Report

Posttranslational N-Myristoylation of BID as a Molecular Switch for Targeting Mitochondria and Apoptosis

See allHide authors and affiliations

Science  01 Dec 2000:
Vol. 290, Issue 5497, pp. 1761-1765
DOI: 10.1126/science.290.5497.1761

Abstract

Many apoptotic molecules relocate subcellularly in cells undergoing apoptosis. The pro-apoptotic protein BID underwent posttranslational (rather than classic cotranslational) N-myristoylation when cleavage by caspase 8 caused exposure of a glycine residue. N-myristoylation enabled the targeting of a complex of p7 and myristoylated p15 fragments of BID to artificial membranes bearing the lipid composition of mitochondria, as well as to intact mitochondria. This post-proteolytic N-myristoylation serves as an activating switch, enhancing BID-induced release of cytochrome c and cell death.

Localization of proteins to distinct subcellular compartments, including membranes, is a critical event in multiple cellular pathways such as apoptosis. Discrete topogenic sequence elements within proteins function as an address for unidirectional targeting to select membrane sites (1). Alternatively, lipid modification of proteins, including isoprenylation, myristoylation, palmitoylation, or modification by glycosyl-phosphatidylinositol, enables targeting and permits stable membrane association (2, 3). One drastic cell fate decision, apoptosis, follows signal transduction events and results in the redistribution of proteins, which often initiates their effector activity. Phosphorylation, a well-documented mechanism that can relocate proteins (4), regulates the movement of pro-apoptotic BAD from cytosol to mitochondria (5) and the movement of Forkhead transcription factor (FKHRL1) from cytosol to nucleus (6). InCaenorhabditis elegans, the pro-apoptotic molecule Egl-1 releases Ced-4 from mitochondria, which then travels to nuclear membranes (7). Site-specific cleavage of several hundred death substrates by dedicated proteases, called caspases, is a critical step in the execution phase of apoptosis (8). For example, cleavage of the chaperone ICAD releases its partner CAD (caspase-activated deoxyribonuclease), which translocates to the nucleus to degrade DNA (9, 10). Other caspase substrates include DNA repair enzymes, structural components of the cytoskeleton or nuclear scaffold, and BCL-2 family proteins that affect mitochondrial dysfunction (8,11–14). This includes the pro-apoptotic molecule BID, a member of the “BH3 domain only” subset that links proximal signals from death receptors to the common apoptotic pathway (11–13). Engagement of the receptor Fas (CD95) or of tumor necrosis factor receptor 1 (TNFR1) activates caspase 8, which cleaves the inactive cytosolic form of BID (p22), generating a truncated 15-kd fragment (tBID) (11–13) that relocates to mitochondria within 1 hour. The exposed BH3 domain of tBID (15,16) binds and oligomerizes BAK, a resident mitochondrial family member with multiple BH domains, resulting in mitochondrial dysfunction, including the release of cytochrome c (17). How BID rapidly and selectively targets the mitochondrial outer membrane remains unresolved.

Multidimensional nuclear magnetic resonance analysis indicated that uncleaved and cleaved BID have approximately the same conformation in solution, suggesting that the p7 and p15 fragments remain in a noncovalent complex after cleavage by caspase 8 (15). Consequently, we explored the mechanism by which this complex translocated to and inserted into the mitochondrial membrane. We confirmed that the p15 tBID fragment was not released when recombinant full-length p22 was cleaved by caspase 8. When NH2-terminal histidine-tagged p7 (his-p7) was released from a nickel agarose column by imidazole, the p15 tBID fragment always coeluted (Fig. 1A, lane 1), suggesting a tight noncovalent complex. The solution structure of p22 BID suggests that a hydrophobic interaction between α1 and α3 helices may be responsible (15, 16). In support of this idea, p15 was only dissociated from his-p7 when the nonionic detergentn-octyl glucoside reached its critical micelle concentration (0.6% w/v), indicating a strong hydrophobic interaction (Fig. 1A) (15, 18). We next compared the ability of intact p22, cleaved p7-p15 complex, and p15 tBID to target isolated mitochondria. The cleaved p7-p15 complex displayed little improvement in targeting over p22, whereas the isolated p15 fragment bound the mitochondrial membrane more efficiently (Fig. 1B). Treatment of the targeted mitochondria with alkali revealed that only the p15 appeared to be integral to the membrane, whereas any mitochondrial p22 was only loosely associated.

Figure 1

Analysis of the noncovalent BID complex (p7-p15). (A) Dissociation of the p7-p15 complex with octyl glucoside (26). The histidine-tagged p22 BID was cleaved by recombinant caspase 8 (lane 1) and applied to a nickel-affinity column. p15 was eluted from the column-bound p7 with the indicated concentrations of octyl glucoside in the wash buffer. His-p7 was then eluted with 1 M imidazole. The fractions were analyzed by SDS-PAGE and stained with Coomassie blue. (B) In vitro targeting of recombinant p22, p7-p15 complex, or p15 BID to isolated liver mitochondria (26). The mitochondrial pellets (P) and the supernantant (S) were separated by centrifugation. Samples were also taken from mitochondrial pellets treated with an alkaline solution [0.1 M Na2CO3 (pH 11)]. The samples were size-fractionated by SDS-PAGE and analyzed by protein immunoblot analysis with an Ab to BID.

Despite the high stability and low targeting efficiency of the p7-p15 complex in vitro, tBID localizes to mitochondria very rapidly after cleavage in vivo (13). Thus, we searched for possible posttranslational modifications of BID that would promote its translocation and membrane insertion, as it lacks the typical hydrophobic COOH-terminus that serves as a signal/anchor sequence for other BCL-2 family members (19). Phosphorylation of BID does not appear to play a role, because neither Fas nor TNFR1 activation affected the phosphorylation of BID. Subsequently, we assessed BID for lipid modification and found that BID was not palmitoylated. However, we noted that an NH2-terminal glycine would be generated upon cleavage by caspase 8 and that this conserved site on the p15 fragment (GSQASR) (20) conformed to a consensus myristoylation site (3). Although N-myristoylation of mammalian proteins has previously been described only as a cotranslational modification of nascent polypeptides (3), we tested whether BID could be myristoylated posttranslationally after cleavage by caspase 8. Jurkat cells were metabolically labeled with [3H]myristic acid, then treated with an antibody (Ab) to Fas. Immunoprecipitation of proteins from cell extracts with an Ab to BID revealed that p15 BID, but not p22, was myristoylated (Fig. 2A). Furthermore, the linkage of myristic acid to p15 was resistant to treatment with 0.2 M KOH and 1 M hydroxylamine, indicating an amide linkage characteristic of myristoylation (Fig. 2A), which can occur at NH2-terminal glycines or occasionally at internal lysine residues. We developed an in vitro myristoylation assay using rabbit reticulocyte lysate to map the site(s) of myristoylation in the p7-p15 complex. Substitution of Gly60 with Ala (G60A) abolished myristoylation of the p15 fragment completely, whereas mutation of internal lysine residues had no effect (Fig. 2, B and C). Caspase 8 cleaved recombinant G60A p22 as efficiently as it did wild-type (wt) p22 BID in vitro (Fig. 2D).

Figure 2

Characterization of BID myristoylation and identification of the modification site. (A) Metabolic labeling of Jurkat cells with [3H]myristic acid and analysis of fatty acid linkage to BID (27). Jurkat cells were labeled with [3H]myristic acid overnight and subsequently activated with an Ab to Fas. Proteins from cellular lysates prepared 1 hour after activation were immunoprecipitated with an Ab to BID and fractionated by SDS-PAGE. Gel slices were treated with 0.2 M KOH, methanol, 1 M hydroxylamine (pH 7.5), or 1 M tris (pH 7.5), and autoradiograms were developed. (B) In vitro myristoylation of the p7-p15 complex. The wt or G60A p7-p15 complex was treated with rabbit reticulocyte lysate in the presence of 20 μCi [3H]myristic acid at 30°C for 30 min. The samples were fractionated by SDS-PAGE, and an autoradiogram was developed. The input wt and G60A p7-p15 complexes were stained with Coomassie blue. (C) Mapping of the myristoylation site within BID (20). Site-directed mutagenesis was used to generate p15 BID mutants in which potential myristoylation sites were replaced. The p15 BID mutants were translated in vitro in the presence of 20 μl of [3H]myristic acid, using rabbit reticulocyte lysate. The translated products were immunoprecipitated with an Ab to BID, separated by SDS-PAGE, and exposed for autoradiography. (D) Cleavage of wt and G60A p22 by recombinant caspase 8 (20:1, w/w) at 30°C for 1 hour. The full-length p22 and the cleavage products were separated by SDS-PAGE and stained with Coomassie blue.

To study the functional significance of BID myristoylation, we tested whether N-myristoyltransferase (NMT), the enzyme responsible for cotranslational myristoylation (3), would also catalyze posttranslational myristoylation of BID. Recombinant NMT myristoylated the cleaved p7-p15 wt complex but not the G60A complex in vitro (Fig. 3A). Furthermore, myristoylation of the p7-p15 complex markedly enhanced targeting of myr-p15 to mitochondria. Whereas <30% of the p7-G60A p15 complex would associate with mitochondria, essentially 100% of p7-myr-p15 associated with the sedimented mitochondria (Fig. 3B). The myr-p15 proved alkaline-resistant (as above), which supports an integral membrane position. We next tested whether myristoylation might promote dissociation of the p7-p15 complex. A noncovalent complex of wt or G60A BID was myristoylated, and the interaction between his-p7 and p15 assessed by using nickel-agarose beads to capture p7. All of the [3H]myristoylated p15 was associated with p7 (Fig. 3C), indicating that myristoylation did not in and of itself cause dissociation of the complex. Alternatively, myristoylation of the p7-p15 complex might improve targeting to selected membranes, in particular the outer mitochondrial membrane, with its unique lipid composition (21). The p7-p15 complex did not associate with standard liposomes in the absence of myristoylation, but the p7-myr-p15 complex did (Fig. 3D). This suggests that no specific protein target is absolutely required for targeting of myr-BID. Liposomes containing cardiolipin [to reflect the lipid composition determined for the outer mitochondrial membrane at contact sites (21), where tBID clusters] displayed a modestly increased binding of the myristoylated complex (Fig. 3D). However, liposomes in which phosphatidylglycerol was substituted for cardiolipin to maintain charge showed similar binding, suggesting that the overall net negative charge, rather than individual lipids, may account for the improvement (Fig. 3D). Alternatively, a study that targeted the nonmyristoylated form of p15 tBID to liposomes proposed that cardiolipin provides a unique structure to the membrane (22). Much of the myr-p15 used here proved resistant to dissociation under alkaline conditions, whereas associated p7 was released, consistent with the nonintegral location noted for p7 at mitochondria (12). Therefore, the improved targeting of p7-myr-p15 BID to mitochondria is achieved, at least in part, by a selective interaction between myr-p15 and the mitochondrial membrane. To determine whether the improved targeting of myr-p15 enhanced mitochondrial dysfunction, we assessed the release of cytochrome c. The p7-myr-p15 complex was also more efficient at releasing cytochrome c from isolated mitochondria than was a nonmyristoylated p7-G60A p15 complex in both a time course (Fig. 3E) and dose response (Fig. 3F) assessment. Release was initiated with 0.01 ng (9 fmol) of p7-myr-p15 complex per 1 mg of purified mitochondria. Similarly, 900 fmol of p7-myr-p15 BID will maximally release cytochrome c from 1 mg of mitochondria, whereas 320 pmol of nonmyristoylated p15 is required.

Figure 3

Targeting of the myristoylated p7-p15 complex of BID to mitochondria and liposomes. (A) In vitro myristoylation of the p7-p15 complex by NMT. The wt or G60A p7-p15 complexes were treated with recombinant NMT (28) at 30°C for 30 min in the presence of 4.3 μM [3H]myristoyl CoA and were subsequently used in an in vitro targeting assay with isolated mouse liver mitochondria. The mitochondrial pellet (P) and supernatant (S) were separated by SDS-PAGE, and an autoradiogram was developed. (B) Targeting of BID to mitochondria. The mitocondrial pellet and supernatant from the same experiment shown in (A) were analyzed by protein immunoblot analysis with an Ab to BID. (C) Myristoylation does not cause the dissociation of the p7-p15 complex (28). The histidine-tagged p7-p15 complex was myristoylated in vitro by NMT in the presence of [3H]myristoyl CoA. His-p7 was sedimented with nickel agarose beads as the pellet (P) separated from the supernatant (S). Proteins were fractionated by SDS-PAGE, followed by exposure for autoradiography or protein immunoblotting. (D) Targeting of BID to liposomes (28). p7/myr-p15 (Myr-BID) and nonmyristoylated BID complexes were incubated with liposomes bearing the lipid composition of the mitochondrial outer membrane (0% CL) or of outer membrane contact sites, which include CL (25%) (21,22) or substitution of phosphatidylglycerol for CL to maintain negative charge (26% PG). The liposomes were pelleted and separated from the supernatant (S1) by ultracentrifugation, and then washed once more by either a neutral buffer [150 mM KCl and 20 mM Hepes (pH 7.0)] or an alkaline solution [0.1 M Na2CO3 (pH 11)] and ultracentrifuged again. Samples of the liposome pellet (P) and the second supernatant (S2) were fractionated by SDS-PAGE and stained with Coomassie blue. (E) Effect of myristoylation of BID-induced release of cytochrome c from mitochondria. The wt or G60A p7-p15 complex was treated with recombinant NMT in the presence of myristoyl CoA and was subsequently incubated with isolated mitochondria at 25°C for 15, 45, and 60 min (28). The supernatant was separated from the mitochondrial pellets and analyzed with an Ab to cytochrome c (Pharmingen). (F) Effect of myristoylation on dose response of BID-induced cytochrome c release from mitochondria. Increasing amounts of myristoylated wt or G60A p7-p15 complex (freshly prepared as above) were incubated with isolated mitochondria at 25°C for 45 min (28). The supernatant was separated from the mitochondrial pellets and analyzed for released cytochrome c by enzyme-linked immunosorbent asssay (Quantikine, R&D Systems).

We next addressed the functional significance of BID myristoylation in subcellular trafficking and cell death in vivo. Either wt or G60A p22 BID, each bearing a COOH-terminal green fluorescent protein (GFP) tag, was stably expressed in a bulk population of MCF7 cells, which also expressed human Fas. A similar amount of p15-GFP fragment was generated from wt or G60A BID cells activated with Ab to Fas (Fig. 4A). Confocal microscopy revealed that most cells expressing wt BID-GFP displayed a redistribution of GFP after Fas activation from a diffuse cytosolic distribution to a clustered localization that was coincident with mitochondria, as assessed with Mitotracker stain (Fig. 4B). In contrast, most cells bearing G60A BID-GFP retained a diffuse cytosolic localization of GFP after Fas activation (Fig. 4B). Furthermore, only the additional wt BID-GFP, but not comparable amounts of the G60A mutant, accelerated apoptosis of MCF7 cells after Fas activation (Fig. 4C). Taken together, these data indicate that myristoylation of BID promotes its targeting to mitochondria as well as enhances its pro-apoptotic activity in vivo.

Figure 4

Decreased translocation to mitochondria and reduced pro-apoptotic activity of the nonmyristoylatable mutant of BID. (A) Protein immunoblot analysis of BID in Fas-activated MCF7 cells. MCF7 cells expressing human Fas were infected with recombinant retrovirus possessing either full-length wt or G60A BID bearing a COOH-terminal GFP tag and a puromycin-resistance gene. The bulk population of puromycin-resistant MCF7 cells was treated with Ab to Fas and cyclohexmide for 4 hours, and total cell lysates were analyzed by protein immunoblotting with an Ab to BID. The positions of full-length BID-GFP (FL) and p15-GFP (cleaved) are indicated by arrows. (B) Subcellular localization of wt versus G60A BID. BID-GFP–expressing MCF7 cells were either treated with Ab to Fas and cycloheximide for 4 hours or left untreated. Mitochondria were identified with 150 nM Mitotracker, which was incubated with the cells for 30 min before fixation. The subcellular colocalization of BID-GFP (green fluorescence) and mitochondria (red fluorescence) was assessed by confocal microscopy. (C) Viability of the retrovirus-infected MCF7 cells after Fas activation. The MCF7 cells with wt BID, G60A BID, or the empty vector control (puro) were treated with an Ab to Fas and cycloheximide. At indicated time points, cells were stained with propidium iodide (PI, 1 μg/ml), and the percentage of the viable cells (PI-negative population) was determined by flow cytometry. Data reflect duplicate assays and are representative of two independent puromycin-selected populations of MCF7 cells expressing wt or G60A BID-GFP (29).

BID provides an example of an unexpected posttranslational N-myristoylation, resulting in a selective pathway of subcellular trafficking. Sixteen of ∼60 identified caspase substrates expose an NH2-terminal glycine upon cleavage (8), which, combined with the loose consensus motif for myristoylation, suggests that this paradigm could prove a common modification in apoptosis and perhaps other proteolytic pathways. Assessment of the p7-myr-p15 BID complex indicates that N-myristoylation has a strong influence on BID selecting mitochondria, inserting into the outer membrane, releasing cytochrome c, and killing cells. Specifically, the enhanced movement of myristoylated BID from cytosol to mitochondria in vivo is apparently compounded by improved insertion of myristoylated BID into membranes. Myristoylation has been associated with proteins in other membrane compartments and also been observed to modulate protein-protein or protein-lipid interactions. Cotranslational myristoylation of NADH (the reduced form of nicotinadimide adenine dinucleotide) cytochrome b5 reductase does appear to be required for its mitochondrial localization (23). Taken together, the myristic acid moiety itself is unlikely to be the selective targeting motif; instead, the N-myristoylation of the BID complex appears to promote a protein conformation that favors the mitochondrial outer membrane. Thus, the N-myristoylation of BID after proteolytic processing represents a molecular switch that helps ensure the next critical step in apoptosis: the release of cytochrome c and subsequent cell demise.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: stanley_korsmeyer{at}dfci.harvard.edu

REFERENCES AND NOTES

View Abstract

Navigate This Article