Cytosolic Ca2+ Homeostasis in Ehrlich and Yoshida Carcinomas A NEW, MEMBRANE-PERMEANT CHELATOR OF HEAVY METALS REVEALS THAT THESE ASCITES TUMOR CELL LINES HAVE NORMAL CYTOSOLIC FREE Ca2+*

The intracellularly trappable fluorescent Caz+ indi- cator quin-2 was used to measure free cytosolic Ca2+, [Caz+],, in the two highly dedifferentiated tumor cell lines, Ehrlich and Yoshida ascites carcinomas. was apparently significantly lower than in any normal cell tested previously with this method.

4 To whom correspondence should be addressed. surface receptors (7), intracellular enzyme activation (81, and mitogenic stimulation (9-11). In a preliminary report (12), we applied quin-2 to two highly dedifferentiated cell lines, Ehrlich and Yoshida carcinomas, to test the possibility that a permanent modification of [Caz+]i homeostasis is involved in the uncontrolled proliferation of malignant cells; this hypothesis was supported by much indirect evidence (13-27). We have already shown (12) that the apparent [CaZ++li of Ehrlich and Yoshida carcinomas ranges between 10 and 80 nM, the values of [Ca2+Ii being apparently higher at higher Ca2+ indicator trapped inside the cells. The interpretation of such result was intriguing. On the one hand, the possibility that the highly dedifferentiated ascites carcinomas had a low [Ca2+Ii raised a critical biological question, on the other hand the dependence of [Ca2+]i on the intracellular quin-2 content suggested looking for a technical artifact in the measurement of [Ca2+Ii.
For some time we had been aware that endogenous heavy metals were a potential source of artifact in the measurement of [Ca2+Ii with quin-2 (11, but we assumed that heavy metals were so tightly bound in the cytoplasm that they would not interfere seriously with the dye. Rightly or wrongly the same assumption has been tacitly made by nearly all investigators measuring [Ca"+Ii by any method be it photoproteins, dyes, or ion-sensitive microelectrodes. We have thus reconsidered this assumption since the strong dependence of [Ca2+Ii on intracellular quin-2 concentration may be explained by a partial quenching of the dye by a fixed amount of endogenous heavy metals. In order to tackle this problem we thought that a direct test was to measure [Ca2+]i after intracellular heavy metals were removed by chelation, translocation, or both. The consequent goal of a membrane-permeant and heavy metal-specific chelator suggested looking for an uncharged polydentate ligand using nitrogens rather than oxygens as donor atoms. Examination of review compilations immediately revealed that a likely candidate, TPEN,' was already known in the chemical literature (28). This new chelator permeates membranes and has an extraordinary high affinity for heavy metals but a low affinity for Ca" and M$+. This unusual and useful combination of physicochemical properties can be demonstrated by the ability of externally applied TPEN to strip heavy metals from quin-2 trapped within liposomes. The use of TPEN has allowed the demonstration that a variable proportion of intracellular quin-2 is quenched by heavy metals both in normal as well as in neoplastic cells. The underestimation of [Ca2+Ii, however, is negligible in normal cells above an intracellular quin-2 concentration of 1 mM, while in the ascites carcinomas [CaZ++li was underestimated significantly up to 5-6 mM intracellular quin-2. After correction for heavy metal interference, [Ca2+]i was found to be similar in both normal and neoplastic cell lines, i.e. between 100-200 nM at any quin-2 concentration used. Data on cytoplasmic Ca2+ homeostasis in the ascites carcinomas will be presented. The relevance of these results to the problem of malignant transformation will be also briefly discussed.

EXPERIMENTAL PROCEDURES
Cells-Ehrlich ascites cells were grown in BALB/c mice and Yoshida cells in Wistar albino rats. Cells were harvested 6-10 days after transplantation, washed twice, and resuspended in Hank's minimal essential medium buffered with HEPES at pH 7.4. Yoshida cells in our hands were more variable than Ehrlich cells. In particular, in some batches of cells a large increase in cell size was observed. For this reason most of the experiments shown in this paper were performed with Ehrlich cells.
Mouse thymocytes were obtained from BALB/c mice as previously described (1). Human granulocytes were obtained from healthy donors as previously described (3). Viability, always above 90%, was assessed by eosin red exclusion.
[Caz+]i Measurement-&in-2 acetoxymethyl ester, quin-a/AM, was synthesized according to Tsien et al. (1). The loading procedure was similar to that described by Tsien et al. (1). Briefly, cells at a concentration of 8-10 X 10' ml were incubated at 37 "C in HMEM buffered with HEPES at pH 7.4 with various concentrations of quin-2/AM and after 15 min the cells were diluted to approximately 2 X 10'/ml. Bovine serum albumin, 0.5%, was included to increase the loading efficiency. After 40-45 min, the cells were washed with fresh medium and kept at room temperature. Immediately before use the cells were centrifuged down and resuspended in a basic physiological saline medium containing: 140 mM NaC1, 5 mM KCl, 1.1 mM CaClZ, 0.5 mM MgCl,, 1 mM N&PO,, 5.5 mM glucose, 0.1 mM diethylenetriaminepentaacetic acid buffered with 20 mM HEPES at pH 7.4 (37 "C). Unless otherwise specified, all the experiments were performed in this medium. &in-2 cell content was calculated on the basis of a cell volume of 0.85 and 0.7 pl X (10' cells)" for Ehrlich and Yoshida cells, respectively. Cell volume was measured using 3Hz0 and ["Clinulin.
As described previously (l), calibration of the dye responses to give absolute [Ca2+Ii values requires determinations of the minimal and maximal fluorescences, Fand F-, of the dye in very low and high Ca", respectively. Fand F,. were determined in a different order than previously (see "Results"). Excitation and emission wavelengths were 339 f 2 and 492 f 10 nm, respectively.
Heauy Metals-The total cytoplasmic concentration of heavy metals was measured as follows. Cells (1.5 X l@ X ml-') were suspended in Caz+-free medium in the presence of the membrane-impermeant heavy metal chelator DTPA, then 300 p~ digitonin was added. Cells were kept at 37 "C for 15 min to allow the release of most cytoplasmicsoluble components; after a first centrifugation to eliminate nuclei and organelles the supernatant was further spun at 300,000 X g for 2 h to remove the heaviest proteins. The heavy metal content in the supernatant was measured by atomic absorption.
TPEN was synthesized according to Anderegg and Wenk (28). The identity of the product was confirmed by its melting point and proton magnetic resonance spectrum. Its affinities for Ca2+ and Mg2+ were measured by comparing the pH-titration curves of 1 mM TPEN in 100 mM KCl, 22 "C, in the presence and absence of 10 mM CaC12 or MgClz. The multiple equilibria (29) were analyzed by the computer program SCOGS (30) to extract least squares estimates of the protonation and metal-binding constants and their statistical uncertainties. The protonation constants, deduced from the divalent cation-free titrations, agreed closely with those reported by Anderegg and Wenk (28).
Liposomes were prepared essentially as described by Olson et al. (31). Briefly, egg lecithin and phosphatidylserine, dissolved in chloroform, were dried on the bottom of a glass tube by a stream of N2.
A solution containing quin-2 and the appropriate concentrations of buffer, CaClZ, MgS04, ZnCl2, or MnClZ were then added and the phospholipids were then carefully resuspended by gentle mixing. The suspension was left at room temperature for 30-40 min and then mixed by vortexing for 10 periods of 20 s at 30-s intervals. The liposome suspension thus obtained was kept at room temperature until used. Before using, an aliquot of the liposomes was spun at 15,000 X g for 15 min and the pellet resuspended in fresh medium.
Enzyme and CeU Metabolism Assays-Glutamate dehydrogenase, acid phosphatase, and glucose 6-phosphatase were measured according to standard methods (32). Mitochondrial Caz+ uptake was measured with a Caz+-sensitive electrode as described by Bernardi and Pietrobon (33). Acridine orange uptake by lysosomes was measured spectrophotometrically as described by Dell'Antone (34). ATP and 0 2 consumption rate were measured as previously described (35).
Materials-Ionomycin was a kind gift of Dr. Liu from Hoffman La Roche (Nutley, NJ). A23187 was obtained from Calbiochem. HMEM was purchased from Flow Lab, United Kingdom. All other chemicals were of analytical or highest available grade.

Quin-2 Loading and Intracellular Location of the Indicator-
We have previously shown (1-7) that the specific fluorescent Ca2+ indicator quin-2, structure 3b of Tsien (36), can be trapped in the cytoplasm of intact lymphocytes and platelets by incubating the cells with the nonpolar ester derivative quin-2/AM, which is hydrolyzed back to the original molecule by cytoplasmic esterases. The tumor cell lines, Ehrlich and Yoshida ascites carcinomas, behave like normal cells as far as loading efficiency and location of the indicator inside the cell is concerned. The loading efficiency is typically about 15-25%, slightly lower than that observed in lymphocytes. These cell lines, loaded with up to 6-10 mmol of quin-2lliter of cell water, remain viable with minor changes in cellular ATP levels (15% decrease at quin-2 loading above 3 nmol X (10' cells)". Intracellular location of quin-2 was assessed as previously described for lymphocytes (l), i.e. by finding a digitonin concentration which releases all quin-2 from the cells without affecting the functional integrity of the major membranous intracellular organelles. Table I shows that digitonin at the concentration of 3 pg X (10' cells)" releases all quin-2 and the cytoplasmic enzyme lactic dehydrogenase even though the organelles remain largely intact as judged by the sedimentability of the marker enzymes of mitochondria (glutamate dehydrogenase), lysosomes (acid phosphatase), and microsomes (glucose 6-phosphatase). Furthermore, we have observed that mitochondria and lysosomes remain functionally intact after digitonin treatment as demonstrated by their  respective abilities to accumulate Ca'+ or the fluorescent dye, acridine orange. This insensitivity of lysosomes to this digitonin concentration was observed only in the intact cells. In fact digitonin is very potent in releasing acid phosphatase and/or in inhibiting acridine orange uptake in isolated lysosomes. We presume that the lack of effect of digitonin on lysosomes in situ is due to the preferential interaction of the detergent with the cell plasma membrane. It was also demonstrated that the esterases which hydrolyze quin-2/AM are present in the supernatant of cells treated with digitonin, but not in the pellet containing organelles, nuclei, and membranebound enzymes. Quin-2/AM was added to the supernatant or to the resuspended pellet of digitonin-treated cells. After 1 h at 37 "C, quin-2/AM was largely hydrolyzed to quin-2 (acid) in the sample containing the supernatant but not in that containing the pellet of digitonin-treated cells.
The excitation and emission spectra of quin-2 trapped by Ehrlich and Yoshida cells, after correction for cells autofluorescence, have the same shape and peaks of quin-2 in simple saline medium. As previously discussed (1) the chromophore of quin-2 is similar to a dansyl moiety and one would expect some changes of the spectral characteristics if the dye was bound to intracellular hydrophobic molecules.
The observation that quin-2 in these tumor cell lines, as in other cells, is located in the cytoplasm and not detectably bound to intracellular components, justifies the use of the calibration procedure previously described for lymphocytes (1). However, as discussed by Tsien et al. (l), the effective affinity of the indicator for Ca2+ is somewhat affected by [Mg"']i and by pHi. Therefore, an absolute quantitation of [Ca'+Ii requires knowledge of these two variables. pHi of Ehrlich cells has been measured by several groups and it has been shown to range between 7.2 and 7.4, at pH, 7.4, depending on the experimental conditions and methodology (37-39).
Di Virgilio (40) has recently measured the pHi of Ehrlich and Yoshida cells by the methods recently described by Rink et al. (41). Under the same conditions as those used in these experiments, pHi was found to be 6.93 f 0.03 and of 6.95 k 0.02 for Ehrlich and Yoshida cells, respectively, thus slightly lower than the pHi measured with 31P NMR (39) or "Cdimethyloxazolidinedione accumulation (37). The reasons for these discrepancies have been briefly discussed (40) and a more detailed analysis of this problem will be presented elsewhere.' However, it is relevant to mention that the effective & of quin-2 for Ca2+ is practically pH-independent above pH 6.85. Thus even our measurement of pHi, which is the lowest for these cells in the literature at pH, 7.4, is significantly higher than that required for effectively modifying the Kd for Ca'+ of quin-2. ately to the fall in [Ca'+],, whereas intracellular dye should respond only gradually. The cells were then lysed with Triton X-100 to expose all the dye to low [Ca"] and establish Fhn. Finally, CaClz (5 mM) was added to restore high [Ca"] and record F-. This calibration procedure was preferred to that previously used (1) (first Triton then EGTA and Tris), because unlike other cells like lymphocytes and granulocytes, some external quin-2, variable from preparation to preparation, was always present even after repeated washings. This external quin-2 is probably due to cell damage during spinning and resuspension of the cells. Furthermore the spontaneous leakage of trapped quin-2 was higher in these cell lines compared to normal cells like lymphocytes. For example, in simple saline medium at 37 "C Ehrlich cells release up to 4040% of total intracellular quin-2 in 1 h while under the same conditions lymphocytes release only 10-15%. After correction for extracellular dye, quin-2 inside the cells was only about onequarter saturated with Ca", corresponding to a [CaZ+li of nearly 40 nM. By contrast, quin-2 in a variety of normal cell types (for a review see Ref. 42) is generally about halfsaturated with CaZ+, corresponding to resting [Ca2+Ii in the range 100-120 nM. Obviously, it is important to assess the possibility that quin-2 itself may be perturbing [Ca'+Ii levels, especially since quin-2 is a high affinity Ca'+ chelator, generated in millimolar concentrations inside the cell in a calcium-free form. If the cells could not maintain Ca'+ homeostasis one would expect [Ca'+]i to decrease as the cellular concentration of quin-2 increases. A nearly constant [Caz++li as a function of quin-2 cell content was previously obtained with other cell types, and was associated with a net accumulation of Ca' ' from the external medium, enough or more than enough to maintain [Ca'+]i constant. Fig. 2 shows however that in these tumor cells, the apparent [Ca2+Ji actually increases as the quin-2 concentration increases. Moreover, at very low quin-2 cell content, below 0.2 mM, the apparent and/or diluted after detergent addition.

Effects of Heavy Metals on Quin-2 Fluorescence and Characterization of a New Hydrophobic Heavy Metal Chelutor-
Heavy metals like Fe2+, Zn2+, etc., present inside the cells, are putative quenchers of quin-2 fluorescence. Fig. 3 shows that, in free solution, quin-2 fluorescence is nearly completely quenched by stoichiometric concentrations of Znz+ and that this effect can be completely reversed by a heavy metal chelator like DTPA. If the cells contained sufficient heavy metals available to bind to quin-2, its fluorescence inside the cell would be partially quenched. However, after cell lysis, the heavy metals would be diluted and chelated by the EGTA present in the medium. The sequestration of heavy metals by EGTA would therefore restore the Ca2+ sensitivity to quin-2. If this were the case, our calibration procedure, which requires cell lysis in the presence of EGTA, should lead to an underestimation of [Ca2+Ii.
In Table 11 the digitonin-releasable amount of Zn", Fez+, Cuz+, and Mn2+ of Ehrlich and Yoshida cells is reported. While Cu2+ is quite low, and Mn2+ hardly measurable, the concentration of Fe2+ and especially of Zn2+ are rather high, namely E 100 and s 700 p~, respectively. The total content of heavy metals here reported is similar to that already observed by Thiers and Vallee in liver cells (43).  Although we do not know how tightly these cations are bound to proteins and thus how easily exchangeable they are, their content is stoichiometrically high enough to potentially interfere with [Ca2+Ii measurement with quin-2. Hydrophilic heavy metal chelators like EGTA and DTPA have been added to the extracellular medium after the loading procedure but no difference of quin-2 fluorescence was observed between controls and treated cells. This result was not unexpected, since both EGTA and DTPA are unable to cross the plasma membrane.

Heauy metals content of Ehrlich and Y o s h a ceUs
Recently Anderegg and Wenk (28) have reported the synthesis of a new heavy metal chelator TPEN, which has properties theoretically suitable for stripping heavy metals from quin-2 inside the cells. In particular: 1) TPEN is highly soluble in organic solvents, moderately soluble in water, and substantially uncharged at physiological pH. The p K values for mono-through tetraprotonation are 7.19, 4.86, 3.35, and 2.95, respectively (28). 2) TPEN has very high affinities for heavy metals, for example, respectively, TPEN will bind negligible amounts of these cations. Therefore it is expected not to exert any direct buffering effect on their cytosolic levels. 4) The ultraviolet absorption of aqueous TPEN peaks at 260 nm and is insignificant above 300 nm, so that it constitutes negligible fluorescence interference with quin-2 measurements (excitation 339 nm). By contrast, a previous membrane-permeant heavy metal chelator, o-phenanthroline (44), is strongly fluorescent at wavelengths near those of quin-2 (45). Fig. 3B shows the effect of TPEN in relieving Zn2+ quenching of quin-2 fluorescence in solution. Similar results were obtained with Mn2+, Cu2+, and Fez+.
In Fig. 4A, it is shown that externally added TPEN can compete for Zn2+ and Mn2+ binding to quin-2, when quin-2 is trapped within the aqueous phase of liposomes. Addition of TPEN, but not of DTPA, to liposomes containing quin-2, MnC12, ZnC12, and CaC12, at a concentration under which quin-2 fluorescence is about 50% quenched, induces a rapid increase of fluorescence to nearly 100% Caz+ saturation. Addition of EGTA and Tris to reduce [Ca"], to below 10 nM does not induce any rapid decrease of quin-2 fluorescence indicating that the rise of fluorescence was not due to unspecific leakage of the indicator from the liposomes. Furthermore even in the presence of a large Ca2+ concentration gradient across the liposome membrane, quin-2 fluorescence remains constant for several minutes indicating that the liposomes are still impermeant to Ca". Addition of the Ca2+ ionophore ionomycin immediately induces a decrease of quin-2 fluorescence indicating a decrease of [Ca2+Ii inside the liposomes. Fig. 4B shows that TPEN does not increase the fluorescence  quin-2 fluorescence from 28 up to 50% of maximal fluorescence, while further additions of TPEN are ineffective. The effect of TPEN on apparent [Caz+Ii is to raise it from 40 to 110 nM. Fig. 5B shows that the effect of TPEN in rising [Ca2+]i is more dramatic at very low quin-2 cell content. Addition of TPEN rapidly raises the fluorescence of cellular quin-2 from a value near 0% of FmaX to about 50% of F,.,, corresponding to a [Ca2+]i of 115 nM. This last value of [Ca2+]i is identical with that measured at higher quin-2 cell concentration in the presence of TPEN (Fig. 5A). Fig. 6 shows [Ca2+Ii of Ehrlich and Yoshida cells as a function of quin-2 cell content in the presence of TPEN. In contrast to the results of Fig. 2, [Ca2+Ii is now constant at about 110-120 nM.

AA g A A A A A A
It could be argued that TPEN raises [Ca2+Ii by inducing some toxic effect on the cells. However in the short term, 630 min, TPEN, at the concentrations used in these experiments, has no appreciable effect on cell viability (measured by eosin red exclusion), ATP levels, and O2 consumption rate. Only a 2-3% extra release of trapped quin-2 was observed in the first 5 min after TPEN addition, while afterward the spontaneous leakage was indistinguishable from that of controls. We have observed some clear toxicity of TPEN with longer incubations: a 20% reduction of cell viability and a 30% reduction of [3H]thymidine incorporation after 4 h of incubation with 10 PM TPEN. Although these last observations indicate that TPEN is not totally innocuous to the cells, the toxicity of this heavy metal chelator is not dramatic, and, at least in the short term, no major modification of cell metabolism occurs.
Effect of Heavy Metal Removal on [Ca2+]; Measurement in Normal Cells-In normal cells like mouse thymocytes or human granulocytes the effect of TPEN, although qualitatively similar to that observed in the tumor cells, is quantitatively less evident. For example, in mouse thymocytes (Fig.  7A), at 0.45 mM [quin-2];, TPEN raises the Caz+ saturation from 32 to 5076, corresponding to an apparent [Ca2+Ii rise from 55 to 115 nM; in Ehrlich cells, at the same quin-2 cell content, "PEN apparently raises [Ca"]; from 20 to 115 nM. In both thymocytes and granulocytes between 0.2 and 4.0 mM quin-2 cell content (Fig. 8, A and B ) , [Ca2+Ii is constant at E 120-130 nM in the presence of TPEN, thus not significantly different from the values observed in the Ehrlich and Yoshida carcinomas. However, while in mouse thymocytes in the absence of TPEN [Ca2+]; is clearly underestimated below 1 mM [quin-2]i, in human granulocytes the effect of TPEN is hardly significant even at very low [quin-2Ii.
Additional evidence suggesting that no major modification of normal Ca2+ homeostatic mechanisms occur as a consequence of TPEN addition is provided by the experiments of Fig. 7. Pretreatment of mouse thymocytes or human granulocytes with TPEN does not inhibit the rise in [CaZ++li induced by the polyclonal T-cell mitogen concanavalin A in thymocytes (compare Fig. 7, A with B ) or by the chemotactic peptide met-Leu-Phe in granulocytes (compare Fig. 7, C with D) Fig. 7A can be observed, in the absence of TPEN, at [quin-2Ii above 1 mM, when the underestimation due to heavy metal quenching is negligible (results not shown and see also Ref. 9).
As pointed out previously the effect of TPEN in granulocytes is hardly detectable. [Caz+]; Homeostasis in the Carcinoma Cells- Fig. 9A shows the effect of 100 nM A23187 on Ehrlich cells [Caz+]i. There is a rapid rise of [Ca2+]; from 100 nM to 800 nM and then slowly [CaZ+li decreases and settles to about 200 nM. In other cells, mouse spleen and thymus lymphocytes or pig mesenteric lymphocytes (l), the same amount of ionophore (x mg of protein") is able to rise [Ca2+Ii to above 1 PM and [Ca2+]i remains at this level for more than 1 h. The inefficacy of A23187 in rising total cell Ca2+ in ascites carcinomas has already been observed by Cittadini et al. (46). This relative insensitivity to Ca" ionophores is not specific to Ehrlich cells, since it was also observed in Yoshida cells and with the chemically unrelated Ca2+ ionophore, ionomycin, in both cell types (not shown). The cause of the insensitivity to ionophores remains obscure, although one might postulate either a high background rate of both Ca" influx and active extrusion, or a slow movement of the ionophore in the plasma membrane. The observation by Cittadini et al. (47) that the rate of %a uptake is only slightly increased by A23187 tends to favor the last hypothesis. One might speculate that one responsible factor might be a decreased membrane fluidity due to the known abnormally high levels of cholesterol in tumor cell plasma membranes (48). Fig. 9B shows the effect of A23187 in calcium-free medium. The ionophore induces a rapid rise of [Ca"]; which slowly returns to and below the initial level. This experiment indicates that tumor cells possess membrane-enclosed ca" stores which can be rapidly depleted by ionophores. The possibility that the mitochondria are the major intracellular Ca2+ stores has long been debated (49). This problem is critical in tumor cells where an increased Ca2+ uptake capacity of isolated mitochondria is widely documented (18,19). In order to determine whether Ca2+ is sequestered by mitochondria in intact tumor cells, the proton ionophore and uncoupler FCCP, known to release Ca2+ from isolated mitochondria, was added. Both in calcium-free and in calcium medium FCCP causes a small transient increase of quin-2 fluorescence (not shown).
The interpretation of this experiment is complicated by a side effect of FCCP on cytoplasmic pH. We have observed2 that FCCP causes a rapid decrease of cytoplasmic pH, as measured with an intracellularly trapped pH indicator, from the initial value of 6.93 to about 6.65. The effect is not surprising since FCCP is a proton ionophore and pHi should tend to reach electrochemical equilibrium with the plasma membrane po-  Fig. 7C is higher than that of A23187 in Fig. 9B. This is due to the lower quin-2 cell concentration in Fig. 9C (see also the difference in noise). As previously discussed (1, 7) quin-2 will blunt any rise of [Ca2+Ii induced by a release of Ca2+ from intracellular stores, this effect increasing with quin-2 cell concentration. In the same batch of cells, the amount of [Ca2+Ii increase induced by FCCP is about one-third of that induced by ionomycin. Fig. 10 shbws that depolarizing the plasma membrane potential, either by increasing external K+ concentration (left) or with a pore-forming ionophore, gramicidin (right), does not affect [Ca2+]i. It is worth mentioning that gramicidin dramatically increases intracellular Na+ content; its lack of effect on [Ca2+Ii argues against any major role of Na+ in controlling Ca2+ homeostasis in these cells. This last observation is consistent with the recent report by Spitzer et al. (50). In insideout plasma membrane vesicles of Ehrlich cells, only an ATPdependent Ca2+ pump was observed, not a Na+-Ca2+ exchange.
We have also tested several calmodulin inhibitors. Phenothiazines like fluphenazine and chlorpromazine proved unsuitable because at their necessary concentrations, their fluorescence overwhelms the quin-2 signal. Pimozide is not itself fluorescent but seems to cause substantial changes in cell autofluorescence. At 30-50 PM it causes a negligible increase of [Ca2+Ii in the short term (~1 0 min); prolonged incubations cause some increase of [Ca2+Ii, but also an increase of cell death, thus making doubtful any interpretation of this result. The recently synthesized very potent calmodulin inhibitor R24571 (calmidazolium) has no short-term effect on [Ca2+Ii at concentrations up to 2.5 X lo" M. At micromolar concentrations, however, it causes rapid cell death, as revealed by quin-2 leakage from the cells and eosin staining. Whereas R24571 at submicromolar doses seems highly effective at inhibiting calmodulin in homogeneous assay systems, it is not ' T. Pozzan, unpublished

DISCUSSION
Although the measurement of [Ca2+]i with quin-2 is a relatively new method, this indicator has found wide applicability in cell biology (42). Quin-2 has been used mostly qualitatively to demonstrate whether or not [Caz+li changes in response to certain stimuli; little attention had been paid to the calibration of the fluorescence signals into [Caz+Ii, which requires knowledge of parameters like cytoplasmic pH; or [Mg"], and careful checking of possible sources of artifacts, such as incomplete hydrolysis, binding, or sequestration of the dye by cellular organelles. In particular when quin-2 is applied to a new cell type, where these parameters are unknown, and the goal is to determine whether a permanent alteration of Ca2+ homeostasis is involved in the biological phenomenon under investigation, the knowledge of absolute [Caz+li becomes critical. The above mentioned controls have been performed in this paper for the Ehrlich and Yoshida carcinomas. No parameter was found to be significantly different from that of other cells previously tested to justify the apparent low [Caz+li and the strong dependence of [Ca2+Ii on the intracellular quin-2 concentration observed in these cell lines. Two recent observations suggest that heavy metal quenching of intracellular quin-2 is not as negligible as we previously assumed. (a) The apparent decrease of [Ca2+Ii at very low intracellular quin-2 concentrations even in lymphocytes (10); (b) the observation by the use of 5FBAPTA (an analogue of quin-2 which can be followed by NMR) that a significant proportion of the intracellular Ca2+ indicator is bound to heavy metals in thymocytes (51).
A new membrane-permeable heavy metal chelator, TPEN, was thus employed in order to strip heavy metals which might be bound intracellularly to quin-2. TPEN is simply a structural analogue of EDTA in which the four carboxylates have been replaced by pyridine groups of similar steric placement. Although TPEN was known to be an extremely high-affinity heavy metal chelator (28,52) nothing had been reported on its biological activity or its interaction with Ca2+ or M e . We now find that TPEN binds Ca2+ and MgZ+ negligibly at intracellular levels of those cations, yet it readily crosses liposomal membranes and tears divalent metals from trapped quin-2.
Both in Ehrlich and Yoshida cells, as well as in the normal cell types tested here, mouse thymocytes and human peripheral blood granulocytes, the effect of TPEN is qualitatively similar. While [Ca2+Ii apparently rises with increasing [quin-2Ii, in the absence of TPEN, [Ca2+]i is constant around 100-150 nM in the presence of TPEN over the whole range of [quin-21, concentration employed.
However, quantitatively the effect of TPEN on apparent [Caz+li is far less dramatic in these normal cells, especially in the granulocytes. We have too little data to speculate why these tumor cell lines have a greater amount of quin-2perturbing, exchangeable heavy metals than other cells have, nor what this difference might make to the cell's physiology. Several pieces of evidence indicate that the effect of TPEN is not due to a real rise in [Ca2+li, but rather to its ability to tear heavy metals from intracellular quin-2.1) The effect of TPEN is inversely related to the quin-2 concentration inside the cells, i.e. larger at low [quin-2Ii when the heavy metals content of the cells is similar or even higher than that of quin-2. 2) The final effect of TPEN is dose-independent, suggesting that after removal of heavy metals from quin-2 it has no other major effects on [Ca2+Ii homeostasis. 3) In the presence of TPEN, [Ca2+Ii readings are independent of the amount of quin-2 loaded. 4) The final effect of TPEN on [CaZ+li readings is independent of the cell density. 5) No short-term effects of TPEN on cell viability, ATP content, and O2 consumption rate have been detected. 6) The [Caz+li rises induced by concanavalin A or Met-Leu-Phe, in lymphocytes or granulocytes, respectively, are not affected by the pretreatment of the cells with TPEN.
The ability of free TPEN to cross membranes rapidly and preferentially chelate intracellular heavy metals is enough to make it a novel and promising tool for probing the importance of such metals inside intact cells. Its affinities are substantially higher and its stoichiometry is simpler than those of ophenanthroline, a chelator previously used in attempts to mop up heavy metals (44). To prevent TPEN from removing needed heavy metals, it might be necessary to administer it in preformed combination with those metals, just as EDTA is administered clinically as the Ca2+ complex. Further modification of the TPEN structure might also be needed to increase its ability to discriminate between beneficial and toxic heavy metals.
The binding of heavy metals to quin-2 is a significant caution in the use of this probe as a relatively nonperturbing indicator of [CaZ+li. Depending on the cell type, quin-2 might alter some important cell functions by chelating heavy metals, giving rise to pharmacological effects not attributable simply to calcium buffering. Such effects would probably be most noticeable in prolonged incubations (10). In practice, however, most cell types loaded with quin-2 have shown no obvious toxicity and can still perform highly complex functions, like the energy-dependent redistribution of surface receptors in B lymphocytes (7), shape change, aggregation and secretion in platelets (2), exocytosis and superoxide production in neutrophils (3), neurotransmitter release in pheochromocytoma cells (4), activation of phosphorylase in hepatocytes (€9, and mitogen-stimulated DNA synthesis (10). If specific instances of quin-2 toxicity are found, the availability of TPEN offers a means to test whether the perturbation is due to heavy metal chelation and whether the quin-2 readings are being perturbed by heavy metal quenching. In the long run, the best solution would be to use Caz+ indicators with reduced heavy metal affinities. Fortunately, such properties seem available in chelators based on BAPTA (36), in which there is no ring nitrogen as in quin-2 to promote M$+ and heavy metal binding.
The use of quin-2 and TPEN has also allowed an approach to a problem of utmost biological interest, i.e. the relationship between [Ca"+]i and the malignant phenotype. Until now, free cytosolic calcium has never been measured in undifferentiated tumor cells, although there have been several indications of an alteration in calcium homeostasis after malignant transformation (12-26). On the basis of indirect evidence, both opposing possibilities have been proposed, i.e. a free cytosolic Caz+ concentration higher (12) or lower (53) in tumor cells than in normal cells.
Here we demonstrate that once the heavy metal interference is removed, the Ehrlich and Yoshida cells show resting [caZ+li in the same 100-200 nM range as essentially all nontransformed nucleated mammalian cells yet tested with quin-2. Unfortunately, we do not have the exact normal precursor cells to compare with the ascites cells. It would be highly desirable to compare normal and transformed phenotypes in one cell type like the 3T3 fibroblast, but, at present, there are technical problems in calibrating quin-2 fluorescence in cells growing on a solid substrate (1). Although we have not yet compared a single cell type before and after transformation, the normality of [Caz+Ii even in these highly dedifferentiated tumor cells suggests that alterations in [Ca2+Ii cannot be general features of the transformed phenotype as some authors had suggested. The processes maintaining [Ca2+]i also seem rather similar to those in differentiated cell types like lymphocytes (1). The ascites cells possess intracellular Ca2+ stores, including, but not limited, to the mitochondria, but lack voltage-activated Ca2+ channels or a Na+-dependent Ca2+ extrusion mechanism as shown by the insensitivity of [Ca2+Ii to membrane depolarization or internal Na+.
If there is any general difference between normal and transformed cells in their handling of intracellular calcium, it is more likely to be in the sensitivity of cellular processes. to [Ca2+Ii than the actual magnitude of [Caz+li. For example, the high calmodulin content in certain tumor cells might lead to an abnormal activation of some Ca2+-calmodulin-dependent reactions even at normal resting [Caz+]i. Another powerful and ubiquitous transducing pathway activatable by Ca2+ is protein kinase C (54). The phorbol esters, a highly potent class of tumor promoters, either sensitize the kinase to Ca2+ or eliminate the Ca2+ requirement altogether, causing the cellular effects of the kinase to be expressed in the total absence of any elevation in [Ca2+]i above the normal resting level (2,6,9).