Epidermal Growth Factor-activated Calcium and Potassium Channels*

The earliest responses to activation of the epidermal growth factor (EGF) receptor include a transient increase in calcium influx and a transient membrane hyperpolarization. The underlying mechanisms are, however, not well understood as yet. In the present study, we have applied patch clamp recording in the cell-attached and the outside-out mode, and fluorimetric cytosolic Ca2+ determinations, to identify the na- ture of the ion channels involved, to characterize their properties at the level of single channels, and to un- ravel their mechanism of activation. We provide evidence that activation of the EGF receptor results ini- tially in the activation of voltage-independent Ca2+ channels that can be defined as direct receptor-oper- ated channels. This in turn causes the activation of Ca2+-dependent K+ channels, which results in a (de- layed) membrane hyperpolarization and then leads to the activation of a second class of Ca2+ channels that are sensitive to hyperpolarization. An autocatalytic generation of further hyperpolarization and Ca2+ influx is the predicted outcome of this ionic cascade. Based on the observed inhibitory effects of protein kinase C activation on the activity of Ca2+-dependent K* channels, we propose that protein kinase C is in- volved in the negative regulation of this cascade, which explains the transient nature of these responses.

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ T o whom correspondence and reprint requests should be addressed.
The monoclonal antibodies that can interfere specifically with the binding of EGF to either the high or low affinity class were used, have shown that the minor class of high affinity EGF-R (5-10% in A431 cells) is primarily involved in signal transduction (7). The earliest cellular responses to the activation of the EGF-R include a number of ionic changes, such as a transient plasma membrane hyperpolarization and increase in [Ca2+Ii, and a sustained rise in pHi (8)(9)(10). Some of these responses have been linked to the hydrolysis of polyphosphoinositides by phospholipase C, which is probably activated as a direct consequence of phosphorylation on tyrosine residues by the intrinsic EGF-R tyrosine kinase (11). As a result, the second messengers 1,2-diacylglycerol and Ins (1,4,5)P3 are released. 1,2-Diacylglycerol activates protein kinase C (12) that in turn is able to phosphorylate the Na+/H+ exchanger (13), causing a stimulation of this antiporter and thus a rise in pHi (14). It has been well established that Ins (1,4,5)P3 production is responsible for the release of Ca2+ from intracellular stores (15).
In A431 cells, the Ca2+ release constitutes only a minor component of the total increase in [Ca2+Ii, since it has been shown that the major part originates from EGF-induced Ca2+ influx (8,9,(16)(17)(18). In accordance, EGF causes only a minor production of Ins (1,4,5)P3 in these cells (19). This is in contrast with the effect of other agonists, like bradykinin. Addition of bradykinin to A431 cells results in rapid elevation of Ins (1,4,5)P3 levels and subsequent release of Ca2+ from intracellular stores (20). As yet, no convincing clues as to the mechanisms reponsible for the initiation of Ca2+ influx have been provided. EGF does not employ depolarization-activated CaZ+ channels to raise [CaZ+li because depolarization of A431 cells does not induce Ca2+ influx. Furthermore, the EGFinduced Ca2+ signal is not affected by nifedipine, a potent antagonist of depolarization-activated Ca" channels (9). The nature of the Ca2+ channel reponsible for the EGF-induced CaZ+ influx has thus still to be established. In this context, it is of interest to note that of all cellular responses to EGF checked for, Ca2+ influx is the only one that is impaired in A431 cells when EGF binding to the major class of low affinity receptors is prevented by pretreatment of the cells with monoclonal antibody 2E9 ( 7 ) .
Also, the mechanism underlying the observed hyperpolarization is not well understood. A431 cells hyperpolarize transiently upon addition of EGF from their normal V , of -65 mV to -85 mV (8,21). It was shown that this hyperpolarization was carried by K+ ions. Since it occurs concomitantly with the increase in [Ca2+Ii and can be mimicked by raising [Ca"], it has been assumed that Ca2+-activated K+ channels are responsible for this response (8,21).
In the present study, we have applied patch clamp recording and fluorimetric Ca2+ measurements on A431 cells to study the regulation of Ca2+ influx and hyperpolarization at the level of single ion channels. We show that activation of the EGF-R triggers an autocatalytic ionic cascade, in which a voltage-independent activation of Ca2+ channels is the primary response.
The resulting Ca2+ influx causes increased activity of (Ca"'dependent) K+ channels, thus leading to a hyperpolarization which in turn results in further Ca2+ influx due to activation of hyperpolarization-dependent Ca2+ channels. In addition, we provide evidence that protein kinase C activation serves as a negative feedback mechanism by a specific decreasing effect on K+ channel activity. This could explain the observed transient nature of the EGF-induced hyperpolarization and Caz+ influx.

EXPERIMENTAL PROCEDURES
Materiaki-EGF was purchased from Collaborative Research. Bradykinin was obtained from Sigma. In all experiments, the final concentrations of EGF and bradykinin were 100 ng/ml and 1 pM, respectively. These concentrations of EGF and bradykinin elicit maximum responses concerning both hyperpolarization and Ca2+ influx (e.g. Refs. 8 and 19).
Electrical Recording-For the patch clamp analysis (22) of calcium channels in the cell-attached configuration, Ba2+ was used as charge carrier to improve signal resolution, since this ion is a better permeant for these channels (23)(24)(25)(26). Composition of the patch pipette solution was (in mmol/liter) BaCl,, 96; HEPES, 10; adjusted to pH 7.4 with KOH (approximately 6 mmol/liter needed) at 33 "C.
Unitary inward BaZ+, Ca", and outward K+ currents were recorded through a patch clamp amplifier (our own design and construction). The patch pipettes were pulled from thin-walled borosilicate glass pipettes and were heat-polished.
Electrical recordings were continuously sampled at 200 kHz by a HP3565A digital signal analyzer. Averages of time records of 0.9 ms were stored by a computer (HP-9000-330 with the program HP-VISTA) on hard disc. The data were analyzed by the computer. To exclude false events, no automated single channel event detection was used. A second program separated Ba2+ and K+ currents. For each of both currents, the program generated plots of the current/ voltage relationship and plots of popen wrsw holding potential or time domain. Distributions of amplitude, open times, and closed times, fitted with their probability density functions, were produced for both types of currents (27) Measurements of [Ca2+]i-Nearly confluent A431 cells grown on glass coverslips were loaded with indo-1 (28) by incubating them with 2 WM indo-1-ester at 37 "C as previously described for quin-2 loading (9). Using a fluorescence spectrometer (Perkin-Elmer 3000), [Ca"], was determined, using an excitation wavelength of 355 nm (5-nm slit) and an emission wavelength of 405 nm (10-nm slit) according to standard procedures (9,28).

RESULTS
Characterization of K' and Ca2+ Ion Channels-Cell-attached patch clamp recordings on A431 cells, not exposed to EGF, reveal multiple channel openings. Their nature can be identified on the basis of their reversal potentials and on the effects of specific channel blockers. For studies of Ca2' chan-nels, Ba2+ was used routinely as a charge carrier in the pipette solution. Under these conditions, two types of ion channels were detectable ( Figs. 1 and 2a).
The predominant channel was identified as a K+ conducting channel on the basis of its reversal potential -85.6 mV (see Table I), the theoretical reversal potential being -86.4 mV in our configuration, as calculated from the Nernst equation (intracellular [K+] = 166 mM (9), pipette [K' ] = 6 mM). Furthermore, the reversal potential of this channel could be shifted to more positive values by adding K+-aspartate or KC1 to the pipette medium (Table I). The K+ channel showed burst-like gating kinetics (Fig. 2, d and f ) , its conductance was 18 pS.
The minor class of ion channels represented Ca2+ conducting channels, their conductance being 1-3 pS. We determined a reversal potential of 162 mV ( Table I) for these channels, which is close to its theoretical reversal potential of 178 mV ([Ca2+Ii = 200 nM, pipette [Ba2+] = 96 mM). These currents disappeared completely from our recordings when 10 mM Co2+ (a potent inhibitor of Ca2+ currents) was added to the pipette medium (Fig. 2b). The gating kinetics of this Ca2+ channel are shown in Fig. 2, c and e.

K' and Ca2+ Channel Activity Is Dependent on Membrane
Potential-The membrane potential is a common denominator of the activity of many types of ion channels. Under our conditions, the V,,, of A431 cells averaged -65 mV. An imposed depolarization of the patch significantly increased the popen of the K' channels ( Fig. 3b). This property of the K+ channel provides these cells with an efficient V , stabilizing mechanism. The Ca2+ channels could not activated by depolarizing the patch. On the contrary, it appeared that depolarization abolished completely all Ca2+ openings. However, at hyperpolarizing patch potentials, ranging from -70 mV to -85 mV, thep,,, of the Ca2+ channel was markedly increased. This increase was due to (on average) shorter time intervals between successive Ca2+ channel openings. The duration of the openings stayed the same. When the patch was hyperpolarized to -90 mV or even more negative values, the popen decreased again. The V,,, dependence of the Ca2+ channel is shown in Fig. 3a. These results demonstrate that a hyperpolarization of A431 cells will result in an activation of Ca2+ channels, while depolarization inhibits Ca2+ channels. EGF Induces Increased Activity of Both Ca2' Channels and K+ Channels-Earlier studies have shown that adding EGF to A431 cells results in a hyperpolarization (8) as well as a Caz+ influx (e.g. Refs. 9 and 29). Our measurements confirm these observations at the level of single ion channels. The bath application of EGF increased the popen of both K+ and   Table   I). A lag period of variable duration (10-100 s) before the onset of these effects was observed consistently. After this lag period, the popen of the K+ channels increased 2-%fold. Unitary Ba2+ currents showed a 1.5-10-fold increase in popen. The popen of both channels decreased again after 30-60 s and returned eventually to initial values (Fig. 4a). EGF did not affect the amplitude of the unitary K+ or Ba2+ currents (data not shown). The effect of EGF on the popen of the Ca2+ channels could be attributed mainly to a marked increase in the number of openings that showed a 2-10-fold increase per 10-9 period at maximum EGF effect. The EGF-induced increase in the popen of the K+ channels was, however, of a more complex nature. Firstly, EGF induced a (on average) longer duration of K+ channel openings (Table I). Secondly, EGF increased the time that K+ channels are bursting (we noted that the bursts lasted approximately 3 times longer). No effect on the time intervals between two successive K+ channel openings or bursts could be detected.
The increased popen of the K+ and Ca2+ channel evidently reflect the earlier observed EGF-induced hyperpolarization and Ca2+ influx, respectively. Subsequent experiments were carried out to establish the possible cause-effect relationships.
To test this hypothesis, we determined first whether the K' channels could be activated by an increased [Ca"+Ii in the absence of EGF. To this end, we elevated [Ca2+Ii by a stepwise increase of the extracellular Ca2+ concentration to 10-20 mM, or, alternatively, by adding bradykinin to the cells. Both stimuli transiently raised [Ca2'li, as was confirmed in fluorimetric studies. Both conditions resulted in an approximately 10-fold increase of the popen of K' channels ( Fig. 46), demonstrating that the EGF-induced hyperpolarization can indeed be caused by Ca2+-activated K' channels.
Our experimental set-up enabled us to see whether the activation of K' channels is solely mediated by the rise in [Ca2+Ii, or whether other factors are involved. To this end, the effects of adding EGTA (3 mM) to the bath medium were studied. EGTA was applied after formation of the gigaseal. The addition of EGTA did not produce any effect on our recordings (Fig. 5 ) . Importantly, subsequent addition of EGF did not activate K+ channels unless the normal extracellular Ca2+ concentration was restored. After such a restoration, a response to EGF was still present ( n = 6; Fig. 5 ) . The response to bradykinin, which is not influx-dependent, is not affected by extracellular EGTA ( n = 8). In control experiments, no significant effect on the popen of the K' channels was noted when EGTA was added nor when, subsequently, the original extracellular Ca2+ levels were restored (data not shown). These experiments show that the increase in the popen of the K+ channels induced by EGF is completely mediated by an elevation of [Ca2+];. EGF-induced Ca2+ Influx Is Dependent on Hyperpolarization-A prediction from the observed inhibitory effects of an imposed depolarization on Ca2+ channel activity would be that such a depolarization is able to inhibit the EGF-induced Ca2+ influx in A431 cells. That this is indeed the case was confirmed by comparing the effect of EGF on [Ca2+]; in normal bath medium versus a depolarizing bath medium (Fig.  6a). Under depolarizing conditions, the amplitude of the EGFinduced elevation of [Ca2+Ii was greatly reduced. Moreover, the time course of the EGF effect became markedly slower. This corresponded with our electrophysiological measurements (Fig. 66) under these conditions. Addition of EGF did still induce a higher popen of the (Ca2+-activated) K' channels (1.5-3-fold, n = 6), but with a significantly increased lag time (range 200-300 s versus 10-100 s in normal medium). The absence of a hyperpolarization did apparently block [Ca2+Ii influx partially. These results indicate that most of the EGFinduced Ca2+ influx is dependent on a hyperpolarization. EGF Activates Initially Voltage-independent Ca2+ Channels-The interdependent activation of Ca2+-dependent K+ channels and hyperpolarization-dependent Ca2+ channels in response to EGF provides an attractive autocatalytic mechanism to explain the generation of the observed hyperpolarization and Ca2+ influx, yet leaves the mechanism of initiation of these events unresolved. To obtain insight into this mechanism, we have applied the patch clamp method in the outside-out configuration. In this configuration, recordings are made from an isolated patch of plasma membrane of which the original outside is exposed to the bathing medium. This abolishes the possible effects of second messengers to a large extent, and thus allows the detection of possible receptoroperated ion channels (30). It appeared that bath application of EGF caused the immediate activation of a novel 10-pS conductance that could be identified as a Ca2+ channel. The observed reversal potential was 33 mV, as is its theoretical reversal potential (pipette [Ca"'] = 25 mM, bath [Ca"] = 2 mM). Subsequent addition of 23 mM CaC12 to the bath solution shifted the reversal potential to 0 mV (Fig. 7), as can be predicted from the Nernst equation. This Ca2+ channel showed no V,-dependent behavior. This type of ion channel could not be detected in the cell-attached recordings after bath application of EGF and may thus represent an EGF-Roperated Ca2+ channel. These findings indicate that EGF-R activation results initially in the induction of V,-independent receptor-operated Ca2+ channels. As a consequence, the activity of Ca2+-dependent K+ channels will increase, which will  lead to a hyperpolarization. This in turn will act on hyperpolarization-activated Ca2+ channels, and a further autocatalytic Ca2+ influx and hyperpolarization will result.
Involvement of EGF-R Tyrosine Kinase and Protein Kinase C-In the experiments in the outside-out configuration of the patch clamp technique, it was observed that EGF regulates Ca2+ channel activity in the absence of Mg2+ and ATP in the pipette solution. This might indicate that the regulation of ion channel activity occurs in the absence of phosphorylation. To establish further the relationship between EGF-R activation and the increases in ion channel activity, we have determined the effect of the EGF-R tyrosine kinase inhibitor tyrphostin AG 213 (a kind gift of Dr. A. Levitski, Hebrew University, Jerusalem, Israel). Tyrphostin AG 213 (100 M) antagonizes specifically tyrosine kinase activity of the EGF-R, but does not influence receptor dimerization (4). Preincubation of the A431 cells for 24 h with this inhibitor completely blocked all responses of EGF with respect to K+ and Ca2+ channel activity. This demonstrates that the ion fluxes of EGF are mediated by the tyrosine kinase of the EGF-R. Apparently, in the process of making outside-out patches, sufficient Mg2C and ATP remains associated with the patch membrane to sustain the kinase reactions involved.
The transient character of the EGF-induced hyperpolarization and Ca2+ influx requires the existence of an efficient negative feedback mechanism. Activation of protein kinase C is a possible candidate to exert such an effect, since such activation by phorbol esters is known to inhibit various responses to EGF, including the rise in [Ca2+Ii and the hyperpolarization (8,9,16,17). For that reason, we investigated the influence of 100 ng/ml TPA on the properties of the ion channels present in our preparation. TPA did decrease the popen of the K' channels, without affecting the Ca2+ channels. Within 2 min after the addition of TPA, we observed that the popen of the K+ channels was diminished to 20-40% of its initial value (Fig. 8 and Table I kinase C in A431 cells. Protein kinase C could thus play a role as a negative feedback mechanism in regulating the transient nature of the EGF-induced hyperpolarization and Ca2+ influx.

DISCUSSION
Earlier studies have categorized a transient hyperpolarization and Ca2+ influx among the first detectable responses of the activation of the EGF-R in A431 cells (8,9,(16)(17)(18). The underlying mechanisms are, however, not well understood as yet. In the present study, we have applied patch clamp recording in the cell-attached and the outside-out mode, and fluorometric [Ca2+Ii determinations, to identify the nature of the ion channels involved, to characterize their properties at the level of single channels, and to unravel the mechanism of activation. We provide evidence that activation of the EGF-R results initially in the activation of voltage-independent Ca2+ channels that can be defined as direct receptor-operated channels. This in turn causes activation of Ca2+-dependent K+ channels, which results in a (delayed) membrane hyperpolarization and then leads to the activation of a second class of Ca2+ channels that are sensitive to hyperpolarization. An autocatalytic generation of further hyperpolarization and Ca2+ influx is the predicted outcome of this ionic cascade. Based on the observed inhibitory effects of protein kinase C activation on the activity of the Ca2+-dependent K+ channels, we propose that protein kinase C is involved in the negative regulation of this cascade, which explains the transient nature of these responses (see also Fig. 9). These responses provide a satisfying explanation for the well known EGF-induced hyperpolarization and Ca2+ influx in A431 cells (8,9,(16)(17)(18) and confirm the suggested involvement of Ca'+-dependent K+ channels in these responses (8,21).
An important outcome of our experiments is the identification of the primary target of EGF-R activation. While cellattached patch clamp recordings showed the involvement of Ca2+-activated K+ channels and hyperpolarization-activated outside-out configuration revealed indeed the involvement of a separate class of receptor-operated Ca2+ channels as the likely primary target of the activated EGF-R. How the initial receptor-operated Ca" influx is evoked is not clear. Although a preincubation of A431 cells with the tyrosine kinase inhibitor tyrphostin AG 213 caused a full inhibition of EGF-induced K+ and Ca2+ channel activity in cell-attached experiments, the definitive proof that the initial activation of voltage-independent Ca2+ channels is mediated directly by the EGF-R tyrosine kinase awaits measurements in the outside-out mode on patches of cells bearing mutated EGF-R that lacks tyrosine kinase activity. Until then, other possibilities should be kept in mind. In this context, it is of interest that the Ca2+ influx is the only response in A431 cells that is inhibited by the monoclonal antibody 2E9 (7). This antibody specifically inhibits binding of EGF to the low affinity class of EGF-R. Apparently, the low affinity receptor population is involved in the activation of voltage-independent Ca2+ channels, because 2E9 blocks Caz+ influx completely (7).
The presence of an autocatalytic mechanism raises questions about the regulation of the transient nature of these responses. The observed effects of protein kinase C activation could offer an explanation. We were able to demonstrate that the phorbol ester TPA acts as a potent inhibitor of K+ channels in A431 cells. As a consequence, the cell will depolarize and this will terminate Ca2+ influx. Our finding that in A431 cells K+ channels are inhibited by protein kinase C is not surprising. The inhibition of K+ channels by protein kinase C is a well established phenomenon (31). Moreover, Pandiella et al. (8) had shown that A431 cells depolarize upon treatment with phorbol ester. Moreover, it has been reported that protein kinase C activation inhibits both the EGFinduced Ca2+ influx (9, 17) and the hyperpolarization (8), and that, after down-regulation of this enzyme, EGF induces a sustained rise of [Ca2+Ii (17). Since protein kinase C is stimulated by activation of the EGF-R (32), these data indicate that K+ channels are indeed inhibited by protein kinase C in these cells.
The results presented in this study provide insight into the mechanisms underlying the EGF-induced ionic responses. With the indentification of a self-invigorating mechanism, only a very limited number of receptors needs to be activated to cause both hyperpolarization and Ca2+ influx. It is likely that also in other nonexcitable cells such a mechanism may exist. In many nonexcitable cell types, it is reported that depolarization inhibited the receptor-operated Ca2+ influx EGF-controlled Ion Channels (e.g. T lymphoblasts (33), hepatocytes (34), parotid acinar cells (35), neutrophils (36), basophilic leukemia cells (37), and platelets (38)). Moreover, for some of these cell types, it was found that phorbol esters inhibited a rise of [Ca2+Ii. We speculate that in these cell types the receptor-operated Caz+ influx is mediated by a similar transactivation of hyperpolarization-activated Ca2+ channels and Ca+-activated K+ channels. It could well be that such an autocatalytic cascade stands as a model for receptor-operated ionic responses (including [Ca2+Ii oscillations) in many nonexcitable cell types. The mechanism of ion fluxes proposed here may thus have a physiological relevance which exceeds the EGF response in A431 cells.