Propagation of Action Potentials Through Electrotonic Junctions in the Salivary Glands of the Pulmonate Mollusc, Helisoma Trivolvis

The diagnostic feature of electrically excitable systems is a cell membrane responsiveness characterized by voltage dependent, specific ion conductances of the sort described by Hodgkin & Huxley (1952). Classically, such excitable systems were thought to be composed of nerve and/or muscle cells. The increasing sophistication of microelectrode recording techniques and the application of these techniques to a broader spectrum of tissues has made us aware that a diversity of cell types are electrically excitable, e.g. cells in the tadpole skin (Roberts & Stirling, 1971), certain non-neural endocrine tissues (Matthews & Saffran, 1973; Matthews & Sakamoto, 1975), and ciliated protozoa (Naitoh & Eckert, 1969). The present communication is concerned with the electrophysiological properties of exocrine gland secretory cells.

Secretory cell types may be classified into those that display regenerative electrical responses like neurones, and those that have passive membranes with no known voltage dependent ion conductances. The former group includes the cells of the rete mirabile of a coelenterate (Mackie, 1976), mammalian endocrine cells in the adrenal cortex (Matthews & Saffran, 1973), and the β cells of the mammalian pancreas (Matthews & Sakamoto, 1975)-Into the latter group fall all previously investigated pxocrine tissues such as mammalian salivary glands (e.g. Lundberg, 1955; Petersen, 1976), insect salivary glands (e.g. Berridge, Lindley & Prince, 1975; House, 1973; Loewenstein & Kanno, 1964), and mammalian pancreatic acinar cells (Petersen & Ueda, 1975).

This study is one of a series of investigations (Kater, 1977; Kater, Murphy and Rued, 1977) on glandular secretion in the Mollusca, a phylum previously unexplored in this respect. In these studies, it is demonstrated that the cells of gastropod exocrine glands (i.e. the salivary and pedal glands) are electrically excitable. These are the first electrically excitable glandular cells reported in an animal of the protostome line. In addition, this communication examines another important feature of the salivary glands of the snail, Helisoma trivolvis; electrical coupling via low-resistance electrical junctions. Such communication is widespread throughout the Metazoa (McNutt & Weinstein, 1973; Satir & Gilula, 1973; Staehelin, 1974). Though numerous physiological functions of electrical coupling have been proposed (Loewenstein, 1968; Bennett, 1973), the most unambiguous of these is to serve in the conduction of electrical signals between excitable cells such as neurones and muscle cells. Propagation of electrical signals via low-resistance junctions occurs in epithelia of hydrozoan coelenterates (Mackie, 1976) and amphibian tadpoles (Roberts & Stirling, 1971). We report here the propagation of regenerative action potentials throughout the salivary glands of HeHsoma trivolvis.

Salivary glands were dissected from laboratory stocks of the pulmonate snail, Helisoma trivolvis. The dissection procedure was as reported previously (Kater & Kaneko, 1972). Typically we used adult snails (12−16 mm vertical shell diameter) whose salivary glands were linked to one another by connective tissue at the distal ends (Fig. 1). The glands were cut free from the buccal mass leaving small pieces of muscle around the proximal end of the salivary duct. Depending upon the experiment, the oesophageal trunks (from which arise the salivary nerves, Fig. 1) were either cut, or were left intact, maintaining the connexions with the buccal ganglia. Miniature insect pins were placed through the bits of muscle to hold the glands linearly and securely pin them to a Sylgard pad. The Sylgard pad was inserted into a Lucite recording chamber similar to that used by Thomas (e.g. 1972), for a continuous flow of physiological saline. Saline was fed into the chamber via a gravity flow system and was removed by aspiration. An agar bridge was placed in the bath. Both dissections and electrophysiological recordings were performed in physiological saline (pH 7·3) containing (mm); NaCl, 51·3; KC1, 1·7; CaCl₂, 4·1; MgCl₂, 1·5; and Tris Cl, 5·0. The preparation was maintained at room temperature (21 °C ± 2°) for electrophysiological recordings.

Conventional electrophysiological recordings, stimulation, and display techniques were used. Glass fibre-filled micropipettes (2 mm outside diameter) were filled with 3 M potassium acetate (15−50 MΩ resistance). These electrodes were connected via a silver/silver chloride wire to a dual channel, high input impedance, capacity compensated, unity gain amplifier equipped with a bridge circuit for injecting current. The output of the amplifier was displayed on an oscilloscope or a Brush 220 chart recorder. Electrode penetrations were made under visual control by use of a Wild dissection microscope. Current was monitored using a virtual ground system (e.g. Moore, 1971). For critical electrode bridge balance, such as that needed for electrical coupling measurements, the bridge was balanced with the electrode inside a cell. This was accomplished with brief ( < 1 ms) current pulses which did not charge the membrane and thus allowed only electrode impedance to be nulled. This was repeated for each value of current injected. Intracellular staining was obtained by injecting the dye, fast green, iontophoretically (Thomas & Wilson, 1966; Rovainen, 1967).

Living salivary glands were viewed and photographed under bright field optics. For light microscopy, glands were fixed in Camoy’s fixative and prepared by standard histological techniques. Twenty μm paraffin sections were stained with phloxin and haematoxylin. For scanning electron microscopy, material was fixed in 1·5% glutaraldehyde in 0·1 M cacodylate buffer and post-fixed in 1% OsO₄. It was dehydrated through a graded series of alcohols, critical-point-dried using CO₂, and mounted on a stub with copper tape. Specimens were coated with a thin layer of goldpalladium prior to viewing.

The bilaterally arranged salivary glands of Helisoma consist of a pair of yellowish, blind-ending, tubular structures (Fig. 1). Acinar outpocketings ranging from 100 to 500 μm in diameter are clustered along the distal three-fourths of the glands (Figs. 1, 2 A, B). Proximal to these acinar regions the tubes extend as ducts which pass through the muscular buccal mass at points just lateral to the point of exit of the oesophagus and which empty their secretions into the buccal cavity (Fig. 1). The lumen of each acinus is contiguous with that of the central tube and duct. Each of these luminal surfaces is lined with highly motile cilia (Fig. 2D). Viewed under compound Nomarski optics, secretory globules can be seen to be conveyed by the cilia from an acinus and along the duct to be emptied into the buccal cavity.

Initial light and electron microscopical studies confirmed that Helisoma’s salivary glands display variability of secretory cell types as observed in the related snail, Lymnaea stagnalis (Carriker & Bilstad, 1946; Boer, Bonga & Van Rooyen, 1967), but no detailed study of the morphology was made. In external surface view, the cells appear irregularly shaped and in close-packed array (Fig. 2B). The lumina of the acini and the ducts are lined with columnar secretory cells. When sectioned these cells appear approximately 35 μm in width and 75 μm in length (Fig. 2C, D). Distinct granules can be seen in living cells viewed with Nomarski optics, and in fixed and sectioned specimens viewed with light or electron microscopy (c.f. Fig. 2C). These cells typically contain large amounts of rough endoplasmic reticulum and prominent Golgi apparatus.

In addition to the predominant secretory cells, the glands also contain sparse populations of small (0·5−0·2 μm diameter) muscle fibres. Also, when the distal portion of the oesophageal nerve trunk is filled with CoCl₂, a few small neurone-like elements are suggested in the glands. In order to ensure that the intracellularly recorded activity in this and subsequent studies was that of the secretory cells, a survey of the gland was performed employing dye-filled microelectrodes. This survey con-clusively demonstrated that the small size and relative sparsity of the ancillary cell types effectively limit microelectrode recordings to the larger, visually impalable secretory cells. As has been shown to be the case for the cells generating action potentials in the pedal gland of Ariolimax (Kater, 1977), all cells injected with dye displayed the large spherical to ovoid shape characteristic of secretory cell types.

Microelectrode penetration of a salivary gland cell is signalled by a rapid negativegoing shift in potential. Penetration of 228 cells in 9 different animals showed a mean resting potential of − 73 ± 9 mV. Previous authors have suggested different morphological cell types in pulmonate salivary glands on the basis of cytochemistry (Carriker & Bilstad, 1946; Boer et al. 1967); the graphic distribution shown in Fig. 3, however, fails to resolve any such groupings on the basis of differential resting potentials. Occasional low values of resting potential (below 45 mV) were obtained, but these usually were unstable recordings, indicating poor penetrations. Throughout our studies we have arbitrarily divided the salivary gland into four equal regions beginning at the most proximal acinus and terminating at the blind ending of the tubular salivary glands. Our data has prompted us to ask whether any regional specialization of the gland can be found. In a subsequent study (Kater et al. 1977) it will be shown that for some characteristics, but not the resting potential, such regional specialization does exist.

Some variability in the magnitude of resting potentials was detectable between preparations. Consistent low values through a particular gland were frequently traceable to either (1) glands dissected several hours prior to recording (and allowed to partially digest themselves) or (2) glands in which any particular area was mechanically damaged. This latter case will be seen as important for the gland as a whole when the extensive electrical coupling of this system is discussed.

Fig. 4 demonstrates the most striking features of the spontaneous activity recorded in the salivary gland cells. We use the term ‘spontaneous activity’ to refer to activity recorded with no stimulation of the preparation. We do not imply that the activity is endogenously generated in glandular cells. A subsequent communication (Kater et al. 1977) documents the neural activation of most of this ‘spontaneous’ gland cell activity. In the upper record of Fig. 4 there are two rapid transients which overshoot zero potential. The first of these is similar in form to a classical neuronal action potential, and the second has the appearance of an action potential occurring upon a large burst of EPSPs. That such transients may be justifiably designated action potentials will be demonstrated below on the basis of their regenerative nature and of their refractoriness.

Although spontaneous action potentials nearly always overshoot the zero potential, the maximum amplitude and the nature of the waveform vary considerably. Action potentials may arise abruptly out of the base line or ride on one or more, EPSP-like, summating depolarizing potentials. Action potentials arising out of the EPSP-like excitation are frequently of lower peak amplitude than those arising more abruptly from the base line. Similarly, when two or more action potentials occur in rapid sequence such that full repolarization does not occur, the succeeding action potentials typically have lower peak amplitudes than the first. The repolarization phase of spontaneous action potentials also shows considerable variability. Action potentials arising from resting potential usually repolarize with little or no undershoot, in contrast to action potentials arising from depolarizing slow potentials which typically show distinct undershoots. The greatest undershoots, in both amplitude (several mV) and duration (several seconds), typically follow relatively long duration action potentials. A survey of cells along the length of the gland gave no indication of differentiation of cell types on the basis of spontaneous action potential waveforms. Any given cell could generate various forms of spontaneous transients.

Patterned forms of spontaneous gland cell activity could be observed when connexions with the buccal ganglia were left intact. In such cases rhythmic bursts of action potentials occur with similar timing to feeding motor output (Fig. 4B, also cf. Kater, 1974). Evidence of the neural origin of such spontaneous activity will be presented in a subsequent paper (Kater et al. 1977) which demonstrates that high Mg²⁺ solutions, known to block chemical synaptic activity, essentially silence salivary gland cells.

The generation of action potentials by salivary gland cells seems to be a general phenomenon in gastropod molluscs. We have examined some pulmonates, the opisthobranch, Aplysia californica, and the prosobranch, Vivipara sp. The pulmonates examined comprised the aquatic snails, Lymnaea stagnate, Physa glabrata, Planorbis corneus, and Stagnicola reflexa ; the terrestrial snail, Helix pamatia ; and the terrestrial slugs, ArioUmax californica, and Limax sp. In every species at least some proportion of the gland cells we examined displayed spontaneous electrical activity like that in Heteoma.

Simultaneous intracellular recordings from two salivary gland cells within the same acinus reveal extremely close correspondence of spontaneous transient events (Fig. 5 A). Even cells a few acini apart show similar activity with only minor variations in waveform (Fig. 5 B). Such results might be expected for electrically coupled cells.

Electrical coupling is demonstrated readily in this system using the standard technique of d.c. current passage from one cell to another and measuring the voltage recorded in each of these two cells. Throughout the glands current passes readily between cells within an acinus (Fig. 6) and to some extent between cells in neighbouring acini. Coupling coefficients between adjacent cells (measured as V₂/ V₁ with current pulses injected into ‘cell 1’, Bennett, 1966) may be as high as 0 · 9. As would be expected, such coupling decreases considerably with distance. The usual coupling coefficient between cells chosen at random within an acinus was about 0 · 2; in one particular series of 12 pairs of cells the value was 0 ·17 + 0 · 09. The electrical coupling between cells on the apices of two adjacent acini was approximately 0 ·1.

There is no evidence of rectification at the junctional membranes. However, critical distinctions are difficult with bridge injection methods. When current is passed between two cells whose effective input impedance (K/Z) is approximately equal the bidirectional coupling coefficients are nearly identical (Fig. 6). When two cells have substantially different effective input impedances, the coupling coefficient between them varies depending upon which cell receives the current injection. However, this does not demonstrate rectification at junctional membranes because current injected into the two cells will ‘see’ different current sinks (Bennett, 1966).

The tortuous nature of the acinar morphology of these glands precludes direct quantification of the decrease of d.c. current with distance. However, it is clear that within any given acinus sufficient coupling exists to result in extremely high levels of current passage between cells, and accordingly contribute to the high levels of synchrony observed in spontaneous fluctuations of membrane potential. It will be shown below that the electrical coupling is sufficient to allow the propagation of action potentials from acinus to acinus along the tubular gland. Functionally, such synchrony of electrical activity could regulate synchronous secretory activity throughout the gland.

As a consequence of such extensive electrical coupling on multiple surfaces, the impedance of any given cell is usually extremely low; so low that we have for the majority of cases been unable to inject sufficient current to reach threshold for spike generation. The input impedance of these cells averaged 8 · 25 ±4 · 71 M Ω (n = 38). In low frequency we encountered, however, some cells with input impedances of 40 M Ω or more. Such cells may be found scattered throughout the length of the gland, but occur most often in the first or second acinus adjacent to the duct emptying into the buccal mass.

High input impedance cells could be depolarized to suprathreshold levels by intracellular current injection. The response to relatively long current pulses is the generation of an all-or-none action potential (Fig. 7 A). Repetitive firing is rarely observed in response to sustained depolarizations. Short duration current injections readily demonstrate the regenerative nature of these action potentials (Fig. 7B).

Action potentials could rarely be evoked intracellularly in low input impedance cells. In order to circumvent the current-passing limitations of bridge-mediated intracellular current injections, we routinely have employed extracellular field stimulation to evoke action potentials (Fig. 8 A). While it is possible that action potentials generated in this fashion are mediated by neural excitation (see Kater et al. 1977), this was certainly not always the case in these experiments. Field stimulation can evoke action potentials from gland cells at latencies as small as 1 or 2 ms. Furthermore, though synaptic activation through nerve stimulation alway fatigues at repetitive rates in excess of 10 Hz (Kater et al. 1977), field stimulation continues to evoke action potentials at frequencies beyond this level. In Helisoma we have never observed chemical synaptic transmission with latencies less than 4 ms, nor have we observed following frequencies greater than 20 Hz (e.g. Kater, 1974); both of these criteria are always exceeded by field stimulation. Fig. 8B illustrates the refractory nature of salivary gland cell action potentials evoked by field stimulation.

Simultaneous intracellular recordings from secretory cells that are distant from each other indicate that action potentials can propagate throughout the salivary glands. Field stimulation at either end of the gland can evoke an action potential which can propagate in either direction. The delay between action potentials recorded from two cells is a function of the intercellular distance (Fig.9A, B). A quantitative determination of the speed of conduction through this network was precluded by the tortuous geometry characteristic of the glands. It should be emphasized that propagation is not an artifact of field stimulation since even during neural activation action potentials progress from proximal to distal just as seen in Fig. 9.

Fig. 9C shows two superimposed action potentials generated in rapid succession in a cell of the most proximal acinus by intracellular depolarization of that cell. Simultaneous recordings from a cell in the adjacent acinus revealed on the first sweep an EPSP and on the second an action potential rising out of a second EPSP. The electrotonic network in these glands seems to have a sufficiently high effective space constant to allow the network of coupled cells to act as a conducting cable. One-to-one spike correspondence decreases with distance from the cell in which current is injected.

Propagation of intracellularly evoked action potentials does not usually occur between cells with one or more interposed acini. It is possible that our experimental procedures may have reduced the electrical coupling to result in an artificially low safety factor and that under physiological conditions a single action potential could propagate throughout the gland. We suggest, however, that it is necessary for a certain minimum number of cells to fire action potentials essentially simultaneously to ensure propagation of excitation throughout this system.

The investigation of puhnonate exocrine glands indicates that two physiological processes may be much more general and widespread than previously believed:

1. the production of regenerative action potentials in non-neural secretory cells;

2. the propagation of electrical impulses through electrically coupled epithelial cells. The first of these processes has been previously described only in mammals (Matthews & Saffran, 1973 ; Matthews & Sakamoto, 1975) and a hydrozoan coelenterate (Mackie, 1976); the second has been reported only in amphibians (Roberts & Stirling, 1971) and coelenterates (Mackie, 1976). The present work demonstrates that action potentials are regenerated and propagated throughout the secretory salivary gland network of the pulmonate gastropod mollusc Helisoma. The production of action potentials in gastropod salivary glands seems to be a general occurrence. This, coupled with the generation of action potentials in slug mucous glands (Kater, 1977), suggests that exocrine secretion in general is linked to regenerative electrical activity in this class.

A common feature of regenerative electrical events in secretory cells seems to be an influx of calcium ions. The necessity for calcium ions in many secretory processes is well known (e.g. Douglas, 1974, 1976; Katz and Miledi, 1965, 1969a, b and 1970; Llinas, Blinks & Nicholson, 1972 ; Llinas and Nicholson, 1975 ; Matthews & Sakamoto, 1975; Mackie, 1976). Most recently a Ca²⁺ component of the action potential in the pedal exocrine gland of the slug, Ariolimax, was found to be necessary for the secretion of mucus from the gland (Kater, 1977). Preliminary experiments indicate that external Ca²⁺ is necessary for the salivary gland action potentials, but a further investigation is required to make direct comparisons with the slug mucous glands.

Low resistance pathways are a common feature of epithelia (McNutt & Weinstein, 1973; Satir & Gilula, 1973). However, the precise function of these pathways is unclear. One possible function of the electrical coupling in Helisoma salivary glands is to co-ordinate the activity of various secretory cell types which may produce different components of a composite saliva. This co-ordination might include metabolic co-operation between the cells and the regulation of synthetic activities (Burk, Pitts & Subak-Sharpe, 1968; Gilula, Reeves & Steinbach, 1972). However, the propagation of action potentials in this system suggests that a primary function of electrical coupling in this glandular epithelium is to ensure the synchronous release of secretory product from various cell types to form a complex saliva.

The morphology of this system as described in Figs. 1 and 2 represents the typical situation. There exists, however, at low frequencies, individual variation of several morphological features. Among the more interesting atypical features is the occurrence of a supernumerary nerve of unknown function extending from the cerebral or pedal ganglia to the salivary glands. A second variation in innervation that is not quite as rare is the emergence of the salivary nerve directly from the buccal ganglia instead of branching from the oesophageal trunk.

The colour of the glands, though usually yellowish, varies considerably. The glands can be deep yellow or creamy white and may have an opaque, crystalline, or transparent appearance. This variance is suspected to be a function of the physiological state of the glands and of quantitative and qualitative variations in secretory contents.

A final variable feature is the acinar morphology of the salivary glands. Older glands typically have larger, more numerous, and more globular acini than younger glands. In some cases the acini protrude from the salivary glands in a single plane such that the glands have a bilaterally flattened appearance. In other cases the acini protrude in radially symmetrical whorls. The significance of all of these variations is unknown.

One of the more puzzling features of these salivary glands is the variation in frequency of spontaneous action potentials both between preparations and within a given preparation over time. Spontaneous activity may either decrease or increase in a given preparation. Perhaps injury to the glands could cause an overall depolarization leading to increased excitability and ‘injury potentials’. On the other hand, damage may cause localized uncoupling (e.g. Asada & Bennett, 1971) leading to partially isolated groups of cells and making the gland as a whole much more inactive.

The factors underlying the variability in the rising phase, peak amplitude, and repolarization phase of action potentials in this system are difficult to specify. As a result of the extensive electrical coupling, the waveform of transients generated in a given cell will be modified by the electrotonic decay of passive and regenerative electrical events in neighbouring cells. Another factor affecting waveform is the variability in the effective shunt resistance of the cells. Such variability would include changes in membrane permeability due to transmitter effects in those cells that are innervated (see Kater et al. 1977). The effective current sink into neighbouring cells would also contribute to the shunt resistance. Particularly relevant in this situation is Getting’s hypothesis that electrically coupled cells which are isopotential are functionally uncoupled (Getting, 1974). Thus, action potentials generated simultaneously in many cells of a coupled network would be generated across a higher effective resistance (due to less current shunting into adjacent cells) than action potentials evoked in only one or a few cells. This could at least partially explain why evoked action potentials are of lower amplitude than spontaneous action potentials in the same cells.

There seems to be more than one factor involved in whether an action potential can be evoked by intracellular depolarization of a given cell. The effective input impedance is obviously important. Those cells in the 40 M Ω range are always capable of generating intracellularly-evoked spikes, whereas only occasional cells with low input impedance can produce them. The nature of the threshold variability among low input impedance cells is unclear. Furthermore, that all of the cells actually share common ionic bases for their action potentials is not yet known.

We have shown how the generation and propagation of action potentials in the salivary glands may account for the co-ordination of cellular activities within the glands. In a subsequent paper (Kater et al. 1977) we will investigate the integration of the activity of the salivary glands in the feeding behaviour of Helisoma.

We thank Dr Roger Thomas for invaluable technical assistance and Ms J. Kater for histological studies and assistance with the preparation of this manuscript. This work was supported by Public Health Service research grant 1 ROi NS09696 and in part i Roi AM19858. D.M. was supported by training grant no. HD-00152.