BCL2 Is a Downstream Effector of MIZ-1 Essential for Blocking c-MYC-induced Apoptosis*

The c-MYC oncoprotein is among the most potent transforming agents in human cells. Ironically, c-MYC is also capable of inducing massive apoptosis under certain conditions. A clear understanding of the distinct pathways activated by c-MYC during apoptosis induction and transformation is crucial to the design of therapeutic strategies aimed at selectively reactivating the apoptotic potential of c-MYC in cancer cells. We recently demonstrated that apoptosis induction in primary human cells strictly requires that c-MYC bind and inactivate the transcription factor MIZ-1. This presumably blocked the ability of MIZ-1 to activate the transcription of an unidentified pro-survival gene. Here we report that MIZ-1 activates the transcription of BCL2. More importantly, inhibition of the MIZ-1/BCL2 signal is an essential event during the apoptotic response. Furthermore, targeting BCL2 with short hairpin RNA or small molecule inhibitors restores the apoptotic potential of a c-MYC mutant that is defective for MIZ-1 inhibition. These observations suggest that repression of BCL2 transcription is the single essential consequence of targeting the MIZ-1 pathway during apoptosis induction. These data define a genetic pathway that helps to explain historical observations documenting cooperation between c-MYC and BCL2 overexpression in human cancer.

Elevated levels of the c-MYC oncoprotein contribute to the initiation and progression of most human tumors (1,2). Increased expression of c-MYC induces proliferation and inhibits differentiation (3)(4)(5)(6)(7)(8). Functioning as a basic helixloop-helix transcription factor, c-MYC can either activate or repress the expression of specific target genes associated with various biological functions (9,10). Via this transcriptional regulatory activity, c-MYC contributes to diverse aspects of cancer biology, including cell cycle progression (7), angiogenesis (11), metastasis (12), cell adhesion (13), cell growth (14), and genomic instability (15). Under some conditions, c-MYC expression also induces robust apoptosis (16). The induction of apoptosis by c-MYC is thought to provide a safeguard against cancer by eliminating cells that accumulate high levels of this oncoprotein. Oncoproteins such as E1A (17,18) and E2F-1 (19,20) share this capacity for inducing apoptosis. For tumorigenesis to occur, transformed cells must disable the apoptosis pathway normally activated by oncoproteins like c-MYC.
The BCL2 family of proteins represents key regulators of apoptosis (21,22). This family includes both pro-survival and pro-apoptotic members, and the balance of these two types of proteins within the cell determines whether apoptosis will occur. The pro-survival BCL2 family members BCL2, BCLxL, and MCL1 block apoptosis by binding and inactivating pro-apoptotic BCL2 family members (21,23). The proapoptotic proteins include BIM, NOXA, BAX, BAK, and PUMA (23)(24)(25)(26). Although the mechanism by which the distinct pro-apoptotic BCL2 family members trigger cell death is not completely understood, stress-induced oligomerization of BAX and BAK does result in permeabilization of the outer mitochondrial membrane (21). This permeabilization liberates factors, including SMAC, HTRA2, apoptosis-inducing factor, and cytochrome c, that ultimately trigger cellular destruction via caspase activation (27)(28)(29).
Apoptosis induction depends on the ability of c-MYC to regulate multiple factors involved in cell death. For example, by activation of the tumor suppressor p19 ARF, c-MYC promotes p53 stabilization (30). Studies from E-Myc transgenic mice suggest that inactivation of this pathway is the primary mechanism by which the apoptotic potential of c-MYC is debilitated during tumorigenesis. c-MYC can also induce apoptosis independent of p53 stabilization in fibroblasts, primary myeloid cells, and pre-B cells (31,32). An additional mechanism by which c-MYC promotes apoptosis independent of p53 is via increasing steady-state levels of the pro-apoptosis BIM protein (25). Furthermore, transcription of both the BCL2 and BCLxL pro-survival factors is repressed by c-MYC, thus promoting apoptosis (31,33,34). In support of these and other previous studies (35), we recently demonstrated that transcriptional repression by c-MYC contributes to apoptosis induction in human cells (36). Mechanistically, we showed that this repression is mediated by the ability of c-MYC to bind and inactivate the transcription factor MIZ-1 (36). Remarkably, inhibition of MIZ-1 is specifically required during c-MYC-induced apoptosis but not c-MYC-induced transformation or cell cycle progression (36). These data led to the hypothesis that inactivation of MIZ-1 by c-MYC is critical during apoptosis induction because MIZ-1 normally activates the transcription of an unknown, pro-survival gene. Establishing the identity of this pro-survival gene should provide a critical advance in our understanding of this previously unknown pathway. A large number of findings from both human tumors and mouse models have established that expression of the pro-survival BCL2 protein cooperates with c-MYC during tumorigenesis (22,(37)(38)(39)(40)(41). Here we link these two previously unrelated sets of observations by demonstrating that MIZ-1 normally activates the transcription of the BCL2 gene, and BCL2 activation blocks c-MYC-induced apoptosis. In contrast to BCL2, the repression of previously described targets of MIZ-1 (p15 INK4B , p21 CIP1 , NRAMP1, and MAD4) does not contribute to c-MYC-mediated apoptosis. More importantly, using a combination of shRNA 3 and a BCL2 small molecule inhibitor, we show that repression of BCL2 transcription by c-MYC is the essential event that is accomplished by targeting the MIZ-1 pathway. These data establish a novel pathway linking c-MYC, MIZ-1, BCL2, and ultimately apoptosis.
Cell Culture and Transient Transfections-Early passage fibroblasts (2091 and IMR90) and epithelial cells (293T) were obtained from the ATCC and propagated in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (Tissue Culture Biologicals), 2 M glutamine (Invitrogen), 100 g/ml penicillin (Invitrogen), and 100 g/ml streptomycin sulfate (Invitrogen). Transient transfections were performed using Lipofectamine 2000 (Invitrogen) and Opti-MEM I-reduced serum media (Invitrogen) following the manufacturer's protocol. ABT-737 was diluted in Me 2 SO to a working concentration of 5 mM. Following further dilution in culture media, cells were treated at a concentration of 7.5 M ABT-737.
Retroviral and Lentiviral Infections-Retrovirus was produced by cotransfection of retroviral vectors and SV-A-MLV helper plasmid (43) into the 293T packaging cell line. Viruscontaining supernatant was harvested, and three sequential viral infections were performed on target cells 24 -48 h posttransfection in the presence of 8 g/ml hexadimethrine bromide. Lentivirus was packaged in 293T cells transfected with the indicated pLKO.1-puro shRNA plasmid (Sigma) in addition to the pVSV-G and pCMVdelta8.2 helper plasmids (44).
Electroporation-The IMR90 cell strain was electroporated with a total of 1.5 g of DNA according to manufacturer's protocol (Amaxa Nucleofection).
Apoptosis Assays-Subsequent to retrovirus and lentivirus transduction, primary HDFs were seeded at 1.2 ϫ 10 6 cells/ 10-cm dish. The following day, cells were washed thoroughly with phosphate-buffered saline and maintained in Dulbecco's modified Eagle's medium supplemented with 0.2% fetal bovine serum for 24 h. Cells were stained with annexin V-fluorescein isothiocyanate and propidium iodide (Pharmingen).
Immunoblotting and Immunoprecipitation Analysis-Whole cell lysates were prepared in a Nonidet P-40-based buffer (45), and Western blots were performed as described previously (45). The BIM (StressGen), c-MYC (Santa Cruz Biotechnology), p53 (Santa Cruz Biotechnology), ␤-tubulin (Sigma), and BCL2 (Pharmingen) antibodies used for Western blotting were obtained from commercial sources. Endogenous MIZ-1 was detected by immunoprecipitation and subsequent Western blot using two MIZ-1 antibodies (Santa Cruz Biotechnology and M. Eilers, respectively). Immunoblots were quantified using Kodak one-dimensional 3.5.3 imaging software.
Real Time Reverse Transcription-PCR Analysis-Total RNA extraction was performed using TRIzol reagent (Invitrogen), and cDNA was produced with SuperScript II RNase H-reverse transcriptase (Invitrogen) following the manufacturer's protocol. The 7000 sequence detection system (ABI Prism) and SYBR GREEN PCR Master Mix (Applied Biosystems) amplified ϳ50-bp products during quantitative reverse transcription-PCR. Each sample was analyzed in triplicate and normalized to a nonspecific control. Primer sequences are listed in supplemental Table 1.

RESULTS
Primary HDFs engineered to overexpress c-MYC undergo apoptosis upon serum deprivation (36,46). Using a mutant c-MYC protein incapable of interaction with MIZ-1 (c-MYC V394D), we recently suggested that the c-MYC-MIZ-1 interac-tion is necessary for c-MYC to promote apoptosis in HDFs (36). Although the c-MYC V394D mutant has been widely used, there has not been conclusive evidence that it is exclusively defective for binding to MIZ-1 (47)(48)(49)(50)(51)(52)(53)(54). As with any point mutation, it has remained possible that the mutation interferes with other functions of c-MYC. For example, valine 394 is located within the helix-loop-helix domain, an area necessary for interaction with the transcription factor MAX, the ubiquitin ligase SKP2, and the histone acetyltransferase p300 (55)(56)(57)(58). To fully validate the c-MYC/MIZ-1/apoptosis pathway, it was necessary to conclusively establish whether the V394D mutation selectively blocks MIZ-1 binding. For this purpose, we reasoned that if c-MYC V394D is defective in initiating apoptosis because the mutation selectively interferes with MIZ-1 interaction, then loss of MIZ-1 should rescue the ability of c-MYC V394D to promote cell death. To examine whether loss of MIZ-1 can rescue the apoptotic potential of the c-MYC V394D mutant, we introduced an shRNA construct targeting MIZ-1 into primary HDFs. These cells were previously engineered to express either c-MYC WT or c-MYC V394D (Fig.  1A). Following 24 h of serum deprivation, cells were harvested to document MIZ-1 depletion (Fig. 1B) and quantitate apoptosis induction (Fig. 1C). The shRNA treatment resulted in MIZ-1 mRNA reduction by 60%. Consistent with our previous studies, c-MYC WT induced apoptosis in the presence of the control shRNA, and c-MYC V394D was defective in initiating cell death. The loss of MIZ-1 expression completely rescued the apoptotic defect attributed to c-MYC V394D, providing genetic evidence that this mutant is defective in promoting apoptosis solely because it blocks the interaction between c-MYC and MIZ-1. Expression of a second MIZ-1 shRNA construct yielded identical results, suggesting these data are not because of off-target effects (data not shown).
Having validated a role for MIZ-1 in the inhibition of c-MYC-mediated apoptosis, studies were initiated to define the genetic pathway linking MIZ-1 to cell death. Depending on the cellular context, c-MYC overexpression induces apoptosis through either a p53-dependent or p53-independent mechanism (59). In the p53-dependent pathway, c-MYC overexpression activates p19 ARF, thereby promoting MDM2 inhibition

BCL2 Activation by MIZ-1 Blocks Apoptosis by c-MYC
and subsequent p53 stabilization (30). It was therefore of interest to determine whether the requirement for MIZ-1 inactivation during apoptosis induction was because of a link between the MIZ-1 pathway and p53 stabilization. For this purpose, p53 steady-state protein levels and transcription of p53-induced pro-apoptosis target genes were analyzed in serum-withdrawn HDFs expressing either c-MYC WT or c-MYC V394D ( Fig.  2A). Two fibroblast strains derived from lung (IMR90) and foreskin (2091) were analyzed to address strain-specific effects. Expression of both c-MYC WT and c-MYC V394D induced p53 stabilization (Fig. 2A). In addition, similar induction of the p53 pro-apoptotic target genes BAX, NOXA, and PUMA was observed (Fig. 2B). Furthermore, depletion of p53 failed to inhibit c-MYC-mediated apoptosis in HDFs (data not shown). These findings eliminate the possibility that c-MYC WT and c-MYC V394D differ in apoptotic potential because of distinct capacities for p53 induction.
The previously established antagonism between MIZ-1 and c-MYC suggests the proteins regulate apoptosis through controlling expression of a target gene subset that contributes to cell death. Although c-MYC WT is expected to repress the effector gene(s), c-MYC V394D should be defective in transcriptional repression, correlating with the loss of apoptotic induction. Although c-MYC can repress transcription of a large set of genes, only p21 CIP1 , p15 INK4B , NRAMP1, and MAD4 have been empirically identified as also regulated by MIZ-1. Additionally, the CKI p21 CIP1 has been reported to inhibit both p53-independent and p53-dependent apoptosis (60). In serumwithdrawn HDFs, the four previously established targets of c-MYC and MIZ-1 were not differentially regulated by c-MYC WT and c-MYC V394D, suggesting MIZ-1 does not regulate apoptosis through transactivation of these genes ( Fig. 2A and  3). Both p21 CIP1 and p15 INK4B were induced in response to c-MYC WT expression. Because these CKIs are also p53 transcriptional targets (36,61,62), it is likely p53 stabilization is promoting CKI transcription, thereby overwhelming c-MYC repression, as observed previously (63).
Microarray analysis has identified several hundred genes whose transcription is activated by MIZ-1 (64,65). Analysis of the functional information known regarding these MIZ-1 targets revealed four pro-survival genes: BCLxL, BCL2, MCL1, and the apoptosis inhibitor immediate early response 3 (IER3). As mentioned above, both BCLxL and BCL2 have been reported previously as repressed by c-MYC (31,33,34). To determine whether any of these four targets were responsible for the differential apoptotic potential between c-MYC WT and V394D, mRNA induction of each target was analyzed in serum-deprived HDFs expressing wild-type or mutated c-MYC ( Fig. 2A and Fig. 4, A and B). This analysis revealed that BCLxL, MCL1, and IER3 were regulated identically between c-MYC WT and c-MYC V394D. Like these targets, the pro-apoptotic BIM protein was not differentially regulated by c-MYC WT and c-MYC V394D in HDFs (supplemental Fig. 1). In contrast to all other BCL2 family members examined, BCL2 itself was selectively repressed by c-MYC WT (Fig. 4, A and B).
Among the MIZ-1 targets examined, BCL2 was the only pro-survival gene to be repressed by c-MYC WT and not repressed by c-MYC V394D. To confirm that c-MYC V394D was not generally defective in transcriptional repression, endogenous c-MYC transcription was examined (Fig. 4, C and D). As a mechanism of autoregulation, c-MYC binds to its own promoter to repress transcription (66 -72). The V394D mutant repressed endogenous c-MYC transcription, confirming BCL2 repression by c-MYC specifically requires MIZ-1 inactivation.
As the initial observation linking MIZ-1 to BCL2 transcription was made by microarray, empirical verification of these data was warranted. To assess the ability of MIZ-1 to regulate BCL2 expression, exogenous MIZ-1 was expressed in increasing concentrations, and steady-state levels of BCL2 were examined. BCL2 transcription increased with MIZ-1 expression in a dose-dependent manner (Fig. 5A). Additionally, BCL2 protein was elevated correlating with activated transcription (Fig. 5B). This reproducible observation verified the previously reported microarray data. To assess whether this effect held true when MIZ-1 was expressed at the endogenous level, two shRNA constructs targeting MIZ-1 were introduced into HDFs (Fig. 5, C and D). This analysis revealed that depletion of endogenous MIZ-1 reduced the steady-state transcript and protein levels for endogenous BCL2.

BCL2 Activation by MIZ-1 Blocks Apoptosis by c-MYC
To determine whether endogenous c-MYC regulates BCL2, two c-MYC shRNA constructs were utilized. These shRNAs reduced c-MYC transcript levels by 95% (Fig. 6A). Concomitant with this reduction in endogenous c-MYC, BCL2 transcript levels were elevated several fold (Fig. 6A). The increase in BCL2 mRNA levels was accompanied by a similar increase in endogenous BCL2 protein (Fig. 6B).
The pro-survival activity of BCL2 can block c-MYC-mediated apoptosis and inhibit cell death in vitro (73,74). Within the panel of candidate pro-survival target genes analyzed, only BCL2 was differentially regulated by c-MYC WT and c-MYC V394D. If BCL2 is an essential effector protein of c-MYC-mediated apoptosis through MIZ-1, BCL2 inactivation should rescue the apoptotic potential of the c-MYC V394D mutant. Therefore, to determine whether BCL2 is a functionally relevant target of c-MYC and MIZ-1, the apoptotic potential of c-MYC V394D was assessed after shRNA-mediated depletion of BCL2 (Fig. 7). HDFs expressing c-MYC WT and c-MYC V394D were infected with lentivirus encoding either a control shRNA construct or an shRNA sequence targeting BCL2 (Fig. 7,  A and B). Expression of BCL2 shRNA depleted BCL2 mRNA by ϳ80% (Fig. 7B, left panel). As expected, in the absence of BCL2 shRNA, c-MYC WT repressed BCL2 transcription, whereas c-MYC V394D was defective for BCL2 repression (Fig. 7B, left  panel). Expression of the control gene tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein (YWHAZ) was minimally changed between samples, confirming BCL2 regulation by c-MYC, and depletion of BCL2 through shRNA introduction was specific (Fig. 7B, right panel). In cells expressing the nonspecific shRNA, c-MYC V394D was defective in promoting apoptosis (Fig. 7C). However, BCL2 depletion rescued the apoptotic potential of c-MYC V394D to levels comparable with c-MYC WT. These data implicate BCL2 as a critical target of the c-MYC/MIZ-1 apoptosis pathway.
Recently, small molecules capable of specifically inactivating pro-survival BCL2 family members have been developed (75)(76)(77)(78). As an alternative method of BCL2 inhibition, c-MYC V394D-mediated apoptosis was assessed in the presence of the small molecule BCL2 antagonist ABT-737 (79). ABT-737 interacts with the hydrophobic pocket of BCL2 to inhibit pro-survival activity. Treatment with ABT-737 completely restored the ability of c-MYC V394D to induce apoptosis (Fig. 8). Considered together, these data support a model in which apoptosis induction by c-MYC requires the inhibition of the ability of MIZ-1 to activate BCL2 transcription.   BCL2 is activated by MIZ-1. A, to determine BCL2 regulation by MIZ-1, MIZ-1 was expressed in IMR90 fibroblasts through electroporation of a MIZ-1 expression plasmid. Cells were harvested 48 h post-electroporation, and BCL2 transcription was analyzed by quantitative RT-PCR. Fold induction in transcription normalized to actin mRNA levels is shown. Mean detection among triplicate PCRs Ϯ S.D. is indicated. B, Western blot analysis was utilized to determine BCL2 steady-state protein levels. BCL2 levels were quantified and normalized to ␤-tubulin. The fold increase in total BCL2 protein is indicated. C, to examine endogenous MIZ-1 regulation of BCL2, HDFs were infected with lentivirus encoding a control shRNA (green fluorescent protein) or shRNA targeting MIZ-1. BCL2 and MIZ-1 mRNA levels were analyzed by quantitative RT-PCR. Fold repression in transcription normalized to actin mRNA is shown. Mean detection among triplicate PCRs Ϯ S.D. is indicated. D, Western blot analysis was conducted to determine corresponding BCL2 protein regulation. Whole cell extracts were immunoprecipitated (IP) with a MIZ-1 antibody to enrich MIZ-1 detection.

DISCUSSION
The balance of pro-survival and pro-apoptotic signals within transformed cells determines the rate of tumorigenesis and tumor progression. It is essential for the apoptosis signals activated by oncogene overexpression to be inhibited for oncogenesis to occur. Although c-MYC is a driving signal for cell proliferation, cell growth, metastasis, genomic instability, and angiogenesis, this oncoprotein can also induce apoptosis when survival signals are blocked. We previously demonstrated that MIZ-1 inactivation is essential for c-MYC to induce the apoptotic program. We now extend these observations by showing that the mechanism by which MIZ-1 inhibits c-MYC-mediated apoptosis is through activation of the pro-survival gene BCL2.
We have demonstrated here that c-MYC and MIZ-1 antagonistically regulate BCL2 expression. It remains unresolved whether these transcription factors directly control BCL2 transcription or whether c-MYC and MIZ-1 regulate the expression of intermediate proteins that, in turn, affect BCL2 transcription. Although further studies will be necessary, a previous quantitative chromatin immunoprecipitation screen detected c-MYC binding to the BCL2 locus (80).   However, because many genes are bound by c-MYC without apparent changes in transcription (80,81), this observation does not conclusively prove that BCL2 is a direct c-MYC target. Further complicating the issue, BCL2 mRNA levels are controlled by micro-RNA expression (82). Given that c-MYC affects other target genes by controlling micro-RNA cluster expression (83), c-MYC may regulate BCL2 activity through a mechanism involving micro-RNAs. Clearly, a thorough evaluation of each biochemical scenario is warranted to determine whether BCL2 is a direct or indirect target of the c-MYC-MIZ-1 complex.
The importance of the c-MYC/MIZ-1/BCL2 pathway is underscored by observations from transgenic mouse models and human cancer genetics. In several systems, cooperation between c-MYC and BCL2 in promoting oncogenesis leads to high grade malignancies. For example, in Bcl2 transgenic mice, follicular hyperplasias progress to malignant lymphomas upon rearrangement of the c-Myc locus (40). In addition, E-Myc transgenic mice do not present with lymphomas until ϳ6 -8 months of age (22,84), whereas Bcl2/c-Myc double transgenic mice succumb to immature lymphoblastic leukemia within just days of birth (41). Finally, in c-Myc transgenic mice carrying a conditional Bcl2 allele, Bcl2 deletion results in massive apoptosis (38). Data from human cancer also support a critical link between c-MYC and BCL2. In recent years, a variety of human malignancies have been reported to contain rearrangements of both BCL2 and c-MYC. For instance, aggressive germinal center B cell lymphomas with an IGH:BCL2 translocation also contain either an IGH:MYC fusion or a MYC amplification (39). Furthermore, follicular lymphomas, which are defined by a BCL2 translocation, transform to high grade lymphomas upon rearrangement of the c-MYC locus (37,85), reminiscent of the mouse model discussed above. Moreover, the simultaneous overexpression of BCL2 and c-MYC has been reported in supraglottic squamous cell carcinoma (86) and acute lymphoblastic leukemia (87). Together these mouse models and examples from human cancer genetics strongly suggest a critical prosurvival role for BCL2 expression in c-MYC-driven tumors.
Expression of BCL2 proteins varies by cell type, and c-MYC has been shown to selectively regulate the pro-survival gene(s) necessary for apoptosis inhibition (31). Because other BCL2 family members have been identified as MIZ-1 targets by microarray (65), it is plausible that c-MYC repression of MIZ-1 pro-survival activity functions through different effectors in a cell type-specific manner. Future analysis will address the specificity of c-MYC and MIZ-1 in controlling the BCL2 family proteins.
Given the role established here for MIZ-1 in regulating BCL2 expression, it remains possible that deregulation of MIZ-1 contributes to tumorigenesis. Future studies will be aimed at determining whether specific genetic or epigenetic changes in the MIZ-1 pathway contribute to BCL2 expression in human tumors, particularly those driven by c-MYC. In fact, tumors in both humans and mice display increased steady-state levels of BCL2 in the absence of genetic alterations at the BCL2 locus (31,88), raising the possibility that alterations to upstream components of the MIZ-1 pathway contribute to tumorigenesis in these cases.
Cancer progression and resistance to chemotherapy depend on apoptosis impairment, which often results from an increase in the activity of endogenous pro-survival factors. The reversal of apoptotic defects renders cells susceptible to cytotoxic therapy and tumor regression. Such is the rationale of using small molecule inhibitors, like ABT-737, to restore the endogenous apoptosis mechanism in transformed cells. This class of inhibitors mimics the interaction with BH3-only proteins by binding the hydrophobic pocket of BCL2 family members, thus inhibiting their pro-survival activity. Our findings suggest that BCL2 antagonists can bypass defects in the c-MYC/MIZ-1 pathway to promote apoptosis (Fig. 8). This opens the possibility that small molecule inhibitors targeting the BCL2 family may be particularly successful in treatment of c-MYC-driven lymphomas that also exhibit BCL2 overexpression.
The link established here demonstrating that repression of MIZ-1-driven BCL2 transcription defines a previously unknown genetic pathway linking c-MYC to apoptosis. This pathway may provide a genetic explanation for the numerous studies in humans and mice showing that enforced BCL2 expression cooperates with c-MYC in tumorigenesis.