Differential regulation of nuclear and cytosolic Ca2+ in HeLa cells.

The results reported in this study address the controversial issue that nuclear free Ca2+ ([Ca2+]n) may be regulated independently of cytosolic free Ca2+ ([Ca2+]c). We have measured [Ca2+]n and [Ca2+]c with recombinant aequorin targeted to the nucleus and cytosol in HeLa cells. We found that histamine, ATP, and ionomycin increased [Ca2+]c quantitatively more than [Ca2+]n, although the time course of these changes was similar. The difference between [Ca2+]c and [Ca2+]n depended on the stimulus, and the relative difference between [Ca2+]n and [Ca2+]c was less with ionomycin than with histamine or ATP. After depletion of the internal Ca2+ store, restoration of extracellular Ca2+ resulted in only increased [Ca2+]c without a significant increase in [Ca2+]n. Treatment with cyclopiazonic acid resulted in a delayed increases in [Ca2+]n compared to [Ca2+]c. These differences in both timing and magnitude of nuclear Ca2+ signals confirm that the cell can limit or delay increases in nuclear free Ca2+. Taken with the fact that an inositol phosphate signaling system resides in the nucleus and its envelope, our data support the hypothesis that [Ca2+]n may be independently regulated.

Cytosolic Ca 2ϩ is a pivotal regulator of many cytosolic functions (1,2). However, the role of Ca 2ϩ in the nucleus is more controversial. The involvement of nuclear Ca 2ϩ in processes such as mitosis, apoptosis, and gene transcription (reviewed in Ref. 3) is suggested by the presence of inositol 1,4,5-trisphosphate (InsP 3 ) 1 receptors and inositol 1,3,4,5-tetrakisphosphate receptors on the nuclear membrane (4,5) and the presence in the nucleus of an inositol lipid cycle (6) and the Ca 2ϩ binding proteins, calmodulin, calreticulin and calpain (7).
It is widely assumed that the nuclear pore forms a large channel for the free diffusion of ions and macromolecules (8). However, electrophysiological data from patch clamping the nuclear envelope revealed that the nuclear pore is not freely permeable to all ions (9,10) and that transport across the pore may be regulated by the perinuclear Ca 2ϩ store (11) and/or ATP (12). In addition, there was no diffusion of fura-2 (13) or InsP 3 (14) into the cytoplasm after they had been injected into oocyte nuclei. The combined evidence from the literature therefore strongly suggests that the nucleus is capable of independently regulating the transport of ions such as Ca 2ϩ .
Previous reports comparing [Ca 2ϩ ] n and [Ca 2ϩ ] c using Ca 2ϩsensitive fluorescent dyes have been divided on whether [Ca 2ϩ ] n is independently controlled (14 -20). The controversy has arisen primarily from problems with dye compartmentalization and with alterations in Ca 2ϩ sensitivity of the dye by protein binding (16). Reported differences in nuclear and cytosolic fluorescence have therefore not been universally accepted as evidence for differential regulation of [Ca 2ϩ ] n and [Ca 2ϩ ] c (15)(16)(17). Despite this controversy most authors support the hypothesis that changes in [Ca 2ϩ ] n regulate nuclear events (15)(16)(17)(18)(19)(20).
The Ca 2ϩ -sensitive photoprotein aequorin can be localized to specific organelles within the cell by the addition of protein targeting sequences. It is therefore ideally suited to monitoring Ca 2ϩ within subcellular compartments, as unlike the fluorescent indicator dyes it is insensitive to pH or protein binding (21). This strategy has been successfully used to target aequorin to the mitochondria (22), nucleus (23,24), and endoplasmic reticulum (25,26) of live cells. We have previously demonstrated efficient targeting of apoaequorin to the nucleus by addition of the nuclear structural protein, nucleoplasmin, to the N terminus (24). The N-terminal addition of luciferase to apoaequorin targets aequorin to the cytosol and prevents passive diffusion of the 22-kDa apoaequorin into the nucleus (24). The luciferase aequorin is also serendipitously a more stable cytosolic apoaequorin variant (27). cDNAs encoding luciferaseaequorin and nucleoplasmin-aequorin have been inserted into replication-deficient adenovirus (RAd) vectors, which allow efficient expression of the aequorin chimeras in a wide range of cell types (28). A major advantage of this method of gene transfer is that approximately 100% of the cells express recombinant protein, compared with 10 -20% achieved by transfection (29), and it therefore removes the need to create stable cell lines expressing the photoprotein.
We have used the targeted Ca 2ϩ indicator proteins to demonstrate that increases in [Ca 2ϩ ] c are not reflected by an equivalent increase in [Ca 2ϩ ] n in HeLa cells stimulated by agonist, Ca 2ϩ ionophore, and the endoplasmic reticulum Ca 2ϩ ATPase inhibitor, CPA. Our results provide evidence consistent with the independent regulation of [Ca 2ϩ ] n .

EXPERIMENTAL PROCEDURES
Materials-Coelenterazine, zero Ca 2ϩ calibration buffer, and Ca 2ϩ / EGTA calibration buffers were purchased from Molecular Probes (Eugene, Oregon). Tissue culture reagents, ionomycin, and CPA were from Sigma (UK), and all other chemicals were of analytical reagent grade and from either Sigma or Fisons (UK).
Immunolocalization-HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented with 2 mM L-glutamine, 100 g/ml penicillin, 100 g/ml streptomycin, 100 g/ml amphotericin, and 10% * This research was supported in part by grants from the Wellcome Trust and the Arthritis and Rheumatism Council, and by National Institutes of Health Grant HL38918. 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 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. fetal calf serum. The RAd vectors, luciferase-aequorin (RAdLA) and nucleoplasmin-aequorin (RAdNPA) previously developed (28) were used to infect HeLa cells at a multiplicity of infection of 100 plaqueforming units/cell. Cells were trypsinized 24 h postinfection with the appropriate RAd vector, reseeded onto glass coverslips, and examined by indirect immunofluorescence as described for COS7 cells (28).
In Vitro Ca 2ϩ Calibration of Recombinant Aequorin Variants-HeLa cells were infected with RAdLA, RAdNPA, and a RAd vector-expressing recombinant untargeted apoaequorin, RAdCA (28), as described above. Two days postinfection cells were washed twice with phosphate-buffered saline, scraped from the culture plates, and resuspended in lysis buffer (0.5 mM EDTA, 5 mM ␤-mercaptoethanol, 20 mM Tris HCl, pH 7.4). Harvested cells were subjected to three freeze-thaw cycles and centrifuged at 12,000 rpm for 5 min at room temperature. The supernatants were spin dialyzed at 4°C with three exchanges of zero Ca 2ϩ buffer (120 mM KCl, 1 mM MgSO 4 , 10 mM EGTA, 10 mM MOPS pH 7.1 at 37°C). The aequorin variants were reconstituted with 2 M coelenterazine in zero Ca 2ϩ buffer for 3 h on ice, and light production was measured in triplicate in Ca 2ϩ /EGTA buffers with free Ca 2ϩ concentrations ranging from 11.6 nM to 39.8 M as described previously (25). Rate constants were then calculated from fractional luminescence (30).
Intracellular Ca 2ϩ Measurements-HeLa cells were washed, detached with trypsin, and reseeded onto glass coverslips 24 h after infection with the appropriate RAd vector. Following attachment, cells were incubated in medium supplemented with 5 mM sodium butyrate for 16 -18 h in order to enhance expression. Recombinant apoaqeuorin was reconstituted with 2 M coelenterazine in culture medium at 37°C for a minimum of 3 h prior to each experiment. Coverslips were loaded onto a heated chamber (37°C) and perfused with modified Krebs-Ringer Hepes buffer (containing 120 mM NaCl, 4.8 mM KCl, 1.2 mM KH 2 PO 4 , 1.2 mM MgSO 4 , 1.3 mM CaCl 2 , 5.5 mM glucose, 25 mM Hepes, pH 7.4, at 37°C). For experiments performed in the absence of extracellular Ca 2ϩ , 1 mM EGTA was substituted for 1.3 mM CaCl 2 in the Krebs-Ringer Hepes buffer. The unconsumed aequorin was determined at the end of each experiment by exposing cells to 5 mM CaCl 2 in water, which causes the cells to lyse by exposure to the hypotonic solution. Cells were imaged for chemiluminescence via direct contact of the obverse side of the coverslip with a fiberoptic array, onto a three-stage image-intensified photon counting CCD camera (Photek Ltd.) (31). Data were stored as a photon count event file at video rates and analyzed using both supplied (Photek Ltd.) and custom (C. M. Rembold) software.

Intracellular Localization of Aequorin
Variants-cDNAs encoding recombinant aequorin, luciferase-aequorin, and nucleoplasmin-aequorin were inserted into a RAd vectors which has previously been shown to infect HeLa cells efficiently (28). Indirect immunofluorescence of luciferase-aequorin (Fig. 1A) and nucleoplasmin-aequorin (Fig. 1B) in HeLa cells clearly demonstrates localization of nucleoplasmin-aequorin to the nucleus and of luciferase-aequorin to the cytosol. The level of expression in cells was variable and the proportion of cells which were immunopositive was Ͼ95%.
In Vitro Calibration of Recombinant Aequorin Variants-Calibration of the Ca 2ϩ response in vitro was performed on expressed aequorin variants extracted from RAd vector-infected HeLa cells (Fig. 1C). In order to transform the bioluminescence data into [Ca 2ϩ ], the rate constants were determined for each variant at [Ca 2ϩ ] between 11.6 nM and 39.8 M. The rate constants, k, were determined for each [Ca 2ϩ ] either from the slope of the line when log e (counts/s) was plotted versus time or counts/s/total remaining counts (21,25). Calibration of the recombinant aequorins (Fig. 1C) demonstrated that the addition of nucleoplasmin or luciferase to the N terminus of aequorin did not affect the Ca 2ϩ dependent light emission in vitro when compared to unmodified recombinant aequorin.

Comparison of [Ca 2ϩ ] n and [Ca 2ϩ ] c in HeLa Cells
Stimulated with ATP, Histamine, or CPA-HeLa cells expressing either luciferase-aequorin or nucleoplasmin-aequorin were imaged with a three microchannel plate intensified photon counting CCD camera directly via contact with a fiberoptic. The resolution allowed stimulated luminescence from individual cells to be monitored. Cells which contained large amounts of aequorin after pharmacological manipulation but before lysis were chosen for analysis. For accurate measurement of resting [Ca 2ϩ ], the bioluminescent signal from 60 -100 cells on each coverslip was analyzed and the fractional luminescence was calculated taking into account aequorin consumption (30).
We compared the Ca 2ϩ signal in response to ATP or histamine stimulation in cells expressing either nucleoplasmin-aequorin or luciferase-aequorin. Basal [Ca 2ϩ ] was similar in the nucleus and cytosol, with a mean value of approximately 100 nM. Stimulation with 1 M ATP in the presence of extracellular Ca 2ϩ resulted in similar increases in both [Ca 2ϩ ] n and [Ca 2ϩ ] c ( Fig. 2A). Subsequent stimulation with a higher ATP concentrations (10 and 100 M) increased [Ca 2ϩ ] c more than [Ca 2ϩ ] n . Increases in [Ca 2ϩ ] n and [Ca 2ϩ ] c in response to 100 M histamine occurred at the same rate initially (Fig. 2B); however, the increase in [Ca 2ϩ ] n stopped prior to the increase in [Ca 2ϩ ] c , resulting in higher sustained [Ca 2ϩ ] c than [Ca 2ϩ ] n .
A similar pattern was observed with 100 M ATP stimulation in the absence of extracellular Ca 2ϩ . Initially the increases in [Ca 2ϩ ] n and [Ca 2ϩ ] c were similar but were followed by a larger increases in [Ca 2ϩ ] c than [Ca 2ϩ ] n and the nuclear Ca 2ϩ signal returned to baseline more rapidly (Fig. 2C) (Fig. 3A) and absence (Fig. 3B) of extracellular Ca 2ϩ , 2 M ionomycin induced rapid increases in [Ca 2ϩ ] n and [Ca 2ϩ ] c . The response in the presence of extracellular Ca 2ϩ was larger than the response in its absence, consistent with ionomycin-induced release of intracellular Ca 2ϩ stores in both and induction of a calcium release-activated current (I CRAC ) in the former. The sustained increase in [Ca 2ϩ ] c was significantly larger than the increase in [Ca 2ϩ ] n (Fig. 3A). The relative difference between [Ca 2ϩ ] n and [Ca 2ϩ ] c was less than that observed with histamine or ATP stimulation and is demonstrated by an xy comparison of [Ca 2ϩ ] c and [Ca 2ϩ ] n (Fig. 3C), which shows that the slope of the relation observed with ionomycin in the presence of extracellular Ca 2ϩ was significantly greater than that observed with either ATP or histamine. DISCUSSION The results reported here clearly show that HeLa cells can prevent [Ca 2ϩ ] n rising to the same level as that in the cytosol. The magnitude of this effect depended on the conditions presented to these cells. When the source of the cytosolic Ca 2ϩ signal was entirely extracellular, there was no detectable rise in nuclear free Ca 2ϩ (Fig. 2C), which demonstrates that [Ca 2ϩ ] n does not necessarily increase when a Ca 2ϩ signal is generated in the cytosol. These results have important implications for the role of Ca 2ϩ as a specific regulator of nuclear events through Ca 2ϩ binding proteins, such as calmodulin, calreticulin, and calpain.
Evidence suggesting a role for free Ca 2ϩ in the nucleus has centered on the existence of a mechanism for generating and responding to Ca 2ϩ signals in the nucleus (3)(4)(5)(6)(7). The fluorescent Ca 2ϩ indicator dyes have been extensively used to study nucleocytoplasmic Ca 2ϩ gradients at rest and following stimulation in a wide variety of cell types, often with conflicting results (14 -20, 32-34). A major problem with the dyes is in situ calibration. This is because they load differently into various intracellular organelles where the local environment is critical for indicator calibration (16,17). Unlike these previous studies, our results are not dependent on calibration methods. Addition of proteins or peptides to the amino terminus of aequorin has been shown to be an efficient means of targeting the protein to a variety of subcellular organelles without affecting the specific activity of the photoprotein (24), unlike changes to the C terminus (25). Nuclear targeting of aequorin did not alter its Ca 2ϩ sensitivity in vitro (Fig. 1C), and aequorin calibration is independent of indicator concentration since estimating the rate constant is a ratiometric method. In addition, calibration of the nuclear targeted aequorin in situ demonstrated by Brini et al. (35) showed no difference from the in vitro calibration. However, we are not able to exclude the possibility that the differences in [Ca 2ϩ ] reported here were influenced by focal changes in cytosolic Ca 2ϩ . The kinetics of aequorin mean that high local changes in [Ca 2ϩ ] could affect the precise estimation of the mean [Ca 2ϩ ] (36). Although our data cannot rule out the existence of local elevations in [Ca 2ϩ ] c , the major differences in timing and magnitude of Ca 2ϩ signals reported here cannot be explained entirely by such a mechanism.
The dose-response relationship for ATP and nuclear free Ca 2ϩ showed that the lower nuclear Ca 2ϩ relative to cytosolic was much more obvious at high doses of agonist (Fig. 2, A and  B). Stimulation of HeLa cells by agonist (ATP or histamine) always increased both [Ca 2ϩ ] n and [Ca 2ϩ ] c . When the source of cytosolic Ca 2ϩ signal was entirely through the plasma membrane there was no detectable rise in [Ca 2ϩ ] n , which is consistent with the importance of Ca 2ϩ release from internal stores as the mechanism regulating [Ca 2ϩ ] n . The delay in increase and the slower rate of rise of [Ca 2ϩ ] n when endoplasmic reticulum Ca 2ϩ was released by the sarco, endoplasmic reticulum calcium ATPase inhibitor CPA (Fig. 2D), is also consistent with this hypothesis. The question therefore arises as to whether these agonist-induced differences in [Ca 2ϩ ] n and [Ca 2ϩ ] c are a result of differences in the amount of Ca 2ϩ released into these two compartments from the internal Ca 2ϩ store or differences in Ca 2ϩ buffering capacity of the relevant compartment. A direct route for Ca 2ϩ into the nucleus from the ER is supported by the presence of InsP 3 receptors on the inner membrane of the nuclear envelope (37) and the reports that InsP 3 stimulated release of Ca 2ϩ directly into the nucleus from the nuclear envelope (14,38). These conclusions are also supported by a report that depolarization-induced increases in [Ca 2ϩ ] c were attenuated in the nucleus of neuroblastoma cells (18). Although our results showing attenuation of the nuclear Ca 2ϩ signal could be explained by mechanisms in the cytoplasm preventing Ca 2ϩ reaching the nuclear membrane, the possibility also exists that the nuclear pore may have some selectivity against Ca 2ϩ , in spite of its large size observed microscopically. It is now well established that the nuclear pore is not freely permeable to other small molecules and ions (9,10,13,14). The correlation plot between [Ca 2ϩ ] n and [Ca 2ϩ ] c (Fig. 3C) showed a clear difference between Ca 2ϩ ionophore-generated Ca 2ϩ signals and those induced by agonist, indicating that there must be a mechanism independent of cytosolic free Ca 2ϩ by which agonists raise [Ca 2ϩ ] n .
The lack of any detectable rise in [Ca 2ϩ ] n , when extracellular Ca 2ϩ was added back to cells stimulated by ATP and histamine (Fig. 2C), confirms our previous findings (24) that the nucleus is protected from Ca 2ϩ influx across the plasma membrane induced by the membrane attack complex of complement. Inhibition of the SERCA pump with CPA resulted in a delay before [Ca 2ϩ ] n began to increase. This suggests that the cell can limit increases in [Ca 2ϩ ] n induced by CPA, an agent proposed to increase [Ca 2ϩ ] c by stimulating Ca 2ϩ influx across the plasma membrane (I CRAC ) and which does not involve release of Ca 2ϩ directly into the nucleus. This delay in [Ca 2ϩ ] n increase may operate via a mechanism which involves emptying of the perinuclear Ca 2ϩ store as demonstrated for nuclear targeted proteins (39) and intermediate size macromolecules (11). It is not yet clear whether Ca 2ϩ signals generated at the cell membrane are rapidly buffered by cytosolic Ca 2ϩ binding proteins, the endoplasmic reticulum or mitochondria. A key issue now is whether the source of Ca 2ϩ determines whether Ca 2ϩ signals are transmitted to the nucleus.
Our data, demonstrating clear differences in both the timing and magnitude of changes in [Ca 2ϩ ] n and [Ca 2ϩ ] c , show that control of [Ca 2ϩ ] n must involve at least two different mechanisms, direct release of Ca 2ϩ into the nucleus and control of Ca 2ϩ movement from the cytosol into the nucleus either through the buffering of Ca 2ϩ by the cytoplasm, which would prevent a Ca 2ϩ signal generated at the plasma membrane reaching the nuclear envelope, or regulation across the nuclear membrane. Independent regulation does require that the nuclear envelope regulates its Ca 2ϩ permeability and there is now evidence in the literature to support this (18,24,40) and a signal-mediated Ca 2ϩ release directly into the nucleus. This direct release into the nucleoplasm occurs either from the perinuclear space, as demonstrated in isolated liver nuclei (38), or a possible intranuclear Ca 2ϩ store (41). The presence of InsP 3 receptors on the inner nuclear membrane (37), of nuclear InsP 3 generating enzymes in the nucleus (6), the impermeability of the nuclear envelope to InsP 3 (14), and the finding that injection of InsP 3 into the nucleus generates a nuclear Ca 2ϩ signal (14) clearly supports this hypothesis.
Although nuclear Ca 2ϩ has been extensively studied there is currently no clear consensus on its regulation and role in nuclear events. Individual variation between cell types and the mechanisms regulating [Ca 2ϩ ] n may explain the differences in nuclear Ca 2ϩ signaling reported in the literature (13)(14)(15)(16)(17)(18)(19)(32)(33)(34). Our results highlight the value of using targeted indicators in intact cells to monitor changes in organelle [Ca 2ϩ ]. They also highlight the need now to search for the molecular basis of independent control of nuclear free Ca 2ϩ , and whether this is due to a Ca 2ϩ gradient involving the buffering of cytosolic Ca 2ϩ signals before they reach the nuclear envelope or a gated mechanism involving the nuclear envelope.