The N Terminus of the Anti-apoptotic BCL-2 Homologue MCL-1 Regulates Its Localization and Function*

The BCL-2 homologue MCL-1 plays an important role in the regulation of cell fate by blocking apoptosis as well as regulating cell cycle. MCL-1 has an unusual N-terminal extension, which contains a PEST domain and several phosphorylation sites that have been suggested to regulate its turnover. Here we report that the first 79 amino acids of MCL-1 regulate its subcellular localization. Deletion of this domain impairs both its mitochondrial localization and its anti-apoptotic activity. Conversely, expression of the N terminus of MCL-1 promotes both the association of MCL-1 with mitochondria and cell survival in a fashion that is dependent on the presence of endogenous MCL-1. In addition, the N terminus of MCL-1 has an antagonistic effect on proliferation. Although MCL-1 decreases proliferation through binding to proliferating cell nuclear antigen and cyclin-dependent kinase 1 in the nucleus, the N terminus of MCL-1 accelerates cell division. On the other hand, deletion of this region further increases the anti-proliferative activity of MCL-1. These results suggest that the N terminus of MCL-1 plays a major regulatory role, regulating coordinately the mitochondrial (anti-apoptotic) and nuclear (anti-proliferative) functions of MCL-1.

tion of a series of apoptotic proteases, known as caspases, through the formation of the apoptosome, a protein complex composed of APAF-1, caspase-9, and cyt c (1).
Events leading to the release of cyt c from mitochondria are controlled by the BCL-2 family of proteins that are characterized by the presence of at least one BCL-2 homology (BH) domain (1,3). Several BCL-2 homologues also possess a C-terminal transmembrane domain that mediates their association with mitochondria (4,5). Pro-apoptotic BAX and BAK, containing BH1-3, are required for permeabilization of mitochondria, although their exact mechanism of action is still debated. Both proteins are kept inactive in healthy cells and become activated through a change in conformation upon induction of apoptosis, leading to their oligomerization (1,3). This activation is regulated by a balance between two antagonistic classes of BCL-2 homologues: BH3-only and anti-apoptotic proteins. Anti-apoptotic BCL-2 proteins such as BCL-2 itself, BCL-X L , and MCL-1 inhibit apoptosis by blocking either directly or indirectly the activation of BAX and BAK (1,3). Interaction between BCL-2 family members is mediated by a groove formed by the BH1-3 domains in the BCL-2 homology region on the one hand and the ␣-helix formed by the BH3 of the pro-apoptotic proteins on the other hand (2). These interactions play a requisite role in the regulation of BCL-2 family proteins.
MCL-1 is an anti-apoptotic BCL-2 homologue that is necessary for early embryonic development as well as for the generation and maintenance of hematopoietic cell lineages (6,7). MCL-1 has a short half-life, and its protein levels are tightly regulated both transcriptionally and through proteasomal degradation (8). Degradation of MCL-1 after UV irradiation is required for induction of apoptosis (9). Conversely, MCL-1 protein levels are increased in several human cancers, and its overexpression in transgenic mice leads to malignancies (8,10). Paradoxically, MCL-1 has also been reported to delay cell cycle progression in S/G 2 through interaction with proliferating cell nuclear antigen (PCNA) and cyclin-dependent kinase 1 (CDK1) (11,12). This has been associated with a nuclear localization of MCL-1, as opposed to its mitochondrial anti-apoptotic function. MCL-1 is, thus, an important regulatory protein in the context of development and oncogenesis.
Structurally, MCL-1 differs from other anti-apoptotic BCL-2 proteins in two respects. First, differences in amino acid charge distribution in the BCL-2 homology region allow for a different specificity toward BH3-only proteins as compared with BCL-2 and BCL-X L (13). Second, MCL-1 possesses a 170-amino acid extension at its N terminus that contains two PEST domains and several phosphorylation sites. This N-terminal extension is a potential regulatory domain for the degradation of MCL-1 through its PEST sequences (14), which contain phosphorylation sites that promote the recruitment of the E3 ligase ␤-TRCP (15,16) as well as caspase cleavage sites (17).
Here we report that the first 79 amino acids of MCL-1 regulate its subcellular localization, promoting its association with mitochondria. Overexpression of the N terminus of MCL-1 (Nt-MCL) promoted cell survival by recruiting more MCL-1 at this organelle, whereas deletion of this sequence (MCL-1 ⌬N79) partially impaired both its mitochondrial localization and anti-apoptotic activity. Conversely, MCL-1⌬N79 inhibited cell growth more potently than MCL-1, whereas Nt-MCL increased proliferation. These results suggest that that the N terminus of MCL-1 plays an important regulatory role in the function of MCL-1.

MATERIALS AND METHODS
MCL-1 Constructs-To generate Nt-MCL, full-length MCL-1 in pcDNA3.1/V5-His(A) (cloned between BamHI and EcoRI) was digested using BamHI and NotI and cloned into the same vector to give a fusion protein expressing the first 79 amino acids of MCL-1 with a His and a V5 tag at its C terminus. V5-MCL-1⌬N79 and V5-MCL-1⌬N54 were generated by replacing the first 79 or 54 amino acids of MCL-1 by a V5 tag using PCR and cloned into pcDNA3.1/His-V5(A) between BamHI and EcoRI restriction sites. MCL-1⌬TM was generated by removing amino acids 321-350 using PCR and cloned into pcDNA3.1/His-V5. Nt-MCL-GFP was generated by fusing the first 79 amino acids of MCL-1 to the N terminus of GFP using PCR. Both GFP and Nt-MCL-GFP were cloned into pcDNA3.1/His-V5 vector.
Cell Culture, Transfections, and Infections with Adenoviral Vectors-HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum and 20 mM L-glutamine. To generate cell lines stably expressing full-length MCL-1 or the deletion mutants, HeLa cells were transfected with the various constructs using Lipofectamine 2000 (Invitrogen) and selected with G418. For UV irradiation, cells were grown to 95% confluency (or 50% confluency for immunofluorescence), the medium was removed, and the cells were irradiated using 200 mJ/cm 2 UVB. Fresh medium (containing 50 M zVAD-FMK in the case of immunofluorescence) was then added, and the cells were further incubated for the indicated time. Infections with adenoviral vectors coding for tBID and NOXA (a gift of Dr. Gordon Shore) were carried out as described (18) using 50 and 10 plaque-forming units/cell of virus, respectively. All infections were carried in the presence of 50 M zVAD-FMK. For determination of caspase activity, 50 g of cell extract were incubated with 100 M DEVD-p-nitroanilide in 50 mM HEPES, pH 7.5, 150 mM NaCl, 0.5% Triton X-100, and 2 mM dithiothreitol for 30 min at 37°C. Absorbance was measured at 405 nm.
Half-life Determination-To determine the half-life of MCL-1, HeLa cells were transiently transfected with the various constructs for 24 h, after which they were either irradiated using 200 mJ/cm 2 UVB or treated with 10 M cycloheximide (CHX). The experiment was also repeated in the presence of 10 M MG132 as a control. Cell were harvested at the indicated times and analyzed by Western blot using a MCL-1-specific antibody as well as a p85 PI3K antibody as a loading control. The density of the bands was quantified using Image J software and normalized using the p85 PI3K blot.
Immunofluorescence-For immunofluorescence, cells were grown on coverslips and treated as indicated in the figure legends. Cells were then fixed with 4% paraformaldehyde and analyzed by immunofluorescence using AlexaFluor 488 and AlexaFluor 594 secondary antibodies (Molecular Probes). Coverslips were then visualized using an Axioplan 2 microscope (Carl Zeiss MicroImaging, Inc.) with a 100 ϫ 1.30 oil objective (Carl Zeiss MicroImaging, Inc.). Images were captured and overlaid using Northern Eclipse software.
siRNA-Cells were transfected with siRNA specific for GFP, BCL-X, and MCL-1 (Santa Cruz Biotechnology) using Silent-Fect (Bio-Rad) transfection reagent. 10 nM siRNA was used for siRNA GFP and siRNA BCL-X, whereas 6.67 nM was used for siRNA MCL-1 to reduce toxicity. Cells were collected after 24 h analyzed for protein expression by Western blot.
Isolation of Mitochondria and Alkaline Extraction-Mitochondria were isolated as previously described (19) with the following modifications. Cells were harvested and resuspended in HIM buffer (200 mM mannitol, 70 mM sucrose, 10 mM HEPES, pH 7.5, 1 mM EGTA). Cells were broken by passing them 24 times through a 25-gauge needle. Nucleus and cell debris were removed by centrifugation at 1000 ϫ g. The resulting supernatant was centrifuged at 9000 ϫ g to pellet the heavy membrane fraction containing mitochondria. For alkaline extraction, 30 g of heavy membrane were incubated for 30 min on ice with 0.1 M Na 2 CO 3 followed by centrifugation at 50 000 ϫ g to recover the membranes (19).
BrdUrd Incorporation-Cells were transfected with the indicated constructs and 24 h later pulsed for 1 h with 10 M BrdUrd. Cells were then fixed with 4% paraformaldehyde and stained for GFP or MCL-1. After a second fixation with 4% paraformaldehyde, coverslips were treated with 4 M HCl in 1% Triton X-100 for 15 min, washed, and stained for BrdUrd.

RESULTS
The First 79 Amino Acids of MCL-1 Regulate Its Anti-apoptotic Activity after UV Irradiation-MCL-1 has a long N-terminal region containing a PEST domain, which is not required for binding to other BCL-2 homologues (13). Caspase-mediated cleavage of this domain (at Asp-127 and Asp-157) has been reported to impair the anti-apoptotic function of MCL-1 (17,20). To test the anti-apoptotic activity of different N-terminal deletions of MCL-1 that do not disrupt its PEST domain and, therefore, should not affect its half-life, HeLa cells were transiently transfected with several MCL-1 deletion mutants, and cyt c release was analyzed by immunofluorescence after irradiation of the cells with 200 mJ/cm 2 UV. As shown in Fig. 1A, MCL-1 almost completely blocked cyt c release in this context, whereas the activity of mutants lacking the first 54 (MCL-1⌬N54) or 79 amino acids (MCL-1⌬N79) was partially impaired. A similar effect was observed when the C-terminal transmembrane domain was deleted (MCL-1⌬TM). The increased cyt c release observed for the MCL-1 mutants, as compared with wild-type MCL-1, was not a consequence of their lower expression as they were all expressed at similar or higher levels (MCL-1⌬N79) as compared with full-length MCL-1 (Fig. 1B). Similarly, stable expression of MCL-1⌬N79 in HeLa cells, despite being expressed at higher levels than the wild-type protein (Fig. 1C), only partially prevented cyt c release and caspase activation, as compared with the full-length protein, after UV irradiation ( Fig. 1

, D and E).
MCL-1 is a protein with a short half-life, and its degradation after UV irradiation is required for subsequent induction of apoptosis (9). A recent report indicating that deletion of the first 20 or so amino acids of MCL-1 led to an increased stability of the protein (21) as well as the greater expression of MCL-1⌬N79 ( Fig. 1) suggested that MCL-1⌬N79 could have an altered half-life. To test whether the impaired activity of MCL-1⌬N79 is due to an altered half-life, the degradation rate of the different MCL-1 constructs was tested after CHX treatment and UV irradiation. The half-life of overexpressed MCL-1, as determined using CHX, was very short (about 60 min) and was not influenced by the ⌬N79 deletion ( Fig. 2A). Similarly, UV irradiation of HeLa cells provoked the rapid degradation of both MCL-1 and MCL-1⌬N79, with similar kinetics (Fig. 2A).
The caspase-cleaved form of BID (tBID) promotes apoptosis by directly activating BAX and BAK at the mitochondria (22,23). Importantly, apoptosis induced after infection of HeLa cells with an adenovirus expressing tBID (Ad tBID) occurs in the absence of MCL-1 degradation (Fig. 2B). In this context full-length MCL-1, but not MCL-1⌬N79, could reduce signifi-cantly tBID-induced cyt c release (Fig. 2C). Altogether, these results suggest that the altered anti-apoptotic activity of MCL-1⌬N79 is not a consequence of altered sensitivity to proteasomal degradation.
MCL-1 prevents apoptosis at least in part by sequestering pro-apoptotic BAK (24). Although complete deletion of the N-terminal portion of MCL-1 does not affect binding to BH3only proteins (13), the effect of the N79 deletion on the interaction between BAK and MCL-1 was tested. Because fulllength MCL-1 and BAK interacted very weakly in CHAPS-containing buffer (data not shown), the experiment was carried in Triton X-100-containing buffer as a measure of the capacity of these proteins to interact. Wild-type MCL-1 and MCL-1⌬N79 were transiently transfected into HeLa cells along with FLAG-BAK, followed by immunoprecipitation with an anti-FLAG antibody. As shown in Fig. 2D, FLAG-BAK immunoprecipitated both MCL-1 constructs with a similar efficiency, indicat- ing that the N79 deletion does not directly disrupt the interaction between MCL-1 and BAK.
The First 79 Amino Acids of MCL-1 Are Required for Its Mitochondrial Localization-MCL-1 localizes to the mitochondria where it can block apoptosis by interacting with pro-apoptotic BAK (24,25). Because deletion of the C-terminal transmembrane domain of MCL-1 had a similar effect on its anti-apoptotic activity than the ⌬N79 deletion, we analyzed the subcellular localization of the different MCL-1 constructs. This was tested in transiently transfected HeLa cells using a MCL-1-specific antibody, as the levels of endogenous MCL-1 in HeLa cells are below detection levels by immunofluorescence (see the untransfected cells which do not show MCL-1 staining in Fig. 3A). As shown in Fig. 3A, full-length MCL-1 co-localized with the mitochondrial protein TOM20. In contrast, MCL-1⌬N79 localized to the cytosol in a majority of the cells, although a weak association with mitochondria could still be observed (Fig. 3A, V5-MCL-⌬N79 is depicted; similar results were obtained in absence of the tag (not shown)). As expected, deletion of the C-terminal transmembrane domain, which is required to target BCL-2 homologues to intracellular membranes (5), showed no mitochondrial localization (MCL-1⌬TM; Fig. 3A).
To determine whether the N-terminal 79 amino acids of MCL-1 act as a mitochondrial targeting sequence, this domain was expressed as a GFP fusion protein (Nt-MCL-GFP). As shown in Fig. 3B, Nt-MCL-GFP localized to the cytosol. Similar results were obtained when Nt-MCL was tagged with a V5 epitope at its C terminus (not shown). This suggests that although required for the proper mitochondrial localization of MCL-1, this domain is not sufficient.
Nt-MCL Enhances the Mitochondrial Localization of MCL-1-The number of cells demonstrating mitochondrial localization of MCL-1 was quantified for each construct. Only cells in which most of MCL-1 colocalized with TOM20 (as for MCL-1 in Fig. 3A) were counted as mitochondrial. Although fulllength MCL-1 was mitochondrial in most cells, MCL-1⌬N79 co-localized with TOM20 in only 15% of the cells (Fig. 4A). As expected, there was no association of MCL-1⌬TM with mitochondria (Fig. 4A). The effect of the expression of Nt-MCL on the localization of MCL-1 was then tested. HeLa calls stably expressing a V5-tagged version of Nt-MCL (Nt-MCL-V5) were transiently transfected with the different constructs, examined by immunofluorescence using a MCL-1-specific antibody, and scored for mitochon- drial localization of the MCL-1 constructs. As shown in Fig.  4A, Nt-MCL-V5 significantly increased the mitochondrial localization of both full-length MCL-1 and MCL-1⌬N79 but had no effect on MCL-1⌬TM, consistent with the requirement of the transmembrane domain for mitochondrial targeting.
MCL-1 is synthesized on cytosolic ribosomes before being imported into mitochondria. If the cytosolic localization of MCL-1⌬N79 is the consequence of a slower rate of import, it should nevertheless accumulate at the mitochondria given enough time. Therefore, by preventing synthesis of new MCL-1⌬N79 using CHX, it should be possible to observe its accumulation at the mitochondria. As shown in Fig. 4B, the number of cells with mitochondrial MCL-1⌬N79 did increase in a timedependent manner after the inhibition of protein synthesis, whereas the localization of MCL-1 and MCL-1⌬TM was not affected. In addition, Nt-MCL-V5 did increase the rate at which cells with mitochondrial MCL-1⌬N79 appeared (Fig. 4C) but had no effect on the half-life of MCL-1⌬N79 under these con-ditions (Fig. 4D). These results indicate that mitochondrial targeting is less efficient in MCL-1⌬N79 and that the presence of Nt-MCL can at least partially compensate for this defect.
Although the bulk of endogenous MCL-1 is found associated with mitochondria, some can also be found in the cytosol or nucleus under some conditions (9,11,12). To test whether Nt-MCL can enhance the mitochondrial localization of endogenous MCL-1, HeLa cells stably expressing vector alone or Nt-MCL-V5 were fractionated, and the heavy membrane-containing mitochondria were isolated (19). The presence of Nt-MCL resulted in an increase in the amount of MCL-1 associated with heavy membrane (HM; Fig. 5, A and B), suggesting that Nt-MCL stimulates the association between MCL-1 and mitochondria. However, there was also a small increase (Ϸ20%) in the total amount of MCL-1 in Nt-MCLexpressing cells (Fig. 5, A and B) that may partially account for the increase in mitochondrial MCL-1. Therefore, to better establish the role of Nt-MCL in promoting the association between MCL-1 and mitochondria, we looked at its insertion into mitochondria. MCL-1, like other anti-apoptotic BCL-2 homologues, is predicted to insert into the outer mitochondrial membrane through its C-terminal transmembrane domain and, thus, be resistant to extraction by 0.1 M Na 2 CO 3 . As previously reported (25), only a fraction of MCL-1 was resistant to alkaline extraction, and in contrast with total mitochondrial MCL-1, the amount of alkaline-resistant MCL-1 did not increase in the cells expressing Nt-MCL (Fig. 6, A and  B). This suggests that Nt-MCL promotes the association of MCL-1 with mitochondria but not its insertion into the membrane. Again, the effects of Nt-MCL were not the consequence of an altered stability of MCL-1, as the degradation of endogenous MCL-1 after CHX or UV was the same in the presence or absence of Nt-MCL-V5, whereas the addition of the proteasome inhibitor MG132 could delay both CHX-and UV-induced degradation of endogenous MCL-1 (Fig. 5C).
Nt-MCL Promotes Cell Survival-The anti-apoptotic function of MCL-1 is dependent on its association with BAK at the mitochondria (24). The increase in mitochondrial MCL-1 caused by the expression of Nt-MCL should, therefore, promote cell survival after apoptosis induction. To test this hypothesis, HeLa cells stably expressing Nt-MCL-V5 were irra- FIGURE 3. MCL-1⌬N79 is predominantly cytosolic. A, HeLa cells were transiently transfected with the indicated constructs, fixed, and double-stained with antibodies against mitochondrial TOM20 (AlexaFluor 488 (green)) and MCL-1 (AlexaFluor 594 (red)). Cells were then analyzed by immunofluorescence. B, HeLa cells were transfected with Nt-MCL-GFP, fixed, and stained with an antibody against mitochondrial TOM20 (AlexaFluor 594 (red)) along with the GFP fluorescence. Cells were then analyzed as in A. Representative images are shown. diated with 200 mJ/cm 2 UV and analyzed for cytochrome c release by immunofluorescence and caspase activity using the substrate DEVD-p-nitroanilide. Expression of Nt-MCL resulted in a decrease in both cyt c release (Fig. 6A) and caspase activity (Fig. 6B). Similar results were obtained when apoptosis was induced with Ad tBID (Fig. 6A). In addition, transient expression of Nt-MCL-GFP significantly reduced cyt c release from UV-treated HeLa cells (Fig. 6C). These results suggest that Nt-MCL possesses an anti apoptotic activity, consistent with a role in recruiting MCL-1 at the mitochondria. The anti-apoptotic activity of Nt-MCL should, therefore, be dependent on the presence of endogenous MCL-1.
Several approaches were used to test whether Nt-MCL requires endogenous MCL-1 for its function. NOXA is a BH3only protein that specifically binds to MCL-1 and A1 to inactivate them, resulting in a greater sensitivity to apoptosis (18,24,26). Because A1 is absent from HeLa cells (24), an adenovirusexpressing human NOXA (Ad NOXA) was used to selectively inhibit MCL-1 in these cells. Two hours after UV irradiation, HeLa cells do not yet show signs of apoptosis such as cyt c release (Fig. 7A). However, infection of these cells with Ad NOXA, although insufficient to cause cyt c release on its own, sensitized them to UV-induced cytochrome c release, which could be prevented by overexpression of anti-apoptotic BCL-2 (Fig. 7A). Expression of Nt-MCL in this context did not provide any protection (Fig. 7A). The levels of MCL-1 protein were also knocked down using siRNA (Fig. 7B). As with Ad NOXA, loss of  The presence of MCL-1 was analyzed by Western blot using a MCL-1-specific antibody. The blot was reprobed with an antibody against the mitochondrial protein TOM20 as a loading control (right panel). B, the density of the MCL-1 signal in the different fractions as well as in whole cell lysates (Total) was quantified using Image J software. Shown are the results of three independent experiments Ϯ S.D. *, p Յ 0.01. HM, heavy membrane. C, Nt-MCL does not affect the half-life of endogenous MCL-1. HeLa cells stably expressing Nt-MCL or vector alone were treated with either 10 g/ml CHX (left panel) or 200 mJ/cm 2 UV (right panel) for the indicated times. Alternatively, the degradation of endogenous MCL-1 was analyzed under the same conditions in the presence of 10 M MG132. MCL-1 expression was analyzed by Western blot using a MCL-1-specific antibody. The blot was then probed with a p85 PI3Kspecific antibody as a loading control. Quantification was achieved using Image J software. Shown are the results of at least three independent experiments Ϯ S.D.
MCL-1 expression resulted in sensitization of the cells to UVinduced cytochrome c release that was not blocked by Nt-MCL (Fig. 7B). Of note, knocking down a different anti-apoptotic BCL-2 homologue, BCL-X, did sensitize cells to UV-induced apoptosis, but Nt-MCL could still inhibit cytochrome c release in this context (Fig. 7B), indicating that the absence of protection by Nt-MCL is not due to a general sensitization of the cells to apoptosis. Altogether, these results suggest that Nt-MCL specifically requires the presence of full-length MCL-1 to block apoptosis.
To test the effect of Nt-MCL on the anti-apoptotic activity of the various MCL-1 mutants independent of endogenous MCL-1, we used MCL-1 null MEFs (MCL-1 ⌬/Ϫ ). These cells were transfected with the various constructs, and the effect of Nt-MCL-GFP on their survival was tested. Infection with Ad tBID was used as an apoptotic inducer for these experiments because it allows dissociating its anti-apoptotic activity from possible consequence of its rapid turnover. As expected, MCL-1 null MEFs were more sensitive than their wild-type counterparts (Fig. 7C). In addition, although Nt-MCL-GFP could protect the wild-type cells from tBID-induced cyt c release, it had no effect on MCL-1 null cells (Fig. 7C), consistent with the Noxa and siRNA experiments (Fig. 7, A and B). Similar to HeLa cells, MCL-1 did prevent cyt c release in MCL-1 null MEFs transfected with GFP alone, whereas MCL-1⌬N79 and MCL-1⌬TM were much less efficient ( Fig. 7C; expression levels are shown in Fig. 7D). The effect of Nt-MCL-GFP was then tested. As shown in Fig. 7C, Nt-MCL-GFP prevented cyt c release in MCL-1 null MEFs expressing MCL-1⌬N79, but not MCL-1⌬TM, consistent with ability to target MCL-1⌬N79 but not MCL-1⌬TM to mitochondria. Of note, Nt-MCL-GFP did not increase the expression of the MCL-1 constructs in the MEFs (Fig. 7E). In fact, despite MCL-1⌬N79 being more efficient at preventing cytochrome c release in the presence of Nt-MCL-GFP, its expression was lower. These results suggest that Nt-MCL prevents apoptosis by targeting MCL-1 to mitochondria rather than by increasing MCL-1 expression.
The N Terminus of MCL-1 Regulates Its Nuclear Function-MCL-1 has been reported to localize to the nucleus under certain circumstances, where it can slow down the cell cycle by interacting with PCNA and CDK1 (11,12). Although only a small amount of nuclear MCL-1 can be detected in transfected HeLa cells, both MCL-1⌬N79 and MCL-1⌬TM showed significant nuclear localization ( Fig. 8A; the nuclear localization of the mutants can also be observed in Fig. 3A). We then tested whether, by regulating the localization of MCL-1, the N terminus of MCL-1 could negatively regulate its anti-proliferative activity. HeLa cells were transfected with various MCL-1 constructs, and proliferation was assessed using BrdUrd incorporation. As previously reported, expression of MCL-1 in HeLa cells did reduce their proliferation rate (Fig. 8B). Both MCL-1⌬N79 and MCL-1⌬TM showed a more robust inhibition of BrdUrd incorporation than MCL-1. On the other hand, more cells expressing Nt-MCL-GFP than the control GFP incorporated BrdUrd into their DNA. Again, the effect did not correlate with expression levels of MCL-1, as cells expressing Nt-MCL-GFP had more MCL-1 than cells expressing GFP alone (Fig.  8C). Altogether, these results are consistent with a role for the N terminus of MCL-1 in promoting the association of MCL-1 with mitochondria, at the expense of its nuclear localization.

DISCUSSION
MCL-1 is a highly regulated BCL-2 homologue, presumably because of its requisite role in development and maintenance of homeostasis (6,7,27). Its expression is tightly regulated by cytokines and other growth factors (28). In addition, MCL-1 is a short-lived protein as a consequence of its targeting for proteasomal degradation by the BH3-containing E3-ligase Mule (29,30). However, other regulatory mechanisms are likely to play a role in MCL-1 turnover as it is still degraded in the absence of Mule, albeit at a slower rate (30). For example, both the BH3-only protein NOXA and phosphorylation by glycogen synthase kinase 3 promote the degradation of MCL-1, the latter through the E3-ligase ␤-TRCP (15,16,24). The glycogen synthase kinase 3 phosphorylation site is located within the N-terminal half of MCL-1, within the PEST domain, suggesting that this domain has an important regulatory function. In addition, deletion of first 17 amino acids of MCL-1 have been recently suggested to stabilize MCL-1 (21), and although deletion of the first 79 amino acids of MCL-1 did not significantly increase its half-life in this study, the increased expression levels of the mutant were consistent with such a role. Here we show that the N terminus of MCL-1 also plays an important regulatory role in controlling the mitochondrial localization of MCL-1.
All anti-apoptotic BCL-2 homologues, including MCL-1, possess a C-terminal hydrophobic segment that targets them to intracellular membranes such as endoplasmic reticulum or mitochondria (4,5). Information contained within this sequence and the flanking positive charges is believed to be sufficient to target these proteins to specific organelles (5). In fact, deletion of the C-terminal transmembrane domain of MCL-1 disrupts its interaction with mitochondria (MCL-1⌬TM; Figs. 3 and 4), indicating that the transmembrane domain is essential for mitochondrial targeting. However, because deletion of the first N-terminal 79 amino acids of MCL-1 (MCL-1⌬N79) also leads to a predominantly cytosolic localization of MCL-1, this region is likely to carry additional information required for the proper localization of MCL-1. The expression of the N-terminal region (Nt-MCL), although itself cytosolic, can partly revert the cytosolic localization of MCL-1⌬N79 (but not MCL-1⌬TM), suggesting that this domain is not a classical signal sequence. It might, however, modulate the interaction between MCL-1 and other co-factors required for its translocation to the mitochondria. For example, the mitochondrial localization of BCL-2 and BCL-X L has previously been linked to their association with the FK506-binding protein FKBP38 (31).
The activity of anti-apoptotic BCL-2 homologues is dependent on their association with intracellular membranes such as the endoplasmic reticulum and mitochondria. It is, therefore, not surprising that disruption of their association with these organelles would at least partially impair their anti-apoptotic activity. One of the main targets of MCL-1 is pro-apoptotic BAK, which is localized at the surface of the mitochondria (24), making it especially important for MCL-1 to associate with mitochondria to promote survival. Indeed, disruption of MCL-1 interaction with mitochondria by deleting either its N terminus (MCL-1⌬N79) or C-terminal transmembrane domain (MCL-1⌬TM) impaired its anti-apoptotic function. This loss of function could be dissociated from effects on the turnover of MCL-1 since MCL-1⌬N79 is at least as stable as MCL-1, expressed at higher levels, and the effects were still observed when apoptosis was induced with Ad tBID, which does not cause the degradation of MCL-1. Conversely, Nt-MCL promoted both the association of MCL-1 with mitochondria and cell survival after induction of apoptosis with both UV and Ad tBID, in a fashion that is dependent on the presence of MCL-1. The effects of NT-MCL could also be dissociated from the effects on MCL-1 expression as it promoted survival in both HeLa and MCL-1 null MEFs despite having opposite effects on the expression levels of MCL-1 and MCL-1⌬N79 in the different settings. Of note, the increased mitochondrial MCL-1 in Nt-MCL-expressing cells was not reflected in the amount of MCL-1 inserted into the outer mitochondrial membrane (alkali-resistant). Because Nt-MCL-expressing cells are nonetheless more resistant to apoptosis, this could indicate that the alkalisensitive MCL-1 constitutes the active fraction of the protein, similar to BCL-W, which undergoes a change from an active alkali-sensitive form to inactive alkali-resistant form after induction of apoptosis (32).
Several anti-apoptotic BCL-2 homologues possess an antiproliferative activity in addition to their well characterized antiapoptotic function (for review, see Ref. 33). Although BCL-2 and BCL-X L affect G 0 /G 1 transition, MCL-1 acts at the S/G 2 checkpoint. This has been proposed to be through the interaction between MCL-1 and either PCNA or CDK1 (11,12). Of note, both these interactions take place in the nucleus, whereas most of MCL-1 is associated with mitochondria in healthy cells, thus providing a possible mechanism to regulate the anti-proliferative function of MCL-1. The N terminus of MCL-1 indeed seems to regulate the balance between these two seemingly opposite roles of MCL-1. Specifically, expression of the first 79 amino acids of MCL-1 promoted the association of MCL-1 with mitochondria at the expense of its nuclear anti-proliferative function (Fig. 8). Interestingly, proteolytic cleavage of the N terminus of MCL-1 (around amino acid 17) was recently shown to occur in normally growing cells (21). Although this cleavage was shown to enhance the stability of MCL-1, it could also potentially regulate the anti-apoptotic (mitochondrial) versus anti-proliferative (nuclear) activities of MCL-1. It is also tempting to speculate that phosphorylation of serine 64 by CDK1 (34) regulates the balance between these activities. In any case, the studies presented here have revealed a novel and important function for the N terminus of MCL-1 in the regulation of this important protein.