Hormones and neurotransmitters that act through inositol 1,4,5‐trisphosphate (IP3) can induce oscillations of cytosolic Ca2+ ([Ca2+]c), which render dynamic regulation of intracellular targets. Imaging of fluorescent Ca2+ indicators located within intracellular Ca2+ stores was used to monitor IP3 receptor channel (IP3R) function and to demonstrate that IP3‐dependent oscillations of Ca2+ release and re‐uptake can be reproduced in single permeabilized hepatocytes. This system was used to define the minimum essential components of the oscillation mechanism. With IP3 clamped at a submaximal concentration, coordinated cycles of IP3R activation and subsequent inactivation were observed in each cell. Cycling between these states was dependent on feedback effects of released Ca2+ and the ensuing [Ca2+]c increase, but did not require Ca2+ re‐accumulation. [Ca2+]c can act at distinct stimulatory and inhibitory sites on the IP3R, but whereas the Ca2+ release phase was driven by a Ca2+‐induced increase in IP3 sensitivity, Ca2+ release could be terminated by intrinsic inactivation after IP3 bound to the Ca2+‐sensitized IP3R without occupation of the inhibitory Ca2+‐binding site. These findings were confirmed using Sr2+, which only interacts with the stimulatory site. Moreover, vasopressin induced Sr2+ oscillations in intact cells in which intracellular Ca2+ was completely replaced with Sr2+. Thus, [Ca2+]c oscillations can be driven by a coupled process of Ca2+‐induced activation and obligatory intrinsic inactivation of the Ca2+‐sensitized state of the IP3R, without a requirement for occupation of the inhibitory Ca2+‐binding site.
One of the most common and fundamental mechanisms of cell signaling is through changes in the cytosolic free Ca2+ concentration ([Ca2+]c) (Berridge, 1993; Clapham, 1995). The predominant pathway of [Ca2+]c elevation in non‐excitable cells is through the second messenger inositol 1,4,5‐trisphosphate (IP3), which mobilizes Ca2+ from intracellular stores, most often associated with the endoplasmic or sarcoplasmic reticulum. The [Ca2+]c signals generated in response to activation of receptors coupled to this signal transduction pathway are often complex at the single cell level, even when receptor activation occurs in a sustained and continuous manner. Thus, IP3‐linked agonists can give rise to large repetitive [Ca2+]c transients, or oscillations, and each of these [Ca2+]c transients may be organized at the subcellular level as a propagating Ca2+ wave (Toescu, 1995; Thomas et al., 1996). Studies using Ca2+ indicator dyes within the intracellular stores of intact cells have shown that agonist‐induced [Ca2+]c oscillations are accompanied by inverse oscillations of stored Ca2+ (Tse et al., 1994; Chatton et al., 1995). In many cases, the strength of the activating stimulus is conveyed to the cell by the frequency rather than the amplitude of the [Ca2+]c oscillations, which has given rise to the concept of frequency‐modulated Ca2+ signaling (Berridge et al., 1988; Goldbeter et al., 1990). Although it is difficult to measure the output responses to these signals at the level of individual cells, it is clear that [Ca2+]c oscillations represent an important mechanism in regulating cell function (Tse et al., 1993; Hajnóczky et al., 1995). The manifestation of [Ca2+]c oscillations as Ca2+ waves appears to reflect the subcellular organization of the Ca2+ release system and the feedback interactions that give rise to the transient spikes of [Ca2+]c increase (Toescu, 1995; Thomas et al., 1996). Through these mechanisms, oscillatory [Ca2+]c signals can be directed to discrete subcellular domains (Kasai et al., 1993; Thorn et al., 1993), propagate through entire cells (Rooney et al., 1990; Lechleiter and Clapham, 1992) or even coordinate the activities of large numbers of coupled cells in intact tissues and organs (Stauffer et al., 1993; Sanderson et al., 1994; Robb‐Gaspers and Thomas, 1995).
The mechanisms underlying [Ca2+]c oscillations have been the subject of much study and of a number of reviews (Berridge, 1990, 1993; Petersen and Wakui, 1990; Meyer and Stryer, 1991; Thomas et al., 1996). Most of the models describing [Ca2+]c oscillations rely on positive and negative feedback effects on the Ca2+ release system, resulting in either fluctuations in the level of IP3 or changes in the activity of the Ca2+ channels of the intracellular stores. It has been suggested that oscillations of IP3 could arise from feedback activation of PLC by [Ca2+]c (Harootunian et al., 1991; Meyer and Stryer, 1991), and protein kinase C (PKC) has been proposed to function as a negative feedback regulator of IP3 generation (Cobbold et al., 1991). However, [Ca2+]c oscillations can be induced by direct introduction of non‐metabolizable IP3 analogs in both mammalian cells (Wakui et al., 1989) and in Xenopus oocytes (DeLisle and Welsh, 1992; Lechleiter and Clapham, 1992), suggesting that [Ca2+]c oscillations are not secondary to oscillating levels of IP3. An alternative mechanism of generating IP3‐dependent [Ca2+]c oscillations has evolved from the observation that the IP3 receptor Ca2+ channel (IP3R) is itself sensitive to both positive and negative feedback effects of Ca2+ (Iino, 1990; Bezprozvanny et al., 1991; Finch et al., 1991). Thus, it has been proposed that each [Ca2+]c transient is initiated by a local elevation of trigger Ca2+ that activates IP3Rs in the immediate vicinity to yield the rapid rising phase of Ca2+ release and propagation of [Ca2+]c waves. This process is believed to be terminated by a negative feedback effect of the elevated [Ca2+]c that inactivates the IP3R and allows the Ca2+ pumps to re‐sequester the released Ca2+. Experiments in intact cells have provided evidence for both the positive and negative effects of [Ca2+]c on the IP3R (Parker and Ivorra, 1990; DeLisle and Welsh, 1992; Lechleiter and Clapham, 1992; Oancea and Meyer, 1996), and regulation by luminal Ca2+ has also been proposed (Missiaen et al., 1991, 1992; Nunn and Taylor, 1992; Tanimura and Turner, 1996a). However, it is apparent that recovery from the inhibited state of the IP3R depends on factors other than the decline of [Ca2+]c (Ilyin and Parker, 1994; Oancea and Meyer, 1996).
Previous studies of the mechanisms underlying [Ca2+]c oscillations have relied on manipulations in intact cells, or have examined the individual components in isolation using subcellular systems. In the present study, we have established a permeabilized cell system in which Ca2+ oscillations can be evoked by global application of IP3. Repetitive cycles of Ca2+ release and re‐uptake were monitored using low affinity fluorescent Ca2+ indicators localized within the intracellular Ca2+ stores (Hofer and Machen, 1993, 1994), and changes in IP3R permeability were also monitored using the retrograde flux of Mn2+ to quench luminal dye (Hajnóczky and Thomas, 1994; Hajnóczky et al., 1994). This approach has allowed us to define the minimum requirements for intracellular [Ca2+]c oscillations and dissect the mechanism in a single experimental system. IP3‐induced Ca2+ oscillations were found to depend on fluctuations of [Ca2+]c, but Ca2+ re‐uptake and control by the luminal Ca2+ content of the stores were not essential components of the mechanism. While this manuscript was in preparation, Tanimura and Turner (1996b) reported similar findings in salivary epithelial cells. In addition, we found that the IP3R undergoes an obligatory inactivation from the Ca2+‐sensitized state without the need for occupation of the inhibitory Ca2+‐binding site. Although this intrinsic inactivation is likely to occur together with Ca2+‐induced inhibition, experiments utilizing Sr2+ in place of Ca2+ suggest that the coupled processes of Ca2+‐dependent activation and subsequent obligatory inactivation of the IP3R is sufficient to generate [Ca2+]c oscillations in intact cells without utilizing the inhibitory Ca2+‐binding site of the IP3R.
Results and discussion
Ca2+ oscillations in intact and permeabilized hepatocytes
Treatment of hepatocytes with vasopressin causes a dose‐dependent generation of IP3 (Thomas et al., 1984), which is accompanied by [Ca2+]c oscillations at submaximal agonist doses and sustained [Ca2+]c increases with high levels of vasopressin (Figure 1A). The [Ca2+]c oscillations in hepatocytes represent a very clear example of frequency modulation. The interspike period and initial latency decrease as the agonist dose is increased, but the amplitude and kinetics of the individual [Ca2+]c spikes remain constant over a broad range of agonist doses (Woods et al., 1986; Rooney et al., 1989). The permeability of the intracellular Ca2+ release channels during stimulation with vasopressin was monitored in intact hepatocytes by measuring the Mn2+ quench of fura2 compartmentalized within the Ca2+ stores (Glennon et al., 1992; Hajnóczky et al., 1993, 1994; Renard‐Rooney et al., 1993). Addition of submaximal vasopressin after pre‐loading the cytosol with Mn2+ resulted in a series of brief steps of rapid quench reflecting the opening of the intracellular channels, and these steps were separated by extended periods of slow quench where channel permeability was low (Figure 1B). The Mn2+ quench steps and [Ca2+]c oscillations showed similar sensitivities to vasopressin dose for latency and frequency, and high levels of vasopressin caused a sustained Mn2+ quench that also paralleled the [Ca2+]c response. Importantly, the rapid phase of Mn2+ quench occurred with the same rate throughout the effective vasopressin dose range (2.09 ± 0.18%/s and 2.11 ± 0.09%/s at 50 and 0.5 nM vasopressin, respectively). These data show that submaximal vasopressin doses cause synchronized periodic activation and subsequent deactivation of the entire population of intracellular Ca2+ channels that can be activated by saturating levels of vasopressin. Thus, it appears that [Ca2+]c oscillations in the intact cell are driven by cycling between a fully open and a largely closed state of the IP3R channels.
In order to dissect the mechanisms involved in this process, we established a permeabilized cell system that responds to a fixed level of exogenously added IP3 with autonomous oscillations of Ca2+ release and re‐uptake at the single cell level. Hepatocytes were permeabilized with digitonin using a protocol that preserves the functional integrity of the endoplasmic reticulum (ER) Ca2+ stores (Renard‐Rooney et al., 1993; Hajnóczky et al., 1994). Luminal [Ca2+] ([Ca2+]ER) was measured with compartmentalized low affinity Ca2+ indicators (fura2FF or furaptra) (Hofer and Machen, 1993, 1994). Submaximal doses of IP3 evoked oscillations of [Ca2+]ER that were inverted relative to vasopressin‐induced [Ca2+]c spikes in intact hepatocytes (Figure 1C). Thus, each [Ca2+]ER spike consisted of a rapid Ca2+ release phase followed by a slower re‐accumulation of Ca2+. The [Ca2+]ER oscillations occurred in a coordinated manner throughout each cell, but adjacent cells in the imaging field responded asynchronously (Figure 2A). Stepped increases in IP3 concentration increased the oscillation frequency, with little change in the kinetics or amplitude of the individual [Ca2+]ER spikes (Figure 2B). The interspike period varied from 20 to 240 s, which is similar to the range observed for agonist‐induced [Ca2+]c oscillations in intact hepatocytes (Rooney et al., 1989). Maximal doses of IP3 caused a rapid and persistent loss of [Ca2+]ER (Figures 1C and 2B). Mn2+ quench of compartmentalized fura2 was also used to monitor changes in IP3R permeability during IP3‐induced [Ca2+]ER oscillations in the permeabilized hepatocyte preparation. Consistent with the intact cell data (Figure 1B), submaximal doses of IP3 evoked brief bursts of rapid Mn2+ entry into the stores, separated by extended periods where Mn2+ permeability returned close to the basal rate, whereas maximal IP3 caused a sustained and complete quench of the luminal dye (Figure 1D). Taken together, these data demonstrate that the entire process responsible for [Ca2+]c oscillations can be reproduced in a cell‐free system in which the plasma membrane is disrupted and only the Ca2+ fluxes mediated by intracellular organelles remain intact.
Mechanisms of permeabilized cell Ca2+ oscillations
Oscillations of [Ca2+]ER occurred at a constant level of IP3 applied in a bath volume >10 000‐fold in excess of the original intracellular volume, suggesting that oscillatory changes of [IP3] were not required in this system. This is supported by experiments in which [Ca2+]ER oscillations were induced by maximal IP3 in the presence of the competitive IP3R blocker heparin (Ghosh et al., 1988). In contrast to the stimulation of IP3 binding by [Ca2+]c (Pietri et al., 1990), heparin affinity is not affected by Ca2+ (Rouxel et al., 1992), making it a good tool to shift the range of IP3 sensitivity. Heparin addition terminated [Ca2+]ER oscillations induced by submaximal IP3 (Figure 3A). However, in the presence of heparin, [Ca2+]ER oscillations could be observed with micromolar IP3 concentrations that would otherwise cause sustained [Ca2+]ER release (Figure 3B). Cellular formation or breakdown of IP3 is unlikely to contribute significantly under these conditions, because [IP3] is effectively clamped at a high level. These findings are consistent with previous reports in which non‐metabolizable IP3 analogs induced [Ca2+]c oscillations in intact cells (Wakui et al., 1989; DeLisle and Welsh, 1992; Lechleiter and Clapham, 1992). A potential problem with the interpretation of the intact cell experiments is that there may be a contribution from oscillations of IP3 formation secondary to the Ca2+ release triggered by the non‐metabolizable IP3 analog. However, in our permeabilized cell studies, the presence of heparin would greatly reduce the efficacy of any endogenous IP3 formation, which would also be diluted rapidly into the essentially infinite sink of extracellular medium containing high levels of exogenous IP3.
One potential mechanism by which [Ca2+]c oscillations could occur at a constant IP3 level is through feedback regulation of the IP3R channel by [Ca2+]ER (Missiaen et al., 1991, 1992; Nunn and Taylor, 1992; Tanimura and Turner, 1996a). In this model, the cycling between open and closed states is dependent on both Ca2+ release and re‐uptake, such that inhibition of the ER Ca2+ pump with thapsigargin would be expected to terminate the oscillations of [Ca2+]ER. When thapsigargin was added either during IP3‐induced [Ca2+]ER oscillations (Figure 3C) or together with IP3 (Figure 3D), the re‐uptake phase of the [Ca2+]ER spikes was completely prevented. However, Ca2+ release still occurred in a periodic manner and, as a result, [Ca2+]ER declined in a series of discrete steps. The Mn2+ quench approach (see Figure 1D) also showed the same cycling between the high and low permeability states of the IP3R when thapsigargin was added shortly before IP3 under these conditions (data not shown). Thus, feedback regulation by [Ca2+]ER is not an essential component of the Ca2+ oscillation mechanism.
The role of [Ca2+]c in the permeabilized hepatocyte system was investigated using 10 mM BAPTA to buffer the medium Ca2+ ([Ca2+]o). Under these conditions, the bell‐shaped dependence on [Ca2+]c (Iino, 1990; Bezprozvanny et al., 1991; Finch et al., 1991) was clearly apparent at submaximal IP3, with properties similar to those described previously for hepatocytes permeabilized in suspension (Marshall and Taylor, 1993). For example, at 250 nM IP3, the rate of [Ca2+]ER release was stimulated 15 ± 6‐fold when [Ca2+]o was increased from <5 to 500 nM, whereas Ca2+ release rates decreased with higher [Ca2+]o and were barely detectable at >5 μM [Ca2+]o (not shown). IP3 released Ca2+ throughout the range of 1–2000 nM [Ca2+]o, but [Ca2+]ER oscillations were never observed in the presence of the Ca2+ buffer (e.g. Figure 4A), indicating that fluctuations in [Ca2+]o are necessary for [Ca2+]ER oscillations. Although the Ca2+ buffer prevented oscillations, Ca2+ release induced by submaximal IP3 was still transient at [Ca2+]o levels >200 nM (Figure 4A). Since [Ca2+]o was highly buffered, Ca2+ re‐uptake reflects refilling of the same Ca2+ store at steady‐state, which implies an inactivation of the IP3R channel sufficient to allow the ER Ca2+ pump to overcome the IP3‐activated release pathway. This was demonstrated directly by addition of thapsigargin at steady‐state, which revealed a Ca2+ release rate >10‐fold slower than the rate when IP3 was added initially or when IP3 and thapsigargin were added simultaneously to naive cells (Figure 4A). By contrast, Ca2+ release rates recovered when IP3 was washed out and then added again (Figure 4B). The high level of BAPTA makes it unlikely that steady‐state inactivation of the IP3R was mediated by released Ca2+ during sustained incubation with IP3. An alternative to Ca2+ feedback inhibition that might contribute to the decline in Ca2+ release under these conditions is the ligand‐induced inactivation of the IP3R by IP3, which occurs in a time‐dependent manner at fixed [Ca2+]o (Hajnóczky and Thomas, 1994).
Role of IP3‐induced inactivation of the IP3R
The time‐dependent inactivation of the IP3R by IP3 was demonstrated originally by measuring IP3R permeability using retrograde Mn2+ flux through the channel to quench luminal fura2 at various times after addition of IP3 (Hajnóczky and Thomas, 1994). In those experiments, suspensions of permeabilized hepatocytes were pre‐incubated with thapsigargin to deplete the Ca2+ stores. However, Combettes et al. (1996) suggested recently that there may have been sufficient residual Ca2+ within the stores under these conditions to sensitize the IP3R to IP3, and that loss of this [Ca2+]ER during the incubation with IP3 could account for the inactivation of the IP3R. Although we observed no Ca2+ release and no change in [Ca2+]ER measured with luminal fura2 in response to IP3 in thapsigargin‐treated cells in our previous studies (Hajnóczky and Thomas, 1994), we have re‐examined this question using both low and high affinity Ca2+ indicator dyes. In the present experiments, suspensions of hepatocytes were permeabilized in the presence or absence of 2 μM thapsigargin with [Ca2+]o buffered to 400 nM with 15 mM BAPTA. When [Ca2+]ER was monitored with luminal fura2FF (Figure 5A), inclusion of thapsigargin from the start of the experiment (lower two traces of Figure 5A) completely blocked Ca2+ uptake and depleted the Ca2+ stores to the point where IP3 was unable to cause any further loss of [Ca2+]ER. Thapsigargin was less effective in depleting the Ca2+ stores when added after completion of ATP‐dependent Ca2+ uptake (upper traces of Figure 5A), but treatment with ionomycin caused a rapid decrease of [Ca2+]ER to the level measured in cells pre‐treated with thapsigargin. Pre‐incubation with thapsigargin was also found to eliminate the Ca2+ release response to IP3 in experiments using luminal fura2, which has the advantage of being sensitive to [Ca2+]ER in the submicromolar range (Figure 5B).
The same fura2‐loaded cell preparation examined in Figure 5B was also used for Mn2+ quench measurements of IP3R permeability (Figure 5B inset). These experiments demonstrate that the initial fast phase of Mn2+ quenching observed when IP3 was added together with the Mn2+ was greatly reduced when the cells were pre‐incubated with IP3 for 30 s. This reflects the time‐dependent inactivation induced by IP3, as reported previously (Hajnóczky and Thomas, 1994). These findings also provide direct evidence that the IP3R remains permeable to Mn2+ even when the stores are completely depleted of Ca2+, in contrast to the findings of Tanimura and Turner (1996a). Another potential problem with the use of Mn2+ to study IP3R permeability is that Mn2+ may displace Ca2+ from other binding sites, which could then contribute to the inhibition of the IP3R (Combettes et al., 1996). However, the presence of 2 mM Mg‐ATP in our experiments provides additional Ca2+ and Mn2+ buffering capacity, which prevents the substantial Ca2+ changes that might otherwise occur on Mn2+ addition (calculated [Ca2+]o increased from 435 to 515 nM after MnCl2 addition in Figure 5B). Moreover, this [Ca2+]o change occurs only at the time of Mn2+ addition, and so cannot explain the time‐dependent decrease in Mn2+ uptake rate during pre‐incubation of the cells with IP3. It should also be noted that Striggow and Ehrlich (1996) have concluded that the free [Mn2+] used in these experiments (2 μM) is close to the optimum for measuring IP3R permeability.
Since the Ca2+ dependence for sensitization of the IP3R to IP3 and for IP3‐dependent inactivation appeared to be similar in hepatocytes (Marshall and Taylor, 1993; Hajnóczky and Thomas, 1994), we hypothesized that these may be coupled events. Therefore, the Mn2+ quench approach was used to compare the effects of [Ca2+]o on IP3R sensitization and IP3‐induced inactivation in suspensions of fura2‐loaded hepatocytes permeabilized in the presence of thapsigargin. The Mn2+ quench evoked by 125 nM IP3 was taken as a measure of IP3R sensitization, and the inhibition of Mn2+ quench evoked by maximal IP3 after a 20 s pre‐pulse with either 125 nM or 7.5 μM IP3 was used to measure IP3‐dependent inactivation (Hajnóczky and Thomas, 1994). In addition, the [Ca2+]o‐induced increase in the proportion of high affinity IP3Rs was measured using a low level of [3H]IP3 (Pietri et al., 1990; Marshall and Taylor, 1994). There was a very marked increase in IP3R channel activation by 125 nM IP3 as [Ca2+]o was increased in the range 300 nM to 1 μM (Figure 6A), and this was paralleled by a dramatic increase in IP3‐induced inactivation (Figure 6B). The [Ca2+]o dependence of activation and inactivation at submaximal IP3 was shifted to higher [Ca2+]o than the sensitization for IP3 binding (Figure 6C), which may result from the cooperative nature of channel activation at submaximal IP3 (Meyer et al., 1990). Although maximal IP3 caused IP3R activation at all levels of [Ca2+]o, the extent of inactivation was entirely dependent on [Ca2+]o and was closely correlated with the [Ca2+]o‐induced increase in high affinity IP3 binding (Figure 6A–C).
The correlation between IP3R sensitization and inactivation was investigated further by substituting Sr2+ and Ba2+ for Ca2+. Marshall and Taylor (1994) have shown that the sensitizing effects of Ca2+ on IP3 binding and Ca2+ release are mimicked by Sr2+ but not by Ba2+, and that neither Sr2+ nor Ba2+ is effective in mimicking the direct inhibitory effect of Ca2+. To eliminate the potential for feedback by released Ca2+, the effects of Sr2+ and Ba2+ on IP3R channel activation were monitored by the IP3‐induced retrograde flux of these ions into fura2‐loaded stores in the presence of thapsigargin and EGTA. Sr2+ caused a marked sensitization to IP3 (50 nM) compared with Ba2+ (Figure 7A), and this can be explained by the differential effects of these ions on IP3 binding under these conditions (Figure 7C, and see Marshall and Taylor, 1994). These effects on IP3 sensitivity were paralleled by IP3‐induced inactivation, such that Sr2+ but not Ba2+ supported the time‐dependent inactivation during IP3 pre‐incubation (Figure 7B). The fact that Ba2+ was without effect shows that Sr2+ was not acting simply by displacing Ca2+ from other binding sites. Thus, the differential effects of Sr2+ and Ba2+ on IP3‐induced inactivation indicate that the Ca2+ dependence of this process reflects binding to the stimulatory Ca2+ site of the IP3R.
Taken together, the data of Figures 6 and 7 provide evidence that only the Ca2+‐ (or Sr2+‐) sensitized form of the IP3R undergoes ligand‐induced inactivation. Moreover, the fact that Sr2+ is effective in supporting IP3‐induced inactivation indicates that this process does not depend on divalent metal ion binding at the inhibitory site of the IP3R. Finally, the observation that inactivation at submaximal [IP3] follows a similar cooperative Ca2+ dependence to activation (Figure 6A and B) but does not parallel the non‐cooperative Ca2+ dependence of IP3 binding suggests that channel opening is a prerequisite for the inactivation process and hence inactivation may be an obligatory consequence of channel activation in the Ca2+‐sensitized state.
Agonist‐induced oscillations of [Sr2+]c in intact hepatocytes
The data presented above and previous studies comparing the effects of divalent metal ions on IP3R function suggested that Sr2+ could be used to distinguish the roles of IP3‐dependent inactivation of the IP3R and direct feedback inhibition by Ca2+ in the generation of [Ca2+]c oscillations in intact cells. It has been shown that Sr2+ can be accumulated by ATP‐dependent intracellular Ca2+ stores and subsequently released in response to agonist (Montero et al., 1995). In order to eliminate possible contributions to IP3R regulation from residual Ca2+ in the stores (Morgan and Jacob, 1996), hepatocytes were incubated in the presence of EGTA and treated with high vasopressin and the reversible SERCA Ca2+ pump inhibitor cyclopiazonic acid (Figure 8). After washout of these agents, there was no further [Ca2+]c response to vasopressin. Following a further washout period, the cells were exposed to SrCl2, BaCl2 or CaCl2, each of which rapidly appeared in the cytosol, presumably as a result of the activated capacitative Ca2+ entry pathway. The cells were allowed to re‐load with the added divalent cation and, after a steady‐state was achieved, they were challenged again with vasopressin. Vasopressin‐induced oscillations of fura2 fluorescence were observed in the cells re‐loaded with Sr2+ and Ca2+, but not in those loaded with Ba2+. Although Ba2+ did not support oscillations, it did not prevent the induction of oscillations by vasopressin when the medium was supplemented with SrCl2 in the continuing presence of BaCl2.
Loading of hepatocytes with fura2 acetoxymethyl ester (fura2/AM) results in partial compartmentalization of the dye in the ER. This luminal fura2 does not usually contribute to the measured [Ca2+]c changes in intact cells, because [Ca2+]ER remains sufficiently high to saturate the dye unless the cells are treated with agonist in the presence of SERCA pump inhibitors (Glennon et al., 1992). However, the affinity of fura2 for Sr2+ is 30‐fold lower than for Ca2+ and, as a result, the oscillations of fura2 fluorescence recorded in the presence of Sr2+ reflect a mixed signal for the cytosolic and ER compartments. This gave rise to a variety of oscillation patterns, ranging from largely cytosolic signals (Figure 8A, cell #2, and D, cell #1) to those cells in which the luminal changes predominate (Figure 8A, cell #3, and D, cell #2). Microinjected fura2 was used to obtain a pure cytosolic signal (Figure 8B). In these experiments, vasopressin gave rise to baseline spikes in the Sr2+‐loaded cells that propagated throughout the cell and were similar to those observed with Ca2+, except the [Sr2+]c oscillations were 10‐ to 20‐fold smaller in amplitude than the [Ca2+]c oscillations in the same cells. This can be explained by the lower affinity of fura2 for Sr2+, and suggests that the absolute magnitude of [Ca2+]c and [Sr2+]c oscillations are similar in hepatocytes. The Sr2+ oscillations also demonstrated frequency modulation, such that the initial lag time and the interspike period decreased with increasing vasopressin dose (Figure 8C).
The fact that the cells were incubated in the presence of EGTA and depleted of Ca2+ to the point where there was no detectable Ca2+ release to vasopressin should ensure that the oscillations in Sr2+‐loaded cells are due predominantly to Sr2+ fluxes. A number of other lines of evidence support the conclusion that these are Sr2+ oscillations and that they are driven directly by Sr2+ feedback effects rather than as a secondary consequence of residual Ca2+ fluxes. The observation of oscillatory decreases in [Sr2+]ER in fura2/AM‐loaded cells indicates that [Ca2+]ER must have been reduced to the submicromolar range where it was no longer able to saturate the luminal fura2. Furthermore, Sr2+‐dependent oscillations continued for >30 min, often with little change in amplitude through many cycles. This repetitive cycling would be expected to chase out any residual Ca2+ that could play a role in feedback regulation at the IP3R. Finally, the small amplitudes of the oscillations measured with microinjected cytosolic fura2 are consistent with [Sr2+]c spikes, whereas if these reflected [Ca2+]c spikes they would be inadequate to elicit the feedback activation of the IP3R necessary to propagate the release throughout the cell. The small amplitude of these spikes also shows that [Sr2+]c does not achieve the near millimolar concentrations where it might act at the inhibitory Ca2+ site of the IP3R (Marshall and Taylor, 1994). We were unable to determine whether Sr2+ alone could support IP3‐induced oscillations in our permeabilized cell system, because we cannot use chelators in this preparation, and the contaminating Ca2+ and effectively infinite volume of the incubation medium are sufficient to allow substantial Ca2+ loading of the intracellular stores. Nevertheless, the observation of [Sr2+]c oscillations in intact cell experiments and the demonstration that Sr2+ mimics the effects of Ca2+ in sensitizing the IP3R and supporting IP3‐dependent inactivation in permeabilized cells provides strong evidence that Sr2+ is able to substitute effectively for Ca2+ in driving the basic oscillation mechanism. Moreover, the inability of Sr2+ to substitute for Ca2+ at the inhibitory binding site of the IP3R suggests that this form of negative feedback control by Ca2+ is not essential to obtain [Ca2+]c oscillations.
Our findings with permeabilized hepatocytes demonstrate that feedback regulation of the IP3R by [Ca2+]c at a constant level of IP3 represents the minimum requirement for oscillatory Ca2+ release. In addition, it appears that the stimulatory Ca2+‐binding site of the IP3R can effect both activation and termination of Ca2+ release, with the latter process occurring through the intrinsic slow inactivation that follows IP3‐induced activation of the Ca2+‐sensitized state of the IP3R. This does not exclude an additional contribution from direct negative feedback by released Ca2+ at the inhibitory Ca2+‐binding site. It is also difficult to formally exclude the possibility that IP3 binding to the Ca2+‐sensitized IP3R leads to a change in the properties of the inhibitory Ca2+‐binding site that increases its affinity for Ca2+, and perhaps Sr2+. It is possible that other regulatory mechanisms may also contribute to [Ca2+]c oscillations in intact cells, including regulation of the IP3R by [Ca2+]ER, [Ca2+]c stimulation of IP3 formation and enhanced plasma membrane Ca2+ entry. However, our findings in the permeabilized cell preparation demonstrate that these are not essential components of the [Ca2+]c oscillator.
Our studies suggest a basic mechanism of [Ca2+]c oscillations that depends on [Ca2+]c‐dependent interconversion between two modes of IP3R channel activation, a Ca2+‐free basal state that requires high levels of IP3 for activation and a Ca2+‐sensitized state that can be activated at much lower levels of IP3 but undergoes an intrinsic inactivation when IP3 is bound. The Ca2+‐ (and Sr2+‐) dependent interconversions of the IP3R are shown in Figure 9, with the postulated predominant pathway underlying a [Ca2+]c spike shown by the thick blue arrows. At the resting [Ca2+]c between [Ca2+]c spikes, IP3 affinity is low and the IP3R channel does not inactivate, so that submaximal levels of IP3 cause continuous low level Ca2+ release. As [Ca2+]c rises, IP3Rs convert to a conformation with high affinity for IP3, which accelerates Ca2+ release by these channels resulting in positive feedback by [Ca2+]c that effectively recruits all available IP3Rs to the high affinity activated conformation. A key observation of the present study is that there is an obligatory coupling between channel opening and inactivation in the Ca2+‐sensitized state of the IP3R. Thus, in the high affinity conformation, IP3R activation occurs in a phasic manner, whereby channel opening is followed by a time‐dependent inactivation that does not require further Ca2+ binding. An obligatory linkage between the activation and inactivation of the ryanodine receptor by depolarization in skeletal muscle has also been reported (Pizarro et al., 1996), suggesting that this may be a common property of intracellular Ca2+ release channels. This process is ideally suited to generate a stable transient increment of Ca2+ release during each [Ca2+]c spike. In the final phase of the [Ca2+]c oscillation cycle, the intrinsic inactivation, perhaps in combination with direct feedback inhibition by [Ca2+]c, allows a return to basal [Ca2+]c through the action of Ca2+ pumps. Recovery of the IP3R from IP3‐dependent inactivation can occur either by dissociation of Ca2+ from the stimulatory site or by removal of IP3 (Hajnóczky and Thomas, 1994) but, since Ca2+ regulates IP3 affinity, both Ca2+ and IP3 are expected to dissociate from the IP3R during the recovery phase. Although IP3R inactivation reverses more slowly than the direct inhibitory effect of Ca2+ (Finch et al., 1991; Ilyin and Parker, 1994; but cf. Oancea and Meyer, 1996), it is not slow enough to account for the long interspike periods. Therefore, other factors may be involved in resetting the system prior to the next [Ca2+]c spike or, alternatively, the primary determinant of oscillation frequency may be the time required to generate a sufficient Ca2+ trigger signal to initiate the next Ca2+ release spike. Overall, the coupled feedback regulation of the IP3R by Ca2+ and IP3 is likely to play a key role in ensuring that the amplitude and duration of each [Ca2+]c spike is constant over a range of IP3 and agonist doses, to yield an essentially pure frequency‐modulated [Ca2+]c signal.
Materials and methods
Imaging measurements in intact and permeabilized hepatocytes
Hepatocytes were isolated from the livers of Sprague–Dawley rats by collagenase perfusion and maintained in primary culture for 3–24 h in Williams E medium, as described previously (Rooney et al., 1989; Hajnóczky et al., 1993), except that dexamethasone was omitted.
For measurements of [Ca2+], [Sr2+] and [Ba2+] in intact hepatocytes, the cells were loaded with 5 μM fura2/AM for 15 min in the presence of 100 μM sulfinpyrazone or were microinjected with fura2 free acid as described previously (Rooney et al., 1989; Lin et al., 1994). Measurements of Mn2+ quench of compartmentalized fura2 utilized cells loaded with 5 μM fura2/AM for 45–60 min (Renard‐Rooney et al., 1993; Hajnóczky et al., 1994). Fura2 and other Ca2+ indicators were loaded into intact hepatocytes incubated at 37°C in extracellular medium (ECM) composed of 121 mM NaCl, 5 mM NaHCO3, 10 mM Na‐HEPES, 4.7 mM KCl, 1.2 mM KH2PO4, 1.2 mM MgSO4, 2 mM CaCl2, 10 mM glucose and 2% bovine serum albumin (BSA), pH 7.4. Intact cell experiments were carried out in the same buffer with BSA reduced to 0.25%, and CaCl2 was omitted for measurements of Mn2+ quench and intracellular [Sr2+] and [Ba2+]. The free [Ca2+] of this Ca2+‐free ECM was 400 nM measured using fura2 free acid (1.5 μM).
Measurements of [Ca2+]ER in permeabilized hepatocytes were carried out by first loading the intact cells for 60–120 min with 6 μM furaptra/AM or 6 μM fura2FF/AM. These Ca2+ indicators have Kd values of 53 μM for furaptra (Hofer and Machen, 1994) and 35 μM for fura2FF (A.Minta, TEFLABS), making them suitable for measuring changes in [Ca2+]ER. Dye‐loaded hepatocytes were washed with Ca2+‐free buffer and permeabilized by incubation for 6 min with 15 μg/ml digitonin in intracellular medium (ICM) composed of 120 mM KCl, 10 mM NaCl, 1 mM KH2PO4, 20 mM Tris‐HEPES at pH 7.2 with 2 mM MgATP and 1 μg/ml each of antipain, leupeptin and pepstatin. In order to decrease [Ca2+]o, the ICM was passed through a Chelex column prior to addition of MgATP and protease inhibitors. Medium free [Ca2+] was <100 nM after Chelex treatment and did not exceed 300–400 nM after addition of ATP and protease inhibitors. Direct measurement of [Ca2+]o using 250 nM fura2 in the imaging chamber in the presence of permeabilized hepatocytes yielded a free [Ca2+] of ∼400 nM. In some experiments 2 μM CPT‐cAMP was added to facilitate IP3R activation and Ca2+ re‐uptake (Hajnóczky et al., 1993), since this appeared to increase the percentage of responsive cells, but all findings were reproduced in the absence of this agent. After permeabilization, the cells were washed into fresh buffer without digitonin. There was no detectable metabolism of IP3 in the permeabilized cell preparation.
Individual cells were examined by digital imaging fluorescence microscopy at 35°C (Rooney et al., 1989, 1990; Renard‐Rooney et al., 1993; Hajnóczky et al., 1993, 1995). [Ca2+]c in intact cells was calculated from the fluorescence ratio derived from image pairs obtained using 340 and 380 nm excitation. Mn2+ quench of compartmentalized fura2 fluorescence was measured using the Ca2+‐insensitive excitation wavelength of 360 nm (Glennon et al., 1992; Hajnóczky et al., 1993, 1994). Calibration of fura2FF signals in permeabilized hepatocytes gave values of ∼500 μM for [Ca2+]ER after completion of ATP‐dependent Ca2+ uptake. Experiments were carried out with at least three different cell preparations, and 30–50 cells were monitored in each experiment. Submaximal IP3 evoked [Ca2+]ER oscillations in 50–60% of cells and stepwise Mn2+ quench of compartmentalized fura2 in 25–30% of cells. Traces represent single cell responses unless indicated otherwise.
Fluorometric measurements of ion fluxes in suspensions of permeabilized hepatocytes
Suspensions of fura2‐loaded hepatocytes were permeabilized with 25 μg/ml digitonin in the presence of 2 μM thapsigargin and 1 μM ruthenium red for 10 min at 37°C in a fluorometer cuvette (Deltascan, PTI), as described previously (Hajnóczky et al., 1993, 1994; Renard‐Rooney et al., 1993; Hajnóczky and Thomas, 1994). For experiments where [Ca2+]o was varied, EGTA (5–23 μM) and CaCl2 (8 μM) were included during the pre‐incubation. Actual [Ca2+]o values were measured with the small amount of fura2 (∼0.2 μM final) released from the cells. MnCl2 (usually 60 μM) was added together with EGTA to give a constant level of Ca2+ and Mn2+ during the Mn2+ uptake phase for all [Ca2+]o pre‐incubation conditions. The quench of luminal fura2 by Mn2+ was monitored with 360 nm excitation. Dual wavelength excitation (340/380 nm) of compartmentalized fura2 fluorescence was used to monitor the entry of Sr2+ and Ba2+ into the stores, whereas single wavelength excitation at the appropriate isofluorescence wavelengths (365 nm for Sr2+ and 370 nm for Ba2+) was used to monitor Mn2+ quench in the presence of these divalent cations. Although it might be expected that Sr2+ or Ba2+ in the stores would interfere with the Mn2+ quenching of fura2, this effect was found to be negligible both by calculation of the expected binding of Mn2+ in the presence and absence of these ions, and by direct measurements with fura2 in vitro. For example, addition of 0.8 μM MnCl2 to fura2 (nominally 1.5 μM) quenched the fluorescence by 31.1 and 29.3% in the presence and absence of 60 μM SrCl2, respectively. It should also be noted that the basal leak of Sr2+ and Ba2+ into the stores was sufficient to approach equilibration during the 90 s pre‐incubation period, even in the absence of IP3 (see Figure 7).
[3H]IP3 binding measurements
Hepatocytes were permeabilized at 2 mg of protein/ml in ICM. All studies with Sr2+ and Ba2+ and the Ca2+ dependence of IP3 binding were carried out in the presence of 40 μM EGTA and at 22°C. Incubations with [3H]IP3 (1.4 nM for 90 s) were carried out with [Ca2+]o values set by addition of CaCl2 (0–45 μM) or in the presence of 160 μM Sr2+ or Ba2+. The bound and free fractions were separated by filtration (GF/B filters, transit time 2–3 s). Non‐specific binding determined in the presence of 10 μM unlabeled IP3 was <5% of total binding. Specific binding was normalized to maximum binding attained in the presence of Ca2+ (14.8 ± 1.5 fmol/mg cell protein). The time course of increased IP3 binding induced by Sr2+ was measured after 4 min pre‐incubation with 50 nM [3H]IP3.
This work was supported by grants DK38422 and DK51526 from the NIH. G.H. is a recipient of a Burroughs Wellcome Fund Career Award in the Biomedical Sciences.
- Copyright © 1997 European Molecular Biology Organization