BimEL is an important determinant for induction of anoikis sensitivity by mitogen-activated protein/extracellular signal-regulated kinase kinase inhibitors

Adhesion to appropriate extracellular matrix is essential for survival of many normal cells, and loss of adhesion induces programmed cell death, a phenomenon termed anoikis (1–3). This mechanism prevents cells from colonizing outside of their correct tissue context and is essential for maintenance of tissue homeostasis and architecture. Acquisition of resistance to anoikis is a critical step in cancer progression. The ability to survive and proliferate without proper positional signals is believed to be a prerequisite for invasion and metastasis to ectopic locations. Understanding the molecular biology that confines normal cells to their natural environment and the mechanisms by which cancer cells escape that restraint has implications in cancer research. Agents that can reestablish anoikis sensitivity would represent novel therapeutic strategies. We reported that the mitogen-activated protein (MAPK)/extracellular signal-regulated kinase (ERK) kinase (MEK) inhibitor U0126 sensitized HBC4 and MB231, two breast cancer cells with constitutively activated ERK to anoikis (4). In this article, we show that the BH3-only protein BimEL plays a decisive role in the induction of sensitivity to anoikis by MEK inhibitors.

Apoptosis signaling of the mitochondrial pathway is regulated by proapoptotic and antiapoptotic Bcl-2 family proteins (5, 6). These proteins are distinguished by the presence of at least one of four Bcl-2 homology domains. The antiapoptotic proteins such as Bcl-2, Bcl-x_L, Mcl-1, and Boo/Diva share three or four Bcl-2 homology domains. The proapoptotic members can be divided into two subgroups: multidomain members such as Bax and Bak and BH3-only proteins that share homology only within the short BH3 domain. The BH3-only proteins in mammals include Bad, Bid, Bik, Bim, Blk, Bmf, Hrk, Noxa, and Puma. The BH3-only proteins act as sensors of various stresses. Each BH3-only protein recognizes distinct stimuli and is activated by diverse mechanisms, but all provoke release of cytochrome c from mitochondria by binding to and neutralizing antiapoptotic Bcl-2 family proteins, which in turn activates the proapoptotic proteins Bax and Bak (5, 6). In healthy cells, various transcriptional and post-translational mechanisms operate to maintain BH3-only proteins in a latent form to prevent inappropriate cell death. For example, Bim and Bmf are sequestered to the microtubules and actin cytoskeleton, respectively, away from antiapoptotic Bcl-2 family proteins in mitochondria (7, 8).

Bim was discovered by an expression screen for proteins that bind to Bcl-2 (9) and also by a yeast two-hybrid system using Mcl-1 as bait (10). Three major isoforms BimEL, BimL, and BimS were initially identified, and several other splice variants have been reported recently (11–13). The apoptotic activity of BimEL and BimL is suppressed by binding to dynein light chain DLC-1, a component of the dynein motor complex on the microtubules (7). BimS lacks the binding domain to DLC-1 and thus is constitutively activated. Bmf was also identified as a Mcl-1 interacting protein using a yeast two-hybrid screen (8). Bmf is sequestered to the myosin V motor complex associated with the actin cytoskeleton through interaction with DLC-2. Bim and Bmf both seem to be proteins that check cytoskeletal integrity but monitor distinct damage. Bim is released by paclitaxel treatment (8), which polymerizes microtubules, and also by cytokine deprivation and abnormal calcium flux (7, 14, 15). Bmf, on the other hand, is activated by actin-depolymerizing agents and cell detachment (8).

Here, we show that Bim is also activated on detachment in some cancer cells. These cancer cells escaped anoikis most likely by rapidly degrading BimEL, a major isoform of Bim. Degradation depended on phosphorylation of BimEL by ERK, and MEK inhibitors blocked the phosphorylation and hence degradation. As a consequence, BimEL accumulated and sensitized these cells to anoikis. Our results explain in part the mechanism of induction of anoikis sensitivity by MEK inhibitors and the reason why only certain cells are sensitized. In addition, our data also provide implications for predicting the efficacy of MEK inhibitors.

GATEWAY vectors and enzymes were from Invitrogen (Carlsbad, CA). Recombinant ERK2, c-Jun NH₂-terminal kinase (JNK), p38MAPK, and RSK2 were from Upstate (Charlottesville, VA). U0126 was from Promega (Madison, WI). Fugene 6 was from Roche Diagnostics (Indianapolis, IN). Rabbit anti-phospho-BimEL(Ser⁶⁹) antibody was raised against a synthetic peptide corresponding to amino acids 62 to 75 (C-GPLAPPApSPGPFAT) of human phospho-BimEL(Ser⁶⁹) and was affinity purified by standard procedures. Other antibodies used were from commercial sources: Bim from ProSci, Inc. (Poway, CA); Bcl-x from PharMingen (San Diego, CA); COX4 from Clontech (Palo Alto, CA); ERK1, ERK2, and α-tubulin from Santa Cruz Biotechnology (Santa Cruz, CA); phospho-ERK from Cell Signaling Technology (Beverly, MA); X-press tag from Invitrogen; and ubiquitin from MBL (Nagoya, Japan). Poly-2-(hydroxyethyl methacrylate) was from Sigma Chemical Co. (St. Louis, MO).

Human BimEL and Bmf cDNAs were isolated by PCR from HeLa cDNA library and subcloned into pDONR201 entry vector (Invitrogen). Mutations were introduced in GATEWAY entry clones using QuickChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). Primers used were BimEL(S59A): forward 5′-GCTGCCCCCACGGCGCCCCTCAGGGCCCGC, reverse 5′-GCGGGCCCTGAGGGGCGCCGTGGGGGCAGC; BimEL(S69A): forward 5′-GGCCCCACCTGCCGCCCCTGGCCCTTTTGC, reverse 5′-GCAAAAGGGCCAGGGGCGGCAGGTGGGGCC; BimEL(S77A): forward 5′-GGCCCTTTTGCTACCAGAGCCCCGCTTTTCATCTTTATGAG, reverse 5′-CTCATAAAGATGAAAAGCGGGGCTCTGGTAGCAAAAGGGCC; and BimEL(L152A/D157A): forward 5′-GGATCGCCCAAGAGGCGCGGCGTATTGGAGCCGAGTTTAACGC, reverse 5′-GCGTTAAACTCGGCTCCAATACGCCGCGCCTCTTGGGCGATCC. Genes were moved into GATEWAY destination vectors according to the method described by the manufacturer. Plasmid encoding active MEK1 (pFC-MEK1) was purchased from Stratagene.

For expression in Escherichia coli, genes were moved into pEXP1-DEST vector (Invitrogen), and expression clones were introduced into BL21 Star (DE3)pLysS strain E. coli (Invitrogen) to produce proteins with NH₂-terminal polyhistidine and X-press tags. Proteins were purified with Ni-NTA resin (Qiagen, Valencia, CA) according to the method recommended by the manufacturer and used as substrates for in vitro kinase and ubiquitination assays. To express exogenous BimEL proteins in mammalian cells, two conserved residues in the BH3 domain (L152 and D157) were mutated to alanines. Genes were transferred into pDEST26 vector to produce BimEL protein with NH₂-terminal polyhistidine tag. Expression clones were transfected into cells using Fugene 6 and expressed proteins were purified with Ni-NTA resin and subjected to immunoblotting.

To immunoprecipitate Bim from MCF7 cells, anti-Bim antibody was cross-linked to immobilized protein G using a Seize X Mammalian Immunoprecipitation Kit (Pierce, Rockford, IL). Cells were lysed in buffer containing 20 mmol/L HEPES (pH 7.4), 150 mmol/L NaCl, 1 mmol/L EDTA, 1 mmol/L Na₃VO₄, 10 mmol/L NaF, 1% Triton X-100, and protease inhibitor cocktail (Sigma P8340). After removing insoluble material, the extract was incubated with conjugated anti-Bim antibody. Immunoprecipitated proteins were eluted with ImmunoPure Elution Buffer (Pierce). Mitochondria fractions were isolated using an ApoAlert Cell Fractionation Kit (Clontech). Immunoblotting analysis was done as described (4).

Bacterially expressed Bim and Bmf proteins were phosphorylated in vitro using recombinant protein kinases according to protocols provided by Upstate. The reaction was terminated by addition of concentrated SDS-PAGE sample buffer. Phosphorylation was analyzed using a Bioimage Analyzer BAS-1800 (Fujifilm, Tokyo, Japan) after SDS-PAGE.

In vitro ubiquitination assay of bacterially expressed BimEL proteins was done using MDA-MB231 cell lysate as described by Cockman et al. (16).

Human breast cancer cells used in this study have been described (4). Poly-2-(hydroxyethyl methacrylate)–coated plates were prepared as described (17, 18).

We reported that MEK inhibitors trigger apoptosis in human breast cancer cells HBC4 and MDA-MB231 when cells are deprived of anchorage but not when anchored (4). The phenomenon resembled anoikis, a form of apoptosis that occurs in various normal cell types on detachment from their proper substratum. To elucidate the mechanisms by which MEK inhibitors restore anoikis sensitivity, we examined the inhibitors' effects on Bcl-2 family proteins.

Treatment of HBC4 and MDA-MB231 with MEK inhibitors U0126 and PD184352 caused a rapid increase of the proapoptotic BH3-only protein, BimEL (Fig. 1). Levels of BimL and BimS, the two other major isoforms of Bim, slightly increased in MDA-MB231 but did not change in HBC4. Inhibitors of JNK, p38MAPK, or PI3K did not seem to have any effect on the amount of BimEL (data not shown). In other human breast cancer cells in which anoikis sensitivity was not induced by MEK inhibitors, the increase in BimEL level following MEK inhibitor treatment was less extensive (Fig. 2 and data not shown).

Because the reestablishment of susceptibility to anoikis and the increase in the proapoptotic protein BimEL seemed to occur concurrently, it was tempting to assume a connection between these two events. However, BimEL increased on MEK inhibitor treatment in HBC4 and MDA-MB231 cells regardless of the state of anchorage. Although only nonanchored cells entered apoptosis, the amount of BimEL was comparable in anchored and nonanchored cells.

To address this issue, we examined the possibility that Bim, like Bmf (8), is released from the dynein motor complex and activated in response to cell detachment. Mitochondrial fractions from anchored and nonanchored MDA-MB231 and HBC4 cells with or without U0126 treatment were isolated and analyzed by immunoblotting. As shown in Fig. 3, significantly higher levels of BimEL were detected in the mitochondrial fractions from U0126-treated nonanchored cells compared with those of anchored cells. The results indicate that in HBC4 and MDA-MB231 cells BimEL translocates to mitochondria from microtubules on detachment.

To explore whether Bim plays a role in determining sensitivity to MEK inhibitors, we also examined the levels of Bim proteins in eight breast cancer cells. MDA-MB231 and HBC4 had exceedingly low basal Bim levels compared with other breast cancer cells (Fig. 4). This suggests that these two cancer cells promote survival by keeping the level of Bim low and are thus vulnerable to its augmentation and sensitized to anoikis by MEK inhibitors. Other cells probably have acquired mechanisms to resist higher levels of Bim and are less affected by MEK inhibition.

Transcriptional mechanisms contribute to Bim expression in hematopoetic (15, 19–21) and neuronal (22–25) cells and also in MCF7 (26). Recent reports showed that inhibition of the MEK-ERK pathway increased Bim mRNA in MCF10A human mammary epithelial cells (27, 28) and Rat-1 fibroblasts (29). In MDA-MB231 and HBC4 cells, however, we did not observe substantial induction of the BimEL transcript by MEK inhibitors using reverse transcription-PCR (data not shown), and the level of BimEL seemed to be controlled by post-translational mechanisms. BimEL is a phosphoprotein (15, 30). We have also noticed that, in addition to elevation in protein level, treatment with MEK inhibitors increased the electrophoretic mobility of BimEL, indicative of reduced phosphorylation. We thus investigated the possibility that BimEL is a substrate of ERK or a kinase downstream of ERK and that the BimEL level is regulated by phosphorylation.

To test this hypothesis, recombinant BimEL was produced in E. coli and subjected to in vitro kinase reaction using ERK or its downstream kinase RSK2. As shown in Fig. 5, BimEL was effectively phosphorylated by ERK but very weakly by RSK2. ERK phosphorylates serine and threonine residues with proline at +1 position. Of the six possible sites in BimEL (Fig. 6A), we focused on three serine residues in the BimEL-specific domain because BimL and BimS were not phosphorylated by ERK (data not shown). The three potential target serines, Ser⁵⁹, Ser⁶⁹, and Ser⁷⁷, were mutated to alanine residues and tested for phosphorylation by ERK. Whereas the S59A and S77A mutants could still be phosphorylated by ERK as efficiently as the wild-type BimEL, phosphorylation of the S69A mutant was barely detectable (Fig. 6B). Another BH3-only protein Bmf was not phosphorylated. The results show that BimEL is phosphorylated by ERK in vitro and that Ser⁶⁹ is the sole ERK phosphorylation site.

The proteins were also subjected to JNK and p38MAPK reactions (Fig. 6B). All proteins including the S69A mutant and Bmf were phosphorylated, indicating that these two kinases phosphorylate residues not phosphorylated by ERK. JNK has been reported to phosphorylate BimL on Thr⁵⁶ (31), which corresponds to Thr¹¹⁶ in BimEL. The phosphorylation level of the S69A mutant was somewhat lower than the wild-type or other mutant BimEL proteins. This suggests that, although not a major site, Ser⁶⁹ can also be phosphorylated by JNK and p38MAPK at least in vitro.

To examine whether Ser⁶⁹ is actually phosphorylated in cells, we raised an antibody against a peptide surrounding phospho-Ser⁶⁹. This antibody detected recombinant BimEL only when phosphorylated by ERK (Fig. 6C). For detection of Ser⁶⁹ phosphorylation of endogenous BimEL, we used MCF7 because it expressed the highest level of BimEL among the breast cancer cells examined (Fig. 4). Because phosphorylated/activated ERK was undetectable in MCF7, cells were stimulated with phorbol 12-myristate 13-acetate (PMA) to activate ERK (Fig. 7A). The samples were subjected to immunoprecipitation with cross-linked anti-Bim antibody followed by immunoblotting with antibodies against phospho-BimEL(Ser⁶⁹) and Bim. As shown in Fig. 7B, PMA induced phosphorylation of BimEL on Ser⁶⁹, which accompanied its mobility shift. U0126 completely blocked activation of ERK (Fig. 7A) and phosphorylation of BimEL (Fig. 7B). Subsequent analyses identified the presence of ERK in the immunoprecipitates (Fig. 7B). ERK was detected in samples from PMA-stimulated and unstimulated and U0126-treated cells, regardless of its activation state.

To further confirm phosphorylation of BimEL on Ser⁶⁹ by ERK in cells, HEK293T cells were transfected with NH₂-terminal polyhistidine-tagged BimEL or BimEL(S69A) with or without active MEK. Because wild-type BimEL could not be expressed to a level sufficient for immunoblot analysis, two conserved residues in the BH3 domain (L152 and D157) were mutated to alanines to reduce apoptotic toxicity. The basal ERK activity of HEK293T cells was also low, and phosphorylated ERK was detected only when active MEK was transfected (Fig. 8A). The exogenously expressed BimEL proteins were collected on a chelating resin and subjected to immunoblotting. As shown in Fig. 8B, phosphorylated BimEL was detected using phospho-BimEL(Ser⁶⁹) antibody when control BimEL (BimEL without S69 mutation) was transfected together with active MEK. No signal was observed in samples without cotransfection of active MEK or in samples transfected with the S69A mutant.

Reprobing of the filter with antibody against Bim showed that expression of active MEK caused a marked mobility shift of control BimEL, whereas the shift of BimEL(S69A) was modest (Fig. 8C). The position of the slower migrating form matched that of the band detected by the anti-phospho-Bim antibody. Longer exposure of the filter probed with anti-Bim antibody revealed the presence of a series of higher-molecular-weight bands or smear, indicative of polyubiquitination. Again, these were detected only when control BimEL was expressed together with active MEK and were not seen without MEK or when the putative ERK phosphorylation site was mutated. The filter was further probed with anti-ubiquitin antibody (Fig. 8D), and the result indicated that the higher-molecular-weight bands and smear are actually polyubiquitinated BimEL (Fig. 8C).

Wild-type BimEL and BimEL(S69A) were also transiently expressed in HBC4 and MDA-MB231 cells and analyzed as above (Fig. 9). The level of expression was not sufficient for direct detection of phosphorylated BimEL using the antibody against phospho-BimEL(Ser⁶⁹). However, in the immunoblot analysis using anti-Bim antibody, wild-type BimEL resolved as a doublet with an additional band with decreased electrophoretic mobility, whereas the S69A mutant migrated as a single band, suggesting that Ser⁶⁹ is also phosphorylated in these cells. In addition, the S69A mutant was expressed at a higher level, indicating that this mutant is more stable in cells.

The results obtained strongly suggested that BimEL is phosphorylated on Ser⁶⁹ by ERK and that the phosphorylation cues its polyubiquitination, which would lead to degradation of the apoptotic protein. Recent reports also indicate that polyubiquitination of BimEL is promoted by phosphorylation on Ser⁶⁹ (32, 33). To obtain additional evidence that Ser⁶⁹ phosphorylation signals polyubiquitination, recombinant BimEL and BimEL(S69A) were subjected to in vitro ubiquitination assay using MDA-MB231 extract. ERK is constitutively activated in MDA-MB231 cells (34), and incubation of BimEL with MDA-MB231 extract supplemented with an ATP-regenerating system resulted in phosphorylation of BimEL on Ser⁶⁹ (data not shown). As shown in Fig. 10, when ubiquitin was added to the mixture, a high-molecular-weight smear was detected in the wild-type but not in the S69A mutant BimEL sample, further supporting that phosphorylation of BimEL on Ser⁶⁹ triggers its polyubiquitination.

The major conclusion from the present study is that the main mechanism for induction of anoikis sensitivity by MEK inhibitors in MDA-MB231 and HBC4 cells is increase in BimEL level due to inhibition of phosphorylation-dependent degradation. Elusion of anoikis is believed to be a requirement for malignant growth in various cancers, although the mechanisms by which cancers acquire resistance to detachment-induced cell death may vary greatly. Agents that reestablish anoikis sensitivity can be envisaged to selectively inhibit aberrant survival and proliferation of neoplasms at ectopic locations and have significant clinical benefit. Our studies here suggest that MEK inhibitors have the potential to develop into this class of promising therapeutic strategy.

In this study, we provided direct evidence using phosphospecific antibody that Ser⁶⁹ is actually phosphorylated in cells. Bim has been shown to be a phosphoprotein in various cellular settings (15, 30–32, 35–38), and while this article was in preparation, several groups reported that phosphorylation of BimEL on Ser⁶⁹ by activation of the MEK-ERK pathway promotes its proteasome-mediated degradation (28, 29, 32, 33, 36). Our results are in accord with those reports. Although Ser⁶⁹ seemed to be the major phosphorylation site and ERK seemed to be the only BimEL kinase in HBC4 and MDA-MB231, lines of evidence show the existence of other phosphorylation sites. BimL was phosphorylated by JNK on Thr⁵⁶ (Thr¹¹⁶ in BimEL) and also on Ser⁴⁴ and/or Ser⁵⁸ (31). JNK phosphorylation on Thr⁵⁶ released BimL from the dynein motor complex and increased its apoptotic activity. We observed that p38MAPK also phosphorylated BimEL and BimL in vitro. As shown in Fig. 2, BimEL in MDA-MB453 and SKBR3 cells resolved as multiple bands. In HEK293T cells, cotransfection of BimEL(S69A) and active MEK resulted in some electrophoretic mobility retardation, and recombinant BimEL(S69A) mutant protein incubated with cell extracts also resolved as multiple bands (data not shown). All of these results are indicative of phosphorylation on other residues by other kinases. Sites of Bim phosphorylation and responsible kinases may differ with cells and with the cellular environment or the types of signal sensed. It is also possible that the consequences of phosphorylation may differ with cell types, and further work remains to fully establish the biological significance of Bim phosphorylation.

In MCF7 cells, detachment activated Bmf but not Bim (8). Although recent reports implied the involvement of Bim in anoikis (27, 28), Bim had not been shown to be released from the cytoskeleton on detachment. The results presented here indicate that, in HBC4 and MDA-MB231 cells, BimEL is activated and translocated to mitochondria on loss of adherence, which explains the reason why MEK inhibitors induce apoptosis in these cells only when deprived of anchorage. The precise nature of different Bim behavior is yet to be determined, but as shown in Fig. 4, the level of Bim differed surprisingly from cell to cell. HBC4 and MDA-MB231 expressed an exceedingly low level of Bim, and MCF7 expressed an exceedingly high level of Bim. Quantitative analysis of the immunoblot shown in Fig. 4 suggested that the amounts of Bim proteins in MCF7 and MDA-MB231 cells differ by at least 3 orders of magnitude, and it is conceivable that these two cancer cells suppress Bim activity by different mechanisms. We speculate that, in HBC4 and MDA-MB231, BimEL is activated by detachment but phosphorylation by ERK promotes its proteasomal degradation and protects cells from apoptotic death.

Our preliminary observations suggest that a phosphorylation-initiated BimEL degradation system also exists in MCF7 and other cancer cells. However, cells such as MCF7 most likely have acquired mechanisms to tolerate a high level of Bim, and additional increase might not impact cell survival. It is apparent that different strategies should be employed for treatment of cancers that counter Bim action with different mechanisms. Experiments with knockout mice have shown that loss of Bim impairs cytotoxic response to microtubule perturbation (39). Conversely, cells with elevated Bim expression can be expected to have high sensitivity to drugs that target microtubules. Actually, it was reported that MCF7 cells are more susceptible to paclitaxel than MDA-MB231, and the difference in sensitivity was attributed to Bim (26). Paclitaxel has been reported to activate ERK, and MEK inhibitors enhanced the toxicity of paclitaxel (40–42). Bim may in part explain this synergy: paclitaxel frees sequestered Bim from the microtubules, but the concomitant activation of ERK compromises the apoptotic activity of released BimEL by promoting its phosphorylation and degradation. MEK inhibitors prevent the decrease of activated BimEL and enhance paclitaxel toxicity.

Molecular target drugs pinpoint specific pathways involved in malignancy and have a theoretical advantage over conventional cytotoxic compounds. However, to fully exploit the potential of molecular targeted therapies in clinical situations, the prognosis of susceptible tumors and responding patients would be critical. In the case of MEK inhibitors, tumors exhibiting high ERK activity were more susceptible, and it was proposed that assays to monitor ERK activation using phosphospecific antibodies could be exploited as tools for prognosis as well as assessment of target inhibition (14). We extended our study to a large panel of human cancer cells and observed the tendency that cells vulnerable to MEK inhibitors have high levels of ERK phosphorylation. However, the correlation was not perfect, indicating that the degree of dependence on ERK for survival does not necessarily parallel ERK activity. On the contrary, thus far, we have observed superior association between induction of BimEL and sensitization to anoikis. Although measurement of ERK phosphorylation undoubtedly will be the prime assay to predict and monitor the efficacy of MEK inhibitors, Bim may also serve as an additional marker.

In conclusion, we identified BimEL as an important determinant for induction of anoikis sensitivity by MEK inhibitors. Our results also imply a role for BimEL as a biochemical marker for prognosis and diagnosis of the efficacy of MEK inhibitors.

We thank Yuusuke Zakabi and Mari Fukuyama for technical assistance.