett et al. 1987a; Mason & Sikdar, 1988). In these situations, therefore, Na⁺ action potential firing is not necessary for hormone release caused by releasing factors, and because single, distinct Ca²⁺ action potentials are not observed in the presence of TTX, it must be held in doubt whether action potentials play a role in hormone secretion.

What is the evidence for receptor-operated ion channels? We should consider two different types of agonist effect on channels: primary-and secondary-ligand-activated channels. Meldolesi & Pozzan (1987) have suggested that these should be referred to as receptor-operated (ROC) and second-messenger-operated channels (SMOC), respectively. Evidence for direct ligand activation of ion channels in normal pituitary cells comes from the work of Inenaga & Mason who studied the effect of γ-aminobutyric acid (GABA) on a GABA_a receptor-Cl⁻ channel complex in normal bovine lactotrophs (Fig. 5) (Inenaga & Mason, 1987). Likewise, Kehl et al. (1987) and Taleb et al. (1987) have shown GABA to activate chloride channels in rat pars intermedia cells. GABA has been shown to inhibit PRL release from lactotrophs in vitro (Enjalbert et al. 1979; Grossman, Delitala, Yeo & Besser, 1981), and its effects on Cl ⁻ channels may be the underlying mechanism. This type of action is typical of neurotransmitter actions on neurones.

There are a number of ways in which releasing factors might act through SMOCs to alter [Ca²⁺]_i. The intracellular messenger inositol trisphosphate appears to function in some cells by stimulating the release of Ca²⁺ into the cell cytoplasm from intracellular stores. Other more indirect effects could be via the closing of K⁺ channels, in turn depolarizing membrane potential, opening of voltage-dependent Ca²⁺ channels or non-specific cation channels and permitting the influx of Ca²⁺ into the cell cytoplasm. This depolarization may be sufficient to open voltage-activated Ca²⁺ channels, but below the threshold for stimulating action potential firing. These factors may also modulate the opening or closing of voltageindependent Ca²⁺ channels in the cell membrane, again leading to the influx of extracellular calcium. Finally, there may exist effects, via changes in intracellular messengers, influencing active pumps or passive ion exchange mechanisms in the plasma and intracellular organelle membranes that normally regulate intracellular Ca²⁺ levels.

We thus can reach a tentative conclusion that under some conditions hypophy seal factors may regulate hormone release via changes in membrane potential (GABA actions), whereas other factors may influence release in a non-voltage-dependent manner (e.g. GRF action on GH release, GnRH action on LH release). In the former case these cells behave in a manner normally associated with neurones, whereas the non-voltage-dependent regulation of release is more akin to that in other cells within the periphery. Endocrine cells seem then to possess properties of both these cell types.

Much evidence with clonal and normal pituitary cells suggests that when cell membrane receptors are stimulated by hypothalamic peptides, changes in resting levels of a variety of soluble cytoplasmic molecules are observed. The formation or release of these molecules occurs through the regulation of membrane-bound enzymes such as adenylate cyclase, guanylate cyclase and phospholipase C. The molecules formed can be grouped under the heading of intracellular messengers, and include the cyclic nucleotides cyclic AMP and cyclic GMP, diacylglycerol, inositol trisphosphate, and Ca²⁺.

One mechanism for intracellular regulation of intracellular events is via the phosphorylation of proteins. Many important regulatory molecules such as ion channels and enzymes are proteins, and their activation or inactivation by phosphorylating processes plays a critical role in the control of cell activity. Many of these ‘intracellular messengers’ appear to activate specific enzymes (such as protein kinases and calcium-binding proteins), which in turn phosphorylate other proteins. An example of the consequences of phosphorylation is that when voltage-activated Ca²⁺ currents are recorded, the currents decrease in amplitude very rapidly as the cytoplasmic contents are dialysed by the pipette solution. However, this Ca²⁺ current rundown is to a large extent prevented if millimolar concentrations of MgATP are included in the pipette (DeRiemer & Sakmann, 1986; Cobbett et al. 1987b), or if other high-energy-phosphate-generating enzyme systems are employed.

However, the first electrophysiological evidence that intracellular messenger substances are involved in modulation of anterior pituitary ion channels came from work on gonadotrophin-secreting cells (Mason & Waring, 1985). First, when gonadotrophs were stimulated to release LH or FSH by the hypothalamic peptide GnRH, intracellular recordings showed that action potentials were not generated, the membrane was only slightly depolarized (by 2 –5 mV) and, significantly, that membrane voltage ‘noise’ increased markedly -implying the activation of underlying ion channels in the absence of voltage changes. Second, when single Ca²⁺ channels from an isolated membrane patch attached to the gonadotroph were recorded, GnRH applied to the membrane outside the patch caused activation of Ca²⁺ channels inside the patch (Fig. 6A). GnRH applied directly to the extracellular face of the membrane patch inside the electrode did not activate channels. Third, if the patch was excised from the cell after GnRH had been observed to activate Ca²⁺ channels, the channel activity rapidly disappeared (Fig. 6B), even if Ca²⁺ concentration on the cytoplasmic face was elevated to very high levels. This evidence taken together suggested that GnRH probably caused the formation of intracellular messenger substance(s) which acted on the cytoplasmic membrane face to open calcium-selective channels, and that the messenger substance in gonadotrophs was something other than Ca²⁺.

Other more recent work using whole-cell recording to dialyse single pituitary cells has shown that the hypothalamic peptides GnRH, GRF and TRH will not, in general, activate their target cell if the cell interior is perfused, and this has been taken to indicate that soluble messengers are removed from the intracellular milieu during this process. Some good evidence for this has been the recent work of Dufy, Jaken & Barker (1987) who showed that dialysed GH3 cells would not respond to TRH unless the pipette contents contained both protein kinase C (PKC) and 100nmoll⁻¹ Ca²⁺ (but not if Ca²⁺ was buffered to 10nmoll⁻¹), suggesting a possible synergy of action in supporting action potential firing (Fig. 7).

How then do these intracellular messengers regulate ion channels in anterior pituitary cells, and what is their role in secretion?

Work on gonadotrophs suggested (Mason & Waring, 1985, 1986) that Ca²⁺ channels in the plasma membrane activated by GnRH were not dependent on voltage, but rather it was inferred that channel activation occurred, at least in part, because of modulation by some soluble intracellular messenger. A more direct test of this hypothesis is now possible, using patch-clamp methodology to both clamp the membrane voltage and dialyse the cell interior with a Ca²⁺-sensitive dye (Indo-1) allowing intracellular Ca²⁺ concentration to be monitored by microspectro-fluorimetry. This combined technique has been used to study the action of TRH in single prolactin-secreting GH3 cells held under voltage-clamp.

Under these conditions, a depolarizing voltage step pulse clearly activates a rise in intracellular Ca²⁺ concentration (Benham, 1988). For instance, in the presence of only 1mmoll ⁻¹ extracellular Ca²⁺, resting intracellular Ca²⁺ level rises from about 100 to about 400 nmol I⁻¹ during a 0-5-s voltage pulse which evokes an inward Ca²⁺ current and decays over about 3 –6 s. This demonstrates the presence of voltage-activated Ca²⁺ channels in the plasma membrane.

Fig. 8 shows the results of TRH application when membrane voltage is maintained close to resting potential, at –70 mV, with 1 mmol I⁻¹ extracellular calcium. In this experiment, a large transient rise in intracellular Ca²⁺ from about 80 to 400 nmol I⁻¹ followed by a sustained plateau level slightly above the basal value was observed, despite the fact that membrane potential did not change. A small transient inward current was also noted, which may have been a Ca²⁺-activated chloride current. In another experiment, smaller Ca²⁺ transients also occurred when all extracellular Ca²⁺ had been removed, suggesting the source of the rise in intracellular Ca²⁺ concentration to be partly intracellular, but also with a significant contribution from plasma membranes which are probably activated close to membrane resting potential.

Accumulation of intracellular cyclic AMP following adenylate cyclase activation in pituitary cell cultures has been observed following activation by such factors as GnRH, GRF and vasoactive intestinal peptide (VIP) (Borgeat et al. 1972; Kaneko et al. 1973; Bilezikjian & Vale, 1983; Naor, Snyder, Fawcett & McCann, 1980), and is associated with gonadotrophin and growth hormone release. How then is the rise in cyclic AMP level coupled to hormone release in these cells?

Recent studies in our laboratory on ovine gonadotrophs indicate that cyclic AMP could influence the release process by changing cellular excitability through alteration of ionic channel gating. In particular, we have focused on Camdependent K⁺ channels of about 100 pS conductance, which probably play a major role in control of membrane potential during rises in intracellular Ca²⁺ concentration (Fig. 9A). Activation of the adenylate cyclase enzyme by forskolin applied extracellularly causes closure of Ca²⁺-dependent K⁺ channels, shown using cell-attached, patch-clamp recordings. Elevation of intracellular cyclic AMP concentration by dibutyryl cyclic AMP (a membrane-permeable analogue of cyclic AMP) produces similar results (Fig. 9B). The fact that this is a direct action of cyclic AMP on the intracellular face of the membrane was confirmed by application of cyclic AMP to the cytoplasmic face with recordings from inside-out patches.

We have additionally explored the biochemical basis for this inhibition and found it to occur as a result of a two-stage process. First, the cyclic AMP binds to a cyclic AMP-dependent protein kinase (cAMPPK) present in the membrane. This kinase has two components, a regulatory subunit which is bound by cyclic AMP and causes its detachment from the second, catalytic, subunit. The latter phosphorylates the Ca²⁺-dependent K⁺-channel protein and thus closes it. Direct application of the catalytic subunit of the cyclic AMP-dependent protein kinase (CS-cAMPPK) to inside-out patches decreased channel activity in a manner similar to that of cyclic AMP, and this can be reversed by the protein kinase inhibitor applied on the cytosolic face (Fig. 10B) following closure of the channel by 10 μmoll⁻¹ cyclic AMP (Fig. 10A). As for hormone secretion, cyclic AMP potentiates the release of LH in response to GnRH, probably by causing a shift in the membrane potential to a more depolarized state, since K⁺ channels normally participate in maintaining the cell in a hyperpolarized state. This would ensure further Ca²⁺ influx through voltage-activated Ca²⁺ channels leading to more hormone release (Sikdar et al. 1987).

The second-messenger role of cyclic GMP in the secretory phenomenon is less than clear. For instance, although GnRH stimulates the formation of cyclic GMP in cultured pituitary cells, increased cyclic GMP production provoked by sodium nitroprusside does not lead to increased LH release (Naor & Catt, 1980), suggesting that cyclic GMP production is not an obligatory step in LH release. In rat somatotrophs, however, cyclic GMP stimulates GH secretion (Peake, Steiner & Daughaday, 1972). We have recently found in cell-attached, patch-clamp recordings that cyclic GMP, presumably via a protein kinase molecule, opens a Ca²⁺-permeable channel in ovine gonadotrophs, similar to the action of GnRH. Cyclic GMP clearly acted on these channels at the cytoplasmic face in inside-out patches, but its overall role in modulation of hormone release remains obscure at present.

The key enzyme activating the phosphatidylinositol system is phosphoinositi-dase C (PIC), which causes the formation of phosphoinositides (such as InsP₃ and InsP₄) as well as diacylglycerol (DAG). In anterior pituitary cells the factors known to stimulate this system include acetylcholine (ACh) action on muscarinic receptors, bombesin and, as described below, TRH.

TRH causes hydrolysis of a phosphoinositide precursor, phosphatidylinositol 4,5-bisphosphate [PtdIns(4,5)P₂] into cytoplasmic (InsP₃) and membrane-bound (DAG) components. The concentration of InsP₃ reaches a maximum in 15 s (Rebecchi & Gershengorn, 1983) and possibly contributes to an initial rise of intracellular Ca²⁺ level following TRH activation. The membrane-bound component, DAG, activates protein kinase C (PKG) which is thought to cause a second phasic rise in intracellular Ca²⁺ level based on findings in the GH cell. These events may play a role in transduction of the TRH signal into a secretory response.

Electrophysiological recordings of bovine lactotrophs and GH3/B6 cells using conventional intracellular recording techniques, reveal that TRH induces an increase in Ca²⁺-dependent K⁺ conductance leading to cell hyperpolarization, followed by a period of sustained spiking activity but with only very slight depolarization (Ingram, Bicknell & Mason, 1986; Ozawa & Kimura, 1979L Phorbol myristyl acetate (PMA) mimics the TRH response (Dufy et al. 1987) presumably by activation of PKC. However, the electrophysiological responses to both TRH and PMA were lost when the whole-cell recording configuration of the patch-clamp technique was employed -presumably because of the dialysis of important compounds from the cell cytoplasm. The responses were restored, however, when products of cell lysates were included in the patch electrode. The active cellular components needed to produce the response were identified as protein kinase C and InsP₃. The phosphoinositide Ins/, or calcium, was necessary to allow the TRH-induced hyperpolarization, and PKC, in combination with calcium, allowed the TRH stimulation of the slight depolarization and increase in spiking rate. PKC alone in the pipette failed to elicit either the TRH or the PMA electrical response, implying an essential role of Ca²⁺ in the TRH-stimulated electrophysiological response. In fact, protein kinase C and Ca²⁺ synergistically increase GH, LH and TSH secretion from anterior pituitary cells (Judd, Koike, Yasumoto & MacLeod, 1986).

Ca²⁺ entering the cell cytoplasm can activate a number of Ca²⁺-binding proteins such as calmodulin (CaM), which is associated with a number of intracellular events provoked by a rise in intracellular Ca²⁺ concentration (Cheung, 1980). The Ca²⁺ –CaM complex can then act directly on other proteins, e.g. the ATPase pump found in many cell membranes, that pumps Ca²⁺ out of the cell, or indirectly via phosphorylation of other enzymes such as the protein kinases.

In gonadotrophs, the physiological releasing factor GnRH induces redistribution of calmodulin from the cytosol to the plasma membrane (Conn, Chafou-leas, Rogers & Means, 1981), and there is evidence for an association between calmodulin and membrane patch areas containing receptors for GnRH (Jennes, Bronson, Stumpf & Conn, 1985). However, the cellular protein kinases, which bridge the events between calmodulin and cytoskeletal activation and translate these to granule movement, extrusion and exocytosis, still remain to be characterized. Hormone release from the pituitary seems to be dependent on calmodulin since it can be attenuated by antagonists of calmodulin (Conn, Rogers & Sheffield, 1981).

Ca²⁺ also has other actions, independent of calmodulin, on other Ca²⁺-binding proteins. For example, it can act directly upon the Ca²⁺-dependent K⁺ channel to cause its opening and thus influence cell excitability as described above.

There exist many interrelationships between these intracellular messengers in the cell. For example, cyclic AMP can act to influence the influx or extrusion of Ca²⁺ from the cytoplasm. Ca²⁺ may act, via calmodulin, to activate or inactivate adenylate cyclase (which produces cyclic AMP) or phosphodiesterase (which breaks down cyclic AMP). In smooth muscle cells the ability of the Ca²⁺ –CaM complex to phosphorylate (and activate) myosin light chain kinase (MLCK) is Inhibited by prephosphorylation of another site on MLCK by cyclic AMP-dependent protein kinase.

In anterior pituitary cells an example of the interrelationship between cyclic AMP and the phosphatidylinositol system has been suggested to occur in somatotrophs (Ingram & Bicknell, 1986). The GRF receptor is known to be coupled to adenylate cyclase and stimulation of the receptor with GRF leads to a rise in intracellular cyclic AMP concentration. TRH, however, acts via the stimulation of PIC. Application of GRF and TRH together in these cells will release very significantly more GH than the sum of the individual releasing factors alone, implying that activation of more than one intracellular pathway may promote, in this case, a synergistic response. Another interrelationship of intracellular messenger action in anterior pituitary cells is seen with the Ca²⁺ dependent K⁺ channel, which is activated by Ca²⁺ but closed by cyclic AMP.

How do these anterior pituitary cells differ from one another? It seems likely that release-modifying factors act on different cells in a similar manner, e.g. the TRH stimulation of lactotrophs and thyrotrophs is via rises in Ins P₃ and DAG in both cell types. What may be different, however, is the fact that each cell type within the pituitary has its own receptors for specific peptides and non-peptides, and the interrelationship between these extracellular and intracellular stimuli will be unique for a particular cell. However, it seems likely that a common action is to change the concentration of Ca²⁺ inside the cell, enabling the role of Ca²⁺ as a second messenger and allowing it to act in synergy with PKC and other systems.

Anterior pituitary hormone secretion is a Ca²⁺-dependent process. Depriving cells of extracellular Ca²⁺ by incubation in media containing Ca²⁺ buffers such as EGTA inhibits GnRH-induced LH secretion (Bates & Conn, 1984), GRF-induced GH secretion (Sheppard, Kraicer & Milligan, 1980) and TRH-induced PRL secretion (Thorner, Hackett, Murad & MacLeod, 1980). Furthermore, this inhibition can be reversed by elevation of extracellular calcium. Agonist-induced hormone secretion can be inhibited by various Ca²⁺ channel blockers such as cobalt and verapamil (Conn, McArdle, Andrews & Huckle, 1987; Merritt & Brown, 1984; Ingram et al. 1986). Secretion of LH, GH and PRL can be enhanced in several ways, such as by the Ca²⁺ ionophore A23187 (Sheppard et al. 1980; Conn, Rogers & Sandhu, 1979; MacLeod, Schettini & Canónico, 1984), by raising the extracellular potassium concentration to depolarize the cell membrane and promote influx through voltage-dependent Ca²⁺ channels (Sheppard et al. 1980; Hopkins & Walker, 1978; Nakano, Fawcett & McCann, 1976) or by maitotoxin, a Ca²⁺ channel activator (Schettini et al. 1984; Rasmussen & Barret, 1984). In normal bovine lactotrophs, the Ca²⁺ channel blockers (Co²⁺, Mn²⁺) decrease the basal and TRH-stimulated secretion (Cobbett et al. 1987b). This and other evidence strongly suggests that alterations of intracellular ionized Ca²⁺ concentration may also play a major role in anterior pituitary cell stimulus-secretion coupling. Further, other experiments also suggest that intracellular Ca²⁺ may serve to couple the different signal transduction pathways present in these cells.

Using the photoproteins aequorin and quin2 (Tsien et al. 1982; Grynkiewicz et al. 1985) it has been shown that TRH stimulates a rapid elevation of intracellular Ca²⁺ concentration in suspensions of GH3 cells (Gershengorn, 1985) and bovine pituitary cells (Schofield, 1983). Several studies have shown that the increase is biphasic. The first phase consists of a rapid several-fold transient increase which peaks within 6 –8 s and returns rapidly towards prestimulation levels. The increase is unaffected by prior removal of extracellular Ca²⁺ or the Ca²⁺ channel blockers nifedipine and verapamil, and is therefore believed to be the result of release from intracellular stores. This is followed by a prolonged phase dependent on the influx of external Ca²⁺ since it is abolished by the removal of extracellular Ca²⁺ and is markedly reduced by nifedipine and verapamil (Albert & Tashjian, 1984; Gershengorn & Thaw, 1985). Similar changes in gonadotrophs are stimulated by GnRH (Limor et al. 1987).

A similar biphasic pattern of PRL and GH secretion in response to TRH has been observed. The transient phase involves the release of stored hormone and appears to depend on Ca²⁺ mobilization from intracellular stores, whereas the second phase, involving the release of newly synthesized hormone, is dependent on Ca²⁺ influx from outside the cell (Aizawa & Hinkle, 1985). It is believed that TRH binding to its receptor stimulates PtdIns(4,5)P₂ hydrolysis to Ins(l,4,5)P₃ and DAG. The former then diffuses to a non-mitochondrial intracellular Ca²⁺ store and causes the first-phase transient rise of intracellular Ca²⁺ concentration which in turn leads to the release of stored hormone. It is also suggested that TRH binding initiates an inhibitory process involving PKC that rapidly reverses the rise in intracellular ionized Ca²⁺ level. The second phase of extracellular Ca²⁺ influx is believed to be responsible for the prolonged phase of GH and PRL secretion, although the mechanisms and pathways involved remain largely unresolved.

We have recently begun to monitor intracellular free Ca²⁺ concentration at the single-cell level in enriched cultures of normal bovine lactotrophs using the technique of fura2 ionized Ca²⁺ imaging (Fig. 11A), which provides a pixel-by-pixel dynamic image of intracellular ionized Ca²⁺ concentration. Challenging cells with 10 –100 nmol I⁻¹ TRH, maximal for PRL secretion, induces a rapid transient increase of intracellular Ca²⁺ concentration in most cells within seconds of TRH application (Fig. 11 A,B). This consists of a several-fold increase with a rise time of the order of 1 to several seconds followed by an exponential decay which lasts between 8 and 25 s. Similar transient responses have also been observed in the presence of extracellular cadmium and cobalt or when extracellular Ca²⁺ was substantially reduced just prior to stimulation. This is strong evidence that the transient response is the result of Ca²⁺ release from intracellular stores. These results are in general agreement with the first-phase Ca²⁺ response reported in GH3 cells.

In summary, factors acting to cause hormone release from anterior pituitary cells also cause increases in free cytoplasmic Ca²⁺ concentrations. Is this Ca²⁺ rise linked directly to secretion? To begin to answer these questions we have employed new techniques to record the secretory event ‘exocytosis’, and the preliminary results we, and others, have obtained are described in the following section.

Exocytotic release is a fundamental process in the function of neurones and other secretory cell types. Secretory products (neurotransmitters, hormones, various peptides, proteins, purinergic substances, etc.) are packed in vesicles. A number of these vesicles fuse with the surface membrane, during both nonactivated and activated phases, to release secretory product into the extracellular space. This process is called exocytosis (see review by Knight & Baker, 1987). In addition to regulated exocytosis that is triggered by a stimulus, continuous fusion of vesicles with the plasma membrane probably occurs. This constitutive process is thought to be part of the continuous cycling of cell membranes.

Stimulated secretion from nerve terminals as well as many other cells requires extracellular Ca²⁺ (Douglas, 1968). Katz & Miledi (1967) first suggested that presynaptic action potentials open Ca²⁺ channels in the presynaptic terminal, causing an influx of Ca²⁺, which triggers the secretory machinery. This proposal has been extended from neural tissue to other secretory cells, where extracellular Ca²⁺ is essential for long-term release of hormone.

A major obstacle to elucidating the mechanism of the stimulus-secretion pathway is that the events controlling exocytosis are occurring at the inner face of the plasma membrane, which is relatively inaccessible to experimental manipulation. However, the technique of measuring cell membrane capacitance developed by Neher & Marty (1982) enables measurements on a single cell during exocytosis. Such recording is usually made in a whole-cell, patch-clamp recording configuration, in which the intracellular space is dialysed with the solution used to fill the patch pipette. Membrane potential is ‘clamped’ to a desired level to which an additional sinusoidal voltage is added. Analysis of the resulting sinusoidal current, that flows through the membrane, can reveal conductive and capacitive properties of the cell membrane using a phase-sensitive detector.

Is elevated [Ca²⁺]_i necessary to trigger exocytosis? Although calcium ions play a pivotal second messenger role in many secretory cell types (Knight & Baker, 1987), there is growing evidence for a number of cells that elevated [Ca²⁺]_i alone cannot fully explain the observed secretory response. For example, in the platelet, secretion is only triggered by doses of the Ca²⁺ ionophore ionomycin which raise [Ca²⁺]_i to micromolar levels. Sub-micromolar levels of Ca²⁺ do not trigger secretion (Rink & Hallam, 1984). The natural agent thrombin, however, only raises [Ca²⁺]_i to sub-micromolar levels (0·1 μmoll⁻¹) but does not trigger secretion. Furthermore, collagen-triggered secretion occurs with no detectable change in [Ca²⁺]_i from quin2-loaded platelets (Rink, Sanchez & Hallam, 1983). The fact that secretory exocytosis need not be regulated by a rise in [Ca²⁺]_i is demonstrated in tissues such as juxtaglomerular apparatus (Keeton & Campbell, 1981) and parathyroid gland (Nemeth & Scarpa, 1987). In these cells a rise in [Ca²⁺]_i is inhibitory to secretion. In addition, protein kinase C will activate physiological processes related to secretion at very low levels of cytosolic Ca²⁺ (DiVirgilio, Lew & Pozzan, 1984).

In the case of bovine lactotrophs our preliminary experiments have shown that it is possible to induce cell degranulation (measured as an increase in capacitance) by dialysing the intracellular space with solutions containing raised [Ca²⁺]_i, (Fig. 12A), and to observe small step-wise fluctuations in membrane capacitance consistent with both fusion and retrieval of single vesicles (see Fig. 12B). Assuming the specific membrane capacitance of the lactotroph cell membrane to be about 1 μF cm⁻², these step changes of about 5 fF reflect fusion (or retrieval) of granules with diameters of 0·5 μm, similar to the diameters of granules observed in these cells with electron microscopy. Using pipettes filled with a solution known to support exocytosis in chromaffin cells (Neher & Marty, 1982) and elevating [Ca²⁺]_i by activating voltage-dependent Ca²⁺ currents is not sufficient to evoke time-locked capacitance changes (a measure of exocytosis), which seems to suggest that other second-messenger systems in addition to Ca²⁺ are probably involved in controlling secretion in these cells. There is other evidence which supports this tentative speculation. In GH3 cells it was observed that the sustained phase of secretion is Ca²⁺-independent (Gershengorn, 1986); in rabbit neutrophils secretion can be stimulated at extremely low levels of [Ca²⁺]_i (<10 ^−l0mol I ⁻¹) (Barrowman, Cockcroft & Gomperts, 1986). Furthermore, using the whole-cell, patch-clamp technique, it was found that mast cells would not generally degranulate even if [Ca²⁺]_i was maintained in the range 0·2 –10 μmol I⁻¹ (Fernandez, Neher & Gomperts, 1984).

The unifying factor linking all of these examples is that second messengers other than Ca²⁺ could play an equally important role in stimulus-secretion coupling. Possible candidates could include products of phosphoinositide metabolism, cyclic nucleotides, guanine-nucleotide-binding proteins (G-proteins), and other molecules as yet undefined. In the example of the platelet the available evidence suggests that it is a product of phosphoinositide breakdown, diacylglycerol, which plays a key role in the secretory response, presumably by activating the enzyme PKC (Nishizuka, 1984). In the case of mast cells, degranulation initiated by the addition of GTP analogues (in the presence of ATP) was sufficient to stimulate secretion (Fernandez et al. 1984). In these cases it is likely that the main role of the additional second messenger is to modulate the Ca²⁺-sensitivity of the exocytotio machinery since buffering of [Ca²⁺]_i to very low levels in platelets resulted in the failure of PKC activators to trigger secretion (Knight, 1986), and a rise in [Ca²⁺]_i accelerated, but did not initiate, the degranulation of mast cells, judged from the capacitance signal that is complete in about 3 –5 min (Neher, 1988).

Interestingly, rabbit neutrophils can be stimulated to secrete by GTP analogues at <10 ⁻¹⁰moll⁻¹ Ca²⁺ (Barrowman et al. 1986). Under these conditions activation of polyphosphoinositide phosphodiesterase (the enzyme responsible for the production of diacylglycerol and inositol trisphosphate which is under the control of the G-protein, G_p) does not occur (Cockcroft, 1986). In this case it is unlikely that secretion is truly Ca²⁺-independent. GTP-analogue-induced Ca²⁺-indepen-dent secretion has also been reported in rat insulinoma RINm5F cells (Vallor, Biden & Wollheim, 1987) and in bovine adrenal chromaffin cells (Bittner, Holz & Neubig, 1986). There is evidence for the the existence of pertussis-toxin-sensitive G-proteins in anterior pituitary cells (Journot et al. 1987), but whether their activation is a necessary prerequisite for secretion from these cells remains to be investigated.

A variety of new methods for studying the properties of single cells are providing us with new insight into the control of secretion. In the case of anterior pituitary cells which secrete a wide range of physiologically important hormones, the clear challenge is to understand how brain peptides and neurotransmitters regulate electrical activity (i.e. ionic permeability) of both plasma membrane and intracellular membrane compartments, and how these events are coupled to intracellular biochemistry and Ca²⁺ homeostasis and ultimately to exocytosis and hormone release. Already, several unexpected challenges to conventional dogma have come to light.

First, we now know that rises in intracellular Ca²⁺ concentration may be dynamically modulated by both intracellular Ca²⁺ stores and by extracellular sources. The gates, or channels, controlling Ca²⁺ entry into the cytosol are almost certainly regulated in different ways, although some features may be common. We might class one pathway of control as the conventional voltage-regulated pathway, and the second as chemical control, by ligand-channel complexes or by intracellular messengers. Intracellular messengers almost certainly regulate the intracellular compartment, but it appears they may also modulate plasma membrane Ca²⁺ channels in the absence of any large voltage shifts. Both are probably important in optimizing levels of cell Ca²⁺ under different conditions. This information will enable us to focus on key differences between pituitary cells and neurones in respect of Ca²⁺ entry; for example, peptide agonists, unlike certain transmitters, do not alter membrane potential markedly, although they do affect plasma membrane Ca²⁺ permeability.

Second, agonist binding to plasma membrane receptors activates a diversity of intracellular metabolic cascades. We are discovering that a further class of receptors also exists, these being associated with intracellular binding sites and some intimately coupled to cytosolic receptors linked to plasma membrane ion channels. In this way, both voltage-activated channels and second-messenger-activated channels may be modulated, with consequences for control of secretion.

Third, using techniques for measuring the electrical consequences of exocytosis, we are beginning to establish in pituitary cells, as found for mast cells secreting histamine, that membrane fusion may depend on factors other than intracellular Ca²⁺ concentration and that the fusion event is not an ‘all-or-nothing’ process. The ability to dialyse the cell cytosol of anterior pituitary cells with candidates for messenger substances is enabling reconstitution studies which should reveal whether calcium ions are really the prime mover in secretion, or merely a modulator acting in synergy with other molecules -cyclic nucleotides, phosphoinositides and so forth. This area of work is likely to provide us with important information not only about the events preceding secretion but also about what follows in membrane recovery -a much slower event whose dependence on Ca²⁺ is only now beginning to be explored.

Finally, it is clear that we are now faced with more questions and challenges about control of the secretory event than may have been obvious only a year or two ago. Although we are rapidly developing the tools to explore and understand secretion at the level of molecular interactions, the excitement for the future is really about what we do not know! Undoubtedly, the best we can expect is that the answers will be unexpected.

Various aspects of this work have been generously supported by the Wellcome Trust, Beit Memorial Trust, Kabi Vitrum AB, Nuffield Foundation, Burroughs Wellcome Foundation, Medical Research Council, Lalor Foundation, the Meat and Livestock Commission of Great Britain, and the Research Council of Slovenia.