Muscarinic Response In Rat Lacrimal Glands

Conductance changes elicited by muscarinic agonists are one of the best studied examples of ‘slow synaptic responses’, which are characterized both by slow kinetics and by the intervention of several molecular processes between agonist-receptor binding and cell response (reviewed by Purves, 1978; Kehoe & Marty, 1980; Hartzell, 1981). It was shown 30 years ago that electrical effects of vagal stimulation on heart muscle appear with a marked delay (Del Castillo & Katz, 1955). Direct application of muscarinic agonists to heart muscle cells or to parasympathetic ganglion cells similarly produces delayed conductance changes with a lag of at least 100 ms (Purves, 1976; Hartzell, Kuffler, Stickgold & Yoshikami, 1977). Such effects mark a striking contrast with the fast nicotinic transmission found in vertebrate neuromuscular junctions and sympathetic ganglia, where the synaptic delay is less than 1 ms (Katz, 1969). Activation of muscarinic agonists also generally leads to an alteration of the cell metabolism, which is manifested at the level of the plasma membrane, where the turnover of phosphatidylinositol derivatives is increased (Berridge & Irvine, 1984), and in the cytoplasm, where the level of intracellular solutes such as Ca²⁺ or cyclic GMP is changed (Purves, 1978).

Recently, a great deal of effort has been devoted to understanding the conductance changes elicited by muscarinic agonists in various vertebrate preparations. Four broad types of response were found. In certain bullfrog sympathetic neurones, called B neurones, the main action of muscarinic agonists is to inhibit a special category of K⁺ channels responsible for the ‘M current’ (Brown & Adams, 1980; Adams, Brown & Constanti, 1982a,b). The same effect has also been observed in sympathetic neurones of vertebrates (Brown & Selyanko, 1985), in neurones of the central nervous system (Halliwell & Adams, 1982; Nowak & Macdonald, 1983) and in smooth muscle cells (Sims, Singer & Walsh, 1985). In heart muscle cells, on the other hand, muscarinic agonists increase the opening probability of another class of K⁺ channels which are already present at rest (Sakmann, Noma & Trautwein, 1983). These channels have faster kinetics than M current channels. Furthermore, the heart channels are activated by membrane hyperpolarization, whereas M currents are enhanced by depolarization. It is possible that the same channel as that described by Sakmann, Noma & Trautwein regulates the K⁺ conductance increase found in C neurones of certain bullfrog sympathetic ganglia (Dodd & Horn, 1983) as well as some recently described inhibitory responses of central neurones (McCormick & Prince, 1986; Egan & North, 1986).

A third effect of muscarinic agonists, also observed in heart muscle cells, is to inhibit Ca²⁺ currents. As a matter of fact, the long-held view that acetylcholine (ACh) decreases the Ca²⁺ currents observable at rest is now questioned by a recent study using isolated, dialysed heart muscle cells, which suggests that at least part of the current shift previously attributed to a decrease of Ca²⁺ currents is instead due to an increase of K⁺ currents (lijima, Irisawa & Kameyama, 1985). Nevertheless, after activation of Ca²⁺ currents with beta-adrenergic agonists, ACh clearly decreases the Ca²⁺ currents back to their resting level under conditions where K⁺ currents are effectively blocked (Breitwieser & Szabo, 1985).

It is not known whether the action of muscarinic agonists on the M current involves a second messenger mechanism. Experiments on cell-attached patches showed that the activation of the K⁺ current in the heart does not require any soluble cytoplasmic substance (Soejima & Noma, 1984). By analogy with other systems, it seems likely that the reversion of adrenergic activation of Ca²⁺ currents in the heart is due to a decrease of intracellular cyclic AMP, but direct evidence is still lacking. In the fourth main type of muscarinic response, on the other hand, there is a very strong case in favour of the involvement of a second messenger substance, in this case calcium. The best documented example is that of exocrine glands, where the calcium rise evoked by ACh leads to an increase of membrane conductance, to exocytosis and to intercellular uncoupling (Petersen, 1980). The conductance effect was recently shown to result from the activation of several classes of Ca²⁺-dependent channels, and it is thought to be the main factor regulating extrusion of electrolyte (Petersen & Maruyama, 1984; Marty, Tan & Trautmann, 1984).

Finally, yet other types of responses have been described, which do not readily fall into the above categories. In some neurones, activation of muscarinic receptors leads to an inhibition of Ca²⁺-dependent currents (North & Tokimasa, 1983; Cole & Nicoll, 1984; Pennefather, Lancaster, Adams & Nicoll, 1985). In smooth muscle cells, one of the actions of muscarinic agonists is to increase a cation-selective conductance pathway (Benham, Bolton & Lang, 1985). However, these effects have yet to be fully characterized.

The present paper focuses on the Ca²⁺-dependent currents found in exocrine glands. The main features of the channels and of their regulation by muscarinic agonists are reviewed. New results concerning the effects of ACh on the channels, the kinetics of the response and the influence of extracellular Ca²⁺ are presented.

Three types of Ca²⁺-dependent channels are found in exocrine glands (reviewed in Petersen & Maruyama, 1984; Marty et al. 1984).

Single-channel activity of the first type of channel, which we called the ‘BK channel’ for Big K⁺ channel, is illustrated in Fig. 1. The records were obtained in an outside-out patch from a cell isolated from a rat lacrimal gland. The internal side of the membrane was bathed with a high-K⁺ solution containing 10 nmol l⁻¹ Ca²⁺. The patch contained two channels, which displayed a marked sensitivity to membrane potential. Single-channel currents are remarkably large (Fig. 1B).

BK channels are found in salivary and lacrimal glands of rodents, and in pig exocrine pancreas (Petersen & Maruyama, 1984). Inside-out patch experiments demonstrate that BK channels are activated by very low concentrations of Ca²⁺, between 10 and 100 nmol l⁻¹ (Maruyama, Gallacher & Petersen, 1983; Findlay, 1984). Above 1μmoll⁻¹, Ca²⁺ blocks the channels in a voltage-dependent manner (Marty et al. 1984).

Single-channel currents due to BK channels may readily be identified in cell-attached experiments, owing to their characteristic single-channel conductance and voltage dependence. Comparison of the open state probability with results from inside-out patches indicates that the resting level of calcium is very low, of the order of 10 nmol l⁻¹ (Findlay, 1984). Upon application of ACh in the bath, a clear increase of the channel open state probability is observed (Fig. 2). The effects of ACh on the cell resting potential could be simultaneously estimated by following the size of the BK unitary currents. There was no general rule concerning the evolution of the resting potential. In the experiment shown, it seemed rather stable, at a value which was estimated at –60 mV by comparison with isolated patch results. In other cases,a hyperpolarization, or a depolarization followed by a hyperpolarization, were observed. Thus, the increase of BK channel activity is not a direct consequence of a membrane potential change. It is most probably due to an increase of intracellular Ca²⁺ ions.

In rat lacrimal glands, BK channels account entirely for the conductance of the secreting cells at rest. Due to the voltage dependence of the channels, cell l/V curves display a pronounced outward rectification (Trautmann & Marty, 1984). This cellular current, as well as the unitary currents shown in Fig. 1, may be blocked by bath application of 2 mmol l⁻¹ tetraethylammonium (Trautmann & Marty, 1984).

Ca²⁺-dependent Cl⁻ channels of low unit conductance were demonstrated in rat lacrimal glands (Marty et al. 1984). Owing to the small value of this conductance, the channels were mostly studied using noise analysis rather than direct, single-channel measurements. In cell-attached patches, application of the Ca²⁺ ionophore, A 23187, to the bath solution resulted in a noisy inward current response with occasional single-channel steps of about 0·4 pA at an estimated membrane potential of – 115 mV (Marty et al. 1984). We now report on new cell-attached experiments which address two questions. First, can the inward current response be stimulated by ACh as well as by the Ca²⁺ ionophore? And, if so, is this response only present in isolated cells, where luminal and basolateral membranes are exposed, or does it also occur when recording from cell clusters, which presumably expose only their basolateral membrane to patch-clamp pipettes?

At least two types of inward unitary currents which were likely to be due to Cl⁻channels were observed at hyperpolarized potentials. The first type consisted of well-defined single events of very small amplitude, around 0·1 pA near –100 mV (Fig. 3). The second type of event had a larger amplitude, of 0·3–0·6pA near –100 mV, and was accompanied by large, open-channel noise (Fig. 3). In many instances, the larger events seemed to break down into smaller steps, suggesting that the two types of events may have represented two gating modes of a common channel population. (Similar results were previously obtained with Ca²⁺ ionophore applications: Marty et al. 1984, p. 312.) The identification of the single-channel events of Fig. 3 as being due to Ca²⁺-dependent Cl⁻ channels rests on three arguments. First, the single-channel steps are very similar to the events previously described when stimulating with the Ca²⁺ ionophore. Second, their size varied qualitatively as expected from Cl⁻ channels, with a reversal potential somewhat above the cell resting potential.

(Note, however, that a clear reversal was obtained only with the larger type of channel.) Third, whole-cell recordings indicate that only Ca²⁺-dependent Cl⁻channels have a single-channel conductance compatible with the single-channel observations.

In the patch illustrated in Fig. 3, activity was so infrequent that it was impossible to ascertain whether ACh had any effect on the channels. In other examples, a clear channel activation was obtained, but it was then difficult to distinguish single steps as overlaps were the rule (Fig. 4). Such a response is observable both near –60 mV and near –120 mV in the experiments of Fig. 2 and Fig. 4A. All responses had a delay of about 1 s, and they outlasted the ACh response by 10–30 s, after which the currents returned to control values. The recordings of Fig. 4A and 4B show examples of single steps of about 0·1 pA which seem to merge into larger events.

Ca²⁺-dependent channels require a Ca_i level in excess of 100 nmol l⁻¹ for their activation and they are fully closed in a resting cell (Trautmann & Marty, 1984; Marty et al. 1984; Evans & Marty, 1986b). Cellular currents due to these channels have been studied in cells dialysed with a K⁺-free solution buffered with N-hydroxyethylenediamine triacetic acid (HEDTA) (Evans & Marty, 1986b). The currents are activated by depolarization, and display slow exponential relaxations in response to voltage jumps. They are blocked by high concentrations of furosemide (Evans, Marty, Tan & Trautmann, 1986).

A third type of Ca²⁺-dependent channel was described in pancreatic acinar cells by Maruyama & Petersen, who showed that it is activated by internal Ca²⁺ upon stimulation by cholecystokinin (Maruyama & Petersen, 1982a,b). This channel has a unit conductance near 25 pS (Fig. 5), is selective for monovalent cations and is largely independent of voltage. In lacrimal glands, activation of this channel requires micromolar concentrations of Ca²⁺ (Marty et al. 1984). It is therefore questionable whether ACh is actually able to raise the internal Ca²⁺ concentration to a high enough level to activate these channels. In fact, cell-attached experiments often failed to show such activation. That the channels were nevertheless present in some of these patches was indicated by the observation of single channels with a very low opening probability throughout the experiment. In some other cases (e.g. in Fig. 4C), ACh did open the channels, but the effects were limited to a single opening, or they were poorly reversible. These experiments are consistent with our previous suggestion that the cationic channels participate in the ACh response (Marty et al. 1984), but further work will be needed to ascertain the exact experimental conditions needed for their activation.

It would be obviously important to know whether any of the three types of channels are preferentially localized on the basolateral or on the luminal membrane.

Since K⁺-and cation-selective channels may be recorded on the superficial membrane of cell clusters which have retained an acinar structure, it is likely that these two channel types belong to the basolateral membrane (Petersen & Maruyama, 1984). As the experiments illustrated by Figs 2, 3 and 4A were performed on such aggregates, the same also applies to Ca²⁺-dependent Cl⁻ channels. However, it should be stressed that in all cases, the glands were subjected to partial dissociation using enzymatic and mechanical treatment. Even though care was taken to select aggregates with a compact appearance, the possibility exists that these treatments may have damaged the tight junctions and thus allowed luminal membrane proteins to diffuse to the basolateral domain. To sum up, the present evidence suggests, but does not prove conclusively, that all three types of Ca²⁺-dependent channels are located on the basolateral membrane. This is in line with the model of electrolyte secretion of Marty et al. (1984), which assumes that the basolateral membrane possesses the three kinds of channel. There is no direct information on the presence of channels on the luminal membrane.

The results of cell-attached experiments on isolated cells and on cell clusters are summarized in Table 1. Most patches showed a noisy inward current response, probably due to Cl⁻ channels. A few patches also displayed activation of BK channels and/or of cationic channels. No difference appears between isolated cells and cell clusters. The failure of 4/5 of the patches to show the BK response is most probably due to the comparatively low density of BK channels (one in 5–15μm², Trautmann & Marty, 1984), since the corresponding patches also failed to show any BK channel opening when the membrane was depolarized before ACh application. (Rather small pipettes, with an input resistance of about 5 MΩ in normal saline, were selected in these experiments, and a minimum amount of suction was employed, such that the patch areas were probably of a few μm².)

At low concentrations (typically 0·1μmoll⁻¹ ACh, or 0·5μmoll⁻¹ carbamylcholine, CCh) muscarinic agonists increase the voltage-dependent K⁺ current (Trautmann & Marty, 1984; Marty et al. 1984). This effect results from an increase in the opening probability of BK channels, rather than from recruitment of new channels (Trautmann & Marty, 1984). It can be blocked by increasing the internal Ca²⁺-buffering capacity (Trautmann & Marty, 1984; Marty et al. 1984). On the other hand, removal of all external Ca ions does not abolish the current (Trautmann & Marty, 1984; Marty et al. 1984). These experiments suggest that ACh activates Ca²⁺-dependent channels by releasing Ca²⁺ from internal stores.

Whereas low agonist concentrations only activate BK channels, stronger stimulations result in the additional development of a CP-selective conductance (Trautmann & Marty, 1984; Marty et al. 1984; Findlay & Petersen, 1985). This is due to the activation of the low unitary conductance, Ca²⁺-dependent Cl⁻ channels described above.

CK-selective channels may also be activated by stimulating the cell with the Ca²⁺ ionophore A 23187, or by dialysing the cell with a buffered high-Ca²⁺ solution (Marty et al. 1984; Evans & Marty, 1986b). Such studies showed that the Cam dependent Cl⁻ channel has a low unitary conductance (1–2 pS) and that it is activated by membrane depolarization. It is totally inactive at 10⁻⁷ mol l⁻¹ Ca²⁺, and entirely activated at 10⁻⁶moll⁻¹ Ca²⁺ or above (Evans & Marty, 1986b).

Some of the features of the electrical response to ACh are shown in Figs 6–8. Under standard recording conditions, the holding current measured at –60 mV is very small; depolarizing pulses to 0 mV elicit outward relaxations due to BK channels (Trautmann & Marty, 1984). Upon application of ACh, activation of BK channels leads to an increase of the outward current observed at 0 mV, and activation of Ca²⁺-dependent Cl⁻ channels leads to inward current observable at –60mV. [The current at 0 mV is entirely carried by K⁺-selective channels, as Cl⁻-and cation selective channels have their reversal potential at 0 mV under the conditions of Fig. 6. At –60 mV, on the contrary, contributions from K⁺ currents are small due to the proximity of the K⁺ equilibrium potential (–84mV). The main current carriers are Cl⁻ ions at this potential, since cation-selective channels have a high threshold for activation by Ca²⁺.] Both Cl⁻ and K⁺ current responses display damped oscillations with a period of 5–10 s (Fig. 6A). The oscillations are particularly marked at – 60 mV, where six periods may be recognized on the example shown. The holding current comes back to zero between successive waves, and it eventually subsides entirely in spite of the continued presence of ACh. At 0 mV, synchronous waves are observed, but their relative amplitude is less pronounced than at –60 mV. After 40 s of ACh application, the outward current is more than three times larger than before stimulation, and shows only a very slow decline. Even after 2min of continuous agonist application, a substantial residual activation was observed (not shown).

Inward tails which are observed after voltage jumps are also due to Cl⁻ currents. Comparison with whole-cell results with fixed internal Ca²⁺ solutions (Evans & Marty, 1986b) suggests that, during such waves, the internal Ca²⁺ level rises from about 0·lgmoll⁻¹ to nearly 1μmoll⁻¹. At the end of the stimulation, the inward current dropped to a very low level, indicating that Ca, had decreased to 0·1 μmol l⁻¹ or below. The fact that K⁺ channels were still activated suggests that Ca, was in fact close to 0·1 μmol l⁻¹.

In the experiment of Fig. 6B, ACh (1 μmoll⁻¹, as in Fig. 6A) was applied in a Ca²⁺-free solution containing 0·5 mmol l⁻¹ EGTA. As may be seen, a large K⁺ and Cl⁻ response was obtained. The differences between Fig. 6A and B are well within the range of cell-to-cell variations observed with either experimental condition. These results confirm our previous observations that external Ca²⁺ does not markedly affect the response (Trautmann & Marty, 1984; Marty et al. 1984).

However, it was reported that the whole-cell response of pig pancreatic cells to cholecystokinin, which has many similarities with the muscarinic response studied here, strongly depends on extracellular Ca ions (Maruyama & Petersen, 1984). To retest this point a different experimental approach was taken (Fig. 7). ACh was applied for a long period with a double microperfusion system (see figure legend). The same ACh concentration was placed in each microperfusion solution. One solution contained 1 mmol l⁻¹ Ca²⁺ and the other contained 0·5 mmol l⁻¹ EGTA instead. When the external solution was abruptly changed from one solution to the other, no discontinuity was observed in the evolution of the response (Fig. 7). We conclude from these experiments that external Ca²⁺ does not affect the muscarinic responses in our recording conditions. The reason for the difference from the response to cholecystokinin in the pancreas remains to be elucidated.

The results of Figs 6 and 7 contrast with those obtained with radiolabelled Rb⁺ efflux, where muscarinic agonists induce an early component which does not depend on extracellular Ca²⁺ and a late component which does (Putney, 1979; Parod, Leslie & Putney, 1980). From such evidence, it was concluded that the decline of the Rb⁺ efflux in Ca²⁺-free solution is due to an exhaustion of Ca²⁺ stores, which need external Ca²⁺ to be replenished. Direct comparison with the present experiments is difficult since the time span of flux studies is much longer than that of a whole-cell recording. Nevertheless, the decline of the currents in Figs 6 and 7 is probably not due to depletion of Ca²⁺ stores, since dialysis of the cell with inositoltrisphosphate elicits a response which lasts for many minutes (Evans & Marty, 1986a). On the other hand, unknown intracellular factors necessary for the effect of external Ca²⁺ may be lost in the whole-cell recording experiments.

The microperfusion system used in the present study allows very rapid changes of the bath solution. After a step application of ACh, a lag of about 1 s was always observed before the rise of Ca²⁺-dependent currents (Figs 6–8; Trautmann & Marty, 1984; Marty et al. 1984). The delay is longer for Cl⁻ currents than for K⁺ currents. If several applications of ACh are performed on the same cell, the delay is found to increase (Trautmann & Marty, 1984; Marty et al. 1984). The time required for the new external solution to equilibrate around the cell is of the order of 0-1 s, as estimated in control experiments using tetraethylammonium (TEA) as a blocker of BK currents. Thus, the delay is a genuine property of the response, and it must be attributed to the processes occurring in the cell membrane and cytoplasm between agonist binding and the final activation of ion channels by internal Ca²⁺.

It was also of interest to study the return of the current to the baseline after removal of ACh in order to gain insight into the last events governing Ca²⁺ release. This could not be achieved with a simple microperfusion system, and two separate microperfusions were used instead (Fig. 8). The results indicate that the Ca²⁺-induced currents return to the baseline with a time constant of 0·7 s. Again, this is longer than the delay for effective solution exchange as assayed in control experiments using TEA. The results of Fig. 8 suggest that the termination of the response is more rapid than its onset, provided that the agonist is removed quickly enough. The time of decay of the current is too short to represent the diffusion time of a water-soluble Ca²⁺-releasing substance such as inositoltrisphosphate (IP3, see following section) out of the cell. Possible rate-limiting steps for this decay are the elimination of IP3 by endogenous phosphatases or the rate of inactivation of the phosphodiesterase (phospholipase C) which produces IP3.

Previous work on permeabilized pancreatic cells and on isolated cellular fragments suggests that the endoplasmic reticulum is the store from which Ca²⁺ is released, and that 1,4,5-inositoltrisphosphate (IP3) is the agent directly responsible for Ca²⁺ release (Streb, Irvine, Berridge & Schulz, 1983; Prentki et al. 1984). When testing the effects of IP3 in our preparation, we found a rapid and sustained activation of Ca²⁺-dependent channels (Evans & Marty, 1986a). Furthermore, we examined the effects of various nucleotides on the ACh response and found that dialysis with GTPyS, a non-hydrolysable analogue of guanosine triphosphate, is able to potentiate the effects of ACh (Evans & Marty, 1986a). In addition, GTPγS alone was able to elicit occasional waves of Ca²⁺-dependent currents. These results confirm the role of IP3 as a Ca²⁺ releasing factor in the present preparation, and they provide evidence that a GTP-binding protein is involved in the production of IP3. The latter suggestion is in line with earlier findings on insect salivary glands and in blood platelets (Litosch, Wallis & Fain, 1985; Cockcroft & Gomperts, 1985).

It was shown recently that in the presence of GppNHp, another non-hydrolysable analogue of GTP, the muscarinic activation of K⁺ currents in the heart becomes irreversible (Breitwieser & Szabo, 1985). The same is true of the decrease of Ca²⁺ currents elicited by ACh in the same preparation. It was furthermore shown that the K⁺ current response is potentiated by GTP and blocked by pertussis toxin, an inhibitor of one of the G proteins characterized so far (Pfaffinger et al. 1985). Taken together, the results obtained in the heart and in exocrine glands suggest that all muscarinic responses may involve the activation of a G protein as an early step occurring next to the agonist-receptor binding. Some basic differences may, however, be noted between the two preparations. In the heart, non-hydrolysable analogues of GTP are ineffective in the absence of muscarinic agonists, whereas GTPγS elicits Ca²⁺-dependent currents by itself in exocrine glands. Second, GppNHp uncouples the receptor from the response in heart muscle cells, whereas Ca²⁺-dependent currents may still be regulated by ACh in the presence of GTPγS in exocrine glands. The exact molecular origin of these differences has yet to be characterized.

During the past 4 years, application of new electrophysiological methods has led to a detailed description of muscarinic responses in exocrine glands. The work was first devoted to characterization of the molecular components of the conductance modulated by ACh. Attention is now shifting to the kinetic properties of the response and to the chain of events taking place in the membrane, in the cytosol and in the endoplasmic reticulum, which links the agonist-receptor interaction to Ca²⁺ release. Application of patch-clamp techniques to these problems will undoubtedly be difficult since the accessible measurements (cell currents) are the final consequence of a complex set of reactions, which cannot easily be manipulated individually.

Fortunately, a reasonable working hypothesis may be put forward for this chain of events since it is now realized that Ca²⁺ release is the final consequence of a receptor-protein G interaction and of the subsequent production of IP₃ by phospholipase C. As expected from this broad scheme, it is possible to stimulate Ca²⁺-dependent currents by dialysing the cell with GTPγS or with IP₃. These new developments show that it is possible to enter the system at various levels, and they indicate what could be a step-by-step approach to the intracellular events put into play by muscarinic agonists.

This work was supported by grants from the Direction des Recherches Etudes et Techniques, the Centre National de la Recherche Scientifique and the Université Pierre et Marie Curie.