Electrophysiological and Morphological Identification of Action Potential Generating Secretory Cell Types Isolated from the Salivary Gland of Ariolimax

The electrophysiology of exocrine secretory cells has received increasing attention over the last decade (see reviews by Peterson, 1976, 1980; and House, 1980). One of the goals of these studies has been to elucidate the relationship between electrical characteristics and the process of secretion. In the exocrine secretory cells of molluscs, regenerative electrical activity has been shown by research at this laboratory (Kater, 1977; Kater, Murphy & Rued, 1978; Kater, Rued & Murphy, 1978; Hadley, Murphy & Kater, 1980). Stimulus-secretion coupling in these exocrine cells may thus by mechanistically similar to that found at the neuromuscular junction and the squia giant synapse. There are, however, potential pitfalls in this interpretation, since exocrine tissues are composed of cells which may communicate electrically with one another via low resistance junctions (e.g., Kater, Rued & Murphy, 1978; Kater & Galvin, 1978). The situation is further complicated when there are morphologically distinct secretory and nonsecretory cell types composing such an organ system (e.g., Boer, Wendelaar Bonga & van Rooyen, 1967 ; Walker, 1970; Beltz & Gelperin, 1979; Kater, Rued & Murphy, 1978). Thus, the electrical activity recorded from a gland cell may reflect the membrane properties of a closely communicating, yet morphologically different, neighbouring cell. In such a situation, the electrical activity of the cells can only be studied by recording from identified cells that have been isolated from one another. In the present investigation we have found that at least three different types of electrical activity can be recorded from the intact salivary gland of the slug.

Ariolimax were provided by Mr Robert Daub of Coos Bay, Oregon and could be maintained in our laboratory for 5–6 months. The slugs were kept at 13 °C in plastic boxes lined with moist paper towels and were fed zucchini and lettuce.

To remove the salivary gland the slug was first pinned, dorsal surface up, through the head and tail to a wax-covered dish. An incision was made along the dorsal midline from tail to head and the flaps retracted laterally. The gland, located just posterior to the buccal mass, is easily removed after cutting attachments to the underlying tissue and oesophagus, and the ducts to the buccal mass. The gland was stored in Ca²⁺-free, high Mg²⁺ saline (see Table 1) in the refrigerator for up to 48 h.

A small piece of whole gland was minced with a razor blade in a few drops of Ca²⁺-free, Mg²⁺-free saline and was digested in 3 ml of 1· 0 % collagenase (Sigma, Type I) at 30 °C in Ca²⁺-free, Mg²⁺-free saline in a shaker bath for 10–20 min. The precise duration was determined by visual inspection of samples. At the end of the digestion period the suspension was centrifuged at 50g for 5 min. After aspirating the supernatant, the cell pellet was suspended in 3 ml of Ca²⁺-free, Mg²⁺-free saline containing 2 mm-EDTA and incubated for 10–20 min in a shaker bath at room temperature. This suspension was centrifuged at 50 g for 5 min and the pellet then suspended in 3 ml of normal saline (Table 1).

The above procedure was the most successful of many variations tried, including the use of trypsin, hyaluronidase and various combinations of these enzymes with collagenase. Digestion was carried out at 30 °C because the electrophysiological results suggested that higher temperatures damaged the tissue. Best results were obtained when the enzyme solution was prepared just prior to use. Since the cells adhere to glass, cell yields were increased by using only plastic vessels.

A small piece of gland was stretched and pinned to a black Sylgard (Dow Corning) block which was then fitted into a Lucite chamber (Kater, Rued & Murphy, 1978) and was continuously perfused with saline solutions. In some cases, the tissue was first soaked in 0· 2 % collagenase at 30°C for 10–15 min; this seemed to aid penetration of the cells without noticeably affecting their electrical properties. Solutions could be changed rapidly, allowing manipulation of the ionic composition of the media. The time-course for a solution change was indicated by the time required to abolish the action potential in Na⁺-free, Ca²⁺-free media. This could be as little as 30 s for surface cells or up to 3 min for deeper cells. One to two additional minutes were then allowed before recordings were made.

Cells were impaled with microelectrodes filled with 3 M-potassium chloride or 4 M-potassium acetate. Intracellular recording and stimulation were accomplished by conventional techniques (Kater, Rued & Murphy, 1978). Current traces in Figs 2–6 are used to indicate only the time course of current injection and relative magnitude in any given series.

Cells were deposited on coverslips coated with polylysine to promote adhesion. Coverslips were prepared by soaking them in a solution of poly-L-lysine (Sigma Type VII-B) in distilled water (2mg/ml) for 18–24 h, washing in distilled water for 1 h and air drying. The coverslip was placed in a chamber whose bottom was also a glass coverslip since the cells were to be viewed from below on an inverted compound microscope (Zeiss, ICM) equipped with Nomarski optics.

Cells adhering to the coverslip were impaled with 3 M-potassium acetate-filled glass microelectrodes (d.c. resistance = 30–60 MΩ) bent within a few millimetres of their tips to facilitate movements within the condenser working distance. Recording and stimulation through individual microelectrodes were made by use of a high input impedance electrometer with integral current monitor (Dagan, Preamp-clamp). Signals were displayed on a Tektronix oscilloscope.

The cells were continuously perfused with saline by a Buchler Poly-staltic pump running at approximately 1 ml/min. Solutions were drawn off by a second channel of the same pump.

The composition of the various solutions used is shown in Table 1.

Tissue was fixed for 6 h at 4 °C in a modified Karnovsky’s fixative (Karnovsky, 1965), containing 1 % paraformaldehyde and 3% glutaraldehyde in 0· 06 M-cacody-late buffer (pH 7· 4) and postfixed in 1 % OsO₄ in cacodylate buffer. After dehydration in an ethanol series and embedding in Spurr’s medium (Spurt, 1969), sections were cut with glass knives on a Porter Blum MT2 ultramicrotome. Thin sections were stained with 3 % aqueous uranyl acetate and with lead citrate (Venable & Coggeshall, 1965) and were viewed and photographed with a Phillips EM 300 electron microscope. Semi-thin sections (0· 5 μm) were mounted unstained on glass slides and photographed with phase contrast optics on a Zeiss Photomicroscope.

Dissociated cells, suspended in a drop of saline, were deposited on polylysine-coated coverslips (notched in one corner to serve as a landmark in subsequent processing) and left undisturbed in small Petri dishes for 20 min to allow cells to settle and adhere. Glutaraldehyde fixative (1 % in 0 · 05 M-cacodylate buffer, pH 7· 2) was then applied dropwise to fill the dish, and cells were fixed for 1 h at 4 °C. After a buffer rinse the coverslips were inverted onto a drop of saline on a glass slide for viewing and photography on a Zeiss Universal Microscope equipped with a calibrated mechanical stage, Nomarski optics, and set up for Polaroid Type 105 photography. When an area of interest was located, its stage coordinates were recorded along with the coordinates of a landmark such as the notch, and the area was photographed at low power. This Polaroid print then served as a map of the area, allowing single cells to be numbered and photographed at higher power. The coverslip, removed from the slide, was then placed, cell-side up, in a Petri dish and was carried through osmification, alcohol dehydration and infiltration. Beem capsules were filled with Spurr’s embedding medium so that there was a positive meniscus, and the coverslips were inverted onto this. After polymerization the approximate location of the cells of interest could be measured relative to the notch using the recorded coordinates. A dissecting microscope with magnification to 60 × allowed exact localization of the cells by comparison with the Polaroid map. Marks whose extensions would intersect at the cell’s location were made on the block itself and the coverslip was removed after spraying with a quick-freezing aerosol (Cryokwik). After trimming, the boundaries of the block face were drawn on the Polaroid map to allow easy identification of sectioned cells at the microscope.

Semi-thin and thin sections were cut, stained and photographed as described above, with the exception that thin sections were mounted on formvar-coated, slotted grids.

The transparency of the living gland allows ready visualization with Nomarski optics of many focal planes through a thin piece of tissue. Apparent secretory activity and passage of material through a duct system can be observed clearly in such preparations. The living gland consists of a variety of morphologically distinct cell types (Fig. 1). While the total number of cell types is as yet unknown, three were readily defined on the basis of the appearance and size of their secretory granules: (1) small granule (l–2μn granule diameter, which at 40× magnification have a rough-textured appearance, but individual granules are not generally resolved); (2) medium granule (3–6μm granule diameter); and (3) large granule (8–12μm granule diameter) cell types (Fig. 1). Cell size also varies with cell type; large granule cells can exceed 100 μm in at least one dimension, while small and medium granule cells are considerably smaller (although some in both classes reach 85 μm in diameter).

Cells impaled by blind penetration into the gland displayed three classes of electrical activity. In response to depolarizing current injection many cells were electrically inexcitable. These cells often had large resting potentials, and thus did not appear to be damaged. Other cells generated overshooting, all-or-none action potentials. These action potentials were of two classes (Fig. 2), being distinguished primarily on the basis of their duration and the presence of a plateau on the falling phase of the slower action potential. In some cells, depolarizing current pulses elicited graded responses, but it was not possible to determine if the response was of a third distinct class or the result of impalement injury. The ionic basis of the more frequently encountered fast action potential was investigated to serve as a basis for later comparison with enzymatically isolated cells (Fig. 3). When Na⁺ and Ca²⁺ were removed from the bathing medium and replaced with Tris and Mg²⁺ respectively, the action potential was abolished. If Ca, but not Na⁺, was added back to the perfusion solution, the action potential was restored, although its amplitude was somewhat smaller. On the other hand, if Na⁺, but not Ca²⁺, was returned to the bathing medium, only a small règenerative response was elicited by depolarizing current pulses. In other preparations, where Ca²⁺ carries current during the action potential, Co²⁺ ions have been shown to be an effective blocker of Ca²⁺ currents (Geduldig & Junge, 1968; Hagiwara & Takahashi, 1967). The addition of 10mm-CoCl₂ to normal saline (both Na⁺ and Ca²⁺ present) greatly reduced the amplitude of the action potential (Fig. 3). Taken together, these data suggest that during the rising phase of the action potential Ca²⁺ ions are the predominant current carriers in this cell type.

In order to correlate directly the electrical activity of a particular cell type with its morphology, further electrophysiological experiments were performed on single cells isolated from the salivary gland by enzymatic dissociation. This provided a degree of resolution not obtainable in the intact gland since it is not possible to discern the morphology of cells in the whole gland with the optical configuration required for electrophysiology.

In preparations of dissociated cells, three predominant cell types were clearly distinguished (Fig. 4). No attempt was made to determine the viability or yield of the preparation quantitatively. However, based on casual inspection, approximately 30 % of the cells were viable (as judged from the correlation of the microscopic appearance with subsequent electrophysiological experiments). This relatively low yield of cells posed no problems in the present study since bulk properties of the cell suspension were not studied and one could readily select viable cells for electrophysiological study. Among the cell types observed with Nomarski optics were three types that were similar to those observed in the intact gland, with large, small and medium granules (Fig. 4). The largest of the large granule cells apparently do not survive dissociation. While these too might be isolated in viable condition with a more refined dissociation protocol, our simplified procedure was adequate for our purposes, yielding cells of 80 ftm or greater.

Isolated gland cells were studied by impaling cells with a single microelectrode for passing current and recording membrane potential. Both large and small granule cells gave overshooting action potentials in response to depolarization (Fig. 4A, B) while the medium granule cell was electrically inexcitable and displayed only delayed rectification (Fig. 4C). The two types of action potentials resembled those observed in the intact gland ; the spike of the large granule cell was longer in duration than the spike of the small granule cell, and the large granule cell spike had a plateau on the falling phase. Membrane potential, input resistance and spike overshoot were measured in a number of large and small granule cells (Table 2). Small granule cells had an average resting potential of 73 mV, an average input resistance of 158 MΩ and an average spike overshoot of 29 mV. Large granule cells had an average resting potential of 60 mV, an average input resistance of 306 MΩ and an average spike overshoot of 38 mV. The values of input resistance are quite high in relation to the large size of the cells. Some large and small granule cells did not generate spikes at the end of the stimulus pulse; however, the input resistance of these cells was low and it was assumed they were damaged. No data were taken from such cells.

The ionic basis of the rising phase of the action potential was investigated in both small (Fig. 5) and large (Fig. 6) granule cells. When both Na⁺ and Ca²⁺ were removed from the bathing medium, small granule cells were incapable of generating an action potential (not shown). If Na⁺, but not Ca²⁺, were present only a small response could be elicited by the stimulus pulse (OCa, Fig. 5A). However, when Ca²⁺, but not Na⁺, was present the cell could still generate an overshooting spike (ONa, Fig. 5A). When 10mm-CoC1₂ was added to normal saline solution, the amplitude of the spike was greatly reduced (NR & Co, Fig. 5B). Responses of the large granule cells to similar changes of the bathing medium (Fig. 6) were similar to those of the small granule cells. The only difference noted was that the large granule cells displayed no regenerative activity when the cell was bathed in a Ca²⁺-free medium. One puzzling finding was that addition of 10 mm-Co²⁺ to normal saline did not completely abolish the response of the large granule cell to depolarization as evidenced by the nonexponential fall of potential at the end of the current pulse. This was a routine observation. These data suggest that Ca²⁺ ions are the predominant current carrier during the action potential in both small and large granule cells. There appears to be a small Na⁺ component to the spike of the small granule cells, but it is not certain whether Na⁺ ions carry significant current during the spike of the large granule cell. Since the morphological appearance and spike characteristics of the isolated, small granule cell were similar to those observed in the intact gland, it appears that the isolation procedure has not significantly compromised the functional integrity of the cell.

Histological and ultrastructural observations were made on the spiking secretory cell types of the salivary gland in order to provide further data for cell type classification. Dissociated cells were identified under Nomarski optics as large (Fig. 7A) and small (Fig. 7C) granule cells. The same cells shown in Fig. 7A and C were then embedded and sectioned and are shown in the phase photomicrographs in Fig. 7B (large) and Fig. 7D (small). (These same cells were also thin sectioned and viewed with the electron microscope (not shown), and while identification of cell type was possible, the quality of preservation in these cells after the lengthy procedures of identification and photography was not adequate to provide ultrastructural details.) The morphology revealed by sectioning these isolated cells clearly corresponds to cell types seen in sections of the whole gland (Fig. 7E, F). Here adjacent semi-thin (Fig. 7E) and thin (Fig. 7F) sections allow the same cells to be viewed with both the light and electron microscope.

The large granule cell (Fig. 7A, B, E, F) is characterized by the density and large size of its granules (up to 10 μm diameter) and the large cell size, which can exceed 100 μm in length (some shrinkage in processing is unavoidable and measurements of processed cells and granules are not as large as those of living cells). The small granule cell (Fig. 7C, D, E, F) is smaller in size (≃35 μm in diameter) and contains small (1–2 μn), electron-lucent granules with characteristic intragranular densities. These two cell types are easily identifiable under Nomarski optics prior to intracellular recordings and can be linked reliably to the corresponding dense large granule cell type and small granule cell type seen in plastic sections of these glands. The medium granule cell has not been carried through this rigorous identification procedure and so cannot yet be linked with certainty to a cell type seen in sectioned material.

The action potential of both large and small granule salivary gland cells was shown here to be primarily Ca²⁺-dependent, as with the ionic basis of the action potential describedin the whole gland of Helisoma (Hadley et al. 1980). Spike activity also has previously been reported in the pedal gland of the slug (Kater, 1977) and this has been shown to be dependent both on Na⁺ and Ca²⁺. Although action potentials have not been observed in mammalian salivary glands (Kater & Galvin, 1978) they have been observed in isolated mammalian adrenal medullary cells (Biales, Dichter & Tischler, 1976; Brandt, Hagiwara, Kidokoro & Miyazaki, 1976); the spike here is primarily Na⁺-dependent although a small Ca²⁺ component can be observed under appropriate conditions (Biales et al. 1976). Pancreatic islet cells show slow depolarizing wave with superimposed spikes in response to applied glucose (Dean & Matthews, 1970, Matthews & Sakamoto, 1975; Meissner & Schmelz, 1974); both the depolarizing wave and spike are Ca²⁺-dependent.

It is important to recognize that the individual cell types, which make up any glandular system, normally function as an integrated ensemble (for Helisoma see Kater, Murphy & Rued, 1978 and Senseman & Salzburg, 1980). The present communication, however, speaks only to the individual properties of particular cell types rather than attempting to ask about interrelationships. Such interactions deserve investigation in their own right. This fact notwithstanding, it is clear that the kind of information obtained by partitioning the complexity of this glandular tissue may be masked in studies which rely on data solely from intact glands. The finding that morphologically distinct cell types display different electrical activity may indicate that different secretory products are released only under a specific set of conditions associated with changes in membrane potential.

The purpose of the morphological studies described here is to establish a method of reliable classification of isolated cells and not to provide an ultrastructural analysis of these cells such as has been done on pancreatic acinar cells (Amsterdam & Jamieson, 1974) and adrenal medullary cells (Fenwick, Fajdiga, Howe & Livett, 1978). While the quality of the preservation of our material is certainly adequate for our present purposes, it is lower than that obtained in fixing a freshly dissociated cell pellet. A thorough ultrastructural and histochemical analysis is presently in progress (J. W. Kater, R. G. Kessel, S. B. Kater and J. G. Assouline, in preparation). These studies indicate not only that different cell types are present in this gland but that the secretory products elaborated by these cell types are also histo-chemically different.

Advantages of working with dispersed single cells have been apparent since Douglas’ early work on the mast cell (Kanno, Cochrane & Douglas, 1973). With great technical skill Douglas and coworkers were able to employ these small cells for direct visual observation of exocytosis. An important feature of the molluscan cells in the present study is the fact that a single granule in the large granule cell type can be 10 ^m in diameter, i.e., as large as an entire mast cell. We have made preliminary observations of exocytosis in these living cells during microelectrode penetration in order to assess the feasibility of examining the problem of stimulus-secretion coupling (Llinas & Heuser, 1977). Under appropriate stimulus and cinematographic conditions the secretory event might be directly observed in these living exocrine cells under direct voltage control, thus providing an opportunity to study the relationship of such important parameters as Vm, ΔCa²⁺_m and exocytosis in a single system.

We thank Robert Hadley and Drs A.D. Murphy and Brian Salzberg for their critical reading of this manuscript and F. Hunter, M. Jolly and P. Gade for assistance in its preparation. This research was supported by NIH grant 5 ROI AM19858 to S. B. Kater and NIH fellowship 1 F32 NS05797 to J. M. Goldring.