Optical Studies of the Secretory Event at Vertebrate Nerve Terminals

The study of the secretory event perse, as well as the excitation that provokes it, has proved especially difficult at the nerve terminals of vertebrates. This, of course, is largely the result of the small size of the structures of interest and, consequently, much of our mental imagery, if not our understanding, related to the release of transmitter substances comes from the electron microscope (Heuser et al. 1979; Heuser & Reese, 1981). As far as electrophysiology is concerned, we have depended for our understanding largely upon a series of elegant studies of synaptic transmission at the frog neuromuscular junction (Katz, 1969), in which most measurements were actually made on the postsynaptic element, and upon analogies drawn from that most useful of invertebrates, the Atlantic squid, Loligo.

Patch-clamp methods have begun to provide important tools for the study of secretion (Fernandez, Neher & Gomperts, 1984; Lindau & Fernandez, 1986) in some systems (e.g. mast cells), but their reliable application to vertebrate nerve terminals is only beginning (Mason & Dyball, 1986; Lemos & Nowycky, 1987). Optical techniques for the measurement and analysis of transmembrane electrical events have also found a wide range of applications over the past decade because, although they often provide less detailed information than does the voltage clamp, they can offer certain advantages over more conventional measurements. Because the membranes of interest are not mechanically violated, the methods are comparatively non-invasive. Spatial resolution is limited only by microscope optics and noise considerations; it is possible to measure changes in membrane potential from regions of a cell with linear dimensions of the order of 1 μm, and the use of optical sectioning, with elimination of scattering by means of confocal imaging, may soon permit optical recording from small regions of cells in intact ganglia or brains. Temporal resolution is limited by the physical response of the probes and by the bandwidth imposed upon the measurement, again usually by noise considerations, and response times faster than typical membrane time constants may be achieved. Because mechanical access is not required, unusual latitude is possible in the choice of preparation, and voltage changes may be monitored in membranes which are otherwise inaccessible. Also, since no recording electrodes are employed, and the measurements are actually made at a distance from the preparation - in the image plane of an optical apparatus-it is possible to record changes in potential simultaneously from a large number of spatially separated sites. Several recent reviews have surveyed the literature on optical measurement of membrane potential (Cohen & Salzberg, 1978; Waggoner, 1979; Freedman & Laris, 1981; Salzberg, 1983; Grinvald, 1985) and they should be consulted by the reader interested in the details of these techniques.

The evidence that optical methods are equivalent, at least in a limited sense, to electrode measurements has been detailed elsewhere (Cohen et al. 1974; Ross et al. 1977; Salzberg et al. 1977; Cohen & Salzberg, 1978; Gupta et al. 1981). Indeed, under certain conditions (Salzberg & Bezanilla, 1983), an optical measurement of membrane potential is superior to an electrode measurement. Our purpose here, however, is to illustrate the application of optical techniques to the coupled problems of excitation and secretion at vertebrate nerve terminals, problems for which optical methods are particularly well suited. The direct optical recording of the action potential from the nerve terminals of a vertebrate and the study of its ionic basis (Salzberg, Obaid, Senseman & Gainer, 1983; Obaid, Orkand, Gainer & Salzberg, 1985) provide information that is presently unobtainable by other means. A variety of linear potentiometric probes can now be used to make this kind of measurement, by exploiting the extrinsic absorption or fluorescence signals that result when the potential varies across a membranes stained with these dyes. We shall begin by describing some of the properties of the nerve terminal action potential, as revealed in the light of the tungsten-halogen bulb. Analysis of optical recordings of the action potential in the nerve terminals of a mammal has disclosed the presence of at least two large and rapid changes in the intrinsic optical properties of the neural lobe, in addition to the extrinsic absorption changes that report changes in membrane potential, and one of these, a change in light scattering, has now been shown (Salzberg, Obaid & Gainer, 1985) to be intimately associated with the physiological events that attend the release of the neuropeptides arginine vasopressin (AVP) and oxytocin. After describing the properties of these intrinsic signals, and detailing some of the evidence that this optical signal is related to the secretory event, we will conclude this chapter by discussing some of the pharmacology of the calcium channels that are required for the release of neuropeptides in the neurohypophysis.

The vertebrate hypothalamo-neurohypophyseal system represents an extremely attractive model for the study of excitation-secretion coupling at nerve terminals. Magnocellular neurones located in the hypothalamus (supraoptic and paraventricular nuclei in mammals, preoptic nucleus of lower vertebrates) project their axons as bundles of fibres through the median eminence and infundibular stalk to terminate in the neurohypophysis, where the neurohypophyseal peptides (oxytocin and arginine vasopressin, or their homologues) and associated proteins (neurophysins) are secreted into the circulation. Indeed, the neurohypophysis has been a classical in vitro preparation for measuring the calcium-dependent release of peptide hormones under various conditions of stimulation (Douglas, 1963; Douglas & Poisner, 1964), and extracellularly recorded compound action potentials are readily measured in this organ (Dreifuss, Kalnins, Kelley & Ruf, 1971). The analysis of excitation-secretion coupling in the neurohypophysis has been compromised, however, by the circumstance that these nerve terminals are too small for intracellular measurement of the electrophysiological events that affect release.

Resting and slowly changing transmembrane voltages in synaptosomes prepared from rat brain homogenates had already been measured using cyanine (Blaustein & Goldring, 1975) and merocyanine (Kamino & Inouye, 1978) dyes a decade ago, and action potentials have been recorded from the growth cones of tumour cells in tissue culture using an oxonol dye (Grinvald & Farber, 1981). However, it was not until 1983 that it became possible to record the action potential from intact nerve terminals in a vertebrate, with good fidelity to the time course of the transmembrane potential change (Salzberg et al. 1983). This was achieved in the neurohypophysis of Xenopus, where we took advantage of the extraordinary qualities of this preparation for optical recording. In particular, the homogeneity of the neurohypophysis, the tremendous proliferation of nerve terminals, and the relative lack of excitable membrane that is not involved in secretion make this tissue nearly ideal for optical measurement. In addition, there is no postsynaptic membrane to confound the interpretation of the optical signals. Because the optical recordings of the action potential in the nerve terminals of the frog neurohypophysis could be obtained with extraordinary signal-to-noise ratios, we could manipulate the shape of the action potential by changing extracellular calcium concentration and with other agents known to alter the release of neurohormones and neurotransmitters (Salzberg et al. 1983). Organic, as well as inorganic calcium channel modifiers could be studied (Salzberg et al. 1983; A. L. Obaid, R. Flores & B. M. Salzberg, in preparation), and we could also show that when voltage-sensitive Na⁺ channels were blocked by tetrodotoxin (TTX) and K⁺ channels were blocked by tetraethylammonium (TEA⁺) (Katz & Miledi, 1967), direct electric field stimulation of the nerve terminals evoked large active responses that reflected the same inward Ca²⁺ current that is associated with hormone release from these nerve terminals (Obaid et al. 1985).

Fig. IB shows an optical recording that represents the intracellular potential changes during the action potentials in a population of synchronously activated nerve terminals in the neurohypophysis of the frog, Xenopus laevis. This recording was photographed from an oscilloscope screen without signal averaging. The technique used here, multiple site optical recording of transmembrane voltage (MSORTV), is an extension of the method used to record extrinsic absorption changes in the squid giant axon and in many other preparations (Salzberg et al. 1977,1983; Grinvald, Cohen, Lesher & Boyle, 1981; Senseman, Shimizu, Horwitz & Salzberg, 1983; Salzberg, 1983) and has been described in detail in these reports and in an article by Cohen & Lesher (1986). The MSORTV system is illustrated schematically in Fig. 1A. Briefly, light from a tungsten-halogen lamp is collimated, made quasi-monochromatic with a heat filter (KG-1, Schott Optical Co., Duryea, PA) and an interference filter (700nm; 70nm full width at half maximum), and focused by means of a bright-field condenser onto the neurohypophysis (pars nervosa). The neurohypophysis had been excised together with the hypothalamus and the infundibulum, and the entire preparation was vitally stained by incubating it for 25 min in a 100 μ gml⁻¹ solution of the merocyanine-rhodanine dye, NK2761 (Nippon Kankoh Shikiso Kenkyusho Inc., Okayama, Japan) (Kamino, Hirota & Fujii, 1981; Gupta et al. 1981) in Xenopus Ringer’s solution (112mmoll⁻¹ NaCl; 2 mmol I⁻¹ KC1; 2mmoll⁻¹ CaCl₂; 15mmoll ^{− l} Hepes, 33 mmol I⁻¹ glucose; pH adjusted to 7·35). Light transmitted by the stained preparation was collected by a high numerical aperture water-immersion objective, which formed a real image on a 12 × l2-element silicon photodiode matrix array (MD 144-0; Integrated Photomatrix Inc., Mountainside, NJ) located in the image plane of a compound microscope (UEM, Carl Zeiss, Inc., Oberkochen, FRG). The photocurrents generated by the central 124 array elements were separately converted to voltages and amplified as described earlier (Salzberg et al. 1977; Grinvald et al. 1981). In most of the experiments, all the amplifier outputs were passed to a data acquisition system based on a PDP11/34A computer (Digital Equipment Corp., Maynard, MA) capable of acquiring a complete frame every 0·8 ms with an effective resolution of 18 bits. This data acquisition system is similar to that described previously by Grinvald et al. (1981) and employed by Grinvald, Manker & Segal (1982), Senseman et al. (1983) and Salzberg et al. (1983). In some experiments, the temporal resolution was markedly improved (17 μ s per point for a single detector element) when selected outputs of the current-to-voltage converters were also passed in parallel to a 16-channel signal averager (TN 1500; Tracor Northern Inc. Middleton, WI). These digitized output signals could be stored on magnetic tape for later display and analysis. A map of the preparation was obtained by superimposing a transparent overlay representing the photodiode array elements on a photograph taken through the trinocular tube of the microscope after removing the photodiode array housing. The experiment shown in Fig. 1A employed a 20 ×, 0·33 numerical aperture water-immersion objective (Nikon) to image part of the neurointermediate lobe (neurohypophysis plus pars intermedia) onto the photodetector array. Fig. 1C is a drawing of the projected region of the gland, prepared from the photomicrograph. Fig. ID shows the MSORTV display of the action potentials recorded optically, in a single sweep, from different areas of the pars nervosa, following a single brief stimulus applied to the hypothalamus. The optical signals are invariably confined to elements of the grid that correspond to the neurohypophysis; note that in this instance, a small portion of the neurohypophysis lies under the pars intermedia (row 9). The optical signal illustrated in Fig. IB represents the analog output from channel E5. This record, which is typical, illustrates the large signal-to-noise ratio that may be achieved in these experiments. A variety of control experiments showed that these signals represent voltage changes; e.g. no signals could be recorded in the absence of illumination or in white light; the signals could be inverted at shorter wavelengths (Senseman & Salzberg, 1980).

Evidence that the optical signals shown in Fig. 1 do, in fact, represent the membrane potential changes associated with the invasion of the nerve terminals of the neurohypophysis by the action potential was provided indirectly by some morphometric observations. In the rat, it is estimated that the 18000 magnocellular neurones in the hypothalamus give rise to approximately 40000000 terminal endings in the neurohypophysis, and morphometric data suggest that here secretory endings and swellings comprise 99% of the excitable membrane area, with axons making up the remaining 1% (Nordmann, 1977). Comparably detailed morphometric studies do not yet exist for the neurohypophysis of Xenopus, but the structural resemblance of the anuran neurohypophysis to that of the rat (Gerschenfeld, Tramezzani & De Robertis, 1960; Rodriguez & Dellman, 1970; Dellman, 1973) suggests that, in Xenopus, it is also very heavily enriched with nerve terminals. Thus, we concluded that the optical signals shown in Fig. 1, originating in the lateral regions of the neurohypophysis, are, at the least, dominated by the membrane potential changes in the terminals of the magnocellular neurones (for additional arguments, see Salzberg et al. 1983).

The nerve terminal action potentials shown in Fig. 1 exhibit two rather unexpected features. First, they have relatively long durations (i.e. full width at half height of about 6 ms) compared with those recorded from vertebrate axons (e.g. ⪡ 2 ms in the frog, Stämpfli & Hille, 1976). There must be, of course, some temporal dispersion in the population recorded by a single photodetector element, and 6ms represents, therefore, an upper bound on the spike width. However, internal evidence suggests that, in fact, very little temporal dispersion is present, since the spike width is constant over a considerable range of objective magnifications. Also, the value for the width of the action potential is in good agreement with the 5 – 6 ms duration reported for the action potentials measured in magnocellular neurone somata (Poulain & Wakerly, 1982) and for the compound action potential in the neurohypophysis (Dreifuss et al. 1971). Perhaps the more remarkable characteristic of the nerve terminal action potential illustrated in Fig. 1 is its conspicuous after-hyperpolarization that is reminiscent of the afterhyperpolarization frequently associated with a calcium-mediated potassium conductance (e.g. snail neurones, Meech & Strumwasser, 1970; rat myotube, Barrett, Magleby & Palotta, 1982; guinea pig Purkinje cells, Llinas & Sugimori, 1980a,b). We found that it was possible to alter dramatically the after-hyperpolarization recorded optically, by changing the ionic composition of the bath. Fig. 2 shows the results of a series of experiments designed to examine the calcium dependence of this long-lasting hyperpolarization. The traces shown here represent typical outputs from single channels of the 124-channel MSORTV system; each trace was recorded in a single sweep. In each row, the optical signal on the left was recorded in normal Ringer’s solution which contained 2 mmol I⁻¹ Ca²⁺. These control records are normalized to the same peak height. The middle trace in Fig. 2A Shows the effect of a 10-min exposure to elevated (5 mmol I⁻¹) extracellular Ca²⁺ concentration; the after-hyperpolarization increased by approximately 90%. This probably represents a lower limit since it is likely that some dye was bleached during the measurement, and some diminution of the signal is almost inevitable. We always found (Salzberg et al. 1983) a large increase in the size of the afterhyperpolarization, although the magnitude of the effect was variable. An increase in the height of the action potential (14% in Fig. 2A) was often noted; this finding suggests that an inward Ca²⁺ current might contribute to the normal rising phase of the action potential (see below). The record on the right in Fig. 2A shows the optical signal from the same region of the neurohypophysis 10 min after the return to 2 mmol I⁻¹ Ca²⁺. The optical traces in Fig. 2B show the effect of reducing extracellular Ca²⁺ to 0·1 mmol I⁻¹ by replacing the Ca²⁺ with Mg²⁺. The middle trace shows that after 10 min the undershoot is completely absent, replaced by a long-lasting depolarization. The magnitude of this effect also was variable from one experiment to the next; however, a reduction of extracellular calcium to 0·1 mmoll⁻¹ or less never failed to reduce the after-hyperpolarization. Further lowering of the extracellular Ca²⁺ concentration occasionally eliminated the action potential entirely, but we consider this to be a consequence of steady-state depolarization to which the optical measurements are insensitive. The right-hand trace in Fig. 2B again illustrates the recovery (10 min following return to 2 mmol I⁻¹ Ca²⁺). In this experiment, reduction of the magnitude of the signal resulting from bleaching was more apparent. Although the effects of changes in extracellular Ca²⁺ concentration were always reversible, this was not the case with other ionic substitutions. Cadmium, for example, is known to block calcium currents in a variety of preparations (e.g. Standen, 1981) and Fig. 2C shows that after 2 min in 200 μmoll⁻¹ Cd²⁺ the after-hyperpolarization was almost entirely eliminated. This effect was only partially reversible. Longer exposure to cadmium eliminated the after-hyperpolarization entirely. The peak height was also considerably reduced but, in the absence of recovery, it is impossible to know whether this is, at least partially, an effect of bleaching. Fig. 2D shows that nickel, another inorganic calcium blocker (Kaufmann & Fleckenstein, 1965) at millimolar concentrations, also blocks the spike after-hyperpolarization. In some experiments, Ni²⁺ produced a small but consistent increase in the spike height. The origin of this effect is not clear, and, although a decrease in sodium inactivation is possible, a change in resting potential cannot be ruled out. All the experiments illustrated in Fig. 2 suggest that calcium plays a significant role in shaping the action potential in the vertebrate nerve terminals studied here. Experiments with other inorganic calcium antagonists such as manganese, cobalt and lanthanum are consistent with this idea, although their potency seems to be less than that of cadmium.

Because the entry of calcium plays an important role in excitation-secretion coupling, we attempted to determine whether an inward calcium current contributed measurably to the depolarizing phase of the action potential in the terminals of the neurohypophysis. A priori, calcium is unlikely to be the dominant carrier of the inward current in these nerve terminals because the very large surface-to-volume ratio in these small secretory endings would then result in the imposition of a large, probably intolerable calcium burden. Fig. 3 illustrates an experiment designed to examine this point. Because calcium antagonists leave the depolarizing phase of the action potential largely intact, sodium appears to be carrying the bulk of the inward current. If so, it should be blocked by tetrodotoxin (TTX), and a positive result would provide evidence that the rising phase of the action potential is due primarily to the opening of voltage-dependent sodium channels. To test this possibility, it was necessary to change the stimulation conditions, since, in the presence of TTX, block of the sodium action potential in the axons of the infundibulum would prevent the excitation of the nerve terminals in the neurohypophysis by blocking spike conduction. We were able to circumvent this by resorting to a form of field stimulation in which the lateral tips of the pars nervosa were held by suction electrodes, and very brief (200 – 500 μ s) currents were passed between the two electrodes. In this fashion, the terminals could be excited directly, without the requirement for axonal transmission. This technique was employed in the experiment shown in Fig. 3. The MSORTV record on the left in Fig. 3A shows the control, in 2 mmol I⁻¹ Ca²⁺ Ringer’s solution. The amplitude of the optical spike increased with field strength, as expected, since a large and potentially variable number of nerve terminals were stimulated and monitored. A maximal response was obtained typically with stimuli of 100 – 200 V and was largely independent of stimulus polarity. As with the optical signals elicited by stimulation of the hypothalamus, the amplitude of the response was approximately 0·3% of the transmitted light intensity at 700 nm. The middle trace was recorded from the same population of nerve terminals following 18 min of exposure to 200 μmoll⁻¹ Cd²⁺ to block any inward calcium current. The reduction in the magnitude of the action potential is obvious, along with a considerable broadening of the spike and the complete loss of the after-hyperpolarization, the latter effects probably resulting from the elimination of the large calcium-mediated potassium conductance. The extent of the reduction in the height of the action potential may be exaggerated, however, by factors other than blockage of calcium channels (e.g. dye bleaching). The righthand trace in Fig. 3A shows that 6min after application of 5 μ mol 1 ^{− 1} tetrodotoxin (TTX), the Cd²⁺-resistant portion of the terminal action potential was completely eliminated. The residual signal seen in this record is the passive depolarization produced by the field stimulation; it was reversed with reversal of stimulus polarity, and it exhibited the same wavelength-dependence as that of the optical recording of the normal action potential. (It should be noted, however, that the size of the passive electrotonus was very variable from element to element.)

The complementary experiment is shown in Fig. 3B. Here, the sodium current was first blocked with TTX (middle trace), and the small residual active response was eliminated by Cd²⁺ (righthand trace), and substantially decreased in size by 1 mmol I⁻¹ Ni²⁺ (not shown). Thus, the experiments illustrated here, and others, suggest that the rising phase of the nerve terminal action potential, in the neurohypophysis of Xenopus, is mediated predominantly by sodium, although these data also suggest the presence of at least a small calcium component. The magnitude of this calcium component is difficult to assess quantitatively, particularly because the TTX blockade of the normal regenerative sodium depolarization may limit the opening of voltage-dependent calcium channels. The important role for calcium entry in excitation-secretion coupling is fully consistent with its apparently small contribution to the terminal action potential (Katz & Miledi, 1969); in adrenal chromaffin cells, for example, where calcium influx is the critical event in excitation-secretion coupling (Douglas, 1978), the Ca²⁺ component of the action potential is also small (Brandt, Hagiwara, Kidokoro & Miyazaki, 1976).

Nonetheless, under conditions in which the terminal membrane depolarization is prolonged, sufficient inward calcium current may develop to give rise to a relatively large calcium action potential. The records shown in Fig. 4 are from one of many neurohypophyses in which tetraethylammonium (TEA⁺) was added, in addition to TTX, to block the voltage-dependent potassium conductance and, thereby, prolong the depolarization of the nerve terminal membrane. Fig. 4A shows a control action potential from a population of nerve terminals. The action potential exhibits the characteristic rapid upstroke that we attribute primarily to the effect of a fast inward sodium current, a rapid repolarizing phase resulting from K⁺ efflux, largely through the delayed rectifier, and the typical afterhyperpolarization resulting from a Ca²⁺-mediated increase in K⁺ conductance. When 5 mmol I⁻¹ TEA⁺ was added to the Ringer’s solution (not shown), the spike was prolonged and, in most experiments, a hump appeared on the falling phase, possibly due to repetitive firing of some of the terminals. The addition of a high concentration of TTX, sufficient to block voltage-sensitive sodium channels, abolished the hump and unmasked a large after-hyperpolarization that presumably results from the enhanced activation of a Ca²⁺-mediated K⁺ conductance. Fig. 4B shows the optical response recorded from the same population of terminals represented in Fig. 4A, following 20 min of exposure to 5 mmol I⁻¹ TEA⁺ and 2 μ moll⁻¹ TTX. The regenerative response illustrated is smaller and slower than the normal action potential, and appears to depend entirely upon the influx of calcium and the subsequent rise in a TEA⁺-insensitive K⁺ conductance. The further addition of 0·5 mmol I⁻¹ Cd²⁺ to the bath (Fig. 4C) completely blocked the remaining active responses. (We have already seen, in Figs 2, 3 that such low levels of cadmium do not block the action potentials in the nerve terminals of the neurohypophysis in the absence of TTX.) The residual optical signals shown in Fig. 4C represent the purely passive electrotonus, as indicated by their symmetrical response to stimulus polarity. The combined effects of TEA⁺ and Cd²⁺ could often be reversed, but it was never possible to reverse completely the effect of TTX.

Although the optical signal recorded in Fig. 4B has the appearance of a calcium action potential, it is possible that a major contribution to the rising phase of the response, in TTX, comes from the influx of Na⁺ through Ca²⁺ channels, which are not blocked by the toxin. It was possible, in these experiments, to assess qualitatively the relative contributions of Ca²⁺ and Na⁺ to the active response by recording optical signals at different [Ca²⁺]₀ at high and low [Na⁺]₀, and at different [Na⁺]₀ at high and low [Ca²⁺]₀. These experiments (Obaid et al. 1985), all carried out in TTX/TEA⁺, suggested that, in low [Ca²⁺]₀ (0·2 mmol l^{− 1}), Na⁺ might contribute significantly to the inward current conducted by cadmium-sensitive, TTX/TEA⁺-insensitive channels. However, at normal [Ca²⁺]₀, varying [Na⁺]₀ had little effect on the rate of rise of the optical signal, as if sodium ions were unable to compete effectively with normal concentrations of Ca²⁺.

Although the evidence presented above (Figs 2– 4) indicates the presence of a prominent calcium-mediated potassium conductance in the terminal membrane, the inhibition of this conductance by a high-affinity blocker would provide important complementary evidence. Charybdotoxin (CTX), a protein toxin derived from scorpion venom, blocks calcium-sensitive potassium channels from skeletal muscle in planar lipid bilayers (Miller, Moczydlowski, Latorre & Phillips, 1985) with a K_i of 10 nmol I⁻¹. We have observed (Obaid & Salzberg, 1985; A. L. Obaid, R. Flores & B. M. Salzberg, in preparation) that charybdotoxin, at 50 nmol I⁻¹, specifically blocks the after-hyperpolarization of the nerve terminal action potential in Xenopus. In addition, both the width and height of the action potential are increased in the presence of CTX. Fig. 5 shows the effect of 50 nmol I⁻¹ CTX on the nerve terminal action potential recorded optically. Fig. 5A is the action potential elicited in normal Ringer’s solution by direct-field stimulation, and Fig. 5B shows the complete elimination of the after-hyperpolarization following an 8-min exposure to CTX. (Note that there is a 10% increase in the height of the action potential, in addition to the loss of the undershoot.) Evidence that charybdotoxin acts directly on a calcium-mediated potassium conductance, rather than indirectly by blocking calcium entry, is provided in Fig. 6. Fig. 6A shows the action potential in normal Ringer’s solution, and Fig. 6B illustrates the calcium response obtained 8 min following the addition of 5 mmol I⁻¹ TEA⁺ and 1μ mol I⁻¹ TTX to the Ringer’s solution. This response also exhibits an after-hyperpolarization, similar to that seen in Fig. 3. Fig. 6C demonstrates the effect of 23 min of exposure to 50 nmol I⁻¹ charybdotoxin (in TEA⁺/TTX Ringer’s solution). The after-hyperpolarization is again eliminated, out calcium entry is actually enhanced. At the same time, the loss of an outward current prolongs the duration of the response. Evidence that the active response is mediated by calcium entry is given in Fig. 6D, where a 2-min exposure to 0·5 mmol I⁻¹ Cd²⁺ entirely abolishes the active response.

The calcium-mediated active responses, recorded optically and described here, appear to share some properties with the local subthreshold action potentials described by Hodgkin (1938) in unmyelinated crab nerve; with TTX blocking any propagated action potential in the axons of the infundibulum, the small calcium currents cannot excite a sufficient length of nerve to produce a propagated action potential. Nonetheless, their dependence on [Ca²⁺]₀ and sensitivity to Cd²⁺ block indicates that they probably result from a voltage-sensitive calcium influx into the terminals (Kostyuk & Krishtal, 1977; Hagiwara & Byerley, 1981; Obaid et al. 1985) which mediates hormone release.

Aminoglycoside antibiotics, including neomycin, streptomycin and gentamicin, depress evoked transmitter release at the neuromuscular junction by competing with extracellular calcium (Fiekers, 1983). We were able to show (Parsons, Obaid & Salzberg, 1986) that neomycin (220μ moll⁻¹) reduced the height of the upstroke of the nerve terminal action potential, increased the spike duration and abolished the after-hyperpolarization, in the neurohypophysis of Xenopus. These effects were reversible upon washing, and could be antagonized by raising [Ca²⁺]₀ from 2 to 5 mmol I⁻¹ (Parsons et al. 1986). When the effects of these agents on the regenerative calcium response were examined, we found that both neomycin (220μmol I⁻¹) and gentamicin (190 μ mol I⁻¹) decreased the height of the calcium spike, and practically eliminated the after-hyperpolarization. These observations all supported the idea that aminoglycoside antibiotics antagonize the entry of calcium into vertebrate nerve terminals during excitation.

Other organic modifiers of calcium channel behaviour have been used by several researchers to distinguish different types of calcium channel in a variety of preparations (Miller, 1987). Previously, however, it has not been possible to use these pharmacological tools to study calcium channels in the intact nerve terminals of a vertebrate. Using the extrinsic absorption changes exhibited by the terminals of the frog neurohypophysis stained with the merocyanine-rhodanine dye NK2761, we carried out a series of experiments (Salzberg, Obaid & Flores, 1987; A. L. Obaid, R. Flores & B. M. Salzberg, in preparation) that help to define the pharmacological profile of the calcium channels present in intact neurosecretory terminals of vertebrates. Because the g_K(Ca)-dependent afterpotential of the spike such a sensitive indicator of calcium entry into the terminals of the neurohypophysis (Salzberg et al. 1983), we used this component of the optical signal to examine the sensitivities of the calcium channels that are present in this population of vertebrate nerve terminals. We used the dihydropyridine compounds nifedipine and Bay-K8644, at concentrations ranging from 2 to 5 μmoll⁻¹, and the peptide toxin ω-conotoxin GVIA (Olivera et al. 1985). In the frog neurohypophysis, neither dihydropyridine compound had any effect on the calcium-dependent components of the action potential (Figs 7, 8), the spikes remaining absolutely unchanged for up to 1h in the presence of either nifedipine or Bay-K8644. ω-Conotoxin GVIA (l/rmolP¹), however, rapidly abolished the after-hyperpolarization of the normal action potential (Fig. 9) by blocking calcium entry (and not by reducing g_K(Ca))-These results demonstrated clearly the effect of ω -conotoxin on the calcium channels involved in the release of neuropeptides (see below) from the intact nerve terminals of vertebrates and, although this is the first direct evidence from measurements on intact nerve terminals, it is consistent with the findings of others using patch-clamp techniques on neuronal preparations (McCleskey et al. 1987; Fox, Nowycky & Tsien, 1987). The insensitivity to dihydropyridines exhibited by these same calcium channels is apparently at variance with some reports of dihydropyridine block of neuronal calcium channels (for a review, see Tsien, 1986). However, Rane, Holz & Dunlap (1987) concluded that chronic depolarization is required to achieve nifedipine block of calcium channels in chick DRG neurones, but is not required for Bay-K8644 action (Holz, Dunlap & Kream, 1988). Our results with nifedipine are consistent with this interpretation, although our failure to observe any enhancement of calcium channel activity in the presence of Bay-K8644 is not.

The anatomy of the amphibian posterior pituitary provides a unique opportunity for the study in vitro of excitation-secretion coupling in the vertebrate: as we pointed out earlier, there is no postsynaptic excitable membrane to confound the interpretation of the optical signals (Salzberg et al. 1983; Obaid et al. 1985) and, more important for biochemical studies, no barrier to the collection of its secretory products. However, relatively little is known about the secretion of amphibian neuropeptides and neurohormones, although an extensive literature has accumulated on the biochemistry of secretion in the mammalian, and particularly the rodent, neurohypophysis. Together with Hal Gainer, we felt that it would be important to be able to correlate the electrical events in the terminals with the kinetics of release of the secretory products, and we thought that this might be accomplished by means of optical recording of membrane potential combined with sensitive radioimmunoassays using the CD-I mouse.

Fig. 10A shows an optical recording of a train of seven action potentials, stimulated at 16 Hz, in a population of nerve terminals in the neurohypophysis of the CD-I (Charles River Breeding Laboratories) mouse. It represents the change in transmitted light intensity at 675 nm, monitored in a single sweep by one channel of the 124-channel MSORTV system. The preparation had previously been incubated in a Ringer’s solution (in mmol I⁻¹: NaCl, 154; KC1, 5·6; MgCl₂,1; CaCl₂, 2·2; glucose, 10; Hepes, 20, adjusted to pH 7·4 with NaOH) containing 0·1mgml⁻¹ of the merocyanine-oxazolone dye NK2367 (Salzberg et al. 1977). The method used here was virtually identical to that used to record electrical activity from the neurosecretory terminals of the frog neurohypophysis. The trace in Fig. 10A exhibits a gradual decline in baseline, followed by recovery, with the appearance of a transient hyperpolarization. The constant size of the upstroke, however, suggested to us that this apparent hyperpolarization might not, after all, represent a change in membrane potential. Fig. 10B shows a single trial recorded by the same detector element, except that the measuring wavelength was 540 nm. At this wavelength, the potential-dependent absorption change exhibited by membrane stained with NK2367 is known to be opposite in sign to that observed at 675 nm (Senseman & Salzberg, 1980). Notice, however, that although the optical spikes are reversed in sign, the slow component of the signal is essentially unchanged. This weak dependence on wavelength of the slow component of the optical trace suggested that it might reflect a change in light scattering. Fig. 10C illustrates an identical experiment, carried out on another mouse neurohypophysis, before staining. Here, the dye-dependent extrinsic absorption signal that depends upon membrane potential is absent, revealing a series of increases in transmitted light intensity (downward deflections of the trace), each preceded by a transient decrease in transparency (upward deflection) occurring at the time of the action potential. This experiment employed a relatively long a.c. coupling time constant (3 s) to minimize the distortion of the time course of the optical signal. Nonetheless, the tendency towards recovery following the train of stimuli is largely an artefact of the coupling time constant. A similar experiment is illustrated in Fig. 10D, in which a digital sample-and-hold circuit, consisting of 12-bit analog-to-digital and digital-to-analog converters connected in series is used instead of capacity coupling. This method avoids any a.c. distortion of the signal, which is observed to remain nearly constant for seconds following the cessation of the stimuli.

Fig. 10E illustrates the wavelength-dependence of these intrinsic optical signals between 600 and 860 nm, corrected for the spectral efficiency of the apparatus. The wavelength range was selected to exclude major absorption peaks of intrinsic pigments, e.g. haemoglobin. The fractional change in transmitted intensity exhibits a gradual decline with increasing wavelength, consistent with a change in light scattering. All components of the optical signal were eliminated when the neurohypophysis was depolarized with KC1 or when the Ringer’s solution contained micromolar concentrations of TTX.

It should be clear to the reader that optical techniques which employ visible light, especially those that rely upon intrinsic optical changes that require no exogenous probes, offer unique capabilities in terms of resolution in time and space, combined with a gentleness that permits the non-destructive measurement of many cellular and subcellular processes. In particular, light-scattering methods have been applied to biological systems, in vitro and in vivo, since the experiments of Tyndall (1868) and have played a role in a variety of recent efforts to study changes in neurone structure during action potential propagation and synaptic transmission (Cohen, 1973). Cohen & Landowne (1970) reported small (8 × l0 ^{− 7}) changes in light scattering in the skate electric organ that appeared to be presynaptic in origin, and Shaw & Newby (1972) detected calcium-dependent movement in a thoracic ganglion of a locust (Schistocerca gregaria) by monitoring the power spectrum of ‘twinkling’ (light beating) produced by scattered laser light (see also Piddington & Sattelle, 1975; Englert & Edwards, 1977). These preliminary reports, however, have not been expanded.

In the absence of potentiometric dye staining of the terminals of the neurohypophysis, two intrinsic optical signals, illustrated in Fig. 10C, were evident following each stimulus. The rapid upward deflection of the trace (decrease in transparency), which we have termed the E-wave (Salzberg et al. 1985) appears to coincide with the arrival of the action potential at the terminals (cf. Fig. 10A), and the subsequent increase in transparency, the S-wave, appears to correspond towards slower and longer-lasting process occurring in the nerve terminals. Several lines of evidence suggested that this latter intrinsic optical signal (Fig. 10C,D) is closely correlated with the secretory process. If this association is more than accidental, we would anticipate that the long-lasting changes in transparency (but not the E-wave) would exhibit certain properties that characterize neurosecretory systems in general, and the release of neuropeptides in particular; viz, dependence on stimulation frequency, with marked facilitation; dependence on [Ca²⁺]₀; and sensitivity to Ca²⁺ antagonists and to pharmacological procedures known to influence secretion (Douglas & Poisner, 1964).

The frequency-dependence of the transparency changes (intrinsic optical signals), which we have referred to as ‘light scattering’ (Salzberg et al. 1985), is illustrated in Fig. 11. In each case, the stimuli were delivered to the neurohypophysis for 400ms, but at different frequencies. Fig. 11A – D shows, not surprisingly, a monotonic increase in the total size of the light-scattering change with the number of stimuli. The size of the fractional change in light scattering per stimulus (Δ_50–460ms/N), however, varied with frequency, reaching a broad maximum at between 8 and 16 Hz. Fig. HD demonstrates the tendency of the total optical change to saturate at very high stimulation rates. In each instance, a marked facilitation of the optical response is evident between the first and second stimulus. Fig. HE was obtained under identical conditions (16 Hz) to Fig. 11C, except that 1 mmol I⁻¹ CdCl₂ was added to the bath for 15 min before the single trial recorded here. In Fig. HE, Cd²⁺ is seen to block more than 90% of the S-wave (downward deflection of the trace; decrease in opacity) of the light-scattering change without significantly affecting the rapid upstroke (E-wave) that occurs coincident with the action potential. The plot in Fig. 11F summarizes the dependence of the magnitude of the light-scattering change on the rate and number of stimuli and also includes the effect of barium ions added to the normal calcium concentration (see below).

Calcium profoundly influences neurosecretory activity (Dodge & Rahamimoff, 1967; Katz, 1969; Douglas, 1978) and the extracellular concentration of this ion would be expected to modulate the size of any intrinsic optical signal related to secretion from the terminals of the neurohypophysis. Fig. 12 shows the effect of [Ca²⁺]₀ on the light-scattering signal evoked by stimulation of the infundibulum at 10 Hz for 400 ms. Traces A – F, which are each the average of 16 sweeps, were obtained in alphabetical order. The effect of calcium concentration is apparent, and the blockade of the optical signal by 0·5 mmol I⁻¹ Cd²⁺ in Fig. 12F is dramatic. Calcium, of course, has other effects, including a direct effect on excitability (Frankenhaeuser & Hodgkin, 1957) and these must be considered. To eliminate the effects of threshold variation, the stimuli were supramaximal in each instance. Examination of the records in Fig. 12 (compare, for example, B and D) reveals significant differences in the amplitudes of the E-waves that precede each S-wave in the light-scattering signal and coincide in time with the terminal action potential. The E-wave evidently has a different origin from the large S-wave that is blocked by cadmium. We have suggested (Salzberg et al. 1985) that this very each intrinsic optical change (E-wave) reflects the arrival of the action potential at the terminals of the neurohypophysis, and we assumed that, as a compound optical signal, its size is roughly proportional to the number of terminals activated synchronously under a given set of conditions. On this assumption, we have compensated for changes in the invasibility of the tissue, resulting somehow from changes in extracellular calcium concentration, by normalizing the S-wave according to the size of the initial E-wave. This procedure has two potential difficulties. The first has to do with the interpretation of the E-wave. The peak of the E-wave occurs slightly before the peak of the voltage change in the terminals, as determined by the response of voltage-sensitive dyes. And, although the extrinsic absorption signals could be distorted if the dye response exhibited a slow component, the merocyanine-rhodanine and merocyanine-oxazolone dyes, at the concentrations used in these experiments, respond in less than 2 μs to step changes in voltage (B. M. Salzberg, A. L. Obaid & F. Bezanilla, unpublished observations) and their overall response is dominated by this fast component. If, instead, the E-wave depended on the inward current, it might be expected to reach a maximum at the time of the maximum rate of rise of the extrinsic absorption (voltage) signal. In fact, the time-to-peak of this early component of the opacity change falls somewhere between the time-to-peak of the voltage change and the time-to-peak of the time derivative of the absorption (voltage) change. This result suggests that the ultimate interpretation of the E-wave will be complicated and that, although this portion of the intrinsic optical signal is related to excitation, it may exhibit both current-and potential-dependence (Cohen, Keynes & Lan-downe, 1972a, b).

A second possible difficulty involves our assumption that the size of the E-wave, whether related primarily to current or to voltage, is proportional to the number of active terminals. This can only be approximately correct, since it will be affected by changes in the magnitude of the upstroke of the terminal action potential, particularly those resulting from changes in the resting membrane potential. It will also be sensitive to changes in the temporal dispersion of excitation of the terminals, and small changes in the size of the factor used for normalizing the optical signals might seriously distort the quantitative analysis of the optical signals. On balance, however, under constant stimulation conditions, the height of the E-wave should provide a rough comparative estimate of the degree of invasion of the tissue and thereby offer a convenient normalization for the larger lightscattering changes that appear to be related to secretion (S-wave). In any event, we have taken care that none of the qualitative conclusions that we have drawn (Salzberg et al. 1985) are affected by this procedure. Accordingly, Fig. 12G shows records obtained in normal Ringer’s solution (A), together with records obtained in low [Ca²⁺]₀ (B), high [Ca²⁺]₀ (D), and normal Ringer’s solution to which 0·2 mmol I⁻¹ Cd²⁺ had been added (F), all normalized so as to equalize the sizes of the initial E-waves. In this way, the effect of [Ca²⁺]₀ on the light-scattering signal per active terminal is demonstrated most convincingly.

Heavy water (deuterium oxide) is known to depress excitation-contraction coupling in different muscle types (Kaminer, 1960; Svensmark, 1961; Bezanilla & Horowicz, 1975) and excitation-secretion coupling in a variety of systems including pancreatic beta cells (Lacy, Walker & Fink, 1972). In the neurohypophysis of the mouse, complete (>98%) replacement of the water in the Ringer’s solution by D₂O resulted in a reduction in the size of the light-scattering signal by about 60%. Fig. 13 illustrates the effect of D₂O substitution on the optical response to stimulation of the terminals of the mouse neurohypophysis. Fig. 13A –C shows light-scattering changes at 675 nm obtained during single trials in normal Ringer’s solution (A), 30min in D₂O-substituted Ringer’s solution (B), and 40 min following the return to normal Ringer’s solution (C). The depression produced by deuterium oxide seems to reflect primarily a decrease in the light scattering from each active terminal, consistent with earlier observations of the effects of D₂O (Kaminer, 1960; Svensmark, 1961; Lacy et al. 1972; Bezanilla & Horowicz, 1975), rather than a decrease in the number of active terminals, as judged by the very small variation in the size of the E-wave. This is illustrated in Fig. 12D, where the averages of 16 trials are shown, each record obtained immediately following the corresponding single trial. These records have been normalized to the height of the first peak, to compensate for small differences in the number of active terminals. It should be noted that replacement of water with D₂O produces a shift in the true pD, compared to the value measured with a glass pH electrode (Glascoe & Long, 1960) of approximately 0·4 pD units in the ‘alkaline’ direction. To control for the possibility that the depression of the lightscattering change shown in Fig. 13 resulted from a pD shift, we repeated these experiments in normal Ringer’s solution with its pH adjusted to 8·0. This procedure had no effect on the normal light-scattering signals.

Rendering the Ringer’s solution bathing the neurohypophysis hypertonic is expected to increase the volume of the extracellular space by shrinking the terminals. We might be able to detect this effect qualitatively in several different ways. Perhaps simplest would be to note the effect of hypertonicity on the appearance of extracellular potassium accumulation (Frankenhaeuser & Hodgkin, 1956). Fig. 14 illustrates this behaviour indirectly in a neurohypophysis that had been stained with the merocyanine-oxazolone dye NK 2367 (Salzberg et al. 1977). Fig. 14A shows the sum of the extrinsic absorption change, representing membrane potential, and the light-scattering signal, during stimulation at 16Hz for 400ms in normal Ringer’s solution. The decrease in the overall size of the action potential (20% between the first and last spike in the train) suggests that, in this preparation, sufficient potassium is accumulating extracellularly to affect the undershoots of the action potentials, but that this effect is masked by the downward deflection produced by the S-wave of the light-scattering signal. When the solution bathing the preparation is made 20% hypertonic by the addition of sucrose, the reduction in spike amplitude is less pronounced (8%), as though the effective extracellular volume was decreased by the shrinking of the terminals. The light-scattering signal, however, is not appreciably reduced.

The intrinsic optical signals described above are closely correlated with the secretory activity of the magnocellular neurone terminals located in the neurohypophysis of the mouse. A mechanical artefact associated with contraction of vascular smooth muscle cannot be ruled out unequivocally and, indeed, such a effect would be expected to exhibit many of the same properties as an intrinsic signal related to secretion. However, the time course of the S-wave, the component of the light-scattering change that we associate with secretion, would seem to be too fast by at least an order of magnitude to have its origin in the contraction of smooth muscle. The S-wave reaches its half-maximal value within 7 –8 ms of the application of the stimulus pulse to the region of the pars nervosa where the infundibular stalk enters (and 3 –4 ms following the peak of the extrinsic signal). Even neglecting any conduction latency, this is considerably faster than any known mechanical response of vascular smooth muscle. Similarly, we feel justified in ruling out changes related to the terminal action potential alone, in accounting for the major component of the intrinsic optical signal. For example, extracellular accumulation of potassium ion might give rise to a change in light scattering, mediated by a transport number effect (Girardier, Reuben, Brandt & Grundfest, 1963; Barry & Hope, 1969; Cohen et al. 1972b) resulting from an increased salt concentration in the restricted space immediately outside the terminals. At least two kinds of evidence argue against such a mechanism. First, the addition of 2·2 mmol I⁻¹ barium to the normal complement of extracellular calcium increases the size of the S-wave, despite the fact that barium is unlikely to replace calcium in activating a potassium conductance (Standen & Stanfield, 1978; Vergara & Latorre, 1983) and barium is likely, in fact, to reduce the voltagedependent potassium conductance (Eaton & Brodwick, 1980; Armstrong, Swenson & Taylor, 1982; Armstrong & Taylor, 1980). Thus, the accumulation of potassium is expected to be severely decreased under these conditions. However, extracellular barium does replace calcium in some secretory systems, (Dicker, 1966; Rubin, 1974), although it is not as effective, and this is consistent with our observation that the increase in the size of the light-scattering signal with barium is smaller than that contributed by the addition of a similar amount of calcium. In this context, it would be helpful to demonstrate directly, using ion-selective electrodes, that potassium accumulation is, in fact, reduced in the presence of barium, but this has not yet been done. The experiment using sucrose (Fig. 14B) to increase the extracellular volume fraction also suggests that the light-scattering signal does not depend primarily on an alteration of extracellular space, since the effect of repetitive stimulation should be less under conditions in which the terminals are shrunk by the hypertonicity of the bath, but the light-scattering change was, in fact, not diminished. Indeed, the S-wave transparency changes that seem to reflect rapid alterations in light scattering are dramatically affected by many of the same interventions that are known to alter the release of neuropeptides from these same terminals. The dependence on stimulation frequency (Fig. 11) is in good accord with evidence obtained in several laboratories suggesting that neuropeptide secretion increases with frequency of stimulation (Dreifuss et al. 1971; Nordmann & Dreifuss, 1972; Poulain & Wakerley, 1982), and the facilitation of the intrinsic optical response seen in virtually all of the records seems particularly significant. Our own experiments, carried out in collaboration with Seth Wolfe and Hal Gainer (Gainer, Wolfe, Obaid & Salzberg, 1986) have used radioimmunoassays (RIA) for arginine vasopressin (AVP) to demonstrate the dependence upon stimulation frequency of vasopressin release in the CD-I mouse. Fig. 15 shows the RIA measurements of arginine vasopressin release from mouse neural lobes, measured during two complete cycles of 600 stimuli delivered at 1 Hz and at 20 Hz. The data were highly reproducible (Gainer et al. 1986), and exhibited an enhancement of approximately 20-fold between 20 Hz and 1 Hz. Release at 1 Hz was not distinguishable from basal levels, so that it was not possible to calculate an accurate ratio of specific AVP released per pulse at 20 Hz versus 1 Hz. However, it was apparent that the AVP released per pulse in the mouse neurohypophysis is highly frequency-dependent, as is the case in the rat neurohypophysis (Dreifuss et al. 1971; Dutton & Dyball, 1979; Nordmann & Dreifuss, 1972). Further, the amount of AVP released during a second cycle of electrical stimulation was virtually identical to that released during the first cycle, suggesting that the stimulation paradigm was not deleterious to the release mechanism, and did not deplete the terminals to any significant degree.

The effect of extracellular calcium concentration on the magnitude of the S-wave of the intrinsic optical signal (Fig. 12) is also in remarkably good agreement with the classical behaviour of secretory systems (Douglas, 1978). The effect of replacing all the water in the Ringer’s solution with D₂O, however, had not been reported in the mammalian neurohypophysis before our experiments (Salzberg et al. 1985). Under these conditions, we found (Fig. 13) a depression of approximately 60% in the S-wave of the light-scattering signal. Radioimmunoassay measurements of secreted vasopressin (Gainer et al. 1986), revealed a nearly identical decrease (Table 1) in vasopressin release from the mouse neural lobes when D₂O was substituted for H₂O under comparable stimulation conditions. These experiments also provided direct corroboration of our observations on the effects of [Ca²⁺]₀ on AVP release, estimated from the variation in the size of the S-wave, and reproduced the saturation that we found at concentrations above 5 mmol I⁻¹ (Table 1).

Another series of experiments, carried out in our laboratory together with T. D. Parsons, at the suggestion of David Quastel, has shown that neomycin and other aminoglycoside antibiotics depress the S-wave of the light-scattering signal (Parsons, Obaid & Salzberg, 1985). This finding is significant because, at low concentration, these drugs act presynaptically at the snake neuromuscular junction and depress evoked transmitter release by competing with [Ca²⁺]₀ (Fiekers, 1983). Our observation that, for example, neomycin at 220 μmoll⁻¹ in normal Ringer’s solution depressed the magnitude of the S-wave by 60% after 11 min provides additional evidence that these agents act presynaptically, since there are no postsynaptic elements in the neurohypophysis, and that they act as competitive antagonists of [Ca²⁺]₀ (Fiekers, 1983; Parsons et al. 1986). Further, the compatibility of the aminoglycoside effects on the light scattering from the neurohypophysis, with voltage-clamp data reported from the neuromuscular junction (Fiekers, 1983) reinforces our interpretation (Salzberg et al. 1985) that the light-scattering signals reflect processes closely related to neuropeptide release.

With these results in mind, it is perhaps worthwhile returning to our earlier discussion of the pharmacological profile of the calcium channels that are present in the nerve terminals of vertebrate neurohypophyses. The evident correlation between the intrinsic optical changes in the terminals, on the one hand, and the release of neuropeptides, on the other, allows us to pose the question of whether the calcium channels in the terminals which are sensitive to ω -conotoxin and insensitive to dihydropyridines are the same calcium channels that are required for normal secretion. Apparently, they are. Fig. 16 shows an experiment in which we attempted to observe the effects of dihydropyridine calcium channel modifiers on the light-scattering signal. Fig. 16A shows the intrinsic optical change recorded during a single stimulus train applied in normal Ringer’s solution. Fig. 16B shows the result of a stimulus train following 22 min in 5 μ mol I⁻¹ nifedipine. No effect observed, and this result is consistent with the absence of any effect of nifedipine on the action potential in the neurohypophyseal nerve terminals of the frog (Fig. 7). Similarly, Bay-K8644, at 5 μ mol I⁻¹ for at least 40 min (not shown) had no effect on the intrinsic optical signal exhibited by the nerve terminals of the CD-I mouse. The calcium channels required to produce the normal S-wave and, by extension, secretion of neuropeptides, are blocked by cu-conotoxin, however. This is illustrated in Fig. 17, where the inhibition of the S-wave of the light-scattering signal in the presence of 5 μ mol I⁻¹ ω -conotoxin GVIA is shown.

We have, as yet, no reason to implicate any particular step among the sequence of events that couples excitation to secretion in the generation of these large intrinsic optical signals, and the identity of their physiological basis remains unclear. The fusion of secretory vesicles during exocytosis should result in the loss of relatively high refractive index particles, and should thereby reduce the refractive index gradients in the terminal. However, the very early onset and fast rise of the S-wave suggest that the optical signal might arise as a result of some calcium-dependent process prior to the fusion of secretory vesicles and the release of their contents. For example, it is possible that the light-scattering changes described here reflect a swelling of secretory vesicles, or they may reflect, instead, changes in the cytoskeleton, or rapid alterations in the state of intracellular calcium stores following calcium entry but prior to secretion (Neering & McBurney, 1984). In the latter case, these intrinsic optical changes may be related to the transparency changes that precede tension development which have been observed in cut skeletal muscle fibres (Kovacs & Schneider, 1977; Rios, Melzer & Schneider, 1983). Identification of the physiological origin of the intrinsic optical changes must await precise measurement of light scattering per se. Quasi-elastic light-scattering measurements, particularly on simplified model systems such as secretosomes or vesicle suspensions (see Sattelle, 1988; this volume) may assist in the interpretation of these very interesting signals.

In any case, the weak dependence on wavelength of the intrinsic signal related to secretion, contrasted with the strong wavelength-dependence of the extrinsic absorption signals provided by linear potentiometric probes (e.g. merocyanines) (Ross, Salzberg, Cohen & Davila, 1974; Ross et al. 1977; Cohen & Salzberg, 1978; Salzberg, 1983) should permit one to monitor simultaneously, in a stained preparation, the voltage changes in the nerve terminals and the time course of events intimately associated with the release of secretory products. The inherently fast responses of the optical measurements may then improve our ability to resolve early events in the coupling of excitation to secretion.

We have chosen the vertebrate neurohypophysis to illustrate the technique of optical recording of membrane potential for the study of excitation at nerve terminals. These experiments led to the use of light-scattering methods for the investigation of excitation-secretion coupling, and it seems very likely that intracellular indicators for the measurement of [Ca²⁺]_i-transients will also find applications in the optical study of the secretory event in this system.

We are grateful to Dr H. Gainer for introducing us to the neurohypophysis as an experimental system and for his participation in many of the experiments described here, to Drs C. M. Armstrong, L. B. Cohen, M. Morad, R. K. Orkand D. B. Sattelle and R. W. Tsien for valuable discussions, and to Drs Orkand, Parsons, Sattelle and D. M. Senseman for their participation in some of the experiments discussed. We are also grateful to Dr Chris Miller for his generous gift of charybdotoxin, and to L. B. Cohen and S. Lesher for providing the software used for data acquisition and display. Supported by USPHS grant NS 16824.