Increased Expression of Mcl-1 Is Required for Protection against Serum Starvation in Phosphatase and Tensin Homologue on Chromosome 10 Null Mouse Embryonic Fibroblasts, but Repression of Bim Is Favored in Human Glioblastomas *

Inactivating mutations in the tumor suppressor gene phosphatase and tensin homologue on chromosome 10 (PTEN) result in elevated levels of phosphatidylinositol (3,4,5)-trisphosphate, activation of protein kinase B (PKB), and protection against apoptotic insults such as withdrawal of survival factors. Protection may arise through the inhibition of the pro-apoptotic protein Bim, which is normally repressed by a PKB-dependent mechanism. Here we show that PTEN (cid:1) / (cid:1) immortalized mouse embryonic fibroblasts (MEFs) exhibit elevated PKB phosphorylation and are resistant to serum withdrawal-induced death, but exhibit normal Bim expression fol-lowingwithdrawalofserum.Incontrast,expressionofMcl-1,apro- survival member of the Bcl-2 family, was elevated in PTEN (cid:1) / (cid:1) MEFs. Transient or stable overexpression of Mcl-1 in PTEN (cid:2) / (cid:1) MEFs conferred resistance to serum withdrawal, whereas ablating expression of Mcl-1 in PTEN (cid:1) / (cid:1) MEFs, using RNA interference, abolished their resistance to serum withdrawal-induced apoptosis. To determine if Mcl-1 is selected for overexpression in human tumors we examined human glioblastoma cell lines but found that loss of PTEN had no effect on Mcl-1 expression. In contrast, two of three PTEN (cid:1) glioblastoma cell lines low expression of Bim, was refractory to serum that the resistance of PTEN (cid:1) / (cid:1) MEFs to serum is due to the up-regulation of Mcl-1 but that loss of PTEN in tumor cell lines is more complex and may favor de-regulation of different apoptotic regulators such as Bim.

A large number of early and advanced human cancers contain mutations in the tumor suppressor Phosphatase and TENsin homologue on chromosome 10 (PTEN). 3 Loss of heterozygosity (LOH) at the human PTEN locus (10q23) is associated with many high grade glioblastomas and endometrial tumors as well as some melanoma, prostate, and breast carcinomas (1). Germ line mutations of PTEN are also found in rare autosomal dominant syndromes such as Cowden disease, which are characterized by benign hyperplastic outgrowths (2). Re-introduction or overexpression of wild type PTEN in human carcinomas results in suppression of tumor growth, apoptosis, or both (3)(4)(5)(6). Complete loss of PTEN in mice is embryonic lethal, whereas PTEN ϩ/Ϫ mice are viable but develop various neoplasms by adulthood (7).
Resistance to apoptosis can be achieved via the up-regulation and/or down-regulation of anti-apoptotic or pro-apoptotic Bcl-2 family members, respectively. Anti-apoptotic family members share four defined regions of homology, BH1-4, and include Bcl-2, Bcl-X L , A1, and Mcl-1. Pro-apoptotic members are divided into two subfamilies: those that contain domains BH1-3 (Bak and Bax) and those that contain only the BH-3 domain such as Bim, Bad, and Bmf (15). Recent studies have drawn attention to Bim as a target for PKB-dependent survival signals. For example, cytokine withdrawal in mouse Ba/F3 cells results in de novo expression of Bim and apoptosis (16). Bim expression is induced by the transcription factor FOXO3A that is normally inhibited by PKB-dependent phosphorylation (17). Indeed, inhibition of the PI3K pathway is sufficient to induce Bim expression in a variety of cells (16,18,19).
In this study we sought to address whether Bim is repressed following PTEN LOH. We found that PTEN Ϫ/Ϫ immortalized mouse embryonic fibroblasts (MEFs) are resistant to serum withdrawal-induced apoptosis, but this protection was not associated with suppression of Bim but rather with elevated levels of Mcl-1. Indeed, the increase in expression of Mcl-1 was required for PTEN Ϫ/Ϫ MEFs to resist serum starvationinduced apoptosis. In contrast, loss of PTEN in human glioblastomas (GBMs) did not affect Mcl-1 expression, but targeted Bim for repression. These results identify Mcl-1 as a physiologically relevant downstream target expressed following loss of PTEN but also indicate that the biological outcome of deleting PTEN is complex, affecting different proand anti-apoptotic Bcl-2 family proteins in different cells.
Transient Overexpression of Mcl-1-HA-tagged human Mcl-1 cDNA (Hugh Brady, Institute of Child Health, London) was subcloned into pcDNA3.1/Zeo (Invitrogen). PTEN ϩ/Ϫ cells were transfected with 10 g of pcDNA:HA:Mcl-1 or empty vector and 2 g of pcDNA:EGFP-spectrin (Paul Coffer, University of Utrecht), using a calcium phosphate precipitation protocol to ensure that EGFP-expressing cells were also expressing HA:Mcl-1. 20 h later the cells were subjected to a 24-h serum starvation. Cells were stained with propidium iodide and the EGFP-spectrin-expressing population was analyzed for sub-G 1 DNA content by flow cytometry.
Western Blot Analysis and Assay of Apoptosis-Cells were lysed and analyzed by Western blot as described previously (18,20). Caspase activity was monitored by appearance of cleaved caspase-3 and DEV-Dase assay, and cells with hypo-diploid DNA were assayed by propidium iodide staining and flow cytometry (18,20).
Conformational Changes in Bax-Immunoprecipitation of conformationally active Bax using the N-20 antibody was carried out as previously described (18).
Real-time RT-PCR-Preparation of total RNA was performed as previously described (18). (Q)RT-PCR was performed according to the protocol supplied with the TaqMan® reverse transcription reagents (Applied Biosystems) as described previously (18). For mouse Mcl-1 we used 5Ј-TGTAAGGACGAAACGGGACT-3Ј as the forward primer and 5Ј-AAAGCCAGCAGCACATTTCT-3Ј as the reverse primer. For mouse glyceraldehyde-3-phosphate dehydrogenase we used 5Ј-TCAA-CGACCCCTTCATTGAC-3Ј as the forward primer and 5Ј-ATGCA-GGGATGATGTTCTGG-3Ј as the reverse primer. For human Bim we used 5Ј-ATCTCAGTGCAATGGCTTCCA-3Ј as the forward primer and 5Ј-CTCCTGTGCAATCCGTATCT-3Ј as the reverse primer. For human ␤-Actin we used 5Ј-TGTTTGAGACCTTCAACACC-3Ј as the forward primer and 5Ј-TAGGAGCCAGAGCAGTAATC-3Ј as the reverse primer.

PKB Phosphorylation Is Elevated in Immortalized PTEN Ϫ/Ϫ Mouse
Embryonic Fibroblasts-We sought to characterize the effect of serum withdrawal on immortalized MEF cell lines derived from mice heterozygous or homozygous null for PTEN (13). In the first instance, we confirmed by immunoblotting that PTEN was expressed in the PTEN ϩ/Ϫ cells but not in the PTEN Ϫ/Ϫ cells (Fig. 1). PTEN LOH resulted in elevated PKB phosphorylation in cycling cells compared with PTEN ϩ/Ϫ cells (9, 13), whereas total PKB and total Bax protein levels did not differ and served as a loading control.
The difference in PKB phosphorylation was more pronounced following serum starvation ( Fig. 2A). In PTEN ϩ/Ϫ MEFs PKB phosphorylation declined rapidly following withdrawal of serum and remained low. In contrast, in PTEN Ϫ/Ϫ MEFs, the decline was less pronounced, and the levels of PKB phosphorylation recovered from 3 h onwards. ERK1/2 phosphorylation also declined rapidly following serum starvation in the PTEN ϩ/Ϫ cells and only started to recover after 24 h; however, we did observe in some experiments that this recovery was more rapid in PTEN Ϫ/Ϫ cells, perhaps suggesting that loss of PTEN may impact on the ERK pathway in some instances (21). No differences in total ERK1/2, PKB, or Bax levels were observed ( Figs. 1 and 2A).
PTEN Ϫ/Ϫ MEFs Are Resistant to Serum Withdrawal-induced Apoptosis-Apoptosis following serum withdrawal was initially assessed by immunoblotting total cell lysates with an antibody that recognizes the cleaved (17 kDa) active form of caspase-3. Following withdrawal of serum from cycling PTEN ϩ/Ϫ cells, levels of cleaved caspase-3 protein increased rapidly, being apparent by 1 h and maximal by 3 h (Fig.  2A). Although serum withdrawal did promote an increase in cleavage of caspase-3 in PTEN Ϫ/Ϫ cells, levels were substantially reduced when compared with the corresponding PTEN ϩ/Ϫ samples. To further quantify the reduction in caspase activation we compared the ability of whole cell extracts to cleave the synthetic caspase substrate Ac-DEVD-AMC (a DEVDase assay). In PTEN ϩ/Ϫ cells DEVDase activity was clearly elevated 4 h after serum withdrawal and was maximal by 24 h, whereas samples derived from serum-starved PTEN Ϫ/Ϫ cells exhibited significantly reduced DEVDase activity (p Ͻ 0.05) such that it barely increased above that at time zero (Fig. 2B). In addition, PTEN Ϫ/Ϫ cell cultures contained significantly fewer cells with sub-G 1 DNA than the equivalent PTEN ϩ/Ϫ population following 12 and 24 h (p Ͻ 0.05) of serum withdrawal (Fig. 2, C and D). However, despite the strong inhibition of caspase activation, we did observe that protection conferred by loss of PTEN was progressively lost after 24 h. This may indicate that cells can also undergo cell death by a parallel PTEN-independent pathway at later time points, perhaps reflecting a later necrotic response. A much smaller, but still significant (p Ͻ 0.05), reduction in sub-G 1 DNA content was also observed in PTEN Ϫ/Ϫ cells following a 24-h treatment with the DNA-damaging agent etoposide (Fig. 2D). These data clearly show that PTEN LOH in MEFs prevents the activation of caspases and the appearance of downstream markers of apoptosis following serum withdrawal.
Serum Withdrawal-induced Apoptosis in PTEN ϩ/Ϫ MEFs Occurs in the Absence of de Novo Bim Expression-The up-regulation or activation of pro-apoptotic BH3-only proteins serves to link cellular stress to initiation of the cell death response (15). Bim, in particular, is strongly implicated in promoting cell death following withdrawal of survival factors (16,18,19,(22)(23)(24)(25)(26). In viable cells Bim expression is repressed, in part, by PKB-dependent phosphorylation and inactivation of FOXO transcription factors (17,19), so we investigated whether Bim expression was down-regulated in PTEN Ϫ/Ϫ cells. Western blot analysis revealed that Bim expression did increase following serum withdrawal (Figs. 3A and 4A). However, there was no difference in the magnitude of Bim expression between PTEN ϩ/Ϫ and PTEN Ϫ/Ϫ cells.
Death following withdrawal of survival factors is blocked by cycloheximide in many cell types, indicating a requirement for de novo synthesis of a pro-apoptotic protein, such as Bim (18,23). Indeed, the increase in expression of Bim following serum withdrawal was blocked by cycloheximide (Fig. 3A). However, cycloheximide did not block serum withdrawal-induced caspase activation in PTEN ϩ/Ϫ cells (Fig.  3B); PTEN Ϫ/Ϫ cells again failed to activate caspases. Thus, even though  Bim expression increases following serum withdrawal, this is not required for cell death in these immortalized MEFs. Rather, serum withdrawal-induced death proceeds through pre-existing components, and the ability of PTEN status to determine this response is unlikely to be due to differences in de novo Bim expression.
Mcl-1 mRNA and Protein Levels Are Elevated in PTEN Ϫ/Ϫ MEFs-Because changes in Bim expression were unlikely to be the PTEN-dependent changes in cell death in MEFs, we examined the expression of other Bcl-2 family members (Fig. 4A). Following serum withdrawal from PTEN ϩ/Ϫ and PTEN Ϫ/Ϫ cells we observed no difference between the cell lines with respect to the expression of Bim and Bmf (BH3-only subfamily) or Bax and Bak (Bax sub-family) (Fig. 4A). Levels of the anti-apoptotic protein Bcl-2 steadily decreased upon serum starvation in both cell lines, although absolute levels were higher in PTEN ϩ/Ϫ cells. Bcl-X L protein levels remained largely unchanged throughout the experiment, although again, absolute levels were higher in PTEN ϩ/Ϫ cells. However, we did observe changes in the expression of Mcl-1 that were consistent with it conferring protection from apoptosis. First, we observed that the basal expression of Mcl-1 in PTEN Ϫ/Ϫ cells was typically 2.5-fold greater than that observed in PTEN ϩ/Ϫ cells (Fig. 4, A and  B). Second, the level of Mcl-1 declined following serum withdrawal in PTEN ϩ/Ϫ cells but was maintained in serum-starved PTEN Ϫ/Ϫ cells (Fig. 4A).
Quantitative real-time RT-PCR was used to determine levels of Mcl-1 mRNA throughout a time course of serum withdrawal (Fig. 4C). Mcl-1 mRNA levels were normalized to glyceraldehyde-3-phosphate dehydrogenase mRNA levels and then plotted as -fold change over basal levels (time ϭ 0) in PTEN ϩ/Ϫ cells. The basal level of Mcl-1 mRNA in PTEN Ϫ/Ϫ cells was 2-fold higher than that observed in PTEN ϩ/Ϫ cells and increased still further following serum starvation. In contrast, Mcl-1 mRNA levels remained unchanged in PTEN ϩ/Ϫ cells so that from 9 h Mcl-1 mRNA levels were typically 3.5-fold higher in PTEN Ϫ/Ϫ cells versus PTEN ϩ/Ϫ cells.
Overexpression of Mcl-1 Protects PTEN ϩ/Ϫ Cells from Serum Withdrawal-We deemed that elevated expression of Mcl-1 was a plausible mechanism to account for the observed protection against serum withdrawal in PTEN Ϫ/Ϫ cells. To test this directly, we transiently transfected PTEN ϩ/Ϫ cells with plasmids expressing HA:Mcl-1 and an EGFP: spectrin fusion protein, before serum starving for 24 h (Fig. 5, A and B). Cells transfected with empty vector and EGFP:spectrin served as a negative control. Immunoblots confirmed expression of Mcl-1 and EGFP: spectrin (Fig. 5A). When EGFP-positive cells were analyzed for sub-G 1 DNA content we observed that transient expression of Mcl-1 caused a significant (3-fold) reduction in serum withdrawal-induced cell death (Fig. 5B). Immunoblotting confirmed that this Mcl-1-dependent protection did not block the serum starvation-induced increase in Bim expression (Fig. 5A).

Mcl-1 Is Required for Resistance to Serum Withdrawal in PTEN Ϫ/Ϫ
Immortalized MEFs-To test whether the increase in Mcl-1 expression in PTEN Ϫ/Ϫ MEFs was required for their protection against serum withdrawal, we used a small hairpin RNA interference strategy to reduce Mcl-1 expression in PTEN Ϫ/Ϫ MEFs. Two independent PTEN Ϫ/Ϫ clonal cell lines (Ϫ/Ϫ1 and Ϫ/Ϫ8) were derived from the parental PTEN Ϫ/Ϫ MEFs. Western blot analysis showed that endogenous Mcl-1 protein levels were reduced in both clones, relative to that of the parental PTEN Ϫ/Ϫ cell line (Fig. 7A). PTEN Ϫ/Ϫ clones 1 and 8, in which Mcl-1 expression was reduced, were sensitized to serum withdrawal-induced apoptosis relative to parental PTEN Ϫ/Ϫ cells, as judged by an increase in DEVDase activity and an increase in sub-G 1 DNA content (Fig. 7, B and C). Furthermore, the cells that displayed the most effective knockdown of Mcl-1 (Ϫ/Ϫ8) exhibited the greatest sensitivity to serum starvation. Thus, Mcl-1 is necessary for most, if not all, of the resistance to serum withdrawal-induced apoptosis observed in immortalized PTEN Ϫ/Ϫ MEFs.

Deletion of PTEN in Human Glioblastoma Cells Does Not Affect Mcl-1 Expression but Does Inhibit Bim Expression-Prompted by our
results in MEFs, we wished to address whether Mcl-1 expression was also increased in PTEN Ϫ/Ϫ human GBMs (1,27). In the first instance we compared expression of Mcl-1 in the two MEF cell lines with that in three GBM cell lines, including those that were wild type (LN18 and LN229) or homozygous null (U87-MG) for PTEN (Fig. 8A). Once again, Mcl-1 levels were elevated in PTEN Ϫ/Ϫ MEFs compared with PTEN ϩ/Ϫ MEFs. The PTEN Ϫ/Ϫ GBM cell line U87-MG exhibited elevated levels of PKB phosphorylation in comparison to the two PTEN ϩ/ϩ GBM cell lines (LN18 and LN229), whereas total PKB levels remained unchanged. Side-by-side analysis of the MEFs and the human GBMs revealed that human Mcl-1 resolved at a slightly higher molecular weight than murine Mcl-1 on SDS-PAGE gels. More importantly it revealed that Mcl-1 levels in the three GBM cell lines were identical, regardless of PTEN status This analysis was expanded by the inclusion of two additional PTEN Ϫ/Ϫ GBM cell lines, U251-MG and U373-MG (Fig. 8B). All three PTEN Ϫ/Ϫ GBMs were characterized by very high levels of basal PKB phosphorylation, which was refractory to serum starvation. Differences in ERK1/2 phosphorylation were observed between some GBM cell lines, but these were independent of PTEN status. Again, we did not observe any increase in expression of Mcl-1 in any of the GBM cell lines regardless of PTEN status (Fig. 8B). Indeed, in one case (U373-MG, PTEN Ϫ/Ϫ ) we observed a reduction in Mcl-1 expression. Thus, in contrast to the situation in MEFs, loss of PTEN does not favor increased expression of Mcl-1 in glioblastoma cell lines.  In the course of this analysis, however, we did observe substantial differences in the expression of Bim that correlated well with PTEN status (Fig. 8B). Serum starvation caused a clear increase in Bim expression in both PTEN ϩ/ϩ LN18 and LN229 cell lines, which was apparent between 3 and 7 h and analogous to that seen in CCl39 and Rat-1 fibroblasts (18,25). In contrast, two of the three PTEN Ϫ/Ϫ cell lines, U87-MG and U373-MG, expressed only very low levels of Bim. U251-MG cells maintained very high levels of Bim protein, even in the presence of survival factors. Actin was used as a loading control.
Further analysis focused on LN229 and U87-MG as representative of PTEN wild type or PTEN null GBM cell lines, respectively. Quantitative real-time PCR showed that serum starvation resulted in a clear increase in Bim mRNA in LN229 cells (3.5-fold above basal), whereas U87-MG cells exhibited only a modest increase in Bim mRNA (Fig. 8C). Neither U87-MG nor LN229 cells underwent apoptosis following serum withdrawal as measured by immunoblotting for cleaved caspase-3, DEV-Dase activity or sub-G 1 DNA content. 4 However, when we examined the activation of Bax, using the appearance of the conformationally active N-20 epitope (18), we observed that the PTEN null U87-MG cells exhibited no activation of Bax following serum withdrawal, whereas the LN229 cells exhibited clear activation of Bax. The total, activable Bax was the same in the two cell lines as judged by the ability of triton X-100 to fully expose the N-20 epitope.
Taken together, these data suggest that in at least two independent human GBM cell lines loss of PTEN results not in the up-regulation of Mcl-1 but rather the down-regulation of Bim mRNA and protein. Furthermore, the reduction in Bim expression in a PTEN Ϫ/Ϫ GBM correlates with decreased Bax activation following serum starvation.

Increased Expression of Mcl-1 Is Required for Resistance to Serum
Starvation in PTEN Ϫ/Ϫ MEFs-Deletion of PTEN is associated with increased resistance to multiple death stimuli, orchestrated primarily through activation of the key survival kinase PKB and its downstream targets (9, 13, 28 -30). PKB can regulate cell survival indirectly through phosphorylation of the pro-apoptotic transcription factors FOXO3A (31) and YAP (32), the JNK scaffold proteins JIP1 and POSH (33,34), and the p53-binding protein MDM2 (35). The relative importance of these many PKB targets may vary in different cell types. Furthermore, some PKB targets may be influenced by other pathways making interpretation more difficult; for example, Bim is regulated by the PKB (16), ERK (18,35), and JNK (24) pathways.
To simplify interpretation we examined serum withdrawal-induced apoptosis, because (a) this is a relatively benign form of cell stress that doesn't involve DNA damage, (b) serum growth factors would be expected to determine in part the activation state of the PKB pathway, and (c) loss of growth factor dependence for survival is a hallmark of cancer cells (36). Our results confirm that PTEN LOH in MEFs causes constitutive activation of PKB and confers considerable protection against serum starvation (Figs. 1 and 2) (9). Analysis of pro-and antiapoptotic Bcl-2 family members showed that Mcl-1 protein and mRNA levels were elevated in PTEN Ϫ/Ϫ cells. Furthermore, RNA interferencemediated knockdown showed that this increase in Mcl-1 expression in PTEN Ϫ/Ϫ MEFs was required for resistance to serum withdrawal, whereas transient or stable overexpression of Mcl-1 alone was sufficient to protect the otherwise sensitive PTEN ϩ/Ϫ MEFs. Thus, of all the changes that could result from loss of PTEN, increased expression of Mcl-1 alone is necessary and sufficient to account for the resistance of PTEN Ϫ/Ϫ MEFs to serum starvation.
Mcl-1 differs from the majority of Bcl-2 proteins in that it does not possess a BH-4 domain, has a short half-life, and is degraded by the proteasome (37). Indeed, the PI3K/PKB pathway has previously been implicated in increasing Mcl-1 protein translation (38), whereas granulocyte macrophage-colony stimulating factor has been shown to increase Mcl-1 protein stability via PKB and ERK-dependent phosphorylation (39). However, we observed no difference in the half-life of either Mcl-1 protein or mRNA in PTEN ϩ/Ϫ and PTEN Ϫ/Ϫ cells, 4   Increased Mcl-1 expression certainly has the capacity to confer a malignant phenotype (44) and is often associated with increased resistance to apoptosis (45)(46)(47). In addition, Taniai and co-workers (48) have recently shown that RNA interference-mediated down-regulation of Mcl-1, but not Bcl-2 or Bcl-X L , was sufficient to confer sensitivity to tumor necrosis factor-related apoptosis-inducing ligand-induced death in cholangiocarcinoma cells. Our results suggest that increased expression of Mcl-1 alone is necessary and sufficient to account for the resistance of PTEN Ϫ/Ϫ MEFs to serum starvation, and we believe it represents the first demonstration of a clear genetic link between PTEN LOH and an increase in Mcl-1 expression.
Prompted by these observations we anticipated that loss of PTEN in MEFs would protect cells from serum withdrawal by blocking the increase in Bim expression. However, two observations argue against this. First, there was no difference between PTEN ϩ/Ϫ and PTEN Ϫ/Ϫ MEFs in the magnitude or kinetics of Bim expression following serum withdrawal, indicating that de-regulation of the PKB pathway following loss of PTEN does not impair Bim expression in MEFs. Second, although serum starvation did promote an increase in Bim expression, this was not required for cell death in PTEN ϩ/Ϫ cells, because cycloheximide blocked the increase in Bim expression but did not inhibit cell death. This is in striking contrast to the situation in CCl39 fibroblasts (18), sympathetic neurons (50), and lymphocytes (16), where death requires de novo protein synthesis. Indeed, it is notable that following serum withdrawal the appearance of cleaved, activated caspase-3 was very rapid in MEFs, being apparent within 1 h (Fig. 2), whereas in CCl39 fibroblasts caspase activation is not apparent until 5-6 h after serum withdrawal and can be blocked by cycloheximide (18). Thus, in contrast to many cell types, serum withdrawal-induced cell death must proceed through the post-translational modification of pre-existing components in MEFs. This doesn't rule out a role for Bim in cell death following serum starvation in these MEFs, but it does argue that Bim would need to be regulated by some post-translational mechanism under these circumstances (25,51,52). PTEN mutation and/or LOH in human GBMs is well documented (1,27,53,54). Despite our data showing that PTEN status was a clear FIGURE 8. PTEN ؊/؊ human glioblastoma cell lines fail to up-regulate Mcl-1 but display altered Bim expression following serum withdrawal. A, the indicated MEFs and human GBM cell lines were maintained in complete medium (C, 10% FBS) or switched to serum-free (SF) medium for the times indicated. Normalized total protein was resolved by SDS-PAGE and then immunoblotted with antibodies for Mcl-1, total PKB, and phospho-PKB (Thr-308). B, human GBM cell lines were maintained in complete medium (C, 10% FBS) or switched to serum-free (SF) medium for the times indicated. Normalized cell lysates obtained from cycling cells was resolved by SDS-PAGE and then immunoblotted with antibodies for phospho-PKB (Ser-473), total PKB, phospho-ERK1/2 (Thr-202/Tyr-204), ERK2, Mcl-1, Bim, and actin. Data were taken from a single experiment that is representative of three. C, total RNA was derived from cycling or serum-starved LN229 or U87-MG GBMs. Levels of Bim transcript were normalized to those of ␤-actin transcript, both quantified using real-time RT-PCR, and expressed as -fold change over basal. Data are presented as the mean Ϯ S.D. of triplicate assays and are representative of two independent experiments giving similar results. D, LN229 or U87-MG GBMs were maintained in complete medium (C) or were serum-starved (SF) for 16 h before lysing in CHAPS buffer. Cells maintained in complete medium but lysed in the presence of 1% Triton X-100 (TX) served as a positive control and effectively set the maximum in the assay. Conformationally active Bax was immunoprecipitated with the N-20 antibody and compared with total Bax levels from whole cell extracts. Immunoblotting whole cell lysates with an Actin antibody served as a loading control. determinant of Mcl-1 protein levels in MEFs, we found that Mcl-1 levels were not elevated in any of the PTEN Ϫ/Ϫ GBMs we tested, even though the PKB pathway was constitutively active. In contrast, Bim expression was markedly reduced in two of three PTEN Ϫ/Ϫ GBMs compared with PTEN ϩ/ϩ GBMs, showing for the first time that loss of PTEN in human GBMs results in reduced Bim expression.
Regardless of differences in Bim expression, both PTEN Ϫ/Ϫ and wildtype GBMs were equally resistant to serum withdrawal-induced apoptosis as measured by caspase activation and sub-G 1 DNA content. 4 This is not surprising because it is well known that human GBM cell lines exhibit a cell-intrinsic resistance to a wide range of pro-apoptotic stimuli due to the up-regulation of IAP (inhibitor of apoptosis) proteins, which confer a broad protective effect by preventing activation of caspases (55,56). Indeed, this may explain why one PTEN Ϫ/Ϫ GBM cell line (U251-MG) was able to withstand quite high levels of Bim. Despite this, we did observe that PTEN null U87-MG cells failed to activate Bax following serum withdrawal, whereas Bax activation was observed in the PTEN wild type LN229 cells. This suggests that loss of PTEN in GBMs can limit or inhibit the apoptotic pathway at a point upstream of the mitochondria; a result consistent with reduced Bim levels.
The reason for the high Bim levels in the U251-MG cell line is unclear and could reflect changes in the Bim gene itself or changes in mRNA or protein half-life, although it would seem surprising for high Bim expression to be selected for in a tumor cell line. Indeed, Bim actually exhibits many of the hallmarks of a tumor suppressor gene as might be expected if it is a determinant of cell death at limiting concentrations of survival factor. Indeed, loss of Bim in mice accelerates Myc-induced B-cell leukemia in a mouse model (57) and is frequently deleted in mantle cell lymphoma (58). In solid tumor cells one study has demonstrated repression of Bim expression and resistance to anoikis in the human breast cancer cell line MDA-MB-231 that overexpresses epidermal growth factor receptor (59). Here we have shown for the first time that in naturally occurring PTEN Ϫ/Ϫ human GBM cell lines Bim expression is repressed at the level of mRNA and protein. It remains to be seen whether this holds true for other PTEN null human cancers.
In summary, the molecular consequence of abrogating PTEN gene function is complex and depends on the cell type. Indeed, our work highlights the dangers associated with extrapolating from one cell type to another. Deleting PTEN in MEFs results in the up-regulation of Mcl-1, which is both necessary and sufficient to mediate resistance to serum withdrawal-induced apoptosis. In contrast, loss of PTEN in human glioblastomas causes the down-regulation of Bim without changing the expression of Mcl-1. The cellular mechanisms that determine which Bcl-2 family member is affected following PTEN deletion for any given cell are unknown. With this in mind it is interesting to note that the anti-apoptotic effect of Mcl-1 in B and T lymphocytes is through selective binding and inhibition of Bim, suggesting a physiological link between the two proteins in vivo (60). Thus, if Mcl-1 expression is increased (as in PTEN Ϫ/Ϫ MEFs) there may be no requirement for repression of Bim, whereas in PTEN Ϫ/Ϫ GBMs repression of Bim may negate the need to up-regulate Mcl-1.
In the labeling of the y axes on Figs. 2, 3, 5, 6, 7, and 8, the Greek lowercase epsilon was transformed into an exclamation point. The corrected figures are shown below. Additions and Corrections DECEMBER  Increased expression of Mcl-1 is required for protection against serum starvation in phosphatase and tensin homologue on chromosome 10 null mouse embryonic fibroblasts, but repression of Bim is favored in human glioblastomas.

PAGE 33281:
Under "Experimental Procedures," subheading "Real-time RT-PCR," the human Bim and ␤-actin primers described are wrong. The published primers actually amplify murine Bim and ␤-actin transcript rather than the stated human transcript. The real-time RT-PCR data presented in Fig. 8C still remains factually correct, as it was derived using the human Bim and ␤-actin primers. The correct primers used in this study are described as follows: Human Bim forward, 5Ј-TGC AGA CAT TTT GCT TGT TCA A-3Ј, and reverse, 5Ј-GAA CCG CTG GCT GCA TAA TAA-3Ј; human ␤-actin forward, 5Ј-CTC CTC CTG AGC GCA AGT ACT C-3Ј, and reverse, 5Ј-CGG ACT CGT CAT AGT CCT GCT T-3Ј. "Convertases" in the title was printed as "Covertases." The correct title is listed above.