Two Types of Action Potentials in A Secretory Epithelium of A Clam Mantle

An increasing number of studies of secretory tissues has revealed a considerable diversity in electrical response to stimulation. Many exocrine glands have been shown to be electrically inexcitable (Petersen, 1976; Ginsborg & House, 1980). Other secretory cells have been reported to be excitable: spontaneous as well as electrically- and chemically-elicited action potentials have been observed in adrenal chromaffin cells (Biales, Dichter & Tischler, 1976; Brandt, Hagiwara, Kidokoro & Migazaki, 1976) and in the adrenal cortex (Matthews & Saffran, 1973), and bursts of spikes have been observed as a response to glucose in pancreatic beta cells (Dean & Matthews, 1970; Matthews & Sakamoto, 1975). Among the invertebrates, propagated action potentials have been demonstrated in the salivary gland of a snail, Heliosoma trivolvis (Kater, Rued & Murphy, 1978) and in the rete mirabilis of Hippopodius (Mackie, 1976). Evidence suggests that the action potentials in these tissues initiate the secretory process, but further experimentation is still required.

Recently, Sorenson, Wood & Kirschner (1980) reported the presence of action potentials in the outer epithelium of the mantle of the clam. This epithelium, which secretes the shell, presents features that make it of interest for electrophysiological studies. It is a flat sheet consisting of a single layer of uniform cells, interconnected with gap junctions (Neff, 1972), and suitable for conventional microelectrode studies. The resting potential of its cells averages −60 mV and depends primarily on the external K concentration: a 10-fold increase in K concentration at the apical surface causes a 59 mV depolarization (at elevated levels of K) as expected for a potassium electrode. High extracellular Ca decreases the resting K permeability. Cl⁻ was found to be permeant and apparently distributed passively (Sorenson et al. 1980).

Little is known about the active electrical properties of this secretory epithelium. Our purpose here is to make a more complete description of these properties of the clam mantle. Preliminary reports of some of our results have been published elsewhere (Beirāo & Sorenson, 1981, 1982).

The central region of the clam mantle is composed of two epithelia separated by an interstitial space containing haemolymph (Fig. 1C). The inner epithelium faces the animal’s mantle cavity and is normally bathed by sea water. It is a single layer of ciliated, non-ciliated and gland cells. The outer (shell-facing) epithelium, responsible for shell secretion, is bathed by a thin layer of an extrapallial fluid that exists between the mantle and the shell (the extrapallial space) whose composition is very similar to haemolymph (Crenshaw, 1972). The outer epithelium of the clam Mercenaria mercenaria is a uniform single layer of columnar cells whose morphology shows seasonal variation. During the period of maximal shell secretion they are tall and narrow with an average height of 30 pm. The membrane that faces the shell (apical membrane) shows densely packed microvilli, 2-3pm long. Tight and gap junctions join each cell to its neighbour near its apical border. The cells show some degree of lateral separation below the apical junction. The lateral and basilar membranes (here referred to as basolateral membrane) are convoluted and are bathed by the haemolymph of the interstitial space (Neff, 1972).

Specimens of Anomalocardia brasiliana were obtained from Guanabara Bay in Rio de Janeiro, or sometimes from Sāo Vicente Bay near Santos. These animals were kept either in small containers or in a 100·1 balanced aquarium with aerated sea water containing plankton and algae as a food source. Animals stored in small containers showed signs of loss of vitality after about 2 weeks and were discarded; animals kept in the large aquarium lasted well for several months. Of the animals used in this paper, the longest storage period was 43 days.

An animal was selected for use only if it actively pumped water when undisturbed and if it closed its valves quickly when handled (Sorenson et al. 1980). It always offered a strong opposition to opening its valves. The adductor muscles of the animal were cut, the valves were opened and its visceral parts removed. The largest possible area of the mantle was cut free from the adductor muscle and was carefully detached from the shell at the pallial attachment near its border. The fragment thus obtained had an area of about 1 cm². It was then removed and placed in the experimental chamber.

Two different chambers were employed. In one method, the mantle was pinned down onto a Plexiglas chamber with a soft plastic bottom, the secretory epithelium facing upward. Isolated current pulses were delivered by suction electrodes filled with normal saline or by stainless steel electrodes pressed lightly against the mantle. In some instances the stimulation was carried out with an intracellular glass microelectrode filled with 2 mol 1⁻¹ K-citrate, whose intracellular position could be verified by switching it to the input of the electrometer preamplifier. Action potentials and resting potentials were recorded by conventional techniques (Sorenson et al. 1980) (Fig. 1A).

In the second method, the mantle was mounted between two compartments of Plexiglas which were separated by two thin plastic plates (3·5×5 cm) with coincident windows (1 mm diameter). The mantle was placed on one of the plates (previously smeared with Vaseline), covering its window. A small piece of filter paper with a 6-mm diameter hole was often used to keep the mantle extended. The other half of the chamber, with its plate also smeared with Vaseline, was then used to clamp the mantle between the two plates, thus shutting off the only pathway between the two compartments (Fig. 1B).

Ionic substitutions could be made independently in either compartment, the anterior one facing the outer (secretory) epithelium and the posterior one facing the inner epithelium. Isolated current pulses were passed across the mantle via Ag/AgCl electrodes and the transverse resistance of the mantle could be calculated by measuring the voltage drop across it. Upon stimulation, the 0-008 cm² exposed area of the secretory epithelium could be activated and ‘membrane’ action potentials could be recorded by an intracellular microelectrode.

Resting and action potentials were recorded with conventional glass microelectrodes pulled from 0·7 mm tubing, filled with 3 mol 1⁻¹ KC1 and with resistances of 8–30MΩ They were connected via an Ag/AgCl electrode to high input impedance capacity-compensated preamplifiers (Wide Band Electrometer, BAK,

Instruments for Life Sciences, Rockville, Md; Model NF1 amplifier modified for unity-gain, Bioelectric Instruments, Hastings-on-Hudson, NY). Recordings were displayed and photographed upon Tektronix oscilloscopes (Tektronix, Beaverton, Oregon) models 5111 and 564B. Resting potentials were recorded on a digital voltmeter (ECB-Equipamentos Cientificos do Brasil, Sāo Paulo, SP). Grounding was achieved with an Ag/AgCl electrode connected to the bath through a 0·5 mol 1⁻¹ KC1 bridge.

In some experiments differential recording between two microelectrodes was performed using unity-gain preamplifiers fed into the differential inputs of the oscilloscope. The microelectrodes could be placed either inside or outside the cell, and in the latter case they could be either near the basolateral membrane (in the interstitial space) or near the apical membrane. A reliable placement of these electrodes was obtained by the following procedure, (i) To place the microelectrode near the apical membrane, it was moved toward the mantle until a small deflection due to touching the membrane was observed, and it was then withdrawn by about 50/xm. (ii) To place the microelectrode inside the cell, instead of withdrawing after the initial deflection, a small tap on the electrode holder or advance of the electrode resulted in a sharp drop of the potential toward stable negative values. After the experiment, the electrode was slowly withdrawn, causing a loss of the measured potential to zero, (iii) The electrode was considered to be in the interstitial space when, once inside the cell, it was further advanced until the potential suddenly approached near-zero values (usually less than 4mV). After the experiment, the electrode was slowly removed, causing a sharp decrease of the potential to the vicinity of the resting potential values. A further withdrawal caused a return to zero. If this sequence was not observed, the experiment was abandoned.

The ratio of the resistance of the basolateral membrane to the resistance of the apical membrane (Rb/Ra) was estimated by the voltage drop measured across the apical and the basolateral membranes, when pulses of 0·2-1·0μA were passed across the mantle (Froemter, 1972).

The mantle was bathed in normal saline whose composition, similar to the extrapallial fluid (cf. Crenshaw, 1972), was (in mmol 1⁻¹): NaCl, 360; KC1, 10; CaCl₂, 10; MgCl₂, 50; Na₂SO₄, 25; and Tris-maleate buffer, 5 (pH 7·2). In 0-Ca saline, Ca was replaced by Mg, and in 0-Na saline, Na was replaced by either choline or tetramethylammonium. In some solutions, HEPES buffer was used with similar results. All reagents were analytical grade. All experiments were carried out at room temperature (21–26°C).

For measuring parameters of action potentials, three lists of the recorded action potentials were prepared: one of 24 clams that showed spike-like action potentials, one of 39 clams with cardiac-like action potentials, and one of 10 clams with cardiac like action potentials obtained in the two-chamber method. By random sampling, five action potentials of different specimens were taken from each list and their overshoot, duration, maximal rate of rise, and rates of repolarization were measured with transparent graph paper and transparent ruler. The duration of the action potential was measured from the upstroke to the point of intersection of the tangent fitted to the final phase of rapid repolarization with the resting potential level (Niedergerke & Orkand, 1966; Gadsby & Cranefield, 1979). The calculated average and standard deviation of each group are unbiased estimates of the population of each kind of action potential (Steel & Torrie, 1960).

When data obtained from different mantles were compared, two-tailed t-tests were used; a probability level of 0·05 was used as a criterion for statistical significance.

Most of the data shown are typical experiments from five or more. The experiments shown containing a comparison between control and experimental procedures were always performed in the same mantle and the same impalement. They were followed by a return to control conditions to rule out the possibility of irreversible effects.

Intracellular injection of depolarizing current pulses could elicit a regenerative response (Fig. 2B) with the characteristics of an electrically excitable one, for it continued but little changed during a constant current flow. Furthermore the response was independent of the stimulus strength and duration, with a threshold between −40 and −35 mV. Finally, the action potential propagated in a non-decremental fashion throughout the epithelium with an average conduction velocity of 15·5 ± 2·5cms⁻¹(N = 17).

The injection of hyperpolarizing current pulses or subthreshold depolarizing pulses demonstrated an electrotonic spread of the passive voltage transients (Fig. 2A). These responses decreased markedly with the distance from the current electrode. Preliminary experiments using the procedure described by Josephson & Schwab (1979) and carried out on four mantles (17 measurements) gave an estimated value of 260 µm for the two-dimensional space constant.

According to their active electrical response, two populations of clams could be distinguished. In one group, stimulation of the mantle elicited action potentials with a prominent plateau, resembling some cardiac action potentials. In the second group, the plateau was diminished or missing altogether. We have correlated this difference with the state of the animal with regard to shell secretion (see below).

A typical propagated cardiac-like action potential of a freshly collected clam is shown in Fig. 3A. The average values of its overshoot, duration and rates of rise and repolarization are given in Table 1. Repolarization took place in three phases, as for cardiac action potentials (Cranefield, 1975). Phase 1, the initial rapid depolarization, occurred at an average rate of 9·3 ±4·8 Vs⁻¹. During the plateau (phase 2) repolarization had slowed to 1·4 ± 0·7 Vs⁻¹. This was followed by a phase of faster repolarization (4·7 ± 1·1 Vs⁻¹, phase 3). A hyperpolarizing afterpotential was sometimes present in mantles which were presumably somewhat depolarized. Finally, the action potential was followed by a long-lasting (15 s) depolarizing afterpotential of small size (3·0 ± 0·5 mV).

Fig. 3B shows a typical propagated spike-like action potential, obtained with the same procedure as used in Fig. 3A, but from the mantle of a clam kept for 2 weeks in a small container of aerated sea water. The overshoot and rate of rise were not statistically different from those of the cardiac-like action potentials, but a much shorter duration of the action potential was evident, lasting but 11·1 ± 1·4ms. Repolarization took place at a single rate, the plateau being absent (Table 1).

When the mantle was placed between two chambers and current passed across it, the ‘membrane’ action potential generated showed essentially the same features as the propagated action potentials (Fig. 3C for cardiac-like and Fig. 3D for spike-like responses). The rate of repolarization during the plateau (phase 2) was more rapid than that of the propagated response. In these experiments, short (0·10–0·15ms) pulses were used to minimize the artifact generated in the recording electrode due to the current pulse.

Clams exhibiting a prominent plateau phase were mostly freshly collected clams and those kept for up to a few weeks in a balanced aquarium. Their shells usually had glistening and smooth inner surfaces and the mantles were clean. The second group, with spike-like action potentials, consisted of animals kept for longer periods in the aquarium or for a few weeks in small containers, but a few freshly collected animals could be placed in this category. Their inner shell surfaces were chalky and rough, with an etched appearance. Their mantles usually had a white membranous plaque covering most of the surface. Of the 24 mantles with spike-like action potentials, five were from freshly collected clams.

Mantles of 39 specimens with cardiac-like action potentials had an average resting potential of 54·8 ±5·4 mV, while 24 specimens with spike-like action potentials had a significantly greater resting potential of 62·5 ± 4·9 mV (Table 1).

Differences were also found in the steady-state voltage response to current (Fig. 4). Mantles from clams with cardiac-like action potentials displayed a linear relationship between current and voltage up to the threshold value of the action potential (Fig. 4A). On the other hand, mantles from clams with spike-like action potentials gave a non-linear response (Fig. 4B). No attempt was made to compare the two types of mantles with exactly the same electrode spacing.

The two-compartment method was used to verify the existence of a refractory period. Twin pulses of the same amplitude were used and the time between them was varied in such a way that the first pulse would generate a normal action potential and the second one would stimulate the epithelium at different times during and shortly after the first one. Currents in these experiments were greater than five times that usually needed to elicit action potentials. Fig. 5 shows that during the action potential a second stimulus was unable to elicit a second response, a characteristic of refractory periods. In five spike-like action potentials the refractory period lasted 12·8 ±1·8 ms. In cardiac-like action potentials it was mostly dependent on the duration of the action potential. Normal cardiac-like action potentials could only be elicited by a second stimulus which occurred 5–10 ms after the end of the first response. Another important feature is that stimuli given at later parts of the plateau elicited a small and slow response with about the same duration but with a slightly smaller amplitude than the plateau (Fig. 5A). When the same procedure was carried out using a mantle with spike-like action potentials, the presence of a refractory period was also observed, but no such slow response could be obtained, suggesting that the slow response observed during the refractory period of the cardiac-like action potential was not merely an attenuated version of the rising phase (Fig. 5B).

Under certain conditions, the plateau of the action potential was inhibited without much effect on the overshoot or rate of rise. When Ca in the bathing solution was replaced with Mg, the plateau was reversibly inhibited within 10–15min (Fig. 6), suggesting the existence of a Ca-dependent mechanism important for the generation of the plateau.

The plateau and duration of the action potential were progressively reduced by repetitive stimulation as seen in the fifth and tenth action potentials in Fig. 7A.

A small decrease of peak amplitude and rate of rise can also be seen. With stimulation at frequencies higher than 1 Hz, action potentials occasionally failed to be elicited; the subsequent action potential would then show partial recovery from the inhibitory effect. It is noteworthy that the resting potential showed little depolarization during repetitive stimulation. Mantles with spike-like action potentials, or mantles in solutions without Ca, were insensitive to repetitive stimulation.

There was no summation of the negative afterpotential (Fig. 8B), its amplitude remaining unchanged during and soon after a train of pulses.

Substitution of 0-Ca saline for normal saline had a significant effect on the plateau, but almost no effect on the spike (Fig. 6). Substitution of 0-Na saline for the normal solution also had no significant effect on the spike, even when the epithelium had been soaked for 40 min. Simultaneous replacement of both Na and Ca had the same effect as replacement of Ca alone. These observations suggest that the spike of the action potential is generated in a membrane that faces a compartment not readily accessible to the ionic substitutions.

The site of action potential generation was examined by recording alternately across the basolateral and apical membranes. The recording electrode was intracellular; the reference electrode was placed either near the outer face of the secretory epithelium, or across it, in the interstitial space (see Materials and Methods for details). The records obtained in both recording situations were very similar, the only possible difference being that the overshoot of the action potential recorded across the apical membrane was consistently about 4 mV smaller than that recorded across the basolateral membrane (Fig. 9).

The mantle was then clamped between two compartments and current pulses of 0·2–1·0μA were passed across it. When the reference electrode was placed extra-cellularly near the apical membrane and the recording electrode was inside the cell, the voltage drop across the apical membrane (ΔVa) averaged 500 ± 120mVμA⁻¹(N = 10). When the recording electrode was placed in the interstitial space, the voltage drop across the epithelium was 620 ± 140mVμA⁻¹(N = 10). The average drop across the basolateral membrane (ΔVb) was 120 ± 40 mV μA⁻¹(N= 10). The average of the ratios ΔVa/ΔVb obtained in each of the 10 experiments was 4·6 ± 1·7, which is the ratio of the resistances Ra/Rb (Froemter, 1972). Thus, for a given current that crosses the epithelial cell, the voltage transient in the apical membrane is, on average, 4·6 times greater than that in the basolateral membrane.

Stronger current pulses were then passed in an attempt to elicit action potentials. Currents that depolarized the basolateral membrane (and hyperpolarized the apical membrane) were effective in eliciting an action potential (Fig. 10A). Depolarization of the apical membrane did not produce an action potential unless the pulses were ten times longer and even then the action potential arose as an anodal break (Fig. 10B). We conclude that the basolateral membrane is the excitable one.

The voltage drop across the secretory epithelium gives its resistance. Assuming that no significant current crossed the epithelium in any part other than the area that covers the window, the transepithelial resistance is calculated to be 5·0 ±1·1 Ωcm². This is a very low value and therefore the secretory epithelium can be viewed as a ‘leaky’ one. The resistance across the whole mantle is nearly as low (13·4 ± 1·4Ωcm²). This explains why an electrogenesis occurring in the basolateral membrane can be easily recorded between the inside of the cell and the outer bath.

The experiments described in this paper show that the shell-facing epithelium of the mantle is electrically excitable, i.e. that it generates an all-or-nothing propagated action potential, that a threshold exists, and that the generated action potential exhibits a refractory period.

Many other epithelia exhibit propagated action potentials (Anderson, 1980). In hydrozoan coelenterates, epithelial conduction of impulses has been observed in many species and these impulses seem to be important for the control of movements such as swimming, crumpling and tentacle contraction (Mackie & Passano, 1968). In amphibian tadpoles, as well as in tunicates, epithelial propagation of impulses is related to sensory functions and participates in protective reflexes (Roberts, 1971; Bone & Mackie, 1975). Epithelial conduction exists in tadpoles as a provisional means of conducting sensory impulses, thus providing them with the ability to respond to external stimuli before the skin receives its definitive innervation (Roberts & Smyth, 1974). Mackie (1970) has suggested that epithelia preceded the nervous system in the conduction of impulses.

Non-neural functions of epithelial action potentials have also been reported. Mackie (1976) described a relationship of epithelial action potentials and secretion in

Hippopodius rete mirabilis, in which bursts of action potentials are followed by the appearance of secretion droplets on the surface of the epithelium. This was correlated with the presence of a secondary depolarization of long duration. A similar phenomenon may be present in the clam mantle, as a long-lasting depolarizing afterpotential was observed. However, an important qualitative difference between the two polarizations is that, in the clam mantle, there is no cumulative or facilitatory effect of the afterpotentials.

A distinct feature of this action potential is that it can occur as a spike or with a plateau whose duration is variable in different specimens. Mantles whose action potentials have an evident plateau have, on average, lower resting potentials than those without plateaus, but the presence of the plateau cannot be explained by the lower resting potential. An increase of K concentration to 20 mmol 1⁻¹, although leading to a depolarization of 8 mV, does not increase or induce the plateau and further depolarization often decreases it (data not shown). Furthermore, removal of Ca in the bathing medium and repetitive stimulation promote a marked reduction of the plateau without hyperpolarizing the cells.

The absence of the plateau cannot be attributed to a degeneration of the animals’ condition during storage because: (a) some (21%) of the clams that showed spike-like action potentials were freshly collected; (b) although the storage conditions were artificial, many animals lived for up to 8 months under those conditions, whereas 43 days was the longest time of storage for the animals used in the experiments described in this paper; (c) the clams were selected for the experiments only if they actively pumped water, closed their valves quickly when handled and opposed strongly the opening of their valves.

Our observations suggest that the presence or absence of the plateau is related to the secretion of the shell, which is not a continuous process. Under conditions of active growth, an alternation of shell secretion and shell resorption may occur. This turnover of shell material probably takes place when the valves are closed and may be looked upon as a normal aspect of the metabolism of the clam (Crenshaw & Neff, 1969). During long periods of adverse environmental conditions, in which the clams keep their valves closed for long periods and engage in anaerobic metabolism, shell resorption may well predominate over growth. Seasonal cycles of growth and shell resorption have also been reported (Wilbur, 1972). Dugal (1939) and Wilbur (1972) have reported that the chalky and etched appearance of the shell and the presence of white plaques on the mantle are typical of clams resorbing shell. These mantles always show spike-like action potentials. On the other hand, clams which are actively secreting shell have glistening and smooth inner shell surfaces and no plaques on the mantle (Wilbur, 1972). These mantles show cardiac-like action potentials. We conclude, therefore, that the plateau occurs in clams that are secreting shell, and we suggest that the action potential participates in the control of secretion.

The control of secretion is being investigated in many laboratories, and several glands have been shown to be controlled by action potentials. In several glands the participation of Ca as a trigger for the secretory process has been revealed (Ginsborg & House, 1980). It may not be fortuitous that the plateau of the action potential of the mantle is a Ca-dependent process. We are currently investigating this aspect of its ionic mechanism.

The plateau of the action potential may result from a retarded repolarization or from a long-lasting inward current, or both. Whatever the mechanism, it must explain the fact that experimental modifications of the plateau can occur without significant interference with the spike of the action potential, for example in the experiments of Ca removal and of repetitive stimulation. This means that the spike is essentially independent of the conditions which determine whether the plateau exists or not.

On the other hand, Fig. 5 shows a typical experiment in which a slow depolarizing response is not preceded by a spike. This slow response is very similar to the plateau. The occurrence of a plateau without a preceding spike may mean that the plateau has a mechanism distinct from that of the spike and is not just a prolongation of it; otherwise, inhibition of the spike should also block the plateau. It might also take place at a different site.

Omission of Ca from the solution bathing the apical membrane caused a reversible inhibition of the plateau. The rate of rise and the overshoot of the action potential, on the other hand, proved to be insensitive to the replacement of sodium in the external solution. This suggested that the apical membrane is not important for the electrogenesis of the spike. The results presented in Figs 9 and 10 are strong evidence that the basolateral membrane is the site of the electrogenesis. This membrane faces the interstitial space, a compartment not readily accessible to changes in the ionic composition of the extrapallial solution, which provides the explanation for the observed insensitivity of the spike to these changes in the external solution.

The effect of Ca on the plateau suggests that there is a separate Ca-dependent inward current taking place at the apical membrane, but we cannot yet rule out other possible indirect effects of Ca.

Repetitive stimulation at 1 Hz led to a reversible decrease of the plateau. It may be that the mechanism of the plateau inactivates during repolarization and that the removal of this inactivation takes longer than 1 s to be accomplished. This possibility can be ruled out, because when two stimuli are given at a 1-s interval, the second action potential is not very different from the first. With shorter intervals, the spike of the second action potential is much more affected than the plateau (Fig. 5). On the other hand, the experiment of Fig. 7 shows that there is a significant difference between the fifth and tenth action potentials, although they have in common the fact that they are preceded by Is by another action potential.

All experiments showed that there is a progressive decrease in the plateau. This implies a cumulative effect. It is not due to a summation of the negative afterpotentials, which was shown not to occur (Fig. 8). Furthermore the depolarization induced by the afterpotential is too small to account for the modification induced in the plateau as discussed previously. The results suggest, therefore, an accumulation, or depletion, of a substance as the cause of the decrease in the plateau.

The authors are indebted to Dr H. Masuda for his generous supply of clams. This work was supported in part by grants 40. ·1526/80 and 30·0303/78-BF-07 from the Conselho Nacional de Desenvolvimento Científico e Tecnológico-CNPq, and by Convenio FINEP-Bioquimica 323.