Action Currents Generate Stepwise Intracellular Ca2+Patterns in a Neuroendocrine Cell*

It is believed that specific patterns of changes in the cytosolic-free calcium concentration ([Ca2+] i ) are used to control cellular processes such as gene transcription, cell proliferation, differentiation, and secretion. We recently showed that the Ca2+ oscillations in the neuroendocrine melanotrope cells of Xenopus laevis are built up by a number of discrete Ca2+ rises, the Ca2+ steps. The origin of the Ca2+ steps and their role in the generation of long-lasting Ca2+ patterns were unclear. By simultaneous, noninvasive measuring of melanotrope plasma membrane electrical activity and the [Ca2+] i , we show that numbers, amplitude, and frequency of Ca2+ steps are variable among individual oscillations and are determined by the firing pattern and shape of the action currents. The general Na+ channel blocker tetrodotoxin had no effect on either action currents or the [Ca2+] i . Under Na+-free conditions, a depolarizing pulse of 20 mm K+ induced repetitive action currents and stepwise increases in the [Ca2+] i . The Ca2+ channel blocker CoCl2 eliminated action currents and stepwise increases in the [Ca2+] i in both the absence and presence of high K+. We furthermore demonstrate that the speed of Ca2+ removal from the cytoplasm depends on the [Ca2+] i , also between Ca2+ steps during the rising phase of an oscillation. It is concluded that Ca2+ channels, and not Na+ channels, are essential for the generation of specific step patterns and, furthermore, that the frequency and shape of Ca2+ action currents in combination with the Ca2+ removal rate determine the oscillatory pattern.

It is proposed that the temporal and spatial aspects of the Ca 2ϩ signal determine (encode) which (sub)cellular process will be regulated (20). It seems that not only frequency modulation but also amplitude modulation can encode cellular effects (20,21). According to this principle, first messengers can regulate specific cellular activities by inducing distinct types of Ca 2ϩ signals. Consequently, the mechanism(s) cells use to generate different types of Ca 2ϩ signals are of special interest.
The present study concerns the relationship between plasma membrane electrical activity and Ca 2ϩ signaling in an excitable secretory cell, the neuroendocrine melanotrope cell of Xenopus laevis. This pituitary intermediate lobe cell releases ␣-melanophore-stimulating hormone, a peptide that causes skin darkening in animals adapted to a black background (22). The cell displays intracellular Ca 2ϩ oscillations that are regulated by neurotransmitters and neuropeptides involved in the regulation of ␣-melanophore-stimulating hormone release (23)(24)(25)(26)(27). This observation has led to the conclusion that the oscillations are the driving force for secretion in this cell (24,27). In addition, Ca 2ϩ signaling is assumed to be involved in neurotransmitter-controlled biosynthesis of proopiomelanocortin, the precursor of ␣-melanophore-stimulating hormone (23, 28 -30). The Ca 2ϩ oscillations depend on the activity of -conotoxin GVIA-sensitive Ca 2ϩ channels in the plasma membrane (25,26). Spatio-temporal studies using confocal laser-scanning microscopy have shown that each oscillation starts at the plasma membrane and is subsequently propagated as a wave to the nucleus (30,31). The high temporal resolution of the linescanning mode of the confocal laser-scanning microscopy has revealed that the rise phase of each oscillation is built up by a number of discrete increases referred to as Ca 2ϩ steps (30,31). It has been suggested that the steps are building blocks for Ca 2ϩ signaling in the Xenopus melanotrope cell (31). So far, no detailed information is available on how the steps contribute to the generation of distinct Ca 2ϩ patterns. Xenopus melanotrope cells have also been shown to display bursting electrical activity (32,33). This raises the possibility that the action potentials are the driving force for local Ca 2ϩ influxes that give rise to the stepwise build up of Ca 2ϩ to form distinct Ca 2ϩ patterns. To test this hypothesis we have performed simultaneous measurements of electrical plasma membrane activity (cell-attached patch clamping) and Ca 2ϩ signaling (microfluorometry). We show that the membrane action currents are Ca 2ϩ currents, that each Ca 2ϩ step is created by a single action current, and that the bursting pattern of Ca 2ϩ currents, in combination with the Ca 2ϩ removal rate, determines the shape of each oscillation.

EXPERIMENTAL PROCEDURES
Animals-Young-adult (8 months of age) male and female specimens of X. laevis, raised in our department under standard laboratory conditions, were adapted to a dark background for at least three weeks before the experiments, under continuous illumination, at 22°C. The animals were fed weekly with beef heart. All experiments have been carried out under the guidelines of Dutch laws concerning animal welfare.
Cell Culture-Animals were anesthetized in a solution containing 0.1% (w/v) MS222 (3-aminobenzoic acid ethyl ester; Sigma). To remove blood cells, the animals were perfused with Xenopus Ringer's solution containing 112 mM NaCl, 2 mM KCl, 2 mM CaCl 2 , 15 mM Ultral-HEPES (Calbiochem), 10 mM glucose, and 0.025% (w/v) MS222 (pH 7.4). After decapitation, neurointermediate lobes of the pituitary gland were rapidly dissected and rinsed four times in XL L15 culture medium consisting of 76% (v/v) L15 medium (Life Technologies, Inc.), 1% (v/v) kanamycin solution (Life Technologies, Inc.), 1% (v/v) antibiotic/antimyotic solution (Life Technologies, Inc.), 2 mM CaCl 2 , and 10 mM glucose (pH 7.4). After an incubation period of 45 min in Xenopus Ringer's solution without CaCl 2 and with 0.25% (w/v) trypsin (Life Technologies, Inc.), the lobes were dissociated by gentle trituration with a siliconized Pasteur's pipette. Then, the resulting cell suspension was filtered, followed by centrifugation for 10 min at 500 rpm. The pellet was resuspended in XL L15 culture medium (80 l/lobe-equivalent), and the cells were plated on coverslips coated with poly(L-lysine) (Sigma; M r Ͼ 300 kDa) in aliquots equivalent to 1 lobe/coverslip, yielding approximately 10,000 cells/coverslip. The cells were allowed to settle for 1-1.5 h in an incubator, at 22°C. Then 2 ml of XL L15 medium containing 10% fetal calf serum was added to each coverslip, and the cells were incubated for another 3 days at 22°C before use.

Measurements of Action Currents and Changes in the [Ca 2ϩ
] i -A microscopic setup for combined, time-coordinated microfluorometric and electrophysiological experiments was used to simultaneously measure changes in the [Ca 2ϩ ] i and action currents (current waveforms that represent action potentials).
To measure changes in the [Ca 2ϩ ] i , cells were loaded with 4 M fura-2/AM (Molecular Probes, Leiden, NL) in Xenopus Ringer's solution containing 1 M Pluronic F127 (Molecular Probes) (34) for 30 min at 22°C. After loading, cells were washed with Xenopus Ringer's solution to remove nonhydrolyzed fura-2/AM. Thereafter, cells were placed under continuous superfusion with Xenopus Ringer's solution (1 ml/min) on the stage of an upright microscope (Zeiss Axioskop FS, Göttingen, Germany). Unattached cells were removed, and attached cells were allowed to equilibrate for 30 min before the start of an experiment. During an experiment, cells were alternately exposed to excitation light from a multiwavelength illumination system (T.I.L.L. photonics, polychrome II, Planegg, Germany) at wavelengths of 355 and 380 nm. The fluorescence emission at 510 nm was measured with a photomultiplier tube (Hamamatsu Photonics K.K., R928, Japan) at a sampling rate of 50 Hz. The ratio of the emission intensities (355 nm/380 nm) was used as a measure for changes in the [Ca 2ϩ ] i .
Electrophysiological recordings were performed using the cell-attached recording configuration of the patch-clamp technique (35). In this cell-attached configuration, biphasic wave forms, the action currents reflecting action potentials, were recorded without disturbing the intracellular environment. All recordings were made with a pipette potential of 0 mV using an EPC-9 patch-clamp amplifier (HEKA, Lambrecht/Pfalz, Germany). Data were filtered with the built-in 4-pole Bessel filter of the EPC-9 at 3 kHz. Synchronized acquisition of both microfluorometric and electrophysiological data was performed with an Apple Macintosh PowerPC 8200/120 with Pulse/Pulsefit software (version 8.07; HEKA). Patch electrodes with a resistance of 4 -6 megaohms were pulled from borosilicate glass capillaries (GC150 -15; Clark Electromedical Instruments, Pangbourne, UK) using a Narishige PP-83 pipette puller (Narishige Scientific Instrument Laboratories, Tokyo, Japan). They were filled with Xenopus Ringer's solution.
To study the calcium dynamics during loading, cells (n ϭ 4) were loaded with fura-2 via a patch pipette in the whole-cell voltage-clamp configuration as described by Helmchen et al. (36). The pipette solution contained 100 M fura-2, 112 mM CsCl, 1.8 mM MgCl 2 , 0.2 mM MgATP, 10 mM Ultral-HEPES (pH 7.4, adjusted with CsOH). The extracellular solution consisted of 93 mM tetraethylammonium chloride, 2 mM CaCl 2 , 5 mM CsCl, 15 mM Ultral-HEPES, 2 mM MgCl 2 , 10 mM glucose (pH 7.4, adjusted with tetraethylammonium hydroxide). Background fluorescence intensity was determined when the pipette was still in the cellattached configuration. After break-in, Ca 2ϩ influx was triggered every 15 s by a depolarizing pulse from Ϫ80 to 0 mV with a duration of 100 ms. At the same time fura-2 fluorescence emission intensities were measured at 355 and 380 nm excitation as described above. The ratio of the emission intensities (355 nm/380 nm) was used to determine the decay time constants () and the relative amplitude (A r ) 1 of the evoked Ca 2ϩ transients at different fura-2 concentrations.
Chemicals-Drug-containing solutions were applied by local perfusion from a wide-mouthed glass pipette (inner diameter 0.8 mm) placed about 100 m from the recorded cell. The level of the bath solution was kept constant by means of a suction device. In Na ϩ -free conditions, NaCl was replaced by an equiosmotic amount of N-methylglucamine. To block Na ϩ channels, 1 M tetrodotoxin (TTX) was used, whereas Ca 2ϩ channels were blocked with 2 mM CoCl 2 . To keep the osmolarity constant, the concentration of NaCl was adjusted when high K ϩ concentrations (20 mM) were used. All chemicals were from Sigma, unless stated otherwise.

RESULTS
Calcium Dynamics and Fura-2-loading-Because it is known that the concentration of exogeneous Ca 2ϩ buffers like fura-2 can alter the amplitude and kinetics of Ca 2ϩ transients (36,37) we checked whether this also holds for Xenopus melanotropes. Cells (n ϭ 4) were loaded with fura-2 via patch pipettes in the whole-cell voltage-clamp configuration. After break-in, Ca 2ϩ influx was triggered every 15 s by depolarizing pulses from Ϫ80 to 0 mV with a duration of 100 ms (Fig. 1A, circles). The Ca 2ϩ -insensitive fluorescence (F 355 ) was used to monitor the diffusion of fura-2 into the cell (Fig. 1A). We assumed that the concentration of fura-2 in the cell was equal to the fura-2 pipette concentration when F 355 reached a plateau. As F 355 followed an exponential time course, the fura-2 concentration during loading could be calculated at any given time. The ratio of the emission intensities (355 nm/380 nm; eg. Spontaneous Ca 2ϩ Patterns and Electrical Activity-About 80% of the single melanotrope cells derived from the pituitary gland of X. laevis appear to display spontaneous Ca 2ϩ oscillations (26). In the present study, a total of 42 oscillating cells were studied to investigate the detailed nature and origin of the Ca 2ϩ patterns in individual cells loaded with the Ca 2ϩ indicator fura-2/AM. When recording the [Ca 2ϩ ] i at a sampling rate of 6 s, smooth oscillations with a fixed frequency and amplitude were observed (e.g. Fig. 2A). With a much higher temporal resolution of 20 ms, cells showed highly complex Ca 2ϩ oscillation patterns with strong inter-and intracellular differences (Fig. 2, B-D). In most cells (37 of 42), oscillations did not appear smooth but showed stepwise increases, the Ca 2ϩ steps (Fig. 2, B and C). In only a few cases (5 of 42), the oscillations reached the peak amplitude after one discrete rise in the [Ca 2ϩ ] i (Fig. 2D). Oscillations of different cells not only varied in frequency and relative amplitude but also in the number of steps building up an oscillation, which ranged from 1 ( Fig. 2D) to 17 (Fig. 8A). Within a given cell, the number of Ca 2ϩ steps building up a Ca 2ϩ oscillation can also vary (Fig. 2E). Fig. 2E shows that the amplitude of the oscillatory pattern may not necessarily be determined by the number of steps building up an oscillation. For example, the first and second oscillation shown in Fig. 2E have the same relative amplitude (Ar ϭ 1.39), whereas in the first oscillation two more steps are required to reach this amplitude. The amplitude of the oscillation displayed in Fig. 2D is even bigger (Ar ϭ 1.69) than that shown in Fig. 2E, whereas only one discrete rise in the [Ca 2ϩ ] i can be observed. Moreover, steps are not only present during the rising phase of an oscillation but also on top of an oscillation (Fig. 2, B and C). The number of steps on top of an oscillation is variable (e.g. Fig. 2B; first versus second oscillation), and this parameter determines the duration of an oscillation within a given cell. In addition, the average step interval between cells can vary. For example, the step interval calculated from the first 15 steps in Fig. 2B is 1.88 Ϯ 0.24 s (mean ϮS.E.). This is significantly higher (Student's t test; p Ͻ 0.0001) than the step interval calculated from the first 15 steps in Fig. 2C, which is 0.56 Ϯ 0.09 s (mean Ϯ S.E.).
To test whether fura-2 loading influences the electrical membrane activity, the firing pattern was checked in unloaded melanotropes. Using the cell-attached patch configuration, clear bursts of spontaneous "action currents," representing action potentials, were observed in unloaded, single melano- , and variation in the step interval (compare traces B and C). E, the relative amplitude (Ar ϭ R top /R basal in arbitrary units) of the oscillation is not necessarily determined by the number of steps. In the first oscillation, five steps are needed, whereas in the second oscillation, only three steps are sufficient to reach the same relative amplitude. F, bursting electrical activity of an unloaded melanotrope measured using the cell-attached patch-clamp configuration. tropes ( Fig. 2F; n ϭ 6). These bursts were similar to the bursts observed during combined measurements with fura-2-loaded cells (see Fig. 3A), indicating that loading the cells with fura-2 does not alter the firing behavior.
Relation between Ca 2ϩ Patterns and Action Potential Firing-To study the relation between the electrical activity and Ca 2ϩ oscillations of the same cell, electrophysiological measurements were combined with simultaneous measurements of changes in the [Ca 2ϩ ] i ( Fig. 3A; n ϭ 42). To check for possible pipette-induced changes in the original Ca 2ϩ signal, each combined experiment was preceded by a Ca 2ϩ measurement alone. Fig. 3A shows that the bursts of electrical activity are directly related to the Ca 2ϩ oscillations. From looking in detail (Fig.  3B), it is evident that each action current is accompanied by a discrete rise in the [Ca 2ϩ ] i . After a burst of action currents, the [Ca 2ϩ ] i smoothly returns to the basal level.
During some measurements, action current firing changed from a bursting mode into continuous firing (Fig. 4A), resulting in the disappearance of Ca 2ϩ oscillations and a steady high [Ca 2ϩ ] i . Because this phenomenon was never observed during the Ca 2ϩ measurements preceding the combined measurements, we conclude that these particular changes in the firing pattern could have been induced by the pipette. Therefore, such recordings were discarded. Nonetheless, it is interesting to note that even at this high Ca 2ϩ level, the tight relationship between action current firing and the occurrence of Ca 2ϩ steps was maintained (Fig. 4B).

Involvement of Na ϩ and Ca 2ϩ in
Step Generation-To determine the nature of the inward currents, the Na ϩ channel blocker TTX and the inorganic Ca 2ϩ channel blocker Co 2ϩ were added. 1 M TTX did not have an effect on either action currents or Ca 2ϩ oscillations in any of the cells measured ( Fig. 5A; n ϭ 11). On the other hand, applying 2 mM CoCl 2 clearly abolished both action currents and Ca 2ϩ oscillations in every cell studied (Fig. 5B; n ϭ 15). In the complete absence of extracellular Na ϩ (Na ϩ replaced by N-methyl-D-glucamine), no action currents or Ca 2ϩ oscillations were observed ( Fig. 6A; n ϭ  4). Under this Na ϩ -free condition, action currents and a rise in the [Ca 2ϩ ] i could still be induced by a depolarizing K ϩ (20 mM) pulse ( Fig. 6A; n ϭ 4). The rise in the [Ca 2ϩ ] i was clearly built up by a number of Ca 2ϩ steps, each accompanied by an action current (Fig. 6B).
Relation between the Shape of an Action Current and the Amplitude of a Ca 2ϩ Step-Because action current measurements were performed simultaneously with Ca 2ϩ measurements, a link between the shape of an action current and the amplitude of a Ca 2ϩ step could be demonstrated. The upper trace of Fig. 7A shows the action currents reflecting the Ca 2ϩ steps shown in the lower trace of Fig. 7A. Whereas the amplitude of the successive action currents decreased, the amplitude of the accompanying Ca 2ϩ steps increased. The relative difference in the ratio values between the start and the top of a Ca 2ϩ step (Ar ϭ R top /R basal ); lower trace of Fig. 7A) was taken as the step amplitude. To determine the amount of charge entering the cell during an action current, the peak areas of the downward action currents were integrated. In Fig. 7B the peak area of each action current was plotted against the relative amplitude of the Ca 2ϩ steps in Fig. 7A. The peak areas of the three successive action currents increased, and the amplitude of the Ca 2ϩ steps increased. A clear linear relation was found between the current peak area and the Ca 2ϩ step size.
Kinetics of Ca 2ϩ Removal during a Ca 2ϩ Oscillation-The calcium removal kinetics have been studied to investigate the role of this removal in shaping the calcium oscillations. The presence of a discontinuous (i.e. stepping) Ca 2ϩ oscillation presents a unique opportunity to analyze the speed of removal of Ca 2ϩ from the cytoplasm (V d ) during the rising phase of a Ca 2ϩ oscillation. The speed of Ca 2ϩ removal was analyzed both during the rising phase following each Ca 2ϩ step and during the declining phase, bringing Ca 2ϩ back to basal level. Fig. 8A shows an example of a Ca 2ϩ oscillation built up by steps. In total, 27 steps can be distinguished distributed over three individual oscillations (numbered I, II, and III).
Curve fitting was applied to model the kinetics during different phases of the Ca 2ϩ oscillation. An exponential function can be used to adequately describe first order processes. By using a standard nonlinear regression algorithm (38), the de-cline after the top of an oscillation was modeled with a first order exponential function, where R x is the fura-2 emission ratio at time x, and A, B, x 0 , and are constants obtained from the fit. For peak I, the parameters in this equation were estimated from a fit of 628 data points: A ϭ 1.04 Ϯ 0.000166, B ϭ 0.737 Ϯ 0.00151, x 0 ϭ 25.08, and ϭ 4.75 Ϯ 0.0312. Given the quality of the fit ( 2 ϭ 0.00008, S.D. Ͻ 10%) we considered the exponential model to be valid (38). However, between steps, each decline consists of a limited amount of data points (varying between 24 and 104 for Fig. 8A), and a reliable exponential fit could not be obtained. Because over a small interval an exponential function can be approximated by a straight line, we used a linear approximation for quantifying the calcium re- moval speed after each step during the rising phase of the oscillation.
The data of Fig. 8A (60 s) was analyzed by introducing the following parameters (see Fig. 8B): R b,1 , the resting fura-2 emission ratio; R t,n the fura-2 emission ratio at the top of step n; R b,n , the fura-2 emission ratio just before occurrence of step n, and dx n , the time needed for the fura-2 emission ratio to drop from R t,n to R b,nϩ1 .
By using a linear least squares algorithm, a line was fitted for each interval (R t,n ,R b,nϩ1 ), solving C and D in where R x is the fura-2 emission ratio at time x, and C and D are constants. D can be taken as a measure for V d,linear , the removal speed of Ca 2ϩ from the cytoplasm determined by the linear fit. V d,linear for the n th step is defined as dR n /dx n (with dR n being (R t,n Ϫ R b,nϩ1 )). Only fits with an r value of at least 0.90 were analyzed. In Fig. 8A, one step (marked with a star in Fig. 8A) was omitted from the analysis, because this criterion was not met.
By plotting V d,linear as a function of the step number, the dynamic properties of the Ca 2ϩ -regulating mechanisms in the cell could be visualized (Fig. 8C). To reduce noise, a moving average was calculated from the data (window size of 3) to produce the line shown in Fig. 8C. For the analyzed trace, V d,linear was dynamically up-and down-regulated during the different phases of the Ca 2ϩ oscillation, such that steps near the top of the oscillation were associated with a higher V d,linear .
To determine whether V d is a function of R t,n (i.e. if the removal speed of Ca 2ϩ is a function of [Ca 2ϩ ] i ), the decline speed V d,linear was plotted against R t,n (Fig. 8D). A clear linear correlation between R t,n and V d,linear was found (A ϭ Ϫ0.23 Ϯ 0.020, B ϭ 0.22 Ϯ 0.013, r value ϭ 0.92), reflecting a linear increase of V d,linear with R t,n . A saturation of V d,linear was not observed during analysis (Fig. 8D, no deviation from the linear model at the highest R t,n values).
To compare Ca 2ϩ removal kinetics during the rising phase (after each step) with the Ca 2ϩ removal kinetics during the declining phase of the oscillation, V d,linear was compared with V d,exp , the velocity of Ca 2ϩ removal given by a first order exponential function (Eq. 1). By substituting R t,n and R b,nϩ1 (which were directly estimated from the original recording) in Eq. 1, x t,n and x b,nϩ1 can be calculated by writing In this way, substituting R t,n gives x t,n , and R b,nϩ1 gives x b,nϩ1 . By calculating (R t,n Ϫ R b,nϩ1 )/(x b,nϩ1 Ϫ x t,n ), V d,exp can be obtained for each n. Fig. 8E clearly shows that V d,linear and V d , exp are linearly correlated (A ϭ 0.0016 Ϯ 0.00064, B ϭ 0.87 Ϯ 0.054, r value ϭ 0.92). The low value of A and a value of B near unity show that the linear correlation is almost ideal (for a 45°line, consistent with a perfect linear correlation, A ϭ 0 and B ϭ tan 45°ϭ 1). In general, the correlation between V d and R t,n was not affected at all by occurrence of steps during the rising phase of the oscillation.

DISCUSSION
Calcium Dynamics and Fura-2 Loading-Because it is known that the concentration of exogeneous Ca 2ϩ buffers like fura-2 can alter the amplitude and kinetics of Ca 2ϩ transients in presynaptic terminals of neurons (36,37), we checked whether this holds also for the neuroendocrine melanotrope cells of Xenopus. The values of the decays of the Ca 2ϩ transients are independent of the [fura-2] and are in the range of 2-5 s. Only a very small decrease in relative amplitude (Ar) was found during loading. However, absolutely no relationship between and Ar is present. Therefore, we conclude that the intrinsic Ca 2ϩ removal in Xenopus melanotropes is rather slow and not altered by the loading with fura-2.
Ca 2ϩ Patterns-Ca 2ϩ oscillations in Xenopus melanotropes have a frequency of about 1-3/min, and within each cell, have a rather stable amplitude (24 -27). They have a smooth appearance when observed during Ca 2ϩ measurements using a sampling interval of 6 s (see Fig. 1A). In the present study, these oscillations have been studied in detail, with particular attention to their shape and the mechanisms that generate this shape. By performing continuous measurements with a high temporal resolution (20 ms) clear inter-and intracellular variations in Ca 2ϩ patterns were observed. The Ca 2ϩ oscillations did not appear smooth anymore but were built up by discrete changes in the [Ca 2ϩ ] i , the Ca 2ϩ steps. These findings extend recent confocal line scanning experiments (30,31), which suggested that the rising phase of oscillations are built up by three to four of these discrete steps. The present study, where the microfluorometric method allows continuous Ca 2ϩ measurements with a high temporal resolution, shows that the number of steps during an oscillation is variable and can be as high as 17. The oscillations had discrete rises not only during the rising phase of an oscillation but often also on top of the oscillation. This has not been reported for melanotropes or other secretory cells before. FIG. 7. Relation between the shape of action currents and the amplitude of Ca 2؉ steps. Combined measurements of action currents (upper trace) and Ca 2ϩ steps (lower trace) carried out as in Fig. 3. A, the peak areas of the action currents were determined by integration, and the relative amplitude (Ar in arbitrary units (AU)) of the Ca 2ϩ steps was calculated by dividing R top /R basal . B, graph showing the positive correlation between the action current peak areas and the amplitude of the Ca 2ϩ steps.

The Role of Action Potentials in the Generation of Specific
Ca 2ϩ Oscillation Patterns-To study the ability of a cell to transfer information in the form of complex [Ca 2ϩ ] i patterns, it is important to understand how the Ca 2ϩ patterns are generated. From fast line scanning experiments, it is known that the Ca 2ϩ oscillations in Xenopus melanotropes can travel through the cell as multiple waves from the plasma membrane into the nucleus (30,31). The Ca 2ϩ oscillations are inhibited by the N-type Ca 2ϩ channel blocker -conotoxin GVIA (26) and not by the L-type Ca 2ϩ channel blocker nifedipine (26). This indicates that they originate at the membrane by Ca 2ϩ influx through N-type Ca 2ϩ channels. It has been hypothesized that action potentials cause the opening of the voltage-operated N-type Ca 2ϩ channels (32,33,39,40). However, a direct relation between action potentials and Ca 2ϩ oscillations has never been shown in pituitary melanotropes. Until now, the action potentials in Xenopus melanotropes observed have been induced by depolarizing pulses using the whole-cell patch-clamp technique (32,33), or they have been measured using the perforatedpatch configuration (33). In the present study, the cell-attached configuration of the patch-clamp technique was used to prevent any disturbances of the intracellular environment. In this Step marked by an asterisk (*) was omitted from analysis (bad fit). The box indicates the part that is enlarged to form panel B. B, parameters used to describe the step kinetics. R b,1 , the resting fura-2 emission ratio; R t,n , the fura-2 emission ratio at the top of step n; R b,n , the fura-2 emission ratio just before occurrence of step n; and dx n , the time needed for the fura-2 emission ratio to drop from R t,n to R b,nϩ1 . An asterisk (*) marks the time needed for the fura-2 emission ratio to rise from R b,n to R t,n . C, V d,linear , the velocity of Ca 2ϩ removal after a Ca 2ϩ step determined by a linear fit (for details, see "Results"), plotted against the step number. D, a linear correlation was found between V d,linear and R t,n , the fura-2 emission ratio at the top of step n. E, a clear linear correlation was found between V d,linear and V d,exp . V d,exp is the velocity of Ca 2ϩ removal during the declining phase of oscillations determined using a first order exponential function (see Eq. 1 under "Results"). patch configuration, bursts of action currents representing bursts of action potentials were observed. The bursting behavior was not induced by fura-2 loading as both loaded and unloaded cells displayed this firing behavior. The combined patch-clamp and Ca 2ϩ microfluorometry showed that each action current is generating a discrete, stepwise increase in the [Ca 2ϩ ] i . In this way, a burst of action currents builds up a full-sized Ca 2ϩ oscillation. Also the Ca 2ϩ steps on top of an oscillation appeared to be generated by action currents. After each burst of action currents, the [Ca 2ϩ ] i smoothly returned to the basal level. In some combined experiments, the Ca 2ϩ signal went up and stayed high. At the same time, action current firing changed from a bursting pattern into continuous firing, as observed previously in Xenopus melanotropes (32,33). These changes may have been induced by the patch pipette because such sustained Ca 2ϩ elevations were never observed in the control Ca 2ϩ measurements preceding the combined experiments.
A relation between action potentials and Ca 2ϩ transients has been established in other pituitary cells (41)(42)(43)(44), but in none of these studies have such complex Ca 2ϩ patterns been seen to be generated spontaneously as in the Xenopus melanotropes. In the former studies, either single, small amplitude Ca 2ϩ transients were generated by single action potentials firing at a low frequency (41)(42)(43), or large transients lacking discrete rises (44) were generated by bursts of action potentials. In rat corticotrophs stimulated by vasopressin, a huge transient rise in the [Ca 2ϩ ] i is coupled to action potential firing (45).
We have found a relationship between the peak area of an action current and the size of a Ca 2ϩ step. This indicates that Ca 2ϩ influx represents most of the inward current. Furthermore, this observation confirms the idea that the melanotrope cell generates its Ca 2ϩ pattern by changing the amount of Ca 2ϩ influx and not by Ca 2ϩ release from intracellular stores. Changing the duration of Ca 2ϩ influx i.e. by tetraethylammonium, a K ϩ channel blocker known to increase action potential duration (46), drastically changes the Ca 2ϩ oscillations in Xenopus melanotropes. 2 The amplitude becomes higher, the duration becomes longer, and the frequency becomes lower. The contribution of Ca 2ϩ -induced Ca 2ϩ release to our Ca 2ϩ signal seems to be minor, although we can not rule out some involvement. If Ca 2ϩ -induced Ca 2ϩ release would have a big influence on the shape of Ca 2ϩ transients, of the Ca 2ϩ transients should increase at higher [Ca 2ϩ ] i levels; a higher [Ca 2ϩ ] i would induce further Ca 2ϩ release and thereby sustain the decay phase (47). However, we found a constant in each phase of a Ca 2ϩ oscillation. Furthermore, evidence for the involvement of thapsigargin-sensitive stores or ryanodine-sensitive stores in the generation of Ca 2ϩ oscillations is lacking because thapsigargin and ryanodine do not inhibit the Ca 2ϩ oscillations (26). Also treatments with the Ca 2ϩ ionophores ionomycine and Br-A23187, which empty intracellular Ca 2ϩ stores, do not block the Ca 2ϩ oscillations. 3 Currents Underlying Action Potentials-Mathematical models containing Na ϩ channels (48,49) predict that changes in the Na ϩ current lead to other bursting patterns. However, TTX, which blocks most of the Na ϩ currents in Xenopus melanotropes (33), had no effect on either the electrical firing or the Ca 2ϩ oscillations. Blocking the Ca 2ϩ channels with CoCl 2 on the other hand fully blocked both phenomena. Under Na ϩ -free conditions, electrical activity and [Ca 2ϩ ] i changes were absent, probably because of a general hyperpolarization. Although TTX-insensitive Na ϩ currents are present in Xenopus melano-tropes (33), we showed that even in the complete absence of Na ϩ , a K ϩ -evoked depolarization induced Ca 2ϩ action currents and stepwise increases in the [Ca 2ϩ ] i , which looked similar to those seen under normal conditions. This observation indicates that the machinery for generating the action currents and steps is still fully functional. Although these findings suggest a role for Na ϩ only in setting the excitability of the plasma membrane, with Ca 2ϩ determining the bursting behavior, we cannot rule out a contribution by TTX-insensitive Na ϩ channels.
The Role of Ca 2ϩ Removal in Generating Specific Ca 2ϩ Patterns-In this study we show for the first time that during the rising phase of a Ca 2ϩ oscillation, the speed of Ca 2ϩ removal is not constant but depends on the [Ca 2ϩ ] i . With increasing [Ca 2ϩ ] i the removal speed linearly increases. This up-regulation plays a role in shaping the Ca 2ϩ oscillation. During the rising phase of an oscillation, each Ca 2ϩ influx episode leaves the [Ca 2ϩ ] i at a higher level than the preceding episode. At the top of an oscillation the influx and removal of Ca 2ϩ become balanced because of the up-regulation of the removal speed, and consequently influx episodes no longer give a net increase in the [Ca 2ϩ ] i . The relatively slow removal at a low [Ca 2ϩ ] i leads to the stepwise increase. Interestingly, this process fits with mathematical models (48) that predict that, although the electrical bursting patterns look the same, the shapes of Ca 2ϩ oscillations could be very different depending on how fast Ca 2ϩ is removed from the cytoplasm. If the Ca 2ϩ removal rate is low, the [Ca 2ϩ ] i increases, reaching a maximum near the end of a burst of action potentials. If the Ca 2ϩ removal is very fast, each action potential gives rise to a single Ca 2ϩ transient. In these models fixed values for the speed of Ca 2ϩ removal were used. However, we now have demonstrated that the Ca 2ϩ removal rate is not fixed during an oscillation.
When a burst of action currents is terminated, the [Ca 2ϩ ] i smoothly declines in a single exponential way, indicating the presence of one Ca 2ϩ removal process. The mechanism of termination of a burst may involve Ca 2ϩ -dependent K ϩ channels (50,51) or voltage-or Ca 2ϩ -dependent inactivation of Ca 2ϩ channels (50,52,53). The single exponential function describing the decay after the top of an oscillation exactly fits the decline phases of the Ca 2ϩ steps during the rising phase of the oscillation. This indicates that the same Ca 2ϩ removal process is active during the rising phase and the decay phase of an oscillation.
Conclusion-This study shows that bursting firing in combination with the calcium removal rate determine the shape of a Ca 2ϩ oscillation. In this way individual Xenopus melanotrope cells can generate different Ca 2ϩ oscillatory patterns. We propose that not only the frequency and amplitude of the Ca 2ϩ oscillations serve to encode cellular regulatory information but also other oscillation pattern variables such as number of Ca 2ϩ steps, the step amplitude, the step frequency, and the speed of Ca 2ϩ removal. Therefore, the information storage capacity of the Ca 2ϩ signaling seems much more complex than previously thought. Currently, studies are being carried out on the modulation of the Ca 2ϩ patterns in Xenopus melanotropes by various regulatory neurotransmitters and under different conditions of adaptation of the animal to background light intensity.