Fortilin binds Ca²⁺ and blocks Ca²⁺-dependent apoptosis in vivo

Abstract

Fortilin, a 172-amino-acid polypeptide present both in the cytosol and nucleus, possesses potent anti-apoptotic activity. Although fortilin is known to bind Ca²⁺, the biochemistry and biological significance of such an interaction remains unknown. In the present study we report that fortilin must bind Ca²⁺ in order to protect cells against Ca²⁺-dependent apoptosis. Using a standard Ca²⁺-overlay assay, we first validated that full-length fortilin binds Ca²⁺ and showed that the N-terminus (amino acids 1–72) is required for its Ca²⁺-binding. We then used flow dialysis and CD spectropolarimetry assays to demonstrate that fortilin binds Ca²⁺ with a dissociation constant (K_d) of approx. 10 μM and that the binding of fortilin to Ca²⁺ induces a significant change in the secondary structure of fortilin. In order to evaluate the impact of the binding of fortilin to Ca²⁺ in vivo, we measured intracellular Ca²⁺ levels upon thapsigargin challenge and found that the lack of fortilin in the cell results in the exaggerated elevation of intracellular Ca²⁺ in the cell. We then tested various point mutants of fortilin for their Ca²⁺ binding and identified fortilin(E58A/E60A) to be a double-point mutant of fortilin lacking the ability of Ca²⁺-binding. We then found that wild-type fortilin, but not fortilin(E58A/E60A), protected cells against thapsigargin-induced apoptosis, suggesting that the binding of fortilin to Ca²⁺ is required for fortilin to protect cells against Ca²⁺-dependent apoptosis. Together, these results suggest that fortilin is an intracellular Ca²⁺ scavenger, protecting cells against Ca²⁺-dependent apoptosis by binding and sequestering Ca²⁺ from the downstream Ca²⁺-dependent apoptotic pathways.

INTRODUCTION

Fortilin is a 172-amino-acid polypeptide originally identified in our laboratory as a molecule that specifically interacts with MCL1 (myeloid cell leukaemia sequence 1) [1], a member of the Bcl-2 family of anti-apoptotic molecules [2]. Fortilin is also known as TCTP (translationally controlled tumour protein).

The amino acid sequence of fortilin is highly conserved among species ranging from human to rice [3]. Its message is distributed ubiquitously in normal tissue, most abundantly in the liver and kidney [3]. Fortilin is present in both the nucleus and cytosol [3], inducible by serum stimulation [4] and heavy metals [5], and expressed at much higher levels in cancerous cell lines than in non-cancerous cell lines [3]. Fortilin is overexpressed in a variety of human malignancies, including liver, thyroid, laryngeal, skin, uterine, breast, ovarian, prostate and rectal cancers [6]. Fortilin has a potent anti-apoptotic function: its overexpression protects HeLa [3] and U2OS [7] cells against apoptosis. In addition, the depletion of fortilin from cells induces spontaneous cell death in MCF-7 [3] and U937 cells [6]. Although fortilin is likely to make cells cancerous and cancerous cells more resistant to chemotherapy by preventing these cells from undergoing apoptosis, the exact mechanism by which fortilin blocks apoptosis remains unknown.

That fortilin binds calcium was originally shown by Haghighat and Ruben in 1992 [8]. In that work, a large quantity of homogenate of Trypanosoma brucei, the parasitic cause of West African sleeping sickness, was fractionated and probed by ⁴⁵Ca²⁺ in a Ca²⁺-overlay assay to identify Ca²⁺-binding proteins. One of the proteins shown by the Ca²⁺-overlay assay to bind Ca²⁺ was a 22-kDa protein, whose N-terminal sequence was found to be identical with that of T. brucei fortilin [8]. The fact that fortilin binds Ca²⁺ has since been validated by Ca²⁺-overlay assays in several laboratories including those of Sanchez et al. [9], Kim et al. [10], Rao et al. [11] and Arcuri et al. [12]. In 1999, Xu et al. [13] reported that Ca²⁺ up-regulated fortilin transcription, suggesting a role for fortilin in intracellular calcium homoeostasis. However, the biochemistry of the fortilin–Ca²⁺ interaction has not been extensively studied. In addition, the exact biological consequence of fortilin binding to Ca²⁺ remains unknown. Furthermore, it is unknown whether fortilin is required to bind Ca²⁺ in order to block the Ca²⁺-dependent apoptosis.

The cytosolic concentration of free Ca²⁺ is maintained at very low levels (∼100 nM) by the continuous pumping of Ca²⁺ out of the cytosol into the ER (endoplasmic reticulum) by SERCAs (sarcoplasmic/endoplasmic reticulum Ca²⁺-ATPases) [14]. This explains why the ER, among all organelles, contains the highest concentration of Ca²⁺ and why the inhibition of SERCAs, for example by thapsigargin, immediately and drastically elevates Ca²⁺ levels in the cytosol. Even slight elevation of the Ca²⁺ concentration beyond the tightly regulated range could lead to the most drastic biological phenotype, death of the cell by apoptosis. Cytosolic Ca²⁺ beyond the normal range could attack and injure mitochondrial membranes, leading to the release of pro-apoptotic molecules such as cytochrome c and AIF (apoptosis-inducing factor) [14,15], and to programmed cell death. Ca²⁺ can also induce apoptosis by activating pro-apoptotic Ca²⁺–CaM-dependent enzymes [16–18], Ca²⁺-dependent endonucleases [19], Ca²⁺-binding cysteine proteases [20–22], calcineurin [23] and Ca²⁺-sensitive NO synthases [14,24]. Once Ca²⁺ levels go beyond a critical threshold to activate one of these death-bound pathways, apoptosis is inevitable.

It is possible, however, that there is a cellular mechanism by which the cell sequesters cytosolic Ca²⁺ before it activates the downstream cell-death pathways described above. Based on the fact that fortilin blocks apoptosis [3] and the fact that fortilin binds Ca²⁺[8], we hypothesized that fortilin binds Ca²⁺ in the cytosol, functions as a Ca²⁺ scavenger and sequesterer, prevents Ca²⁺ levels from going beyond the apoptosis threshold and protects cells against Ca²⁺-dependent cell death. We hypothesized that the anti-apoptotic activity of fortilin was due to its binding to, and consequent sequestration from downstream apoptotic pathways of, Ca²⁺ that was released in response to apoptotic stimuli. In the present study we report that fortilin indeed binds Ca²⁺ and scavenges Ca²⁺ released in response to thapsigargin, preventing cytosolic Ca²⁺ levels from increasing and activating Ca²⁺-dependent apoptosis pathways.

MATERIALS AND METHODS

Materials

Calmodulin was a gift from Dr John A. Patkey (The University of Texas Health Science Center at Houston, Houston, TX, U.S.A.). Fortilin peptides were chemically synthesized by GenScript. The sequence of each peptide is shown in Figure 8(A). For all procedures described below, plastic and glass instruments were rinsed several times with 0.1 M HCl and then washed extensively with ultra-pure water to remove contaminating Ca²⁺.

Figure 8. Short peptides of fortilin, fortilin(43–62), fortilin(57–76) and fortilin(127–146), bind Ca²⁺ in equilibrium dialysis assays.

(A) Fortilin peptides covering various regions of fortilin were synthesized. Peptides in the highlighted rows are fortilin peptides that exhibited significant Ca²⁺-binding activity in equilibrium dialysis assays. (B) Equilibrium dialysis of fortilin peptides showed significant binding of fortilin(43–62), fortilin(57–76) and fortilin(127–146) to Ca²⁺. ****P<0.001, ***P<0.005, **P<0.01, *P<0.05, ⁺P=0.05. CPM, counts per min.

Molecular cloning

The cDNA of fortilin and its deletion and point mutants were synthesized by PCR-based strategies as we have described previously [25]. In all cases, the authenticity of cloned constructs was confirmed by automated dideoxynucleotide sequencing (Lark Technologies).

Expression and purification of recombinant fortilin and its mutants

Expression and purification of recombinant fortilin protein and its mutants were performed as previously described [26], using Escherichia coli BL21 cells transfected with pGEX-fortilin (or its mutants; Amersham–Pharmacia) or pQE41-fortilin (Qiagen).

Ca²⁺-overlay assay

Ca²⁺-overlay assays were performed as described previously [9,10,12,13,27]. In brief, proteins were size-fractionated by SDS/PAGE and transferred on to a PVDF membrane (Immobilon-P). The membrane was air­dried, incubated for 2 h at room temperature (25 °C) with ⁴⁵CaCl₂ (MP Biomedicals) in CBB [Ca²⁺-binding buffer; 60 mM KCl, 5 mM MgCl₂ and 10 mM imidazole (pH 6.8)] at a final concentration of 40 μCi/ml, rinsed in CBB, washed in 50% ethanol for 10 min, air-dried, and exposed to a phosphor imager screen overnight. Signals were detected by a Molecular Scanner-FX and Quantity One software system (Bio-Rad), according to the manufacturer's instructions. Proteins transferred on to the membrane were semi-quantitatively assessed by Ponceau S staining.

CD spectropolarimetry

Far-UV CD spectra of GST (glutathione transferase)–fortilin samples (0.36 μg/μl) were collected on a Jasco J715 spectropolarimeter using a 1-mm-path-length quartz cuvette at room temperature. Spectra were recorded from 260 to 195 nm in 0.5 nm steps at 100 nm/min. Data represent the average of 10 independent spectra. All protein samples were passed through a PD-10 column (Sephadex-25; Pharmacia) equilibrated in 0.1 M phosphate buffer (pH 7.4) for desalting. The samples were then passed twice through a Chelex 100 chelating ion-exchange resin column (Bio-Rad). Data were expressed as mean residue ellipticity (degrees cm²/dmol).

Flow dialysis assay

A flow dialysis assay was performed as previously described [28], at 25 °C, using ⁴⁵CaCl₂ and Spectrapore 6 cellulose dialysis membrane (M_r cut-off 1000 Da, Spectrum Medical Industry). The flow rate was 3 ml/min and the response time of the apparatus was about 0.9 min. Initially, 2.25 μCi of ⁴⁵Ca²⁺ was added into 1 ml of the GST–fortilin protein solution in the upper chamber. After the successive titration with unlabelled Ca²⁺, an 800 μl portion of the dialysate fraction (1 ml/tube) which flowed out through the lower chamber (0.66 ml) was taken, and the radioactivity was determined using a Beckmann LS-600C liquid scintillation counter. The resulting Ca²⁺-binding data were analysed by fitting to the Adair equation for the 2-site model:

where y is the number of bound Ca²⁺ (mol/mol-protein), x is the concentration of free Ca²⁺ and K₁ and K₂ are the macroscopic dissociation constants.

Cell, cell lines and culture conditions

U2OS cells and MEFs (mouse embryonic fibroblasts) were maintained in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% FCS (fetal calf serum) and, when appropriate, antibiotics. MEF cells from passages 4–9 were used in all experiments.

Isolation of MEFs

MEFs were isolated as previously described [29–31].

Western blot analysis

Cells were harvested by the direct addition of Laemmli SDS gel loading buffer [1,3]. When appropriate, cells were harvested into RIPA buffer [50 mM Tris/HCl (pH 7.4), 150 mM NaCl, 1% NP40 (Nonidet P40), 0.1% SDS, 0.5% sodium deoxycholate, and protease inhibitors (Complete Protease Inhibitor Cocktail Tablets; Roche Biochemicals)] for the determination of protein concentrations (using the Bradford method; Bio-Rad). Western blot analysis was performed as described previously [3,25], using anti-fortilin and anti-actin (Roche Molecular Biochemicals) antibodies.

In vivo Ca²⁺ release assay

U2OS (1×10⁴) or MEF (2×10⁴) cells were seeded in each well of a 96-well plate. The next day, the cells were transfected with either siRNA (small interfering RNA) against luciferase, a non-mammalian protein from Photinus pyralis (American firefly) (siRNA_luciferase, control), or siRNA against human fortilin (siRNA_fortilin) at final concentrations of 100 nM. Both siRNAs [32,33] were synthesized by Dharmacon Research. Transfection was carried out using the DharmaFECT™ two transfection reagent (Dharmacon) as described previously [1]. The siRNA_fortilin consisted of a mixture of four siRNA duplexes targeting four different regions of fortilin mRNA, namely 5′-AGATGTTCTCCGACATCTA-3′, 5′-CGAAGGTACCGAAAGCACA-3′, 5′-GGGAGATCGCGGACGGGTT-3′ and 5′-GGTACCGAAAGCACAGTAA-3′. At 48 h after transfection, cells were exposed to 5 μM Fura 2/AM (fura 2 acetoxymethyl ester) for 1 h at 37 °C under 5% CO₂, allowed to recover in DMEM with 10% FCS for 20 min, and transferred to assay buffer consisting of HBSS (Hanks balanced salt solution; Cambrex Bioscience) supplemented with 10 mM Hepes, 200 μM Ca²⁺, 0.1% BSA, 2.5 mM probenecid and pluronic-F127 (Molecular Probes). A baseline relative intracellular Ca²⁺ concentration (termed r-[Ca²⁺]_i-base), was calculated by first obtaining a fluorescent signal by excitation at 355 nm and emission at 505 nm [termed RFU_355/505 (where RFU is the relative fluorescence unit)] and a fluorescent signal by excitation at 363 nm and emission at 512 nm (termed RFU_363/512) on a SpectraMax M2 plate reader (Molecular Probes) and then dividing RFU_355/505 by RFU_363/512, as described previously [34]. The Fura 2 fluorescence emission ratio [i.e. the ratio of RFI (relative fluorescence intensity) at an excitation at 355 nm and an emission at 505 nm (RFI_355/505) to an excitation at 363 nm and emission at 512 nm (RFI_363/512)] correlates with intracellular Ca²⁺ concentration ([Ca²⁺]_i) [34]. At zero time, thapsigargin (Sigma) was added to a final concentration of 400 nM. Then, fluorescent signals (RFU_355/505 and RFU_363/512) were obtained every 10 s for the next 10 min. The r-[Ca²⁺]_i at a given time point was calculated by dividing RFU_355/505 by RFU_363/512 minus r-[Ca²⁺]_i-base. The free intracellular Ca²⁺ index represents the area under the curve between zero time and 600 s and was expressed as an arbitrary unit. Thapsigargin, a sesquiterpene γ­lactone derived from the plant Thapsia garganica, is a specific and irreversible inhibitor of the SERCAs, including SERCA1, SERCA2a, SERCA2b and SERCA3 [35]. Conversely, thapsigargin has no effect on plasma membrane Ca²⁺-ATPase, Na,K-ATPase, or other enzymes [35]. In addition, the administration of thapsigargin results in the immediate and drastic elevation of [Ca²⁺]_i [36]. Experiments were performed in quadruplicate and repeated at least three times, with consistent results. After the determination of r-[Ca²⁺]_i, cells were subjected to Western blot analysis as described above.

Cell death assay in the presence and absence of BAPTA-AM [acetoxy-methyl-1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid]

In brief, 1.5×10⁴ U2OS cells were seeded in each well of a 96-well plate and transfected with siRNA_fortilin or siRNA_luciferase as described above. A cell death assay using EthD-1 (ethidium homodimer-1; Molecular Probes) was performed as described previously [37,38], with the following modifications. At 12 h after transfection, cells were treated with 15 μM BAPTA-AM-Ca²⁺-chelator (Molecular Probe) or vehicle (DMSO) for 45 min at 37 °C, washed twice with PBS, challenged with 1 μM thapsigargin or vehicle, and incubated for 18 h at 37 °C. After incubation, cells were washed with PBS and stained with 8 μM EthD-1 for 30 min at room temperature. EthD-1 is excluded from living cells but can cross the compromised plasma and nuclear membranes of dying cells and interact with nucleic acids to give red fluorescence. EthD-1 positivity has been observed in the late-phase of apoptosis or in necrosis [37,38]. Fluorescent signals were obtained by 528 nm excitation and 617 nm emission (RFU_528/617). The EthD-1 index, reflecting cell death rate, was calculated using the following equation:

where [RFU_528/617]_TG0 is the RFU_528/617 from non-TG-treated cells (background) and [RFU_528/617]_saponin is RFU_528/617 from cells treated with 0.1% saponin (Sigma). Typically, EthD-1 indices were calculated for two groups of cells, one treated and one not treated with BAPTA-AM. The EthD-1 indices of BAPTA-AM-treated cells represented Ca²⁺-independent cell death rates, whereas those of non-BAPTA-AM-treated cells represented total cell death rates both Ca²⁺-dependent and -independent. Data from caspase 3 activity assays (Figures 6C and 7C) show that most of thapsigargin-induced Ca²⁺-dependent cell death represents necrosis, without significant activation of caspase 3. For MEF, 1.5×10⁴ MEF_{fortilin+/−} or MEF_fortilin+/+ cells were used. Experiments were normally performed in quadruplicate and repeated at least three times, with consistent results.

Figure 6. Fortilin inhibits thapsigargin-induced Ca²⁺-dependent cell death in U2OS cells.

(A) Morphology of BAPTA-AM-treated and ­untreated U2OS cells transfected with either siRNA_fortilin or siRNA_luciferase. BAPTA-AM-treatment reduces the percentage of dead cells (rounded and detached), especially in siRNA_fortilin-treated cells. Cell death preventable by BAPTA-AM (a potent Ca²⁺-chelator), represented Ca²⁺-dependent cell death. (B) Fortilin protected cells against thapsigargin-induced cell death (column 1 compared with column 3), most of which was Ca²⁺-dependent cell death (CDC). **P<0.01; #, cell death blocked by fortilin. (C) The siRNA_fortilin-treated cells, but not siRNA_luciferase-treated cells, exhibit caspase 3 activation upon thapsigargin stimulation, suggesting that most of the cell death of siRNA_luciferase-treated cells in (B) (columns 3 and 4) represented necrosis (cell death lacking caspase 3 activation). **P<0.01.

Figure 7. Fortilin inhibits thapsigargin-induced Ca²⁺-dependent cell death in MEF cells.

(A) Morphology of BAPTA-AM-treated and -untreated MEF_{fortilin+/−} and MEF_fortilin+/+ cells. BAPTA-AM-treatment reduces the percentage of dead cells (rounded and detached), especially in MEF_{fortilin+/−} cells. Cell death preventable by BAPTA-AM (a potent Ca²⁺-chelator), represented Ca²⁺-dependent cell death. (B) Fortilin protects cells against thapsigargin-induced, Ca²⁺-mediated (BAPTA-AM-preventable), cell death (CDC). ***P<0.005; #, cell death blocked by fortilin. (C) MEF_{fortilin+/−} cells, but not MEF_fortilin+/+ cells, exhibit caspase 3 activation upon thapsigargin stimulation, suggesting that most of cell death of MEF_{fortilin+/−} cells in (B) (columns 3 and 4) represented necrosis (cell death lacking caspase 3 activation). *P<0.05.

Caspase 3 activity assay

Caspase 3 assays for MEFs and siRNA-treated U2OS cells were performed as previously described [3]. In brief, cells were treated with 1 μM thapsigargin or vehicle as described above. Cytosolic proteins were extracted by three cycles of freezing and thawing in hypotonic cell lysis buffer [25 mM Hepes (pH 7.5), 5 mM MgCl₂, 5 mM EDTA, 5 mM dithiothreitol and 0.05% PMSF; all from Sigma]. Cytosolic extracts (20 μg) were added to caspase assay buffer [312.5 mM Hepes (pH 7.5), 31.25% sucrose and 0.3125% CHAPS] with R110 (benzyloxycarbonyl-DEVD-rhodamine 110) as substrates (Molecular Probes). Release of R110 by caspase-3-like activity was quantified, after 2 h of incubation at 37 °C, using a SpectraMax M2 plate reader (Molecular Probes) set to an excitation value of 498 nm and emission value of 521 nm. The results were expressed as relative fluorescence units/μg of protein.

Equilibrium dialysis assay of fortilin peptides and GST–fortilin and its mutants

An equilibrium dialysis assay to assess qualitatively the binding of candidate proteins and peptides to Ca²⁺ was performed as described previously [39–41], with modifications. Briefly, 100 μl each of fortilin peptides, GST, GST–fortilin, its mutants or His₆-tagged fortilin [20 μM in EDB (equilibrium dialysis buffer; 10 mM Mops, pH 7.4 and 100 mM KCl)] was placed in a Tube-O-Dialyzer tube (M_r cut off 1000 Da; Genotech), which was then placed in a beaker containing 200 ml of EDB with 1 μM Ca²⁺ and ⁴⁵Ca²⁺ at 1000 c.p.m./μl. The radioactivity of each 15 μl aliquot was determined in triplicate.

Generation of cells stably overexpressing fortilin and its mutant

U2OS cells were transfected, using FuGENE™ 6 reagent (Roche Molecular Biochemicals), with empty pcDNA4-His-Max (pcDNA4; Invitrogen) or pcDNA4 containing the cDNA encoding either wild-type fortilin (pcDNA4_Fortilin) or its mutant (pcDNA4_FortilinΔ2). FortilinΔ2 contained two point mutations: the glutamate residues at amino acid positions 58 and 60 were both mutated to alanine residues and sometimes referred to as fortilin(E58A/E60A). Transfected cells were selected in medium containing 400 μg/ml Zeocin (Invitrogen) and characterized by Western blot analyses. The resulting cell lines were named U2OS_empty, U2OS_fortilin and U2OS_fortilinΔ2 respectively.

Statistical analysis

Experiments were performed in duplicate, triplicate or quadruplicate to obtain S.D. Variations in flow dialysis were expressed as S.E.M. Differences between two experimental groups were analysed using the Student's t test. P<0.05 was considered statistically significant.

RESULTS

Fortilin binds Ca²⁺ in a Ca²⁺-overlay assay

To test the hypothesis that fortilin binds to Ca²⁺, standard Ca²⁺-overlay assays were performed where proteins of interest were resolved by SDS/PAGE and transferred on to PVDF membranes, which were in turn incubated with ⁴⁵Ca²⁺, washed and exposed to a phosphor imager screen. As shown in Figure 1(A), ⁴⁵Ca²⁺ was not bound at all by albumin (lane 1), was bound robustly by the established Ca²⁺-binding protein calmodulin (lane 2), and was bound by fortilin–His₆ (lane 3). Using the same system, we tested different portions of fortilin for their importance in Ca²⁺ binding. As shown in Figure 1(B), full-length fortilin [GST–fortilin(1–172)], but not BSA or GST alone, bound ⁴⁵Ca²⁺ (lane 3 compared with lanes 1 and 2). In this system, GST–fortilin(1–70), but not GST–fortilin(71–120) or GST–fortilin(121–172), bound ⁴⁵Ca²⁺. These results suggest that fortilin in fact binds Ca²⁺ and that the 70 amino acid residues at the N-terminus of fortilin are critical for the binding of fortilin to Ca²⁺ (Figure 1C).

Figure 1. Ca²⁺-overlay assay shows that fortilin binds Ca²⁺ through its first 70 amino acids.

(A) Ca²⁺-overlay assay of fortilin–His₆. (B) Ca²⁺-overlay assay of GST–fortilin and its deletion mutants. Exactly 10 μg of proteins were size-fractionated by SDS/PAGE and transferred on to a PVDF membrane, which was then incubated with ⁴⁵Ca²⁺, washed and exposed to a phosphor imager screen. Neither albumin nor GST-only bound Ca²⁺, whereas calmodulin (a known Ca²⁺-binding protein) did bind Ca²⁺. In this system, both fortilin–His₆ (A) and GST–fortilin (B) bound Ca²⁺. In addition, GST–fortilin(1–70), but not GST–fortilin(71–120) or GST–fortilin(121–172), bound Ca²⁺ (B and C). *Degradation products.

Flow dialysis assays suggest the presence of two high-affinity Ca²⁺-binding sites in fortilin

In order to characterize further the binding of fortilin to Ca²⁺, flow dialysis assays were performed. Results of three independent measurements were shown in Figure 2(A). GST–fortilin showed a biphasic Ca²⁺ binding, binding to approx. two sites with high affinity and to multiple sites with lower affinity. The high-affinity sites were almost saturated at approx. 10 μM Ca²⁺. Ca²⁺ binding to the lower-affinity sites disturbed the measurement of the high-affinity sites and the Ca²⁺-binding data to the latter were scattered. One of the three experimental results was less scattered in this range, and was subjected to curve-fitting analysis (Figure 2B), which revealed two high-affinity sites with macroscopic dissociation constants of K₁=1.75×10⁻⁵ M (±3.06×10⁻⁶ M) and K₂=7.58×10⁻⁶ M (±1.64×10⁻⁶ M). Although the exact nature of the Ca²⁺ binding to the low-affinity sites is not clear from these experiments, these results again suggest that fortilin binds Ca²⁺ and that there are two high-affinity binding sites with dissociation constants of 7.58 and 17.5 μM.

Figure 2. Flow dialysis confirms that fortilin binds Ca²⁺.

(A) Ca²⁺ binding to GST–fortilin. Ca²⁺ binding was measured at 25 °C by flow dialysis as described in the Materials and methods section. The results of three independent experiments were shown. Concentrations of GST–fortilin were 24.7 μM (○), 18.4 μM (□) and 13.8 μM (●) in 0.1 M NaCl and 20 mM Mops/NaOH (pH 7.0). (B) Ca²⁺ binding to the high-affinity sites of fortilin. One of the results in (A) (24.7 μM GST–fortilin) was quantitatively analysed by fitting to an Adair equation (see the Materials and methods section) revealing two Ca²⁺-binding sites with K₁=1.75×10⁻⁵ M (±3.06×10⁻⁶ M) and K₂=7.58×10⁻⁶ M (±1.64×10⁻⁶ M).

Fortilin changes its secondary structure in the presence of Ca²⁺

To determine whether fortilin changes its secondary structure in the presence of Ca²⁺, we subjected fortilin–His₆ to CD spectropolarimetry analyses in the absence and presence of Ca²⁺. As shown in Figure 3(A), fortilin did change its secondary structure in the presence of 10 μM Ca²⁺, most significantly at a wavelength of ∼200 nm. More important, the secondary structure of fortilin remained the same in the presence of 10 μM, 100 μM and 1 mM Ca²⁺, suggesting that Ca²⁺ saturated fortilin at concentrations of approx. 10 μM or greater (Figure 3B). In addition to the fact that the CD spectrum change again suggests the presence of an interaction between fortilin and Ca²⁺, these results also suggest that the dissociation constant (K_d) of the fortilin–Ca²⁺ interaction is in the vicinity of 10 μM (Figures 3A and 3B), a finding that is consistent with the flow dialysis data (Figure 2B).

Figure 3. The binding of fortilin to Ca²⁺ changes its secondary structure as assessed by CD spectra.

Far-UV CD spectra of GST–fortilin (0.36 mg/ml each) in the presence of no Ca²⁺ and 10 μM, 100 μM, and 1 mM Ca²⁺, were collected and recorded from 260 to 200 nm in 0.5-nm steps at 100 nm/min. The samples had been passed through a Chelex 100 chelating ion-exchange resin column (Bio-Rad) to remove residual Ca²⁺ from samples. Data are expressed as mean residue ellipticity (degrees cm²/dmol). The CD spectra of fortilin changed in the presence of 10 μM Ca²⁺ (A) but changed no further at Ca²⁺ concentrations greater than 10 μM (B), suggesting that Ca²⁺-binding sites on fortilin were saturated by approx. 10 μM Ca²⁺.

siRNA depletion of fortilin augments the thapsigargin-induced increase in [Ca²⁺]_i in U2OS cells

To determine whether the presence of fortilin changes the intracellular Ca²⁺ concentration in vivo, we transfected U2OS cells with either siRNA_fortilin or siRNA_luciferase. Treatment with siRNA_fortilin reduced the intracellular fortilin concentration to undetectable levels (siRNA_fortilin, Figure 4A). Cells were then incubated in the presence of the Ca²⁺-sensor dye Fura 2 and stimulated with thapsigargin to induce the release of Ca²⁺ from the ER into the cytosol. The r-[Ca²⁺]_i in both groups of U2OS cells increased rapidly after thapsigargin stimulation (Figure 4B). Strikingly, in this system, the r-[Ca²⁺]_i of siRNA_fortilin-treated U2OS cells increased significantly more than that of siRNA_luciferase-treated U2OS cells (Figures 4B and 4C). These results suggest that the presence of fortilin blunted the increase in [Ca²⁺]_i in thapsigargin-challenged U2OS cells.

Figure 4. Fortilin blocks the thapsigargin-induced elevation of [Ca²⁺]_i in U2OS cells.

(A) The treatment of U2OS cells with siRNA_fortilin (siRNA against fortilin), but not siRNA_luciferase (siRNA against luciferase, negative control), decreased cellular fortilin levels. (B) [Ca²⁺]_i of U2OS cells treated with either siRNA_fortilin or siRNA_luciferase upon thapsigargin challenge (400 nM). U2OS cells lacking fortilin (siRNA_fortilin) exhibit the exaggerated elevation of [Ca²⁺]_i in response to thapsigargin. See the Materials and methods section for experimental details. (C) Free intracellular Ca²⁺ index, the area under the curve, for U2OS cells treated with siRNA_luciferase, was significantly smaller than that for U2OS cells treated with siRNA_fortilin [160±5.7 and 201.5±4.9 (arbitrary units) for cells treated with siRNA_luciferase and siRNA_fortilin respectively.]. ****P<0.001.

MEF_{fortilin+/−} with lower levels of fortilin exhibit a greater increase in [Ca²⁺]_i upon thapsigargin challenge than MEF_fortilin+/+

To determine whether our observation in thapsigargin-challenged U2OS cells was true in primary culture cells, we compared the r-[Ca²⁺]_i of thapsigargin-challenged MEF_{fortilin+/−} and MEF_fortilin+/+. MEF_{fortilin−/−} has not been successfully isolated. The fortilin expression level in MEF_{fortilin+/−} was approx. half that in MEF_fortilin+/+, as determined by Western blot analysis (Figure 5A). Upon thapsigargin challenge, MEF_{fortilin+/−} exhibited a significantly higher r-[Ca²⁺]_i than did MEF_fortilin+/+ (Figures 5B and 5C). These results suggest that the presence of fortilin interfered with the thapsigargin-induced elevation of [Ca²⁺]_i in MEF. In our previous work [7], we demonstrated that fortilin is present in both the cytosol and the nucleus. In addition, thapsigargin has been shown to increase [Ca²⁺]_i by releasing Ca²⁺ that is stored in the ER into the cytosol [36,42,43]. Taken together, our present results suggest that fortilin scavenges Ca²⁺ released into the cytosol from the ER and blunts the elevation of [Ca²⁺]_i in vivo.

Figure 5. Fortilin blocks thapsigargin-induced elevation of free intracellular Ca²⁺ ([Ca²⁺]_i) in MEF cells.

(A) MEF_{fortilin+/−} cells expressed a significantly smaller amount of intracellular fortilin than MEF_fortilin+/+ as measured by Western blot analysis. (B) r-[Ca²⁺]_i concentration of MEF_{fortilin+/−} or MEF_fortilin+/+, upon thapsigargin challenge (400 nM). MEF_{fortilin+/−} showed more exaggerated elevation of [Ca²⁺]_i to thapsigargin stimulation than did MEF_fortilin+/+. See the Materials and methods section for experimental details. (C) Free intracellular Ca²⁺ index, the area under the curve for MEF_fortilin+/+ cells was significantly smaller than that for MEF_{fortilin+/−} cells (134.09±6.4 arbitrary units and 172.1±3.7 abitrary units for MEF_fortilin+/+ and MEF_{fortilin+/−} cells respectively). ****P<0.001.

Fortilin blocks Ca²⁺-dependent apoptosis in U2OS cells

To evaluate the biological consequence of fortilin-blunted [Ca²⁺]_i elevation in thapsigargin-challenged cells, we treated U2OS cells with either siRNA_fortilin or siRNA_luciferase, stimulated them with thapsigargin in the presence or absence of the potent cell-membrane-permeant intracellular Ca²⁺ chelator BAPTA-AM, and then evaluated these cells for cell morphology (Figure 6A), cell death rate (Figure 6B) and for caspase 3 activation (Figure 6C). First, fortilin protected cells against thapsigargin-induced cell death because we observed far more rounded dead cells in siRNA_fortilin-treated cells that those treated by siRNA_luciferase (Figure 6A; top panels) and because the EthD-1 index, representing the late phase of apoptosis or necrosis, was significantly higher in siRNA_fortilin-treated cells than in siRNA_luciferase-treated cells upon thapsigargin challenge (Figure 6B, column 1 compared with column 3; 5.10±0.94% compared with 1.65±1.28%; **P<0.01). Secondly, most of the cell death caused by the lack of fortilin was preventable by BAPTA-AM and therefore Ca²⁺-dependent in nature. This was because the frequency of rounded dead cells in siRNA_fortilin-treated cells were reduced by BAPTA-AM almost to that in siRNA_luciferase-treated cells (Figure 6A, lower left-hand and upper right-hand panels respectively) and because most of cell death preventable by fortilin (indicated with #, Figure 6B) was preventable by BAPTA-AM and thus Ca²⁺-dependent (Figure 6B). We then measured the caspase 3 activities of these cells. As shown in Figure 6(C), siRNA_fortilin-treated cells, but not siRNA_luciferase-treated cells, exhibited significant caspase 3 activation upon thapsigargin stimulation (1219±359 compared with 28±213 relative units/μg of protein; **P<0.01). The absence of significant thapsigargin-induced caspase 3 activity in siRNA_luciferase-treated cells (Figure 6C, siRNA_luciferase) suggests that most of cell death of siRNA_luciferase-treated cells in Figure 6(B) (column 3) represented necrosis (cell death lacking caspase 3 activation). The results in turn suggest that most of BAPTA-AM-preventable cell death in siRNA_fortilin-treated cells is apoptosis. Taken together, the results presented in Figure 6 suggest that fortilin prevents the thapsigargin-induced elevation of [Ca²⁺]_i from inducing apoptosis in U2OS cells.

Fortilin blocks Ca²⁺-dependent apoptosis in MEF cells

To determine whether the inhibition by fortilin of Ca²⁺-dependent apoptosis seen in U2OS also occurs in primary culture cells, we compared the Ca²⁺-dependent apoptosis rates in MEF_{fortilin+/−} and MEF_fortilin+/+. MEF cells were stimulated with thapsigargin in the presence or absence of BAPTA-AM and evaluated for cell morphology (Figure 7A), cell death rate (Figure 7B) and caspase 3 activation (Figure 7C). Fortilin again protected cells against thapsigargin-induced cell death because we observed far more rounded dead cells in MEF_{fortilin+/−} cells than in MEF_fortilin+/+ cells (Figure 7A, upper left-hand and upper right-hand panels respectively) and because the EthD-1 index was significantly higher in MEF_{fortilin+/−} cells than in MEF_fortilin+/+ cells upon thapsigargin challenge (Figure 7B, column 1 compared with column 2; 15.28±2.0% compared with 3.02±1.75%; ***P<0.005). Secondly, most of the cell death caused by the lack of fortilin was again preventable by BAPTA-AM and therefore Ca²⁺-dependent in nature. This was because the frequency of rounded dead MEF_{fortilin+/−} cells were reduced by BAPTA-AM almost to that of rounded dead cells in MEF_fortilin+/+ (Figure 7A, lower left-hand and upper right-hand panels respectively) and because most of cell death preventable by fortilin (indicated by #, Figure 7B) was preventable by BAPTA-AM and thus Ca²⁺-dependent (Figure 7B). We then measured the caspase 3 activities of these cells. As shown in Figure 7(C), MEF_{fortilin+/−} cells, but not MEF_fortilin+/+ cells, exhibited significant caspase 3 activation upon thapsigargin stimulation (Figure 7C; 1110±15.3 compared with 26±22.5 relative units/μg of protein; *P<0.05). The absence of significant thapsigargin-induced caspase 3 activity in MEF_fortilin+/+ cells (Figure 7C, fortilin^+/+) suggests that most of the cell death of MEF_fortilin+/+ cells in Figure 7(B) (column 3) represented necrosis. The results also suggest that most of BAPTA-AM-preventable cell death in MEF_{fortilin+/−} cells is apoptosis. Taken together, these results suggest that fortilin robustly protects MEFs against thapsigargin-induced apoptosis. Consistent with the data presented in Figure 6, the data presented in Figure 7 suggest that fortilin blocks the apoptosis resulting from thapsigargin-induced elevation of [Ca²⁺]_i.

Glu⁵⁸ and Glu⁶⁰ of fortilin are critical for binding to Ca²⁺

To determine what region(s) of fortilin are critical for Ca²⁺ binding, we synthesized 12 overlapping fortilin peptides (Figure 8A) and subjected them to equilibrium dialysis. As shown in Figure 8(B), CBB alone in the tube had the exact same radioactivity as CBB in the beaker (Buffer and thick line; Figure 8B). CBBs containing calmodulin, fortilin–His₆, and GST–fortilin were all significantly more radioactive than the CBB control (Figure 8B), consistent with the data presented in Figure 1(A). In this system, only three polypeptides, namely fortilin(43–62), fortilin(57–76) and fortilin(127–146), exhibited significantly more radioactivity than did the CBB control, suggesting that the binding of fortilin to Ca²⁺ occurred through amino acids within these three regions. There are eleven charged amino acids within these regions. We systematically mutated these charged amino acids to generate several mutants of full-length fortilin: fortilinΔ1 (D44A/D45A), fortilinΔ2 (E58A/E60A), fortilinΔ3 (E55A/E58A/E60A/E63A), fortilinΔ4 (D71A), fortilinΔ5 (E138A) and fortilinΔ6 (D143A) (Figure 9A). These six mutant proteins of fortilin were expressed as GST-fusion proteins, purified and characterized (Figure 9B) and subjected to the same equilibrium dialysis. In this experiment, the radioactivities of CBB-only and GST-only solutions were identical with the radioactivity of the CBB in the beaker (Buffer, GST-only; Figure 9C). As expected, calmodulin and GST–fortilin exhibited significantly more radioactivity than did CBB alone and GST-only (Figure 9C). In this system, the solutions containing GST–fortilinΔ5 and Δ6 were significantly more radioactive than the controls (P<0.005 and P<0.05 respectively), suggesting that neither Glu¹³⁸ nor Asp¹⁴³ played a critical role in the Ca²⁺ binding of fortilin (Figure 9C). Strikingly, however, the mutations introduced to Glu⁵⁸ and Glu⁶⁰ abolished almost completely the Ca²⁺-binding capability of fortilin (GST–fortilinΔ2, Figure 9C). The finding that Glu⁵⁸ and Glu⁶⁰, but not Glu¹³⁸ or Asp¹⁴³, were critical to the Ca²⁺ binding of fortilin is entirely consistent with our finding in the Ca²⁺-overlay assay that the 70 amino acid residues at the N-terminus of fortilin were critical for Ca²⁺ binding (Figure 1B). Based on these observations, we chose fortilinΔ2 to test the biological consequence of the inability of fortilin to bind Ca²⁺ in vivo. Intriguingly, the crystal structure of Schizosaccharomyces pombe fortilin (MMDB: 16785; PDB: 1H6Q) [44] shows that Glu⁵⁸ and Glu⁶⁰ were both localized in the loose loop of the molecule that protrudes from the surface of the more tightly packed portion of the molecule (Figure 9D).

Figure 9. FortilinΔ2(E58A/E60A) fails to bind Ca²⁺ in equilibrium dialysis assays.

(A and B) Generation and characterization of GST–fortilin mutants. Fortilin mutants containing point mutations within full-length fortilin, namely fortilinΔ1 (D44A/D45A), fortilinΔ2 (E58A/E60A), fortilinΔ3 (E55A/E58A/E60A/E63A), fortilinΔ4 (D70A), fortilinΔ5 (E138A) and fortilinΔ6 (D143A) (A), were generated, expressed with a GST-fusion tag, and characterized by SDS/PAGE and Coomassie Blue staining (B). The mutant in the yellow row (fortilinΔ2) exhibited significant Ca²⁺-binding activity in equilibrium dialysis assays. (C) Equilibrium dialysis of GST–fortilin containing mutations showed the lack of significant binding by fortilinΔ2. *P<0.05, ⁺P=0.073, ***P<0.005. (D) Structure of S. pombe fortilin (PDB: 1H6Q) showing the localization of Glu⁵⁸ and Glu⁶⁰. C, C-terminus; N, N-terminus.

FortilinΔ2(E58A/E60A), a fortilin point mutant lacking Ca²⁺-binding activity, is more susceptible to thapsigargin-induced cell death than its wild-type counterpart

To establish that the protection by fortilin of thapsigargin-challenged cells against cell death is mediated via the binding of fortilin to Ca²⁺ and sequestration of Ca²⁺ from Ca²⁺-dependent downstream apoptosis pathways, we established U2OS cell lines stably expressing wild-type fortilin (U2OS_{fortilin-wild}) and fortilinΔ2 (U2OS_fortilinΔ2) as well as U2OS cells harbouring empty plasmids (U2OS_empty). Western blot analysis showed that exogenous fortilin and fortilinΔ2 were robustly expressed in U2OS_{fortilin-wild} and U2OS_fortilinΔ2 respectively (columns 2 and 3, Figure 10A). We then subjected these cell lines to thapsigargin and determined the EthD-1 indices. As expected, the EthD-1 index was significantly lower in U2OS_{fortilin-wild} than it was in U2OS_empty cells (U2OS_{fortilin-wild} compared with U2OS_empty; 2.31±0.57% compared with 4.58±0.39% respectively; *P<0.05) (Fortilin compared with Empty, Figure 10B). Strikingly, the EthD-1 index of U2OS_fortilinΔ2 cells was no different than that of U2OS_empty cells (U2OS_fortilinΔ2 compared with U2OS_empty; 3.96±1.11 compared with 4.58±0.39% respectively; not significant) (FortilinΔ2 compared with Empty, Figure 10B). These results suggest that fortilin, but not fortilinΔ2, was capable of protecting cells against thapsigargin-induced cell death and that the ability of fortilin to bind Ca²⁺ was required for fortilin to protect cells against thapsigargin-induced cell death. Since most of the thapsigargin-induced cell death was Ca²⁺-dependent (Figures 6B and 7B) and represented apoptosis (Figures 6C and 7C), it is suggested that the ability of fortilin to bind Ca²⁺ is required for fortilin to protect cells against Ca²⁺-dependent apoptosis. These results further suggest that fortilin is a cytosolic Ca²⁺ scavenger and that anti-apoptotic activity of fortilin is at least partly due to its ability to sequester ER-derived Ca²⁺ from downstream Ca²⁺-dependent apoptotic pathways.

Figure 10. FortilinΔ2 (E58A/E60A), a non-Ca²⁺-binding mutant of fortilin, fails to prevent thapsigargin-induced cell death.

(A) Western blot analysis of U2OS cells. Actin, loading control; Native fortilin, fortilin natively expressed in U2OS cells, Fortilin/fortilinΔ2, over-expressed fortilin or fortilinΔ2(E58A/E60A) detected by an anti-fortilin antibody. (B) EthD-1 indices of U2OS cells stably harbouring empty plasmid (Empty), stably expressing wild-type fortilin (Fortilin) and stably expressing fortilinΔ2(E58A/E60A)(FortillinΔ2). *P<0.05 in comparison with Empty; NS, no statistically significant difference in comparison with Empty. The overexpression of fortilin, but not fortilinΔ2, a non-Ca²⁺-binding mutant of fortilin, protects cells against thapsigargin-induced cell death.

DISCUSSION

For over a decade now, fortilin has been known to bind Ca²⁺ [8–12], but its biological significance has remained purely speculative. Since fortilin is an anti-apoptotic protein and Ca²⁺ plays a critical role in apoptosis, we hypothesized that fortilin binds and scavenges Ca²⁺, thus preventing the ion from activating downstream apoptotic execution pathways. In the present study, we set out to test the hypothesis in a series of experiments. There are four key findings in the present study that are novel. First, we have employed four different assays, namely Ca²⁺-overlay (Figure 1), flow dialysis (Figure 2), CD spectropolarimetry (Figure 3) and equilibrium dialysis (Figure 9C) assays to unequivocally demonstrate the presence of the binding of fortilin to Ca²⁺. This was significant because previous studies relied entirely on Ca²⁺-overlay assays to show the binding of fortilin to Ca²⁺ [8–12]. Importantly, the binding of fortilin to Ca²⁺ was associated with the change in secondary structure of fortilin as detected by CD spectropolarimetry (Figure 3). Secondly, we have been able to show, using both U2OS cells and MEFs and a standard Fura 2 Ca²⁺-detection system, that the lack of fortilin led to an exaggerated elevation of free intracellular calcium levels ([Ca²⁺]_i) upon thapsigargin stimulation (Figures 4 and 5). This is significant because it has not been reported previously that fortilin can block the elevation of [Ca²⁺]_i in vivo. Thirdly, we have demonstrated that fortilin protects cells against thapsigargin-induced Ca²⁺-dependent apoptosis (Figures 6 and 7) and that the binding of fortilin to Ca²⁺ is required for fortilin's protection against such apoptosis (Figure 10), both of which have not been reported in the literature. Finally, we have identified amino acids of fortilin critical for its binding to Ca²⁺, namely Glu⁵⁸ and Glu⁶⁰ and generated a double-point mutant of fortilin [fortilin(E58A/E60A) or fortilinΔ2] lacking Ca²⁺-binding (Figure 9). Strikingly, wild-type fortilin, but not fortilin(E58A/E60A) (fortilinΔ2), protected cells against thapsigargin-induced cell death; fortilin is required to bind Ca²⁺ in order to block Ca²⁺-dependent apoptosis. Based on these findings, we conclude the hypothesis above to be correct.

Others besides ourselves have also tried to identify the Ca²⁺-binding site of fortilin. Using fortilin–His₆ and fortilin fragments in a ⁴⁵Ca²⁺-overlay assay, Kim et al. [10] found that wild-type fortilin and fortilin(1–112), but not fortilin(1–80) or fortilin(1–52), bound ⁴⁵Ca²⁺. They therefore concluded that the Ca²⁺-binding site of fortilin resides in amino acid residues 81–112. However, our experiments with GST–fortilin and its mutants suggest that the Ca²⁺-binding site of fortilin resides in the polypeptide consisting of amino acids 1–70 (Figures 1B and 1C). In addition, our equilibrium dialysis assays showed no significant binding between Ca²⁺ and any of several fragments including fortilin(71–90), fortilin(85–104), fortilin(99–118) or fortilin(113–132) peptide. Furthermore, it did show very significant Ca²⁺ binding to fortilin(43–62) and fortilin(57–76) polypeptides (Figure 8). Consistently, there was no binding between Ca²⁺ and fortilinΔ2, a double-point mutant in which glutamate residues at amino acid positions 58 and 60 were mutated to alanine residues (Figure 9C). The apparent discrepancy between our findings and those of Kim et al. [10] may possibly be due to their use of fortilin(1–80), which carries a His₆-tag at its C-terminus and may not have folded properly. It is also possible that Glu⁵⁸ and Glu⁶⁰ consist in one of multiple Ca²⁺-binding pockets within fortilin. The presence of at least two Ca²⁺-binding pockets within fortilin is supported by our flow dialysis data presented in Figure 2. The co-crystallization data of fortilin and Ca²⁺ would allow us to determine the relative spatial contributions of these amino acid residues to Ca²⁺ and help to resolve this apparent discrepancy.

In the present study, flow dialysis assays provided a new insight into the binding of fortilin to Ca²⁺ where fortilin exhibited a biphasic Ca²⁺ binding (Figure 2), suggesting the presence of more than one, most probably two, high-affinity Ca²⁺-binding sites, and the possible presence of multiple lower-affinity Ca²⁺-binding sites. The high-affinity sites were almost saturated at approx. 8–18 μM Ca²⁺, a finding consistent with CD spectropolarimetry data (Figure 3). Ca²⁺ binding to the lower-affinity sites seemed to disturb the measurement of the high-affinity sites and the Ca²⁺-binding data to the latter were scattered. Although the high-affinity binding sites are likely to be responsible for the conformational change of fortilin observed in CD spectra (Figure 3), the exact nature of the Ca²⁺ binding to the low-affinity sites remains unclear. It is tempting to speculate that fortilin binds Ca²⁺ in a step-wise fashion in vivo: as the intracellular Ca²⁺ concentration ([Ca²⁺]_i) starts to rise in response to apoptotic stimuli, fortilin binds and scavenges Ca²⁺, using its higher-affinity Ca²⁺-binding sites, to maintain [Ca²⁺]_i at a baseline level. The ligation of higher-affinity Ca²⁺-binding sites may result in the conformational changes, as detectable by CD spectropolarimetry (Figure 3), to make lower-affinity Ca²⁺-binding sites available for further Ca²⁺ binding. Nevertheless, the exact biological role of lower-affinity Ca²⁺-binding sites calls for further biochemical studies.

Several groups of investigators have also investigated the link between the thapsigargin-induced increase in [Ca²⁺]_i and apoptosis. Thastrup et al. [45] showed that thapsigargin induces rapid and dose-dependent release of stored Ca²⁺ from ER-rich microsomes, resulting in a robust increase in [Ca²⁺]_i, and that thapsigargin does so by inducing the acute and highly specific arrest of SERCA in the ER and consequent rapid leakage of Ca²⁺ into the cytosol [45]. Srivastava et al. [46] demonstrated that [Ca²⁺]_i elevation is the primary mechanism that thapsigargin uses to induce apoptosis by showing that the ability of thapsigargin to induce apoptosis was effectively blocked by buffering [Ca²⁺]_i with BAPTA-AM [46]. They also observed that [Ca²⁺]_i elevation was associated not only with the release of cytochrome c from mitochondria but also with caspase 3 activation [46]. In the present study, we consistently observed that BAPTA-AM drastically reduced the rate of cell death in both fortilin-deficient cells (U2OS cells treated with siRNA_fortilin and MEF_{fortilin+/−}) and control cells, suggesting the majority of thapsigargin-induced cell death is Ca²⁺-dependent. Importantly, thapsigarin-induced cell death in siRNA_luciferase-treated cells and MEF_fortilin+/+ (column 3 of Figures 6 and 7) was not associated with caspase activation (Figures 6C and 7C), suggesting that EthD-1 positivity in these cells represented necrosis, rather than apoptosis. In that light, most of the cell death blocked by the presence of fortilin was both Ca²⁺-dependent and apoptotic in nature. Strikingly, fortilin(E58A/E60A) (fortilinΔ2), a double-point mutant of fortilin lacking Ca²⁺-binding activity, was not capable of protecting cells against thapsigargin-induced cell death (Figure 10). It follows that the binding of fortilin to Ca²⁺ is required if fortilin is to function as an anti-apoptotic agent in thapsigargin-induced, Ca²⁺-dependent apoptosis. In addition, the difference seen in the indices of EthD-1 for control samples among experiments (Figures 6 and 10) most probably represents the fact that U2OS cells used in Figure 6 were treated with siRNA_luciferase, whereas U2OS_empty cells used in Figure 10 had been stably transfected with pcDNA4-empty plasmid vector. It is thus safe to conclude that the primary mechanism by which fortilin blocks thapsigargin-induced Ca²⁺-dependent apoptosis is through its binding to Ca²⁺. In future investigations of possible roles of fortilin in Ca²⁺-independent apoptosis, a double-point mutant fortilin(E58A/E60A) will prove to be a highly effective reagent.