Characterization of fortilin, a novel antiapoptotic protein.

Apoptosis is meticulously controlled in living organisms. Its dysregulation has been shown to play a key role in a number of human diseases, including neoplastic, cardiovascular, and degenerative disorders. Bcl-2 family member proteins and inhibitors of apoptosis proteins are two major negative regulators of apoptosis. We report here the characterization of novel antiapoptotic protein, fortilin, which we identified through yeast two-hybrid library screening. Sequence analysis of fortilin revealed it to be a 172-amino acid polypeptide highly conserved from mammals to plants. Fortilin is structurally unrelated to either Bcl-2 family member proteins or inhibitors of apoptosis proteins. Northern blot analysis showed the fortilin message to be ubiquitous in normal tissue but especially abundant in the liver, kidney, and small intestine. Western blot analysis using anti-fortilin antibody showed more extensive expression in cancerous cell lines (H1299, MCF-7, and A549) than in cell lines derived from normal tissue (HEK293). Immunocytochemistry using HeLa cells transiently expressing FLAG-tagged fortilin and immunohistochemistry using human breast ductal carcinoma tissue and anti-fortilin antibody both showed that fortilin is predominantly localized in the nucleus. Functionally, the transient overexpression of fortilin in HeLa cells prevented them, in a dose-dependent fashion, from undergoing etoposide-induced apoptosis. Consistently, U2OS cells stably expressing fortilin protected the cells from cell death induced by etoposide over various concentrations and durations of exposure. In addition, fortilin overexpression inhibited caspase-3-like activity as assessed by the cleavage of fluorogenic substrate benzyloxycarbonyl-DEVD-7-amido-4-(trifluoromethyl)coumarin. Furthermore, the antisense depletion of fortilin from breast cancer cell line MCF-7 was associated with massive cell death. These data suggest that fortilin represents a novel antiapoptotic protein involved in cell survival and apoptosis regulation.

Apoptosis represents a highly efficient and extremely sophisticated system for removing cells from the surrounding microenvironment (1)(2)(3)(4)(5)(6)(7)(8). As deadly as it may be, apoptosis is essential for the elimination of aberrant cells and the survival of the living organism as a whole. It is not surprising, therefore, that apoptosis is meticulously controlled. Among major regulatory cellular mechanisms described for other systems, ever increasing numbers of them are also being shown to be involved in the regulation of apoptosis. These include (a) proteolytic processing and activation by co-factors (3,9,10); (b) covalent modification such as phosphorylation (11,12), S-nitrosylation (13), thioester bond formation (14), and ubiquitination (15,16); and (c) compartmentalization (10,17,18) of pro-and antiapoptotic molecules.
In addition, there exist molecules specifically designed to inhibit the various stages of apoptosis. Bcl-2 family member proteins and inhibitor of apoptosis proteins (IAPs) 1 represent two major negative regulators of apoptosis. First, Bcl-2 inhibits the release of cytochrome c from mitochondria in the presence of various proapoptotic stimuli (19). Also, it has been shown that Bcl-xL binds apoptotic protease activating factor-1, thereby interfering with its recruitment of caspase-9 (20). In addition, Bcl-2 heterodimerizes with Bax, a proapoptotic protein, thereby preventing Bax from undergoing the conformational changes required for the formation of cytotoxic pores in the mitochondrial membrane (21). Furthermore, Bcl-2 can associate with Raf-1, a protein kinase that phosphorylates and inactivates the proapoptotic protein BAD (22). The inhibitor of apoptosis proteins (IAPs), originally identified as a family of proteins in baculovirus that inhibited the apoptotic response of insect cells to viral infection (23), represent a group of antiapoptotic proteins structurally distinct from Bcl-2 family proteins, consisting of two N-terminal repeats (baculovirus IAP repeats) and a C-terminal RING finger domain. Several human IAPs have been described, including survivin (24), X-linked inhibitor of apoptosis protein (25), neuronal apoptosis inhibitor protein (26), and others. Antiapoptotic function of IAPs appears to be mediated through direct binding and inhibition of effector caspases (caspases 3, 7, and 9) (25,(27)(28)(29)(30).
In order to explore the apoptosis regulatory mechanism related to Bcl-2 family member proteins, we searched for proteins that interacted with myeloid cell leukemia protein-1 (MCL1), a Bcl-2 homologue, using the yeast two-hybrid system. The amino acid sequence of one positive clone that specifically interacted with MCL1 was identical to that of human translationally controlled tumor protein (31)(32)(33), the function of which was unknown. Analysis of the promoter region of this gene has shown the presence of the binding sites for multiple transcription factors, implying that this gene, like other genes, is under transcriptional control (34). We then discovered that this mol-* This study was supported in part by National Institutes of Health Grants HL04015 and HL68024 and the MacDonald General Research Fund at St. Luke's Episcopal Hospital (Houston, TX) (to K. F.). 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  ecule is ubiquitously expressed in normal tissue. In addition, we found that it prevented cells from undergoing apoptosis. On the basis of these observations, it is now designated fortilin (from fortis, meaning strong and robust in Latin). In the study reported here, we focused on characterizing the specific function of fortilin as an antiapoptotic protein. The data we present indicate that fortilin is a novel nuclear antiapoptotic protein, structurally distinct from both Bcl-2 and IAP family antiapoptotic proteins.

MATERIALS AND METHODS
Cell Lines and Culture Conditions-The ML1a cell line was maintained in RMPI1640 medium supplemented with 10% fetal calf serum (Mediatech, Herndon, VA) and antibiotics. The A543, U2OS, HeLa, 293, MCF-7, and NIH 3T3 cell lines were maintained in Dulbecco's modified Eagle's medium (Mediatech) supplemented with 10% fetal calf serum and antibiotics.
Molecular Cloning-The cDNA fragments of full-length fortilin, p21, ␤-galactosidase (LacZ), Bcl-xL, and MCL1 were obtained by standard polymerase chain reaction techniques, using appropriate primer sets, and were ligated in frame to the appropriate bacterial, yeast, and mammalian expression vectors. In all cases, the authenticity of cloned constructs was confirmed by automated dideoxynucleotide sequencing (SeqWright Co., Houston, TX).
Yeast Two-hybrid Library Screening and Sequence Analysis-Fulllength MCL1 was cloned into pAS2.1 (CLONTECH, Palo Alto, California), a vector that encodes the GAL4 DNA-binding domain, and used as bait. A human fetus liver library was screened, according to the manufacturer's instructions (CLONTECH) and as described previously (36,37). The full-length human fortilin sequence was aligned to those of other species by the Omega nucleic acid and a protein analysis program (Genetics Computer Group, Madison, WI). The hydrophilicity score was determined with the same program according to the methods of Goldman, Engelberg, and Steitz (GES) (38) and von Heijne (39).
Northern Blotting-Northern blotting was performed using multipletissue Northern membranes (OriGene, Rockville, MD) and a 32 P-labeled fortilin probe generated by the random prime method (Roche Molecular Biochemicals) (40). Blots were visualized using a Bio-Rad Quantity-One phosphorimaging system.
Generation of Anti-fortilin Antibody-Antiserum specific for fortilin was prepared in rabbits with the synthetic peptide NH 2 -CKYIKDYMK-SIKGKLEEQRPER-COOH, corresponding to amino acids 90 -111 conjugated to maleimide-activated keyhole limpet hemocyanin, and an appropriate adjuvant. Generated antiserum was purified by affinity chromatography on a peptide-Sepharose matrix and tested by enzymelinked immunosorbent assay against a control or the fortilin peptide.
Western Blotting of Fusion Proteins-Full-length fortilin cDNA was cloned in frame to the pQE-30 bacterial expression vector (Qiagen, Valencia, CA). The plasmid was co-transfected into BL21 Escherichia coli with the pREP4 plasmid (Qiagen). Polyhistidine-tagged fortilin (MRGS-His 6 -fortilin) was induced by the addition of isopropyl-␤-D-thiogalactopyranoside and purified to near homogeneity under native conditions according to the manufacturer's instructions (Qiagen). The integrity of the purified MRGS-His 6 -fortilin was confirmed by SDSpolyacrylamide gel electrophoresis (SDS-PAGE) and Coomassie Blue staining, with a single band appearing at around 30 kDa. To characterize the rabbit anti-fortilin antibody, serially diluted RGS-His 6 -fortilin (100 to 12.5 ng by weight) was subjected to SDS-PAGE and Western blot transfer. Transferred proteins were probed first with anti-RGS-His monoclonal antibody (Qiagen), after which the membrane was stripped and reprobed with rabbit anti-fortilin antibody. To evaluate the specificity of the antibody, SDS-PAGE and Western blot transfer were performed in duplicate. Before the antibodies were added, the membranes were incubated with either a control peptide consisting of amino acids 210 -230 of MCL1 (NH 2 -LETLRRVGDGVQRNHETVFQG-COOH) or the fortilin peptide (NH 2 -CKYIKDYMKSIKGKLEEQRPER-COOH) used to raise the antibody, both at the concentration of 100 ng/ml. The immunoprobing was performed as previously described (36).
Western Blotting of Total Cell Lysates-Total cell lysates were prepared as previously described (40). Briefly, cells were harvested, washed with cold phosphate-buffered saline, counted, centrifuged, and immediately frozen in liquid nitrogen. SDS loading buffer (5 mM Tris⅐HCl, pH 6.8, 100 mM dithiothreitol, 2% SDS, 0.1% bromphenol blue, and 10% glycerol) was then added to the frozen pellet (150 l/1 ϫ 10 6 cells), and the samples were incubated at 45°C for 1 h. The genomic DNA in the lysates were sheared by passing the lysate through 27-gauge needles three times. Then, 7.5-l samples, corresponding to 2.5 ϫ 10 4 cells, were loaded in each well. SDS-PAGE, Western blot transfer, and immunoprobing were performed as described previously (40). Rabbit anti-fortilin, preimmune sera, and horseradish peroxidase-conjugated goat anti-rabbit IgG (Southern Biotechnology Associates, Inc., Birmingham, AL) were used at dilutions of 1:2000, 1:2000, and 1:5000, respectively. Densitometric analysis was performed using a Bio-Rad chemiluminescence screen and Quantity One software system, according to the manufacturer's instruction. The signal intensities of fortilin bands were divided by the signal intensity of actin band from the same cell line and expressed as the relative expression index.
Immunocytochemistry-For the intracellular localization of fortilin, HeLa cells were seeded in four-well Lab-Tek chamber slides (Nalge Nunc International, Rochester, NY) and transfected with pFLAG containing fortilin or other cDNAs by using FuGENE6 (Roche Molecular Biochemicals). Twenty-four hours after transfection, the cells were fixed with 4% paraformaldehyde in phosphate-buffered saline, permeabilized at Ϫ20°C with an acetone-methanol solution (1:1, v/v), blocked with 10% normal goat serum, and probed with anti-FLAG (M2; Sigma). Bound primary antibodies were detected with goat secondary antibody conjugated to rhodamine red X (Jackson ImmunoResearch Laboratories, West Grove, PA). The nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI; Sigma). Slides were examined under a Zeiss Axioskop fluorescent microscope (Carl Zeiss Ltd., Herts, UK) equipped with a Zeiss image processing system and the appropriate filter sets.
Immunohistochemistry-Human breast tissue specimens were obtained from the Novagen Human Disease Tissue Archive (Novagen, Madison, WI). First, the sections were dewaxed, rehydrated, and subjected to antigen retrieval with Citra Plus (BioGenex, San Ramon, CA) using the microwave method according to the manufacturer's instructions. Endogenous peroxidase was then quenched, and tissue avidin and biotin were blocked. Tissues were then incubated with an affinitypurified rabbit anti-fortilin antibody (9 g/ml) for 30 min at room temperature. Bound antibodies were detected with goat anti-rabbit antibody conjugated to avidin, streptavidin-horseradish peroxidase complex (BioGenex), and diaminobenzidine. Finally, tissues were lightly counterstained with hematoxylin. A brown stain indicated positive fortilin immunoreactivity.
Assay of Etoposide-induced Cell Death-HeLa cells were seeded in four-well Lab-Tek chamber slides (Nalge Nunc International) at 1 ϫ 10 4 cells/well and transfected with pFLAG containing fortilin or other cDNAs by using FuGENE6 (Roche Molecular Biochemicals). Twentyfour hours after transfection, the cells were challenged with 5 g/ml etoposide. Cells were then stained for the FLAG epitope using anti-FLAG antibody (M2; Sigma) and anti-mouse IgG conjugated to rhodamine red X (Jackson ImmunoResearch Laboratories). The nuclei were counterstained with DAPI, and the cells were examined under a Zeiss Axioskop fluorescent microscope (Carl Zeiss Ltd.) equipped with the appropriate filter sets. Cells that emitted red fluorescence were evaluated for nuclear morphology. Condensed small nuclei or fragmented nuclei were counted as apoptotic. The apoptotic index was calculated as follows: (number of red-fluorescing cells with apoptotic nuclei)/(number of total red-fluorescing cells counted) ϫ 100. All experiments were performed in duplicate.
Generation and Characterization of U2OS Cells Stably Expressing Fortilin-U2OS cells were transfected with empty pcDNA6, a mammalian expression vector containing a blasticidine resistance gene (Invitrogen, Carlsbad, California), or with pcDNA6 vector encoding wild-type fortilin, by using FuGENE6 (Roche Molecular Biochemicals) according to the manufacturer's instructions. Transfected cells were selected for ϳ3 weeks and characterized by Western blot analysis. For the cytotoxicity assay, U2OS cells stably expressing wild-type fortilin (U2OS.F) and U2OS cells stably harboring empty pcDNA6 (U2OS.E) were seeded in a 96-well plate in quadruplicate. For the investigation of dose response, cells were challenged with various concentrations of etoposide (0 -20 g/ml) for 48 h. For the time course study, cells were challenged with 5 g/ml of etoposide and harvested after various incubation periods (0 -96 h). In both cases, the cell media were assayed for the lactate dehydrogenase (LDH) activity released from the cytosol of damaged cells, using a cytotoxicity detection kit (Roche Molecular Biochemicals) according to the manufacturer's instructions. The cytotoxicity index was calculated as follows: (LDH activity in the media Ϫ background LDH activity)/(LDH activity in the medium of cells lysed by 1% Triton X-100 Ϫ background LDH activity) ϫ 100.
Assay of Caspase-3-like Activity-Approximately 2 ϫ 10 6 U2OS stable transfectants, either expressing fortilin (U2OS.F) or harboring empty expression vector (U2OS.E), both of which have been character-ized above, were challenged either with Me 2 SO (vehicle) or with etoposide at a final concentration of 10 g/ml for 48 h. Caspase-3-like activity was then determined as previously described (13). In brief, cytosolic proteins were extracted in hypotonic cell lysis buffer (25 mM HEPES, pH 7.5, 5 mM MgCl 2 , 5 mM EDTA, 5 mM dithiothreitol, 0.05% phenylmethylsulfonyl fluoride; all from Sigma) by three cycles of freezing and thawing. The protein concentration of samples was determined by using a Bio-Rad Bradford protein assay kit (Bio-Rad). Ten micrograms of cytosolic extracts were added to caspase assay buffer (312.5 mM HEPES, pH 7.5, 31.25% sucrose, 0.3125% CHAPS) with benzyloxycarbonyl-DEVD-7-amido-4-(trifluoromethyl)coumarin as substrates (Calbiochem). Release of 7-amido-4-(trifluoromethyl)coumarin (AFC) was quantified, after 2 h of incubation at 37°C (or, in the case of time course experiment, every 10 min), using a Fluoroskan system (Thermo-Labsystems, Helsinki, Finland) set to an excitation value of 355 nm and emission value of 525 nm. The results were expressed as relative fluorescence units/g of protein.
Assay of Antisense-treated MCF-7 Cell Survival-For the Western blot analysis to evaluate intracellular fortilin concentration with antisense treatment, 1 ϫ 10 6 MCF-7 cells, a malignant breast ductal carcinoma cell line, were seeded on a six-well plate. The next day, benzyloxycarbonyl-DEVD-fluoromethylketone (Kamiya Biomedical Company, Seattle, WA), a caspase-3 inhibitor, was added to the medium at the concentration of 100 M. Twenty-four hours after the addition of the caspase inhibitor, cells were transfected with pFLAG-antisense fortilin by FuGENE6 (Roche Molecular Biochemicals). Cells were harvested immediately after the transfection and 8 h after the transfection by the direct addition of radioimmune precipitation buffer (50 mM Tris-HCl, pH 7.2, 150 mM NaCl, 1% Nonidet P-40, 1% sodium deoxycholate, 0.1% SDS), supplemented with phenylmethylsulfonyl fluoride and aporotinin. Exactly 10 g of protein extracts were then resolved by SDS-PAGE and subjected to Western blot analysis using anti-actin (Chemicon, Temecula, CA) and anti-fortilin antibodies with appropriate secondary antibodies conjugated to horseradish peroxidase (Southern Biotechnology Associates, Inc.). The rest of the immunoprobing was performed as previously described (36). For the cell survival assay, MCF-7 cells were seeded onto 24-well plates in triplicate. The next day, cells were transfected by FuGENE6 (Roche Molecular Biochemicals) with either pFLAG-antisense fortilin or empty pFLAG FIG. 1. Sequence and structure of fortilin. A, sequence alignment of fortilin. The amino acid sequence of human fortilin was aligned to the sequences of rabbit, mouse, chicken, Drosophila melanogaster, Caenorhabditis elegans, Saccharomyces cerevisiae, and rice, using a sequence analysis program. Black and gray boxes represent amino acids that are identical and homologous, respectively, to those of human fortilin. The numbers in parentheses indicate GenBank TM accession numbers. As shown here, fortilin is highly conserved among species. B, the domain structure of fortilin. The amino acid sequence of fortilin was evaluated for its hydrophilicity using the GES (black line) and von Heijne (gray line) methods. The middle one-third of fortilin (domain 2; amino acids 71-120; isoelectroic point (pI) ϭ 9.01) is highly hydrophilic and sandwiched by domain 1 (amino acids 1-70; pI ϭ 4.13) and domain 3 (amino acids 121-172; pI ϭ 4.75).
vector, with pFLAG-LacZ used as a transfection survival marker. Cells were harvested 12, 24, 36, 48, or 72 h after the transfection and assayed for ␤-galactosidase activity with the Galacto-Light Plus Assay Kit (Tropix, Bedford, MA). The loss of exogenous ␤-galactosidase activity in the assay reflected the death and loss of transfected cells, while the remaining exogenous ␤-galactosidase activity represented the survival and retention of transfected cells. The survival index of the antisense-treated cells was calculated as follows: (␤-galactosidase activity of antisense-treated cells at a given time point)/(␤-galactosidase activity of control cells at the same time point) ϫ 100.
Statistical Analysis-Dunnett's or two-sample t tests were used to evaluate the difference in apoptotic and survival indices among cells transfected with different plasmids. To evaluate linear trends between the amount of plasmids used for the transfection and apoptotic indices, analysis of variance regression analysis was employed. A p value of less than 0.05 was considered to be statistically significant. For Dunnett's t test, the confidence intervals were 0.95, and the ␣ value was 0.05.

RESULTS
The fortilin cDNA encodes a 172-amino acid polypeptide with no significant homology to any known proteins, in whole or in part. Fortilin is highly conserved not only in mammalian species but also in nonmammalian species (Fig. 1A). The degree of conservation is unusually high in fortilin; human and mouse Bcl-2 are 72% identical, while human and mouse fortilin are 95% identical. Fortilin does not contain typical nuclear localization signals (41), or a hydrophobic transmembrane anchor or signal sequence (Fig. 1, A and B). Overall, fortilin is a hydrophilic protein whose central portion (designated domain 2) is most hydrophilic (Fig. 1B). In addition, fortilin contains two conserved cysteine residues, one in domain 1 (28th amino acid; Fig. 1A) and the other at the very C-terminal end (172nd amino acid; Fig. 1A). It has not been studied whether these cysteine residues are engaged in a disulfide bond formation. Fortilin also contains a potential glycosylation site at Asn 51 -X 52 -Ser 53 , which is highly conserved. The glycosylation status of fortilin has not been previously investigated and awaits elucidation.
Northern blot analysis using 32 P-labeled human fortilin cDNA as probe revealed the ubiquity of fortilin in normal human tissues (Fig. 2). Fortilin signals were especially abundant in the liver, kidney, small intestine, skeletal muscle, and testis (Fig. 2). The presence of a weaker band at 1.2 kilobase pairs, just above the fortilin 1.0-kilobase pair band, may represent the message of a fortilin-like molecule that remains unidentified. The wide tissue distribution of fortilin, taken together with strikingly high sequence conservation among species, suggests that fortilin plays an essential role in basic cellular function.
Next, a polyclonal rabbit anti-fortilin antibody was raised against amino acids 99 -113 of fortilin within domain 2 (Fig.  1B) and was characterized for its ability to bind full-length fortilin (Fig. 3). The anti-fortilin antibody was capable of detecting at least 12.5 ng of His 6 -fortilin (Fig. 3A). When blocked with a peptide consisting of amino acids 99 -113 of fortilin used as antigen, the antibody no longer bound to the fortilin (Fig.  3B). In summary, the rabbit anti-fortilin antibody recognized full-length recombinant fortilin with good sensitivity (Fig. 3A) and specificity (Fig. 3B).
This antibody was then used to evaluate the expression of native fortilin in various cell lines, using SDS-PAGE under denaturing conditions. Immunoblotting with rabbit preimmune serum produced no signal in any cell lines, but the use of the anti-fortilin serum yielded one discrete band at about 28 kDa (Fig. 3C). The calculated size of fortilin is about 19 kDa. Since the recombinant human fortilin, tagged with polyhistidine and expressed in E. coli, exhibited a band at about the same size (Fig. 3C, 6xHis-Fortilin), the discrepancy between the calculated size and apparent size on the immunoblot may have been due to the aberrant migration of fortilin on the gel, although the possibility of covalent modifications cannot be totally eliminated. Consistently, two groups of investigators who expressed fortilin in bacteria reported the size of the recombinant protein on the SDS gel to be 23-25 kDa (42,43).
The degree of fortilin expression varied significantly among the cell lines tested, being elevated in cancerous cell lines such as H1299 (non-small cell lung cancer), MCF-7 (breast cancer), HeLa (cervical cancer), and A549 (lung adenocarcinoma) and less so in soft tissue tumor cell lines such as ML1a (promyelocytic leukemia) and U2OS (osteosarcoma). Cell lines derived from normal tissues such as NIH-3T3 (mouse fibroblasts) and human embryonic kidney (HEK) 293 expressed very little fortilin (Fig. 3C). Thus, the expression of fortilin appeared to be up-regulated in tumor cell lines, especially those of epithelial origin.
Next, the intracellular localization of fortilin was determined by immunostaining HeLa cells that had been transiently transfected with a pFLAG vector containing the cDNA of p21, ␤-galactosidase (LacZ), or fortilin. When probed with anti-FLAG antibody, the FLAG-tagged p21 was detected mostly in the nucleus and LacZ in the cytosol (Fig. 4A), consistent with previous reports (44,45). In this system, the FLAG-tagged fortilin was found predominantly in the nucleus (Fig. 4A). Moreover, immunostaining of human breast cancer tissue, using the rabbit anti-fortilin antibody described above (Fig. 3,  A-C), showed fortilin to be mainly localized in the nucleus (Fig.  4B). Thus, fortilin was shown to be a predominantly nuclear protein.
Next, the specific function of fortilin was determined. Speculating that fortilin might play a role in apoptosis regulation, we tested whether fortilin prevented cell death induced by etoposide, a chemotherapeutic agent (46). For these experiments, HeLa cells were transiently transfected with a pFLAG vector containing the cDNA of fortilin, Bcl-xL, MCL1, or LacZ and were then challenged with etoposide. Cells expressing the FLAG-tagged proteins were identified by immunostaining with anti-FLAG antibody. Nuclei were stained by DAPI and examined for apoptotic nuclei. Cells were considered apoptotic when the nucleus was condensed and small or fragmented (47). Consistent with previous reports (48,49), the overexpression of Bcl-xL and MCL1 prevented HeLa cells from undergoing apo-ptosis. In this system, fortilin exhibited an antiapoptotic effect comparable with those of MCL1 and Bcl-xL (Fig. 5A). In addition, the greater amounts of the fortilin exerted more prominent antiapoptotic effects in the same system (Fig. 5B). Together, these data suggested that fortilin is an antiapoptotic protein that inhibits etoposide-induced apoptosis in HeLa cells. Because fortilin has no structural homology with either Bcl-2 (50) or IAP (inhibitors of apoptosis) family proteins (51) (Fig.  1A), fortilin may represent a new class of antiapoptotic proteins.
Next, the apoptotic activity of fortilin was confirmed using a different cell line and a different cell death detection system. To this end, U2OS cells stably expressing fortilin (U2OS.F) were generated and compared with control U2OS cells stably possessing empty pcDNA6 vector (U2OS.E) for their ability to withstand etoposide-induced cytotoxicity, using a standard LDH cytotoxicity assay as described under "Materials and Methods." As shown in Fig. 6A, the stable expression of fortilin in U2OS cells was associated with significantly less cytotoxicity over the wide range of etoposide concentrations (2.5, 5, 10, and 20 g/ml; for all, p Ͻ 0.001) (Fig. 6A). Furthermore, U2OS.F exhibited significantly less cytotoxicity at various time points after the etoposide challenge (48 h, p Ͻ 0.05; 72 and 96 h, p Ͻ 0.001) (Fig. 6B). Taken together, these results suggested that fortilin is an antiapoptotic protein that prevents etoposideinduced cell death.

FIG. 4. Intracellular localization of fortilin.
A, transient expression system in HeLa cells. HeLa cells were transfected with mammalian expression vectors encoding either FLAG-epitope-tagged p21, ␤-galactosidase (LacZ), or fortilin. Cells were then fixed, permeabilized, and stained with anti-FLAG monoclonal antibody and anti-mouse IgG conjugated to rhodamine red X. The nuclei were counterstained with DAPI. As predicted, p21 and LacZ were visualized in the nucleus and cytosol, respectively. Fortilin was present predominantly in the nucleus. The magnification was ϫ 400, with the scale bar representing 25 m. B, native fortilin expression in human tissue samples. Tissue sections from a patient with ductal breast carcinoma were stained with rabbit anti-fortilin antibody for native fortilin expression. Fortilin was predominantly found in the nucleus. Purified rabbit IgG was used at the same concentration as the anti-fortilin antibody as control. The magnification was ϫ 200, with the scale bar representing 50 m.
To further explore the role of fortilin in apoptosis regulation, the status of caspase-3-like activity in fortilin-overexpressing cells upon etoposide challenge was examined. In brief, the caspase-3-like activity of the U2OS cells stably expressing fortilin (U2OS.F) as described above was compared with that of the control U2OS cells (U2OS.E) when challenged with etoposide. As shown in Fig. 7A, in the absence of etoposide challenge (Fig. 7A, Etoposide Ϫ), the caspase-3-like-activities of U2OS.E (Fig. 7A, C) and U2OS.F (Fig. 7A, F) were similarly low (158.2 Ϯ 18.0 and 283 Ϯ 22.6 for U2OS.E and U2OS.F, respectively; not statistically significant (NS)) (Fig. 7A). When these cells were challenged with etoposide (Fig. 7A, Etoposide ϩ), however, U2OS.E exhibited a 9.4-fold increase in the activity with the caspase-3-like activity index reaching 1488.1 Ϯ 58.7, while the caspase-3-like activity index of U2OS.F remained at 713.9 Ϯ 62.6 (p Ͻ 0.05) (Fig. 7A, asterisk). Consistently, the caspase-3-like activities of etoposide-challenged U2OS.E were always higher than, and diverged continuously from, those of U2OS.F during the 2-h period of the kinetics assay (Fig. 7B). Together, these data suggested that the protection by fortilin against etoposide-induced cytotoxicity, as assessed by LDH assay (Fig. 6, A and B), was at least partially due to the prevention of caspase-3 activation by fortilin, an antiapoptotic molecule.
Next, we tested the antiapoptotic role of fortilin in MCF-7 cells, a malignant human ductal carcinoma cell line. To this end, the effect of fortilin depletion by antisense fortilin treatment on the survival of MCF-7 cells was studied. First, we evaluated the intracellular concentration of fortilin in response to antisense fortilin treatment, using Western blot analysis. As is shown in Fig. 8A, the antisense fortilin treatment, using a pFLAG vector containing the antisense fortilin polynucleotide sequence (pFLAG-antisense-fortilin), significantly reduced the intracellular concentration of fortilin. We then transiently transfected MCF-7 cells with either an empty pFLAG vector or pFLAG-antisense-fortilin that we characterized above (Fig.  8A), along with pFLAG-LacZ used as a transfection survival marker. Cells were harvested at various time points and assayed for ␤-galactosidase activity. MCF-7 cells that had been transfected with the empty pFLAG vector and pFLAG-LacZ accumulated ␤-galactosidase intracellularly over time, as evidenced by the increase in ␤-galactosidase activities at 24, 36, 48, and 72 h (Fig. 8B, lower panel, closed bars). In contrast, MCF-7 cells that had been transfected with pFLAG-antisense fortilin and pFLAG-LacZ had significantly less ␤-galactosidase activity than did the control cells at these time points (Fig. 8B,  lower panel, open bars, p Ͻ 0.01, for 24, 36, 48, and 72 h). The survival of the antisense-treated cells, calculated as the ratio of ␤-galactosidase activities of the antisense-treated and control cells, dropped drastically from 100 to 29% over 72 h (Fig. 8B, upper panel, p Ͻ 0.01, for all time points in comparison with the 12-h time point). Thus, the antisense depletion of fortilin caused MCF-7 cells to undergo spontaneous and massive cell death. The presence of high intracellular levels of fortilin may allow aberrant cells to escape the normal tumor surveillance system to propagate and form tumors, perhaps through the inhibition of protective apoptosis. DISCUSSION We report here the first functional characterization of fortilin as an antiapoptotic protein. We show that fortilin is a hydrophilic nuclear protein, which is unusually well conserved across species. Fortilin exists in various normal tissues, yet its expression is much higher in cancerous cell lines than in cell lines derived from normal tissues. Strikingly, fortilin inhibited etoposide-induced apoptosis in both HeLa cells and U2OS cells. In addition, fortilin blocked the caspase-3-like activity in etoposide-challenged U2OS cells. Furthermore, antisense depletion of fortilin from MCF-7, a breast cancer cell line, caused massive cell death. Since the amino acid sequence of fortilin does not resemble that of Bcl-2 family member proteins or that of IAPs, fortilin may represent a new class of apoptosis modulator that may play a role in basic cellular function and, in case of its dysregulation, tumorigenesis.
Fortilin, also known as translationally controlled tumor protein, has been described in literature only sparsely. In fact, human fortilin has never been expressed for functional study in eucaryotic cells. Fortilin was originally cloned as a human homologue to the mouse p21 protein (32), which is abundantly expressed in a mouse cell line derived from fibroblasts (31). Using quantitative polymerase chain reaction, Sturzenbaum et al. (52) investigated fortilin expression in the earthworm Lumbricus rubellus after exposure to various heavy metals, including lead, zinc, and cadmium, and showed that fortilin message was up-regulated in the presence of heavy metal. Bhisutthibhan et al. (42) investigated Plasmodium falciparum proteins covalently modified by the antimalarial drug artemisinin and found that fortilin was one of the most heavily modified proteins. Yet, although the up-regulation of fortilin by heavy metal and the covalent modification of fortilin by artemisinin were clearly demonstrated in these studies (50,51), the function of fortilin still remained unclear. Intriguingly, our current finding that fortilin is an antiapoptotic protein is consistent with the above findings; the fortilin up-regulation in response to heavy metal may represent a built in survival mechanism of organisms such as L. rubellus against noxious environmental conditions, while the covalent modification of fortilin by artemisinin in P. falciparum may lead to functional inactivation of fortilin and consequently to the apoptotic death. Thus, in the future, artemisinin and certain heavy metals may be useful reagents for further elucidating the apoptotic function of fortilin.
Haghighat et al. (53) reported that fortilin from Tryponosoma brucei, the causative agent of African trypanosomiasis, possesses a calcium-binding property. An increase in intracellular calcium has been implicated in glucocorticoid-induced Base-line caspase-3-like activities were not different between control and fortilin-expressing cells (NS, not statistically different). Upon etoposide challenge, fortilin significantly inhibited the caspase-3like activity in U2OS cells (asterisk, p Ͻ 0.05). B, kinetics of caspase-3 like activities in control and fortilin-overexpressing cells. Approximately 2 ϫ 10 6 U2OS, either expressing fortilin (F) or harboring empty expression vectors (C), were challenged with 10 g/ml of etoposide and subjected to the assay of caspase-3-like activity as described above. The fluorescence signals were measured every 10 min for 2 h. Fortilin significantly inhibited the caspase-3-like activity in U2OS cells at all time points (p Ͻ 0.05).
FIG. 8. The anti-cell death activity of fortilin in MCF-7, a human breast ductal cardinoma cell line. A, the reduction of intracellular fortilin concentration by antisense fortilin treatment. The intracellular concentration of fortilin was evaluated in MCL-7 cells that were treated (Antisense, ϩ) and compared with that from control cells (Antisense, Ϫ), using Western blot analysis of total cell lysates by anti-actin and anti-fortilin antibodies. The fortilin signal of treated cells was significantly weaker than that of control cells. B, the induction of spontaneous MCF-7 cell death by antisense fortilin treatment. MCF-7 cells were transfected with either pFLAG-antisense fortilin or empty pFLAG, along with pFLAG-LacZ used as a transfection survival marker. Cells were harvested at the indicated times, assayed for ␤-galactosidase, and the survival index was calculated. L.U., light units. Asterisks in the survival curve indicate statistically significant (p Ͻ 0.01) reduction in survival rate relative to that at 12 h. Asterisks in the ␤-galactosidase activity indicate statistically significant (p Ͻ 0.01) differences in the activity between antisense-treated cells (open bars) and control cells (closed bars).
apoptosis (54,55). In addition, thapsigargin, a Ca 2ϩ -ATPase inhibitor that increases intracellular calcium by inhibiting the uptake of calcium into the endoplasmic reticulum, also induces apoptosis in certain cell types (56). Furthermore, the rise in intracellular calcium has been shown to be necessary for DNA degradation in anti-Fas-induced apoptosis of Jurkat cells (57). On the other hand, antiapoptotic proteins have been shown to modulate intracellular calcium levels. For example, Bcl-2 inhibits thapsigargin-induced apoptosis in WEHI7.2 cells by blocking the increase of intracellular Ca 2ϩ (35). It is possible that fortilin, by binding and scavenging Ca 2ϩ released in response to apoptotic stimuli, modulates the intracellular calcium concentration and blocks apoptosis mediated by fluctuating Ca 2ϩ concentration. The ability of fortilin to bind Ca 2ϩ and its functional significance must be tested in mammalian cells.
Further investigation of fortilin may reveal additional regulatory mechanisms of apoptosis, which is already proving itself to be ever more sophisticated and complex. The anti-fortilin antibody we have characterized here may prove to be a useful reagent for further investigating the role of fortilin in human tumorigenesis and other proliferative diseases.