Lipid-Dependent Bimodal MCL1 Membrane Activity - American

Oct 14, 2014 - subclasses based on functional criteria as well as on the presence of up to four BCL2 homology (BH) domains: (1). Multidomain BCL2-like...
0 downloads 0 Views 7MB Size
Articles pubs.acs.org/acschemicalbiology

Lipid-Dependent Bimodal MCL1 Membrane Activity Olatz Landeta,† Juan Garcia Valero,† Hector Flores-Romero, Itsasne Bustillo-Zabalbeitia, Ane Landajuela, Miguel Garcia-Porras, Oihana Terrones, and Gorka Basañez* Unidad de Biofisica, Centro Mixto Consejo Superior de Investigaciones Científicas (CSIC)-Euskal Herriko Unibertsitatea/Universidad del Pais Vasco (EHU/UPV), Barrio Sarriena s/n, Leioa 48940, Spain S Supporting Information *

ABSTRACT: Increasing evidence indicates that the mitochondrial lipid membrane environment directly modulates the BCL2 family protein function, but the underlying mechanisms are still poorly understood. Here, we used minimalistic reconstituted systems to examine the influence of mitochondrial lipids on MCL1 activity and conformation. Site-directed mutagenesis and fluorescence spectroscopic analyses revealed that the BCL2 homology region of MCL1 (MCL1ΔNΔC) inhibits permeabilization of MOM-like membranes exclusively via canonical BH3-into-groove interactions with both cBID-like activators and BAX-like effectors. Contrary to currently popular models, MCL1ΔNΔC did not require becoming embedded into the membrane to inhibit membrane permeabilization, and interaction with cBID was more productive for MCL1ΔNΔC inhibitory activity than interaction with BAX. We also report that membranes rich in cardiolipin (CL), but not phosphatidylinositol (PI), trigger a profound conformational change in MCL1ΔNΔC leading to membrane integration and unleashment of an intrinsic lipidic poreforming activity of the molecule. Cholesterol (CHOL) reduces both the conformational change and the lipidic pore-forming activity of MCL1ΔNΔC in CL-rich membranes, but it does not affect the interaction of MCL1ΔNΔC with proapoptotic partners in MOM-like liposomes. In addition, we identified MCL1α5 as the minimal domain of the protein responsible for its membrane-permeabilizing function both in model membranes and at the mitochondrial level. Our results provide novel mechanistic insight into MCL1 function in the context of a membrane milieu and add significantly to a growing body of evidence supporting an active role of mitochondrial membrane lipids in BCL2 protein function.

T

Several models have been proposed to explain how protein− protein and protein−membrane interactions dictate BCL2 protein family function. From the perspective of BCL2-like proteins, pioneering structural studies revealed a hydrophobic groove in soluble BCLXL encompassing BH1, BH2, and BH3 domains of the molecule that can act as a surface for binding to and sequestering BH3 domains of proapoptotic partners.1 Shortly thereafter, the “Direct Activation” model was built stating that BCL2-like proteins principally inhibit apoptosis by sequestering cBID-like activator ligands, rather than BAX-like effectors.4 In contrast, the “Indirect Activation” model proposed that BCL2-like proteins prevent apoptosis by constitutively binding to and neutralizing BAX-like effectors, not BH3-only proteins.5 More recently, the “embedded together” model incorporated an active role of the membrane in the BCL2 interactome.6 Here, BCL2-like proteins exert their antiapoptotic function by neutralizing either BAX-type effectors or BH3-only ligands, with the hierarchy of interactions depending on the relative binding affinities between the different BCL2 family members that universally change when they are localized to the membrane. Importantly, according to this model the prosurvival family members must become

he most common form of programmed cell death in biology and disease is the mitochondrial pathway of apoptosis, wherein a pivotal event is the mitochondrial outer membrane permeabilization (MOMP) allowing for release of cytochrome c (cyt c) and other mitochondrial apoptogenic factors into the cytosol.1 The BCL2 protein family has emerged as the master regulator of MOMP and therefore constitutes a critical control point in the intracellular apoptotic cascade. Members of this family are classically divided into three subclasses based on functional criteria as well as on the presence of up to four BCL2 homology (BH) domains: (1) Multidomain BCL2-like antiapoptotic proteins, including BCL2 itself, BCLXL, and MCL1, which inhibit MOMP; (2) multidomain BAX-like proapoptotic effectors (BAX, BAK, and perhaps BOK), which once activated directly cause MOMP; and (3) monodomain “BH3-only” proapoptotic ligands such as cBID and BAD, which trigger functional activation of BAX-type effectors. Despite its utility, the abovedescribed classification does not take into account that the functional phenotype of many multidomain BCL2 family proteins can be reversed under certain physiological conditions.2 In addition, it is not definitively proven that all members within each BCL2 family subgroup share the same number of BH domains or even an identical mechanism of action.3 © XXXX American Chemical Society

Received: July 27, 2014 Accepted: October 14, 2014

A

dx.doi.org/10.1021/cb500592e | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology

Articles

Figure 1. MCL1ΔNΔC inhibits MOM-like membrane permeabilization through a dual BH3-into-groove interaction mechanism. (a) Representative kinetics of ANTS/DPX release elicited by BAX/BAKΔC plus cBID from MOM-like LUV composed of 45%PC/35%PE/10%PI/10%CL (mol/mol) in the presence or absence of MCL1ΔNΔC. All proteins were added at 150 nM. (b, c) Dose-dependence of ANTS/DPX release elicited by combinations of wt or mt BAX, BAKΔC, cBID, or MCL1ΔNΔC proteins from MOM-like LUV. ANTS/DPX release values were normalized to those obtained without MCL1ΔNΔC. Data shown as mean ± standard error of the mean (S.E.M.), n = 4−7. All IC50 values had 95% confidence intervals.

permeability can be reproduced in a highly simplified system, composed of recombinant proteins and MOM-like liposomes containing low amounts of CL and encapsulating the fluorophore/quencher pair ANTS/DPX.8,13 Consistently, we found that adding BAX or BAKΔC plus cBID to MOM-like liposomes containing 10 mol % of CL (10%CL LUV) led to an increase in ANTS/DPX fluorescence due to permeabilization of the vesicle membrane, which was inhibitable by pretreatment of the liposomes with an equimolar amount of MCL1ΔNΔC relative to proapoptotic proteins (Figure 1a). By contrast, none of these proteins alone produced significant vesicular ANTS/ DPX release under these conditions (see below). It is currently debated whether BCL2-type proteins predominantly inhibit BAX/BAK-mediated membrane permeabilization by binding to and neutralizing BAX-type proteins, their BH3-only protein activators such as cBID, or both types of molecules.1,4−7 Another related question that is currently under debate is whether BCL2-type proteins inhibit membrane permeabilization solely through canonical BH3-into-groove interactions with proapoptotic partners akin to what is observed in solution-based binding studies, or also via noncanonical interactions involving additional binding surfaces not detected in studies performed in membrane-free environments. To attempt answering these mechanistic questions for the specific case of MCL1, we first used a set of BH3 mutants (mt) of BAX (BAXD68R), BAK (BAKΔCI82AN83A), and cBID (cBIDM97AD98A) that selectively remove BAX, BAK, or cBID binding to BCL2type proteins in cellular extracts.14,15 Reliably, BAXmt+cBID, BAX+cBIDmt, and BAKΔCmt+cBID mixtures behaved as combinations of wild-type proteins in our liposome permeabilization assay (Supporting Information FigureS1a). In contrast, the BAKΔC+cBIDmt mixture displayed a greatly reduced ability to permeabilize MOM-like LUV, and thus, the latter combination was discarded from this analysis. Next, the relative effect of each mutant was tested by examining the concentration-dependent inhibition of vesicular ANTS/DPX release mediated by MCL1ΔNΔC. Selective removal of MCL1ΔNΔC:BAXBH3 interaction by mutagenesis decreased MCL1ΔNΔC inhibitory potency approximately by half (1.9-fold increase in IC50), but it did not abolish

embedded into the MOM to effectively block the action of BAX and BAK. Last, the “unified model” also contemplated that antiapoptotic proteins act by dual sequestration of BAXtype effectors or BH3-only ligands at the plane of the membrane but puts forward the idea that the former interaction is uniformly stronger and therefore harder to overcome than the latter one.7 On top of this, several groups provided evidence that certain mitochondrial membrane lipids, including CL, modulate the function of BAX-type proteins, although it is still unclear whether CL is required for apoptosis.8−12 Nevertheless, the potential contribution of CL and other mitochondrial lipids to the function of antiapoptotic BCL2 family members remains virtually unexplored. Here, we used reductionist mechanistic investigations to examine the membrane activities and conformations of MCL1, a BCL2-type protein with a prominent role in cancer development and relapse.1 On the one hand, we show that in the presence of a MOM-like environment, the BCL2 homology region of MCL1 (MCL1ΔNΔC) ties up both BAX-type effectors and cBID ligands in complexes nonproductive for membrane permeabilization solely via canonical BH3-intogroove interactions and without MCL1ΔNΔC adopting a membrane-embedded conformation. On the other hand, we found that CL-rich membranes convert MCL1ΔNΔC into a pore-forming molecule sharing mechanistic features with activated BAX-type effectors. MCL1α5 was identified as the minimal domain of the protein responsible for its pore-forming activity both in model and native mitochondrial membranes. We also report that two additional mitochondrial lipids, PI and CHOL, differentially affect MCL1ΔNΔC membrane activities and conformations. Altogether, our results provide novel mechanistic insight into MCL1 function in the context of a membrane environment.



RESULTS AND DISCUSSION

MCL1ΔNΔC Inhibits Permeabilization of MOM-like Membranes through BH3-into-Groove Interactions with cBID-like Ligands and Activated BAX-like Effectors. We and others previously reported that the physiologically relevant ability of BCL2 family proteins to modulate membrane B

dx.doi.org/10.1021/cb500592e | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology

Articles

Figure 2. MCL1ΔNΔC contains an intrinsic lipidic pore-forming activity that is modulable by mitochondrial lipids. (a) Extents of vesicular contents release elicited by indicated BCL2 family proteins from liposomes containing increasing doses of CL. Liposome compositions were 55%PC/35%PE/ 10%PI/0%CL mol/mol (0CL), 52%PC/35%PE/10%PI/3%CL mol/mol (3CL), 45%PC/35%PE/10%PI/10%CL mol/mol (10CL), 30%PC/35% PE/10%PI/25%CL mol/mol (25CL), 20%PC/20%PE/10%PI/50%CL mol/mol (50CL), and 100%CL (100CL). In all cases, protein concentration was 300 nM (n = 3−4). (b) Dose-dependent effect of MCL1ΔC in the release of ANTS-DPX (ANTS), FD10, and FD70 from liposomes with different CL content. Data shown as mean ± SEM (n = 2−3). (c) Effect of different BCL2 family proteins in pyPC transbilayer distribution in LUV with different CL content (n = 3). (d) Effect of lipids modulating membrane intrinsic monolayer curvature (LPC, DAG) or bending elasticity (CHOL) on MCL1ΔNΔC permeabilizing-activity in CL LUV. Concentrations of MCL1ΔNΔC were 150 nM (LPC) or 600 nM (DAG, CHOL). Data shown as mean ± SEM (n = 2−4). (e) Representative ANTS-DPX release kinetics induced by MCL1ΔNΔC (300 nM) and alamethicin (100 nM) in CL LUV or PI LUV. (f) Destabilization of membrane lipid bilayer structure elicited by indicated proteins assessed by 31P NMR.

MCL1ΔNΔC-mediated inhibition of liposome permeabilization (Figure 1b). Similarly, selective removal of MCL1ΔNΔC:BAKΔCBH3 interaction increased MCL1ΔNΔC’s IC50 by 3.7-fold, but it did not abrogate MCL1ΔNΔC inhibition of vesicular contents release (Figure 1c). On the other hand, when MCL1ΔNΔC:cBID interaction was selectively removed through mutagenesis, the potency of MCL1ΔNΔC to inhibit BAX-mediated membrane permeabilization was reduced by 5.8-fold (Figure 1b). These results indicate that the MCL1ΔNΔC:BIDBH3 interaction is more productive for MCL1ΔC inhibitory activity than the MCL1ΔNΔC:BAXBH3 interaction. Thus, our results contradict the idea promulgated in the “Unified Model” stating that BCL2-type proteins unanimously interact more strongly with activated BAX-type effectors than with cBID-like ligands.7 The “Unified Model” was based on results obtained with cBID chimeras containing BH3 domains of different BCL2 family members, but the interaction between MCL1 and cBID− BAXBH3 chimera was not specifically tested in that study because it was undetectable. It appears reasonable to argue that such cBID−BAXBH3 chimera did not recapitulate the full MCL1-binding potential of the intact BAX protein, because several groups have detected MCL1:BAX and MCL1:BAXBH3 complexes in cellular and cell-free environments.14,16

To further evaluate the importance for MCL1ΔNΔC inhibitory activity of canonical and noncanonical interactions with proapoptotic partners, we tested (i) the R244E variant of MCL1ΔNΔC (MCL1ΔNΔCmt1) containing a mutation in a critical residue of the canonical BH3-binding groove1 and (ii) the I163E variant of MCL1ΔNΔC (MCL1ΔNΔCmt2) containing a mutation in helix α1 whose equivalent in BCL2 (V15E) disrupts a noncanonical BAX:BCL-2 binding surface.17 Relative to native MCL1ΔNΔC, the MCL1ΔNΔCmt1 mutant virtually lost all ability to inhibit vesicular contents release elicited by cBID-activated BAX or BAKΔC, while the MCL1ΔNΔCmt2 variant had a much more restricted effect (Figure 1b,c). Of note, intrinsic fluorescence spectra of MCL1ΔNΔC R244E and MCL1ΔNΔCI163E variants were virtually indistinguishable from that of MCL1ΔNΔC, indicating the two mutants preserved the global folding of the native protein (Supporting Information Figure S1b). These results strongly suggest that MCL1ΔNΔC inhibits permeabilization of MOM-like membranes primarily, if not exclusively, via canonical BH3-into-groove interactions with proapoptotic partners. It is noteworthy that although BCL2 and MCL1 share three conserved BH1−BH3 domains delimiting their canonical grooves, the former but not the latter protein contains a BH4 domain at helix α1.3 C

dx.doi.org/10.1021/cb500592e | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology

Articles

phosphatidylcholine (PC) analogue pyPC exclusively in the external monolayer of the membrane, and changes in the ratio of pyPCexcimer to monomer fluorescence intensity signals (IE/IM) were used to monitor the outward−inward transbilayer movement of the PC analogue.8 As shown in Figure 2c, adding MCL1ΔNΔC to 100%CL LUV led to a pronounced decrease of pyPCIE/IM ratio, indicating transfer of the analogue to the internal monolayer of the liposomal membrane. In sharp contrast, MCL1ΔNΔC addition to MOM-like LUV caused negligible transbilayer movement of pyPC. Membrane physical properties such as monolayer intrinsic curvature and bending elasticity are known modulators of lipidic pores formed by BAX and BAK.8,11 To test whether any of these parameters contributes to MCL1ΔNΔC membranepermeabilizing activity, CL LUVs were prepared containing increasing amounts of LPC, which induces positive membrane monolayer curvature, DAG, which induces negative membrane monolayer curvature, or CHOL, which decreases membrane bending elasticity.11 As shown in Figure 2d, increasing the proportion of LPC in CL LUV gradually enhanced the extent of membrane permeabilization elicited by MCL1ΔNΔC, while incrementing the DAG or CHOL content above a threshold level progressively reduced MCL1ΔNΔC permeabilizing activity. Thus, inducing positive monolayer curvature stress promotes membrane permeabilization by MCL1ΔNΔC, while inducing negative monolayer curvature stress or reducing membrane bending elasticity inhibits MCL1ΔNΔC permeabilizing activity. These results support the notion that MCL1ΔNΔC permeabilizes CL-rich liposomes by forming lipidic pores. Aside from CL, the main acidic lipid of mitochondrial membranes is phosphatidylinositol (PI).24 Thus, we also tested whether PI can substitute for CL in triggering MCL1ΔC membrane-permeabilizing activity. However, MCL1ΔNΔC induced minimal ANTS/DPX release from 100%PI LUV, while the channel-forming peptide alamethicin efficiently permeabilized both 100%CL LUV and 100%PI LUV (Figure 2e). To further examine the relation between membrane lipid composition and MCL1ΔNΔC permeabizing activity, and as a further test of the MCL1ΔNΔC lipidic pore model, we conducted 31P NMR studies. The 31P NMR spectrum of MCL1ΔNΔC-treated liposomes lacking CL (MCL1ΔC, 0% CL) showed the high-field peak and low-field shoulder typical of a bilayer arrangement of membrane lipids (Figure 2f). The addition of MCL1ΔNΔC to membranes containing increasing amounts of CL led to a change in the shape of the spectrum: the bilayer-type signal markedly decreased while a prominent peak appeared around the chemical shift position of phospholipids experiencing isotropic motion, which is typical for highly curved nonbilayer type lipid dispositions. The contribution of the isotropic component to the spectrum gradually increased with the proportion of CL in a manner that correlated qualitatively with CL’s impact on MCL1ΔNΔC membrane-permeabilizing activity. Further agreement between the lipid-dependence of the appearance of the isotropic resonance and membrane permeabilization was found in that (i) MCL1ΔNΔCdid not induce the appearance of an isotropic signal in PI liposomes and (ii) incorporation of CHOL in CLrich membranes significantly decreased the MCL1ΔNΔCinduced isotropic signal (Figure 2f). Collectively, these results indicate that MCL1ΔNΔC possesses the intrinsic capacity to destabilize the membrane

Solution-based binding studies have identified several small molecules that can insert into the MCL1ΔNΔC canonical groove to block entrance therein of proapoptotic BH3 domains. Of these compounds, TW37, AT101, and GX15-037 (Obatoclax) have progressed into clinical studies.18 However, specific proof is lacking that these small molecules can actually suppress the inhibition exerted by MCL1 on membrane permeabilization. Thus, we decided to examine the effect of these clinical drugs in our reconstituted liposome system. Consistent with their purported BH3-mimetic nature, TW37 and AT101 dose-dependently suppressed MCL1ΔNΔC inhition of BAX- and BAKΔC-mediated membrane permeabilization (Supporting Information Figure S1c). Contrastingly, obatoclax had minimal effect in MCL1ΔNΔC inhibitory activity. We next examined the effect of the three clinical drugs on MCL1ΔNΔC inhibitory activity in the presence of MCL1-binding-deficient mutants of BAX and cBID. As shown in Supporting Information Figure S1d, TW37/AT101 overcomes less readily the inhibition exerted by MCL1ΔNΔC on BAXmt+cBID than on BAX+cBIDmt. Therefore, in our experimental system, MCL1-targeting BH3 mimetic drugs block MCL1ΔNΔC:BAX complexes more readily than MCL1ΔNΔC:cBID complexes. Once again, this is in contradiction with the “Unified Model” proposed for BCL2 family proteins.7 However, further studies are clearly necessary to determine to what extent MCL1 or other BCL2-type proteins can inhibit MOMP by dominantly sequestering cBID-like ligands relative to activated BAX-like effectors. MCL1ΔNΔC Possesses an Intrinsic Lipidic PoreForming Mechanism Regulatable by Mitochondrial Lipids. Previous studies by different groups revealed that CL plays an important role in membrane permeabilization by BAX and BAK.8,9,11 To analyze the posible implication of CL in MCL1ΔNΔC membrane activity, LUVs loaded with ANTS/ DPX were prepared containing increasing amounts of CL, followed by the addition of MCL1ΔNΔC alone or combined with BAX or BAKΔC plus cBID (Figure 2a). Remarkably, increasing the liposomal CL concentration gradually diminished the inhibitory activity exerted by MCL1ΔNΔC on BAX or BAKΔC-mediated membrane permeabilization and also progressively augmented vesicular ANTS/DPX release elicited by MCL1ΔNΔC alone. The latter phenomenon was specific for MCL1ΔNΔC, since BAX, BAKΔC, or cBID alone did not permeabilize liposomes with high CL content. Next, we explored the size of the membrane permeabilization pathway created by MCL1ΔNΔC using vesicles containing ANTS/DPX (≈0.4 kDa), or self-quenching concentrations of 10 kDa or 70 kDa fluorescein-dextrans (FD10 and FD70, respectively). In pure CL liposomes (100%CL LUV), MCL1ΔNΔC induced the release of vesicle-encapsulated solutes of all sizes in a dosedependent manner, but with different efficiencies: the larger the encapsulated molecule, the higher the MCL1ΔNΔC concentration required for release (Figure 2b, filled symbols). Contrastingly, MCL1ΔNΔC did not release substantial amounts of fluorescent markers from MOM-like liposomes (10%CL LUV) at any concentrations tested (Figure 2b, empty symbols). Accumulating evidence indicates that BAX and BAK permeabilize membranes by forming lipidic pores.8,11,19−23 One implication of the lipidic pore model is that such structure should allow the movement of lipid molecules from one monolayer of the bilayer to the other. To test this possibility, LUVs were labeled with the pyrene-containing fluorescent D

dx.doi.org/10.1021/cb500592e | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology

Articles

Figure 3. Functional and structural characterization of NBD-labeled monocysteine MCL1ΔNΔC mutants. (a) Sequence-alignment of MCL1, BCL2, and BAX, indicating specific residues mutated to Cys for subsequent labeling with the environmentally sensitive NBD fluorophore (black circles). (b) Structural representation of MCL1ΔNΔC depicting specific residues mutated to Cys for subsequent labeling with NBD (black spheres). (c) Representative Trp fluorescence spectra of NBD-labeled monocysteine MCL1ΔNΔC mutants. (d) Assesment of cyt c release by NBD-labeled MCL1ΔNΔC mutants.

play important roles in BCL2, BAX, or BAK membrane activities.17,26−28 All of the monocysteine MCL1ΔNΔC mutant proteins were expressed at aceptable levels with the exception of MCL1ΔNΔC mutant V224C, which showed virtually no expression, probably because structural perturbations at this site are not well tolerated and induce MCL1ΔNΔC misfolding. The structural integrity of NBD-labeled monocysteine MCL1ΔNΔC mutants was assessed by comparing their Trp fluorescence spectrum with that of unlabeled wild-type MCL1ΔNΔC. The Trp fluorescence spectra of all NBDlabeled monocysteine mutants did not differ significantly from that of wild-type MCL1ΔNΔC (Figure 3c). To test whether Cys mutation and/or NBD labeling affect MCL1ΔNΔC function, we examined the ability of each NBD-labeled monocysteine MCL1ΔNΔC mutant to inhibit the release of mitochondrial cyt c elicited by BAX+cBID. As in native MCL1ΔNΔC, all NBD-labeled monocysteine MCL1ΔNΔC mutants effectively inhibited cyt c release (Figure 3d). We next analyzed whether MOM-like liposomes (10%CL LUV) change spectral properties of NBD-labeled MCL1ΔNΔC mutants, either alone or in combination with different proapoptotic proteins. Insignificant changes were observed when MCL1ΔNΔC mutants were incubated with MOM-like LUV alone, except for MCL1ΔNΔCG200‑NBD, which showed a slight increase in NBD fluorescence intensity (Figure 4a,b). Co-addition of BAX or BAKΔC together with MOM-like LUV did not produce further changes in spectral properties of any NBD-labeled monocysteine MCL1ΔNΔC mutant. However, coaddition of cBID together with MOM-like LUV produced a prominent increase in the fluorescence intensity of MCL1ΔNΔCR229‑NBD, accompanied by a ∼6 nm blue-shift in the λmax(em) of this variant (Figure 4a, left). Importantly, this site is localized at the canonical BH3-binding-groove of

bilayer structure by forming lipidic pores, in a manner that is regulatable by specific mitochondrial lipids. Assessing MCL1ΔNΔC Membrane Conformations and Mechanisms Using Site-Specific Fluorescence Labeling and Fluorescence Spectroscopy. Fluorescence spectroscopy coupled to site-specific fluorescence labeling is a powerful methodology to obtain information about membrane-associated protein conformations and mechanisms. 25 In this approach, a single amino acid in a cysteine (Cys)-free protein is replaced with a Cys residue, and a fluorescent dye is then covalently attached to the sulfhidryl group of the newly introduced Cys residue. Spectral properties of fluorescently labeled monocysteine proteins are then analyzed and compared in the absence and presence of membranes. NBD (7-nitrobenz2-oxa-1,3,-diazolyl) was chosen as the fluorescent probe in this study because it is a relatively small and uncharged dye, and its spectral properties are dramatically different in aqueous and nonaqueous environments. For example, when NBD moves from an aqueus milieu (i.e., the surface of a soluble protein) to a hydrophobic environment (i.e., the nonpolar interior of the membrane or a protein), its emission intensity increases while its wavelength of maximum emission shifts to the blue. The nonpolar membrane interior can be distinguished from a nonpolar environment inside a folded protein or at the interface of oligomerized proteins, by examining the NBD quenching ability of lipids containing doxyl (Dox) labels that are restricted to locations within the nonpolar core of the bilayer. Figure 3a displays the primary sequence of MCL1 aligned to those of BCL2 and BAX with MCL1 residues mutated to Cys highlighted as black circles, while Figure 3b displays MCL1ΔNΔC three-dimensional structure with the same residues highlighted as black spheres. MCL1ΔNΔC residues were primarily selected taking into account that equivalent sites E

dx.doi.org/10.1021/cb500592e | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology

Articles

Figure 4. continued Binding potencies (EC50 values) for MCL1ΔNΔCR229C‑NBD and indicated proapoptotic proteins in MOM-like LUV (10CL), 35% PC/35%PE/30%PI mol/mol LUV (20PI), or 20%PC/35%PE/10% PI/10%CL/25%CHOL mol/mol LUV (25CHOL; n = 3−7). (f) Dox5 (20 mol %) mediated NBD quenching of MCL1ΔC mutants (n = 2−3).

MCL1ΔNΔC, suggesting this groove residue has been likely buried by the BH3 domain of cBID (Figure 4c, right). In addition, the MOM-like LUV plus cBID mixture also significantly increased and decreased the fluorescence intensities of MCL1ΔNΔCG200C‑NBD and MCL1ΔNΔCT282C‑NBD mutants, respectively, which localize in the vicinity of the BH3-binding groove (Figure 4b,c). In contrast, no significant changes in NBD fluorescence were observed at α1 residues L160 and Q170, even though equivalent residues in BCL2ΔC purportedly constitute a noncanonical BAX binding-surface (Figure 4c, left).17 Equally, when MOM-like LUVs were combined with cBID plus BAX-like proteins, NBD fluorescence changed prominently for MCL1ΔNΔCR229C‑NBD, and more moderately for MCL1ΔNΔCG200C‑NBD and MCL1ΔNΔCT282C‑NBD, but not for any other NBD-labeled MCL1ΔNΔC monocysteine mutant (Figure 4b). When MOM-like LUVs were prepared including 25 mol % CHOL, the same profile of NBD fluorescence changes was observed as well (Supporting Information Figure S2). Thus, these results support the notion that in the presence of MOM-like membranes with or without CHOL, cBID, cBID-activated BAX, and cBID-activated BAKΔC, all bind to MCL1ΔNΔC via canonical BH3-into-groove interactions. The increase in MCL1ΔNΔCR229‑NBD fluorescence elicited by cBID alone or together with BAX/BAKΔC occurred in a concentration-dependent and saturatable manner, allowing to estimate apparent binding potencies (EC50 values) between MCL1ΔNΔC and proapoptotic partners in a MOM-like lipid environment (Figure 4d). Interestingly, cBID bound MCL1ΔNΔC with a potency that was significantly stimulated upon coaddition of BAKΔC, but not BAX (Figure 4d). Moreover, MCL1-binding-deficient mutants of BAX and cBID showed that MCL1ΔNΔCR229C‑NBD interacts more strongly with cBID than with BAX, which agrees qualitatively with the distinct effects exerted by those same mutants on the ability of MCL1ΔNΔC to inhibit membrane permeabilization (Figure 4e). Additional experiments with MCL1-binding-deficiency showed that supplementing MOM-like LUV with 25 mol % CHOL had no significant effect on MCL1ΔNΔC binding to cBID or to BAX while substitution of 10 mol % CL by 20 mol % PI reduced MCL1ΔNΔC binding to both proapoptotic partners (Figure 4e). We also examined the degree of NBD fluorescence quenching by 1-palmitoyl-2-stearoyl-(5-doxyl)-sn-glycero-3phosphocholine (PC-Dox5), a phospholipid derivative with an uncharged NO group attached to the fifth carbon of the hydrocarbon chain that is localized in a shallow location at the hydrophobic core of the lipid bilayer.29 As can be seen, Dox5 caused negligible quenching of NBD-labeled MCL1ΔNΔC mutants incubated with MOM-like membranes and proapoptotic proteins (Figure 4f). Thus, we conclude that neither cBID nor cBID-activated BAX-like proteins trigger MCL1ΔNΔC embedding into MOM-like membranes. The latter set of results is at odds with an important postulate of the “Membrane-

Figure 4. Analysis of MCL1ΔNΔC conformation in MOM-like membranes by site-specific NBD labeling and fluorescence spectroscopy. (a) Representative NBD fluorescence emission spectra of indicated monocysteine MCL1ΔC mutants incubated with or without MOM-like LUV (10CL) and proapoptotic proteins. In all cases, NBD fluorescence was normalized to the peak intensity of the NBD spectrum in buffer alone. (b) Average NBD fluorescence intensity ratios of monocysteine MCL1ΔNΔC mutants incubated with MOMlike LUV in the absence or presence of indicated proapoptotic proteins. Data shown as mean ± SEM (n = 3−5). (c) Structural model of MCL1ΔNΔC complexed with BID BH3 domain (blue helix) depicting MCL1 residues that undergo significant NBD fluorescence changes in the presence of MOM-like LUV plus cBID (blue circles). (d) Dose-dependent MCL1R229C‑NBD fluorescence changes elicited by indicated proapoptotic proteins in MOM-like LUV (n = 3). (e) F

dx.doi.org/10.1021/cb500592e | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology

Articles

Figure 5. Analysis of MCL1ΔNΔC conformation in CL-rich membranes by site-specific NBD labeling and fluorescence spectroscopy. (a) Average NBD fluorescence intensity ratios of monocysteine MCL1ΔC mutants incubated with 100%CL LUV in the absence or presence of indicated proapoptotic proteins. Data shown as mean ± SEM (n = 3−5). (b) Representative NBD fluorescence emission spectra of indicated monocysteine MCL1ΔNΔC mutants incubated with or without 100%CL LUV and proapoptotic proteins. (c) MCL1ΔNΔC residue exposure to lipid in the presence of 100%CL LUV alone or combined with BAX, BAKΔC, or cBID, assessed by quenching of NBD fluorescence with Dox5-PC (n = 2−5). (d, e) Effect of lipid composition in Dox5-mediated quenching of indicated NBD-labeled MCL1ΔC mutants (n = 3−4).

MCL1ΔNΔC mutants underwent substantial changes in NBD fluorescence in the presence of 100%CL LUV (Figure 5a,b). Moreover, coaddition of cBID, BAX, or BAKΔC, alone or combined, together with 100%CL LUV did not produce further changes in the NBD fluorescence intensity of any MCL1ΔNΔC mutant examined. Figure 5c is a histogram showing the extent of NBD fluorescence quenching by Dox5 for the set of nine MCL1ΔNΔC mutants incubated with CL-rich LUV, alone or in the presence of BAX, BAKΔC, or cBID. In all cases, Dox5 potently quenched NBD fluorescence at A256, A261, and T282 sites, located at MCL1ΔNΔCα5 and -α6, suggesting that this region becomes at least partially embedded in the bilayer. To assess the extent of membrane immersion of MCL1ΔNΔCα5 and -α6, we also analyzed the extent of NBD fluorescence quenching at A256, A261, and T282 sites by 1-palmitoyl-2stearoyl-(14-doxyl)-sn-glycero-3-phosphocholine (Dox14), a lipid carrying a doxyl quencher group that is localized more deeply embedded within the membrane hydrophobic matrix than Dox5.29 Quenching levels for MCL1ΔNΔCA256C‑NBD,

Embedded” model stating that BH3-only proteins or activated BAX-type proteins universally trigger membrane embedding of the α5α6 “hairpin” region of BCL2-type proteins, and subsequently noncanonical BCL2 hairpin:BAX hairpin interactions functionally invalidate BAX-type effectors.6,26 On the other hand, because solution NMR studies indicate that MCL1ΔNΔC displays a “ready-to-bind” BH3-binding groove,30 it is plausible that MCL1ΔNΔC requires minimal structural reorganization at the membrane level to interact with proapoptotic partners. Based on this collective evidence, we propose that MCL1 inhibits MOMP primarily, if not solely by adopting a solutionlike conformation at the MOM level that allows it to sequester cBID-like BH3-only proteins as well as activated BAX-type effectors via canonical BH3-into-groove interaction mechanisms. We next analyzed the behavior of the same set of NBDlabeled monocysteine MCL1ΔNΔC mutants in the presence of 100%CL LUV, with or without proapoptotic proteins. Remarkably, except for MCL1ΔNΔC Q170C‑NBD , all the G

dx.doi.org/10.1021/cb500592e | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology

Articles

Figure 6. A synthetic peptide representing MCL1α5 helix efficiently permeabilizes mitochondria-like model membranes and isolated organelles. (a) Schematic representation of human MCL1 helical regions (black boxes) according to mouse MCL1ΔNΔC NMR structure (PDB: 1WSX), with residues delimiting each helix annotated below the corresponding box (gray numbers) and residues encompassing synthetic peptides used in this study depicted above (black numbers). (b) Effect of MCL1-derived peptides (1 μM) in LUV permeability (n = 2−3). (c) Analysis of MCL1-derived peptide insertion into CL monolayers. (d) Model of MCL1α5 depicting two highly conserved Lys residues with potential for CL interaction. (e) Dose-dependent effect of MCL1α5 and MCL1α5K276E,K279E in ANTS/DPX release from 100%CL LUV (n = 3−4). (f) Effect of MCL1α5 and MCL1α5K276E,K279E on membrane bilayer structure assessed by P31 NMR. (g) Effect of MCL1-derived peptides (10 μM) in the release of cyt c from WT MEF or BAX/BAK DKO MEF mitochondria. Data are representative of three independent experiments. (h) Role of PTP in the release of cyt c elicited by BAX plus cBID, CaCl2, or MCL1α5.

MCL1ΔNΔCS261C‑NBD, and MCL1ΔNΔCT282C‑NBD mutants were all considerably lower with Dox14 than with Dox5, indicating that MCL1ΔNΔCA256C‑NBD, MCL1ΔNΔCS261C‑NBD, and MCL1ΔNΔCT282C‑NBD mutants are only partially inserted within the lipid bilayer of CL-rich LUV (Supporting Information Figure S3). Further experiments showed that increasing liposomal CL content led to a progressive enhancement of Dox5-mediated MCL1ΔNΔCA256C‑NBD, MCL1ΔNΔCS261C‑NBD, and MCL1ΔNΔCT282C‑NBD fluorescence quenching (Figure 5d). On the contrary, the fluorescence of NBD-labeled MCL1ΔNΔC mutants was not quenched by Dox5 incorporated in LUV containing increasing amounts of PI. Moreover, supplementing CL LUV with 25 mol % CHOL considerably reduced Dox5-mediated quenching of MCL1ΔNΔC A 2 5 6 C ‑ N B D , MCL1ΔNΔC S 2 6 1 C ‑ N B D , and MCL1ΔNΔCT282C‑NBD fluorescence (Figure 5e). Thus, a

correlation exists between shallow (nontransmembrane) membrane insertion of the MCL1 α5α6 region and breaching of membrane permeability barrier. Based on these collective findings, it can be argued that the mitochondrial lipid composition can modulate a process whereby amphipatic MCL1 α5α6 helices insert into the external monolayer of the MOM to create therein the lateral tension and curvature strain required to open a lipidic pore. MCL1 α5 Peptide Autonomously Permeabilizes Mitochondrial-like Membranes and Induces MOMP. We next attempted to identify the region that encompasses the membrane-permeabilizing activity of MCL1. To this aim, we screened the membrane activity of an overlapping array of ∼20−25 amino acid peptides representing all α-helical segments of MCL1 (Figure 6a). First, the abilities of the different MCL1-derived peptides to release ANTS/DPX from H

dx.doi.org/10.1021/cb500592e | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology

Articles

Figure 7. MCL1α5 peptide induces mitochondrial depolarization and cyt c release in semi-intact cells. (a) Effect of MCL1-derived peptides (10 μM) in the release of cyt c from semi-intact BAX/BAK DKO fibroblasts permeabilized with 0.005% digitonin. Data are representative of three independent experiments. (b) Effect of indicated treatments in the mitochondrial ΔΨm of semi-intact BAX/BAK DKO MEF assessed by Mitotracker fluorescence. Representative confocal micrographs are shown with the time of each frame relative to the time of peptide addition. Hoescht (Hst) dye allows visualization of the nucleus. (c) Mitochondrial depolarization elicited by indicated treatments, assessed as the loss of corrected total cell fluorescence per 103 cells. (d) Effect of indicated treatments in HeLa cells stably expressing cyt c-GFP and transfected with mitoRFP. Representative confocal micrographs are shown with the time of each frame relative to the time of peptide addition. Bars on the right represent averaged Pearson’s coefficient of cyt c-GFP and mito-RFP colocalization (n ≥ 10).

substantially permeabilize 100%CL LUV suggests that the permeabilizing activity of MCL1α5 peptide is not exclusively due to its membrane-inserting capacity. We noticed that the C-terminal part of MCL1α5 harbors two conserved Lys residues (K276 and K279) with potential to establish electrostatic interaction with dianionic CL that may contribute to MCL1α5 membrane activities (Figure 6d). In agreement with this possibility, generation of MCL1α5K276E,K279E variant dramatically reduced peptide liposome-permeabilizing and monolayer-penetrating capacities (Figure 6e, and Supporting Information Figure S4a,b). Interestingly, the C-terminal part of BCLXLα5 contains two negatively charged residues (E153 and D156) instead of two lysines, and a synthetic BCLXLα5 peptide was much weaker than the MCL1α5 peptide permeabilizing CL LUV (Supporting Information Figure S4b). Further experiments showed that

CL-containing LUV were compared. As shown in Figure 6b, MCL1α5 permeabilized the liposomes efficiently and in a CLconcentration dependent manner, while other MCL1-derived peptides were practically inactive. The modest permeabilizing activity of MCL1α9 in CL-containing LUV is consistent with a recent study.31 Next, the ability of MCL1-derived peptides to penetrate into CL monolayers was analyzed. For MCL1 α2, α5, and α9, the change in surface pressure (Δπ) upon peptide addition decreased linearly as a function of increasing initial surface pressure (π0), giving critical surface pressure (πc) values of 30.8, 43.4, and 37.9 mN/m, respectively. Considering that the surface pressure of phospholipid membranes is ∼30 mN/m, these data suggest that the MCL1α5 peptide, and to a lower extent the MCL1α9 and MCL1α2 peptides, but not other MCL1-derived peptides, can penetrate into 100%CL membranes. The fact that MCL1α2 and MCL1α9 peptides do not I

dx.doi.org/10.1021/cb500592e | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology

Articles

quantitative analysis of spatial correlation between cyto c-GFP and Mito-RFP fluorescence (Figure 7d). In summary, these results show that the region encompassing the MCL1α5 helix sustains a lipidic pore-forming activity allowing to permeabilize CL-rich model membranes and to release mitochondrial cyt c. Of note, BAXα5 has been previously shown to form lipidic pores in model membranes and to efficienly induce MOMP.34,35 Moreover, evidence is accumulating that BAX and BAK breach the MOM permeability barrier by forming lipid-containing pores.11,19−23 However, the lipidic pore-forming activity of BAX and BAK seems to rely not only in helix α5, but also in helices α4 and α6.20,22,23,35 Irrespective of this apparent discrepancy between the membrane-permeabilizing activities of MCL1 and BAX/ BAK, it is not unreasonable to hypothesize that CL accumulated at the MOM during advanced stages of the apoptotic process could unleash the lipidic pore-forming activity of MCL1, that would then contribute to MOMP either autonomously or through cooperation with activated BAX-type proteins. It is also conceivable that under some physiological conditions, the mitochondrial CHOL content may modulate MCL1 permeabilizing-activity. Furthermore, it is tantalizing to speculate that in addition to apoptosis-related lipids, other physiological conditions known to trigger MCL1 conversion into a proapoptotic molecule (i.e., MCL1 proteolytic cleavage36 or MCL1 gene-splicing37) may also uncover the lipidic poreforming activity of MCL1. Because MCL1 plays important roles in cancer development and relapse, our novel mechanistic information and tools may be useful for anticancer drug discovery.38

the MCL1α5 peptide, but not its K276E,K279E variant, produced an isotropic peak in CL liposomes, suggesting MCL1α5 destabilizes the membrane bilayer structure to form lipidic pores, as observed for MCL1ΔNΔC (Figure 6f). Moreover, LPC promoted and DAG inhibited membrane permeabilization induced by MCL1α5 peptide, also as in the case of MCL1ΔNΔC (Supporting Information Figure S4b). However, including 25 mol % CHOL in CL LUV did not reduce MCL1α5 permeabilizing activity, suggesting the MCL1α5 pore may be less sensitive to membrane bending elasticity than the MCL1ΔNΔC pore. Next, we tested the ability of different MCL1-derived peptides to induce cyt c release from isolated mitochondria. To this aim, mitochondria were isolated from wild-type (WT) or BAX/BAK DKO MEFs, followed by treatment with MCL1derived peptides, separation of supernatant (s) and pellet (p) fractions, and immunoblotting assesment of their cyt c and TOM20 contents. MCL1α5 peptide induced a dose- and timedependent depletion of cyt c in both WT and DKO MEF mitochondria, while the MCL1 α5K276E,K279E mutant peptide, and most other MCL1-derived peptides failed to release mitochondrial cyt c (Figure 6g, and Supporting Information Figure S4c,e). The BCLXLα5 peptide did not induce mitochondrial cyt c release, consistent with a previous study (Supporting Information Figure S4d).32 We also tested whether opening the mitochondrial permeability transition pore (mPTP) is a major contributor of MCL1α5-induced mitochondrial cyt c release. Speaking against this possibility, the mPTP inhibitor cyclosporine A (CsA) did not block MCL1α5-induced cyt c release, while at the same time CsA completely attenuated calcium-induced mPTP opening (Figure 6h). We then examined whether MCL1-derived peptides induce MOMP in semi-intact cells permeabilized with very low amounts of digitonin (0.005% w/v digitonin).33 First, we performed immunoblotting analyses of cyt c release of membrane and cytosol fractions separated after incubating semi-intact DKO MEFs with MCL1-derived peptides. Once again, among all MCL1-derived peptides examined, only MCL1α5 induced substantial mitochondrial cyt c release (Figure 7a). As an important control for nonspecific membrane damage, MCL1α5 failed to permeabilize the plasma membrane of DKO MEFs (Supporting Information Figure S5). Mitochondrial membrane potential (ΔΨm) is typically reduced early during apoptosis due to cyt c release. Loss of ΔΨm can therefore be used as an indicator of MOMP. To analyze whether MCL1α5 directly affects mitochondrial ΔΨm, we treated semi-intact DKO MEF cells with the peptide, and subsequently the cells were monitored for changes in the fluorescence of the CMXRos variant of Mito-Tracker Red fluorochrome by real-time fluorescence microscopy. Mitochondria of semi-intact DKO cells maintained a high ΔΨm, which decreased to low levels in a time- and dose-dependent manner when cells were treated with MCL1α5 peptide (Figures 7b,c). In contrast, minimal ΔΨm loss was observed in a mock reaction or with MCL1α5K276E,K279E peptide. Last, we imaged by real-time confocal microscopy the effect of MCL1α5 on HeLa cells stably expressing cyt c-GFP which had also been transfected with mito-RFP. Treatment of such cells with MCL1α5 (but not MCL1 α5K276E,K279E) caused a timedependent release of cyto c-GFP, as shown by direct visualization of the loss of cyt c-GFP fluorescence and by a



METHODS

Purification and Labeling of BCL2 Family Proteins. Recombinant MCL1ΔN151ΔC23 (MCL1ΔNΔC), BAX, BAKΔC21 (BAKΔC), and caspase-8-cleaved BID (cBID) were expressed and purified as described earlier.8,11,39 Mutant DNAs were generated by PCR-based mutagenesis using the Quickchange mutagenesis kit (Stratagene) or purchased at GenTech (Montreal, Canada). Details of labeling reactions provided in the Supporting Information. Liposome Preparation. Lipid mixtures at the indicated ratios were codissolved in chloroform/methanol (2:1), and organic solvents were removed by incubation under a vacuum for 2 h. Dry lipid films were resuspended in 100 mM KCl, 10 mM Hepes, 0.1 mM EDTA (KHE buffer), except in experiments for membrane permeabilization were 20 mMKCl, 10 mM Hepes pH 7.0, 0.1 mM EDTA, 12.5 mM ANTS, and 45 mM DPX. Liposomes were then subjected to five freeze/thaw cycles and subsequently extruded 10 times through two polycarbonate membranes of 0.2-μm pore size (Nucleopore, San Diego, CA) to obtain large unilamellar vesicles (LUV). Fluorescence Spectroscopy. Fluorescence intensity and spectral analyses were done in an 8100 Aminco-Bowman luminescence spectrometer (Spectronic Instruments, Rochester, NY) in a thermostatically controlled 1 cm path length cuvette with constant stirring, at 37 °C. Assays of vesicular contents release and PyPC motion were pereformed as described before.8 NBD fluorescence spectra (averaging three spectra) were recorded between 500 and 600 nm at a scan rate of 1 nm/s, using an excitation wavelength of 465 nm (slits 4 nm). In all cases, the contribution of buffer or liposomes alone to sample fluorescence was subtracted as blank. Unless otherwise specified, lipid concentration was 100 μM and protein concentration was 150 nM. Cyt c Release Assays. Assays with isolated mitochondria were done as described previously.34 Semi-intact cells were obtained by resuspension in a buffer consisting of 120 mMKCl, 10 mMNaCl, 1 mM KH2PO4, and 20 mM HEPES-Tris (pH 7.2) supplemented with 0.005% w/v digitonin. HeLa-cyt c-GFP cells were observed in an inverted confocal microscope (TE2000 U, Nikon, Melville, NY) J

dx.doi.org/10.1021/cb500592e | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology

Articles

equipped for epifluorescence and with a Nikon D-eclipse C1si confocal spectral detector, using an x60, 1.45 numerical aperture, oil immersion objective. The signal from each fluorophore was isolated by performing spectral unmixing with the EZ-C1 3.20 software (Nikon). Pearson’s correlation coefficient was estimated using ImageJ software. For analysis, images were converted to 8-bit, cell perimeter was drawn manually, and the background was subtracted automatically. Mitochondrial Membrane Potential Analysis. Semi-intact BAX/BAK DKO MEF cells were incubated with MitoTracker Red (50 nM) and Hoescht 33342 (0.5 μg/mL) to examine mitochondrial ΔΨm and nuclear morphology, respectively. Confocal images were obtained as described above. Images for time-lapse analysis were collected on an inverted fluorescence microscope (Leica DMI300B) with a minimum of 700 cells included per condition. ImageJ software was used to calculate the corrected total cell fluorescence (CTCF) with corrections for background fluorescence (CTCF = integrated density − (area of selected cell × mean fluorescence of background readings)).



unified model of mammalian BCL-2 protein family interactions at the mitochondria. Mol. Cell 44, 517−531. (8) Terrones, O., Antonsson, B., Yamaguchi, H., Wang, H. G., Liu, J., Lee, R. M., Herrmann, A., and Basañez, G. (2004) Lipidic pore formation by the concerted action of proapoptotic BAX and tBID. J. Biol. Chem. 279, 30081−30091. (9) Lucken-Ardjomande, S., Montessuit, S., and Martinou, J. C. (2008) Contributions to Bax insertion and oligomerization of lipids of the mitochondrial outer membrane. Cell Death Differ. 15, 929−937. (10) Montero, J., Morales, A., Llacuna, L., Lluis, J. M., Terrones, O., Basañez, G., Antonsson, B., Prieto, J., García-Ruiz, C., Colell, A., and Fernández-Checa, J. C. (2008) Mitochondrial cholesterol contributes to chemotherapy resistance in hepatocellular carcinoma. Cancer Res. 68, 5246−5256. (11) Landeta, O., Landajuela, A., Gil, D., Taneva, S., Di Primo, C., Sot, B., Valle, M., Frolov, V. A., and Basañez, G. (2011) Reconstitution of proapoptotic BAK function in liposomes reveals a dual role for mitochondrial lipids in the BAK-driven membrane permeabilization process. J. Biol. Chem. 286, 8213−8230. (12) Chipuk, J. E., McStay, G. P., Bharti, A., Kuwana, T., Clarke, C. J., Siskind, L. J., Obeid, L. M., and Green, D. R. (2012) Sphingolipid metabolism cooperates with BAK and BAX to promote the mitochondrial pathway of apoptosis. Cell 148, 988−1000. (13) Billen, L. P., Kokoski, C. L., Lovell, J. F., Leber, B., and Andrews, D. W. (2008) Bcl-XL inhibits membrane permeabilization by competing with Bax. PLoS Biol. 6, e147. (14) Fletcher, J. I., Meusburger, S., Hawkins, C. J., Riglar, D. T., Fairlie, W. D., Huang, D. C., and Adams, J. M. (2008) Apoptosis is triggered when prosurvival Bcl-2 proteins cannot restrain Bax. Proc. Natl. Acad. Sci. U S A 105, 18081−18087. (15) Kim, H., Rafiuddin-Shah, M., Tu, H.-C., Jeffers, J. R., Zambetti, G. P., Hsieh, J. J. D., and Cheng, E. H. Y. (2006) Hierarchical regulation of mitochondrion-dependent apoptosis by BCL-2 subfamilies. Nat. Cell Biol. 8, 1348−1357. (16) Ku, B., Liang, C., Jung, J. U., and Oh, B.-H. (2011) Evidence that inhibition of Bax activation by BCL-2 involves preferential interaction with Bax BH3 domain. Cell Res. 4, 627−641. (17) Ding, J., Zhang, Z., Rooberts, G. J., Falcone, M., Miaao, Y., Shao, Y., Zhang, X. C., Andrews, D. W., and Lin, J. (2010) Bcl-2 and Bax interact via the BH1-3 groove-BH3 motif interface and a novel interface involving the BH4 motif. J. Biol. Chem. 285, 28749−28763. (18) Brumatti, G., and Ekert, P. G. (2013) Seeking a MCL-1 inhibitor. Cell Death Differ. 20, 1440−1441. (19) Kushnareva, Y., Andreyev, A. Y., Kuwana, T., and Newmeyer, D. D. (2012) Bax activation initiates the assembly of a multimeric catalyst that facilitates Bax pore formation in mitochondrial outer membranes. PLoS Biol. 10, e1001394. (20) Czabotar, P. E., Westphal, D., Dewson, G., Ma, S., Hockings, C., Fairlie, W. D., Lee, E. F., Yao, S., Robin, A. Y., Smith, B. J., Huang, D. C., Kluck, R. M., Adams, J. M., and Colman, P. M. (2013) Bax crystal structures reveal how BH3 domains activate Bax and nucleate its oligomerization to induce apoptosis. Cell 152, 519−531. (21) Bleicken, S., Landeta, O., Landajuela, A., Basañez, G., and García-Sáez, A. J. (2013) ProapoptoticBax and Bak proteins form stable protein-permeable pores of tunable size. J. Biol. Chem. 288, 33241−33252. (22) Brouwer, J. M., Westphal, D., Dewson, G., Robin, A. Y., Uren, R. T., Bartolo, R., Thompson, G. V., Colman, P. M., Kluck, R. M., and Czabotar, P. E. (2014) Bak Core and Latch Domains Separate during Activation, and Freed Core Domains Form Symmetric Homodimers. Mol. Cell 55, 938−946. (23) Westphal, D., Dewson, G., Menard, M., Frederick, P., Iyer, S., Bartolo, R., Gibson, L., Czabotar, P. E., Smith, B. J., Adams, J. M., and Kluck, R. M. (2014) Apoptotic pore formation is associated with inplane insertion of Bak or Bax central helices into the mitochondrial outer membrane. Proc. Natl. Acad. Sci. U. S. A. 111, E4076−4085. (24) Horvath, S. E., and Daum, G. (2013) Lipids of mitochondria. Prog. Lipid Res. 52, 590−614.

ASSOCIATED CONTENT

S Supporting Information *

Supporting Figures S1−S5 and methods. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Author Contributions †

Both authors contributed equally to the work

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by grants BFU2011-28566 from the Ministerio de Economia y Competitividad, IT838-13 from Gobierno Vasco, and EHU13-74 from UPV/EHU. O.L. and J.G-V. are recipients of postdoctoral fellowships from the UPV/ EHU. I.B.Z. and H.R.-F. are recipients of a predoctoral fellowship from the Basque Government and Ministerio de Educación, respectively.



REFERENCES

(1) Czabotar, P. E., Lessene, G., Strasser, A., and Adams, J. M. (2014) Control of apoptosis by the BCL-2 protein family: implications for physiology and therapy. Nat. Rev. Mol. Cell. Biol. 15, 49−63. (2) Hardwick, J. M., Chen, Y. B., and Jonas, E. A. (2012) Multipolar functions of BCL-2 proteins link energetics to apoptosis. Trends Cell Biol. 22, 318−332. (3) Aouacheria, A., Rech de Laval, V., Combet, C., and Hardwick, J. M. (2013) Evolution of Bcl-2 homology motifs: homology versus homoplasy. Trends Cell Biol. 23, 103−111. (4) Letai, A., Bassik, M. C., Walensky, L. D., Sorcinelli, M. D., Weiler, S., and Korsmeyer, S. J. (2002) Distinct BH3 domains either sensitize or activate mitochondrial apoptosis, serving as prototype cancer therapeutics. Cancer Cell 2, 183−192. (5) Chen, L., Willis, S. N., Wei, A., Smith, B. J., Fletcher, J. I., Hinds, M. G., Colman, P. M., Day, C. L., Adams, J. M., and Huang, D. C. (2005) Differential targeting of prosurvival Bcl-2 proteins by their BH3-only ligands allows complementary apoptotic function. Mol. Cell 17, 393−403. (6) Kale, J., Liu, Q., Leber, B., and Andrews, D. W. (2012) Shedding light on apoptosis at subcellular membranes. Cell 151, 1179−1184. (7) Llambi, F., Moldoveanu, T., Tait, S. W., Bouchier-Hayes, L., Temirov, J., McCormick, L., Dillon, C. P., and Green, D. R. (2011) A K

dx.doi.org/10.1021/cb500592e | ACS Chem. Biol. XXXX, XXX, XXX−XXX

ACS Chemical Biology

Articles

(25) Johnson, A. E. (2005) Fluorescence approaches for determining protein conformations, interactions and mechanisms at membranes. Traffic 6, 1078−1092. (26) Kim, P. K., Annis, M. G., Dlugosz, P. J., Leber, B., and Andrews, D. W. (2004) During apoptosis Bcl-2 changes membrane topology at both the endoplasmic reticulum and mitochondria. Mol. Cell 14, 523− 529. (27) Aluvila, S., Mandal, T., Hustedt, E., Fajer, P., Choe, J. Y., and Oh, K. J. (2013) Organization of the mitochondrial apoptotic BAK pore: oligomerization of the BAK homodimers. J. Biol. Chem. 289, 2537−2551. (28) Dewson, G., Ma, S., Frederick, P., Hockings, C., Tan, I., Kratina, T., and Kluck, R. M. (2012) Baxdimerizes via a symmetric BH3:groove interface during apoptosis. Cell Death Differ. 19, 661−670. (29) Kyrychenko, A., and Ladokhin, A. S. (2013) Molecular dynamics simulations of depth distribution of spin-labeled phospholipids within lipid bilayer. J. Phys. Chem. B 117, 5875−5885. (30) Day, C. L., Chen, L., Richardson, S. J., Harrison, P. J., Huang, D. C., and Hinds, M. G. (2005) Solution structure of prosurvival Mcl-1 and characterization of its binding by proapoptotic BH3-only ligands. J. Biol. Chem. 280, 4738−4744. (31) Andreu-Fernández, V., Genoves, A., Lee, T. H., Stellato, M., Lucantoni, F., Orzáez, M., Mingarro, I., Aguilar, M. I., and Pérez-Payá, E. (2014) Peptides Derived from the Transmembrane Domain of Bcl2 Proteins as Potential Mitochondrial Priming Tools. ACS Chem. Biol. 9, 1799−1811. (32) Guillemin, Y., Lopez, J., Gimenez, D., Fuertes, G., Valero, J. G., Blum, L., Gonzalo, P., Salgado, J., Girard-Egrot, A., and Aouacheria, A. (2010) Active fragments from pro- and antiapoptotic BCL-2 proteins have distinct membrane behavior reflecting their functional divergence. PLoS One 5, e9066. (33) Roy, S. S., Ehrlich, A. M., Craigen, W. J., and Hajnóczky, G. (2009) VDAC2 is required for truncated BID-induced mitochondrial apoptosis by recruiting BAK to the mitochondria. EMBO Rep. 10, 1341−1347. (34) Valero, J. G., Sancey, L., Kucharczak, J., Guillemin, Y., Gimenez, D., Prudent, J., Gillet, G., Salgado, J., Coll, J. L., and Aouacheria, A. (2011) Bax-derived membrane-active peptides act as potent and direct inducers of apoptosis in cancer cells. J. Cell Sci. 124, 556−564. (35) Garcia-Saez, A. J., Coraiola, M., Dalla Serra, M., Mingarro, I., Muller, P., and Salgado, J. (2006) Peptides corresponding to helices 5 and 6 of Bax can independently form large lipid pores. FEBS J. 273, 971−981. (36) Weng, C., Li, D., Xu, D., Shi, Y., and Tang, H. (2005) Specific cleavage of Mcl-1 by caspase-3 in TRAIL-induced apoptosis in Jurkat leukemia cells. J. Biol. Chem. 280, 10491−10500. (37) Kim, J.-H., and Bae, J. (2014) MCL1-ES induces MCL-1Ldependent BAX- and BAK-independent Mitochondrial apoptosis. PLoS One 8, e79626. (38) Juin, P., Geneste, O., Gautier, F., Depil, S., and Campone, M. (2013) Decoding and unlocking the BCL-2 dependency of cancer cells. Nat. Rev. Cancer 13, 455−465. (39) Terrones, O., Etxebarria, A., Landajuela, A., Landeta, O., Antonsson, B., and Basañez, G. (2008) BIM and tBID are not mechanistically equivalent when assisting BAX to permeabilize bilayer membranes. J. Biol. Chem. 283, 7790−7803.

L

dx.doi.org/10.1021/cb500592e | ACS Chem. Biol. XXXX, XXX, XXX−XXX