Phorbol Esters, Protein Phosphorylation and the Regulation of Neuronal Ion Channels

One well characterized effect of synaptic stimulation of a neurone is to produce a transient depolarization, or hyperpolarization, which triggers or inhibits the occurrence of action potentials. A second, and more pervasive, way that synaptic or hormonal stimulation influences many neurones is to produce long-lasting changes in their electrical properties - changes which substantially outlast the duration of stimulation. Such changes, which frequently result from modifications in the properties of voltage-dependent ion channels in the plasma membrane of a neurone, are of cardinal importance to the functioning of the nervous system and allow it to generate and control long-lasting changes in animal behaviour.

A biochemical mechanism that produces prolonged changes in neuronal excitability is the post-translational modification of ion channels, or of proteins that regulate ion channel activity, by the activation of protein kinases. The nervous system is the richest source of many of the different classes of protein kinases. The activity of several types of protein kinase can be stimulated by second messenger substances which are formed as a result of synaptic stimulation. This article deals with the characteristics of one class of protein kinase, the calcium/diacylglycerol/ phospholipid-dependent protein kinase, also termed protein kinase C, and its possible roles in the regulation of neuronal excitability. These topics will be illustrated using the example of the bag cell neurones of Aplysia, a group of neurones which are particularly suited to combined biochemical-electrophysiological studies and which undergo a series of long-lasting changes in their electrical properties in response to brief synaptic stimulation.

Protein kinase C was first described as an enzyme, present in soluble fractions prepared from rat and bovine brain, which could catalyse the phosphorylation of serine residues in histones (Takai, Kishimoto, Inoue & Nishizuka, 1977; Inoue, Kishimoto, Takai & Nishizuka, 1977). It was shown to be distinct from both the cyclic-AMP-dependent and the cyclic-GMP-dependent protein kinases. The activity of protein kinase C was subsequently shown to depend on its lipid environment and, in particular, to require phosphatidylserine and diacylglycerol, as well as calcium ions, for its full activation (see Fig. 1). Diacylglycerols are generally found in significant quantities in cellular membranes only when cells are exposed to stimuli which cause the hydrolysis of membrane phosphatidylinositides. The transient production of diacylglycerol in response to such stimuli is believed to be the major physiological pathway for the activation of protein kinase C (for reviews see Nishizuka, 1984a,b;Berridge, 1984).

In many different species the highest concentrations of protein kinase C have been found in the nervous system (Kuo et al. 1980; Minakuchi, Takai, Yu & Nishizuka, 1981; Wrenn, Katoh & Kuo, 1981; Girard, Mazzei, Wood & Kuo, 1985). The enzyme has been purified from rat brain and found to comprise a single polypeptide chain with an apparent M_r of approximately 82000 (Kikkawa et al. 1982). Diacylglycerol has been shown to increase the apparent affinity of the enzyme for phospholipids and to decrease the concentration of calcium ions that is needed to activate the enzyme fully. Thus, in the presence of diacylglycerol and phosphatidylserine, the activity of the enzyme can be stimulated maximally at a calcium concentration of about 0·lμmol l⁻¹, which is within the range of basal calcium concentrations of many cells.

Protein kinase C can be cleaved into two fragments by an endogenous calciumdependent protease in rat brain (Kajikawa, Kishimoto, Shiota & Nishizuka, 1983). The larger fragment generated by such proteolysis is catalytically active in the absence of phospholipids, diacylglycerol or calcium ions. Although the proteolytic activation of protein kinase C was known before its activation by lipids came to be appreciated, the physiological significance of proteolytic activation is not yet known. One attractive possibility is that such irreversible activation may be responsible for some prolonged neuronal responses to stimulation (Lovinger et al. 1985).

It has been reported that protein kinase C can also be activated in cell-free extracts, and in intact cells, by unsaturated long-chain fatty acids such as arachidonic acid, linolenic acid, linoleic acid and oleic acid. This has led to the suggestion that the stimulus-induced production of these fatty acids within cells may constitute an alternative physiological pathway for the activation of this kinase (McPhail, Clayton & Snyderman, 1985; Murakami, Chan & Routtenberg, 1985).

Protein kinase C activity can be detected in both cytoplasmic and membrane fractions from vertebrate and invertebrate nervous tissue (Kikkawa et al. 1982; DeRiemer, Greengard & Kaczmarek, 1985a). In intact cells the activation of this enzyme by phorbol esters (see below) results in its translocation from cytoplasm to the plasma membrane (Kraft & Anderson, 1983). Such translocation may also be promoted by an increase in the intracellular calcium ion concentration (Wolf et al. 1985; May, Sahyoun, Wolf & Cuatrecasas, 1985). It has recently been reported that electrical stimulation of an afferent input to the hippocampus results in significant translocation of protein kinase C into membrane fractions (Akers et al. 1985).

A number of agents can inhibit the activity of protein kinase C. These include several polypeptide toxins (Kuo et al. 1983) as well as many antipsychotic agents (Mori et al. 1980; Schatzman, Wise & Kuo, 1981). Most of the agents which bind to calmodulin, and are thereby effective antagonists of calmodulin-dependent enzymes, also inhibit protein kinase C activity. For example, the ‘calmodulin antagonists’ trifluoperazine and W7 inhibit protein kinase C activity in the Aplysia nervous system at the same concentrations that inhibit calcium/calmodulin-dependent protein phosphorylation in these cells (DeRiemer et al. 1984, 1985a). The calmodulin antagonist, calmidazolium, is however a slightly more effective inhibitor of calmodulin-dependent protein phosphorylation than of protein kinase C (DeRiemer et al. 1985a). Because of the lack of specificity of most of the inhibitors of protein kinase C, pharmacological inhibition has not yet proved to be a useful tool for investigation of the role of protein kinase C in neuronal function.

Interestingly, protein kinase C can also be inhibited by calmodulin. Such inhibition has been demonstrated in cell-free extracts of rat brain (Wu, Walaas, Nairn & Greengard, 1982) and of Aplysia nervous system (DeRiemer et al. 1985a) and has also been demonstrated using purified protein kinase C and several of its substrate proteins (Albert et al. 1984b). These data have raised the possibility that calmodulin may be a physiological modulator of protein kinase C. A recent report has also suggested that there may exist other endogenous protein inhibitors of this enzyme (Chan, Murakami & Routtenberg, 1985).

As described above, under normal conditions protein kinase C is believed to be activated by physiological stimuli which result in the catabolism of phospholipids in plasma membranes. There also exist pharmacological techniques for the activation of protein kinase C which can stimulate this enzyme directly in intact cells or in broken cell preparations. These can activate protein kinase C without influencing other second messenger systems which may be normal components of the full physiological response. Such pharmacological agents, which include phorbol esters and synthetic diacylglycerols, have therefore been used to test the possible roles of this enzyme in a variety of physiological processes, including changes in neuronal excitability.

Phorbol esters, such as TPA (12-O-tetradecanoylphorbol-13-acetate), were first described as potent tumour promoters and inflammatory agents when applied to mouse skin. (For an account of the structure—activity relationships of phorbol esters see Hecker, 1971.) Within mouse skin different classes of phorbol esters have been shown to exert different biochemical and cellular effects. In particular, the inflammatory and tumour-promoting effects can be distinguished by differing structureactivity relationships. This heterogeneity of biological response has been correlated with the existence of multiple classes of binding sites for phorbol esters in preparations of mouse skin (Dunn & Blumberg, 1983).

In contrast to mouse skin, equilibrium binding data from many other tissues, including particulate fractions and cytoplasmic fractions prepared from the nervous system, are consistent with a homogeneous class of high affinity binding sites for phorbol esters (Shoyab & Todaro, 1980; Dunphy, Delclos & Blumberg, 1980). In common with the activation of protein kinase C, the binding of phorbol esters requires the presence of phospholipids. The distribution of phorbol ester binding sites in different tissues (Shoyob, Warren & Todaro, 1981; Hergenhahn & Hecker, 1981) closely matches that of protein kinase C activity (Minakuchi et al. 1981; Wrenn et al. 1981), highest concentrations being found in the nervous system (Murphy et al. 1983). Furthermore the phorbol ester binding sites in brain tissue have been shown to co-purify with protein kinase C (Leach, James & Blumberg, 1983; Niedel, Kuhn & Vandenbark, 1983). These data strongly suggest that protein kinase C is the major target of phorbol esters in many tissues and, in particular, in the nervous system (Blumberg et al. 1984). It should be noted, however, that the kinetics of binding of phorbol esters to brain membranes display two distinct phases (Dunphy, Kochenburger, Castagna & Blumberg, 1981). It is not known whether these phases represent binding to protein kinase C in different lipid environments, complex kinetics of binding to the enzyme, or whether they imply the existence of other binding proteins.

Active phorbol esters contain a diacylglycerol-like structure. These agents have been shown to bind to protein kinase C, substituting for diacylglycerol to produce a potent stimulation of enzyme activity (Castagna et al. 1982; Kikkawa et al. 1983). Because phorbol esters are lipophilic, readily entering membranes of intact cells, and because they are not rapidly catabolized, they have been widely used as activators of protein kinase C and their effects on neuronal excitability have recently begun to be investigated. In contrast to exogenous phorbol esters, the endogenously produced diacylglycerols, which generally contain arachidonate at the 2-position, are rapidly degraded in cells, cannot readily cross cellular membranes and cannot be used as exogenous activators of protein kinase C. A number of synthetic diacylglycerols, including l-oleoyl-2-acetyl glycerol, have however been found to activate the enzyme when applied to intact cells (Kaibuchi et al. 1983; Lapetina, Reep, Ganong & Bell, 1985). Although these synthetic analogues may also be metabolized within cells and generally appear to be less potent activators than phorbol esters, they have been used in studies on intact non-neuronal and neuronal cells.

How specific are phorbol esters and synthetic diacylglycerols for the activation of protein kinase C in tissues such as the nervous system? Although this enzyme is, at present, the only known target for these agents, the concentrations that are used in physiological experiments are frequently greater than those needed to activate protein kinase C in cell-free extracts. It is likely, therefore, that these agents will turn out to have other cellular effects. For example, it has been proposed that changes in diacylglycerol concentrations may directly influence the association of plasma membranes with components of the cytoskeleton (Burn, Rotman, Meyer & Burger, 1985). It is also possible that some of these agents can be metabolized to other biologically active substances. At present, however, they are the most powerful probes available for testing the involvement of protein kinase C in biological responses.

In common with the cyclic-AMP-dependent protein kinase and the calcium/ calmodulin-dependent protein kinases, protein kinase C phosphorylates substrate proteins at serine and threonine residues and undergoes autophosphorylation when activated (Kikkawa et al. 1982). The substrate specificity of this enzyme is, however, distinct from that of the other protein kinases and substrates have been found that are phosphorylated only by protein kinase C.

Proteins which influence neuronal excitability, and which have been shown to act as substrates for protein kinase C in vitro, include the nicotinic acetylcholine receptor (Huganir, Albert & Greengard, 1983), the voltage-dependent sodium channel (Costa & Catterall, 1984) and tyrosine hydroxylase (Albert et al. 1984a). It should be remarked, however, that the demonstration that a kinase can phosphorylate a given protein in vitro does not necessarily imply that the enzyme acts on the protein in intact cells. Other protein kinase C substrate proteins of specific interest to neurobiologists include a 87 kDa phosphoprotein in brain synaptosomes which does not appear to act as a substrate for any of the other known kinases (Wu et al. 1982) and a 47kDa protein, named Fl, which undergoes an increase in phosphorylation during long-term potentiation of synaptic responses in the hippocampus (Routtenberg, Lovinger & Steward, 1985). The latter protein is apparently identical to a previously described protein, termed B50, which may play a role in phosphoinositide metabolism (Oestreicher, Van Dongen, Zwiers & Gispen, 1983).

The cellular effects of agents that activate protein kinase C have been investigated in a wide variety of cell types. These effects include an elevation of intracellular pH (Hesketh et al. 1985) and the promotion of cell proliferation (Nishizuka, 1984a). In both non-neuronal and neuronal cells, these agents may evoke or potentiate secretion (Knight & Baker, 1983; Rink, Sachez & Hallam, 1983; Kojima, Lippes, Kojima & Rasmussen, 1983; Nichols, Haycock, Wang & Greengard, 1985; Publicover, 1985; Zurgil & Zisapel, 1985). The following brief account of their effects, however, is confined to effects that have been observed on the electrical properties of neurones.

The effects of activators of protein kinase C on endogenous ionic currents in neurones are either to enhance or suppress excitability, depending on the type of neurone under investigation. Examples of neurones in which a net enhancement of excitability is observed include hippocampal pyramidal cells, photoreceptors of Hermissenda and the bag cell neurones of Aplysia (see below).

Phorbol esters evoke a marked change in the pattern of firing of pyramidal neurones recorded in slices of hippocampus (Baraban, Gould, Peroutka & Snyder, 1985a; Baraban, Snyder & Alger, 1985b). A prominent effect at lower concentrations is the depression of a slow afterhyperpolarization that follows a burst of action potentials, and which may be generated by a calcium-activated potassium current. When sodium currents and potassium currents have been attenuated pharmacologically, an enhancement of the calcium action potentials by phorbol esters is observed. This may therefore represent an enhancement of calcium current. In addition to these actions on intrinsic membrane currents, phorbol esters influence several of the synaptic responses observed in hippocampal neurones and, in particular, produce a large increase in the amplitude of the fast excitatory postsynaptic potential (EPSP) evoked by orthodromic electrical stimulation (Nicoll, Madison, Malenka & Andrade, 1985). It has been reported that the application of an active phorbol ester in vivo can prolong the duration of the long-term potentiation of synaptic responses that follows repetitive stimulation of the perforant path (Lovinger et al. 1985).

In photoreceptors of the mollusc Hermissenda, the application of a phorbol ester, as well as intracellular injection of protein kinase C, have been reported to reduce the amplitudes of the calcium-activated potassium current and the transient inactivating potassium current (A-current) (Fisher, Auerbach & Farley, 1985; Farley & Auerbach, 1986). The same manipulations also increase the magnitude of the voltage-dependent calcium current. Similar changes in the amplitudes of these ionic currents can be evoked by application of serotonin to the photoreceptors.

A suppressive effect of activators of protein kinase C on a neuronal calcium current occurs in primary cultures of neurones from the dorsal root ganglion. In these cells, norepinephrine causes calcium action potentials to shorten in duration by attenuating the amplitude of the voltage-dependent calcium current. This effect is mimicked by l-oleoyl-2-acetyl glycerol, as well as by phorbol esters (Rane & Dunlap, 1985; Werz & Macdonald, 1985). A suppression of calcium current, as well as changes in some other voltage-dependent currents which produce an inhibition of cell firing, has also been observed in neurones of the rat dorsal raphe when exposed to phorbol esters (Freedman & Aghajanian, 1985).

A number of effects of phorbol esters on transmission at peripheral synapses have been noted (Baraban et al. 1985b). One interesting observation, in view of the fact that the acetylcholine receptor acts as a substrate for protein kinase C in vitro, is that a phorbol ester and glyceryl dioleate reduce the sensitivity of cultured myotubes to acetylcholine (Eusebi, Molinaro & Zani, 1985).

The abdominal ganglion of mature Aplysia contains two large clusters of neurones, situated at the junctions of the pleuroabdominal connective nerves and the ganglion. The neurones in these clusters appear to be entirely homogeneous by several electrical and biochemical criteria. Moreover, they are coupled electrically by gap junctions on their processes, which ensures that, when stimulated, these neurones always discharge in synchrony (Kaczmarek, Finbow, Revel & Strumwasser, 1979; Blankenship & Haskins, 1979). These cells have been termed the bag cell neurones and they play a pivotal role in the generation of the prolonged sequence of behaviour that constitutes egg-laying behaviour in the animal.

In response to brief (approx. 15 s) electrical stimulation, or to peptides from the reproductive tract, the bag cell neurones generate a prolonged discharge of action potentials (Kupfermann & Kandel; 1970; Heller et al. 1980). This discharge continues after the stimulation has stopped and generally lasts for about 30 min. During this afterdischarge the action potentials of these neurones can be observed to increase in height and in width (Kaczmarek, Jennings & Strumwasser, 1982). This enhancement of action potentials promotes calcium entry into the cells. The discharge leads to the release of several neuroactive peptides which act on other neurones within the nervous system, as well as on peripheral targets to induce the sequence of egg laying behaviour (Fig. 2) (Arch, 1972; Dudek, Cobbs & Pinsker, 1979; Mayeri, Brownell, Branton & Simon, 1979; Schlesinger, Babirack & Blankenship, 1981; Rothman et al. 1983; Scheller et al. 1983).

The clusters of bag cell neurones can readily be dissected from the remainder of the abdominal ganglion as a homogeneous population of neurones for biochemical analysis. For the analysis of transmembrane ionic currents, individual neurones can also be isolated and maintained in primary cell culture, where they retain many of the properties of neurones within intact clusters (Kaczmarek et al. 1979; Kaczmarek & Strumwasser, 1981).

The pharmacological properties of protein kinase C in the bag cell neurones have been characterized in cell-free extracts both by studying the phosphorylation of endogenous substrates for this enzyme (Fig. 1) and by measuring the incorporation of phosphate into the M_r ≈ 87000 protein from rat brain, which appears to be a specific substrate for this enzyme (DeRiemer et al. 1985a). Using these techniques the enzyme has been shown to be present in both membrane and cytosolic fractions from these neurones. Its activity can be stimulated by TPA at concentrations of 10–100 nmol l⁻¹ and is unaffected by inactive phorbols such as 4-α-phorbol.

Although, as in other types of cells, a number of substrates for protein kinase C can be detected in cell-free extracts of bag cell neurones, it is unlikely that all of these proteins act as physiological substrates for the enzyme in intact cells. An alternative approach to identifying substrates in intact cells is to label intact clusters of cells with radiolabelled inorganic phosphate that is incorporated into ATP and is then transferred by the kinase into phosphoproteins. In such ‘pre-labelled’ cells, the enzyme may be activated by TPA and those proteins whose phosphorylation state is enhanced can subsequently be identified after separation by gel electrophoresis. Fig. 3 shows a two-dimensional autoradiogram of [³²P] phosphoproteins labelled in intact cell clusters. Fig. 4 shows enlargements of the region corresponding to molecular weights of ≈50000–190000 from control clusters and from clusters treated with 100 nmol l⁻¹ TPA. A dramatic increase in ³²P incorporation into a protein of apparent M_r 140000, labelled X, can be detected. This protein cannot be detected in control autoradiograms but appears as a relatively intense spot in autoradiograms from TPA-treated cells. A smaller change in incorporation may also occur in some other proteins (e.g. those marked B in Fig. 4), but the vast majority of phosphoproteins labelled in this homogeneous population of neurones are unaffected by TPA.

Because, in intact cells, activation of protein kinase C may be associated with indirect alterations of other protein kinases (Gilmore & Martin, 1983; Haycock, Browning & Greengard, 1985) it is not yet known whether the M_r 140 000 phosphoprotein is a direct substrate for protein kinase C. These data indicate however that, in contrast to results obtained using cell-free extracts, the number of phosphoproteins affected by activation of this enzyme in intact cells is likely to be rather small.

At the onset of an afterdischarge in bag cell neurones, there is an increase in the height of action potentials. An increase in the height of action potentials evoked by depolarizing current pulses is also observed when an isolated bag cell neurone maintained in cell culture is exposed to TPA at concentrations which activate protein kinase C in these neurones (10–100 nmol l⁻¹) (DeRiemer et al. 1985b). A similar effect is seen within minutes of exposure of the cells to 1,2 dioctanoyl glycerol (K. Bley & L. K. Kaczmarek, unpublished results). Inactive analogues, such as 4-α-phorbol, are without effect. The enhancement of action potential amplitude by TPA and the synthetic diacylglycerol can be mimicked by direct microinjection of purified protein kinase C into such isolated bag cell neurones (DeRiemer et al. 1985b). Both the activators of protein kinase C and the enzyme itself exert their effects without a significant change in the width of the evoked action potentials. The data are therefore consistent with the hypothesis that the phorbol ester exerts its effects through the activation of protein kinase C. A change in the shape of action potentials can, in theory, be brought about by changes in the properties of any one of the voltage-dependent ionic conductances which are activated on depolarization. The ionic basis of the enhanced action potentials has been examined in isolated bag cell neurones, using the whole-cell, patch-clamp technique to dialyse the cells internally with solutions of different ionic composition. When internal solutions which abolish all potassium conductances are used, the inward calcium current can be measured in these neurones without interference from voltage-or calcium-activated outward currents. Under these conditions TPA can be shown to induce a significant increase in the amplitude of the calcium current without any marked change in its voltage-dependence or kinetics (Fig. 5) (DeRiemer et al. 1985b). In contrast to the effects of activation of the cyclic-AMP-dependent second messenger system (see below), TPA has no effect on the voltage-dependent potassium currents in these neurones.

One interesting aspect of the modulation of the calcium current by TPA is its sensitivity to internal dialysis of the cells. Although TPA consistently increases the calcium current when cells are treated with TPA before internal dialysis is begun, and consistently enhances action potentials in cells that have been penetrated by microelectrodes, no effects of TPA are observed if TPA is administered after internal dialysis has begun. Internal dialysis may therefore disrupt some intracellular process which is required for an initial step in the increase of the calcium current by TPA but does not affect the current once it has become enhanced. One plausible hypothesis is that intracellular dialysis prevents the translocation of protein kinase C from cytoplasm to the plasma membrane. An enhancement of calcium current in a neurone may, in theory, be brought about either by some alteration in the properties of pre-existing calcium channels or by the recruitment of additional channels. This question has been examined in the bag cell neurones using cell-attached patch pipettes to record the activity of single calcium channels (Strong, Fox, Tsien & Kaczmarek, 1986). Fig. 6 shows recordings made in control and TPA-treated cells using 330 mmol l⁻¹ Ba²⁺ within the pipettes to amplify the currents flowing through calcium channels. In control cells, relatively small unitary currents, with an open channel conductance of ≈ 14 pS, are observed. This species of calcium channel appears to be sparsely distributed over the plasma membrane of the cell, but patches containing multiple channels are occasionally recorded, suggesting that these channels form clusters in the membrane. In contrast, TPA-treated cells contain, in addition to this small-conductance channel, a second class of channel with a larger unitary conductance (≈23 pS). This TPA-induced channel is found in 80% of patches recorded on cells exposed to 100 nmol l⁻¹ TPA but is rare or absent in control cells. The voltage dependence of the control and TPA-induced channels is similar. It appears likely therefore that the appearance of the covert calcium channel in response to TPA fully accounts for the enhancement of calcium current seen in whole-cell recordings. The actions of activators of protein kinase C on ionic currents in the bag cell neurones are distinct from those of any other second messenger pathway. The onset of afterdischarge is associated with an elevation of cyclic AMP levels (Kaczmarek, Jennings & Strumwasser, 1978). Activation of the cyclic-AMP-dependent protein kinase in these neurones results in a prolongation of action potential duration and repetitive firing through the modification of the properties of three distinct voltagedependent potassium currents (Kaczmarek & Strumwasser, 1984; Strong, 1984; Strong & Kaczmarek, 1986). These changes occur with no apparent alteration of calcium current. Conversely, TPA has no effect on voltage-dependent potassium currents (DeRiemer et al. 1985b). It is possible, therefore, that the cyclic-AMPdependent protein kinase and protein kinase C act synergistically to transform the electrical properties of the bag cell neurones at the onset of afterdischarge.

The high concentrations of protein kinase C in the nervous system relative to those in many other tissues suggest that this enzyme plays an important role in the regulation of neuronal activity. Studies using activators of this enzyme have recently begun and several effects on excitability have been described in different types of neurones. It remains to be determined whether these changes result directly or indirectly from alterations in the function of ion channels by protein kinase C and to what extent the activators of this enzyme also act on other, as yet unknown, metabolic pathways.

Photographs taken at the Discussion Meeting held at Titisee in March 1986