Calcium induces cell survival and proliferation through the activation of the MAPK pathway in a human hormone-dependent leukemia cell line, TF-1.

Survival and proliferation of cells of a human myelo-erythroid CD34+ leukemia cell line (TF-1) depend on the presence of granulocyte-macrophage colony-stimulating factor or interleukin-3. Upon hormone withdrawal these cells stop proliferating and undergo apoptotic process. In this report we demonstrate that a controlled increase in [Ca2+]i induces hormone-independent survival and proliferation of TF-1 cells. We found that moderate elevation of [Ca2+]i by the addition of cyclopiasonic-acid protected TF1 cells from apoptosis. Furthermore, a higher, but transient elevation of [Ca2+]i by ionomycin treatment induced cell proliferation. In both cases caspase-3 activity was reduced, and Bcl-2 was up-regulated. Higher elevation of [Ca2+]i by ionomycin induced MEK-dependent biphasic ERK1/2 activation, sufficient to move the cells from G0/G1 to S/M phases. Meanwhile, activation of ERK1/2, phosphorylation of the Elk-1 transcription factor, and, consequently, a substantial elevation of Egr-1 and c-Fos levels and AP-1 DNA binding were observed. Moderate elevation of [Ca2+]i, on the other hand, caused a delayed monophasic activation of ERK1/2 and Elk-1 that was accompanied with only a small increase of Egr-1 and c-Fos levels and AP-1 DNA binding. The specific MEK-1 kinase inhibitor, PD98059, inhibited all the effects of increasing [Ca2+]i, indicating that the MAPK/ERK pathway activation is essential for TF-1 cell survival and proliferation. Based on these results we suggest that the elevation of the [Ca2+]i may influence the cytokine dependence of hemopoietic progenitors and may contribute to pathological hematopoiesis.

The tight control of proliferation, survival, and apoptosis of bone marrow progenitor cells has a major role in allowing normal hematopoiesis. A variety of cytokines, including GM-CSF 1 and IL-3, regulate viability, proliferation, differentiation, and function of hemopoietic cells (1). Hemopoietic cells undergo apoptotic cell death in the absence of growth factors, whereas an escape from this regulation may result in leukemogenesis.
The human GM-CSF/IL-3-dependent myelo-erythroid cell line, TF-1, has been reported to undergo apoptosis upon hormone deprivation (2). Apoptosis of these cells can be suppressed by re-administration of GM-CSF, IL-3, or IL-4 (3) or overexpression of Bcl-2 (4). In this report we demonstrate that cytoplasmic free calcium ([Ca 2ϩ ] i )-increasing agents can also suppress the apoptotic process and result in long term cell survival.
A number of endogenous substances (adrenergic, purinergic agonists, and cytokines) are able to evoke different [Ca 2ϩ ] i signals. To examine the effect of calcium signals on cell survival and proliferation, we have applied various [Ca 2ϩ ] i -mobilizing agents. SERCA inhibitors, e.g. cyclopiasonic acid (CPA) and thapsigargin deplete sarco-endoplasmic reticulum Ca 2ϩ stores, which in turn induce capacitative influx of Ca 2ϩ (5). Ca 2ϩ ionophores like A23187 and ionomycin mobilize Ca 2ϩ from nonspecific sources (6). In parallel with the distinct mechanism of Ca 2ϩ elevation, the level of [Ca 2ϩ ] i and its physiological effects may also differ among these [Ca 2ϩ ] i -increasing reagents. We demonstrate that endogenous substances induce similar [Ca 2ϩ ] i increases as CPA. Thus the elevation of [Ca 2ϩ ] i caused by CPA is within the range of physiological [Ca 2ϩ ] i signals.
The alteration of intracellular Ca 2ϩ concentration control cellular processes as diverse as proliferation, development, contraction, or secretion. However, when exceeding its normal spatial and temporal boundaries, Ca 2ϩ can result in cell death through both necrosis and apoptosis (7).
The purpose of this study was to delineate how Ca 2ϩ can have multiple roles in cell survival and growth. Ca 2ϩ functions directly through regulation of gene expression, e.g. c-fos (8) or via activating Ca 2ϩ -dependent enzymatic action, e.g. calcium/ calmodulin-dependent kinases II and IV in neurons, calcineurin in lymphocytes, and in conjunction with other signaling pathways such as those regulated through MAPK and PI3K * This work was supported by the Hungarian Research Foundation (Grant OTKA T029291) by the Scientific Research Council, Ministry of Health (Grant ETT 206/2001), and by NRDP 1/024, Ministry of Education, Hungary. 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.
(for review see Ref. 7). Another possible mechanism to consider is the activation of signaling pathways similar to those activated by GM-CSF. GM-CSF induces several pathways, e.g. STAT-5, MAPK, PI3K, protein kinase C, and phospholipase C ␥. Because the major route activated by GM-CSF is the MAPK pathway, our primary goal was to determine whether all biological events are the result of this activation or whether additional pathways are responsible.
In T lymphocytes and PC12 cells, elevated Ca 2ϩ was reported to stimulate MAPK activity (9,10). Stimulation of the MAPK cascade causes the phosphorylation and activation of additional downstream targets, such as p90Rsk, CREB, and Elk-1 (11)(12)(13). Activated CREB and Elk-1 can induce the transcription of immediate early genes such as egr-1 and c-fos (14,15) followed by other genes that control apoptotic processes (16,17) or promote cytokine production (18), all modulating proliferation and survival.
In this report we show that agents (CPA and ionomycin) that generate distinct elevation of [Ca 2ϩ ] i from different sources result in the activation of the MEK/ERK/Elk-1 pathway. Ionomycin induces similar cellular responses as GM-CSF, the natural growth factor of TF-1 cells. In both cases activation of the MEK/ERK/Elk-1 displayed similar kinetics and resulted in cell survival and proliferation. In contrast, CPA induces a delayed MEK/ERK/Elk-1 activation, sufficient only to trigger cell survival. These results point to the importance of the kinetics of the Ca 2ϩ signal.
Taken together we report that the increase of [Ca 2ϩ ] i by activating the MAPK/ERK results in an escape of CD34-positive myelo-erythroid leukemia cells from apoptosis induced by hormone-deprivation. These results suggest that an elevation of [Ca 2ϩ ] i may rescue hemopoietic progenitor cells from apoptosis induced by hormone deprivation in the bone marrow. These effects may protect normal progenitor cells but may also lead to the development of pathological hematopoiesis.

EXPERIMENTAL PROCEDURES
Chemicals and Reagents-Unless otherwise specified, all chemical reagents were purchased from Sigma. The tissue culture media were purchased from Invitrogen, whereas GM-CSF, IL-3, IL-4, PD98059 (MEK1 inhibitor), and UO126 (MEK1/2 inhibitor) were purchased from New England BioLabs.
Cell Line, Culture Conditions, and Stimulation-The factor-dependent human myelo-erythroid leukemia cell line, TF-1 (3), was kindly provided by Dr. C. Braun, Heinrich Pette Institute for experimental Virology and Immunology (Hamburg, Germany). Cells were grown in RPMI medium without nucleosides, supplemented with 10% (v/v) fetal calf serum and 2.5 ng/ml GM-CSF. In all experiments prior to stimulation cells were maintained overnight in GM-CSF-free medium. The addition of the inducers was considered as the t ϭ 0 h of the experiments. Cell viability was determined by trypan blue exclusion and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide incorporation (19) or FACS analysis (propidium iodide) assay (20).
Cell Cycle Analysis-Cell cycle analysis by propidium iodide staining was performed according to a previous study (21). Briefly, 10 6 cells were washed with phosphate-buffered saline, fixed overnight in 70% ethanol, washed, and incubated with 100 g/ml RNase (Sigma) and 10 g/ml propidium iodide in phosphate-buffered saline for 30 min at 37°C. Cells were analyzed for DNA content by a FACSCalibur (BD Biosciences) flow cytometer. Data were analyzed by ModFit software.
Preparation of Cell Extracts and Analysis by Western Blotting-Cells were lysed as described previously (2). 50 g of lysed cell protein was separated by SDS-PAGE gel and transferred to PVDF membrane using the Mini-Protean II system (Bio-Rad). The primary antibodies anti-(Egr-1), anti-(c-Fos), anti-(ERK1), anti-(PARP), anti-(Bcl-xl), anti-(Bax) rabbit polyclonal IgG and anti-(phospho-ERK), anti-(phospho-Elk-1), and anti-(Bcl-2) mouse monoclonal IgG were purchased from Santa Cruz Biotechnology. Anti-rabbit IgG horseradish peroxidase conjugate and anti-mouse IgG horseradish peroxidase conjugate were obtained from Jackson ImmunoResearch as secondary antibodies. Membranes stained at first, with monoclonal antibodies anti-(phospho-ERK), and anti-(phospho-Elk-1), were incubated in TBS/0.1% Tween 20 containing 0.5% H 2 O 2 at room temperature for 30 min, than washed three times with TBS/0.1% Tween 20, and developed by the polyclonal anti-ERK1 antibody. The ECL system was used for chemiluminescence detection (Amersham Biosciences). For quantitative analysis of Western blots, image analyses of x-ray films were performed using Bioscan version 1.0 software following digitalization with a Hewlett Packard 5100C scanner.
Preparation of Nuclear Extracts and Electromobility Shift Assays-DNA-binding proteins were extracted, and the band-shift assay was performed by applying 5 g of protein and a radiolabeled consensus oligonucleotide probe as reported previously (22). Santa Cruz Biotechnology produced the oligonucleotide probe (AP-1).
Electrophoretic Analysis of Oligonucleosomal DNA Degradation (Apoptosis)-In experiments addressing the onset of apoptosis during GM-CSF withdrawal and after the re-administration of GM-CSF, ionomycin, or CPA, TF-1 cells were seeded (5 ϫ 10 5 cells/ml) into RPMI 1640/10% FCS for 16 h, than supplemented with 2.5 ng/ml GM-CSF, 1 M ionomycin, or 7.5 M CPA. After various incubation times, total cellular DNA was isolated from 5 ϫ 10 6 cells, based on the method described previously (23), and loaded into the dry wells of 1.8% agarose gel containing 0.1 g/ml ethidium bromide.
Caspase-3 Activity Determinations-Cells were treated with the test substances as described in the figure legends. At various time points 3 ϫ 10 6 cells were harvested and cell extracts were prepared by the freeze-thawing method in the following lysis puffer: 20 mM HEPES-NaOH, pH 7.4, 50 mM NaCl, 3 mM MgCl 2 , 10 mM ␤-mercaptoethanol, 0.4% Nonidet P-40, 2 mM phenylmethylsulfonyl fluoride, 100 M Na 3 V0 4 , 10 g/ml aprotinin, 10 g/ml leupeptin, and 10 g/ml antipain. The protein content of the samples was determined using the Bradford dye-assay (reagent obtained from Sigma), and aliquots containing 50 g of protein were used in a fluorometric caspase-3 enzyme activity assay, carried out in a microplate reader (24). In a final volume of 100 l, containing 10 mM HEPES-KOH, pH 7.5, 10% sucrose, 1 mM dithiothreitol, 0.025% CHAPS, and 50 M Ac-DEVD-AMC (acetyl-Asp-Glu-Val-Asp-7-amido-4-methylcoumarin) as substrate, the liberation of AMC was followed in time (excitation at 360 nm and emission at 460 mm). The fluorescence intensities were converted into AMC concentrations using an AMC calibration curve. Activities were standardized to the caspase-3 activity obtained in GM-CSF-deprived cells (3.3 Ϯ 0.5 M ACM calculated from three independent experiments).
[Ca 2ϩ ] i Measurements-Cells (0.5-1 ϫ 10 6 cells/ml) were loaded for 30 min at 37°C with 1.5 M fura-2AM in fresh culture medium containing 1% FCS. Than extracellular fura-2AM was removed by centrifugation, and the cells were resuspended in fresh culture medium containing 10% FCS. Cells were treated for various incubation times with the test substances, 2.5 ng/ml GM-CSF, 1 M ionomycin, or 7.5 M CPA. In certain experiments 7.5 M CPA, 1 mM ATP, 1 unit/ml thrombin, 250 ng/ml trypsin, 5 units/ml erythropoietin, 5 ng/ml IL-4, 5 ng/ml IL-3, and 10 ng/ml nerve growth factor were added to untreated cells. Fluorescence was measured in a Hitachi F4000 fluorescence spectrophotometer at 37°C under gentle agitation, as described previously (25). Cytoplasmic free calcium concentration was calculated by using the method of Tsien et al. (26). The presence of extracellular fura-2AM fluorescence was checked by the addition of EGTA (0.5 mM), followed by CaCl 2 (1 mM), to the control cells. The leakage of fura-2AM proved to be small during the experimental period. Nevertheless, this leakage was taken into consideration in the calibration procedure.

Effect of Intracellular Ca 2ϩ
Level on Cell Survival, Proliferation, and Apoptosis-TF-1 cells were synchronized by GM-CSF hormone deprivation and thereafter treated with CPA (7.5 M), ionomycin (1 M), or GM-CSF (2.5 ng/ml). These concentrations were chosen after optimization, considering the viability of treated cells.
The effects of the above reagents on the cell growth were determined (Fig. 1A). The GM-CSF deprived (control) cells did not multiply, and within 72 h their cell number gradually decreased to about 26% of the original. When GM-CSF was re-administered after a leg period of about 24 h, the hormonedeprived cells started to grow exponentially. Interestingly, the addition of ionomycin also induced a gradual increase in the cell number, producing a cell growth of about 250% in the first 48 h. Addition of CPA did not increase significantly the cell number, but prevented its decline, relative to the control cells.
As documented in Fig. 1B, the viability of the control (hormone-deprived) cells decreased to about 45% within 72 h. In contrast, the viability of the GM-CSF-, ionomycin-, or CPAtreated cells did not change notably within this time frame. Measuring the viability by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide incorporation or propidium iodide gave similar results (data not shown).
To test how the traverse of the cell cycle was affected, the DNA content of cells subjected to various treatments was analyzed. Samples of GM-CSF-, ionomycin-, and CPA-treated cells were collected every 3 h during an interval of 33 h. The ratio of cells in the three cell cycle phases, G 1 /G 0 , S, and G 2 /M was determined by flow cytometry analysis by using propidium iodide staining (Fig. 2). These experiments indicated that readministration of GM-CSF forced the majority of cells to enter S phase within 18 h ( Fig. 2A) while ionomycin induced a moderate access of cells to S phase (Fig. 2B). In contrast, CPA treatment caused cell cycle arrest (Fig. 2C). These results, along with the proliferation measurements, reveal that ionomycin, similarly to GM-CSF, induces cell growth.
Both CPA and ionomycin induce store depletion followed by an activated Ca 2ϩ influx. Therefore, we tested whether store depletion and/or an increased level of [Ca 2ϩ ] i were responsible for CPA-induced cell survival. To buffer changes in [Ca 2ϩ ] i , cells were loaded with a relatively high concentration of BAPTA-AM (160 M). High concentration of the chelator was needed due to the inability of lower concentrations to abolish CPA-induced increase in [Ca 2ϩ ] i in the presence of external Ca 2ϩ (not shown). In another approach, extracellular Ca 2ϩ was chelated with 3 mM EGTA. Both treatments caused cell death. BAPTA-AM caused a drastic drop of cell viability, with cells dying within 12 h irrespective of the particular treatment. EGTA decreased the cell number by 75.1 Ϯ 2.1%, 43.9 Ϯ 5.2%, 87.2 Ϯ 3.2%, and 94.6 Ϯ 2.8% of control, GM-CSF-, CPA-, and ionomycin-treated cells, respectively, within 24 h. CPA and ionomycin presumably caused more excessive store depletion, which may account for the more rapid cell death compared with control or GM-CSF-treated cells. Thus, sustained store depletion induces cell death rather than survival, and Ca 2ϩ influx appears to be crucial in CPA-and ionomycin-dependent survival of TF-1 cells.
Because non-excitable cells generally do not endure high elevations of [Ca 2ϩ ] i , the signs of apoptosis were also examined. As shown in Fig. 3A, long term hormone deprivation induced a delayed increase of caspase-3 activity and the appearance of PARP cleavage product (Fig. 3B). A marked DNA ladder formation (Fig. 3C) also appeared within 24 h. In contrast, GM-CSF, CPA, and ionomycin treatment suppressed caspase-3 activity to the level observed in the normally cultured cells (41.5 Ϯ 3.65%) within 6 h. Furthermore, no cleavage of PARP or DNA ladder formation was detected even after 48 h (Fig. 3). These results, reinforced by the viability measurement, show that the treatments described above protected the cells from apoptosis.
As a summary, hormone-deprived cells stopped growing and lost their viability within 72 h. GM-CSF, the natural growth hormone for TF-1 cells, suppressed apoptotic processes and Certain agents, like IL-3, IL-4, erythropoietin, and nerve growth factor, which induce Ca 2ϩ fluxes in other cells, had no effect on intracellular Ca 2ϩ levels in TF-1 cells (data not shown). Thus, based on these experiments, both endogenous and pharmacological agents may modify [Ca 2ϩ ] i in TF-1 cells.
In   (Table I).
All in all, CPA-and ionomycin-induced [Ca 2ϩ ] i alterations were comparable with the effect of endogenous substances. CPA and ionomycin induced controlled [Ca 2ϩ ] i elevation, which was down-regulated within 1 h. GM-CSF, on the other hand, did not increase [Ca 2ϩ ] i in these cells.
Effect of [Ca 2ϩ ] i on the Proteins of the Bcl-2 Family-To identify some of the possible mechanisms allowing hormonedeprived cells to avoid apoptosis, changes in the well-known apoptotic protein family, Bcl-2 were examined. Synchronization by overnight hormone withdrawal was taken as the starting point of all investigations (t ϭ 0 h). Whole cell extracts were analyzed by Western blotting for anti-apoptotic (Bcl-2 and Bclxl) and pro-apoptotic (Bax) proteins. The level of Bcl-xl doubled in 24 h after re-administration of GM-CSF, increasing to 2.5fold by 48 h. Unlike GM-CSF, ionomycin treatment induced only a minor increase in Bcl-xl expression (1.5-fold) by 24 h, without further augmentation. Bcl-xl expression remained unaltered in CPA-treated cells (Fig. 5A). In contrast, GM-CSF, ionomycin, and CPA invariably triggered a comparable expression of Bcl-2, each doubling its level. However, the rate of expression induced by CPA was slower, reaching the level detected in GM-CSF-treated cells with a 24-h delay (Fig. 5B). None of these treatments yielded any change in the level of the pro-apoptotic protein, Bax (Fig. 5C). The finding that the antiapoptotic proteins dominated the apoptotic proteins correlated with the viability measurements (Fig. 1B), suggesting that the regulation of the Bcl-2 family plays a major role in the survival of TF-1 cells. Meanwhile, among the gradually dying control cells (Fig. 1B), a moderate decrease in the level of Bcl-xl and Bcl-2 (Fig. 5, A and B) was seen, accompanied by a small increase of Bax (Fig. 5C).
To summarize, controlled elevation of [Ca 2ϩ ] i induced the up-regulation of the Bcl-2 anti-apoptotic protein without the alteration of other proteins of the Bcl-2 family. This Bcl-2 up-regulation enabled TF-1 cells to evade apoptosis.
Level and Time Course of ERK1/2 Phosphorylation Induced by a Variety of Agents-The activation of the MAPK pathway is one of the major routes leading to cell survival and/or proliferation. Specific monoclonal anti-pERK antibodies enabled us to follow the kinetics and level of ERK1/2 kinase phosphorylation. As shown in Fig. 6A, in a good correlation with the [Ca 2ϩ ] i signal, a rapid activation of ERK1/2 was observed 5 min following treatment with the [Ca 2ϩ ] i -increasing agents. Thus, CPA resulted in a minor stimulation, whereas ionomycin concentration above 500 nM induced marked phosphorylation. Below this ionomycin concentration no activation of ERK1/2 was detected. Elk-1 activation showed a correlation with ERK1/2 stimulation. As revealed, a threshold of ERK1/2 phosphorylation has to be reached for Elk-1 activation. This finding ex-plains the activation of Elk-1 only above 500 nM ionomycin and the failure of activation by CPA.
In terms of the physiological outcome, the kinetics of signaling is probably as critical as the extent of ERK1/2 phosphorylation. To better characterize the activation processes, the time course of ERK1/2 phosphorylation was studied.
Two phases of phosphorylation were observed in the case of GM-CSF, ionomycin, and CPA. The first appeared 5 min after the exposure to these agents, and the second followed between 2 and 3 h (Fig. 6B). At 5 min, CPA elicited only a moderate activation, while ionomycin and GM-CSF caused 13-and 28fold elevations, respectively. In all cases, phosphorylation returned to control levels within 1 h. Intriguingly, a second 11-fold elevation appeared at 2 h in the case of ionomycin, whereas GM-CSF and CPA induced only a 7-fold increase later, at 3 h. Elk-1 phosphorylation followed ERK1/2 phosphorylation

FIG. 5. Western blot analyses of Bcl-xl (A) and Bcl-2 (B) antiapoptotic proteins and Bax (C) pro-apoptotic protein from whole cell extracts of GM-CSF-, ionomycin-, CPA-treated and control cells.
Cells were supplemented with GM-CSF, ionomycin, or CPA or not supplemented, and cellular extracts were made at the time points indicated. 40 g of total cellular protein was run on 12.5% SDS-polyacrylamide gel, transferred to a PVDF filter, and analyzed by immunoblotting using antibodies against Bcl-xl (A), Bcl-2 (B), and Bax (C) proteins. Results represent the mean Ϯ S.E. of three independent experiments normalized to the protein levels of 16 h GM-CSF-deprived cells. Protein loading was standardized to ERK1/2 levels. when ERK1/2 activation reached the critical level, described above (Fig. 6C). As expected, Elk-1 phosphorylation did not take place in control cells. Elk-1 phosphorylation followed similar kinetics as ERK1/2 in treated cells. GM-CSF induced a 40-fold elevation of Elk-1 phosphorylation at 5 min, followed by a 4-fold second peak 3 h later. Ionomycin induced phosphorylation at the same time intervals as GM-CSF, causing 9-and 4-fold increases, respectively. As expected, CPA induced only the second peak of 9-fold phosphorylation at 3 h, where the phosphorylation of ERKs surpassed the threshold levels. No change in phosphorylation level was detected in control cells within 6 h (data not shown).
To test the effect of extracellular Ca 2ϩ on ERK1/2 and Elk-1 activation, cells were pre-treated for 5 min with 3 mM EGTA before the addition of CPA, ionomycin, or GM-CSF. Samples were taken when phosphorylation of ERK1/2 and Elk-1 reached its maximum level. Fig. 6D shows that EGTA pretreatment abolished completely the phosphorylation of ERK1/2 and Elk-1 induced by the [Ca 2ϩ ] i -increasing agents, CPA or ionomycin, but did not affect the receptor-mediated signal induced by GM-CSF.
In summary, GM-CSF induces two phases of activation, an extensive initial ERK1/2 and Elk-1 activation, followed by a smaller second peak. Because GM-CSF did not promote a detectable calcium signal, it is highly probable that in this case the induction of phosphorylation proceeds through different pathways. [Ca 2ϩ ] i -increasing agents also induced two phases of ERK1/2 activation, although in the case of CPA the first peak of activation was insufficient for Elk-1 activation. Considering that both ionomycin and GM-CSF induces not only cell survival but also proliferation, the primary, or rather the biphasic, activation of the ERK1/2-Elk-1 pathway may serve as a key signal for proliferation. The experiments with EGTA pre-treatment highlighted the fundamental role of Ca 2ϩ influx in ERK1/2 phosphorylation induced by [Ca 2ϩ ] i -increasing agents.

Effect of [Ca 2ϩ ] i on the Expression of c-Fos and Egr-1 Transcription Factors and on AP-1 DNA Binding-We have examined the expression of two transcription factors, c-Fos and
Egr-1, that can be stimulated in an Elk-1-dependent manner and are thought to be involved in survival and proliferation of hematopoietic cells. According to our experiments, in both GM-CSF-and ionomycin-treated TF-1 cells mRNA (data not shown) and protein levels (Fig. 7A) of Egr-1 and c-Fos were elevated after 30 and 60 min, respectively. As discussed above, although rapid and strong initial activation of ERK1/2 was observed with GM-CSF and ionomycin, CPA induced only a faint phosphorylation at the same time (Fig. 6B) and could not promote Elk-1 phosphorylation (Fig. 6C). In accordance with the notion that Elk-1 is the major stimulator of c-Fos and Egr-1 expression, none of these were expressed within 1 h of CPA treatment. Accordingly, the activation observed by 3 h of CPA treatment appeared strong enough to drive Elk-1 phosphorylation and trigger the expression of the transcription factors examined (Fig. 7B). As expected, c-Fos and Egr-1 expression did not take place in control cells.
We have examined whether the increased c-Fos expression is accompanied by the increased DNA binding of AP-1. Our results provide evidence that the elevated c-Fos expression manifests in the elevation of AP-1 DNA binding (Fig. 7C).

Effect of MEK Inhibition on the Phosphorylation of ERK1/2 and Elk-1 and on the Expression of c-Fos and Egr-1 Transcription Factors-
The specific MEK-1 inhibitor, PD98059, abolished the ionomycin-and GM-CSF-induced ERK1/2 and Elk-1 phosphorylation as well as the following expression of Egr-1 and c-Fos (Fig. 7A). Similarly, it suppressed the CPA-induced phosphorylation of ERK1/2 and Elk-1 and the expression of the above transcription factors. This difference was most apparent at 3 h, where phosphorylation reaches its maximal level in CPA-treated cells (Fig. 7B). In cells treated with [Ca 2ϩ ] i -increasing agents, the increase in the AP-1 DNA binding level can also be blocked by PD98059 (Fig. 7C). Similar results were obtained by using another MEK1/2 inhibitor, UO126 (data not shown). Altogether, elevation of [Ca 2ϩ ] i is able to trigger c-Fos and Egr-1 expression via the MEK-ERK1/2-Elk-1 pathway.
Effect of MEK-1 Kinase Inhibition on Ca 2ϩ -dependent Survival-To elucidate the role of the MAPK pathway in TF-1 cell survival and proliferation, the long term effect of MEK-1 inhibition was investigated. Table II shows that inhibition of MEK by PD98059 significantly reduced the CPA-or ionomycin-induced Bcl-2 expression measured at 12 h after treatment. PD98059 slightly reduced the weak Bcl-2 level in control cells also, demonstrating that Bcl-2 expression is MEK/ERK-dependent. After 48 h, inhibition by PD98059 caused a drastic drop in cell number and viability that was more pronounced in cells treated with ionomycin than CPA. This could be explained by the prominent elevation of [Ca 2ϩ ] i of ionomycin-treated cells. In contrast, cell survival was not affected in GM-CSFtreated cells, only the rate of proliferation was attenuated. This suggests that in GM-CSF-treated cells the MAPK pathway is mainly critical for proliferation but not for survival, whereas Ca 2ϩ -induced survival principally goes through the MAPK route.

DISCUSSION
In this report we have demonstrated that a controlled increase of [Ca 2ϩ ] i by CPA or ionomycin can induce cell survival and the latter agent stimulates proliferation in a human hemopoietic cell line, TF-1 (Fig. 1). Previous studies have implicated that, in hormone-dependent hemopoietic mouse cell lines (Baf3, FDCP.P2, and IC-2), calcium ionophores and SERCA pump inhibitors may moderate hormone-deprivation induced cell death (27)(28)(29)(30). However, even in these cases, within 24 h cell viability fell significantly. In contrast, we found that TF-1 cells exposed to CPA or ionomycin showed no deterioration of cell viability for up to 72 h.
In search of the mechanisms that lead to cell survival, we have examined the expression pattern of the Bcl-2 family. We have found that both [Ca 2ϩ ] i -raising treatments caused the elevation of Bcl-2 level, without altering the Bcl-xl or Bax levels (Fig. 5). It has been reported that ectopic expression of Bcl-2 in TF-1 cells prevents apoptosis without changes in morphology or phenotype in the absence of GM-CSF (4). Thus, it is plausible that in TF-1 cells the elevation of Bcl-2 level induced by [Ca 2ϩ ] i is sufficient to repress the apoptotic processes without any changes of Bcl-xl levels. This means that Bcl-xl is not essential in promoting survival of TF-1 cells after hormone withdrawal.
The Bcl-2 gene is known to be regulated in an MAPK-dependent manner (16,17). We have shown here that both CPA and ionomycin (Fig. 6) activate the MAPK pathway, suggesting that the increase of Bcl-2 is a result of MAPK activation. Indeed, the inhibition of MEK-1 by PD98059 decreased the level of Bcl-2 expression measured at 12-h treatment. Longer exposure to PD98059 and [Ca 2ϩ ] i -increasing agents caused such a drastic drop in cell number that Bcl-2 expression could not be tested. In parallel to [Ca 2ϩ ] i -raising agents, as shown in an earlier investigation (2), re-administration of GM-CSF causes MAPK activation and the augmentation of Bcl-2 with a raise in Bcl-xl levels.
It is widely accepted that both CPA and ionomycin induce Ca store depletion followed by Ca 2ϩ influx. We demonstrated that store depletion, induced by either an intracellular (BAPTA) or by an extracellular Ca 2ϩ chelator (EGTA) is toxic, cells die rapidly. This is consistent with the notion that store depletion induces cell death rather than survival. It has been demonstrated that the release of Ca 2ϩ from endoplasmic reticulum leads to apoptosis in Chinese hamster ovary (31) and in SH-SY5Y neuronal cells (32). As demonstrated in this work short incubation with EGTA blocked Ca 2ϩ influx and ERK1/2/Elk-1 activation induced by CPA or ionomycin. As also shown, inhibition of ERK activation abolished cell survival. Thus, the activated Ca 2ϩ influx and moderate increase in [Ca 2ϩ ] i are involved in the pro-survival mechanisms of [Ca 2ϩ ] i -increasing agents in TF-1 cells. Similar results were obtained in excitable neuronal cells, where apoptosis could be blocked by a sustained increase of steady state free Ca 2ϩ (33).
It is worth mentioning that no changes in the level of expression of the sarco-endoplasmic reticulum calcium ATPase and plasma membrane Ca 2ϩ ATPase proteins were observed after 24 h of CPA or ionomycin treatments. Moreover, these cells do . Cells were treated, and when indicated were exposed 30 min previously to MEK-1 inhibitor, PD 98059, as described under "Experimental Procedures." 50 g of total cellular protein was separated by electrophoresis on 10% SDS -PAGE, transferred to PVDF filter, and analyzed by immunoblotting using monoclonal antibodies against p-ERK1/2, p-Elk-1, and polyclonal ERK1/2 protein. Phospho-protein levels were detected from cells 5 min following exposure to drugs (A) or as indicated (B). Nuclear extracts of ionomycin-or CPA-treated cells, with or without inhibition by PD 98059, were quantified for AP-1 DNA binding activity by means of electrophoretic mobility shift assay at the times indicated. Specific DNA binding was evaluated with an excess of unlabeled oligonucleotide probe for AP-1 (C). For more details see "Experimental Procedures." not display any detectable Na/Ca exchanger activity. 2 Therefore, it is unlikely that these Ca 2ϩ transporters are involved in CPA or ionomycin induced cell-survival.
Another finding of our investigation was that transient, high elevation of [Ca 2ϩ ] i can trigger the proliferation of hemopoietic cells through the activation of ERK1/2. The time course of ERK1/2 activation determines the fate of several cell responses, including cell proliferation (34). We have determined the delicately controlled time course of ERK1/2 activation in TF-1 cells in response to GM-CSF, ionomycin, and CPA exposure (Fig. 6B). GM-CSF and ionomycin prompted cells to enter S phase (Fig. 2, A and B) and cell division (Fig. 1A) and caused a biphasic ERK1/2 activation. With CPA there is only one significant peak of ERK1/2 activation, which is sufficient to promote cell survival, but falls short of making cells re-enter the cell cycle. This observation correlates well with the recent proposal of Jones and Kazlauskas (35) that growth factor-dependent mitogenesis requires two distinct phases of signaling in NIH3T3 cells. The occurrence of the double phases of ERK1/2 phosphorylation under our experimental conditions may be a result of the transient activation of the MAPK phosphatases, which are regulated by ERKs and/or calcium (36). Interestingly, PD98059 was able to selectively inhibit the second peak also, 2 demonstrating that the activation of both peaks were MEK-dependent.
The expression of c-Fos and Egr-1 proteins, reportedly involved in cell proliferation, are regulated through Elk-1 via SRE and other pathways, e.g. CRE and sis inducible element (38). Furthermore, increase in nuclear Ca 2ϩ has been reported to control CRE-mediated transcription, whereas a rise of cytoplasmic Ca 2ϩ is believed to activate SRE-driven transcription (39). We have found that c-Fos and Egr-1 expression, like Elk-1 activation, is induced by [Ca 2ϩ ] i elevation and is MEK-1-dependent (Fig. 7). This suggests that the expression of these proteins is regulated mainly through Elk-1-dependent mechanisms, although the role of CREB cannot be ruled out. Interestingly, GM-CSF induced only a minor elevation of c-Fos, in comparison to ionomycin and CPA. This may be explained by the involvement of Ca 2ϩ -dependent intragenic regulatory elements involved in c-Fos regulation (40). Because GM-CSF does not induce [Ca 2ϩ ] i elevation, it is unlikely that these regulatory elements become activated.
It has been reported that Egr-1 is important in regulating cell growth, differentiation, and development (41), depending on the cell type. In our investigation, elevation of Egr-1 expression correlated with cell growth. The highest level of Egr-1 expression was in the case of GM-CSF treatment, followed by that caused by ionomycin. CPA only resulted in a low Egr-1 level, and no Egr-1 protein was detected in control cells.
The activation of the MEK-1/ERK1/2/Elk-1 pathway and the following transcription of immediate early genes (c-Fos and Egr-1) could be completely blocked by the MEK-1 inhibitor PD98059 in all treatments applied here. This inhibition in the GM-CSF-treated cells caused an extensive reduction of proliferation rate without any change in cell viability, indicating that PD98059 is not toxic for TF-1 cells in the concentration used. This implies that the MAPK/ERK pathway is central for promoting proliferation in GM-CSF-treated cells, and activation of other pathways is mostly involved in cell survival. In the case of cells treated with [Ca 2ϩ ] i -increasing agents, PD98059 induced a drastic drop of cell number and viability, which was most significant in ionomycin-treated cells. This can be attributed to the toxicity of high elevation of [Ca 2ϩ ] i caused by ionomycin in the absence of MAPK/ERK activation.
It has been suggested (28,29) that the limited suppression of apoptosis in different hemopoietic mouse cells induced by [Ca 2ϩ ] i -increasing agents may be a consequence of autocrine IL-4 production. In our experiments TF-1 cells survive hormone deprivation in the presence of IL-4 (data not shown) in a good agreement with the results of Kitamura et al. (3). Despite this, no production of IL-4 was detected by enzyme-linked immunosorbent assay, neither was any increase in the DNA binding of nuclear factor of activated T cells, 2 which is involved in the induction of IL-4 expression (42). Nonetheless, the presence of cytokine production cannot be ruled out.
In this work we have demonstrated that Ca 2ϩ induces a unique, MEK-dependent activation of ERK/Elk/c-Fos/AP-1 pathway in CD34-positive myelo-erythroid leukemia cells. Elevation of Ca 2ϩ is known to stimulate proliferation-associated MAPKs in different non-hemopoietic cells. In vascular smooth muscle a Ca 2ϩ -dependent isoform of protein kinase C and calmodulin kinase II are upstream activators of ERK1/2 (43). In rat fibroblast functional calcium-sensing receptors are required for activation of signal-regulated kinase kinase and MAPK in response to extracellular calcium (44). Calmodulin binding to an IQ motif in Ras-guanine nucleotide releasing factor 2 exchange factor in kidney epithelial cells stimulates GTP binding by Ras and potentiates elevated Ca 2ϩ concentration-induced ERK1/2 activation (45). Similar calmodulin-dependent Ras-GRF activation was suggested in the regulation of neuronal function by Ca 2ϩ signals (46), but the role of a protein kinase A-dependent activation of the small G-protein Rap-1 can not be ruled out (47). Moreover, the requirement of the Ca 2ϩ -activated proline-rich tyrosine kinase 2 for the activation of the MAPK cascade induced by Ca 2ϩ in PC 12 cells (48) or in pulmonary vein endothelial cells (49) was also reported. Any of the above mentioned pathways may be involved in the Ca 2ϩ -induced TABLE II Effect of the MEK-1 inhibitor, PD98059, on cell proliferation, viability, and Bcl-2 level Proliferation and cell viability was determined by Trypan-Blue exclusion test, following 48 h of treatment. Results of cell growth were normalized to the initial cell numbers (100%). Bcl-2 expression was analyzed by immunoblotting using an antibody against Bcl-2 protein after 12 h of treatment. Results were normalized to the protein levels of 16-h GM-CSF-deprived control cells. Protein loading was standardized to ERK1/2 levels. Treatments were as described under "Experimental Procedures." Data represent at least three independent experiments (mean Ϯ S.D. ] i caused by CPA is within the range of physiological calcium signals, as confirmed by ATP, thrombin, or trypsin treatments of TF-1 cells (Fig. 4). ATP and thrombin are known cellular modulators, and trypsin was found to be expressed at low levels in vascular endothelial cells and leukocytes (50) and overexpressed in certain cancer cells and tissues (51,52). All these observations reinforce our hypothesis that increases in [Ca 2ϩ ] i similar to those caused by CPA or ionomycin can occur in vivo.
In summary, we have shown that the elevation of [Ca 2ϩ ] i in a human hemopoietic cell line induced the activation of the MEK/ERK/Elk/c-Fos/AP-1 pathway. This unique activation led to hormone-independent survival and limited proliferation. Moreover, we showed that the kinetics of calcium signaling influenced the kinetics of ERK1/2 activation and cellular response.
Our results point toward a possible mechanism through which bone marrow cells may escape from hormone-deprivation induced cell death. The [Ca 2ϩ ] i -induced, hormone-independent survival may be leading to clone selection and leukemogenesis. Accordingly, a report on a Phase I clinical trial of carboxyamidotriazole, a synthetic inhibitor of non-excitable calcium channels, suggested that carboxyamidotriazole reversibly inhibits angiogenesis, tumor cell proliferation, and metastatic potential of refractory solid tumors (53). Other pharmacological agents, inhibiting receptor-linked Ca 2ϩ entry in vitro, might slow prostate cancer proliferation in vivo (54). Thus, elevation of [Ca 2ϩ ] i may not only facilitate the development of leukemia but in certain cases may play a role in solid tumor development and progression.