|Home | About | Journals | Submit | Contact Us | Français|
Bcl-2 can be converted into a pro-apoptotic molecule by nuclear receptor Nur77. However, the development of Bcl-2 converters as anti-cancer therapeutics has not been explored. Here we report the identification of a Nur77-derived Bcl-2 converting peptide with 9 amino acids (NuBCP-9) and its enantiomer, which induce apoptosis of cancer cells in vitro and in animals. The apoptotic effect of NuBCPs and their activation of Bax are not inhibited but rather potentiated by Bcl-2. NuBCP-9 enantiomers bind to the Bcl-2 loop, which shares the characteristics of structurally adaptable regions with many cancer-associated and signaling proteins. NuBCP-9s act as molecular switches to dislodge the Bcl-2 BH4 domain, exposing its BH3 domain, which in turn blocks the activity of anti-apoptotic Bcl-XL.
We report that a Nur77-based peptide and its enantiomer bind the Bcl-2 loop, converting Bcl-2 from a protector to a killer of cancer cells. Our results provide mechanistic insight into Bcl-2 conversion and identify a new direction for Bcl-2-based drug leads and cancer drug development. These findings should appeal to a broad audience of clinicians, cancer biologists, molecular biologists and drug designers who have followed the therapeutic approaches targeting Bcl-2 and the underlying molecular mechanisms. In addition, peptide and protein chemists and structural biologists will be intrigued by the activity of an enantiomer, which has important mechanistic implications with broad potential for cancer therapeutics.
Members of the Bcl-2 family are critical regulators of apoptosis, an important biological process that eliminates cells with increased malignant potential such as those with damaged DNA or aberrant cell cycling (Cory et al., 2003; Green and Kroemer, 2004; Reed, 1998; Vander Heiden and Thompson, 1999). They possess at least one of the four conserved motifs called Bcl-2 homology (BH) domains. The family is divided into three subclasses based on numbers of BH domains and function: the anti-apoptotics, including Bcl-2 and Bcl-XL, possess sequence conservation through BH1-4, and pro-apoptotics, which are further divided into multidomain members such as Bax and Bak possessing BH1-3, and BH3-only molecules such as Bid, Bim and Bad (Cory et al., 2003; Green and Kroemer, 2004; Reed, 1998; Vander Heiden and Thompson, 1999). They regulate apoptosis through interactions between the pro-apoptotic and antiapoptotic Bcl-2 family members. BH3-only proteins convey diverse death signals by directly or indirectly activating Bax and/or Bak, which can induce permeabilization of the outer mitochondrial membrane and release apoptogenic factors needed to activate the caspases (Kuwana et al., 2005; Leber et al., 2007; Letai et al., 2002; Willis and Adams, 2005). Anti-apoptotic family members inhibit death by restraining Bax and Bak activity and/or by sequestering BH3-only members. Recently, approaches targeting pro-survival Bcl-2 family members, such as BH3 domain-derived peptides or chemical inhibitors such as ABT-737 (Bouillet and Strasser, 2002; Degterev et al., 2001; Oltersdorf et al., 2005; Reed, 2002; Walensky et al., 2004), are being developed, which show significant anti-cancer activities. These BH3 peptides and chemical inhibitors act by binding to the hydrophobic groove formed by the BH1-3 domains of the pro-survival proteins and antagonizing their survival function, resulting in release of pro-apoptotic members that activate apoptosis.
The functional phenotype of some Bcl-2 family members such as Bcl-2 can be reversed in some cellular contexts. For example, mutants of the Bcl-2-homolog, Ced-9, appear to promote rather than prevent apoptosis in C. elegans (Xue and Horvitz, 1997). Likewise, Bcl-2 homologs in Drosophila can manifest either cytoprotective or cytodestructive phenotypes, depending on cellular context (Colussi et al., 2000; Igaki et al., 2000). The mechanisms responsible for the phenotypic conversion of Bcl-2 are largely undefined. However, the unstructured loop of Bcl-2, which links the BH3 and BH4 domains, appears important (Moll et al., 2006; Zhang, 2007). When the Bcl-2 loop is cleaved by caspase-3, Bcl-2 is converted to a pro-apoptotic protein similar to Bax (Cheng et al., 1997; Grandgirard et al., 1998). Phosphorylation of the loop has also been speculated to convert Bcl-2 to a pro-apoptotic form (Blagosklonny, 2001). It inhibits binding of Bcl-2 to multidomain and BH3-only pro-apoptotic family members (Bassik et al., 2004), and autophagic protein Beclin 1 (Wei et al., 2008). We recently reported that nuclear receptor Nur77 converts Bcl-2 into a killer by binding its loop (Lin et al., 2004). Nur77 (also called TR3 or NGFI-B) is a potent pro-apoptotic member of the nuclear receptor superfamily (Moll et al., 2006; Zhang, 2007). It often translocates from the nucleus to mitochondria in response to different death signals, where it binds Bcl-2 inducing a conformational change (Li et al., 2000; Lin et al., 2004). Nur77 translocation to mitochondria and its induction of a Bcl-2 conformational change has also been implicated in the negative selection of thymocytes in vitro and in animals (Thompson and Winoto, 2008), indicating a physiological role for the Nur77-Bcl-2 interaction. Interestingly, p53 also binds the Bcl-2 loop (Moll et al., 2006) which then acts like a BH3-only protein to activate Bax or Bak by releasing BH3-only proteins like Bid (Chipuk et al., 2005; Leu et al., 2004; Mihara et al., 2003).
The apoptotic effect of Nur77 appears to be clinically relevant, as the expression of the Nur77 subfamily member Nor-1 is positively correlated with survival of diffuse large B-cell lymphoma patients (Shipp et al., 2002) and Nur77 downregulation is associated with metastasis of several primary solid tumors (Ramaswamy et al., 2003). Thus, targeting the Nur77-Bcl-2 apoptotic pathway is an attractive approach for developing cancer therapeutics. The ability of Nur77 to convert Bcl-2 distinguishes this death protein from pro-apoptotic Bcl-2 family proteins, whose activities are restrained by pro-survival Bcl-2 family members. It also offers an opportunity to design drugs, which are likely to be effective against cancer cells with high Bcl-2 levels. Here, we report the identification of a short Nur77-derived peptide and its enantiomer that act as molecular switches to induce a Bcl-2 conformational change, converting it from a protector to a killer of cancer cells in vitro and in animals.
To develop Bcl-2 converters that induce apoptosis in cancer cells with high Bcl-2 levels, peptides corresponding to subregions of a Nur77 fragment, known to interact with Bcl-2 (Lin et al., 2004), were synthesized and conjugated with the cell-penetrating-peptide (CPP), D-Arg octamer (r8) (Jones et al., 2005) (Figure 1A). Amongst these, a 20 amino acid peptide (aa 480 to 499) (NuBCP-20) exhibited potent apoptotic effect in various cancer cell lines, being more potent than a BH3 peptide derived from the pro-apoptotic Bcl-2 family member Bid (Letai et al., 2002) (Figures S1 and S2). Serial deletion identified a 9-mer Nur77 peptide, NuBCP-9, which effectively induced apoptosis of breast cancer cells (Figure 1B and Figure S2). Neither NuBCP-20 nor NuBCP-9 induced the apoptosis of normal primary mammary epithelial cells (Figure 1B and Figure S1). NuBCP-9 is considerably shorter than the shortest BH3 peptide that binds Bcl-2. Further deletion from either its N-terminal or C-terminal end or substituting Ala for NuBCP-9 terminal amino acids (NuBCP-9/AA) completely abolished its apoptotic effect (Figure S2). NuBCP-9 linked to other CPPs, penetratin or transportan-10 (Jones et al., 2005) as well as r8 by a disulfide bond showed a similar degree of apoptosis (Figure S2). Since disulfide bonds are rapidly reduced in cells, the apoptotic effect of NuBCP-9 is not due to its linkage with CPPs.
While investigating the sequence requirements for NuBCP-9, we discovered that replacing L-amino acids with D-amino acids did not diminish its death effect. Peptide enantiomers have been reported to interact with several proteins (Zhou et al., 2002), including calmodulin, α3β1 integrin, DnaJ (Hsp40) co-chaperone and CXCR4, which like Bcl-2 are characterized by large natively disordered loops. In contrast, the enantiomer of Bad-BH3 peptide was not apoptotic (Figure 1C), demonstrating different mode of action of the NuBCP-9 and Bad-BH3 peptides.
We next examined whether NuBCP-9-induced apoptosis was dependent on Bcl-2 expression. NuBCP-9 and its enantiomer showed little effect on Jurkat cells (<20% apoptosis). However, both peptides induced extensive apoptosis (>50%) in Jurkat cells stably expressing Bcl-2 (Figure 1D). In contrast, apoptosis induced by staurosporine (STS) and Bid BH3 peptide was attenuated by Bcl-2 overexpression in Jurkat cells (Figure S3), demonstrating the dual role of Bcl-2. Stable expression of Bcl-2 in CEM leukemia cells also potentiated the death effect of Nur77 peptide but prevented the killing by STS and Bad BH3 peptide (Figure S4). The apoptotic effect of NuBCP-9 was further evaluated in mouse embryonic fibroblasts (MEFs) and Bcl-2 knockout MEFs (Bcl-2−/−MEFs). Although the death effect of STS and Bad BH3 peptide was enhanced in Bcl-2−/−MEFs (Figure 1E), NuBCP-9 and D-NuBCP-9 induced apoptosis of MEFs but not Bcl-2−/−MEFs in a dose (Figure 1F) and time (Figure 1G) dependent manner. Further, suppression of Bcl-2 expression by siRNA or antisense oligonucleotides reduced the killing effect of NuBCP-9 enantiomers (Figures S5–S6). Thus, Bcl-2 is a major target of NuBCP-9s. NuBCP-induced apoptosis requires Bax or Bak, as the peptides similarly induced apoptosis of wild-type, Bax−/−, and Bak−/− MEFs, but lacked activity in double knockout Bax−/−Bak−/−MEFs (Figure 1H), further demonstrating that the peptides act via the Bcl-2-regulated pathway.
The effects of Nur77 peptides on clonogenic survival of MEFs were determined. After exposure to NuBCP-9 or D-NuBCP-9 peptide, MEFs formed very few colonies compared to Bcl-2−/−MEFs (Figure 2A). For example, MEFs treated with 10 µM D-NuBCP-9 formed only 5% of the colonies compared to Bcl-2−/−MEFs. The suppressive effect of NuBCP-9 (Figure 2B) or D-NuBCP-9 (Figure 2C) was also largely reduced in Bax−/−Bak−/−MEFs. Treatment with 15 µM D-NuBCP-9 resulted in about 80% reduction of colonies in MEFs but did not decrease the number of colonies formed by Bax−/−Bak−/− MEFs (Figure 2C). To further evaluate NuBCPs, their effects on the growth of tumors formed in SCID mice were examined. Injection of L- or D-NuBCP-9 but not control peptide (NuBCP-9/AA) dramatically suppressed the growth of MDA-MB435 cancer cell xenografts in mice (Figure 2D and Figure S2C), and potently induced apoptosis of tumor cells, revealed by TUNEL staining (Figures 2E–F). Furthermore, D-NuBCP-9 induced the regression of tumors (Figure 2G). Thus NuBCP-9 and its enantiomer effectively induce apoptosis in vitro and in animals, demonstrating their therapeutic potential.
To determine whether NuBCPs bind Bcl-2, cDNA encoding the residues 478–504 (Nur77/478-504) and 489–497 (Nur77/489-497, equivalent to NuBCP-9) of Nur77 were fused with cDNA of green fluorescence protein (GFP). When transfected into HEK293T cells, GFP-fused Nur77 fragments were precipitated by anti-Bcl-2 antibody only when Bcl-2 was co-expressed (Figure 3A). The co-precipitation was inhibited by addition of NuBCP-9, but not Smac-peptide (not shown). To study whether D-NuBCP-9 interacted with Bcl-2, we used a competition assay. Nur77 lacking its DNA-binding-domain (DBD), Nur77/ΔDBD, bound Bcl-2 (Lin et al., 2004), and binding was inhibited by NuBCP-9 or D-NuBCP-9 (Figure 3B). GFP-Nur77/478-504 also bound to anti-apoptotic Bcl-2 family members, Bcl-B and Bfl-1, but not Bcl-XL and Mcl-1 (Figure 3C). Like Nur77 protein (Luciano et al., 2007), the killing effect of NuBCP-9 was enhanced by overexpression of Bcl-B in HeLa cells and inhibited by suppression of Bcl-B expression by siRNA in H460 cells (Figure S6). These results suggest that NuBCP-9 also converts Bcl-B into a pro-apoptotic molecule.
We next used fluorescence polarization (FP) assays to determine whether NuBCP-9s bind Bcl-2. GST-Bcl-2, but neither GST-Bcl-XL nor GST, induced a concentration-dependent FP of FITC-NuBCP-9 and FITC-D-NuBCP-9, while the FP of FITC-NuBCP-9/AA was little affected (Figure 3D). In addition, unconjugated NuBCP-9 and D-NuBCP-9 competed with binding of FITC-L-NuBCP-9 or FITC-D-NuBCP-9 to GST-Bcl-2, whereas NuBCP-9/AA did not (Figure 3E). Thus, NuBCP-9 and D-NuBCP-9 bind Bcl-2 directly and competitively. The Bcl-2 domain targeted by NuBCPs is distinct from that targeted by BH3 peptides, as shown by the failure of either a BH3 peptide (Bak BH3) or a potent small molecule inhibitor (ABT-737) that targets the Bcl-2 BH3- binding site to reduce FITC-NuBCP-9 binding to GST-Bcl-2 (Figure 3E).
The GFP-Nur77/489-497 and GFP-Nur77/478-504 fusions co-localized extensively with RFP-Mito, a red fluorescence protein (RFP) fused to a mitochondria-targeting sequence (Figure 3F). FITC-D-NuBCP-9-r8 also displayed significant co-localization with RFP-Mito. Thus, NuBCP-9 and its enantiomer, as Nur77, bind Bcl-2 and target mitochondria.
Conversion of Bcl-2 involves a conformational change that exposes its BH3 domain (Lin et al., 2004). BH3 domain exposure is detectable with antibody against a Bcl-2 BH3 peptide (Figure S7). The effect of NuBCPs on Bcl-2 conformation was examined by immunoprecipitation assays (Figure 4A). The Bcl-2 BH3 domain antibody precipitated endogenous Bcl-2 in cells treated with NuBCP-9 or D-NuBCP-9, but not NuBCP-9/AA, showing that NuBCP-9s induced a Bcl-2 conformational change. In contrast, tBid-BH3 and Bad-BH3 peptides did not induce this effect, despite their induction of apoptosis. This was confirmed by flow cytometry analysis, showing a strong enhancement in fluorescence upon staining cells with BH3 antibody exposed to NuBCP-9s, but not NuBCP-9/AA, tBid-BH3, or Smac peptide (Figure 4B). NuBCP-9-induced Bcl-2 BH3 domain immunofluorescence was observed in the presence of the caspase inhibitor zVAD, excluding the involvement of caspases in the Bcl-2 conformational change (Figure S7D). The NuBCP-induced Bcl-2 conformational change was also observed in solid tumors (Figure 2F). The overlapping of TUNEL staining with Bcl-2 BH3 immunofluorescence in tumor treated with NuBCP-9 (Figure 2F) is consistent with the notion that NuBCP-9-induced Bcl-2-dependent apoptosis is correlated with the exposure of Bcl-2 BH3 domain.
We next determined whether NuBCPs could induce a conformational change of purified GST-Bcl-2 protein using circular dichroism (CD) analysis. Our results showed similar changes in GST-Bcl-2 protein spectra, when it was incubated with NuBCP-9 or D-NuBCP-9, but not NuBCP-9/AA (Figure 4C). The fact that NuBCP-9 and D-NuBCP-9, display mirror image spectra (Figures S8–S10), while inducing identical changes in the Bcl-2 spectra (Figure 4C and Figure S8), indicates that NuBCP-9 and D-NuBCP-9 do not contribute significantly to the spectral changes. Binding is saturable and stoichiometric with a Kd of 2.1 ± 0.2 µM for NuBCP-9 and 2.0 ± 0.1 µM for D-NuBCP-9 (Figures 4D–E), in agreement with FP assays (Figure 3E). In contrast, NuBCP-9, D-NuBCP-9 and NuBCP-9/AA had no effect on CD spectra for GST or GST-Bcl-XL (Figures S9–S10). Thus, the NuBCP-induced Bcl-2 conformational change observed in cells can be accounted for by direct binding of NuBCPs to Bcl-2 in a specific 1:1 complex.
The observations that NuBCP-9 and its enantiomer exhibited similar if not identical effects on Bcl-2 function raised the possibility that their binding interface(s) are structurally adaptive. The large, regulatory loop of Bcl-2 is predicted to be natively unstructured (Figure S11) and like other loops of this class may be structurally adaptive (Dyson and Wright, 2005; Iakoucheva et al., 2002). Cell-based Co-IP (Figure 5A) showed that Bcl-2 N-terminal fragment containing the BH4 domain and loop, Bcl-2/1-90, interacted with Nur77 mutants (Nur77/DC3, Nur77/DC1, and Nur77/478-504) known to bind Bcl-2. In vitro, NuBCP-9 and its enantiomer bound similarly and competitively to GST-Bcl-2/1-90, revealed by FP assays (Figure 5B). CD analysis showed that both NuBCP-9s induced similar changes in GST-Bcl-2/1-90 spectra (Figure 5C and Figure S12). The binding affinities of NuBCP-9s for Bcl-2/1-90 (Kd=1.7 µM for NuBCP-9; Kd=2.5 µM for D-NuBCP-9) (Figure 4E and Figure 5B) are similar to the affinities for Bcl-2 (Figure 4E), as are the stoichiometries (Figure 4D, 4E and Figure 5D). Thus, Bcl-2/1-90 retained the ability of Bcl-2 to bind to NuBCP-9s, excluding the involvement of the hydrophobic groove in Bcl-2, consistent with the inability of BH3 peptide and ABT-737 to compete with the binding of NuBCP-9 to Bcl-2 (Figure 3E). We next examined whether the Bcl-2 loop alone was capable of binding to NuBCP-9s. The Myc-tagged Bcl-2 loop, Myc-Bcl-2/29-90, interacted with GFP-Nur77/DC3 (Figure 5E). The interaction was inhibited by NuBCP-9 or D-NuBCP-9, but not by NuBCP-9/AA (Figure 5F). Mutations of Thr69 and Ser70 in the loop slightly enhanced its interaction with Nur77/DC3, while insertion of 10 amino acids in the loop largely impaired the interaction (Figure 5G). Strong support was provided by FP assays, showing that both peptides bound to GST-Bcl-2/29-90 protein (Figure 5H). Also, NuBCP-9 and its enantiomer but not the mutant peptide competed with FITC-NuBCP-9 for binding to GST-Bcl-2/29-90. Further, CD analysis showed a similar change of the spectra of GST-Bcl-2/29-90 protein by NuBCP-9 and its enantiomer (Figure 5I and Figure S13). Bcl-2 and Bcl-2/1-90 underwent greater NuBCP-induced changes in their CD spectra than Bcl-2/29-90, suggesting involvement of the BH4 domain. Thus, NuBCP-9 and its enantiomer bind the Bcl-2 loop.
Our observation that NuBCP-9-induced apoptosis was dependent on Bax and/or Bak prompted us to investigate whether and how NuBCP-9 activated Bax. In vitro assays using isolated mitochondria showed that both NuBCP-9 and D-NuBCP-9 induced Bax dimerization, trimerization and especially oligomerization (Figure 6A) in a concentration dependent manner (Figure 6B). Such an effect occurred only in the presence of GST-Bcl-2 protein. In DoHH2 lymphoma cells that express high levels of Bcl-2 (Dyer et al., 1996), NuBCP-9 induced Bax dimerization/oligomerization (Figure 6C). Transfection of Bcl-2/1-95 inhibited both NuBCP-9-induced apoptosis (Figure 6D) and Bax activation (Figure 6E). NuBCP-9 also induced Bax activation in H460 cells revealed by flow cytometric analysis of Bax immunostaining with anti-Bax antibody (6A7) that recognizes active Bax conformation (Nechushtan et al., 1999) (Figure 6F). Similar to its effect in DoHH2 cells, Bcl-2/1-95 potently inhibited NuBCP-9-induced apoptosis (Figure 6G) and Bax activation in H460 cells (Figure 6H). Furthermore, expression of Bcl-2/ΔBH3, Bcl-2 mutant lacking its BH3 domain, inhibited NuBCP-9-induced apoptosis (Figure 6G) and Bax activation (Figure 6H). In MEF cells, NuBCP-9-induced Bax activation was observed in the wild-type but not Bcl-2 knockout cells (Figure 6I). However, transfection of Bcl-2 into Bcl-2−/−MEF restored the ability of NuBCP to activate Bax (Figure 6J). Together, these cell-free and cell-based studies demonstrate that NuBCP-9 induces Bax activation in a Bcl-2 dependent manner and that the loop and BH3 regions of Bcl-2 are crucial.
To study the mechanism by which enantiomeric NuBCPs induced Bcl-2 conformational change, we first examined how the anti-apoptotic Bcl-2 conformation was maintained and found that the Bcl-2 BH4 domain could act like a brace to stabilize the C-terminal anti-apoptotic BH3-binding pocket. Co-IP revealed that a Bcl-2 N-terminal sequence containing the BH4 domain bound a Bcl-2 mutant from which the BH4 domain was removed (Bcl-2/ΔBH4) (Figure 7A), suggesting an intra-molecular interaction between the BH4 domain and the C-terminal region. The BH4 domain could not bind full-length Bcl-2, likely due to inaccessibility of a BH4-binding site in the C-terminal region. However, a strong interaction was observed when Nur77/ΔDBD or Nur77/DC3 was coexpressed (Figure 7B), suggesting that binding of Nur77 mutants with Bcl-2 reorganized Bcl-2, leading to the exposure of the BH4-binding site in the C-terminal region. Similarly, addition of NuBCP-9 and its enantiomer induced the binding of BH4 domain to Bcl-2 (Figure 7C). Removal of the BH4 domain through caspase cleavage of Bcl-2 loop converts it into a death molecule (Cheng et al., 1997; Grandgirard et al., 1998). Consistently, a Bcl-2 mutant lacking its BH4 domain was extensively immunostained by anti-Bcl-2/BH3 antibody, while the wild-type Bcl-2 protein was not, indicating that the exposure of the BH3 epitope was blocked directly or indirectly by the BH4 domain (Figure 7D). Thus, our data unravel a mechanism of Bcl-2 conversion, in which binding of NuBCP-9 or its enantiomer to Bcl-2 loop dislodges its BH4 domain, leading to a pro-apoptotic conformation that exposes the BH3 domain.
We next determined how NuBCP-9-induced changes in Bcl-2 conformation resulted in Bax activation and apoptosis. One way that Bcl-2 prevents death is by sequestering activator BH3-only family members (such as Bid) and preventing their interaction with Bax/Bak (Cheng et al., 2001; Kuwana et al., 2005; Letai et al., 2002). We recently showed that the interaction between membrane-bound Bcl-2 and tBid resulted in a conformational alteration in the Bcl-2, which induced permeabilization of liposomal membrane to 0.5-kDa fluorescent dye, Cascade Blue (CB) (Peng et al., 2006). Unlike the interaction of tBid with Bax, the pore sizes in liposomal membranes, which result from the interaction of tBid with Bcl-2, are relatively small, allowing the release of CB but not CB-labeled 10 kDa dextrans. Although the physiological significance of the tBid-induced Bcl-2 membrane-permeabilizing activity is largely elusive, it offered an opportunity to study the effect of NuBCP-9 on tBid/Bcl-2 interaction in liposomes. When NuBCP-9 was added with Bcl-2 and tBid to liposomes, NuBCP-9 but not NuBCP-9/AA inhibited the membrane permeabilization induced by tBid/Bcl-2 interaction in a dose dependent manner (Figure 7E). NuBCP-9, at 1 µM, was sufficient for inhibition, which could not be further increased by using 10 µM peptide. This result suggests that 10 µM is a saturating peptide concentration, consistent with our binding studies (Figure 4E). NuBCP-9 alone had no effect on the membrane permeability even at the highest dose (10 µM). Thus, NuBCP-9 can inhibit the interaction of Bcl-2 with an activator BH3-only protein, which in turn may activate Bax.
Some BH3-only proteins including Bid and Bim (termed activators) induce Bax/Bak oligomerization through their direct interaction with Bax/Bak (Cheng et al., 2001; Kuwana et al., 2005; Letai et al., 2002). We then tested if the NuBCP-9-converted Bcl-2 could also function as a direct activator of Bax. As shown in Figure 7F, while adding NuBCP-9 even at a saturating concentration of 10 µM to 50 nM Bcl-2 induced a marginal but reproducible membrane permeabilization of liposomes, the further addition of 50 nM Bax failed to increase the membrane permeability. In contrast, 5 nM tBid strongly induced membrane permeabilization by 50 nM Bax, releasing a cytochrome c surrogate, 10-kDa CB-dextran (Figure 8E). Thus, NuBCP-9 does not induce Bcl-2 to directly activate Bax.
To study whether NuBCP-9 induced exposure of BH3 domain in Bcl-2 acts indirectly to induce apoptosis by competing with Bax or Bak for the BH3 binding pockets of anti-apoptotic Bcl-2 family members, we mutated two highly conserved BH3 residues (Leu97 and Asp102) in Bcl-2 (Figure 8A) that correspond with residues in the Bak BH3 peptide that are required for displacing BH3 proteins and peptides from the BH3 binding pockets of anti-apoptotic Bcl-2 family members (Sattler et al., 1997). The Bcl-2 double mutant (Bcl-2/L97A/D102A) acted dominant-negatively, suppressing Nur77 peptide-induced apoptosis (Figure 8B). We then tested whether the Bcl-2 BH3 peptide could directly inhibit Bcl-XL blocking of tBid-induced Bax permeabilization of liposomes encapsulating the large 10-kDa CB-dextrans. As shown in Figure 8C, the inhibitory effect of Bcl-XL was reversed by the Bcl-2 BH3 in a dose-dependent manner, while the mutant BH3 peptide had no effect. Also, the Bcl-2 BH3 peptide alone did not induce Bax-dependent membrane permeabilization consistent with the observation that the addition of NuBCP-9 to Bcl-2 did not induce Bax-dependent membrane permeability (Figure 7F). Next, we showed using the FPA that F5M-Cys labeled Bcl-2 BH3 peptide but not the double mutant peptide bound to GST-Bcl-XL protein (Kd=144.6 +/− 11 nM) (Figure 8D). Binding of the Bcl-2 BH3 peptide to Bcl-XL confirmed a previous report (Sattler et al., 1997).
We next determined whether Bcl-2 in the presence of NuBCP-9 could mimic the inhibitory effect of the Bcl-2 BH3 peptide on the activity of Bcl-XL against tBid-activated Bax. A saturating concentration of NuBCP-9 (10 µM) was pre-incubated with Bcl-2 protein to insure full conversion before adding Bcl-XL. Figure 8E shows that Bcl-XL inhibition of tBid induced Bax permeablization was reversed by NuBCP-9 in a Bcl-2-dependent manner. In contrast, the addition of Bcl-2 to Bcl-XL further inhibited the tBid-induced Bax activity. Also, Figure 8E shows that similar to Bcl-XL, Bcl-2 alone inhibited the tBid/Bax-mediated membrane permeabilization. Unlike the Bcl-XL case, the Bcl-2 inhibition is reversed by the addition of NuBCP-9. We have tested extensively, using different ratios of Bcl-2/peptide or Bcl-2/Bax, and failed to observe full release of dye by NuBCP-9, suggesting that it cannot convert Bcl-2 to a direct activator of Bax. Together, our liposome results demonstrate that NuBCP-9-induced Bcl-2 conformational change not only neutralizes Bcl-2’s inhibition of Bax-mediated membrane permeabilization but also exposes the Bcl-2’s BH3 motif neutralizing Bcl-XL’s inhibition of Bax (Figure 8F).
Here we report the identification and characterization of a nine amino acid, Nur77 Bcl-2 converting peptide (NuBCP-9), which potently induces apoptosis through a pathway that is potentiated by Bcl-2 expression in vitro and in animals (Figure 1, Figure 2, Figure S3, and S4). Bcl-2 is an attractive drug target, because its levels are elevated in a majority of human cancers and correlate with the resistance of cancer cells to many chemotherapeutic drugs and γ-irradiation. From a therapeutic viewpoint, Bcl-2 overexpression may be advantageous because it distinguishes many cancer cells from normal cells. In this regard, several strategies taking advantage of this difference are currently being investigated, which may lead to improved cancer treatments. Currently, targeting Bcl-2 has mostly relied on antisense oligonucleotides that inhibit Bcl-2 expression or BH3 peptides and small molecule surrogates that bind the Bcl-2 BH3 binding pocket, antagonizing its anti-apoptotic (Bouillet and Strasser, 2002; Degterev et al., 2001; Oltersdorf et al., 2005; Reed, 2002; Walensky et al., 2004). Although it has been known for some time that Bcl-2 can be converted to a pro-apoptotic form using caspase-3 (Cheng et al., 1997; Grandgirard et al., 1998) or activating proteins including Nur77 (Lin et al., 2004), it has been unclear whether this conversion provides a basis for cancer drug development. NuBCP-9 acts by inducing a Bcl-2 conformational change raising prospects that NuBCP-9-based drugs and small molecule Bcl-2 converters might be developed for treating cancers with elevated levels of Bcl-2. The NuBCP-9 enantiomer is a D-peptide which are protease resistant (Zhou et al., 2002), an important consideration for peptide-based drug development, while short peptides are sometimes forerunners for small molecule drug development. Our observation that NuBCP-9 and its enantiomer effectively induced tumor regression in animals establishes them as potential therapeutic leads for the treatment of Bcl-2-overexpressing cancers.
Our data demonstrates that NuBCP-9 induces a Bcl-2 conformational change by binding the Bcl-2 loop to dislodge its BH4 domain (Figure 7A–C) which exposes the BH3 domain (Figure 4, Figure 7D and Figure S7). Our results provide further support for the Bcl-2 loop as a regulator of its activity. The Bcl-2 loop is predicted to be natively unstructured (Figure S11). Recent studies have shown that a large unstructured loop can bind different proteins by structural adaptation through coupled folding (Dyson and Wright, 2005). The majority of human cancer-associated and signaling proteins are predicted to have large, natively unstructured loops. that may account for their positions at the centers of many biological processes (Dyson and Wright, 2005; Li, 2005; Iakoucheva et al., 2002). The observation that both NuBCP-9 and its enantiomer bind the Bcl-2 loop may be a manifestation of an unstructured, conformationally adaptable loop. The Bcl-2 family of proteins is central to apoptosis. Thus it is not surprising that the Bcl-2 loop shares many of the characteristics of structurally adaptable regulatory loops, including its large size (69 residues), high proline content (22%), several phosphorylation and caspase cleavage sites and at least five different protein binding partners (Bruey et al., 2007; Deng et al., 2006; Kang et al., 2005; Lin et al., 2004; Ueno et al., 2000). Consequently, it might be expected that Bcl-2 conversion is subjected to multiple levels of regulation, depending on cell type and cellular environment. Of particular interest is that the Bcl-2 loop is enriched by proline residues, which are widely distributed in disordered regulatory loops of proteins from prokaryotes to eukaryotes and display promiscuity and versatility in protein-protein interactions (Li, 2005). Structural analysis of the Bcl-2/NuBCP complex will eventually resolve whether these proline-rich sequences are responsible for binding to NuBCP-9 and its enantiomer. As many human-cancer-associated and signaling proteins contain large, natively disordered regulatory loops (Dyson and Wright, 2005; Iakoucheva et al., 2002; Li, 2005), proteolytic stable D-peptides may provide a rich source for new drug leads
Induction of apoptosis by NuBCP-9 required expression of either Bax or Bak (Figure 1H) and was associated with their activation (Figure 6). However, the addition of NuBCP-9 to Bcl-2, unlike tBid, did not induce Bax-dependent permeabilization of mitochondria-outer-membrane liposomes (Figure 7F), arguing against a direct activation mechanism. Consistent with an indirect activation mechanism, we found that NuBCP-9 inhibited Bcl-2 interaction with tBid (Figure 7E), suggesting that NuBCP-9 may indirectly induce Bax activation by inhibiting Bcl-2 interaction with activator BH3-only proteins. Investigating this further, our liposome data showed that inhibition of tBid-activated Bax by Bcl-2 or Bcl-XL was reversed by NuBCP-9 (Figure 8E). Our results are reminiscent of previous studies showing that phosphorylation of the nonstructured loop prevented Bcl- 2 from binding to multidomain pro-apoptotic members (Bassik et al., 2004) and Beclin 1, a BH3-containing autophagic protein (Wei et al., 2008). Thus, NuBCP-9, similar to BH3 peptides or their small molecule surrogates, can prevent Bcl-2 from binding and sequestering pro-apoptotic Bcl-2 family members.
One unique property of NuBCP-9, which distinguished it from Bcl-2 inhibitors, is that NuBCP-9 not only antagonizes the survival function of Bcl-2 but also induces a Bcl-2 conformation that inhibits the survival function of its anti-apoptotic relatives, such as Bcl-XL (Figures 8C–E). Such an effect is likely mediated by its ability to induce exposure of the BH3 domain of Bcl-2 (Figure 4A–B, Figure 7D and Figure S7). Mutagenesis of the Bcl-2 BH3 domain showed that it is required for NuBCP-9-induced Bcl-2- dependent apoptosis. Thus Bcl-2 BH3 mutants acted dominant-negatively to inhibit NuBCP-9-induced Bcl-2- dependent apoptosis (Figure 6G and Figure 8B) and Bax activation (Figure 6H). Consistently, a peptide corresponding to the BH3 domain of Bcl-2 effectively neutralized the anti-Bax effect of Bcl-XL in liposome assays (Figure 8C). Similar to the Bcl-2 BH3 peptide, NuBCP-9 induced exposure of the Bcl-2 BH3 domain also neutralized the inhibitory effect of Bcl-XL on Bax activation (Figure 8E). Thus, NuBCP-9 is distinguished from Bcl-2 BH3 domain inhibitors in that it also converts Bcl-2 into a “BH3-like” molecule that in turn inhibits its anti-apoptotic relative Bcl-XL.
Peptides were synthesized as described in Supplemental data.
GST-Bcl-2, GST-Bcl-XL, or GST protein was briefly incubated with FITC or F5M-conjugated peptide with or without competitors in Greiner Fluotrac 600 96-well microplates. Fluorescence polarization was recorded using an Analyst HT 96–384 microplate reader (Molecular Devices, Sunnyvale, CA) with excitation wavelength set at 485 nm and dynamic polarizer for emission at 530 nm.
CD spectra were determined as described in the Supplemental data.
Nuclear morphological change analysis and flow cytometric analysis of Annexin V binding were described (Li et al., 2000; Lin et al., 2004). For the determination of DNA fragmentation in tumor tissue, the TUNEL assay was used. Peptide treatments were done in medium with 10% FBS unless otherwise specified.
For the Co-IP assay, HEK293T cells transfected with various expression vectors were incubated with appropriate antibody and immunoprecipitates determined by immunoblotting as described (Li et al., 2000; Lin et al., 2004). For all the cell-based experiments, peptides fused with cell-penetrating-peptide (r8) were used unless otherwise stated.
Mitochondria were prepared from HeLa cells as described (Zhai et al., 2005)
Cell-based and in vitro Bax activation assays are described in Supplemental data.
Liposomes of mitochondrial outer membrane (MOM) lipid composition and with Cascade Blue (CB) or CB-labeled 10-kDa dextran encapsulated were prepared by the extrusion method (Peng et al., 2006; Tan et al., 2006) and used for assaying Bcl-2 interactions with tBid, Bax, Bcl-XL and NuBCP-9 as described in Supplemental data.
The assays are described in Supplemental data.
Female SCID mice (6-week-old) (Tacomic) were injected with 106 MDA-MB435 cells. Tumors were palpable on day 7. On days 10 and 13, peptides (620 µg in 50 µL PBS) were injected into the tumor areas of 5 mice. Tumor volumes (l × w2)/2 were determined using calipers. No weight changes were observed. Established tumors in mice were injected with peptides and tumor tissues were excised and sectioned after 3 days. Tissues were fixed (10% buffered formalin), then rapidly paraffin-embedded. Apoptosis was detected by TUNEL assay. All procedures were performed according to protocols approved by the Institutional Animal Care and Use Committee of the Burnham Institute.
We thank S. Korsmeyer, S. Lowe and S. Weintraub for knock-out MEFs, Katherine Webster, Daniel Koch, Jun Peng, Danielle King and Zhi Zhang for excellent technical assistance, David Andrews for comments, and L. Frazer for preparation of the manuscript. This work is in part supported by grants to X-k. Zhang, A.C. Satterthwait, J.C. Reed, and J. Lin from the National Institute of Health (CA109345, GM060554, and GM062964), the US Army Medical Research and Material Command (PC073847), the California Tobacco-Related Diseases Research Program (CTRDRP), the California Breast Cancer Research Program (12IB-0168), the Susan G. Komen Breast Cancer Foundation, and the 985 Project of Xiamen University. SKK is supported by a new investigator award from CTRDRP.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.