|Home | About | Journals | Submit | Contact Us | Français|
Antiapoptotic myeloid cell leukemia 1 (MCL-1) is an essential modulator of survival during the development and maintenance of a variety of cell lineages. Its turnover, believed to be mediated by the ubiquitin-proteasome system, facilitates apoptosis induction in response to cellular stress. To investigate the contribution of ubiquitinylation in regulating murine MCL-1 turnover, we generated an MCL-1 mutant lacking the lysine residues required for ubiquitinylation (MCL-1KR). Here, we demonstrate that despite failing to be ubiquitinylated, the MCL-1KR protein is eliminated at a rate similar to that of wild-type MCL-1 under basal and stressed conditions. Moreover, the degradation of wild-type MCL-1 is not affected when ubiquitin-activating enzyme E1 activity is blocked. Likewise, both wild-type and MCL-1KR proteins are similarly degraded when expressed in primary lymphocytes. Supporting these findings, unmodified, in vitro-translated MCL-1 can be degraded in a cell-free system by the 20S proteasome. Taken together, these data demonstrate that MCL-1 degradation can occur independently of ubiquitinylation.
Apoptosis is a genetically controlled cell suicide program which is regulated by members of the BCL-2 family and is essential for development and maintenance of multicellular organisms (16). Antiapoptotic MCL-1 was identified as a gene induced in a human myeloblastic leukemia cell line treated to differentiate in vitro and found to have homology with antiapoptotic BCL-2 (34, 69). Its germ line ablation results in an early embryonic lethality, and lineage-specific gene ablation has revealed that Mcl-1 is essential for maintaining the survival of lymphocytes, neutrophils, neurons, bone marrow progenitors, and hematopoietic stem cells (4, 25, 50, 51, 54, 58).
MCL-1's short half-life (estimated at less than 1 h) is unique among antiapoptotic BCL-2 family members. Under basal conditions, human MCL-1 undergoes rapid protein turnover, but the control of this constitutive degradation pathway is incompletely understood (48). MCL-1 possesses a distinctive amino terminus and several proline-glutamic acid-serine-threonine (PEST)-rich regions; however, its turnover appears to be unaffected by deletion of these regions (2). MCL-1 can be cleaved by caspases and granzyme B, which proteolytically degrade MCL-1 during cell death (14, 27, 29). In addition, human MCL-1 can be ubiquitinylated and degraded by the proteasome (48). Several levels of degradation control have been postulated. In HeLa cells, MCL-1 undergoes constitutive protein turnover independent of cell death signaling (48). In contrast, in growth factor-dependent cells, regulation of MCL-1 stability is modulated by cytokine signaling (20, 21, 41, 71).
Ubiquitinylation of target proteins is processed through a multienzyme cascade (30). Ubiquitin is activated by one of two ubiquitin-activating enzymes (E1), transferred to one of several ubiquitin-conjugating enzymes (E2), and then finally transferred to one of multiple ubiquitin protein ligases (E3), which covalently attaches ubiquitin to lysine residues within the target protein. Polyubiquitin chains, formed by the successive conjugation of ubiquitin to internal lysine residues of the previously conjugated ubiquitin moiety, are recognized and degraded by the 26S proteasome (56).
Two E3 ligases have been implicated in mediating MCL-1 ubiquitinylation. MULE (also called LASAU1, ARF-BP1, or HUWE1), a HECT-domain (homologous to the E6-AP carboxyl terminus) family E3 ligase, possesses a BH3 domain similar to that of proapoptotic BAK that allows it to selectively target MCL-1 (65, 72). While MULE may target MCL-1 for degradation, it is not solely MCL-1 specific, as it also ubiquitinylates p53, E3Histone, c-Myc, and CDC-6 (1, 9, 26, 40). The other E3 ligase, β-TrCP, is a Skp1-CUL1-F box protein (SCF) family member that requires MCL-1 phosphorylation by GSK3 to mediate recognition (21). Like MULE, β-TrCP has other substrate proteins (including BIM, Emi-1, β-catenin, IκB-α, PERIOD2, etc.) (7, 19, 28, 53, 68, 70). Recently, the ubiquitin-specific peptidase 9 X-linked (USP9X) deubiquitinylase (DUB) was identified as a regulator of MCL-1, as RNA interference (RNAi)-mediated silencing of USP9X led to a loss of MCL-1 without affecting its mRNA expression (57). However, USP9X has other substrates, including ASK1, AMPK, Smad4, MARCH1, and Itch (3, 24, 44, 46, 47). Therefore, it is unclear what the relative contributions of the E3 ligases and DUB are in regulating MCL-1 expression, and the multitude of substrates makes it difficult to unravel how this dynamic network is regulated.
To investigate the role of ubiquitinylation in modulating murine MCL-1 protein levels, we generated an allele of murine Mcl-1 that lacks the sites for ubiquitinylation (Mcl-1KR) by mutating all 14 lysine residues to arginine. Surprisingly, MCL-1KR is turned over essentially identically to wild-type MCL-1 under basal and stressed conditions, suggesting the existence of an alternative degradation pathway. To support these findings, we utilized temperature-sensitive ubiquitin-activating enzyme E1 (UBE1)-expressing cells to abolish ubiquitinylation of endogenous MCL-1. In these cells, MCL-1 protein is degraded at a similar rate when the E1 was either active or inactive, consistent with MCL-1 protein being eliminated in a ubiquitinylation-independent manner. To assess the role of ubiquitinylation in regulating MCL-1 turnover in vivo, we generated transgenic mice that express epitope-tagged versions of wild-type MCL-1 or MCL-1KR. In lymphocytes derived from these mice, both versions of MCL-1 were eliminated at similar rates, demonstrating that ubiquitinylation is not essential for MCL-1 protein turnover. Lastly, in vitro-translated MCL-1 protein can be degraded in a cell-free system by the 20S proteasome in the absence of ubiquitinylation. Together, these data demonstrate that while MCL-1 can be the target of ubiquitinylation, this is not the only pathway through which its steady-state protein levels are regulated.
Simian virus 40 (SV40)-transformed wild-type and Mcl-1-deficient mouse embryonic fibroblasts (MEFs) have been previously described (51). HEK293T cells and mouse NIH 3T3 cells were obtained from the American Type Culture Collection (Manassas, VA). Cells were grown in Dulbecco's modified Eagle's medium (DMEM; Invitrogen, Carlsbad, CA) supplemented with 5% fetal bovine serum (FBS). Temperature-sensitive mouse ts20 Ubiquitin Activating Enyzme E1 (ts20TGR clone) cells, provided by H. Ozer (UMDNJ-New Jersey Medical School, Newark, NJ), were cultured at the 34°C permissive temperature and were shifted to 39°C to inactivate the E1 (12).
Amino-terminal Flag-Mcl-1 was generated by PCR amplification and cloning of the murine Mcl-1 cDNA into the p3XFlag-CMV10 vector (Sigma). The mutant murine Mcl-1KR, in which all 14 lysines were replaced by arginine, was generated by site-directed mutagenesis (Stratagene, La Jolla, CA). The PCR primers used to mutate all 14 lysines in Mcl-1 to arginine are available by request. All expression constructs and mutations were verified by sequencing. pCMV-HA-Ubiquitin (HA-Ubi) was a kind gift from D. Bohman (University of Rochester Medical Center, Rochester, NY). pBabe-puro and pBabe-puro-human UAE E1 were kindly provided by C. Sherr (St. Jude Children's Research Hospital, Memphis, TN). The pcDNA3-HA-MULE (Hect9) plasmid was obtained from K. Helin (University of Copenhagen, Copenhagen, Denmark) (1).
Ecotropic retroviruses were produced by cotransfection of one of the following retroviral expression plasmids by using FuGene 6 (Roche Applied Bioscience, Indianapolis, IN) in 293T cells with packaging plasmids (pMD-old-Gag-Pol and pCAG4-Eco): pMSCV-puro, pMSCV-puro-Mcl-1 (murine), pMSCV-puro-Mcl-1KR, pBabe-puro, or pBabe-puro-UAE E1.
MG-132, lactacystin, epoxomicin, proteasome inhibitor I, cycloheximide, and dexamethasone were purchased from Calbiochem (Gibbstown, NJ). Etoposide, staurosporine, puromycin, and N-ethylmaleimide (NEM) were purchased from Sigma. 35S-methionine was purchased from PerkinElmer (Waltham, MA).
For immunoblot analysis, cells were lysed with radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS) containing 1 mM NaF, 100 μM phenylmethylsulfonyl fluoride (PMSF), and 200 μM Na3VO4 supplemented with complete EDTA-free protease inhibitors (Roche) on ice for 30 min. Lysates were cleared by centrifugation, and protein concentrations were determined by bicinchoninic acid (BCA) protein assay.
For coimmunoprecipitation studies, wild-type MEFs, Mcl-1-deleted MEFs alone, or MEFs stably expressing tagless MCL-1 or MCL-1KR were lysed in flag lysis buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 1% Triton X-100, 1 mM EDTA) containing protease inhibitors on ice for 30 min. Input protein was precleared and then precipitated with one of the following antibodies: anti-MCL-1 rabbit polyclonal (Rockland Immunochemical, Gilbertsville, PA), both anti-BIM short and anti-BIM long rat monoclonal antibodies (Millipore, Billerica, MA), or anti-BAD (2G11 monoclonal; BD Pharmingen, San Diego, CA) precoupled to protein A/G plus agarose (Santa Cruz Biotechnology Inc., Santa Cruz, CA) at 4°C. After 18 h, immunocomplexes were recovered, washed, and resuspended in sample buffer.
Samples were heated to 95°C for 5 min and subjected to electrophoresis by using bis-Tris gels (Invitrogen), transferred to polyvinylidene difluoride (PVDF; Millipore) membranes, and developed using Western Lightning (PerkinElmer). The following antibodies were used to screen Western blots: anti-FLAG M2 mouse monoclonal (Sigma), anti-MCL-1 rabbit polyclonal (Rockland Immunochemical), anti-BAD (Santa Cruz Biotechnology), anti-BAK rabbit polyclonal (Millipore), anti-BIM 22-40 rabbit polyclonal (Millipore), anti-BCL-XL (BD Pharmingen), anti-p53 (IC12) mouse monoclonal (Cell Signaling Technology, Danvers, MA), anti-UAE E1 mouse monoclonal (Millipore), anti-UBA6 rabbit polyclonal (generous gift from J. W. Harper, Harvard Medical School, Boston, MA), and antiactin mouse monoclonal (Millipore) antibodies. Secondary antibodies were anti-rabbit or anti-mouse horseradish peroxidase-conjugated (Jackson Immunochemical, Bar Harbor, ME) antibodies. Protein bands were quantified using UN-SCAN-IT Gel and Graph Digitizing software (Silk Scientific, Orem, UT).
HEK293T cells were transfected with 2.5 μg of Flag-tagged or tagless Mcl-1 or Mcl-1KR and 2.5 μg of hemagglutinin (HA)-ubiquitin plasmid by using FuGene 6 (62). Mcl-1-deleted MEFs were cotransfected with 5 μg of the above-described DNA plasmids by using Lipofectamine 2000 (Invitrogen). Cells were treated at 48 h posttransfection with 10 μM MG-132 for 4 h and lysed in buffer A (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, 1% SDS) containing 1 mM sodium fluoride, 200 μM Na3VO4, and 10 mM NEM supplemented with protease inhibitors on ice for 30 min and were then sonicated and boiled for 5 min at 95°C. For immunoprecipitation of HA-ubiquitinylated MCL-1, 200 μg of lysate was diluted to 1 ml with buffer B (lacking SDS) for a final concentration of 0.1% SDS. Precleared samples were immunoprecipitated with 2 μg anti-HA (clone 3F10; Roche) precoupled to protein A/G plus agarose. After 18 h, immunocomplexes were washed five times in buffer B and eluted by boiling in buffer A containing 1× NuPage LDS buffer and 100 mM dithiothreitol (DTT) at 95°C for 5 min. Denatured samples were resolved by SDS-PAGE and immunoblotting.
Mcl-1-deficient MEFs stably expressing MCL-1 or MCL-1KR were washed with labeling medium (methionine- and cysteine-free DMEM supplemented with 5% dialyzed FBS, 48 mg/liter l-cysteine, and 2 mM l-glutamine) for 30 min to deplete intracellular stores of methionine. Labeling medium containing 0.2 mCi/ml 35S-methionine was then added for 2 h and then terminated by removal of medium and addition of labeling medium containing 1 mM nonradiolabeled methionine. Cells were harvested at various time points and lysed on ice in Triton X-100 lysis buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 1 mM NaF, 100 μM PMSF, and 200 μM Na3VO4) supplemented with complete EDTA-free protease inhibitors. Immunoprecipitation (from 200 μg total protein) was performed as described above, and complexes were resolved by SDS-PAGE, transferred to PVDF membranes, subjected to phosphorimagery on a Storm 860 scanner (GE Healthcare, Piscataway, NJ), and immunoblotted with anti-MCL-1 antibody for total protein immunoprecipitated.
In vitro ubiquitinylation experiments of MCL-1 were based on assays previously described (72). Briefly, 2.5 μl of in vitro-translated 35S-labeled MCL-1 or MCL-1KR (Promega TNT coupled transcription/translation systems) was incubated at 37°C for 3 h in a 15-μl reaction mixture with an ATP-regenerating system (50 mM Tris [pH 7.6], 5 mM MgCl2, 2 mM ATP, 10 mM creatine phosphate, 3.5 U/ml creatine kinase), 15 μg methyl-ubiquitin (Biomol, Plymouth Meeting, PA), 10 ng human E1 (Sigma), 100 ng Ubch7 (E2), 300 ng ubiquitin aldehyde (Biomol), and 20 μg HeLa S100 fraction (Biomol). Reactions were terminated in RIPA buffer containing 1× NuPage LDS buffer and 100 mM DTT and boiled at 95°C for 5 min, resolved on 4 to 12% gradient bis-Tris gels (Invitrogen), and then transferred to PVDF membranes. Dried membranes were subjected to autoradiography for 24 h at room temperature by using phosphorimagery.
Mcl-1-deficient MEFs expressing murine stem cell virus (MSCV)-puro vector, tagless MCL-1, or MCL-1KR were plated and were treated with etoposide or staurosporine (Calbiochem, Gibbstown, NJ) as indicated below. Cell viability was determined by staining with annexin V-fluorescein isothiocyanate (FITC) and propidium iodide (PI) (BD Biosciences) and flow cytometry (FACSCalibur; BD Biosciences, San Jose, CA). Each experiment was performed in triplicate, and error bars represent the standard error of the mean (SEM). Transgenic mouse thymocytes were cultured in cRPMI-10 in the presence of dexamethasone (Sigma) or dimethyl sulfoxide (DMSO) vehicle control for 4 h. Cell death was assessed by staining with annexin V and propidium iodide and flow cytometric analyses.
For RNAi, NIH 3T3 cells were plated in Opti-MEM media (Invitrogen) and transfected with respective 70 nM Stealth small interfering RNAs (siRNAs; Invitrogen) specific for mouse UBE1 and UBA6 by using Lipofectamine RNAi Max (Invitrogen). Six hours later, transfection medium was replaced with complete DMEM. Seventy-two hours later, transfected cells were collected, lysed, and immunoblotted. Stealth siRNA sequences were as follows: siUbe1, 5′-GACCTATGGGTTGACTGGATCCCAA-3′; siUba6, GGACCCATATTTACATGGAGGCATA-3′.
Murine full-length cDNAs for amino-terminal Flag-tagged Mcl-1 or Mcl-1KR were cloned into the pBROAD3-mcs vector (InvivoGen, San Diego, CA) containing the mouse ROSA promoter, a 5′ untranslated region (UTR) that contains an engineered intron to increase transcription, and the human β-globin 3′ UTR and polyadenylation sequence. To remove the contaminating backbone vector sequence, the vectors were digested with PacI and gel purified for pronuclear injection of fertilized eggs from FVB mice by the St. Jude Transgenic/Gene Knock-Out Shared Resource. Positive founders were backcrossed to C57BL/6 mice for four generations prior to analysis of transgene expression.
MCL-1, p21, BCL-2, and BCL-XL were synthesized and labeled with 35S-methionine by using the TNT coupled reticulocyte lysate system (Promega) by following the manufacturer's protocol. J. Roberts (Fred Hutchinson Cancer Research Center, Seattle, WA) kindly provided the p21 construct. D. Green (St. Jude Children's Research Hospital, Memphis, TN) kindly provided the BCL-XL expression construct. Briefly, each in vitro-translated protein (0.1 μl/time point) was incubated with 20S proteasomes (1 μg/time point; BioMol, Plymouth Meeting, PA) in 20S assay buffer (20 mM Tris [pH 7.2], 1 mM EDTA, and 1 mM DTT) or with 26S proteasomes (1 μg/time point, BioMol) in 26S assay buffer (20 mM Tris [pH 7.2], 25 mM KCl, 10 mM NaCl, 1 mM MgCl2, 1 mM DTT, 4 mM ATP, 50 mM phosphocreatine, and 17.5 U/ml creatine phosphokinase), respectively, at 37°C for the times indicated (10). At each time point, the reaction was stopped by addition of 4× SDS loading buffer. Samples were resolved by SDS-PAGE, transferred to PVDF membrane, and subjected to phosphorimagery for analysis.
To investigate whether proteasome inhibition affects the levels of endogenous murine MCL-1, MEFs were treated with proteasome inhibitors. By immunoblot analysis, murine MCL-1 appears as a triplet of approximately 36-kDa, 38-kDa, and 40-kDa species (Fig. (Fig.11 A). While the origin of these three MCL-1 species is unclear, several models have been proposed to explain their origin. First, MCL-1 can be phosphorylated, and these modifications may result in differential migration on SDS-PAGE (22, 23, 41, 43). However, when lysates from wild-type MEFs were treated with lambda phosphatase to remove phosphorylation, the migration pattern was unchanged despite efficient dephosphorylation of AKT, indicating that the different MCL-1 species are not the result of differential phosphorylation (data not shown). Second, the lower form of MCL-1 has been proposed to arise from a noncanonical translational initiation downstream of the start codon (66). However, ablation of MCL-1's start codon results in a protein much smaller than the 36-kDa species, appearing to arise from a downstream methionine (data not shown). Third, noncanonical mRNA splicing of a cryptic intron in MCL-1's first exon has been proposed to give rise to the lower species of MCL-1 (33). However, mutagenesis of the putative noncanonical splice donor/acceptor pair does not affect the generation of all three species of MCL-1 (data not shown). Our data are consistent with the observation that human MCL-1 undergoes amino-terminal truncation resulting in multiple protein species (18).
Treatment of wild-type MEFs with MG-132, lactacystin, epoxomicin, and proteasome inhibitor-1 (PI-1) did not alter the level of the MCL-1 40-kDa band but significantly resulted in an increase in the levels of the 38-kDa and 36-kDa bands compared to those of untreated samples (Fig. (Fig.1A).1A). To ascertain whether proteasome inhibitors could antagonize the elimination of MCL-1, we treated MEFs with cycloheximide, an inhibitor of new protein synthesis. After 4 h of incubation with cycloheximide, the 40-kDa and 38-kDa bands of endogenous MCL-1 were eliminated (decreased by 90% compared to the untreated sample) (Fig. (Fig.1A).1A). In contrast, cotreatment with MG-132, lactacystin, epoxomicin, or PI-1 and cycloheximide slowed the degradation of the 40-kDa and 38-kDa bands of MCL-1. Interestingly, the 36-kDa band was retained in all samples cotreated with cycloheximide, suggesting that the 36-kDa form of murine MCL-1 is more stable. Similar increased stability of the smallest species of human MCL-1 has been reported (18). These data indicate that MCL-1 turnover can be antagonized by proteasome inhibition.
To address the role of ubiquitinylation in modulating MCL-1 protein levels, we generated an allele of murine Mcl-1 that is incapable of ubiquitinylation by mutating all 14 lysine residues and conservatively replacing them with arginine (MCL-1KR). To detect ubiquitinylated murine MCL-1, we coexpressed Flag-tagged MCL-1 (Flag-MCL-1) in HEK293T cells or tagless MCL-1 (MCL-1) in Mcl-1-deficient MEFs along with HA-tagged ubiquitin (HA-Ubi). After anti-HA immunoprecipitation, ubiquitin-conjugated proteins were identified by immunoblotting. In cells coexpressing tagged or tagless MCL-1 with HA-Ubi, ubiquitinylated MCL-1 was readily observed (Fig. 1B and C). In contrast, no ubiquitinylation was observed when either the Flag-tagged or tagless MCL-1KR was coexpressed with HA-Ubi (Fig. 1B and C). These data demonstrate that the MCL-1KR protein cannot be ubiquitinylated. Similar data were generated when cells were cotransfected with Mcl-1 constructs and histidine-tagged ubiquitin (His-Ubi) (data not shown). Furthermore, recombinant wild-type MCL-1, but not MCL-1KR, was capable of being ubiquitinylated in vitro (Fig. (Fig.1D).1D). Amino-terminal ubiquitinylation on nonlysine residues can also promote protein degradation (13). However, there is no evidence of amino-terminal ubiquitinylation of tagless MCL-1KR, suggesting that this does not occur for MCL-1 (Fig. (Fig.1B).1B). Ubiquitinylation can also occur on serine and threonine residues by a thiohydroxyl bond, but we do not see any evidence of such modifications of MCL-1KR (Fig. 1B and C) (8, 64). Cysteine residues can also be the targets of ubiquitinylation by a thioester bond, but when tagless MCL-1 or MCL-1KR was stably expressed in Mcl-1-deficient MEFs along with HA-Ubi and the anti-HA immunoprecipitated proteins were eluted without reducing agent (DTT), no ubiquitinylation of the MCL-1KR protein was detected (data not shown) (59, 64). Thus, mutating all 14 lysine residues in MCL-1 to arginine generated a protein that cannot be ubiquitinylated.
We sought to identify the specific lysine residues in murine Mcl-1 used for ubiquitinylation by single-site mutagenesis; however, we found that the sites of murine MCL-1 ubiquitinylation are quite promiscuous, as the presence of even a single lysine residue at any of the 14 positions led to detectable ubiquitinylation (data not shown). Thus, only the Mcl-1KR was incapable of ubiquitinylation, allowing us to address the role of ubiquitinylation in regulating MCL-1 degradation.
To ensure that the Mcl-1KR allele was capable of normal antiapoptotic function, it was stably expressed in Mcl-1-deleted MEFs by using vector, wild-type MCL-1, or MCL-1KR encoding retroviruses. Both the wild-type and Mcl-1KR alleles were expressed at similar levels in Mcl-1-deficient MEFs, and the levels of protein expression were comparable to endogenous MCL-1 expression in wild-type MEF cells (Fig. (Fig.22 A). While Mcl-1-deleted MEFs expressing vector were sensitive to apoptosis, expression of either wild-type MCL-1 or MCL-1KR similarly protected the cells against both etoposide- and staurosporine-induced cell death (Fig. 2B and C).
MCL-1 binds selectively to a subset of proapoptotic molecules, including BIM and BAK, but does not bind the BAD BH3-only molecule (51, 67). Therefore, we tested whether wild-type MCL-1 or MCL-1KR coimmunoprecipitated with BIM or BAK. As expected, wild-type MCL-1 interacted with both proapoptotic BIM and BAK but was incapable of coimmunoprecipitation with BAD (Fig. (Fig.2D).2D). Likewise, MCL-1KR was also able to bind both BAK and BIM, indicating that its ability to interact with proapoptotic molecules is not impaired by the lysine mutations (Fig. (Fig.2D).2D). Furthermore, like wild-type MCL-1, MCL-1KR did not interact with BAD, demonstrating that mutation of the lysine residues to arginine does not alter its specificity for proapoptotic binding partners. These data demonstrate that MCL-1KR is capable of preventing cell death and can bind proapoptotic BCL-2 family members, similar to wild-type MCL-1.
Previously, MCL-1 degradation was shown to occur constitutively in HeLa cells irrespective of cellular stress (48). Therefore, to compare the half-lives of wild-type MCL-1 and MCL-1KR, we monitored the decay of MCL-1 after inhibition of new protein synthesis. Unexpectedly, after cycloheximide treatment, both the 40-kDa and 38-kDa bands of MCL-1 decayed rapidly (half-lives, ~2 h and 1.5 h, respectively) in cells expressing either MCL-1 or MCL-1KR (Fig. (Fig.33 A). The 36-kDa band for wild-type MCL-1 had an extended half-life, decreasing only to 60% by the last time point tested (6 h). Interestingly, the 36-kDa band for MCL-1KR exhibited a short half-life (approximately 2 h), similar to the 40-kDa band, but no bands were present after 4 h. These data imply that modification of the lysine residues may affect the response of the individual MCL-1 protein species to stimuli; ubiquitinylation may affect amino-terminal processing or cellular signaling. However, even though MCL-1KR is capable of mediating protection from death stimuli, the nonubiquitinylatable MCL-1 exhibits a half-life similar to that of the wild-type MCL-1 after cycloheximide treatment.
To determine whether an amino-terminal epitope tag alters the half-life of MCL-1, Flag-MCL-1 and Flag-MCL-1KR were stably expressed in Mcl-1-deficient MEFs and treated with cycloheximide. Both wild-type and MCL-1KR fusion proteins exhibited rates of decay similar to those of the tagless wild type and MCL-1KR after treatment with cycloheximide (Fig. 3A and B). Therefore, the addition of an N-terminal epitope tag does not dramatically alter MCL-1's rate of degradation.
If MCL-1 is degraded by the proteasome, inhibition of the proteasome should stabilize and increase protein levels. When cells were treated with MG-132, the 40-kDa band for both wild-type MCL-1 and MCL-1KR did not change (Fig. (Fig.3C).3C). Interestingly, the 38-kDa band increased with MG-132 treatment in both cell lines. In contrast, the 36-kDa band for wild-type MCL-1 maintained expression, whereas the 36-kDa band for MCL-1KR decreased with MG-132 treatment. Overall, MG-132 treatment increased the levels of the 38-kDa band for both MCL-1 and MCL-1KR and decreased the 36-kDa band for MCL-1KR.
We then compared the stability of wild-type MCL-1 to that of MCL-1KR when protein synthesis and proteasome activity were inhibited. Proteasome inhibition prevented the loss of both the wild type and MCL-1KR (Fig. (Fig.3D).3D). As observed with MG-132 treatment alone, culturing the cells with cycloheximide and MG-132 increased the levels of the 38-kDa band for both wild-type MCL-1 and MCL-1KR (Fig. (Fig.3D).3D). In contrast, the 36-kDa bands of both wild-type MCL-1 and MCL-1KR were similarly extended to 4 h. However, after 6 h, the 36-kDa band for MCL-1KR decreased. Therefore, inhibition of the proteasome decreases the rate of MCL-1 turnover of the 40-kDa band by 2 h, but results in an increase in the 38-kDa band. These data demonstrate that basal turnover of MCL-1 is independent of ubiquitinylation but can be blocked by inhibition of the proteasome.
MCL-1 undergoes rapid degradation in response to a variety of cellular stresses (21, 41, 48, 57). To determine whether MCL-1 turnover under conditions of cellular stress is dependent on ubiquitinylation, Mcl-1-deficient MEFs stably expressing wild-type and MCL-1KR were treated with UV irradiation, and the expression of MCL-1 protein was determined. As has been previously reported, treatment of wild-type MCL-1-expressing cells with UV irradiation led to the rapid elimination of MCL-1 protein, with the 40-kDa band being lost prior to the 36-kDa species (Fig. (Fig.3E)3E) (48, 57). Similarly, the MCL-1KR protein also undergoes rapid elimination in response to UV irradiation, but in this case, all three protein species are eliminated simultaneously (Fig. (Fig.3E).3E). These data are consistent with the ubiquitinylation-independent elimination of MCL-1 protein in response to cellular stress.
Both MULE and β-TrCP E3-ligases have been demonstrated to ubiquitinylate MCL-1, but both ligases have a variety of other target proteins. To determine whether these E3 ligases may affect MCL-1 stability by modulating MCL-1 ubiquitinylation, we sought to test whether modulating the E3 ligase levels affected the stability of wild-type MCL-1 compared to lysineless MCL-1KR. Unfortunately, we were unable to achieve efficient silencing of either MULE or β-TrCP in either MEF or 293T cells. Therefore, we tested whether ectopic expression of MULE or β-TrCP could enhance the degradation of wild-type or lysineless MCL-1 in transient cotransfection experiments. When wild-type Mcl-1 or Mcl-1KR were coexpressed with MULE and subjected to cycloheximide or UV treatment, we could detect enhanced elimination of wild-type MCL-1 but not MCL-1KR (Fig. (Fig.3F).3F). These data indicate that MULE can affect MCL-1 elimination only when lysine residues are present. Interestingly, MULE overexpression actually seemed to prolong the half-life of MCL-1KR, perhaps by its BH3-only domain binding to MCL-1's hydrophobic binding pocket. In contrast, we did not see any alterations in MCL-1 half-life under cycloheximide or UV treatment conditions with overexpression of β-TrCP (data not shown).
The function of β-TrCP has been shown to be modulated by glycogen synthase kinase 3β (GSK3β)-mediated phosphorylation of MCL-1 (21). Therefore, we tested whether blocking or facilitating GSK3β activity would alter wild-type MCL-1 or MCL-1KR elimination. Expression of a constitutively active form of AKT (myristoylated-AKT) has been shown to inhibit GSK3β activity; therefore, we tested whether it could modulate MCL-1 degradation (41). Coexpression of constitutively active AKT was capable of blocking the elimination of both wild-type and MCL-1KR protein in response to both cycloheximide and UV irradiation (Fig. (Fig.3G).3G). These data indicate that besides potentially regulating MCL-1 ubiquitinylation by β-TrCP, AKT activity can modulate MCL-1 turnover in a ubiquitin-independent manner.
One caveat of using cycloheximide treatment to measure protein degradation is that inhibition of protein synthesis may alter the levels of other proteins involved in modulating the half-life of MCL-1. Therefore, we used pulse-chase measurement to directly measure the basal turnover of MCL-1 protein. Mcl-1-deficient MEFs expressing either wild-type MCL-1 or MCL-1KR were labeled in medium containing 35S-methionine and then chased with unlabeled medium. MCL-1 was immunoprecipitated from the MEFs and subjected to autoradiography using a phosphorimager to determine the amount of 35S-labeled protein immunoprecipitated at each time point (Fig. (Fig.4A4A and andB).B). Under basal conditions, the 40-kDa bands for both 35S-labeled wild-type MCL-1 (Fig. (Fig.4A)4A) and MCL-1KR (Fig. (Fig.4B)4B) had similar half-lives of approximately 2 h (Fig. (Fig.4C).4C). In the presence of MG-132, the 40-kDa bands of both labeled wild-type MCL-1 and MCL-1KR were maintained over a 4-h experiment (Fig. (Fig.4D4D and data not shown). These data confirm those generated using cycloheximide in that both wild-type MCL-1 and MCL-1KR undergo degradation at essentially identical rates. Thus, despite lacking lysine residues, the turnover of MCL-1KR occurs at a rate similar to that of wild-type MCL-1, further demonstrating that ubiquitinylation is dispensable for MCL-1 degradation.
Although the MCL-1KR mutant was capable of preventing cell death and antagonizing BH3-only proteins, the mutagenesis of the lysine residues might have some unexpected effect on MCL-1 degradation or affect other potentially positive regulatory modifications. Therefore, we utilized murine ts20 cells (derived from wild-type BALB/c 3T3 clone A31) that express a temperature-sensitive E1 ubiquitin-activating enzyme (12). At the permissive temperature (34°C), E1 ubiquitin-activating enzyme is functional in ts20 cells, but at the nonpermissive temperature (39°C), the E1 is nonfunctional, leading to an increase in proteins that are normally degraded in a ubiquitin-dependent manner (12).
When ts20 cells were shifted to the nonpermissive temperature, expression of short-half-life proteins, such as p53, increased rapidly (Fig. (Fig.5A).5A). However, no change in MCL-1 protein expression was observed under these conditions (Fig. (Fig.5A).5A). To determine whether these effects were due to a loss of functional E1, we retrovirally expressed functional human E1 ubiquitin-activating enzyme in the ts20 cells (Fig. (Fig.5B).5B). Expression of human UAE E1 restored degradation of p53 even after shifting to the nonpermissive temperature; however, ectopic UAE E1 expression did not change MCL-1 levels (Fig. (Fig.5C).5C). As a control for heat shock, no changes in MCL-1 or p53 were observed in NIH 3T3 cells when shifted to 39°C (Fig. 5A and C). Shifting the ts20 cells to 39°C induces a robust increase in p53 protein levels and reduces the presence of ubiquitinylated MCL-1 species by 80%, suggesting that ubiquitinylation is severely impeded by inactivating E1 (data not shown). Similar results were obtained using the hamster cell line tsBN75, which also harbors a thermolabile E1 ubiquitin-activating enzyme (data not shown) (49).
To directly test the importance of ubiquitinylation in regulating degradation of MCL-1, the half-life of MCL-1 in ts20 cells incubated for 2 h at the permissive or nonpermissive temperature was measured. The inactivation of ubiquitin-activating enzyme E1 did not alter the kinetics of degradation of the endogenous MCL-1 protein (Fig. (Fig.5D).5D). The half-life of MCL-1 is ~2 h in ts20 cells at 34°C and remained the same when the temperature was shifted to the nonpermissive temperature. In contrast, the half-life of p53 increased to more than 6 h when the temperature was shifted to 39°C. Together, these data show that inactivation of the E1 ubiquitin-activating enzyme does not result in the stabilization of endogenous MCL-1.
An additional E1 enzyme, UBA6 (also known as UBE1L2 and E1-L2), is expressed in a variety of cell types, including ts20 cells (11, 32, 52). UBA6 has been shown to charge a variety of E2-conjugating enzymes with ubiquitin, but to date, no UBA6-dependent specific substrate proteins have been identified. To discern if UBA6 may regulate MCL-1 protein turnover, we utilized an RNAi strategy. When both UAE1 and UBA6 were knocked down in NIH 3T3 cells by RNAi, no alterations in MCL-1 protein levels were observed (Fig. (Fig.5E).5E). These data demonstrate that MCL-1 degradation can occur in the absence of either functional E1 enzyme and indicate that the elimination of MCL-1 can occur in an ubiquitin-activating enzyme E1-independent manner.
To test the in vivo consequences of ectopically expressing the nonubiquitinylatable allele of Mcl-1, transgenic mice were generated expressing Flag-tagged wild-type Mcl-1 or Mcl-1KR under the control of the ROSA26 promoter. These mice express amino-terminal Flag-tagged MCL-1 or MCL-1KR in a variety of cell lineages, including hematopoietic tissues; the Flag tag allows it to be differentiated from the endogenous murine MCL-1 (Fig. (Fig.6B6B and andC).C). Transgenic mice expressing either Flag-tagged wild-type MCL-1 or MCL-1KR did not exhibit any obvious developmental abnormalities up to 5 months after weaning, indicating that the ectopic expression of either construct does not overtly perturb hematopoietic development and homeostasis (data not shown). However, the ectopically expressed MCL-1 rendered thymocytes from transgenic mice somewhat resistant to dexamethasone-induced apoptosis, indicating that both Flag-tagged MCL-1 constructs were functional (Fig. (Fig.6A6A).
To test whether ubiquitinylation regulates the half-life of MCL-1 in primary cells, the degradation of Flag-MCL-1 and Flag-MCL-1KR was assessed in isolated thymocytes or T lymphocytes cultured with cycloheximide (Fig. 6B and C, respectively). The levels of MCL-1 expression were determined by immunoblot analyses. In both thymocytes and cultured T lymphocytes, ectopically expressed Flag-MCL-1 and Flag-MCL-1KR proteins were eliminated at similar rates in response to cycloheximide treatment (Fig. 6B and C). Therefore, in primary lymphoid cells, the Flag-MCL-1KR protein undergoes degradation at a rate similar to that of wild-type Flag-MCL-1, despite lacking the sites for ubiquitinylation. These data from ex vivo primary cells confirm our findings in cell lines that the degradation of MCL-1 does not require ubiquitinylation.
As the degradation of both wild-type and MCL-1KR proteins can be blocked by proteasome inhibitors, we hypothesized that MCL-1 may be directly targeted to the proteasome in a ubiquitin-independent manner, as has been reported for a growing number of proteins (31). Such ubiquitin-independent proteasome degradation can utilize the classic 26S proteasome (e.g., ODC), the 20S proteasome (e.g., aged or oxidized proteins and p53 in some cases), or other proteasome variants (e.g., the PA28γ proteasome for p21) (10, 36, 45, 55, 60, 63). Thus, it appears that an expanding number of proteins can undergo degradation in a ubiquitin-independent proteasome-dependent manner (6). To test whether MCL-1 was capable of direct recognition by the proteasome in the absence of ubiquitinylation, in vitro-translated proteins were incubated with 20S proteasomes. p21 protein, which has been shown to be degraded in vitro in a ubiquitin-independent manner by the 20S proteasome, was used as a positive control (10, 36). Similar to previously published reports, p21 was efficiently degraded by purified 20S proteasomes, but BCL-2 and BCL-XL proteins were stable (Fig. (Fig.7A7A and andB).B). Wild-type, unmodified MCL-1 was rapidly degraded when incubated with the 20S proteasomes, demonstrating its ability to be degraded without prior ubiquitinylation (Fig. 7A and B). The degradation was dependent on proteasome function, as MG-132 addition impaired the degradation of both p21 and MCL-1 (Fig. (Fig.7A).7A). As a variety of proteins can be degraded by the 26S proteasome in a ubiquitin-independent manner, we tested whether purified 26S proteasomes could degrade in vitro-translated MCL-1 protein. Wild-type, unmodified MCL-1 was poorly degraded (less than 10% degraded after 4 h of incubation) by the 26S proteasome, suggesting that the 26S proteasome does not foster MCL-1's degradation as efficiently as does the 20S proteasome (data not shown). These data indicate that MCL-1 can be degraded in vitro without ubiquitinylation by the 20S proteasome.
The tight regulation of MCL-1 protein expression makes it an ideal regulator of cell survival (48). In response to cellular signaling, MCL-1 protein levels can be rapidly induced by inducing new MCL-1 transcription and by preventing MCL-1 protein turnover. When cells need to be eliminated, MCL-1 levels can be rapidly diminished by blocking new protein synthesis and degrading the existing MCL-1. Dysregulation of this balance, by inappropriately promoting its synthesis or by blocking its elimination, can lead to inappropriate stabilization of MCL-1 and promote cellular survival (72). Furthermore, dysregulated MCL-1 levels can lead to inappropriate cell survival or death; therefore, understanding the regulation of MCL-1 levels is of great importance. Our studies have revealed that MCL-1 degradation may be more complicated than previously anticipated; in the absence of ubiquitinylation, it undergoes normal degradation under basal and stressed conditions.
There is accumulating evidence that proteasome degradation does not systemically depend on prior ubiquitinylation of substrates. Indeed, an expanding number of proteins have been identified to be degraded by the proteasome independent of ubiquitinylation (6, 31). In some cases, proteins (e.g., p21, p53, etc.) can undergo both ubiquitin-independent and ubiquitin-dependent degradation, supporting the growing supposition that proteins may use multiple routes to degradation (31). The relative contributions of these degradative pathways may vary according to cellular contexts. How nonubiquitinylated proteins are recognized and gain access to the proteasome is a topic of fervent research. In the case of p21, it appears that the PA28γ regulator can promote the efficient recognition of nonubiquitinylated p21 (10, 36). However, under in vitro conditions, it appears that nonubiquitinylated p21 can also be directly recognized by the 20S proteasome and potentially by the 26S proteasome (37, 61). How do nonubiquitinylated proteins gain access to the catalytic chamber of the proteasome, which is gated by the flexible N termini of α-subunits? It has been proposed that intrinsically unstructured proteins can themselves promote the opening of the α-ring channel (37).
The interaction of unstructured proteins with binding partners has been hypothesized to mask destabilizing regions or induce folding that renders the protein stabilized and therefore unable to access the proteasome (5, 35). Structurally, MCL-1 can be subdivided into two parts: the carboxyl-terminal 150-amino-acid region that shares homology with other antiapoptotic family members, and the 170-amino-acid amino-terminal region which is unstructured and has been omitted from all structural studies (17, 18). Therefore, the lack of structure in the amino-terminal region of MCL-1 promotes its recognition by the proteasome, as has been suggested for other intrinsically unstructured proteins (5, 38). In support of this model, interactions between MCL-1 and other proteins have been implicated in regulation of its stability. A number of proteins, including proapoptotic BH3-only family members and translationally controlled tumor protein (TCTP), have been reported to modulate MCL-1 levels by regulating its degradation (15, 39, 42, 67). For example, expression of Noxa has been reported to promote MCL-1 degradation (67). In contrast, expression of both proapoptotic BIM and Puma has been reported to promote MCL-1 stabilization (15, 42). The basis for how interactions with BH3-only molecules may modulate MCL-1 stability is still somewhat unclear. Evidence suggests that the binding of BH3-only family members does not induce overt structural changes in the BCL-2-like regions of MCL-1, which are homologous to other antiapoptotic BCL-2 family members (15). However, this study did not investigate whether proapoptotic binding has any effects on the unstructured amino-terminal region of MCL-1. Therefore, the interactions of MCL-1 with binding partners may affect its protein half-life by modulating the structure of its amino terminus. Alternatively, these data are consistent with the different species of MCL-1 resulting from amino-terminal truncation; it is possible that the differences in half-life may represent the loss of unstructured amino-terminal regions on MCL-1. This might explain why the different MCL-1 species are eliminated at different rates.
In HeLa cells, MCL-1 was reported to undergo constitutive protein turnover, and the five amino-terminal lysine residues in human MCL-1 were required for ubiquitinylation; mutagenesis of these residues extended the half-life of MCL-1 (48, 72). Our data from mouse cells and primary tissues stand in contrast to these observations. We observe normal MCL-1 elimination even in the absence of ubiquitinylation. It is possible that human and murine MCL-1 may be regulated differently with respect to the need for ubiquitinylation to facilitate proteasome-mediated degradation. Alternatively, it is possible that the regulation of MCL-1 turnover in HeLa cells is more dependent on ubiquitin-mediated degradation than it is in other cell lines and primary cells. Additionally, the half-life of MCL-1 can be modulated by silencing of the MULE or β-TrCP E3 ligases or USP9X DUB (21, 57, 72). While it appears that these enzymes may directly affect MCL-1 ubiquitinylation and subsequent degradation, they have also been implicated in modifying a host of other cellular proteins (1, 3, 7, 9, 19, 24, 26, 28, 40, 44, 46, 47, 53, 68, 70). Therefore, it is possible that some of their effects may be indirect and through modulation of other MCL-1-interacting proteins.
As has been reported by others, we observe that wild-type MCL-1 undergoes ubiquitinylation. Thus, it appears that MCL-1 molecules represent a heterogeneous protein population which undergoes multiple levels of regulation to ensure that MCL-1 levels are appropriately regulated. Therefore, it is possible that the bulk of MCL-1 proteins is normally degraded in a ubiquitin-dependent manner, while only a minor fraction would be degraded in a ubiquitin-independent manner. However, under conditions in which MCL-1's ubiquitinylation is blocked, our data indicate that turnover would still occur normally by a redundant, alternative ubiquitin-independent mechanism. Further research will be necessary to determine the contribution of this alternative degradation system to maintaining MCL-1 levels in normal and disease biology.
We thank S. Oakes, L. Nutt, D. Green, J. Ihle, G. Zambetti, J. Partridge, P. Brindle, and P. Ney for helpful discussions; the St. Jude Transgenic Core Facility for generating the transgenic mice; and J. Roberts, H. Ozer, D. Bohman, K. Helin, M. Pagano, C. Sherr, D. Green, and J. W. Harper for providing reagents.
This work is supported by the American Lebanese Syrian Associated Charities (ALSAC), and the NIH NCI P30CA021765-30. J.T.O. is a Pew Scholar in the Biomedical Sciences.
Published ahead of print on 12 April 2010.