Physical and Functional Interaction between Myeloid Cell Leukemia 1 Protein (MCL1) and Fortilin

Myeloid cell leukemia 1 protein (MCL1) is an anti-apoptotic protein that is structurally related to Bcl-2. Unlike other Bcl-2 family proteins that are constitutively expressed, MCL1 is inducibly expressed in cells that are recently exposed to growth and differentiation stimuli. Here, we report the identification of fortilin as a novel MCL1-interacting protein by screening of a yeast two-hybrid library with MCL1 as bait. Fortilin specifically interacted with MCL1 both in vitro andin vivo. The intracellular localization of fortilin was predominantly nuclear and identical to that of MCL1, as shown by immunostaining and confocal microscopy analysis. Fortilin, like MCL1, was rapidly inducible in serum-stimulated human aortic vascular smooth muscle cells. Although the depletion of intracellular fortilin by small interfering RNA (siRNA) against fortilin (siRNA-fortilin) did not affect intracellular MCL1 level, the depletion of intracellular MCL1 by siRNA-MCL1 was associated with the significant reduction of the fortilin protein level, without affecting the fortilin transcript numbers. In addition, a pulse-chase experiment showed that the depletion of MCL1 by siRNA-MCL1 was associated with the rapid degradation of fortilin protein, which was found quite stable in the presence of MCL1. Furthermore, the half-life of fortilinR21A, a point mutant of fortilin lacking the binding to MCL1, was significantly shorter than that of wild-type fortilin as shown by a pulse-chase experiment. These data suggest that MCL1, in addition to being an anti-apoptotic molecule, serves as a chaperone of fortilin, binding and stabilizing fortilin in vivo. Taken together with our previous observation that fortilin overexpression prevents cells from undergoing apoptosis (Li, F., Zhang, D., and Fujise, K. (2001) J. Biol. Chem. 276, 47542–47549), it is likely that MCL1, an anti-apoptotic protein inducible by growth and differentiation stimuli, stabilizes another anti-apoptotic protein fortilin maximizing the prosurvival environment in cells.

The regulatory role of the anti-apoptotic protein MCL1 1 in apoptosis needs clarification. MCL1 was originally identified as a protein structurally similar to Bcl-2 from the ML-1 human myeloid leukemia cell line stimulated by phorbol ester (1). Subsequent functional studies confirmed that MCL1 functions as an anti-apoptotic molecule and is capable of blocking apoptosis induced by various apoptotic stimuli, including staurosporine (2), etoposide (3), calcium ionophore A23187 (3), UV irradiation (3), and c-Myc overexpression (4). Other studies showed that antisense depletion of MCL1 in phorbol esterstimulated U937 cells can rapidly trigger apoptosis (5) and that the selective overexpression of MCL1 in hematopoietic tissues of transgenic mice improves the survival of hematopoietic cells and increased the outgrowth of myeloid cell lines (6).
However, despite its well demonstrated anti-apoptotic function, MCL1 distinguishes itself from other members of prosurvival Bcl-2 subfamily such as Bcl-xL, Bcl-2, A1, and Bcl-w (7) by the fact that its intracellular level rapidly and transiently fluctuates in response to a variety of extracellular stimuli. For example, certain cytotoxic agents (8,9) drastically reduce MCL1 levels and then induce apoptosis. On the contrary, a number of cytokines and growth factors have been shown to rapidly up-regulate MCL1 levels, including granulocyte/macrophage colony-stimulating factor (10,11), interleukin (IL)-1␤ (12), IL-3 (11,13), IL-6 (14,15), epidermal growth factor (16), erythropoietin (11), gonadotropin (17), activin A (18), and sera (19). This rapid up-regulation, as well as the down-regulation of MCL1, has been attributed to the presence of PEST sequences (1,20) and to the unusually rapid turnover of MCL1 mRNA (21). MCL1 is in fact the only Bcl-2 family protein that contains PEST sequences (1). Proliferating and differentiating cells are more susceptible to apoptotic stimuli. The fact that MCL1 is rapidly up-regulated in response to a number of growth (10,11,16,17) and differentiating factors (1,5) indicates that MCL1 may be an immediate response molecule that is designed to protect vulnerable cells, which are in the process of adjusting themselves to the changing microenvironment through proliferation and differentiation. Intriguingly, MCL1 has been shown to interact with and negatively regulate the proliferating cell nuclear antigen (PCNA), a cell cycle-regulatory protein essential for G 1 to S phase transition, and thereby retard the cell cycle progression (22). It is likely that MCL1 provides additional protection to vulnerable proliferating and differentiating cells by the slowing of cell cycle progression (23,24).
To further understand the role of MCL1 in apoptosis regulation, we used a yeast two-hybrid system with MCL1 as bait to screen a human cDNA library for proteins that would specifically interact with MCL1. The screening identified a protein of unknown function, previously known as translationally controlled tumor protein (TCTP) (25) or p21 tumor protein (26). Despite its name, TCTP had originally been cloned from the library established from normal connective tissue (26,27) and little evidence had been uncovered to support its specific regulation at the translational level. We then discovered that it inhibited etoposide-induced apoptosis (28). Thus, on the basis of its anti-apoptotic function, we have redesignated this protein fortilin (from the Latin fortis, meaning strong, robust) (28). The data presented here suggest that, in addition to acting as an anti-apoptotic protein itself, MCL1 may function as a protein partner of fortilin, binding to and stabilizing fortilin. These cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum and antibiotics. Human aortic vascular smooth muscle cells were purchased from Cascade Biologics (Portland, OR) and maintained in M231 medium according to the instructions from the manufacturer. Cells from passages 4 -7 were used in all cases.
Molecular Cloning-The cDNA fragments of full-length fortilin and its deletion mutant, MCL1, Bcl-xL, Bak, Bax, and PCNA were obtained by standard polymerase chain reaction (PCR) technique as described previously (22), using appropriate primer sets, and were ligated inframe to appropriate yeast and mammalian expression vectors. A point mutant of fortilin where the arginine at position 21 was replaced with alanine (fortilin R21A ) was generated, using a PCR-based site-directed mutagenesis protocol as described previously (22). In all cases, the authenticity of cloned constructs was confirmed by automated dideoxynucleotide sequencing (SeqWright Co., Houston, TX).
Yeast Two-hybrid Assay-S. cerevisiae SFY526 cells (CLONTECH) were cotransformed with (a) empty pAS2.1 vector or pAS2.1 vector containing a full-length MCL1 construct and (b) empty pACT2 vector or pACT2 vector containing FL1, the original clone isolated from yeast two-hybrid screening and representative of the nearly full-length fortilin molecule (amino acids 5-172). We redesignated FL1 as fortilin ⌬5-172 . Transformed cells were selected on SD plates lacking tryptophan and leucine (SD/ϪTrp/ϪLeu plates) for 7 days and subjected to a X-gal filter lift assay as described above. The blue color that developed within 8 h was considered to represent a positive interaction.
In Vitro Pull-down Assay-Radiolabeled proteins for use in an in vitro binding assay were generated by a TNT quick-coupled transcription/translation system (Promega, Madison, WI) according to the instructions from the manufacturer, using [ 35 S]methionine (Amersham Biosciences) as labeling agent. DNA templates were either circular plasmids or gel-purified PCR products containing a T7 RNA polymerase promoter. The in vitro transcribed-translated, influenza hemagglutinin-tagged fortilin (fortilin-HA) or its mutants and another in vitro transcribed-translated protein (namely MCL1, Bcl-xL, Bak, Bax, or PCNA) (as indicated in Figs. 1B and 2) were added to Buffer A (50 mM HEPES, pH 7.5, 70 mM KCl, 0.5 mM ATP, 5 mM MgSO 4 , 1 mM dithiothreitol (DTT), 0.001% Nonidet P-40, 50 M MG132, 2 g/ml bovine serum albumin (BSA), 2 g/ml aprotinin, 0.5 mM PMSF, and protease inhibitor mixture (Sigma)), and allowed to form complexes at 4°C for 90 min. Fortilin-HA or its mutants were then pulled down with rat anti-HA antibody (clone 3F10; Roche Molecular Biochemicals) and sheep anti-rat polyclonal antibody conjugated to Dynabeads TM (M480; Dynal USA, Lake Success, NY). Immune complexes were then washed five times with Buffer A and once with Buffer B (Buffer A supplemented with 0.01% Nonidet P-40). Finally, precipitated proteins were eluted into SDS gel loading buffer (50 mM Tris-Cl, pH 6.8, 100 mM DTT, 2% SDS, 0.1% bromphenol blue, and 10% glycerol), boiled for 5 min, subjected to 12% SDS-polyacrylamide gel electrophoresis (SDS-PAGE), and visualized by fluorography and imaging with a phosphorimager system (Bio-Rad).
Generation of Stable Fortilin and Mutant Fortilin Transfectants-U2OS cells were transfected with empty pcDNA6, a mammalian expression vector with blasticidine selection marker (Invitrogen), or the same pcDNA6 vector encoding either C-terminally HA-tagged wild-type fortilin, or fortilin R21A, using FuGENE6 (Roche Molecular Biochemicals). Transfected cells were selected at least for 3 weeks by 5 g/ml blasticidine, characterized by immunostaining, Western blot analysis, and in vivo immunoprecipitation assay, and named U2OS empty , U2OS wild-fortilin-HA , and U2OS fortilin-R21A-HA , respectively.
In Vivo Coimmunoprecipitation Assay-For the in vivo coimunoprecipitation assay of MCL1 and fortilin, COS-7 cells (2.5 ϫ 10 6 ) were cotransfected with pcDNA3-fortilin-HA or pcDNA3 and with FLAG-MCL1/pFLAG-CMV or pFLAG-CMV vectors using FuGENE6 (Roche Molecular Biochemicals), according to the instructions from the manufacturer. Thirty-six hours after transfection, cells were harvested by trypsinization, washed with chilled phosphate-buffered saline (PBS), suspended in Buffer A, and lysed by a nitrogen cavitation method (PARR Instrument Co., Moline, IL) (30). After centrifugation, total cell lysates were incubated with either rat monoclonal anti-HA antibody (clone 3F10) or control rat monoclonal antibody, both at a concentration of 2 g/ml. Formed complexes were precipitated by sheep anti-rat antibodies conjugated to Dynabeads TM (Dynal USA); washed with Buffers A and B; eluted into SDS gel loading buffer; and subjected to SDS-PAGE, Western blot transfer, and immunodetection with anti-HA (16B12; Covance, Richmond, CA) and anti-FLAG (M2; Sigma) antibodies. For the in vivo coimmunoprecipitation assay for the stable transfectants of fortilin and its point mutant, U2OS empty , U2OS wild-fortilin-HA , and U2OS fortilin-R21A-HA were used. Cells were harvested by scraping, washed with chilled PBS, suspended in Buffer C (Buffer A supplemented with 0.02% Nonidet P-40), and lysed by a nitrogen cavitation method (PARR Instrument Co). Cleared total cell lysates were incubated with either rat monoclonal anti-HA antibody (clone 3F10) or control rat monoclonal antibody, both at a concentration of 2 g/ml. Formed complexes were precipitated by sheep anti-rat antibodies conjugated to Dynabeads TM ; washed with Buffers C; eluted into SDS gel loading buffer; and subjected to SDS-PAGE, Western blot transfer, and immunodetection with anti-HA (16B12; Covance) and anti-MCL1 (Santa Cruz Biotechnology, Santa Cruz, CA) antibodies.
Indirect Immunofluorescence and Confocal Laser Scanning Microscopy-For the intracellular localization of fortilin and MCL1 and for the assessment of the effect of fortilin overexpression on MCL1 intracellular localization, U2OS cells, seeded in four-well Lab-Tek TM plastic chamber slides (Nalge Nunc International, Rochester, NY), were transiently transfected either with pcDNA6-fortilin-HA or pcDNA6-HA using FuGENE 6 (Roche Molecular Biochemicals). The next day, the cells were fixed with 4% paraformaldehyde in PBS, permeabilized at Ϫ20°C with an acetone-methanol solution (v/v, 1:1), blocked with 10% normal goat serum, and probed with anti-HA monoclonal antibody (16B12; Covance) and rabbit anti-MCL1 polyclonal antibody (Santa Cruz Biotechnology). Bound primary antibodies were detected with goat antimouse secondary antibody conjugated to Cy-2 and goat anti-rabbit secondary antibody conjugated to rhodamine Red X, respectively (Jackson Immunoresearch Laboratories, West Grove, PA). Cell nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI, Sigma). Cells were then examined under a Zeiss Axioskop fluorescent microscope (Carl Zeiss Ltd., Herts, United Kingdom) equipped with a Zeiss image processing system and appropriate filter sets. For the intranuclear localization of fortilin and MCL1, U2OS wild-fortilin-HA cells were seeded on glass coverslides in six-well tissue culture plates, then immunostained as described above, using anti-HA monoclonal antibody (Covance) and rabbit anti-MCL1 antibody (Santa Cruz Biotechnology), along with goat anti-mouse antibody conjugated to Cy-2 and goat antirabbit antibody conjugated to rhodamine Red X (Jackson Immunoresearch Laboratories), respectively, with DAPI nuclear counterstaining. Stained cells were analyzed on a Zeiss 210 confocal laser scanning microscope equipped with argon 488-nm and helium-neon 543-nm lasers, using a 63ϫ objective and appropriate filter sets (Carl Zeiss Ltd.). Green and red colors were assigned to the Cy-2 and rhodamine Red X signals, respectively. Image files were loaded into Adobe Photoshop (Adobe Systems, Inc., San Jose, CA), where superimposed images of Cy-2 and rhodamine Red-X staining were generated.
Western Blot Analysis of Total Cell Lysates-Total cell lysate from human aortic vascular smooth muscle cells was generated as previously described (28). In brief, ϳ1 ϫ 10 6 cells in 10-cm tissue culture dishes were made quiescent by maintaining them for 48 h in Media 231 (Cascade Biologics) containing no serum supplements. Cells were then exposed to Media 231 containing serum supplements and harvested at 0, 1, 2, 4, 8, 24, 48, and 72 h after the medium change. For the harvesting of cells, 500 l of SDS loading buffer was directly added to cell monolayers after two washings with PBS. Collected samples were incubated at 45°C for 1 h. The genomic DNA in the lysate was sheared by passing the lysate through 27-gauge needles three times. Twenty microliters of samples, corresponding to 4.0 ϫ 10 4 cells, were loaded in each lane of a 12% SDS-polyacrylamide gel. The SDS-PAGE, Western blot, and immunoprobing were performed as described previously (22,28). Anti-MCL1 (Santa Cruz Biotechnology), anti-actin (Roche Molecular Biochemicals), and anti-fortilin (28) antibodies were used along with appropriate horseradish peroxidase-conjugated secondary antibodies (Southern Biotechnology Associates, Inc., Birmingham, AL). Bound antibodies were detected, using an enhanced chemiluminescence (ECL) kit (West Pico; Pierce) according to the instructions from the manufacturer. Densitometric analysis was performed using a Bio-Rad chemiluminescence screen, Molecular Scanner-FX, and Quantity One software system, according to the instructions from the manufacturer (Bio-Rad). The signal intensities of MCL1 and fortilin bands were divided by the signal intensity of actin band from the same time point and expressed in terms of -fold increase from the time 0 signals, after normalizing signal intensities at time 0 to one.
Small Interfering RNA in Vivo Gene Silencing Assay-RNA interference (RNAi) is the process of sequence-specific, post-transcriptional gene silencing in animals and plants, initiated by double-stranded RNA (dsRNA) that is homologous in sequence to the silenced gene. It has been shown that 21-nucleotide siRNA duplexes specifically suppress the expression of endogenous and heterologous genes in different mammalian cell lines and that the most efficient silencing is obtained with siRNA duplexes composed of 21-nucleotide sense and 21-nucleotide antisense strands, paired in a manner to have a 19-nucleotide duplex region and a 2-nucleotide overhang at each 3Ј-terminus (31,32). The siRNAs against luciferase, fortilin, and MCL1 were synthesized at Dharmacon Research, Inc. (Lafayette, CO). The siRNA against luciferase, a non-mammalian protein from Photinus pyralis (American firefly), was used to evaluate the potential presence of the nonspecific effects of irreverent siRNA on intracellular MCL1 and fortilin. The sequences of the dsRNAs of luciferase, fortilin, and MCL1 used in the current experiments are as follows. For luciferase, target sequence (cDNA) is 5Ј-CG-TACGCGGAATACTTCGA-3Ј; sense siRNA, 5Ј-CGUACGCGGAAUAC-UUCGAdTdT-3Ј; antisense siRNA: 5Ј-UCGAAGUAUUCCGCGUACG-dTdT-3Ј. For fortilin, target region (cDNA, nucleotides 121-141) is 5Ј-AAGGTAACATTGATGACTCGC-3Ј; sense siRNA, 5Ј-GGUAACAU-UGAUGACUCGCdTdT-3Ј; antisense siRNA, 5Ј-GCGAGUCAUCAAU-GUUACCdTdT-3Ј. For MCL1, target region (cDNA, nucleotides 456 -476) is 5Ј-AATAACACCAGTACGGACGGG-3Ј; sense siRNA, 5Ј-U-AACACCAGUACGGACGGGdTdT-3Ј; antisense siRNA, 5Ј-CCCGUCC-GUACUGGUGUUAdTdT-3Ј. All procedures were performed under RNase-free environment, using RNase-free water, Eppendorf tubes, and pipette tips (Ambion, Austin, TX). The transfection of U2OS cells, either parent U2OS cells or U2OS wild-fortilin-HA , with siRNA duplexes was performed using a TransIT-TKO transfection kit (Panvera, Madison, WI), according to instructions from the manufacturer. Briefly, 5 ϫ 10 4 U2OS cells/well were seeded in 24-well plates. The siRNA duplexes were dissolved in Dilution Buffer (100 mM NaCl, 50 mM Tris-Cl, pH 7.5, in RNase-free water) at the concentration of 2 M. TransIT solution was dissolved into Opti-MEM serum-free medium (Invitrogen) at the concentration of 200 g/ml. The siRNA duplex was then added to the TransIT solution at the concentration of 300 nM. After incubation at room temperature, the mixture was added to the complete media at the final siRNA-duplex concentration of 50 nM. Sixty hours after the transfection, cells were harvested by the addition of 50 l of SDS gel loading buffer. Collected samples were incubated at 45°C for 1 h. The genomic DNA in the lysate was sheared by passing the lysate through 27-gauge needles three times. The samples were then subjected to SDS-PAGE and Western blot analysis, using anti-fortilin (or in case of U2OS wild-fortilin-HA cells, anti-HA), anti-MCL1 (Santa Cruz Biotechnology, Inc), and anti-actin (Roche Molecular Biochemicals) antibodies, as described above and previously (22,28), or to a real-time reverse transcriptase polymerase chain reaction (RT-PCR) as described below. Over 10 experiments were performed using the system described here with essentially same results.
Real-time RT-PCR Assay-U2OS cells were seeded at the density of 2 million/dish onto four 10-cm dishes. The next day, cells were transfected either with siRNA-luciferase or with siRNA-MCL1 using Trans-IT KO transfection reagent, according to the instructions from the manufacturer and as described above under "Small Interfering RNA in Vivo Gene Silencing Assay." Sixty-hours after the transfection, cells were harvested by trypsinization, subjected to mRNA isolation, using RNeasy (Qiagen, Valencia, CA) and to DNase (ABI, Foster City, CA) treatment. The real-time RT-PCR was performed according to the instructions from the manufacturer, using the following primer and probe sets for the detection of fortilin transcripts: forward primer, 5Ј-AT-GACTCGCTCATTGGTG-3Ј; reverse primer, 5Ј-GCTTTCGGTACCT-TCGCCC-3Ј; and probe, 5Ј-TGCCTCCGCTGAAGGCCC-3Ј. The probe was labeled at the 5Ј end with a reporter fluorescent dye (6-carboxylfluorescein, FAM TM ) and at the 3Ј end with a fluorescent dye quencher (6-carboxytetramethylrhodamine, TAMRA TM ) (Biosearch Tech, Novato, CA). For the detection of human glucose-6-phosphate dehydrogenase (G6PDH) transcripts for normalization, the pre-developed assay mixture for G6PDH, consisting of appropriate primers and probe labeled by VIC TM and TAMRA TM (PDAR; ABI) was used. The quantitative realtime RT-PCRs were performed in quadruplicate, using the TaqMan ® RT-PCR kit (ABI) in the Applied Biosystems 7700 Sequence Detector system, according to the instructions from the manufacturer. Both fortilin and G6PDH critical threshold were determined from a single well. Using a standard curve drawn on the serially diluted human total RNA, a fortilin transcript copy number from a well were normalized to the copy number of G6PDH from the specific well and expressed as fortilin mRNA index. The exact same procedure and calculation were carried out for the determination of MCL1 transcript numbers, using the following primer and probe sets: forward primer, 5Ј-GAGGCTGG-GATGGGTTTGT-3Ј; reverse primer, 5Ј-CCAGCAGCACATTCCTG-3Ј; and probe, 5Ј-TCTTCCATGTAGAGGACCTAGAAGGTGGCA-3Ј.
Pulse-Chase Experiments-For pulse-chase experiments performed to assess the effect of MCL1 depletion on the half-life of wild-type fortilin, U2OS wild-fortilin-HA cells were seeded in eight 10-cm dishes. Next day, cells in the first four dishes were treated with siRNA-luciferase while cells in the last four dishes were treated with siRNA-MCL1, as described above. Forty-eight hours after the transfection, cells were washed three times with DMEM without cysteine or methionine (DMEM-Cys Ϫ -Met Ϫ ). Cells were then incubated with DMEM-Cys Ϫ -Met Ϫ containing 150 Ci/ml Tran 35 S L/M metabolic labeling reagents (ICN, Costa Mesa, CA) for 6 h at 37°C. After incubation, cells were washed three times with DMEM with cysteine and methionine (DMEM-Cys ϩ -Met ϩ ) and one dish was harvested from each group into radioimmune precipitation assay buffer (50 mM Tris-HCl, pH7.2, 1% Nonidet P-40, 0.1% SDS, 150 mM NaCl, 1% sodium deoxycholate, 0.5 mM PMSF, 2 g/ml aprotinin, 1 mM bentamidine, and protease inhibitor mixture (Sigma)). Remaining dishes were harvested at 1, 2, and 3 h after the addition of DMEM-Cys ϩ -Met ϩ . Immunoprecipitation was performed as described above under the "In Vivo Co-immunoprecipitation Assay" with the following modifications. After the cell lysate was cleared by centrifugation, 5 l was subjected to scintillation counting to assess the total amount of protein that was metabolically labeled; immunoprecipitants were washed four times with radioimmune precipitation assay buffer before eluted into SDS-loading buffer for SDS-PAGE. The gel was fixed, treated by Amplify (Amersham Biosciences), dried, and exposed to phosphorimager screen (Bio-Rad). Total amount of HA-fortilin protein was assessed by quantifying the signal intensity of HA-fortilin bands and normalizing them to the total amount of proteins labeled. For pulse-chase experiments performed to assess the effect of MCL-1 binding on the half-life of fortilin, U2OS wild-fortilin-HA and U2OS fortilin-R21A-HA cells were used without siRNA treatment. The rest of the procedure was exactly same as described above.

Identification of Fortilin as a MCL1-interacting Protein-To
elucidate the mechanism of action of MCL1, we screened a human cDNA library using a yeast two-hybrid system and MCL1 as bait. Of 1 ϫ 10 6 independent clones screened on SD/ϪAde/ϪHis/ϪLeu/ϪTrp plates, we identified 8 candidate clones. One of these clones (FL1) also activated the ␤-galactosidase reporter gene. Restriction enzyme analysis of FL1 showed it to consist of a ϳ900-bp sequence inserted into the pACT2 vector. Analysis of the amino acid sequence of FL1 revealed it to be identical to the carboxyl-terminal 168 amino acids of human TCTP (GenBank TM accession no. AA056563). On the basis of our observation that this protein functions as an anti-apoptotic protein, this molecule has been redesignated fortilin (from Latin fortis ϭ strong, robust) (28).
Three subsequent sets of experiments confirmed the specific interaction between fortilin and MCL1. First, isolated pACT2-FL1 and/or pAS2.1-MCL1 vectors were cotransfected into SFY 526 yeast cells, a strain different from the ones used for the original library screening (PJ69 -2A and Y187 cells). The expression of FL1 (fortilin ⌬5-172 ) or MCL1 alone did not activate the ␤-gal reporter gene of SFY526 yeast cells, whereas the expression of both molecules did (Fig. 1A). This indicated that the presence of both MCL1 and fortilin ⌬5-172 was necessary and sufficient for ␤-galactosidase reporter gene activation to occur. Thus, MCL1 specifically interacts with the C-terminal 168 amino acids of fortilin in the yeast two-hybrid system.
Second, in vitro coimmunoprecipitation assay showed that MCL1 interacted with full-length fortilin (Fig. 1B). In brief, when we incubated in vitro transcribed-translated fortilin-HA with either in vitro transcribed-translated MCL1 or PCNA, fortilin was coprecipitated with MCL1 ( Fig. 1B, lanes 1 and 3) but not with PCNA (Fig. 1B, lanes 2 and 4). This indicated that MCL1 specifically interacted with full-length fortilin in vitro.
Third, in vivo coimmunoprecipitation assay further confirmed the interaction between MCL1 and fortilin. In brief, COS-7 cells cotransfected with pcDNA3-fortilin-HA or pcDNA3-HA and with pFLAG-MCL1 or pFLAG expressed proteins encoded for by the indicated plasmids, as shown by Western blot analysis of total cell lysate with anti-HA and anti-FLAG antibodies (Fig. 1C, Total lysate (Input)). When the cell lysate was incubated with either anti-HA antibody (Fig. 1C,  lanes 1 and 3) or control antibody (lane 2), fortilin-HA was successfully immunoprecipitated (Immunoprecipitation, bottom panel, lane 3) only in the presence of fortilin-HA in the cell lysate and in the presence of specific antibody against fortilin-HA (Fig. 1C). Moreover, the presence of fortilin was necessary and sufficient for the coprecipitation of MCL1 (Fig. 1C,  Immunoprecipitation). This indicated that MCL1 specifically interacted with full-length fortilin in vivo as well. Taken together, the data from these experiments indicated the presence of a specific interaction between MCL1 and fortilin.
Fortilin Interacts with MCL1 but Not with Other Bcl-2 Family Proteins-To determine whether fortilin interacts with Bcl-2 family proteins other than MCL1, we tested the ability of

FIG. 1. Identification of fortilin as a MCL1-binding protein, as demonstrated in yeast and mammalian cells, in vitro and in vivo.
A, yeast two-hybrid assay. S. cerevisiae SFY526 cells were cotransformed with pAS2.1 or pAS2.1-MCL1 and pGAD GH or pGAD GH-FL1, the C-terminal 168 amino acids (amino acids 5-172) of fortilin (fortilin ⌬5-172 ). Transformed cells were selected on SD/ϪTrp/ϪLeu plates. Colonies grown on these plates were subjected to X-gal filter lift assay. Ϫ, no blue color on the X-gal filter lift assay; ϩ, blue color on X-gal filter lift assay fortilin to physically interact in vitro with Bcl-xL, Bax, and Bak. As shown in Fig. 2, fortilin only interacted with MCL1 and not with Bcl-xL, Bax, Bak, or PCNA (negative control). This indicated that MCL1 is unique among Bcl-2 family proteins in that it specifically associates with fortilin.
Both Fortilin and MCL1 Are Localized in the Nucleoplasm-To determine the intracellular localization of fortilin in relation to that of MCL1, we first transiently overexpressed fortilin-HA in U2OS cells and immunocytochemically evaluated the intracellular distribution of native MCL1, in relation to that of fortilin-HA. Immunostaining using anti-HA antibody revealed fortilin-HA to be always in the nucleus (Fig. 3A, top  left panel). Consistent with previous reports (12,22), MCL1 was typically localized in the nucleus (Fig. 3A, middle panels). These data suggested that both fortilin and MCL1 were present in the same intracellular compartment: the nucleus.
In addition, regardless of the presence of fortilin-HA overexpression, the MCL1 localization was identical, as seen among the cells that were transfected with empty pcDNA6 plasmid (Fig. 3A, bottom panels), the cells that took up the pcDNA6-fortilin-HA plasmid and overexpressed fortilin-HA, and the cells that did not take up the pcDNA6-fortilin-HA or expressed fortilin-HA (Fig. 3A, top panels). Together, these data suggested that fortilin overexpression did not affect the intracellular localization of MCL1.
To define the intranuclear localization of fortilin in relation to that of MCL1, we immunostained U2OS wild-fortilin-HA , using the same anti-HA and anti-MCL1 antibodies. When the crosssections of the double-stained cells were examined under a confocal microscope equipped with appropriate filters, fortilin was found predominantly in the nucleoplasm (Fig. 3B, left  panel, green color). Fortilin appeared to be absent from the nucleoli (Fig. 3B, left panel, arrowhead). In the same analysis, MCL1 was also found predominantly in the nucleoplasm, out-  side the nucleoli (Fig. 3B, middle panel, red color). When fortilin and MCL1 signals were superimposed, the expression patterns of fortilin and MCL1 were found essentially identical, both signals being present for the most part in the nucleoplasm, outside the nucleoli (Fig. 3B, right panel, yellow color). These findings indicated that fortilin and MCL1 were both present in the same intracellular compartment, i.e. in the nucleoplasm, and thus capable of interacting with each other. Furthermore, the data suggest that the MCL1-fortilin interaction might take place in the nucleoplasm.

Fortilin Expression Temporally Correlates with MCL1 Expression in Serum-stimulated Human Aortic Vascular Smooth
Muscle Cells-Although MCL1 is a member of the Bcl-2 family, it is unique in being inducible. MCL1 expression steeply increases in various hematopoietic cell lines upon stimulation by a number of growth factors (10 -18), including sera (19). Therefore, we asked whether MCL1 expression could also be upregulated in serum-stimulated human vascular smooth muscle cells. To this end, cultured human aortic vascular smooth muscle cells that had been made quiescent in serum-free media were stimulated by serum, and the status of MCL1 expression was evaluated at various time points by Western blot analysis using anti-MCL1 antibody. As shown in Fig. 4, MCL1 expression was minimal in quiescent vascular smooth muscle cells (time 0). The expression of MCL1 rapidly increased, however, upon serum stimulation and peaked around 24 h (Fig. 4). We then hypothesized that, if fortilin functionally interacted with MCL1, then the response pattern of fortilin to serum stimulation might be similar to that of MCL1 in this system. Accordingly, the expression status of fortilin was evaluated by Western blot analysis using anti-fortilin antibody. The expression of fortilin was modest in quiescent vascular smooth muscle cells (time 0) but up-regulated upon serum stimulation and maximal around 24 h (Fig. 4). Thus, the temporal expression pattern of fortilin was essentially identical to that of MCL1 in serumstimulated vascular smooth muscle cells. Taken together with the immunostaining data described above (Fig. 3, A and B), these data indicated that MCL1 and fortilin were spatially and temporally colocalized, supporting the hypothesis that MCL1 is a true physical and functional protein partner of fortilin.
MCL1 Stabilizes Fortilin-We then hypothesized that fortilin might function as a chaperone of MCL1, binding to and stabilizing MCL1, or alternatively that MCL1 might function as a chaperone of fortilin, binding to and stabilizing fortilin. To assess the stability of fortilin and MCL1 in the presence and absence of its binding partner, we turned to RNA interference as a way to reduce the total intracellular concentration of these proteins in vivo. When U2OS cells were transfected with siRNA-luciferase, the signal intensities of MCL1 and fortilin were identical to those of cells treated with the transfection reagent alone (Fig. 5A, lane 1 versus lane 2), suggesting that the introduction of irrelevant siRNA duplex would not affect the intracellular concentration of MCL1 or fortilin. As we expected, the introduction of siRNA-MCL1 and siRNA-fortilin caused the substantial depletion of intracellular MCL1 (Fig.  5A, lane 3, MCL1) and fortilin (Fig. 5A, lane 4, fortilin), respectively. In this system, we examined how the siRNA-mediated depletion of MCL1 and fortilin would affect the intracellular concentrations of fortilin and MCL1, respectively. As is shown in Fig. 5A, the fortilin depletion by siRNA-fortilin did not cause the reduction of the MCL1 level (Fig. 5A, lane 4, MCL1), suggesting that the absence of fortilin would not destabilize MCL1. Strikingly, however, the MCL1 depletion by siRNA-MCL1 significantly reduced intracellular fortilin (Fig. 5A, lane 3, fortilin). Actin signal intensities of these lanes were identical, supporting that the equal amounts of protein were loaded in each well (Fig. 5A, lanes 1-4, actin). Next, to evaluate how the depletion of MCL1 would affect the transcription of fortilin gene, U2OS cells were treated either with siRNA-luciferase (control) or with siRNA-MCL1. A real-time RT-PCR assay for MCL1 showed that cells treated with siRNA-MCL1 contained significantly less MCL1 transcripts than cells treated with siRNA-luciferase (Fig. 5B, left panel). In this system, a realtime RT-PCR assay for fortilin showed that fortilin transcripts were identical in number, regardless of the types of siRNA (Fig.  5B, right panel). These data suggest that the depletion of MCL1 protein does not change the fortilin transcript numbers (Fig.  5B) but reduce the amount of fortilin protein (Fig. 5A).
To further validate the role of MCL1 in the stabilization of fortilin protein, U2OS wild-fortilin-HA were transfected with siRNA-luciferase, siRNA-MCL1, or siRNA-fortilin, and subjected to Western blot analysis using anti-MCL1, anti-HA, and anti-actin antibodies as described under "Experimental Procedures." As is shown in Fig. 6A, the introduction of siRNAluciferase to the cells did not change the levels of MCL1 and fortilin, in comparison to cells treated with the transfection reagent alone (Fig. 6A, lane 1 versus lane 2). Cells treated with siRNA-MCL1 and siRNA-fortilin expectedly expressed less MCL1 and fortilin, respectively (Fig. 6A, lanes 3 and 4, MCL1 and fortilin, respectively). In this system, cells treated with siRNA-fortilin expressed the same amount of MCL1, suggesting that the level of fortilin protein did not affect the MCL1 protein level in cell (Fig. 6A, lane 4, MCL1). Consistent with the result shown in Fig. 5A, however, the MCL1 depletion was associated with significant reduction of fortilin-HA (Fig. 6A,  lane 3, fortilin). In a separate assay, U2OS wild-fortilin-HA cells were treated with siRNA-MCL1, harvested into SDS-loading buffer at 0, 24, 48, and 60 h, and subjected to SDS-PAGE and Western blot analysis. The effect of siRNA-MCL1 on intracellular MCL1 level was obvious as early as at 24 h, but was found to be more prominent at 48 and 60 h (Fig. 6B, MCL). The effect of siRNA-MCL1 on intracellular fortilin level followed the re- Total cell lysate from these cells were subjected to co-immunoprecipitation procedure where HA-tagged fortilin was pulled down with anti-HA and the status of co-precipitated FLAG-MCL1 was evaluated by Western blot analysis. NSB, nonspecific bands; IP, immunoprecipitation by anti-HA antibody; Western, Western blot analysis using anti-FLAG and anti-HA antibodies. B, fortilin R21A degrades more quickly than wild-type fortilin. U2OS wild-fortilin-HA cells and U2OS fortilin-R21A-HA cells were subjected to a pulse-chase experiment where cells were pulse-labeled with [ 35 S]methionine/cysteine for 6 h and the amount of intact and radiolabeled fortilin-HA (or was assessed by immunoprecipitation, SDS-PAGE, and quantitative autoradiography. The remaining radiolabeled fortilin level was calculated by the signal intensity of immunoprecipitated fortilin normalized to the total amount of protein labeled and expressed as the percentage of fortilin levels at time 0. Solid line, wild-type fortilin-HA in cells; dotted line, fortilin R21A -HA in cells. duction of MCL1 level; the level of HA-fortilin did not change between 0 and 24 h but showed significant reduction in 48 and 60 h (Fig. 6B, fortilin). Finally, the stability of fortilin in the presence or absence of MCL1 was assessed by a pulse-chase experiment. U2OS wild-fortilin-HA cells, transfected either with siRNA-luciferase or siRNA-MCL1, were pulse-labeled with [ 35 S]cysteine/methionine for 6 h; chased with cold DMEM; harvested at 0, 1, 2, and 3 h; and subjected to immunoprecipitation using anti-HA antibody. Precipitated fortilin-HA was subjected to SDS-PAGE and autoradiography, with the quantification of radioactivity of the bands. As is shown in Fig. 6C (solid line), the level of fortilin-HA did not show significant change over 3 h in the presence of MCL1. On the contrary, the level of fortilin-HA, with MCL1 depletion, showed significant decrease over 3 h (Fig. 6C, dotted line). Taken together, these data suggest that the reduced intracellular level of MCL1 destabilizes fortilin.
A Fortilin Point Mutant with Arg 21 3 Ala Substitution (Fortilin R21A ) Does Not Bind MCL1-We then generated fortilin point mutants and evaluated their ability to interact with MCL1 by in vivo co-immunoprecipitation assay. We found that one of these mutants, fortilin R21A -HA, lacked interaction with MCL1 as shown in Fig. 7A. In this experiment, the U2OS empty , U2OS wild-fortilin-HA , and U2OS fortilin-R21A-HA cells were transiently transfected with pFLAG-MCL1. Cell lysates from these three cell types all contained equal amounts of FLAG-MCL1 (Fig. 7A, Input). Upon immunoprecipitation with anti-HA antibody, the approximately same amounts of fortilin-HA and fortilin R21A were immunoprecipitated (Fig. 7A, IP, bottom panel). As expected, fortilin-HA co-immunoprecipitated FLAG-MCL1 (Fig. 7A, IP, top panel, left lane), whereas the absence of immunoprecipitated fortilin-HA was associated with no FLAG-MCL1 (Fig. 7A, IP, top panel, right lane). In this system, fortilin R21A -HA failed to co-immunoprecipitate FLAG-MCL1 (Fig. 7A, IP, top panel, middle lane). These data suggested that fortilin R21A did not interact with MCL1.
Fortilin R21A Decays Faster than Wild-type Fortilin-Finally, we subjected U2OS wild-fortilin-HA and U2OS fortilin-R21A-HA cells to pulse-chase experiments where these cells were pulse-labeled with [ 35 S]cysteine/methionine for 6 h; chased with cold DMEM; harvested at 0, 1, 2, and 3 h; and subjected to immunoprecipitation using anti-HA antibody. Precipitated fortilin-HA or fortilin R21A -HA was subjected to SDS-PAGE and quantitative autoradiography. As is shown in Fig. 7B, the level of fortilin-HA did not show significant change over 3 h (solid line), consistent with the result shown in Fig. 6C. On the contrary, the level of fortilin R21A -HA exhibited a significant decrease over 3 h, up to nearly 50% (Fig. 7B, dotted line). These data suggest that the interaction with MCL1 is necessary for fortilin to remain stable in the cells. DISCUSSION We have described the specific interaction between MCL1 and fortilin, a novel protein that prevents etoposide-induced cell death in U2OS cells (28). Our results presented here indicate that MCL1 is a protein partner of fortilin that binds and stabilizes fortilin. To our knowledge, this is the first report of an interaction between MC1 and fortilin in the literature.
The interaction between MCL1 and fortilin is likely to be biologically significant. First, the interaction occurs not only in vitro but also in vivo, in both yeast and mammalian cells (Figs. 1 (A-C) and 7A). Second, MCL1 and fortilin are both predominantly localized in the nucleus by immunostaining analysis (Fig. 3, A and B). Third, both MCL1 and fortilin are rapidly induced by serum stimulation and exhibit similar expression kinetics in aortic vascular smooth muscle cell system (Fig. 4). Of all the Bcl-2 family proteins, only MCL1 is inducible (10 -19). Consistently, as suggested by in vitro binding assays (Fig.   2), fortilin apparently interacts with only one member of the Bcl-2 family of proteins: MCL1.
RNAi is the process of sequence-specific, post-transcriptional gene silencing in animals and plants, initiated by dsRNA that is homologous in sequence to the silenced gene (31). It has been shown that 21-nucleotide siRNA duplexes specifically suppress expression of endogenous and heterologous genes in different mammalian cell lines and that the most efficient silencing is obtained with siRNA duplexes composed of 21-nucleotide sense and 21-nucleotide antisense strands, paired in a manner to have a 19-nucleotide duplex region and a 2-nucleotide overhang at each 3Ј-terminus (32). This system would be advantageous over the overexpression system where fortilin or MCL1 is overexpressed and the stability of its partner is evaluated. This is because only a small amount of a native protein may be needed to fully stabilize its binding partner. Using the siRNA system, we first showed that MCL1 does not require fortilin to remain stable. On the contrary, MCL1 depletion quickly destabilized fortilin (Figs. 5A and 6 (A-C)). In addition, a point mutant of fortilin that lacks MCL1 binding was much more prone to degradation than wild-type fortilin (Fig. 7B). It is thus likely that fortilin, through its binding to MCL1, becomes more stabilized and less susceptible to degradation.
In the Introduction, we discussed that the inducibility of MCL1 is unique among anti-apoptotic Bcl-2 family proteins and that MCL1 is rapidly up-regulated upon growth and differentiation stimuli to cells (10 -19). With this in mind, one can speculate that the increase in stability of fortilin by MCL1 may be used by growing and differentiating cells to quickly increase the intracellular pro-survival environment by countering a potential instability associated with growth and differentiation. This stabilization of one anti-apoptotic protein (in this case fortilin) by another anti-apoptotic protein (in this case MCL1) has not been reported in literature and calls for further investigation.