Dynamics of [Ca2+] in the Endoplasmic Reticulum and Cytoplasm of Intact HeLa Cells

We have measured the [Ca2+] in the endoplasmic reticulum ([Ca2+]er) of intact HeLa cells at both 22 °C and 37 °C using endoplamsic reticulum-targeted, low Ca2+ affinity aequorin reconstituted with coelenterazine n. Aequorin consumption was much slower at 22 °C, and this allowed performing a much longer study of the dynamics of [Ca2+]er. The steady-state [Ca2+]er (500–600 μm) was not modified by the temperature, although both the rates of pumping and leak were decreased at 22 °C. The behavior of both [Ca2+]er and cytoplasmic [Ca2+] ([Ca2+]c) after the addition of increasing concentrations of agonists and/or Ca2+-ATPase inhibitors, or following incubation in Ca2+-free medium were compared. We show that agonists induce a fast but relatively small decrease in [Ca2+]er, which is enough to produce a sharp increase in [Ca2+]c. Termination of Ca2+ release is controlled by feedback inhibition of the inositol 1,4,5-trisphosphate receptors by [Ca2+]c, a mechanism that appears to be designed to release the minimum amount of Ca2+ necessary to produced the required [Ca2+]c signal. We also show that Ca2+ release is inhibited progressively when [Ca2+]er decreases below a threshold of about 150 μm, even in the absence of Ca2+ pumping or [Ca2+]c increase. This effect is consistent with a regulation of the inositol 1,4,5-trisphosphate-gated channels by [Ca2+]er.

Monitoring directly the [Ca 2ϩ ] inside the main Ca 2ϩ store of the cell, the endoplasmic reticulum, has been difficult to achieve in intact cells. We have reported recently (1) that using HeLa cells expressing ER 1 -targeted, low Ca 2ϩ affinity aequorin reconstituted with coelenterazine n allows measuring reliably [Ca 2ϩ ] er in intact cells. In that study, however, the high Ca 2ϩ levels reached at steady state in the ER (500 -600 M) led to a fast consumption of aequorin, allowing measurement of [Ca 2ϩ ] er for only a few minutes. We have now observed that aequorin consumption is 1 order of magnitude slower at 22°C, and this allows performing long term comparative studies of the dynamics of [Ca 2ϩ ] er and [Ca 2ϩ ] c under different conditions, e.g. in the presence of extracellular InsP 3 -producing agonists, Ca 2ϩ -ATPase inhibitors, or during incubation in Ca 2ϩfree medium. Our results reveal several unexpected characteristics of the Ca 2ϩ release phenomenon, e.g. that the magnitude of the changes of [Ca 2ϩ ] er after agonist action does not always correlate with the magnitude of the corresponding changes in [Ca 2ϩ ] c . A sharp agonist-induced cytosolic Ca 2ϩ peak can be obtained with only little Ca 2ϩ release from the stores, and further release is feedback inhibited by the increase in [Ca 2ϩ ] c , probably through the generation of microdomains of high [Ca 2ϩ ].

EXPERIMENTAL PROCEDURES
Calibration-Cell lysates of the EM26 cell clone (2) stably expressing the recombinant photoprotein were reconstituted overnight at 4°C with 5 M coelenterazine n, and incubated with solutions containing 110 mM KCl, 10 mM NaCl, 1 mM free Mg 2ϩ , 40 mM HEPES, pH 7.0, at 22°C, and known concentrations of Ca 2ϩ prepared using buffers with 5 mM HEDTA when necessary. The fractional rate of aequorin consumption (luminescence/total luminescence remaining, L/L max ) at steady state was calculated in every case after consuming all aequorin luminescence with 10 mM Ca 2ϩ . The procedures for fitting the curve to the experimental data and other details have been described previously (3).
[Ca 2ϩ ] Measurements-The HeLa cell clone EM26 producing ERtargeted low Ca 2ϩ affinity mutated aequorin (2) was grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and 0.2 mg/ml G418. Cell clones were plated onto 13-mm round coverslips. Before reconstituting aequorin, [Ca 2ϩ ] er was reduced by incubating the cells for 5 min at 37°C with the SERCA inhibitor BHQ (10 M) in KRB (Krebs-Ringer modified buffer: 125 mM NaCl, 5 mM KCl, 1 mM Na 3 PO 4 , 1 mM MgCl 2 , 5.5 mM glucose, 20 mM HEPES, pH 7.4), supplemented with 3 mM EGTA. Cells were then incubated for 1 h at room temperature in KRB containing 0.5 mM EGTA, 10 M BHQ, and either 0.5 M coelenterazine n (for measuring [Ca 2ϩ ] er ) or 4 M fura-2-AM (for measuring [Ca 2ϩ ] c ). The coverslip was then washed for 5 min in KRB containing 0.5 mM EGTA, 5% bovine serum albumin, and 10 M BHQ, and finally placed (for aequorin measurements) in the perfusion chamber of a purpose-built thermostatized luminometer. [Ca 2ϩ ] er values were calculated from the luminescence records using a computer algorithm (3) which follows the curve shown in Fig. 1A. [Ca 2ϩ ] c measurement on cell populations was performed using a Cairn spectrophotometer equipped with a six-filter rotating wheel as described previously (4). Cells used for [Ca 2ϩ ] c measurements were always depleted of Ca 2ϩ in the same way as those used for [Ca 2ϩ ] er measurements, to allow comparison of both types of data in the same conditions. Coelenterazine n, fura-2-AM, and BAPTA-AM were obtained from Molecular Probes. Other reagents were from Sigma (Madrid, Spain) or Merck (Darmstadt, Germany). . 1A shows the calibration curve obtained at 22°C of low Ca 2ϩ affinity mutated aequorin reconstituted with coelenterazine n, a prosthetic group conferring still lower Ca 2ϩ affinity (5). The curve obtained at 37°C, reported previously (1), is also shown for comparison. We can observe that decreasing the temperature reduces the rate of aequorin consumption by nearly 1 order of magnitude at every [Ca 2ϩ ]. This effect allows measuring [Ca 2ϩ ] in the millimolar range for quite a long time (Ͼ15 min) without problems of aequorin consumption. Fig. 1B shows the effect of the addition of 1 mM Ca 2ϩ to Ca 2ϩ -depleted HeLa cells. The luminescence record shows a Ca 2ϩ -triggered luminescence peak that contains most of the aequorin luminescence. Later addition of digitonin produces only a minor release of residual luminescence. The calibrated signal shows that addition of extracellular Ca 2ϩ produces an increase in [Ca 2ϩ ] er which reaches a steady-state level around 600 M (590 Ϯ 90 M, mean Ϯ S.D., n ϭ 44) within 8 -10 min of Ca 2ϩ addition. The rate of increase of [Ca 2ϩ ] er after extracellular Ca 2ϩ addition was 1.8 Ϯ 0.5 M/s (n ϭ 13), nearly 4 times slower than that observed at 37°C ( ] er depends only on the balance between Ca 2ϩ pumping and Ca 2ϩ leak. In fact, the rate of Ca 2ϩ leak from the ER, measured as the rate of [Ca 2ϩ ] er decrease after inhibition of the SERCA with BHQ or after incubation in Ca 2ϩfree medium, increased 2-3-fold within that temperature interval (see below).

Fig
In Fig. 2 The potentiating effect of BAPTA loading could be attributed either to the inhibition of Ca 2ϩ pumping or to a direct effect activating InsP 3 -gated Ca 2ϩ release. In our previous paper (1), we showed that simultaneous addition of 100 M histamine and the SERCA inhibitor BHQ produced only additive effects, suggesting that the main effect of BAPTA was protecting the InsP 3 receptors from inhibition by the increase in [Ca 2ϩ ] c . We have made now a more complete study on this point, which provides further evidence in favor of this hypothesis. Fig. 3 shows the effect on [Ca 2ϩ ] er of the addition of 10 M BHQ either alone or together with 2.5 M, 10 M, or 100 M histamine, at 22°C. In control cells (upper panel), there was no difference in the rate of emptying induced by BHQ or BHQϩ2.5 M histamine. After the addition of BHQ ϩ 10 M histamine, there was an initial drop in [Ca 2ϩ ] er , but the curve then followed the same slope as with BHQ alone. Finally, when we added BHQ ϩ 100 M histamine, we obtained a greater initial drop in [Ca 2ϩ ] er , but the emptying continued afterward at a rate similar to that obtained in the absence of histamine. The effect of histamine was very different in cells loaded with BAPTA. The lower panel of Fig. 3 shows that increasing concentrations of histamine induced Ca 2ϩ release at an increasing rate, and following a near monoexponential decay. These experiments show clearly that the effect of BAPTA loading is not due to a decreased Ca 2ϩ pumping. Similar data could be obtained at 37°C, although at this temperature histamine was found to be more potent inducing Ca 2ϩ release in control cells at every concentration. Fig.  4 (upper panel) shows that addition of increasing concentrations of histamine induced also here a biphasic decrease in [Ca 2ϩ ] er : a fast initial decrease, proportional to the histamine concentration, followed by a much slower phase of release. This second phase, however, was still somewhat faster here at 37°C than that obtained in the absence of histamine. In any case, loading with BAPTA transformed again the curves to a near monoexponential decay at rates proportional to the histamine concentration (Fig. 4, lower panel).
Careful analysis of the decay curves of [Ca 2ϩ ] er obtained after simultaneous addition of histamine and BHQ to cells loaded with BAPTA allowed us to study the possible regulation of the InsP 3 -gated channels by [Ca 2ϩ ] er . Theoretically, if the ER would behave as a single compartment emptying at constant rate through the InsP 3 receptors, the decay of [Ca 2ϩ ] er in the absence of pumping should follow a single exponential. part of the experimental curve. We can see that the experimental data follow the exponential most of the time, but clearly deviate from it when the [Ca 2ϩ ] er decreases below 150 M, no matter what is the histamine concentration used. This finding, which has been reported previously using the same technique but with Sr 2ϩ as a Ca 2ϩ surrogate (6), provides evidence in favor that InsP 3 receptors require a certain level of [Ca 2ϩ ] er for maximal activity.
We have studied finally the effect of histamine on [Ca 2ϩ ] er in the absence of extracellular Ca 2ϩ , as well as the behavior of [Ca 2ϩ ] er during incubation of the cells in Ca 2ϩ -free medium. Fig. 6 (upper panel) shows that when histamine is added to HeLa cells incubated in Ca 2ϩ -free medium, the initial [Ca 2ϩ ] c peak is similar to that obtained in Ca 2ϩ -containing medium, but the plateau phase is largely (though not completely) removed. If the cells are incubated for 10 or 20 min in Ca 2ϩ -free medium, the height of the peak was only reduced to 73 Ϯ 5% (n ϭ 4; 78% in the figure) and 69 Ϯ 12% (n ϭ 4; 81% in the figure) of the control value, respectively. In all the cases, it is apparent that the return of [Ca 2ϩ ] c to the base line is biphasic, with a final slow phase lasting for 1-2 min, which cannot be attributed to Ca 2ϩ entry and must be due to Ca 2ϩ release. The counterpart in the ER is shown in the lower panel. Addition of histamine in EGTA-containing medium produces a decrease in [Ca 2ϩ ] er , which has three consecutive phases from a kinetic point of view: a first rapid drop of about 60 M, which must be responsible for the peak of [Ca 2ϩ ] c ; a second phase of slower decrease, lasting for 1-2 min and getting 80 -90 M lower (coincident in time with the last part of the [Ca 2ϩ ] c peak); and a third phase, in which the rate of decrease in [Ca 2ϩ ] er approaches that obtained just by incubation in Ca 2ϩ -free medium. The same kinetics was observed when histamine was added after 10 or 20 min of incubation in Ca 2ϩ -free medium. It was surprising, however, to find that there was not a strict correlation between the height of the [Ca 2ϩ ] c peak and the [Ca 2ϩ ] er at the moment of histamine addition. The [Ca 2ϩ ] er level decreased to 59 Ϯ 6% (n ϭ 4; 60% in the figure) of the initial value after 10 min (when the [Ca 2ϩ ] c peak was 73 Ϯ 5% of maximum) and to 29 Ϯ 8% (n ϭ 5; 30% in the figure) after 20 min (when the [Ca 2ϩ ] c peak was still 69 Ϯ 12% of maximum). In these experiments, the half-time for emptying of the stores in Ca 2ϩ -free medium was 12 Ϯ 3 min (n ϭ 10).
Similar effects were observed when the experiments were carried out at 37°C, but histamine was in this case more potent, inducing Ca 2ϩ release. Fig. 7 shows that addition of histamine in EGTA-containing medium produced a sharp right). The same kinetics were observed if histamine was added after 3 or 5 min of incubation in Ca 2ϩ -free medium. As in the experiments at 22°C, the initial drop was responsible for the sharp cytosolic peak and the second slower phase was coincident with the last part of the [Ca 2ϩ ] c peak. Again, the height of the [Ca 2ϩ ] c peak did not correlate with the [Ca 2ϩ ] er level, which decreased to 46 Ϯ 10% (n ϭ 4; 48% in the figure) after 3 min in Ca 2ϩ -free medium (the [Ca 2ϩ ] c peak decreased only to 92 Ϯ 6%) and to 35 Ϯ 10% (n ϭ 5; 34% in the figure) after 5 min (the [Ca 2ϩ ] c peak decreased only to 82 Ϯ 17%). The half-time for emptying of the stores in Ca 2ϩ -free medium in these experiments was about 3 min (170 Ϯ 50 s, n ϭ 8). DISCUSSION We have measured directly the dynamics of [Ca 2ϩ ] in the endoplasmic reticulum of intact HeLa cells. Using ER-targeted, low Ca 2ϩ affinity mutated aequorin reconstituted with coelenterazine n, and because of the decrease in the rate of aequorin consumption at room temperature, we have been able to perform long-lasting experiments comparing the effects both in [Ca 2ϩ ] er and [Ca 2ϩ ] c of the addition of extracellular agonists and/or SERCA inhibitors, both in Ca 2ϩ -containing or Ca 2ϩ -free medium. The results obtained, some of them quite unexpected, provide a new view on the role of the intracellular Ca 2ϩ stores in Ca 2ϩ homeostasis.
When using aequorin, the possibility of artifacts due to heterogeneity in [Ca 2ϩ ] or behavior of the pools containing the indicator should be always considered. Native aequorin had a extremely nonlinear sensitivity to Ca 2ϩ , and the signal could be therefore much affected by the presence of small compartments of high [Ca 2ϩ ]. In the mutated aequorin we have used in this paper, the near abolition of one of the Ca 2ϩ -binding sites reduced considerably the slope of the calibration curve, which became near linear with [Ca 2ϩ ] (see Fig. 1 and Ref. 2). Using mutated aequorin, therefore, the possibility of having artifacts due to small compartments with high [Ca 2ϩ ] is much lower. On the other hand, because of consumption, aequorin is very well suited to detect heterogeneity in [Ca 2ϩ ]. We have shown recently using theoretical models (6) that, in experiments such as that of Fig. 1B, we should obtain a peak after Ca 2ϩ addition in the calibrated [Ca 2ϩ ] signal if there were a compartment refilling to a much higher [Ca 2ϩ ] than the rest. The reason for this artifact is that the signal would be dominated initially by the high [Ca 2ϩ ] compartment until aequorin consumption in that compartment was near complete. Afterward, the [Ca 2ϩ ] signal would return down to end monitoring the low [Ca 2ϩ ] compartment isolated. Our experimental results show instead a smooth increase in [Ca 2ϩ ], leading to a steady state that keeps flat until near 90% of aequorin has been consumed (see Fig. 1B). Therefore, we can conclude that there are no gross differences in [Ca 2ϩ ] throughout the bulk of the ER.
Increasing the temperature from 22°C to 37°C accelerated nearly 4-fold the rate of increase in [Ca 2ϩ ] er after addition of extracellular Ca 2ϩ . On the other hand, the rate of Ca 2ϩ leak from the stores, measured from the rate of decrease in [Ca 2ϩ ] er after addition of BHQ in 1 mM Ca 2ϩ -containing medium, also increased with the temperature from 2 Ϯ 0.1 (n ϭ 5) M/s at 22°C to 4.4 Ϯ 0.8 (n ϭ 4) M/s at 37°C. Similarly, emptying of the stores after incubation in Ca 2ϩ -free medium was 4-fold slower at 22°C (half-time 12 min) than at 37°C (half-time 3 min). Given that the steady-state [Ca 2ϩ ] er depends on the balance between these Ca 2ϩ -pumping and leak, it is not surprising to find that the steady-state [Ca 2ϩ ] er does not change with the temperature (590 Ϯ 90 M at 22°C, compare with 550 Ϯ 70 M (Ref. 1), at 37°C).
The experiments shown in Fig. 2 demonstrate that a relatively small and fast decrease in [Ca 2ϩ ] er is enough to produce a maximal [Ca 2ϩ ] c peak. A similar conclusion was reached recently observing the decrease in [Sr 2ϩ ] er and the [Sr 2ϩ ] c peak induced by caffeine in skeletal muscle myotubes (7). Therefore, the height of the [Ca 2ϩ ] c peak appears to be related much more with the rate of Ca 2ϩ release than with the actual amount of Ca 2ϩ released. In fact, 2.5 M histamine produced an almost undetectable decrease in [Ca 2ϩ ] er and a near half-maximum peak in [Ca 2ϩ ] c . Additionally, it is also surprising to see from these experiments that the plateau of elevated [Ca 2ϩ ] c following the main peak coincides with a constant [Ca 2ϩ ] er level, still in the presence of histamine. This phenomenon has two main implications that deserve further discussion. First, the constant [Ca 2ϩ ] er level must be a consequence of the balance between increased Ca 2ϩ -pumping (following the increase in [Ca 2ϩ ] c ) and reduced InsP 3 -activated Ca 2ϩ release. Second, the plateau of [Ca 2ϩ ] c must be due to an increased Ca 2ϩ entry from the extracellular medium due to activation of the store-operated Ca 2ϩ channels under these conditions.
Regarding the first conclusion, clearly activation of Ca 2ϩ release through the InsP 3 -gated channels at maximal rate could never be balanced by the activity of the SERCA, given that ion flux through channels is several orders of magnitude faster than the maximal pumping activity of Ca 2ϩ ATPases. Partial emptying is also not due to heterogeneous distribution of the InsP 3 receptors, because in BAPTA-loaded cells histamine induces an almost complete emptying (see also Ref. 1). The only consistent explanation of these results is therefore that the InsP 3 -gated channels become strongly inhibited after the initial histamine-induced Ca 2ϩ release. The experiments shown in Figs. 3 and 4, where histamine-induced Ca 2ϩ release is observed in the absence of Ca 2ϩ pumping, indicate that the inhibition of the InsP 3 receptors is quite strong at 37°C and near 100% at 22°C. The stronger inhibition at low temperature is probably due to the reduced production of InsP 3 (8). As has been discussed already in Ref. 1, the mechanism for the inhibition most probably depends on the generation of microdomains of relatively high [Ca 2ϩ ] in the vicinity of the InsP 3 receptors, which would inhibit Ca 2ϩ release according to the known bell-shaped dependence of these channels with [Ca 2ϩ ] c (9 -12). This mechanism seems to be perfectly designed from a physiological point of view to produce maximal [Ca 2ϩ ] c peaks with minimum Ca 2ϩ release from the stores. In this way, the energetic cost is considerably reduced and, even more important, the cell is able to respond consecutively to several different stimuli because the Ca 2ϩ store is maintained nearly full most of the time.
Regarding the second point, stimulation of Ca 2ϩ entry by histamine in HeLa cells is believed to occur through the capacitative pathway (SOCC, store-operated calcium channels; see Refs. 13 and 14), a Ca 2ϩ pathway that is activated through an unknown mechanism following the emptying of the intracellular Ca 2ϩ stores. The activation of this pathway has been shown in several cell types to be proportional to the degree of emptying of the stores (15,16). However, if the activation of Ca 2ϩ entry by histamine was just a consequence of the emptying of the stores we observe, it is difficult to imagine a mechanism with such an exquisite sensitivity to detect the minute changes in the bulk [Ca 2ϩ ] er (Ͻ10%) induced by the different histamine concentrations. Alternative explanations could be: (i) selective emptying by the agonist of a store with specific functions in the activation of SOCC, although confocal microscopy studies have shown no specific sites for histamine-induced Ca 2ϩ release in HeLa cells (17); (ii) direct activation of SOCC by the agonist through a mechanism either independent or cooperative with the emptying of the stores (18 -20). Further study will be nec-essary to determine the correlation between the activation of SOCC and the actual level of [Ca 2ϩ ] er .
The results shown in Fig. 5 suggest also that Ca 2ϩ release through the InsP 3 -gated channels may be regulated by [Ca 2ϩ ] er . Ca 2ϩ release follows a single exponential while [Ca 2ϩ ] er is above 150 M, but slows down progressively when [Ca 2ϩ ] er gets below that. An alternative interpretation for this phenomenon could only be found in terms of heterogeneity of the rate of Ca 2ϩ release through the InsP 3 receptors between different pools of the ER, assuming that there was a small portion of the ER having a rate of Ca 2ϩ release several times slower than the rest. We have previously shown the same phenomenon using Sr 2ϩ as a Ca 2ϩ surrogate (6), although in that case the [Sr 2ϩ ] er required to activate maximally Sr 2ϩ release was higher, around 500 M. Regulation by [Ca 2ϩ ] er of the sensitivity of the InsP 3 receptors has been reported previously (21,22), but some other groups failed to detect this effect (23)(24)(25)(26)  This threshold was independent of the histamine concentration and therefore of the rate of Ca 2ϩ release, so excluding alternative explanations such as the build-up of a diffusion membrane potential across the ER membrane. Our results cannot be interpreted either in terms of an effect of luminal Ca 2ϩ acting at the cytosolic inhibitory site (28), because the same effect was observed (6) using cells loaded with Sr 2ϩ , which does not bind to the inhibitory site (1,29,30). Regarding the physiological significance of this mechanism, it is important to note that the stimulatory effect of Ca 2ϩ saturates at [Ca 2ϩ ] er values around 150 M, that is 30% of its steady-state value. This means that inhibition of Ca 2ϩ release by this mechanism will probably take place only when cells are strongly stimulated. In those conditions, this mechanism may be designed as a negative feedback able to stop cell activation, avoiding complete depletion of Ca 2ϩ of the ER.
Finally, another initially unexpected result from this study was the relative lack of correlation between the height of the histamine-induced [Ca 2ϩ ] c peaks and the [Ca 2ϩ ] er level during incubations in Ca 2ϩ -free medium. The height and the shape of the [Ca 2ϩ ] c peak were only slightly modified even when the [Ca 2ϩ ] er had been reduced to about 30% of the initial value (see Figs. 6 and 7). This result, however, could be expected from the data shown in Fig. 2. Given that a decrease in [Ca 2ϩ ] er of 50 -100 M is enough to produce a maximum [Ca 2ϩ ] c peak and that the affinity for Ca 2ϩ of the most abundant Ca 2ϩ -binding proteins in the ER is in the millimolar range (31), we can predict that a similar amount of Ca 2ϩ should be released after a fast [Ca 2ϩ ] er decrease from 500 M to 450 M or from 200 M to 150 M. In fact, the inhibition of the InsP 3 receptors by [Ca 2ϩ ] c limits the total amount of Ca 2ϩ released after histamine action, and therefore the effect of the agonist on [Ca 2ϩ ] c becomes nearly independent of the [Ca 2ϩ ] er , within a certain range of levels (below 150 M, Ca 2ϩ release starts to be inhibited, see above). This may also be important from a physiological point of view, because this mechanism allows cell responses to InsP 3 -producing extracellular agonists to be nearly independent of the Ca 2ϩ content of the stores. On the other hand, these results suggest that the content of the Ca 2ϩ stores cannot be adequately estimated from the height of the peaks of [Ca 2ϩ ] c induced by an agonist. Direct measurement of [Ca 2ϩ ] er appears now essential and will surely throw new light in the next few years on many long-known phenomena related to Ca 2ϩ homeostasis.