Neuron,

Vol. 8, llG9-1125,

June,

1992, Copyright

0 1992 by Cell Press

PhaseDependent Contributions from Ca*+ Entry and Ca*+ Release to Caffeine-Induced [Ca*‘]i Oscillations in Bullfrog Sympathetic Neurons D. D. Friel and R. W. Tsien Department of Molecular and Stanford University School of Stanford, California 94305-5426

Cellular Medicine

Physiology

Summary Sympathetic neurons display robust [Caz+J oscillations in response to caffeine and mild depolarization. Oscillations occur at constant membrane potential, ruling out voltage-dependent changes in plasma membrane conductance. They are terminated by ryanodine, implicating Ca*+-induced CaZ+ release. Caz+ entry is necessary for sustained oscillatory activity, but its importance varies within the oscillatory cycle: the slow interspike rise in [Ca*‘]i requires CaZ+ entry, but the rapid upstroke does not, indicating that it reflects internal Ca2+ release. Sudden alterations in [Ca2+],,, [K+],, or [caffeine], produce immediate changes in d[Ca*+]i/dt and provide information about the relative rates of surface membrane Ca*+ transport as well as uptake and release by internal stores. Based on our results, [Ca*‘]i oscillations can be explained in terms of coordinated changes in Ca*+ fluxes across surface and store membranes. Introduction Oscillations in the intracellular free Ca2+ concentration ([Ca2’],) have been described in a variety of excitable and nonexcitable cells (reviewed in Rink and Hallam, 1989; Berridge, 1990; Jacob, 1990a; Tsien and Tsien, 1990; Meyer and Stryer, 1991). It appears that multiple mechanisms are at work. One generic category of [Ca2+], oscillations, found in many excitable cells, arises from periodic changes in the membrane potential and Ca*+ permeability of the plasma membrane (e.g., Gorman and Thomas, 1978; Schlegel et al., 1987; Hudspeth and Lewis, 1988). Another kind of [Ca*+], oscillation, prominent in nonexcitable cells, involves a dynamic Ca*+ exchange between the cytosol and internal stores (Woods et al., 1986; Harootunian et al., 1988). Caffeine-induced [Ca*+],oscillations in bullfrog sympathetic neurons, first inferred from observations of periodic membrane hyperpolarizations by Kuba and Nishi (1976), provide a good example of neuronal [Ca”]i oscillations. Their mechanism is of interest, since they occur in an excitable cell but are promoted by interventions that mobilize internal Ca*+. These oscillations have been the subject of several studies (e.g., Kuba and Nishi, 1976; Lipscombe et al., 1988), but a number of key questions have remained unanswered. Do the oscillations depend on periodic changes in membrane potential? Is Ca2+ entry necessary for each oscillatory cycle, or does it merely serve to replenish internal Ca*+ stores over the course of

several cycles? What is the mechanism by which caffeine promotes the oscillatory activity? Answers to these questions not only would further our understanding of neuronal [Ca”], regulation, but would also help clarify a recurring issue in signal transduction: how can a maintained stimulus produce a periodic response? We have addressed these questions by measuring [Ca2+], in isolated, fura-2-loaded sympathetic neurons under conditions where oscillations are particularly stable. Fast solution changes were used to alter selectively Ca2+ entry or Ca*+ release at different points in the oscillatory cycle, making it possible to assess the moment-to-moment contributions of different Ca*+ delivery systems to the observed changes in [Ca*+],. The ultimate goal was to understand [Ca*‘], oscillations and their underlying Ca2+ fluxes in much the same way as repetitive electrical activity is understood in terms of underlying ionic currents. It was found that [Ca2+], oscillations can occur without variation in membrane potential but do require Ca*+-induced Ca2+ release, suggesting that the oscillations arise from dynamic Ca2+ exchange between the cytosol and an internal store. Within each oscillatory cycle, Ca2+ is first taken up from and then released into the cytosol by an internal store; this occurs in synchrony with a net surface membranefluxwhosedirection is first inward and then outward. Net Ca*+ uptake and release bythe internal store dominate changes in [Ca*+], throughout the oscillatory cycle, but changes in the rate of Ca2+ entry also occur in phase with changes in [Ca2+],. Results Bullfrog sympathetic neurons responded to either high K+ or caffeine with an elevation in [Ca*+],. When cellswereexposedtocaffeineand high K+incombination (Figures la and lb), [Ca*+], oscillations were consistently observed (21/21 cells exposed to 26 or 30 mM K’ and 5 or 10 mM caffeine). In contrast, oscillations were never seen in the presence of high K+alone and were seen only rarely with caffeine alone (4/79 cells exposed to 10 mM caffeine). The oscillations arose regardless of which stimulus was applied first (compare Figures la and lb) and showed no systematic dependence on the order in which the stimuli were applied (the differences between oscillations in the two traces probably reflect cell-to-cell variability). Oscillations emerged in a stereotypic fashion: [Ca*‘], increased transiently in response to the second stimulus, approached a plateau, and gradually became unstable over a period of several minutes, with excursions above and below the plateau level increasing in size until they became periodic (~1 cycle per min). Several key properties of Ca*+ delivery systems in these neurons are illustrated by the initial [Ca*‘l, responses elicited by high K+ and caffeine (Figures la

NeUKXl 1110

Figure 1. [Caz’], Oscillations ence of High K+ and Caffeine

D

5 30

b

mM

mM c&felle

K

s dJ caffe,ne 30

c

High

d

Coff

m!J K

K

+++ 400

100

nhi

and lb, respectively). These are shown along with the time derivative of [Caz’], (d[Ca*+],/dt) on an expanded time scale (Figures Ic and Id, left) and in phase trajectory plots (d[Ca*‘]i(t)/dt against [Ca*‘],; Figures lc and Id, right). d[Ca*+],(t)/dt is proportional to the total net Ca*+ flux (Jtotar) entering the cytosol at each time (see Experimental Procedures). inward flux (i.e., info the cytosol) is plotted downward to conform with electrophysiological convention. High K’ causes membrane depolarization, voltagedependent Ca*+ entry, and a rise in [Ca*‘], to a new steady-state level in association with a transient inward Ca*+ flux (Figure Ic, left). Caffeine also elicits an inward Ca2+ flux (Figure Id, left), but in this case the source of Ca*+ is intracellular (Lipscombe et al., 1988; Friel and Tsien, 1992; see also Thayer et al., 1988; Kostyuk et al., 1989, 1991; Hernandez-Cruz et al., 1990; Mironov and Usachev, 1991). Based on studies of Ca*+ release channels from muscle (Rousseau et al., 1988)

in the

Pres-

(a) Changes in [Ca2+], induced by high K+(30 mM)and bycaffeine(5mM)in the presence of high K’, which induced [Ca2’], oscillations. (b) Effects of stimulating another cell in the reverse order. In each case, oscillations were associated with periodic changes in both Fs5,,and Fnso. Dotted lines indicate zero [Ca*+], (interval between dots = 9.6 s). (c-e, left) Responses to high K+ (c), caffeine (d), and high K+ in the presence of caffeine (e) from (a) and (b) on an expanded time scale; arrows mark the beginning of each maintained stimulus. Beneath each [Caz’], record is a plot of d[Ca2+],ldt, which gives a measureofthenetCa2+fluxintothecytosol at each time. (c-e, right) Illustration of how d[Ca2+],/dt varies with [Ca*+],, giving the phase trajectory for each response. The high K’-induced inward Ca*+ flux (c) represents voltagedependent Ca*+ entry from the external medium; the caffeine-induced flux (d) reflects releaseof internal Ca*‘. After[Ca2+], returned to the resting level, exposure to high K+ in the maintained presence of caffeine (e) produced two components of inward CaZ’ flux, one resemb!ing high K+-induced entry of external Ca2+ (c), the other resembling caffeine-induced release of internal Ca*+ (d). Squares (e, right) represent the initial portion of the caffeine response with ordinate scaled by0.5; the second flux component may be smaller than the caffeine-induced flux because the store is partially depleted after the initial caffeine response (see b). d[Ca2+],(t)/dt was calculated as {[Ca2+],(t + At/Z) - [Caz’],(t - At/2)}/At, with At = 585 ms (3 sample intervals in this case). This calculation smooths changes in d[Ca”+],/dt that occur rapidly compared with At, e.g., those immediately following application of high K’(C). Dashed lines indicate zero d[Ca*+],/dt. Cell B12X (a and c); cell Bl3F (b, d, and e).

and brain (McPherson et al., 1991), it appears that caffeine releases internal Ca2+ by increasing the [Ca*+], sensitivity of Ca*+-induced Ca*’ release (CICR), increasing the potency with which cytoplasmic Cai’ promotes channel opening and Ca*+ reiease from an internal store. Characteristically, the phase trajectory of the caffeine-induced rise in [Ca2+], shows a rapid acceleration of inward flux with [Ca*‘], (Figure Id, right), possibly reflecting the [Ca*‘], dependence of CICR in the presence of caffeine. After depolarization in the maintained presence of caffeine (Figure le, left), [Ca2+], rises in association with two distinct inward flux components. These components are resolved by plots of d[Ca”‘],/dt against time (Figure le, left, lower trace) or [Ca*&], (right, uninterrupted curve). The first component resembles the flux induced by depolarization in the absence of caffeine (Figure Ic), while the second component is similar in its [Ca*‘], dependence to the flux induced by

[Ca*+], Oscillations

in Sympathetic

Neurons

1111

Figure2. Involvement of CICR feine-Induced [Ca2+], Oscillations 1 uM Ryonodine 30 K + 5 Coff

b

CCoffl

o :

10.

0.

c

10

10

3

I

.sc

i

in

Caf-

(a] [CaZ+], oscillations are blocked by 1 uM ryanodine, after which [Ca2+], stabilizes at a dashed line) that falls belevel UCa*+lrr.ryan, tween the intracycle minimum and maximum. Cell BIZJ. (b) Sudden caffeine removal at different points in the oscillatory cycle caused a rapid fall in [Ca2+],. This occurred either when caffeine was removed following a peak (b, top, 2), or just before a peak while [Ca*+], was still rising (b, bottom, 3). Note that the latency for resumption of oscillations after caffeine was restored depended on where in the cycle it was initially removed. (c] Comparison between phase trajectories from records l-3 in(b). Following each perturbation, [Ca2+], declined approximately exponentially, as revealed by a linear portion of the phase trajectory (dashed line). The slope of this line gives a time constant of -3 s (linear regression). In this cell (BO9P) [Ca*‘], oscillated in the presence of 10 mM caffeine with normal [K+], (2 mM).

(nM)

caffeine (Figure Id, replotted as the square symbols in Figure le after scaling by 0.5). The resemblance is striking, given that the immediate stimulus is high K+ in one case and caffeine in the other. Evidently, application of high K+ in the presence of caffeine produces a combination of voltage-dependent Ca2+ entry and release of internal Ca*+ by the same mechanism that is responsible for caffeine-induced Ca*+ release, namely CICR. Involvement of CICR in the [Ca”]i Oscillations Block by Ryanodine Further support for the involvement of CICR in the [Ca*‘], oscillations comes from the observation that they are inhibited by 1 PM ryanodine (Figure 2a). This drug binds with high affinity and specificity to Ca2+ release channels from brain (McPherson et al., 1991) and muscle (Imagawa et al., 1987; Lai et al., 1988). Ryanodine has been used widely in the study of CICR (Sutko et al., 1985; Marban and Wier, 1985; Meissner, 1986; Lattanzio et al., 1987; Fill and Coronado, 1988) and is thought to inhibit it by either preventing channel opening (McPherson et al., 1991) or stabilizing an open subconductance state (Imagawa at al., 1987; Rousseau et al., 1987; Bezprozvanny et al., 1991), preby the store and venting net Ca*+ accumulation thereby net Ca*+ release. The blocking effect of ryanodine was very consistent (9/9 cells at 1 PM; l/l cell at 10 PM), but somewhat variable in rate (spike amplitude declined over l-8 cycles before [Ca*+], stabilized). The observation that at least one cycle occurred before any indication of block is consistent with use

dependence of ryanodine’s inhibitory effect (Rousseau et al., 1987; Thayer et al., 1988). Since control experiments show that 1 PM ryanodine inhibits caffeine-induced release of internal Ca2+ but does not inhibit voltage-dependent Ca2+ channel current (Friel and Tsien, 1992), we attribute this action of ryanodine to inhibition of CICR. In the presence of ryanodine, [Ca*+], stabilized at a Figure 2, dashed line) that fell belevel (Ka2+15s,ryan; tween the extremes of [Ca2+], reached during the oscillatory cycle. Since [Ca*‘], did not drift under these conditions, we regard [Ca2+]ss,n/an as a steady-state level at which Ca2+ entry and extrusion across the surface membrane balance one another under the prevailing experimental conditions of [K+],, [Ca2+10, and [caff],. In the Discussion, we will consider extending this interpretation of [Ca2+]sS,ryan to the case in which [Ca2+], is oscillating. For now, [Ca2+]5r,lyan will be used simply as a reference level for describing [Ca2+]i changes during the oscillatory cycle. Effect of Sudden Caffeine Removal Caffeine is thought to release internal Ca2+ by increasing the potency with which Ca2+ opens intracellular Ca2+ release channels. If this is true, then the Ca2+ permeability of the store at a given [Ca2+], should be elevated in the presence of caffeine. If caffeine removal reduces this permeability, then it should be possible to assess the activity of CICR while [Ca2+l, is oscillating by removing caffeine: reducing the Ca*+ permeability of the store should create an imbalance between internal Ca2+ uptake and release, a sudden outward net Ca2+ flux, and a drop in d[Ca’+]i/dt.

NWKI” 1112

Figure 3. [Cal+], Oscillations Occur at Constant V, but the Importance of VoltageGated Ca2+ Entry Varies during the Oscillatory Cycle

25 mv

b

CKI,

: 30.

2.

30

c

CCOI,

CmM)

: 2.

0.

2

10 set

Sudden caffeine removal invariably led to a rapid fall in d[Ca2’],/dt, regardless of whether it occurred during the rising or falling phases of the [Ca*+], transient (Figure 2b). Thus, caffeine must influence [Caz’], regulation at these times, presumably by promoting CICR. The changes in [Ca2’], were associated with a prompt increase in net outward CaL+ flux that was several times larger than the outward flux during the declining phase of the spontaneous [CaZ”], transients (Figure 2c, compare traces l-3). The prompt change in d[Ca*‘],/dt is consistent with rapid washout of caffeine (O’Neill et al., 1990) and suggests that its stimulatory effect on the internal store is quickly reversible. After caffeine removal, [Cal’], fell with a fast exponential time constant (r = 3 s) that was independent of when in the oscillatory cycle the perturbation occurred, consistent with a first-order restorative process. The

(a) High K’ (30 mM) depolarized V, and tncreased [Ca2’],. Caffeine (1 mM) produced a small rise in [CaZ-I,, with little or no change in V,. In the presence of 1 mM caffeine, high KL depolarized V, io the same degree but elicited a larger and more transient [W’], response than that elicited by high K’ alone. Increasing [Gaff], to 5 mM produced a transient [Ca2+] elevatton followed by steady [Ca>+], oscillations, but V, remained clamped at -33 mV. Resting potential in this cell (Bl3L) was -66 mV. (band c) Separation of the interspike [Cal’], rise into two intervals with different requirements for external CaL+. Comparison between the effect of lowering [K’],!b) and removing external Cal- (c) in two different phases of the oscillatory cycle. Each column shows the effect of changing the external solution early during the interspike rise in [(Ia’-], (top), when [Ca’+], was not changing very fast, and during the rapid upstroke (center), shown on an expanded time scale at the bottom. Vertical dashed lines indicate when solution changes were made, and the zero [Ca2+], level is indicated by dotted lines. (b, top)When [K’],was lowered from 30 to 2 mM near the interspike minimum, [Ca2’], fell and oscillations stopped, demonstrating the importance of depolarization. (c, top) Removing external Ca2” had the same effect, suggesting that elevated [K’], is important because it promotes entry of extracellular Ca’+. (c, center and bottom) After removing external Ca’* when [Ca?], was increasing more rapidly, [Ca’+],continued to rise, indicatingthat the rise reflects release of internal Cal+. (0, center and bottom) After iowertng [K+], at a similar point in the cycle, [CazL], continued to rise in much the same way, Indicating that depolarizatton per se is not necessary for internal Ca’+ release. [Gaff], = 5 mM. Cell B13G (b); ceil B12Z (c).

overall interpretation is that caffeine promotes a [Ca*+],-dependent leakage of Ca2” via Ca*’ release channels in the internal store and that this leakage opposes ongoing Ca2+ uptake. When caffeine is removed, the leak is promptly abolished and the firstorder restorative process becomes unopposed, leading to net Cal-- uptake by the store. [Caz’li Oscillations Occur at Fixed Membrane Potentials Previous studies have not determined whether variations in membrane potential (V,) play a role in caffeine-induced [Ca*+], oscillations in these ceils (e.g., Kuba, 1980). To address this question, we monitored V,, with an intracellular microelectrode while [Ca2+]: oscillated after stimulation with high K’ and caffeine (Figure 3a). in the combined presence of 30 mM K’

[Ca”], 1113

Oscillations

in Sympathetic

Neurons

voltage-dependent of the plasma

changes membrane.

in the

Ca*+

permeability

Role of Ca2+ Entry from the External Medium during the Oscillatory Cycle Although [Caz’], oscillations do not require fluctuations in V,, sustained depolarization is nevertheless important for continued oscillations, since ongoing oscillatory activity is eliminated by lowering [K’],, from

c

Figure 4. Effects of Terminating the Oscillatory Cycle

Ca *+ Entry

at Different

Points

in

(a) Removing external CaZ’ before [Ca*+], reached [Ca2+],, rVan led to a drop in [Ca2+],, while the same perturbation at a slightly later time, after [Ca*+], exceeded [Ca2+]\r,rvan, failed to arrest the [Ca”], rise. (b) Results from another cell in which the transition was more gradual and occurred at higher [Ca2+],. (c)The effect of reducing [K’], from 30 to 2 mM in another cell. Note the case in which [Caz+], hovered near [Ca2+]ss,,yan for several seconds before rising to give a [Ca2+], transient. Successive perturbations were carried out after restoring [CaZ+]” or [K’], and letting oscillations resume. The dashed line indicates [Ca2+]qr.,an determined after [Ca*+], stabilized following exposure to ryanodine (1 uM). Oscillations were monitored in the presence of 5 mM caffeine and 30 mM K+. Cell 8122 (a); cell BlZX (b); cell B13G (0.

and 5 mM caffeine, stable oscillations occurred even though V, remained fixed at -33 + 1 mV. Similar results were observed in each of 2 cells, indicating that the [Caz+], oscillations do not reflect, or require,

30 to 2 mM (Figure 3b, top). Removing external Ca*+ has a similar effect (Figure 3c, top), indicating that both depolarization and external Ca2+ are required for sustained oscillations. Both of these interventions were made when [Ca2+], was near its lowest value, but sustained oscillatory activity was also suppressed when [K’10 was lowered or external Ca*+ was removed at other stages of the cycle. However, the immediate effects of these interventions depended critically on the point in the cycle when they were imposed. If external Ca2+ was removed during the rapid upstroke, [Ca”], continued to rise, much as it did in the presence of external Ca2+ (Figure 3c, center and bottom). Lowering [K’10 at a similar point in the cycle (Figure 3b, center and bottom) was similarly ineffective in arresting the [Ca2’], rise.Thus,atthisstageofthecycle, neitherdepolarization nor external Ca2+ is required for the [Ca2+]i transient that immediately follows. These results indicate that during the rapid upstroke, Ca2+ is released from an internal store and that release does not require depolarization. Moreover, they show that during the upstroke, Ca2+ release becomes sufficiently strong to raise [Ca”]i without the aid of Ca2+ entry from the external medium. Figure 4 shows the results of systematically varying the point at which voltage-dependent Ca*+ entry was terminated, either by removing external Ca2+ or by lowering [KC],. Each panel shows a representative [Ca*+], transient in the presence of 5 mM caffeine and 30 mM K+ (dotted traces) along with superimposed [Ca2+], records following individual perturbations (continuous traces). After each perturbation (arrows), [Ca2’l, either fell monotonically to a low, steady level, or increased transiently before falling to the same steady level. In each case, there was a transition between the two kinds of responses at a critical point during the upstroke. Figure 4a illustrates a typical Ca2+ removal experiment: the transition was very sharp and OCcurred close to [Ca*+]ss,ryan (dashed line). Figure 4b shows a case in which the transition was more gradual and occurred at a level well above [Ca2+]ss,lyan. Figure 4c shows the effect of lowering [K’],, from 30 to 2 mM at various points in the cycle. Here, as in Figure 4a, [Ca2+], declined monotonically or increased transiently after the perturbation, with the transition OCcurring over a narrow range of [Ca*+],. In one case, [Ca2’], stabilized near the level reached at the time of the perturbation, hovering there for several seconds before finally rising to give a full-blown [Ca2’l, transient.

NWJrCJn 1114

Figure tween

5. Quantitative Net CaJ+ Entry

Comparison and Reiease

be-

(a and 5) Rapid removal of externa! Ca” (a) or caffeine (b) produced a sudden drop in d[Ca’*],/dt (Ad[CaL’]./dtL, and Ad[Ca2’],i (a, top) Orrr interpreta&a,,, respectively). tion of the effects of Cal+ removal. Under steady-state conditions (left), net Ca” entry across the surface membrane (J,,)rhan) is balanced by net CaLi extrusion (J,o.pump), so that 5 coffe1ne 5 coffe1ne the ne? flux across the surface membrane (I,,)) is zero. Removing external CaLL elimrz co, 2 co, leaving J,0,pU,71i>unopposed nates Lh, (right). [Ca”], falls at an initial rate given by Ad[Ca2’j,/dtLc, that is determined by J,o,pump; since I,+],- or time-dependent inactivation of CICR. After [Ca”],, was restored, [Cz?‘], transients were initially prolonged, but gradually returned to their original shape. Cell B12Z

Are Changes in [Ca*+]i Rate Limited by Removal of CICR Inactivation? Inactivation of CICR must be considered as a potential factor in controlling the way that [Ca2+], changes during the oscillatory cycle (Fabiato, 1985). For example, slow removal of [Ca*+],or time-dependent inactivation of Ca2+channels could influence the cycle period. To test this, we removed external Ca2+ when [Ca*+], was near its lowest value and then restored it after varying lengths of time to see whether the latency to next peak was changed (Figure 7~). If the slow rise in [Ca*+], between spikes reflects slow removal of inactivation, then lowering [Ca2+], for a time comparable to the oscillatory period should permit significant recovery. It was found that if Ca2+ was removed for 113 s (ml.5~ cycle period), the latency to the next spike after restoring Ca2+ was no shorter than if Ca2+ was removed for 13.6 s (~0.3 min-‘), even in cells that were impaled with microelectrodes. One possibility is that microelectrodes produce a Ca2+ leak of variable degree at the site of impalement (R,, = 65 MQ in Kuba and Nishi’s experiments versus R,, = 350 MD in the present study). Biological differences between isolated cells and neurons in situ are also possible. Finally, we have presented evidence that the caffeinesensitive store exchanges Ca2+ with the cytosol, but not directlywith the external medium (Friel and Tsien, 1992). This differs from the conclusion of Kuba and Nishi (1976) that the store exchanges Caz+ readilywith the external medium. We argue below that uptake of cytosolic Ca2+ b y the store is crucial for caffeineinduced [Ca*+], oscillations: it drives [Ca2+], below enhances [Ca2+],-dependent deactivation Ka2+lss.nian,

of CICR, store.

and

permits

net

Ca2+

accumulation

Critical Points within the Oscillatory Cycle The oscillatory cycle can be divided into phases by considering several critical points the cycle (Figure 8). These points are defined

by the

distinct within based

on [Ca*+], and how it changes with time, but can be interpreted in terms of the underlying Ca2+ fluxes. This description follows from rather general considerations about a cell with an internal Ca*+ store that lies entirely within the cytoplasm and does not depend upon detailed properties of the underlying transport processes, which will be considered below. Two clearly defined points occur where [Ca2’], is maximal or minimal (labeled 1 and4 in Figure8).Additionally, points 3 and 6 define where [Ca”], crosses the steady-state level reached following exposure to 1 PM ryanodine, after oscillations cease ([Caz+]ss,ryan, dashed line). With some simplifying assumptions (see below),itcan beshownthatthefluxesundergocriticai transitions at these points, as depicted in the Ca2+ flux diagrams in Figure 8b. At any instant during the oscillatory cycle, the total net cytosolic Ca2+ flux (J,,r,~) is the sum of the net Ca2+ flux between the cytosol and the external medium (JIO) and between the cytosol and the store (J,,) (see Figure 8a, inset). In the steady state after ryanodine has rendered the store inoperative, J,O is zero and [Ca*+], = Ka2+Lvan. Since [Ca2+]5z,ryan is stable, JiO must be outward when [Ca*+], >[Ca2+]55,rvan and inward when [Ca2’], for small changes in [Ca”],. If these con< [Ca2+lss,ryan, siderations also apply while [Ca”], oscillates (see below), several conclusions follow. At point 1, d[Ca2’],/dt = -Jtoral/v, = 0, so that Jtota, = 0 and JIs = -JlO. Since [Ca2+], > [Ca2+]ss,Yan, JiO > 0 (i.e., outward), and as a consequence, J,5 < 0 (i.e., inward). Thus, the net flux from the store is just balanced by the net outward flux across the surface membrane. At point 4, similar arguments lead to the conclusion that the net flux from the cytosol into the store just balances the net inward flux across the surface membrane. At point 3, J10 = 0, and since d[Ca2+],/dt is negative, Jl,t,~ = JIs > 0. Here, [Ca2+], falls exclusively under the influence of Ca2’ uptake by the store. At point 6, d[Ca2+],/dt is positive and Jtol,l = J,z. < 0, so that [Ca*+], rises exclusively under the influence of internal Ca*release. Two additional points, 2 and 5, are depicted. Point 2 defines where Ca2+ release gives way to uptake. While its precise location is not known, it must fall between points 1 and 3, since within this interval JIs changes sign. Similarly, point 5 defines where uptake gives way to release. This partitioning of the oscillatory cycle follows directly from the assumptions that (i) [Caz’], is influenced by Ca*+ exchange between the cytosol and two other compartments, one internal, the other external, and (ii) there is a single [Ca2+], level encountered dur-

\c$],

Oscillations

in Sympathetic

Neurons

a

7

50

nM

b

Figure

8. Critical

Points

2 within

the Oscillatory

Cycle

(a) A representative oscillatory cycle (cell B12J), in which the dashed line gives an membrane Ca*+ flux is zero, based on the steady-state level reached in the presence cycle are specified (1, 3,4, and 6) along with two additional points (2 and 5) whose the indicated intervals. (b) Our view of the direction and relative magnitudes of the net Ca I+ fluxes across (J,,) at critical points (l-6) (see inset in [a] for definition of symbols and direction

ing the oscillatory cycle where J,O = 0 and around which small changes in [Cap+], lead to restorative changes in J10 (i.e., aJiJa[Ca*‘], > 0). This level was which is valid as long as surestimated by [Ca2+lss,lyan, face membrane Ca*+ fluxes (iii) depend only on [Ca*+]i, [Ca2+10, V,, and [caff],, variables known to influence surface membrane Ca*+ transport and to have the same values at points 3 and 6 (Figure 8) as they do in the steady-state, (iv) adjust rapidly to changes in these quantities, and (v) are not influenced by 1 uM ryanodine. If assumptions (iii-v) do not hold, then JIO may be nonzero at [Ca2+lss,tyan, leading to an error in the precise location of points 3 and 6. However, as long as assumptions (i-ii) are satisfied, the partitioning of the oscillatory cycle described above remains valid. An Interpretation Caz+ Release The subdivision

Based of

the

on

Ca*+-Induced

oscillatory

cycle

described

estimate of [CaL+], ([Caz+],,) at which the net surface of ryanodine, [Ca2+]rr,rvan. Four critical points in the precise locations are unknown but which fall within the plasma of positive

membrane net flux).

(J,,) and the store

membrane

above is general and does not depend on detailed properties of the underlying transport processes. However, several additional observations can be explained if the hypothesis incorporates CICR-that is, Ca*+ release through channels that are gated by Ca2+ and regulate Ca*+ exchange between an internal store and the cytosol. We propose that, as in other systems, caffeine increases the [Ca*+], sensitivity of release channels and that [Ca*+], oscillations occur under conditions where small changes in [Ca*+], produce significant changes in channel activation. Under such conditions, when the net Ca*+ flux from the store exceeds the net outward flux across the surface membrane, [Ca2+], will rise (e.g., during the upstroke in Figure 8a). The rise in [Ca*‘], leads to further channel opening, enhancement of Ca*+ release, and a regenerative rise in [Ca*+], ([Ca*+],-dependent activation of CICR). Since the store contains a finite amount of Ca*+, its content declines,

NC3lKXl 1120

leading to a reduction in net flux from the store; [Ca*-1, reaches its maximal value (point 1) when the declining net inward flux from the store is balanced by the rising net outward flux across the surface membrane. As the flux from the store declines further, Jut,1 becomes outward & > -JI,) and [Ca2+], declines. This leads to closure of Ca2+ release channels, a further reduction in the net flux from the store, and a regenerative fall in [Ca”]; ([Ca*+],-dependent deactivation of CICR). The net flux from the store declines and changes sign at point 2 to become a net flux into the store; [Ca2+], then falls under the combined influence of uptake by the store and extrusion across the plasma membrane. At point 3, [Ca2+], = [Ca2+]ss,ryan, J10 = 0, and [Ca>+], falls solely under the influence of net Ca2+ uptake by the store. This drives [Ca2+], below [Ca2A]si,ryan, and as a result, J10 becomes inward. As the store refills, the net flux into the store declines; [Ca2+], reaches a minimum (point 4) when the declining flux into the store is balanced by the rising net inward flux across the surface membrane. As net uptake by the store declines further, Jtot,l becomes inward (J,, < -JIs) and [Ca*“], rises. This leads to channel opening, a further reduction in net uptake by the store, and a regenerative rise in [Ca2’], ([Ca2+],-dependent activation of CICR). The net flux into the store falls and changes sign at point 5 to become net release. [Ca2+], then rises under the joint influence of Ca*+ release and Ca2+ entry across the surface membrane. At point 6, [Ca2+], = [Ca2+]ss,rvan, Jln = 0, and [Ca2’], rises solely under the influence of the net Ca2+ release from the store. As [Ca2’], rises above J,” becomes outward, but since J,, < -JIs, Ka2+lis,ryan, Jtot,r is inward and [Ca2+], continues to rise. Since the store is replenished, this leads to a regenerative rise in [Ca*+],. This scenario for [Ca2+], dynamics during the oscillatory cycle provides a framework for discussing the experimental results. Oscillation Frequency Increases with [Ca2+], According to the hypothesis, replenishment of the store after each Ca*+ spike occurs at the expense of cytosolic Ca2+ and thus depends upon slow Ca2+ entry and therefore [Ca*+]“. This explains why the interspike interval becomes shorter with graded increases in [Ca2+],, (Figure 6a). The Caz+ Spike Shows Little Dependence on [CP], TheCa2+ spike itself showed littlevariation with [Ca2+]0 over a concentration range in which the cycle period changed considerably (Figures 6a and 6~). This observation makes sense in terms of the relative magnitudes of Ca2+ delivery via Ca2+ entry and Ca2+ release. As soon as [Ca2’], rises beyond WOO nM, Ca2* release channels open regeneratively, producing a dominant inward flux. CICR then undergoes a stereotyped program of activation, leading to a [Ca”], spike with relatively invariant time course and little dependence on Ca2+ entry or [Ca2+10. The Rise in [Ca2+J between Spikes Undergoes a Transition from Being [Caz+J, Dependent to [Caz+10 lndependen t To explore where in the oscillatory cycle release be-

comes self-sustaining, external Ca2+ was removed at various stages of the cycle and a point of no return, where Ca*+ removal no longer extinguished the Ca2+ spike, was determined (Figure 4). According to the hypothesis, this point must lie somewhere beyond point 5 in Figure 8a. The transition can be understood by separating JiO into its components, Ca2+ entry (Jro,chan) and extrusion (J,o,pump; see Figure 5a, top). Since J,o + JIs and J,o = Jio,chan i- Jio,pump, Jtota~ = Jlo,chan + + J,,. After removing external Ca2+, Jlo,‘han = 0, and [Ca”], now changes under the influence of Jrolal, which at [Ca2+10 = 0 is given by J,o,pump -I Jis. if, at the instant that Ca2+ is removed, the rate of extrusion exceeds the rate of release (J,o,p~mp > -JIs), J,,:,I will be outward and [Ca2’], will fall. On the other hand, if release is more powerful than extrusion (J,, < -J,o,pump), Jtot,r will be inward and [Ca2+], will continue to rise. At the point where release is just balanced by extrusion (Ito,pump = -Jii), Jtomwill be zero; this is where the rise in [Ca”], undergoes the transition from being [Ca2+], dependent to being [Ca*+],, independent (Figure 4~). Oscillation Frequency and Amplitude Are Both [Gaff], Dependent Raising [caff], increased the frequency but decreased the amplitude of the Ca2+ spikes. The shortening of cycle period can be accounted for if increasing [caff],

Jtotal = J,o,pump

causes CICR to activate at lower (Ca2’],, as suggested by the results in Figure 6d, and thereby comes into play at an earlier stage in the interspike interval. Effects on spike amplitude are expected to depend on a combination of changes in the store’s Ca*+ permeabilityandCa2’content.Atlow[caffi,,spikeamplitude should increase with [caff],, because of promotion of CICR; this prediction fits with the experiments of Kuba and Nishi (1976). Over a higher range of [caff],, reduction in the store’s Ca2” content would be expected to limit release; this may explain our observations at 5-10 mM [cafflO. At very high [cafflO, where oscillations are not observed, changes in [Ca*+], may not produce significant changes in CICR, either because the store is empty or because release channels are maximally activated. Another possibility is that at 30 mM, caffeine inhibits voltage-dependent Ca2’ entry (Lipscombe et al.! 1988; Hughes et a!., 1990). Evidence for Variations in Surface Membrane Ca2+ Permeability during the Oscillatory Cycle One unexpected result is that Ca2’ entry across the surface membrane appears to vary significantly during the oscillatory cycle, even though V, and [Ca2+]0 are held constant. Experiments measuring the initiai change in d[Ca2+]i/dt following Ca2+ removal suggest that entry rises with [Ca2+], during the spike, contrary to what might be expected for changes in Ca2+dependent inactivation of Ca2+ channels (Brehm and Eckert, 1978). The rise in /Ad[Ca2’],/dt-c,I with [Ca2+], implies that Ca2+ channel activation increases with moderate [Caz+]i elevations, as found in heart cells (Marban and Tsien, 1982; Gurney et al., 1989). Our finding that Ca2+ entry varies within the oscillatory cycle agrees with the conclusions of Loessberg et al.

[Ca2’], Oscillations 1121

in Sympathetic

Neurons

(1991) for agonist-induced oscillations in AR42) cells, but contrasts with those of Jacob (1990b) for endothelial cells. Our results do not distinguish between rapid [Ca*+],dependent modulation of Ca2+ entry and changes mediated by slow metabolic effects; conceivably, modulation through Ca*‘-dependent protein kinases might take place on the slow time scale of the [Ca*+], oscillations. If Ca2+ entry depends on the previous history of [Ca*+],, then assumption (iv) above is not valid, introducing uncertainty in the location of critical points 3 and 6 in Figure 8. Ultimately, direct measurement of surface membrane Ca 2+ fluxes will be necessary to locate these critical points more precisely.

Assessment of Contributions from Other Stores Our interpretation presumes that caffeine-induced [Ca”]i oscillations reflect transport activity of a single internal store. While CICR appears to be necessary for oscillations, it does not follow that it is sufficient. In addition to the caffeine-sensitive store, bullfrog sympathetic neurons contain stores sensitive to inositol trisphosphate (IP3) and carbonyl cyanide p-trifluoromethoxyphenylhydrazine (FCCP) (Pfaffinger et al., 1988; Friel and Tsien, 1990, Sot. Neurosci., abstract; see also Thayer and Miller, 1990). Compared with caffeine, IP3 agonists have only small effects on [Ca*+],, but the FCCP-sensitive store can strongly influence [Ca2+],. This store is probably not required fortheoscillations described here, since they occur at [Ca2’], levels where the FCCP-sensitive store has little effect on [Ca2’], and continue in the presence of 1 PM FCCP, which appears to reduce or eliminate contributions from this store (D. D. F. and R. W. T., unpublished data). Finally, theoretical work has shown that [Ca”]i oscillations can occur if intracellular Ca2+ transport processes have a nonlinear dependence on [Ca”], (Kuba and Takeshita, 1981). In any case, more information will be required to establish whether other Ca2+ stores contribute to the [Ca2+], oscillations described here.

Relevance to [Ca*+]i Oscillations in Other Systems Our results provide an illustration of the interplay between different Ca2+ delivery systems in the control of the cytosolic free Ca2+ concentration. We find that caffeine-induced [Ca2+], oscillations depend jointly on surface membrane Ca*+ entry and CICR. The physiological relevance of such oscillations remains unknown, but it is interesting to note the growing evidence that CICR can be regulated by several cellular metabolites, including ATP (Endo, 1985; McPherson et al., 1991) and cyclic ADP-ribose (Galione et al., 1991). Since bullfrog sympathetic neurons have received much attention as a model neuron (Adams et al., 1986), additional experiments studying [Ca2’], regulation in these cells may be particularly useful. Studies of [Ca2+], oscillations have traditionally been divided into analysis of membrane potential-driven

[Ca”], oscillations in excitable cells and “subcellular” or “cytosolic”[Ca2+], oscillations in nonexcitable cells. Our experiments show that caffeine-induced [Ca”], oscillations in bullfrog sympathetic neurons are due to a”cytosolic oscillator”that operates at constant V,. This underscores the point that excitable cells are capable of both kinds of behavior (Tsien et al., 1979; Berridge and Rapp, 1979). There are interesting analogies between the oscillations described here and certain examples of agonistinduced [Ca2’], oscillations. Agonists that stimulate production of IPj often induce [Car’], oscillations (Woods et al., 1986; Harootunian et al., 1988; Malgaroli et al., 1990; Wakui et al., 1990; Foskett et al., 1991; Berridge, 1991). It has been argued that in some cases oscillations result from agonist-induced enhancement of Ca2+ entry and periodic activation of CICR (Berridge, 1990). Increased Ca*+ permeability of the plasma membrane seems to be involved, but it is not clear how it arises. One proposal is that permeability is controlled by the content of the IP3-sensitive store: agonists promote Ca2+ entry by emptying the store (Putney, 1986; Hoth and Penner, 1992). According to this view, any treatment that discharges the IP3-sensitive store should promote Ca2+ entry, activate CICR, and produce oscillations much as agonists do. In support of this idea, thapsigargin (TG) induces [Ca2’], oscillations in rat salivary gland cells (Foskett and Wong, 1991) and human Tcells (R. S. Lewis and R. Dolmetsch, personal communication). TG discharges the IPs-sensitive store, apparently by inhibiting specificallythe Ca2’-ATPase mediating Ca2+ uptake into this store (Thastrup et al., 1990). Like agonists, TG promotes Ca2+ entry, but has relatively little effect on IP3 production (Foskett and Wong, 1991). TG-induced oscillations resemble agonist-induced oscillations in terms of their requirement for external Ca2+ and are sensitive to caffeine and ryanodine (Foskett and Wong, 1991). In these respects, TG-induced oscillations resemble the oscillations described in our study, although here Ca2+ entry was promoted bydepolarization rather than by TG. One feature of agonist-induced oscillations that has received great attention is the dependence of oscillation frequency, but not spike shape, on agonist concentration. According to the model proposed by Berridge (1990), this occurs because agonist concentration regulates IP3 concentration, the Ca2+ content of the IP3-sensitive pool, and thereby the surface membrane Ca2+ permeability. Since CICR is insensitive to IP3, spike shape and amplitude are unaffected. This is interesting in view of our finding that increasing [Ca2+10 over a range that speeded the slow interspike rise in [Ca2+], and elevated the oscillation frequency had little or no effect on spike amplitude or shape (Figures 6a and 6~). These resemblances between caffeine-induced [Ca2+], oscillations in sympathetic neurons and many instances of agonist-induced oscillations suggestthat the underlying mechanisms may be quite similar.

Nl3UVJ” 1122

Experimental

Procedures

Microelectrodes mize disruption

The methods used in this study are described (Friel and Tsien, 1992), but are summarized

in detail below.

patch

pipettes

to mini-

elsewhere

Cell Dissociation and Primary Culture Conditions Bullfrogs (Rana catesbeiana) were killed, and the sympathetic chains were removed and placed in sterile amphibian Ca2’-free Ringer’s solution (128 mM NaCI, 2 mM KCI, 10 mM glucose, 10 mM HEPES [pH 7.3 with NaOH]). Chains were incubated for 40 min at 35OC in 2 ml of Ringer’s solution + coliagenase (3 mgimi; Worthington), transferred to Ringer’s solution + trypsin (1.5 mgi ml;Sigma,typeIII),and incubated forlOmin.Afterwashing,they were triturated until the solution became cloudy. After washing again, a drop of the cell suspension was placed in the center of a cover glass coated with poly-t-lysine (Sigma) and affixed with Sylgard (Dow Corning) to the bottom of a 60 mm culture dish, covering a 20 mm hole in the dish. After letting ceils adhere to the substrate, a I:1 mixture of normal Ringer’s solution (which contains 2 mM CaCiJ and modified l-15 medium (CIBCO) was added, and cells were stored at room temperature (18°C-20”C) for up to 1 week. Experiments were carried out at the same temperature with cells having a B cell appearance (Adams et al., 1986). High K’ solutions were equivalent to the Ringer’s solution described above except that 28 mM Na’ was replaced by K+. Nominally Caz+-free solutions contained 0.2 mM added ECTA and 2 mM MgCI,. When [Ca2+10 was varied systematically (Figures 6a and 6c), MgClz was added so that 2 x [Caz+la + [Mg”], = 4 mM. Drugs Sources of drugs were as follows: fura(acetoxymethyl ester and free acid; Molecular Probes), pluronic (BASF Wyandotte), ryanodine (Research Biochemicals Inc.), and caffeine (Sigma). Fresh ryanodine stock solutions (100 mM) were made each day in 95% ethanol and then diluted lOi-fold in 2 mM Ca2+ Ringer’s solution. Control experiments showed that 0.95 x 10e5% ethanol did not have any detectable effect on [Ca2’],. [Caz+], Measurements Cells were incubated for 40 min at room temperature in a solution containing 1 PI of furastock solution per ml of Ringer’s solution. The stock solution contained 1 mM furaacetoxymethyl ester and 25% (w/w) pluronic in dimethyl sulfoxide. Cells were then rinsed with Ringer’s solution and placed on the stage of an inverted microscope (IM 35,Zeiss)and superfused continuously with Ringer’s solution. Cells were illuminated with light from a 150 W Xenon lamp (Muller Electronic-Optics) transmitted by fiber optics to excitation filters (350 and 380 nm) mounted in an eight-position filter wheel that rotated at 20-40 Hz. Excitation light was reflected by a dichroic mirror and transmitted to cells through a 40x objective (Nikon Fluor, NA 1.3). Emitted light passed through the dichroic mirror, a barrier filter, and an aperture centered on the cell body. Light intensity was measured with a photomultiplier tube (PMT, Thorn, OL 30 Series), giving a spatial average of emitted light intensity over the cell soma. The PMT signals associated with the two excitation sources were separately integrated (Cairn Research Spectrophotometer) to give two analog signals (FxjO and F& that were updated once per revolution of the filter wheel. Each signal was sampled every 150-350 ms by digitizing at 2.5 kHr over 12.8 ms intervals and averaging. Results were stored on a laboratory computer. [Ca’+], was calculated from FIsO and FiRD according to Crynkiewicz et al. (1985) after subtracting the light intensity measured in the absence of cells (FIFO,l,k and F IRU,~,~)and computing the ratio R as = (F,. - F~~o,dh - Fim,~ ) R,,, and R,,, were determined described in Friel and Tsien (1992). Electrophysiology Simultaneous measurements impaling cells loaded with fura electrodes (R = 120 MD) filled with a List EPC-7 patch clamp

were used rather than of [Ca2-1, homeostasis.

of [CaLA], and V,, were made by 2.AM with high resistance microwith 3 M KCI. V, was measured under current-clamp conditions.

Solution Changes Solution changes were made with a system of microcapillanes (Drummond Microcaps) as described by Friel and Bean (1988) except that 20 ~1 capillaries were used and solution changes were achieved by moving the capillaries to the cell. Cells were uniformly superfused with the desired solutions, and solution changes were rapid (~100-200 ms), based on the fast and steady change in V,, achieved when cellswere exposed to external solutions with elevated [K‘], (Friel and Tsien, 1992). The internal diameter of the capillaries was approximately 5 times the typical cell diameter. While cells were locally superfused, the recording chamber was also superfused with Ringer’s solution at -5 mlimin. Assessment of Ca2+ Fluxes Ca2+ fluxes were assayed by measuring the initia! change in d[Caz+],/dt (Ad[Caz’],idt) immediately following a perturbation that was chosen to eliminate a specific flux component. To measure Ad[Ca2+]./dt from the [Ca2’], records, data preceding the perturbation were aligned in time with a corresponding stretch of data from a reference cycle. Alignment was done interatively usingaleastsquarescriterion. Pre-perturbation datapointsfrom test and reference records (~100-300 samples) were shifted in time relative to one another, and the alignment that minimized the mean-squared difference in [Ca2+], between the test and reference data sets, giving essentially superimposable records, was used for analysis. d[Ca2’],/dt was then determined by fitting a line to the linear portion of the records (linear regression) just following the perturbation and comparing with d[Ca”],/dt measured in the reference cycle (see Figure SC). The close agreement between test and reference data sets preceding the perturbations suggests that the state of Ca*+ regulation in a given cell was periodic, validating the comparison between the two sets. In general, Ad[Ca2+],/dt will measure the instantaneous change in net Cal+ flux following a perturbation. The reasoning behind this approach is described as follows. We distinguish between the external medium (compartment o), the cyiosol (compartment i), and the internal store (compartment s) and assume that the Cal+ concentration is spatially uniform within each compartment, i.e., that diffusion within compartments 1s fast compared with exchange between compartments during the oscillatory cycle. This is almost certainly valid for [Ca2L]0 (continuous singlecell superfusion) and for [CaZ+], (based on digital imaging experiments), whereas for [Ca*+], this assumption is untested. At each instant in time t, the rate at which [Cal], changes is proportional to the total net flux, Jtol.,, entering the cytosol at that time: d[Ca’+],(t)/dt

= -j,,,,!!t)/v

8,

where v, IS the volume oi the cytosolic compartment and J,,,&t) is the Ca*+ flux density summed over all membranes bounding the cytosol. The negative sign implements the convention that negative J,“,,, represents the net /nward CaZ’ flux, while positive values give the net outward Ca >+ flux. If a perturbation changes A], then i fU,a~by an amount Ad[Ca’Ydt

= d]Ca’-Udt,,,,,,

= -i(Jm,a,+AJ)

pertur~a,,o,,,- dKa'+l,idt!be:ore

perjurhai,onl

- Jmta~j~v,.

= -AJ/v,. Three types of perturbation were used to monitor components of I,,,,, within the oscillatory cycle: removal of external Ca2+ for reduction of [K+],,), to monitor CaLi entry from ihe external medium; caffeine removal, to assay the caffeine-sensitive flux between the store and cytosol; and elevation of the caffeine concentration to a high level to assay the amount of releasible Cal’ In the store. In the remainder of this section, the conditions under which these perturbations assay the desired fluxes will be discussed. It is necessary to separate J,,,Iinto two components, the net

[Ca*+], Oscillations

in Sympathetic

Neurons

1123

Ca2+fluxacross thesurfacemembrane(j,,)and the store membrane (J,,), where subscript direction for positive net flux:

the netfluxacross order indicates the

Note that Ad[Ca2+],/dtmc, and v,, but their ratio is independent net [Ca*+],,- and [caffl,-sensitive

Ad[Ca2+],/dt-,,ft of v, and gives fluxes:

both depend on the ratio of the

Jtota, = Jo0 + Is.

= I ~~,~~.~([Ca*‘l,)/{J,,,~,,,([caffl~)

C.++ Entry from the External Medium To assay the rate at which Ca2+ enters the cytosol nal medium, CaI+ was suddenly removed and d[Ca*+],/dt (Ad[Ca2+],/dt-c,) was measured: Ad[Ca2+],/dt-c,

from the exterthe change in

= {d[Ca2+],idt,c,,,,o} - {d[Ca2+],/dtlc,1,} = -{J&O) - JtO,.I(JCaz’l,])lv,, = - ({J,,(O) + J,,(O)} - {J,O([Ca*tl,)

+ J,~(JCaz+l,)j)~v,,

where d[Ca*+],/dt,c,,=o is evaluated over a small time interval starting at the time of the perturbation, and d[Ca2+],/dtrc,,, is evaluated over a similar interval from a reference cycle. If J. does not depend on [Ca2’], just after Caz+ removal, then J,,(O) = J,$([Ca2’],) and AdKa*+l,k-Jt-c,

= {J10(JCa2’10) - J,,(O)j/v,.

Separating J,O into the restorative net Ca2+ flux generated by pumps and exchangers (Jlo.eump, referred to as Ca2+extrusion) and the passive net flux through channels (J,o,chan, referred to as Ca’+ entry); Ad[Ca2’]/dt-,I

d

Note also that according to this view, Ca*+ and caffeine play fundamentally different roles in Caz+ transport. Ca*+ is regarded as a substrate for a permeability in the plasma membrane (e.g., voltage-dependent Ca*+ channels) such that Ca2+ removal deprives this transport pathway of a substrate. Caffeine, in contrast, is regarded as controlling a Ca2+ permeability of the store membrane, so that caffeine removal eliminates this permeability. Assessment of the Store’s Cd* Content We used the change in d[Ca2+],/dt following a sudden increase to monitor the in bffl, from kaffl,.,,,,~ to [cafflt,.,, (Ad[Ca*‘],ldt+,,~~) Ca*+ content of the store at different points in the oscillatory cycle. One way to interpret Ad[Ca2+],/dt+,,ft is to assume that J,,,c,cr depends on the product of a [Ca*+],- and [caffl,-dependent Ca*+ permeability of the store membrane (P([Cazf],,[caffl,)) and a chemical driving force for net Ca2+ release ([Ca*‘], - [Cal],):

J,~,~,~,([Ca2’l,,[caffl,) Then,

arguing

as with

Ad[Caz+],idt+,,t,

= (~J,o.~~~nKa2+lo) + J,,,um,(Kaz’lo)I

= -~~~Ca2’l,,~caffl,~~~Ca2’l,

-{Jlo,chan(0) + J,o,pump(0)h.

Ad[Ca2+],/dtm,,+f,

-{J,,([caffl,,,,,,I)

= J,o,~~,.(Ka2’l 0)/v I

CaffeineSensitive Cd’+ Release from the internal Store To monitor the rate at which Ca2+ enters the cytosol from the store through the caffeine-sensitive pathway, caffeine was suddenly removed and thechange in d[Ca2+],/dt (Ad[Ca2+],idt&,,) was measured:

Ad[Ca2+],/dtm,,,,

= {dKa2+l,/dtc,t+a)

- {dKa2+l,/dt,,,,,J

= -({J,0(0)+J,~(O)}-{J,0([caffl,)+J,,([caffl,)})/v,

If J,O does then

not depend

strongly

Ad[Ca2+],/dtL,rt

on [caffl,

= {J,.f[caffl,)

(i.e., J,&affl,)

= J&O)),

- J,,(O)]/v,.

Separating J. into the net Ca*+ flux generated by pumps referred to as Ca*+ uptake) and the passive net flux through to as Ca2+ release); nels Lhan, referred

CJ,r.pump, chan-

Ad[Ca*+],/dt-,,t, = (tJ ,s,pum,(kafflo)

+ J,5.&kafflo))

- {J,s,,.m,(0)

+ Jlr,chan(0)Ih

If the rate of Caz+ uptake by the store does not depend on Jcaffl, (i.e., J,~.p&JcafflO) = J,r,pumpfO)) then

Ad[Ca*+],/dt-,,I1

= (J,..&[caffl,)

the net caffeine-sensitive

= {J,.,Acafflo) flux

through

Ca2+ release and if Jlr,teak

If P([Ca*+],,[caffl,,,.,) is much greater than P([Ca*+],,[caffl,,,,,,I) and independent of [Ca*‘], (i.e., if [caffk,.,, raises P([Cazf],,[caffli,,,,) to a saturating level that greatly exceeds P([Ca*+],,[caffl,,,,,,,)) and if [Ca2+]$ >> [Ca2+],, then Ad[Caz+],/dttca(t

= constant

x [Ca*‘],.

[Ca*‘], For kaffl,.,t,al and [caffl,,,,, = 5 and 30 mM, respectively, increased to a peak very rapidly (within 1-3 sample intervals of making the solution change), so Ad[Ca2+],/dt+,,tt must be regarded as approximate. Population results are given as mean f SEM. Acknowledgments The authors would like to thank Drs. R. S. Lewis, A. D. Randall, and F. E. Schweizer for their comments on the manuscript. This work was supported by PHS grants NS24067and HL13156. Correspondence should be sent to D. D. Friel. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 USC Section 1734 solely to indicate this fact. Received

January

8, 1992; revised

March

18, 1992.

References

Berridge, 9583-9586.

M. J. (1990).

Calcium

oscillations.

Berridge, M. J. (1991). Caffeine inhibits induced membrane potential oscillations Proc. Roy. Sot. Lond. (B) 244, 57-62.

- J,,,&O)j/v,, CaZ+ release

+ J,~([caffl,,,,,,,)})/v,.

= -tJ&afflt,.,O - J,~([caffl,,,,,,l)}lv,. = -{J,.,~,~,([cafflf,..I) - J,,.,,,,([caffl,,,,,~l)}/v,. = {P([Ca2+],,[cafflt,,,I) - P([Caz+],,[caffl,,,,,~,)} x {[Ca*+], - [Ca2+],}/v,.

Adams, P. R., Jones, S. W., Pennefather, P., Brown, D.A., Koch, C., and Lancaster, B. (1986). Slow synaptic transmission in frog sympathetic ganglia. J. Exp. Biol. 724, 259-285.

- J,,,,,,,tO)}/v,.

If J,r,chan is separated further into the net flux through leak Ulr,& channels U.,~,~, ) and a caffeine-insensitive does not depend on [caffl,, then Ad[Ca2+],/dt&,,,

directly

- Ka2+l,L

= {d[Caz+l,ldt,,,l,,,,,I) - {d[Ca*+l,/dt[,,ttl,,,l,,ll = -(tJ,,Wfh) + J,,([caffl~,,,~)~

If the net Ca2+ flux through channels is small compared with the rate of extrusion when [Ca2’],=0 (i.e., Jlo,&O) + J,o,pump(0) = does not change very much J ,o,pump(0)) and the rate of extrusion just after removing external Ca2+ (i.e., J,0,,,m,([Ca2’],) = J,o,pump(0)), then

Ad[Ca2’],ldtm,,

- J,,,,,,,O}.

channels.

Berridge,

M. J., and Rapp,

P. E. (1979).

J. Biol.

Chem.

265,

inositol-trisphosphatein Xenopus oocytes. A comparative

survey

of

Neuron 1124

the function, mechanism Exp. Biol. 81, 217-279.

and

control

of cellular

oscillators.

J.

Bezprozvanny, I., Watras, J., and Ehrlich, B. E. (1991). Bell-shaped calcium-response curvesof Ins(l,4,5)PX-and calcium-gated channels from endoplasmic reticulum of cerebellum. Nature 351, 751-754. Brehm, P., and Eckert, R. (1978). Calcium tion of calcium channel in Paramecium. Endo, M. (1975). Mechanism plasmic reticulum of skeletal 484.

of action muscle.

entry leads to inactivaScience 202,1203-1206. of caffeine Proc. Jpn.

Endo, M. (1985). Calcium release from Curr. Top. Membr. Trans. 25, 181-230.

on the sarcoAcad. 51, 479-

sarcoplasmic

R. (1988). Trends

Ryanodine Neurosci.

receptor channel 11, 453-457.

of

Foskett, J. K., and Wong, D. (1991). Free cytopiasmic Ca” concentration oscillations in thapsigargin-treated parotid acinar cells are caffeineand ryanodine-sensitive. J. Biol. Chem. 266,1453514538. Foskett, J. K., Roifman, C., and calcium oscillations bythapsigargin Chem. 266, 2778-2782.

Kostyuk, transients 464.

P. G., Belan, P.V., and Tepikin,A. and oscillations in nerve cells.

V. (T991). Freecalcium Exp. Brain Res. 83,459-

Kuba, K. (1980). Release of calcium ions linked of potassium conductance in acaffeine-treated rone. J. Physiol. 298, 251-269. Kuba, K., and Nishi, S. (1976). Rhythmic depolarization of sympathetic ganglion feine. J. Neurophysiol. 39, 547-563.

reticulum.

Fabiato, A. (1985). Time and calcium dependence of activation and inactivation of calcium-induced release of calcium from the sarcoplasmic reticulum of a skinned caninecardiac Purkinjecell. J. Cen. Physiol. 85, 247-289. Fill, M., and Coronado, sarcoplasmic reticulum.

(1989). Cytoplasmic free Ca in isolated snail neurons as revealed by fluorescent probe fura-2: mechanisms of Ca recovery after Ca load and Ca release from intracellular stores. J. Membr. Biol. 710, 11-18.

Wong, D. (1991). Activation of in parotid acinarcells. J. Biol.

to the activation sympathetic neu-

hyperpolarization ceils induced

Kuba, K., and Takeshita, S. (1981). Simulation oscillation in a sympathetic neuron. J. Theor.

and by caf-

of intracellularCa’+ Biol. 93,1009-1031.

Lai, F. A., Erickson, H. P., Rousseau, E., ILiu, Q. Y., and Meissner, G. (1988). Purification and reconstitution of the calcium release channel from skeletal muscle. Nature 331, 315-319. Lattanzio, F. A., Schlatterer, R. C., Nicar, M., Campbell, K. P., and Sutko, J. L. (1987). The effects of ryanodine on passive calcium fluxes across sarcoplasmic reticulum membranes. J. Biol. Chem. 262, 2711-2778. Lipscombe, D., Madison, D. V., Poenie, M., Reuter, H., Tsien, R. W., and Tsien, R. Y. (1988). Imaging of cytosoiic CaZ+ transients arising from CaL+ stores and Ca?+ channels in sympathetic neurons. Neuron 7, 355-365.

Friel, D. D., and Tsien, R. W. (1992). A caffeine and ryanodinesensitive Ca*+ store in bullfrog sympathetic neurons modulates the effects of Ca*+ entry on [Ca2’],. J. Physiol. 450, 217-246.

Loessberg, P. A., Lhao, H., and Muailem, S. (1991). Synchronized oscillation of Caj+ entry and Cal- release in agonist-stimulated AR42J cells. J. Biol. Chem. 266, 1363-1366.

Galione, A., Lee, H. C., and Busa, W. B. (1991). Ca2+-induced Ca’+ release in sea urchin egg homogenates: modulation by cyclic ADP-ribose. Science 253, 1143-1146.

Malgaroli, A., Fesce, R., and Meldoiesi, J. (1990). Spontaneous [Cal+], fluctuations in rat chromaffin celis do not require inositol 1,4,5-trisphosphate elevations but are generated by a caffeineand ryanodine-sensitive intracellufar Cal’ store. J. Biol. Chem. 265, 300553008.

Gorman, A. L. F., and Thomas, M. V. (1978). Changes in the intracellular concentration of free calcium ions in a pace-maker neurone, measured with the metallochromic indicator dye arsenazo III. J. Physiol. 275, 357-376. Crynkiewicz, G., Poenie, M., and Tsien, R. Y. (1985). A new generation of Ca2+ indicators with greatly improved fluorescence properties. J. Biol. Chem. 260, 3440-3450. Gurney, A. M., Charnet, P., Pye, J. M., and Nargeot, J. (1989). Augmentation of cardiac calcium current by flash photolysis of intracellular caged-Ca2+ molecules. Nature 347, 65-68. Harootunian, A. T., Kao, J. P. Y., and Tsien, R. Y. (1988). Agonistinduced calcium oscillations in depolarized fibroblasts and their manipulation by photoreleased lns(1,4,5)PX, Ca2+, and Cal+ buffer. Cold Spring Harbor Symp. Quant. Biol. L//I, 935-943.

Marban, E., and Tsien, R. W. (1982). Enhancement of calcium current during digitalis inotropy in mammalian heart: positive feed-back regulation by intracellular calcium? J. Physiol. 329, 589-614. Marban, E., and Wier, W. G. (1985). Ryancdine as a tool to determine the contributions of calcium entry and calcium release to the calcium transient and contraction of cardiac Purkinje fibers. Circ. Res. 56, 133-138. McPherson, P. S., Kim, Y-K., Valdivia, H., Knudson, C. M., TakekH., Franzini-Armstrong, C., Coronado, R., and Campbell, K. P. (1991). The brain ryanodine receptor: caffeine-sensitive calcium reiease channel. Neuron 7, 17-25.

ura,

Hernandez-Cruz, A., Sala, F., and Adams, P. R. (1990). Subcellular calcium transients visualized by confocal microscopy in a voltage-clamped vertebrate neuron. Science 247, 858-862.

Meissner, C. (1986). Cal+ release channel 261, 6300-6306.

Hoth, M., and Penner, R. (1992). Depletion activates a calcium current in mast cells.

Meyer, T., and Stryer, L. (1991). Calcium phys. Biophys. Chem. 20, ?53-174.

Hudspeth, resonance bull-frog,

of intracellular stores Nature 355, 353-356.

A. J., and Lewis, R. S. (1988). A model for electrical and frequency tuning in saccular hair cells of the Rana catesbeiana. J. Physiol. 400, 275-297.

Hughes, A. D., Hering, S., and Bolton, T. 8. (1990). The action of caffeine on inward barium current through voltage-dependent calcium channels in single rabbit ear artery cells. Pflugers Arch. 416, 462-466. Imagawa, T., Smith, J. S., Coronado, R., and Campbell, K. P. (1987). Purified ryanodine receptor from skeletal muscle sarcoplasmic reticulum is the Ca2+-permeable pore of the calcium release channel. J. Biol. Chem. 34, 16636-16643. Jacob, R. (1990a). Calcium oscillations in electrically excitable cells. Biochim. Biophys. Acta 1052, 427-438.

non-

Jacob, R. (1990b). Agonist-stimulated divalent cation entry into singlecultured human umbilical vein endothelial cells. J. Physiol. 421, 55-77. Kostyuk,

P. G., Mironov,

S. L., Tepikin,

A. V., and

Belan,

P. V.

Ryanodine activation and inhibitron of sarcoplasmic reticulum. J. Bioi. spiking.

Annu.

of the Chem. Rev. Bio-

Mironov, S. L., and Usachev, J. M. (1991). Caffeine affects Ca uptake and Ca release from intracelluiar stores: furameasurements in isolated snail neurons. Neurosci. Lett. 723, 200-202. Nagasaki, K., and Kasai, M. (1984). Channel selectivity and gating specificity of calcium-induced calcium release channel in isolated sarcoplasmic reticulum. J. Biochem. 96, 176991775. O’Nerll, S. C., Donoso, P., and Eisner, D. A. (1990). The role of [Ca2+], and [Cal’] sensitization in the caffeine contracture of rat myocytes: measurement of [CaL’], and [caffeine],. J. Physiol. 425, 55-70. Pfaffinger, P. J., Leibowitr, M. D., Subers, E. M., Nathanson, N. M., Almers, W., and Hille, B. (1988). Agonists that suppress M-current elicit phosphoinositide turnover and Caz+ transients, but these events do not explain M-current suppression. Neuron 7, 477-484. Putney, J. W. (1986). A model entry. Cell Calcium 7, I-12.

for

receptor

controlled

calcium

[Ca2+], Oscillations 1125

in Sympathetic

Neurons

Rink, T. J., and Hallam, T. J. (1989). excitable cells: notes on oscillations cium 10, 385-395.

Calcium signalling and store refilling.

Rousseau, E., Smith, J. S., and Meissner, modifies conductance and gating behavior channel. Am. J. Physiol. 253, C364-C368.

in nonCell Cal-

C. (1987). Ryanodine of single Caz+ release

Rousseau, E., LaDine, J., Liu, Q. Y., Meissner, G. (1988). Activation of the Ca*+ release channel of skeletal muscle sarcoplasmic reticulum by caffeine and related compounds. Arch. Biochem. Biophys. 267, 75-86. Schlegel, W., Winiger, B. P., Mallard, P., Vacher, P., Wuarin, F., Zahnd, G. R., Wollheim, C. B., and Dufy, B. (1987). Oscillations in cytosolic Ca2+ in pituitary cells due to action potentials. Nature 329, 719-721. Sutko, J. L., Ito, K., and Kenyon, J. L. (1985) Ryanodine: a modifier of sarcoplasmic reticulum calcium release in striated muscle. Fed. Proc. 44, 2984-2988. Tepikin, A. V., Kostyuk, P. G., Snitsarev, V. A., and Belan, P. V. (1991). Extrusion of calcium from a single isolated neuron of the snail Helix promatia. J. Membr. Biol. 723, 43-47. Thastrup, O., Cullen, P. J., Drobak, B. K., Hanley, M. R., and Dawson, A. P. (1990). Thapsigargin, a tumor promoter, discharges intracellularCa’+stores byspecific inhibitionoftheendoplasmic reticulum Ca*+-ATPase. Proc. Natl. Acad. Sci. USA 87,2466-2470. Thayer, 5. A., and Miller, R. J. (1990). Regulation of the intracellular free calcium concentration in single rat dorsal root ganglion neurones in vitro. J. Physiol. 425, 85-115. Thayer, S. A., Hirning, L. D., and Miller R. J. (1988). The role of caffeine sensitive calcium store in the regulation of intracellular free calcium concentration in rat sympathetic neurons in vitro. Mol. Pharmacol. 34, 664-673. Tsien, R. W., and Tsien, R. Y. (1990). Calcium channels, and oscillations. Annu. Rev. Cell Biol. 6, 715-760. Tsien, R. W., Kass, R. S., and Weingart, (1979). Cellular lular mechanisms of cardiac pacemaker oscillations. 81,205-215.

stores,

and subcelJ. Exp. Biol.

Wakui, M., Osipchuk, Y. V., and Petersen, 0. H. (1990). Receptoractivated cytoplasmic Ca2+ spiking mediated by inositol trisphosphate is due to Ca%nduced Ca2+ release. Cell 63, 1025-1032. Woods, N. M., Cuthbertson, K. S. R., and Cobbold, P. H. (1986). Repetitive transient rises in cytoplasmic free calcium in hormone-stimulated hepatocytes. Nature 319, 600-602.