Antiapoptotic protein partners fortilin and MCL1 independently protect cells from 5-fluorouracil-induced cytotoxicity.

Fortilin, a potent 172-amino acid antiapoptotic polypeptide (Li, F., Zhang, D., and Fujise, K. (2001) J. Biol. Chem. 276, 47542-47549), binds MCL1, a protein of the antiapoptotic Bcl-2 family. The fortilin-MCL1 interaction stabilizes and increases the half-life of fortilin but not necessarily of MCL1 (Zhang, D., Li, F., Weidner, D., Mnjoyan, Z. H., and Fujise, K. (2002) J. Biol. Chem. 277, 37430-37438). It is not known to what extent each protein depends on the other for its apoptotic activity. Here, we present evidence that fortilin and MCL1 are capable of functioning as antiapoptotic proteins independently of each other. Using a robust small interfering RNA (siRNA)-mediated gene silencing system developed in our laboratory, we analyzed the cytoprotective effects of fortilin and MCL1 together and apart in U2OS cell lines exposed to 5-fluorouracil (5-FU) in both monoclonal and polyclonal cell populations. When MCL1 was silenced by MCL1-targeted siRNA, fortilin was still able to protect cells from 5-FU-induced cytotoxicity in a dose-dependent manner. Conversely, when fortilin was silenced by fortilin-targeted siRNA, MCL1 was also able to protect cells from 5-FU-induced cytotoxicity in a dose-dependent manner. Together, these data clearly suggest that fortilin and MCL1 can exert their cytoprotective activities independently of each other. The silencing of fortilin and MCL1 did not qualitatively change the subcellular localization of MCL1 and fortilin, respectively. The biological significance of fortilin-MCL1 interaction may be that it increases cellular resistance to apoptosis by allowing MCL1, an independently antiapoptotic protein, to stabilize another independently antiapoptotic protein, fortilin.

Fortilin is a 172-amino acid polypeptide that was originally identified by yeast two-hybrid library screening as a molecule that specifically interacted with MCL1, a protein of the antiapoptotic Bcl-2 family (1). Fortilin is also known as translationally controlled tumor protein (2,3). Early analyses of fortilin in our laboratory revealed that its amino acid sequence is highly evolutionarily conserved; that fortilin is ubiquitous in normal tissues, especially in liver and kidney; and that it localizes in both the nucleus and cytosol. In addition, we found that its overexpression prevents HeLa and U2OS cells from undergoing etoposide-induced apoptosis and that antisense depletion of fortilin can induce MCF-7 cells to die spontaneously. Taken together, these findings have established fortilin as a unique antiapoptotic protein.
Because the amino acid sequence of fortilin does not resemble that of either Bcl-2 family proteins or IAPs (inhibitor of apoptosis proteins) and because fortilin specifically interacts with MCL1, an antiapoptotic Bcl-2 family protein, we first hypothesized that the antiapoptotic function of fortilin is mediated through MCL1. Intriguingly, we found that fortilin interacted only with MCL1, not with other Bcl-2 family proteins, suggesting that fortilin might be an MCL1-specific cofactor in the regulation of apoptosis (4). At that time, we devised our first-generation small interfering RNA (siRNA) 1 system, in which we could specifically and effectively knock down MCL1 or fortilin expression in vivo. Using this system, we unexpectedly found that MCL1 depletion by siRNA targeting MCL1 (siRNA MCL1 ) drastically reduced the intracellular concentration of fortilin, whereas the siRNA targeting fortilin (siR-NA Fortilin )-mediated depletion of fortilin did not affect the intracellular concentration of MCL1 (4). Further investigation revealed that siRNA MCL1 did not affect the amount of fortilin transcripts in the cell in a real-time quantitative reverse transcription-PCR assay and that siRNA MCL1 -induced MCL1 silencing drastically shortened the half-life of fortilin protein in a pulse-chase assay. Finally, a fortilin point mutant that failed to interact with MCL1 (fortilin R21A ) was degraded far more quickly than was wild-type fortilin in vivo. These data suggested that MCL1, through its binding to fortilin, stabilizes fortilin and that the turnover of MCL1 is not affected by fortilin (4).
We then asked whether these protein partners, fortilin and MCL1, could function as antiapoptotic proteins in the absence of each other. We first attempted to address the question by generating cell lines that could stably express fortilin and fortilin R21A . Unfortunately, it was not possible to generate cells that would stably and robustly express fortilin R21A , most likely because of the rapid degradation of fortilin R21A . We then modified the siRNA system further, which eventually allowed us to knock down MCL1 and fortilin expression for extended periods of time (up to 120 h). This second-generation siRNA system, which we used in the current work, is unique in that the MCL1-targeting siRNA represented a mixture of four different siRNAs, directed against four different regions of MCL1 mRNA. The same was true for siRNA targeted against fortilin. This system, together with cells stably overexpressing wildtype fortilin or MCL1, finally provided us with a tool to evaluate whether MCL1 and fortilin could function antiapoptotically in the absence of each other. The data presented here suggest that they can.

EXPERIMENTAL PROCEDURES
Cells, Cell Lines, and Culture Conditions-U2OS cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and antibiotics.
Small Interfering RNA System-Small interfering RNAs (5, 6) were synthesized by Dharmacon Research, Inc. (Lafayette, CO). The siRNA Fortilin actually consisted of a mixture of four siRNA duplexes targeting four different regions of fortilin mRNA, namely, AGATGTTCTCCGACATCTA, CGA-AGGTACCGAAAGCACA, GGGAGATCGCGGACGGGTT, and GGTAC-CGAAAGCACAGTAA. Similarly, the siRNA against MCL1 (siRNA MCL1 ) actually consisted of a mixture of four siRNA duplexes targeting four different regions of MCL1 mRNA, namely, AAACGGGACTGGCTAGTTA, TCA-CAGACGTTCTCGTAAG, CGAGTGATGATCCATGTTT, and GGGACTG-GCTAGTTAAACA. The siRNA against luciferase, a nonmammalian protein from Photinus pyralis (American firefly), was used as a control. All procedures were performed in an RNase-free environment as previously described in detail by us (4). Briefly, the transfection of cells with siRNA duplexes was performed using TransIT-TKO transfection kits (Mirus Corp., Madison, WI), at a final concentration of 1.1%. To minimize the cytotoxicity of the reagent itself, cells were washed once with PBS, and media were changed 6 h after transfection.
Western Blot Analysis of Cell Lysates-Cells were harvested by the direct addition of SDS gel loading buffer (1,4). Lysate samples were collected and incubated at 45°C for 1 h. The genomic DNA in the lysate was sheared by passing the lysate through 27-gauge needles five times. When appropriate, cells were harvested into radioimmune precipitation assay buffer (50 mM Tris-Cl, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.1% SDS, 0.5% sodium deoxycholate, and protease inhibitors (Complete Protease Inhibitor Mixture Tablets; Roche Applied Science)), followed by the determination of protein concentrations using BCA methods (Bio-Rad). The samples were then subjected to SDS-PAGE and Western blot analysis, using anti-fortilin, anti-MCL1 (Santa Cruz Biotechnology), and anti-actin (Roche Applied Science) antibodies, as described previously (1,7).
Trypan Blue Assay-Trypan blue assays of 5-fluorouracil (5-FU)challenged cells were performed as described previously (8,9). All experiments were performed in the absence of Zeocin (Invitrogen). In brief, cells were harvested by brief trypsinization 24 h (and 48 h in the absence of siRNA treatments) after the addition of 5-FU (Sigma). Both floating and attached cells were assayed. At least 150 cells per treatment were counted after being stained with trypan blue at a final concentration of 0.2%. Assays were performed in duplicate.
Generation of Cells Stably Overexpressing Fortilin or MCL1-To generate U2OS cells stably overexpressing either fortilin or MCL1, U2OS cells were transfected, using FuGENE6 reagent (Roche Applied Science), with pcDNA4-His-Max (pcDNA4), a mammalian expression vector containing a Zeocin selection marker (Invitrogen), containing the cDNA encoding either wild-type fortilin (pcDNA4 Fortilin-HA ) or MCL1 (pcDNA4 MCL1-HA ). For control, U2OS cells were transfected with empty pcDNA4. Both the fortilin and MCL1 cDNAs were fused with the nucleotide sequence encoding for influenza virus hemagglutinin (HA) at their 3Ј termini. For the generation of monoclonal populations of cells, transfected cells were clonally selected for at least 3 weeks by using 400 g/ml Zeocin (Invitrogen) and characterized by Western blot analyses using commercial anti-MCL1 antibodies (Santa Cruz Biotechnology) and anti-fortilin antibodies raised in our laboratory (1). The resulting lines were named U2OS empty , U2OS fortilin , and U2OS MCL1 , respectively. To generate polyclonal cell populations, cells transfected with pcDNA4 Fortilin-HA and pcDNA4 MCL1-HA were selected collectively in a large tissue culture dish for 4 weeks with 1500 and 800 g/ml Zeocin, respectively, and characterized by Western blot analyses using commercial anti-MCL1 antibodies (Santa Cruz Biotechnology) and antifortilin antibodies raised in our laboratory (1). The resulting lines were named U2OS Fortilin-poly and U2OS MCL1-poly , respectively. Ten and four clones were pooled for U2OS MCL1-poly and U2OS Fortilin-poly , respectively.
Indirect Immunofluorescence and Fluorescence Microscopy-For the intracellular localization of fortilin and MCL1 in the presence or absence of each other, U2OS cells were transiently transfected with pcDNA4 Fortilin-HA or pcDNA4 MCL1-HA and then treated with siRNA Fortilin , siRNA MCL1 , or siRNA Luciferase and finally subjected to immunocytochemical staining 48 h later, using mouse anti-HA antibody (Covance) and rabbit anti-fortilin or anti-MCL1 antibodies (Santa Cruz Biotechnology). Bound antibodies were detected by goat anti-mouse IgG antibody conjugated to Alexa Fluor 488 and goat anti-rabbit IgG antibody conjugated to Alexa Fluor 568 (Molecular Probes, Eugene, OR) and then subjected to nuclear staining by 4Ј,6-diamidino-2-phenylindole (Sigma). Stained cells were analyzed on a Zeiss Axioskop fluorescence microscope, using a ϫ40 objective and appropriate filter sets (Carl Zeiss Ltd.) as described previously (1). Images were captured using a Spot RT SE6 Slider Camera (Diagnostic Instrument, Inc., Sterling Heights, MI) and the Image Pro Plus software system (Media Cybernetics, Inc., Carlsbad, CA). Subcellular Fractionation Procedure-U2OS cells (1.5 ϫ 10 6 ) were seeded onto three 10-cm dishes. The next day, cells were transfected with siRNA Luciferase , siRNA Fortilin , or siRNA MCL1 as described above. Forty-eight hours after transfection, cells were harvested by trypsinization, and cytosolic and nuclear fractions were separated using the N-PER kit (Pierce), according to the manufacturer's instructions and as described previously (10). Protein concentrations were determined by BCA methods (Bio-Rad). Exactly 10 g of total proteins were resolved by 12% SDS-PAGE and subjected to Western blot analyses. The successful purification of nuclear and cytosolic fractions was confirmed by Western blot analyses using anti-histone H1 (clone AE-4; Santa Cruz Biotechnology) and anti-␣-tubulin (Sigma) antibodies.

A Robust siRNA System for Knockdown of MCL1 and Fortilin Has Been
Developed-To determine the feasibility of creating a microenvironment in which MCL1, fortilin, or both can be selectively depleted, we generated siRNAs against MCL1 (siRNA MCL1 ) and fortilin (siRNA Fortilin ). We then tested these siRNAs for their ability to knock down expression of the target proteins. When U2OS cells were transfected with siRNA MCL1 , MCL1 expression was knocked down to an undetectable level within 12 h (Fig. 1A, top panel). Similarly, siRNA Fortilin was able to knock down fortilin expression, although it took up to 48 h to do so (Fig. 1A, bottom panel). Higher siRNA Fortilin concentrations did not shorten the time required for complete silencing (data not shown). Treatment of cells with TransIT (Mirus Corp.) alone did not affect the intracellular levels of MCL1 and fortilin (Fig. 1B, N versus T). Surprisingly, once knocked down, MCL1 and fortilin expression remained undetectable for as long as 120 h. In addition, in dose-response experiments, we found that 25 nM siRNA MCL1 and 25 nM siRNA Fortilin were sufficient to completely knock down MCL1 and fortilin, respectively (Fig. 1B). Furthermore, we found that it was possible to silence both MCL1 and fortilin at the same time by introducing both siRNA MCL1 and siRNA Fortilin into the cells simultaneously (Fig. 1C). With 25 nM each of siRNA MCL1 and siRNA Fortilin , both MCL1 and fortilin were silenced within 24 h (Fig. 1C). These data suggested that it is possible to quickly and persistently silence MCL1 and/or fortilin expression in U2OS cells by a single transfection with siRNA MCL1 and/or siRNA Fortilin . The concentrations of siRNA needed to achieve silencing in our system were substantially lower than those described previously by us (4) and by others (6).

Depletion of Fortilin or MCL1 Increases the Susceptibility of Cells to 5-FU-induced Cell Death, Whereas Overexpression of Fortilin or MCL1 Protects Cells from 5-FU-induced Cell
Death-We also developed a system in which we could evaluate the viability of cells treated with siRNAs. First, U2OS cells were treated with 25 nM siRNA, incubated for 48 h, and then incubated with 1 mM 5-FU for another 24 h. Then, cells were subjected to trypan blue assay for the assessment of cell viability. As shown in Fig. 2A These data confirmed that fortilin and MCL1 are prosurvival molecules whose overexpression and silencing make cells more and less resistant to cytotoxic stimuli, respectively.
Depletion of Fortilin and MCL1 Does Not Change the Intracellular Localization of Each One's Protein Partner (MCL1 and Fortilin, Respectively)-Considering that fortilin and MCL1 specifically interact with each other (4), we evaluated whether the depletion of intracellular fortilin and MCL1 would change the intracellular localization of MCL1 and fortilin, respectively.
First, we transfected U2OS cells with pcDNA4 Fortilin-HA or pcDNA4 MCL1-HA and then with siRNA Fortilin , siRNA MCL1 , or siRNA Luciferase . Cells were stained with anti-HA, anti-fortilin (in cases of fortilin silencing) or anti-MCL1 (in cases of MCL1 silencing), and 4Ј,6-diamidino-2-phenylindole. The introduction of siRNA MCL1 and siRNA Fortilin into cells was associated with significant reduction of MCL1 and fortilin signals, respectively (Fig. 3A, c versus f and i versus l). In this system, the intracellular localization of fortilin and MCL1, as assessed by signals from anti-HA staining, was identical regardless of the silencing of MCL1 and fortilin, respectively (Fig. 3A, b versus e and h versus k). As shown by immunocytochemistry, fortilin localized predominantly in the nucleus and somewhat in cytosol, whereas MCL1 localized predominantly in the cytosol and somewhat in the nucleus.
Next, we fractionated the lysate from cells that had been treated with either siRNA MCL1 , siRNA Fortilin , or siRNA Luciferase into cytosolic and nuclear fractions. As shown in Fig. 3B, Western blot analyses revealed that ␣-tubulin was detectable only in the cytosolic fraction, and histone H1 was detectable only in the nuclear fraction, suggesting that the cytosolic and nuclear fractions did not cross-contaminate each other. In addition, the treatment with siRNA MCL1 depleted MCL1 in both the nucleus and cytosol (Fig. 3B, lanes 3 and 6). Similarly, treatment with siRNA Fortilin depleted fortilin in both the nucleus and cytosol (Fig. 3B, lanes 2 and 5). In this system, we asked whether the depletion of fortilin and MCL1 changed the predominant intracellular localization of MCL1 and fortilin, respectively. Before fortilin silencing, MCL1 localized predominantly in the cytosol and somewhat in the nucleus (Fig. 3B, lanes 1 and 4, MCL1), consistent with the immunostaining data reported above. Upon fortilin silencing, this pattern of cytosolic predominance persisted (Fig. 3B, lanes 3 and 6, MCL1). Before its silencing, fortilin localized somewhat more in the cytosol than in the nucleus in this system (Fig. 3B, lanes 1 and 4, Fortilin). Upon MCL1 silencing, this pattern again persisted (Fig. 3B, lanes 2  and 5, MCL1). Together, these data suggested that the depletion of MCL1 and fortilin by siRNA does not change the intracellular localization of their protein partners, namely, fortilin and MCL1, respectively.
Fortilin Prevents Cells from Undergoing Cell Death in the Absence of Its Protein Partner, MCL1-Next, we further characterized U2OS Fortilin-8 , one of the clones of U2OS cells overexpressing fortilin, using U2OS Empty-1 as control. Trypan blue assay showed that U2OS Fortilin-8 cells were significantly more resistant to 5-FU-induced cell death than were U2OS Empty-1 cells (Fig. 4A, p Ͻ 0.005 by ANOVA).
We then evaluated our first principal hypothesis, namely, that fortilin was capable of blocking cell death even in the absence of MCL1. Specifically, we transfected U2OS Empty-1 and U2OS Fortilin-8 cells with siRNA Fortilin , siRNA MCL1 , or siRNA Luciferase ; challenged the cells with 1 mM 5-FU for 24 h; and finally determined their viability using the trypan blue assay. Consistent with data reported above (Figs. 2B and 4A), fortilin-overexpressing cells (U2OS Fortilin-8 ) were more resistant to 5-FU-induced cell death than were control cells (U2OS Empty-1 ) ( 0.001 by ANOVA). In other words, in the presence of native MCL1, the higher the intracellular levels of fortilin were, the lower the susceptibility of cells to 5-FU-induced cell death was (Fig. 4B, lanes 4 -6). In still other words, fortilin had a dosedependent anticytotoxic effect on 5-FU-challenged U2OS cells.
We then asked whether the same dose dependence would be present in the essential absence of MCL1. We treated U2OS Fortilin-8 and U2OS Empty-1 cells with siRNA MCL1 to silence MCL1. In some cases, both MCL1 and fortilin were knocked down simultaneously by cotransfection of siRNA Fortilin and siRNA MCL1 as optimized above (Fig. 1C). Importantly, the total amount of siRNA introduced into these cells was kept constant by the addition of siRNA Luciferase , an irrelevant control siRNA. As shown in Fig. 4B, the introduction of siRNA MCL1 into the cells caused the intracellular level of native MCL1 to decrease to levels undetectable by Western blot analyses (Fig. 4B, Western blot, MCL1, lanes 1-3). In cases of double knockdown, Western blot analysis showed no significant signals of MCL1 or fortilin proteins (Fig. 4B, Western blot, MCL1 and Fortilin, lane 1). Strikingly, as assessed by trypan blue assay, the increasing intracellular concentration of fortilin was associated with increasing resistance to 5-FU-induced cell death even in the absence of MCL1 (Fig. 4B, no fortilin (lane 1) ANOVA). In other words, fortilin was equally capable of protecting cells from 5-FU-induced cytotoxicity in both the presence and absence of MCL1 and did so in a dose-dependent fashion (Fig. 4B, lanes 4 -6 versus lanes 1-3). Together, these data suggested that fortilin does not require the presence of MCL1 to exert its antiapoptotic activity.
MCL1 Prevents Cells from Undergoing Cell Death in the Absence of Its Protein Partner, Fortilin-Having established that fortilin does not require MCL1 in order to be antiapoptotic, we then asked whether the reverse was true. We first characterized U2OS MCL1-6 , one of the clones of U2OS cells overexpressing MCL1, using U2OS Empty-1 as a control. Trypan blue assay showed that U2OS MCL1-6 cells were significantly more resistant to 5-FU-induced cell death than were U2OS Empty-1 cells (Fig. 5A, p Ͻ 0.01 by ANOVA, when comparing U2OS Empty-1 and U2OS MCL1-6 cells).
We then evaluated our second principal hypothesis, namely, that MCL1 is capable of blocking cell death even in the absence of fortilin, using the reagents and systems we had already characterized (Figs. 1, 2, and 5A). Specifically, we transfected U2OS Empty-1 and U2OS MCL1-6 cells with siRNA Fortilin , siRNA MCL1 , or siRNA Luciferase ; challenged the cells with 1 mM 5-FU for 24 h; and then determined their viability using the trypan blue assay. In the presence of native fortilin, the higher the intracellular levels of MCL1 were, the lower the susceptibility of cells to 5-FU-induced cell death was ( We then set out to determine whether this dose dependence would hold in the absence of fortilin in the cells by treating U2OS MCL1-6 and U2OS Empty-1 cells with siRNA Fortilin to silence fortilin. In some cases, both MCL1 and fortilin were knocked down simultaneously by cotransfection of siRNA Fortilin and siRNA MCL1 as optimized above (Fig. 1C). As shown in Fig.  5B, the introduction of siRNA Fortilin into the cells caused the intracellular level of fortilin to fall to undetectable levels (Fig.  5B, Western blot, Fortilin, lanes 1-3). In cases of double knockdown, Western blot analysis showed no significant signals for either MCL1 or fortilin protein (Fig. 5B, Western blot, MCL1 and Fortilin, lane 1). Strikingly, the increasing intracellular concentration of MCL1 in these cells was associated with increasing resistance to 5-FU-induced cell death in the absence of fortilin (Fig. 4B, no MCL1 (lane 1) versus native MCL1 (lane 2) versus native plus overexpressed MCL1 (lane 3), 74.6 Ϯ 3.3% versus 60.2 Ϯ 7.4% versus 32.9 Ϯ 0.9%, p Ͻ 0.01 by ANOVA). In other words, MCL1 was equally capable of protecting cells from 5-FU-induced cytotoxicity in both the presence and absence of fortilin and did so in a dose-dependent fashion (Fig. 5B, lanes  4 -6 versus lanes 1-3). Together, these data suggested that MCL1 does not require the presence of fortilin to protect U2OS

Fortilin-MCL1 Interaction in Apoptosis cells from 5-FU-induced cell death.
In Polyclonal Cell Populations, Fortilin Prevents Cells from Undergoing Cell Death in the Absence of Its Protein Partner, MCL1, and Vice Versa-To make certain that the independent cytoprotective effects of protein partners fortilin and MCL1 represented in Figs. 4 and 5 did not originate from the selection process associated with the establishment of monoclonal cell populations, we generated polyclonal populations of U2OS cells that stably expressed fortilin (U2OS Fortilin-poly ) or MCL1 (U2OS MCL1-poly ). Western blot analyses showed that U2OS Fortilin-poly and U2OS MCL1-poly cells robustly overexpressed fortilin and MCL1, respectively (Figs. 6A and 7A).
We then transfected U2OS Fortilin-poly cells with 25 nM siRNA MCL1 and with varying amounts of siRNA MCL1 and siRNA Luciferase , which was added to keep the total amount of siRNAs constant; challenged the cells with 1 mM 5-FU for 24 h; and then determined their viability using the trypan blue assay. With 25 nM siRNA MCL1 , there was no detectable MCL1 in the cells (Fig. 6B). In this setting, increasing the amount of siRNA Fortilin caused the intracellular fortilin concentration to decrease drastically. The reduction of intracellular fortilin was associated with the increase in susceptibility of the cells to 5-FU-induced cell death (p Ͻ 0.0001 by ANOVA). In summary, fortilin protected U2OS cells from 5-FU-induced cytotoxicity in a dose-dependent fashion in the absence of MCL1. These data again suggest that fortilin does not require the presence of MCL1 to protect U2OS cells from 5-FU-induced cell death.
Finally, we transfected U2OS MCL1-poly cells with 25 nM siRNA Fortilin and with varying amounts of siRNA MCL1 and siRNA Luciferase , challenged the cells with 1 mM 5-FU for 24 h, and then determined their viability using the trypan blue assay. With 25 nM siRNA Fortilin , there was no detectable fortilin in the cells (Fig. 7B). In this setting, increasing the amount of siRNA MCL1 caused the intracellular MCL1 concentration to decrease drastically. The reduction of intracellular MCL1 was associated with the increase in susceptibility of the cells to 5-FU-induced cell death (p Ͻ 0.0001 by ANOVA). In summary, MCL1 protected U2OS cells from 5-FU-induced cytotoxicity in A, immunocytochemical analysis. U2OS cells were transfected with either pcDNA4 Fortilin-HA (aϪf) or pcDNA4 MCL1-HA (gϪl) and with either siRNA Fortilin (jϪl), siRNA MCL1 (dϪf), or siRNA Luciferase (aϪc and gϪi) and subjected to triple immunostaining using 4Ј,6-diamidino-2-phenylindole (a, d, g, and j), anti-HA antibody (Covance) (b, e, h, and k), anti-fortilin (i and l), and anti-MCL1 (Santa Cruz Biotechnology) (c and f). siRNA Fortilin and siRNA MCL1 were highly effective at silencing fortilin (i versus l) and MCL1 (c versus f) genes, respectively. The intracellular localization of HA-tagged fortilin (b versus e) and MCL1 (h versus k) was the same in the presence (f and l) and absence (c and i) of MCL1 and fortilin silencing, respectively. pcDNA4 Fortilin-HA , pcDNA4 vector containing the cDNA of HA-tagged fortilin; pcDNA4 MCL1-HA , pcDNA vector containing the cDNA of HA-tagged MCL1; DAPI, 4Ј,6-diamidino-2-phenylindole. B, subcellular fractionation and Western analyses. U2OS cells were transfected with siRNA Fortilin or siRNA MCL1 , lysed, and fractionated into subcellular (cytosolic and nuclear) fractions. After size fractionation of exactly 10 g of proteins by SDS-PAGE and transfer to nitrocellulose membrane, proteins were probed with anti-MCL1, anti-fortilin, anti-␣-tubulin (Santa Cruz Biotechnology), and anti-histone H1 (clone AE-4; Santa Cruz Biotechnology) antibodies. The siRNA Fortilin and siRNA MCL1 robustly depleted fortilin and MCL1 from both cytosolic (lanes 2 and 3) and nuclear fractions (lanes 5 and 6), respectively. The depletion of fortilin and MCL1, however, was not associated with any change in the predominant localization of these proteins. Cytosolic, cytosolic fraction; Nuclear, nuclear fraction; L, siRNA Luciferase ; F, siRNA Fortilin ; M, siRNA MCL1 ; arrowhead, nonspecific Western blot signals at the edges of membranes. a dose-dependent fashion in the absence of fortilin. These data again suggest that MCL1 does not require the presence of fortilin to protect U2OS cells from 5-FU-induced cell death Notably, the data derived from monoclonal cell populations were entirely consistent with those derived from polyclonal cell populations. Taken together, these data strongly suggest that antiapoptotic protein partners fortilin and MCL1 independently protect cells from 5-FU-induced cytotoxicity. DISCUSSION In the current study, we have shown that the protein partners fortilin and MCL1 can function as antiapoptotic proteins, even in each other's absence. In other words, fortilin and MCL1 are independently antiapoptotic. This independence has not been reported previously in the literature. Taken together with our previous observation that MCL1 stabilizes fortilin (4), our current data suggest that MCL1 can (a) function by itself as an antiapoptotic protein and (b) stabilize fortilin, which in turn can function by itself as an antiapoptotic molecule. In addition, although it is known that members of the antiapoptotic Bcl-2 family can heterodimerize with each other (11-17), there have not been any reports of such heterodimerization between a Bcl-2 family protein (in this case MCL1) and a non-Bcl-2 family antiapoptotic protein (in this case fortilin) or reports that such heterodimerization would stabilize dimerizing protein(s). Taken together with the fact that MCL1 is an inducible molecule (18,19), it is likely that fortilin-MCL1 interaction represents a unique cellular mechanism for quickly creating an antiapoptotic microenvironment protective against certain noxious extracellular conditions.
The siRNA system (5, 6) we used in the current work was a powerful tool for investigating the functional dependence between fortilin and MCL1 proteins. In the siRNA system that we have developed, fortilin and MCL1 messages are targeted by multiple kinds of siRNA duplexes. This knockdown strategy, while requiring much lower concentrations, results in much longer silencing of MCL1 and fortilin genes. Consequently, the strategy has allowed us to attain higher cell viability after transfection without compromising the silencing efficiency.
By using this strategy of siRNA-mediated gene silencing to test the independence of MCL1 and fortilin antiapoptotic func-  5 and 6). Western blot analyses showed that both U2OS Empty-1 and U2OS Fortilin-8 cells expressed no MCL1 after siRNA MCL1 treatment (lanes 1-3). Double transfection of siRNA MCL1 and siRNA Fortilin resulted in the reduction of intracellular MCL1 and fortilin to undetectable levels (lane 1). In the presence of MCL1 (lanes 4 -6), higher intracellular fortilin levels were associated with lower cytotoxicity as assessed by trypan blue assay. Strikingly, in the absence of MCL1 (lanes 1-3), higher intracellular fortilin levels were still associated with lower cytotoxicity as assessed by trypan blue assay. ‫,ءءءء‬ p Ͻ 0.001 by ANOVA.  5 and 6). Western blot analyses showed that both U2OS Empty-1 and U2OS MCL1-6 cells expressed no fortilin after siRNA Fortilin treatment (lanes 1-3). Double transfection of siRNA MCL1 and siRNA Fortilin resulted in the reduction of intracellular MCL1 and fortilin to undetectable levels (lane 1). In the presence of fortilin (lanes 4 -6), higher intracellular MCL1 levels were associated with lower cytotoxicity as assessed by trypan blue assay. Strikingly, in the absence of fortilin (lanes 1-3), higher intracellular MCL1 levels were still associated with lower cytotoxicity as assessed by trypan blue assay. ‫,ء‬ p Ͻ 0.05 by ANOVA. tion, we yielded more biologically relevant data than we would have yielded using mutants of MCL1 and fortilin. For example, we attempted to generate multiple monoclonal cell lines of U2OS FortilinR21A , U2OS cells overexpressing the fortilin point mutant R21A, in which the 21st amino acid of fortilin protein had been altered from arginine to alanine. Fortilin R21A does not interact with MCL1 (4). However, this approach proved to be inadequate because the expression levels of fortilin R21A were always low, most likely because fortilin R21A could not be stabilized by MCL1. Because most cell lines express fortilin, the low level of fortilin R21A expression would have prevented its biological activities from being revealed in U2OS cells and other cell lines.
In immunocytochemical and subcellular fractionation experiments using the siRNA system described above, we showed that the intracellular localization of fortilin and MCL1 remained the same in the absence of MCL1 and fortilin, respectively. First, regardless of the presence of fortilin, MCL1 was present to a greater extent in the cytosol and to a lesser extent in the nucleus. Second, regardless of the presence of MCL1, fortilin was shown by immunostaining to predominate in the nucleus but shown by subcellular fractionation to predominate in the cytosol. Despite this apparent discrepancy, it still holds true that the patterns of subcellular localizations of fortilin and MCL1 did not differ depending on the presence of their protein partners. Taken together, these data suggest that it is unlikely that fortilin-MCL1 interaction regulates the subcellular localization of fortilin and MCL1. The fact that fortilin and MCL1 can be located in both the nucleus and the cytosol suggests that fortilin and MCL1 are shuttle molecules, like p53 (20), mdm2 (21), ␤-catenin (22), and many components of the phosphatidylinositol 3-kinase pathway such as the insulin receptor (23), insulin receptor substrates (24), phosphatidylinositol 3-kinase (25), and protein kinase B (26), moving back and forth from the nucleus in response to changes in the cellular microenvironment.
In the previous report, we used a standard pulse-chase assay to show that MCL1 stabilizes fortilin and that the lack of MCL1 leads to the destabilization of fortilin, which is sometimes shown by Western blot analysis (4). In the present study, this destabilization of fortilin in response to the silencing of MCL1 was again seen in Fig. 2A (lanes 1 and 3), Fig. 3B (lanes 1 and   FIG. 6. Fortilin protects cells from undergoing cell death in the absence of its protein partner, MCL1, as shown in studies using polyclonal cell populations. A, characterization of a polyclonal population of U2OS cells stably expressing fortilin (U2OS Fortilin-poly ). Western blot analyses using anti-MCL1, anti-fortilin, and anti-actin antibodies showed that U2OS Fortilin-poly cells contained a significantly higher concentration of fortilin but expressed the same amount of MCL1. Fortilin-HA, fortilin encoded for by pcDNA4 Fortilin-HA . B, dose-dependent protection of 5-FU-challenged U2OS cells by fortilin in the absence of MCL1. U2OS Fortilin-poly cells were transfected with 25 nM siRNA MCL1 to create a fortilin-free environment and with various amounts of siRNA Fortilin and siRNA Luciferase to evaluate the dose-dependent nature of fortilin's cytoprotection. Western blot analyses showed that siRNA MCL1 -treated U2OS Fortilin-poly cells expressed no detectable MCL1. In the absence of MCL1, higher intracellular fortilin levels were associated with lower cytotoxicity as assessed by trypan blue assay. These data were entirely consistent with those obtained using monoclonal populations of U2OS cells (Fig. 4). Difference in cell death was significant at p Ͻ 0.0001 by ANOVA. In the absence of fortilin, higher intracellular MCL1 levels were associated with lower cytotoxicity as assessed by trypan blue assay. These data were entirely consistent with those obtained using monoclonal populations of U2OS cells (Fig. 5). Difference in cell death was significant at p Ͻ 0.0001 by ANOVA.
3), and Fig. 6B (lanes 6 and 7). In addition, 25 nM siRNA Fortilin was capable of knocking down fortilin within 24 h in the absence of MCL1 (Fig. 1C), whereas siRNA Fortilin took more than 24 h to silence fortilin in the presence of MCL1 (Fig. 1A, bottom  panel). It is not entirely clear why fortilin signals occasionally persisted upon the silencing of MCL1, as seen in the Western blot analyses (Figs. 1, A and B). It is possible that fortilin expression was up-regulated by unknown transcriptional factors, thus masking the destabilization of fortilin attributable to the lack of MCL1. The transcriptional regulation of fortilin appears to be very complex (27), and further investigation is needed to evaluate the role of various pathways in it.
The current work places fortilin in a new class of antiapoptotic molecule. Fortilin is not a cofactor of antiapoptotic MCL1, augmenting its function. However, this does not rule out the possible presence of an apoptosis executioner protein that is inhibited by fortilin or an antiapoptotic molecule other than MCL1 that is stimulated by fortilin. The amino acid sequence of fortilin does not resemble that of either the Bcl-2 family or the inhibitor of apoptosis proteins (IAPs) (1). The exact mechanism of action of fortilin as an antiapoptotic molecule is unknown. Thaw et al. (28) have uncovered a structural similarity between fortilin and Mss4 (mammalian suppressor of Sec4). Mss4 is a guanyl nucleotide exchange factor, which facilitates GDP release from and subsequent GTP binding to a subset of the Rab GTPases (29). GTP-bound Rab GTPases function as active forms and recruit effector molecules, such as coiled-coil proteins involved in membrane tethering and docking, enzymes, or cytoskeleton-associated proteins (30). Recently, Cans et al. (31) reported that fortilin interacts with translation elongation factor eEF1A and with its guanyl nucleotide exchange factor, eEF1B␤. Intriguingly, despite its homology to guanyl nucleotide exchange factor, fortilin exhibited guanine nucleotide dissociation inhibitor activity, stabilizing the GDP (inactive) form of eEF1A (31). The up-regulation of eEF1A is reported to be associated with oxidative stress-induced apoptosis (32). Although it is possible that fortilin binds to and keeps eEF1A inactive when eEF1A is up-regulated by apoptotic stimuli, further investigation is needed to define the role of fortilin-eEF1A interaction in the regulation of apoptosis.
The role of fortilin in tumorigenesis has been established. Tuynder et al. (33) performed differential gene expression analyses in which they compared aggressive cancer cell lines with nonaggressive, or reverted, cancer cell lines. Fortilin showed the most striking up-regulation (up to 124-fold on the transcriptional level) in aggressive cell lines. Fortilin overexpression also reduced the sensitivity of cancer cell lines to chemotherapeutic agents such as etoposide (1) and 5-FU (Figs. 2 and  4). On the contrary, the depletion of intracellular fortilin was associated with the spontaneous death of MCF7 cells (1), poly-(ADP-ribose) polymerase cleavage in U937 cells (33), and increased susceptibility to 5-FU in U2OS cells (Figs. 2, 4, and 5). Further dissection of the mechanism of the antiapoptotic activity of fortilin will be important if fortilin is to be developed as a novel target of cancer therapy. The siRNA systems we have described here will be a highly useful tool for such investigations.