Mechanism of apoptosis induction by inhibition of the anti-apoptotic BCL-2 proteins Jerry E. Chipuka, John C. Fisherb, Christopher P. Dillona, Richard W. Kriwackib,c, Tomomi Kuwanad,1, and Douglas R. Greena,1 aDepartment
of Immunology and bDepartment of Structural Biology, St. Jude Children’s Research Hospital, Memphis, Tennessee 38105; cDepartment of Molecular Sciences, University of Tennessee Health Sciences Center, Memphis, Tennessee 38163; and dDepartment of Pathology, University of Iowa, Carver College of Medicine, Iowa City, Iowa 52242
Normal cellular lifespan is contingent upon preserving outer mitochondrial membrane (OMM) integrity, as permeabilization promotes apoptosis. BCL-2 family proteins control mitochondrial outer membrane permeabilization (MOMP) by regulating the activation of the pro-apoptotic BCL-2 effector molecules, BAX and BAK. Sustainable cellular stress induces proteins (e.g., BID, BIM, and cytosolic p53) capable of directly activating BAX and/or BAK, but these direct activators are sequestered by the anti-apoptotic BCL-2 proteins (e.g., BCL-2, BCL-xL, and MCL-1). In the event of accumulated or marked cellular stress, a coordinated effort between previously sequestered and nascent BH3-only proteins inhibits the anti-apoptotic BCL-2 repertoire to promote direct activator proteinmediated MOMP. We examined the effect of ABT-737, a BCL-2 antagonist, and PUMA, a BH3-only protein that inhibits the entire anti-apoptotic BCL-2 repertoire, with cells and mitochondria that sequestered direct activator proteins. ABT-737 and PUMA cooperated with sequestered direct activator proteins to promote MOMP and apoptosis, which in the absence of ABT-737 or PUMA did not influence OMM integrity or cellular survival. Our data show that the induction of apoptosis by inhibition of the anti-apoptotic BCL-2 repertoire requires ‘‘covert’’ levels of direct activators of BAX and BAK at the OMM. BCL-2 family 兩 mitochondria 兩 MOMP 兩 PUMA
T
he mitochondrial pathway of apoptosis requires the release of cytochrome c from the mitochondrial intermembrane space to the cytosol (1, 2). Once released, cytochrome c cooperates with the adaptor protein, APAF-1, to promote the activation of caspases, which are required for the rapid recognition and clearance of the stressed cell. The major function of the BCL-2 family of proteins is to control the integrity of the outer mitochondrial membrane (OMM) (3, 4). The pro-apoptotic multidomain BCL-2 effector proteins, BAX and BAK, oligomerize into proteolipid pores and permeabilize the OMM, allowing the efflux of cytochrome c and other intermembrane space proteins to the cytosol during apoptosis (4, 5). The activation of BAX and BAK, to insert, oligomerize and permeabilize the OMM is a function of the BH3-only proteins, which are further classified into direct activators and derepressors/ sensitizers (6, 7). Direct activator BH3-only proteins, such as BID and BIM, activate BAX and BAK at the OMM leading to cytochrome c release (6–8). BH3 domain peptides derived from BID and BIM behave similarly to the intact proteins as they also induce BAX and BAK oligomerization and pore-forming activity in the absence of additional mitochondrial proteins (6, 7). A few non-BCL-2 family proteins are also described and can have direct activator function, perhaps most clearly demonstrated in the case of cytosolic p53 (9, 10). Conversely, the derepressor/sensitizer BH3-only proteins (e.g., BAD, BIK, BMF, HRK, and Noxa) fail to directly induce BAX and BAK activation, but efficiently release sequestered direct activator proteins from anti-apoptotic BCL-2 members, such as BCL-2, BCL-xL and MCL-1, to promote mitochondrial outer membrane permeabilization (MOMP) (6, 7, 11). However, most of www.pnas.org兾cgi兾doi兾10.1073兾pnas.0808036105
the data concerning derepressor/sensitizer BH3-only protein function were obtained solely from the use of synthesized BH3 domain peptides; there is little information about full-length derepressor/ sensitizer BH3-only proteins. A small molecule BH3 domain peptide mimetic, ABT-737, acts similarly to derepressor/sensitizer BH3-only peptides (e.g., BAD) by rapidly inducing direct activator dependent MOMP in some tumor model systems (12, 13). The potency and selectivity of ABT-737 supports the notion that tumor cells become addicted to anti-apoptotic proteins which sequester direct activators induced during oncogenesis and provides in vivo evidence that a direct activator:anti-apoptotic:derepressor (e.g., BIM:BCL-2:ABT-737) network regulates MOMP (14). As an example, cells derived from chronic lymphocytic leukemia (CLL) constitutively express BIM that must be tonically inhibited by BCL-2 or MCL-1 to escape MOMP and ensure tumor maintenance; the inhibition by BCL-2 can be overcome by ABT-737 treatment leading to BIM release and activity (13). The BH3-only protein PUMA (there are two major isoforms, ␣ and , which share identical BH3 regions and similar kinetics for the induction of apoptosis) promotes MOMP in numerous cellular stress scenarios, such as cytokine deprivation and DNA damage (15–17). Original observations on PUMA indicate that it is a potent inducer of cell death, perhaps solely due to the inhibition of the anti-apoptotic BCL-2 repertoire or through cooperation with direct activator proteins (11, 15–18). Here, we describe several cellular and in vitro derepressor/sensitizer model systems to investigate the synergy between ABT-737 or PUMA with direct activator proteins. In addition, we provide evidence that PUMA can reveal covert direct activator BH3-only protein function on mitochondria derived from healthy tissue. Results and Discussion Following pro-apoptotic stimulation, direct activator BH3-only proteins gain function to promote MOMP and apoptosis; for example, full-length BID is cleaved by caspase 8 to generate an active BID protein (caspase-8-cleaved BID, C8-BID). HeLa cells treated with TNF undergo BID-dependent cytochrome c release and apoptosis (19). In the experiment shown in Fig. 1A, HeLa cells expressing cytochrome c-GFP were continuously treated with sublethal doses of TNF as follows: (Clones 1–6) untreated, 10 g/ml cycloheximide alone, cycloheximide ⫹ 0.1, 0.5, 1 or 2 ng/ml TNF. Cycloheximide is required to abrogate TNF-induced NF-BAuthor contributions: J.E.C., J.C.F., T.K., and D.R.G. designed research; J.E.C., J.C.F., and T.K. performed research; J.E.C., J.C.F., C.P.D., R.W.K., and T.K. contributed new reagents/ analytic tools; J.E.C., J.C.F., C.P.D., R.W.K., T.K., and D.R.G. analyzed data; and J.E.C., J.C.F., R.W.K., T.K., and D.R.G. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. 1To
whom correspondence may be addressed. E-mail:
[email protected] or douglas.
[email protected].
This article contains supporting information online at www.pnas.org/cgi/content/full/ 0808036105/DCSupplemental. © 2008 by The National Academy of Sciences of the USA
PNAS 兩 December 23, 2008 兩 vol. 105 兩 no. 51 兩 20327–20332
CELL BIOLOGY
Edited by Tak Wah Mak, University of Toronto, Toronto, ON, Canada, and approved October 14, 2008 (received for review August 15, 2008)
1
2
3
4
5
6
Clone
2
TNF (ng/ml)
CHX (10 µg/ml) 0
0
0.1 0.5
1
B
- FL-BID
20 kDa15 kDa10 kDa-
25 % AnnexinV-PE
A
- p15 - p7
Short exposure
20 15 10 5
- FL-BID
20 kDa15 kDa10 kDaLong exposure
C
0 Clone 1
- p15 - p7
30 20
+
6
50
2
3
4
5
0 Clone 1
6
2
+ ABT-737
Dextran
5
25
10
Cyto c
4
75 % MOMP
% AnnexinV-PE
40
E
3
D 100
50
0 Clone 1
2
Merge
3
4
5
6
+ PUMA
Clone 1
Cyto c
Merge
Clone 6
Dextran
Fig. 1. Cells tolerate sustained BID activation until the anti-apoptotic BCL-2 repertoire is inhibited by ABT-737 or PUMA. An intact cellular derepression model system was established using sublethal TNF treatment (0.1, 0.5, 1, and 2 ng/ml) with endogenous BID. Cytochrome c-GFP expressing HeLa cells were continually pulsed (6 h treatment, 42 h recovery) with 10 g/ml cycloheximide (CHX) and sublethal doses of TNF. (A) RIPA lysates of the indicated TNF-pulsed clones were subjected to SDS/PAGE and Western blot analysis to observe TNFinduced BID cleavage. (B) The TNF-pulsed clones were treated and analyzed 4 h later for survival with AnnexinV-PE. To ensure the TNF was effective, the parental cytochrome c-GFP expressing HeLa cells were treated with 10 g/ml CHX and 20 ng/ml TNF (indicated with a ‘‘⫹’’). (C) The TNF-pulsed clones were treated with 1 M ABT-737 for 4 h before AnnexinV-PE analysis. (D–E) The TNF-pulsed clones were microinjected with PUMA (0.475 g/l needle concentration; a range of 10 –50 fl injected per cell, approximately 0.25–1.25 nM intracellular concentration) in the presence of 20 M Q-VD-OPh (to prevent cellular morphological changes downstream of MOMP) and cultured for 3 h before confocal imaging. Texas Red dextran was added to the PUMA solution to identify microinjected cells. Cells with permeabilized mitochondria display diffuse cytochrome c-GFP (compare the cytochrome c-GFP pattern in clone 1 to clone 6 in E). The percentage of microinjected cells with diffuse cytochrome c-GFP is indicated in D. Error bars represent the standard deviation from triplicate data.
dependent caspase inhibition (20). Cells were treated for 6 h, then untreated for 42 h and this cycle was repeated through several rounds. After this TNF treatment, cells displayed the accumulation of C8-BID (a complex containing two fragments: amino terminal p7 and a carboxyl terminal p15); however these doses of TNF did not induce cell death (Fig. 1B). The TNF-treated clones were then analyzed for apoptosis 4 h after the addition of 1 M ABT-737. ABT-737 induced marked apoptosis in the TNF-treated clones, with clone 6 showing the greatest response (Fig. 1C). Since ABT737 inhibits only a subset of the anti-apoptotic BCL-2 repertoire (BCL-2, BCL-xL, and BCL-w) (12), the complete inhibition of the anti-apoptotic BCL-2 repertoire [BCL-2-like and MCL-1-like members which are both present in these cells, supporting information (SI) Fig. S1 A] was examined with highly active full-length, recombinant PUMA (the activity was determined as binding to BCL-xL⌬C by NMR spectroscopy, Fig. S2 A–D). The clones were 20328 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0808036105
microinjected with PUMA (and Texas Red dextran to mark the microinjected cells), cultured for 3 h, and analyzed for MOMP by imaging the punctate (mitochondrial) to a diffuse cytosolic (released from mitochondria) transition of cytochrome c-GFP by confocal microscopy (Fig. 1 D-E). Purified heavy membrane fractions (enriched for mitochondria) from these cells released cytochrome-GFP and endogenous cytochrome c with identical direct activator concentrations and kinetics (Fig. S1B); therefore, cytochrome c-GFP is an accurate measure of MOMP. Cells expressing C8-BID released cytochrome c-GFP rapidly, with clones 5 and 6 displaying almost 100% diffuse cytochrome c staining after only 60 min. The microinjected cells were also treated with 20 M quinolylvalyl-O-methylaspartyl-[2,6-difluorophenoxy]-methyl ketone (QVD-OPh), a pan-caspase inhibitor, to block caspase-dependent morphological changes that prevent image capture. In the absence of Q-VD-OPh, cells with diffuse cytochrome c-GFP rapidly blebbed and detached from the coverslip, indicating apoptosis had proceeded under these conditions (data not shown). Heavy membrane fractions from the TNF-treated clones sequestered the active carboxyl terminal p15 fragment of BID, which was partially or completely released by coincubation with 100 nM ABT-737 or 100 nM PUMA, respectively (Fig. 2A). The release of p15 BID from the heavy membrane fractions paralleled the cellular activity of ABT-737 and PUMA to induce MOMP and apoptosis in Fig. 1 C-E. Equal heavy membrane loading was confirmed by similar BCL-2 expression in each lane (Fig. 2 A). The same heavy membrane fractions were also analyzed for cytochrome c release induced by ABT-737, PUMA or the PUMA BH3 domain peptide (Fig. 2B). Again, there was a correlation between expression, sequestration and release of p15 BID with cytochrome c release. Partial inhibition of the anti-apoptotic BCL-2 repertoire by ABT-737 treatment resulted in only partial cytochrome c release; complete inhibition by PUMA or the PUMA BH3 domain peptide resulted in almost complete cytochrome c release (Fig. 2B). As controls, the heavy membranes from each clone responded similarly to recombinant C8-BID, demonstrating equivalent capabilities to release cytochrome c. The requirement for more BH3 domain peptide compared to full-length protein to observe BH3only protein function is consistent with other observations of direct activator and sensitizer BH3-only proteins (6, 7, 10). To determine if the p15 BID was sequestered by anti-apoptotic BCL-2 proteins within the isolated heavy membrane fractions, BCL-2 and MCL-1 were separately immunoprecipitated and analyzed for associated p15 BID. Both anti-apoptotic proteins coimmunoprecipitated with p15 BID, which was not seen when PUMA was present (Fig. 2C). Similarly, ABT-737 prevented coimmunoprecipitation of BCL-2 and p15 BID, but not MCL-1 and p15 BID (Fig. 2C). These data suggest the following scenario: (i) in cells, anti-apoptotic BCL-2 proteins on the OMM can sequester substantial direct activator BH3-only protein function to inhibit MOMP, cytochrome c release, and apoptosis; and (ii) complete inhibition of the anti-apoptotic BCL-2 repertoire by PUMA can release these direct activator proteins from the OMM to rapidly induce MOMP, cytochrome c release and apoptosis. The ability of PUMA to derepress sequestered direct activator molecules at the OMM is similar to a cellular scenario termed, ‘‘primed for death.’’ (14) For example, CLL cells that undergo apoptosis in response to ABT-737 have been shown to harbor the direct activator, BIM, that appears to be derepressed by ABT-737 treatment (13, 14). This primed for death derepression scenario (Fig. 3A) was recapitulated in vitro using C57BL/6 liver mitochondria loaded with different direct activators of BAX and/or BAK and then treated with PUMA or the PUMA BH3 domain peptide (Fig. 3 B–D). Active forms or peptides from three established direct activators were used: BID, BIM, and p53. Mitochondria were incubated with recombinant C8-BID (Fig. 3B, 5–25 pM), BIM BH3 domain peptide (Fig. 3C, 10–100 nM) or cytosolic p53 (Fig. 3D, 10–100 pM) at concentrations too low to induce MOMP, and then Chipuk et al.
Heavy Membranes 1 2 3 4 5
6 1 2 3
4 5 6 1
2 3 4 5 6
25 kDa -
Clone BCL-2
BID
-p15
15 kDa PUMA (100 nM)
ABT-737 (100 nM)
ABT-737
PUMA BH3
PUMA
Total
Buffer
B
C8-BID
DMSO
Clone 1
(s)
2
(s)
3
(s)
4
(s)
5
(s) (s)
6
IP: BCL-2
PUMA
ABT-737
DMSO
PUMA
ABT-737
C
DMSO
WB: Cytochrome c C8-BID (1 nM) ABT-737 (10 & 100 nM) PUMA (10 & 100 nM) PUMA BH3 peptide (100 & 1000 nM)
BCL-2
MCL-1
BID (p15)
BID (p15) IP: MCL-1
Fig. 2. The outer mitochondrial membrane sequesters cleaved BID that promotes cytochrome c release when the anti-apoptotic BCL-2 repertoire is inhibited by ABT-737, PUMA, or the PUMA BH3 domain peptide. (A) Heavy membrane fractions were isolated from the TNF-pulsed clones, treated with 100 nM ABT-737 or 100 nM PUMA (or DMSO vehicle) for 30 min, washed, solubilized, and subjected to SDS/PAGE and Western blot analysis for anti-apoptotic BCL-2 member-associated activated BID (carboxyl terminus p15 fragment). Equal loading of mitochondrial protein is shown by BCL-2 expression. (B) Heavy membrane fractions were isolated from the TNF-pulsed clones and treated C8-BID, ABT-737, PUMA, or PUMA BH3-domain peptide for 60 min at 37 °C. The soluble fraction was subjected to SDS/PAGE and Western blot analysis for cytochrome c. Total cytochrome c was determined by a sample containing mitochondria solubilized in 1% CHAPS. (C) CHAPS solubilized heavy membranes from TNF-treated HeLa clone 6 were subjected to anti-BCL-2 (Left) or anti-MCL-1 (Right) immunoprecipitation in the presence of DMSO, 100 nM ABT-737, or 100 nM PUMA. The coimmunoprecipitated protein complexes were subjected to SDS/PAGE and Western blot analyses for BCL-2, MCL-1 and p15 BID.
washed to remove any unbound direct activator. The treated mitochondria were then resuspended in buffer containing PUMA (500 nM) or PUMA BH3 domain peptide (500 nM). In the p53 samples, 40 nM recombinant monomeric BAX was added along with the derepressor as we have not observed substantial p53induced BAK activation (9). PUMA or the PUMA BH3 domain peptide did not cause cytochrome c release on its own, but when combined with direct activator pretreatment, each induced efficient, complete release (Fig. 3 B–D). Similarly, the addition of PUMA or the PUMA BH3 domain peptide plus recombinant BAX did not induce MOMP unless mitochondria were previously treated with p53 (Fig. 3D). The results support the notion that PUMA can release direct activator proteins or peptides from the OMM to induce MOMP in a manner similar to that observed with mitochondria derived from tumor cells (13, 14). Also, PUMA did Chipuk et al.
not discriminate between the direct activators; C8-BID, BIM BH3 peptide, and cytosolic p53 were equally derepressed from the anti-apoptotic BCL-2 repertoire to activate BAX and/or BAK and induce MOMP. Complementary to the primed for death scenario above is the ‘‘sensitized for death’’ function of BH3-only proteins by which it is hypothesized that inhibition of the anti-apoptotic BCL-2 repertoire increases mitochondrial sensitivity to direct activator BH3-only proteins (Fig. 3E). To examine this scenario, C57BL/6 mitochondria were treated with PUMA (250 nM) or the PUMA BH3 domain peptide (250 nM), washed, and then treated with subMOMPinducing concentrations of C8-BID (Fig. 3F, 5–25 pM), BIM BH3 domain peptide (Fig. 3G, 10–100 nM) or cytosolic p53 (Fig. 3H, 10–100 pM). Pretreatment with either PUMA or the PUMA BH3 domain peptide sensitized mitochondria to direct activator proteins by approximately 100–200-fold (e.g., 1 nM C8-BID is normally required for complete MOMP; 5–10 pM C8-BID were sufficient to induce MOMP from PUMA pretreated mitochondria). In contrast, an additional treatment of PUMA (500 nM final) or the PUMA BH3 domain peptide (500 nM final) without direct activators did not cause cytochrome c release even though this was 50,000-fold excess compared to the 5 pM C8-BID treatment (Fig. 3 F-G, ‘‘2X PUMA’’ and ‘‘2X PUMA BH3’’). Therefore, in addition to a derepression function, PUMA can also regulate MOMP by sensitizing mitochondria to a normally tolerated dose of direct activator protein stimulation. These data biochemically define PUMA as a derepressor/sensitizer BH3-only protein. Next, we used three defined derepression model systems with purified recombinant proteins to observe PUMA cooperation within a direct activator:anti-apoptotic:derepressor network in vitro. In the first, C57BL/6 (⬍ 3 months old) liver mitochondria (which express BCL-2, BCL-xL, MCL-1, and BAK, but no detectable BH3-only proteins, data not shown) were induced to release cytochrome c with C8-BID, and this was inhibited by either recombinant BCL-xL lacking the C terminus (BCL-xL⌬C) or recombinant MCL-1⌬C (Fig. 4A). Addition of PUMA or the PUMA BH3 domain peptide inhibited the anti-apoptotic effects to reveal C8-BID activity (Fig. 4A). As a control, the BAD BH3 domain peptide, which can bind BCL-xL but not MCL-1 (7, 21), derepressed only the C8-BID:BCL-xL⌬C complex and not C8BID:MCL-1⌬C to promote MOMP (Fig. 4A). In the absence of C8-BID, no effects of PUMA, PUMA BH3 peptide, or BAD BH3 peptide were observed. We also used a cellular derepression model system in which HeLa cells stably express cytochrome c-GFP. These cells do not constitutively express and sequester direct activator proteins (the parental cells to the TNF-treated clones in Fig. 1) (7). HeLa cells microinjected with C8-BID underwent complete cytochrome c release, which was defined by the conversion of a punctate (mitochondrial) to a diffuse cytosolic (released from mitochondria) pattern of the cytochrome c-GFP (Fig. 4B). This release was inhibited by coinjection of BCL-xL⌬C; and derepression of the C8-BID:BCL-xL⌬C complex occurred when recombinant PUMA was added (Fig. 4B). Texas Red dextran was combined with all protein solutions to identify cells that were microinjected. Neither full-length BID (FL-BID) nor PUMA alone induced a diffuse cytochrome c-GFP pattern (Fig. 4B). The percentage of cells responding to each microinjection scenario is indicated in the representative cell image. Similar results were also observed for PUMA-mediated derepression of the C8-BID:MCL-1⌬C complex (data not shown). Finally, we used a large unilamellar vesicle (LUV) model system that faithfully mimics BCL-2 family dependent and regulated BAX activation and MOMP (7, 8). LUVs containing fluorescent-dextran (F-dextran) permeabilize and release their contents upon cooperation between BAX and a direct activator protein. N/C-BID, a well-characterized variant of C8-BID (N/C-BID is purified fulllength BID activated via a thrombin cleavage site in place of the PNAS 兩 December 23, 2008 兩 vol. 105 兩 no. 51 兩 20329
CELL BIOLOGY
A
E
“primed for death”
“sensitized for death”
mitochondria sequester direct activators
mitochondria sequester de-repressor/sensitizer
e.g., BCL-2/BCL-xL/MCL-1 sequesters C8-BID, BIM or p53
e.g., BCL-2/BCL-xL/MCL-1 are inhibited by PUMA
add PUMA [excess]
add C8-BID, BIM BH3 or p53 [sub-MOMP] • BCL-2/BCL-xL/MCL-1 cannot inhibit C8-BID, BIM BH3 or p53
• BCL-2/BCL-xL/MCL-1 associate with PUMA
• C8-BID, BIM BH3 or p53 activates BAX and/or BAK
DMSO
C8-BID
PUMA
C8-BID
PUMA BH3
C8-BID
2X PUMA
PUMA
C8-BID C8-BID PUMA PUMA BH3
Buffer
C8-BID DMSO
Total
Buffer PUMA PUMA BH3
B
Pretreatment
F
Pretreatment
PUMA BH3
• C8-BID, BIM BH3 or p53 liberated to activate BAX and/or BAK
Total
• Excess PUMA competes for BCL-2/BCL-xL/MCL-1
2X PUMA BH3
A
(s)
(s)
C57Bl/6
BIM
PUMA BH3
BIM
Total
BIM
PUMA
2X PUMA BH3
DMSO
2X PUMA
PUMA BH3
BIM BIM PUMA PUMA BH3
PUMA
BIM DMSO
Buffer
G Total
C
Buffer PUMA PUMA BH3
C57Bl/6
(s)
(s)
p53
PUMA p53
PUMA BH3 p53
2X PUMA BH3
DMSO
2X PUMA
PUMA BH3
PUMA BH3
PUMA
p53
PUMA
DMSO
p53
Buffer
H p53
Total
Buffer PUMA PUMA BH3
D
Total
C57Bl/6
C57Bl/6
(s) WB: Cytochrome c PUMA (500 nM) PUMA BH3 peptide (500 nM) C8-BID (5, 10 & 25 pM) BIM BH3 peptide (10, 50 & 100 nM) p53 (10, 50 & 100 pM) BAX (40 nM) added to the p53 panel
C57Bl/6
(s) WB: Cytochrome c C8-BID (5, 10 & 25 pM) BIM BH3 peptide (10, 50 & 100 nM) p53 (10, 50 & 100 pM) PUMA (250 nM) PUMA BH3 peptide (250 nM) BAX (40 nM) added to the p53 panel
C57Bl/6
Fig. 3. PUMA releases direct activator proteins sequestered on the OMM to induce MOMP and sensitizes mitochondria to future direct activator protein stimulation. (A) The primed for death derepression model system. In this situation, anti-apoptotic BCL-2 proteins actively sequester occult direct activator proteins, which reveal their activity to engage MOMP after derepressor/sensitizer protein addition. Experimentally, mitochondria are preloaded with evasive levels of C8-BID, BIM BH3 domain peptide, or p53 before derepressor/sensitizer stimulation. (B–D) Isolated C57BL/6 mitochondrial fractions were loaded with direct activators: C8-BID (B), BIM BH3 domain peptide (C) or p53 (D) by incubation at 37 °C for 30 min. Mitochondria were then pelleted, washed twice in MAB, resuspended in MAB containing DMSO (vehicle), PUMA or the PUMA BH3 domain peptide, incubated for 60 min at 37 °C, and fractionated for the supernatant. The entire p53 panel also contains BAX protein in the resuspension step. (E) The sensitized for death model system. In this situation, anti-apoptotic BCL-2 proteins are saturated with derepressor proteins allowing trivial direct activator stimulation to escape inhibition and MOMP is engaged. Experimentally, mitochondria are pretreated with a derepressor/sensitizer before direct activator stimulation. (F–H) Isolated C57BL/6 mitochondrial fractions were pretreated with DMSO (vehicle), PUMA or PUMA BH3 domain peptide for 30 min at 37 °C, then the direct activator proteins, C8-BID (F), BIM BH3 domain peptide (G) or p53 (H) were added, incubated for 60 min at 37 °C and the samples were fractionated for the supernatant. Instead of adding a direct activator protein, another dose of PUMA or PUMA BH3 domain peptide (500 nM final) was added to the ‘‘2X PUMA’’ and ‘‘2X PUMA BH3’’ samples, respectively. The entire p53 panel also contains BAX protein. Total cytochrome c was determined by a sample containing mitochondria solubilized in 1% CHAPS.
caspase 8 site (22)) acted with similar kinetics and concentrations as C8-BID (data not shown). BAX plus 0.045 M N/C-BID efficiently induced the release of F-dextran, which was inhibited by either full-length BCL-xL or MCL-1 (Fig. 4C). The addition of PUMA derepressed both the N/C-BID:BCL-xL and N/CBID:MCL-1 complexes to promote LUV permeabilization (Fig. 4C). To ensure N/C-BID was responsible for LUV permeabilization, the ability of PUMA to activate BAX was also measured. PUMA at concentrations similar to maximal release induced by N/C-BID (0.01–0.1 M) failed to synergize with BAX, and higher concentrations of PUMA (up to 6 M, 133-fold more than 0.045 M N/C-BID) also displayed minimal direct activator function in these assays (Fig. 4C). The PUMA BH3 domain peptide also weakly activated of BAX in this system (7). PUMA only promoted BAX activation when a direct activator:anti-apoptotic complex was present and PUMA could fully derepress BCL-xL or MCL-1mediated inhibition of N/C-BID. These data further support the hypothesis that PUMA functions within a direct activator:anti-apoptotic:derepressor network to reveal direct activator BH3-only 20330 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0808036105
protein function using isolated mitochondrial (Fig. 4A), cellular (Fig. 4B), and LUV (Fig. 4C) derepression model systems. The original descriptions of PUMA suggested that it was the key pro-apoptotic p53 target gene required for p53 dependent, DNAdamage induced apoptosis as exogenous over-expression of PUMA was sufficient to induce cell death in p53 deficient cells (15–17). To determine if PUMA may function via derepression of covert direct activator BH3-only proteins in untreated, healthy, proliferating cells, we transiently transfected SV40 large T antigen expressing wild-type, bid⫺/⫺, bim⫺/⫺ or bid⫺/⫺bim⫺/⫺ mouse embryonic fibroblasts (MEFs) with an increasing amount (0, 10, 25, 50, and 100 ng) of PUMA␣ cDNA before assaying for apoptosis by annexin V staining and flow cytometry. Wild-type MEFs underwent a dosedependent increase in annexin V staining and loss of survival (Fig. S3a), whereas genetic deletion of bid or bim, but more significantly the latter, produced resistance to exogenous PUMA␣ expression. We also compared the effects of PUMA␣ and PUMA transient expression in the same panel of MEFs and found the range of 50–100 ng of PUMA␣ or PUMA cDNA allowed for reproducible Chipuk et al.
50
25
25
SV40-MEFs
FL-BID
(s)
Cyto c-GFP
C8-BID 89.9%
BCL-xL C 5.9%
PUMA 4.6%
C8-BID BCL-xL C 5.9%
0
BIM-EL 0 50 PUMA cDNA (ng)
D
BH3
BID
BID
Buffer C8-BID
C8-BID BCL-xL C PUMA 89.7%
0 50 PUMA cDNA (ng) Protein
Buffer
C
B FL-BID 9.1%
0
C57Bl/6
Total
WB: Cytochrome c C8-BID (1 nM) BCL-xL C & MCL-1 C (25 nM) PUMA (500 nM) PUMA & BAD BH3 peptides (500 nM)
Actin
Protein PUMA
(s)
Wild-type
Wild-type (s)
(s)
bid-/-
bid-/(s)
(s)
Merge
bim-/(s)
bid-/-bim-/WB: Cytochrome c C8-BID (.1, .5 & 1 nM) BID BH3 peptide (100, 250 & 500 nM)
(s)
Normal exposures
bid-/-bim-/-
(s)
80
Wild-type
40
(s)
bid-/-bim-/-
Wild-type
Heavy Membranes
bim-/-
E
60
bid-/-
bid-/(s)
bim-/FL-BID (s)
20
0
BIM-EL BAK
+ + + + + - + - - -
+ + + + + - + + + -
+ + + + + - + + + -
Direct activation
BCL-xL de-repression
MCL-1 de-repression
- - .01 .05 .1 - - - - - - - - -
- - - 6 6 - - + + - - - - -
- - - 6 - - - - - + +
BAX (.16 µM) N/C-BID (.045 µM) 6 PUMA (µM) - BCL-xL (1.5 µM) - MCL-1 (.25 µM)
Fig. 4. PUMA promotes MOMP by functioning within an established BCL-2 family network containing direct activator BH3-only, anti-apoptotic and effector proteins. (A) C57BL/6 mitochondria were incubated with indicated proteins for 60 min at 37 °C and fractionated for the supernatant. Total cytochrome c was determined by a sample containing mitochondria solubilized in 1% CHAPS. (B) Cytochrome c-GFP expressing HeLa cells were microinjected with indicated proteins (microinjection gives an approximate range of 0.054 – 0.27 nM C8-BID, 0.37–1.85 nM BCL-xL⌬C, and 0.25–1.25 nM PUMA) in the presence of 20 M Q-VD-OPh (to prevent cellular morphological changes downstream of MOMP), and cultured for 3 h at 37 °C before imaging. Texas Red dextran was added to protein solutions to identify microinjected cells. The percentages of microinjected cells with diffuse cytochrome c-GFP are indicated and represent at least 200 cells. (C) Large unilamellar vesicles (LUVs) were incubated with indicated combinations of BAX, N/C-BID, BCL-xL, MCL-, and PUMA (all proteins for this assay are full-length) to observe PUMAmediated derepression of BCL-xL (Middle) and MCL-1 (Right). N/C-BID maximally synergized with BAX at 0.045 M. Compared to N/C-BID, similar concentrations of PUMA (.01– 0.1 M) did not induce BAX-mediated LUV permeabilization (Left). A marked excess of PUMA also failed to efficiently activate BAX compared to N/C-BID (0.045 M N/C-BID verses 6 M PUMA, 150⫻ higher concentration yet minimal direct activation). The error bars in C represent the standard deviation from triplicate data.
differences among the indicated MEFs (Fig. 5A, Fig. S3A). Increasing the dose of transfected PUMA␣ or PUMA cDNA ten- to twenty-fold (i.e., 1000 ng) markedly reduced the differences between the genotypes (Fig. S3B), which likely parallels the first observations on PUMA-induced apoptosis. Western blot analysis of whole cell lysates confirmed similar exogenous expression of PUMA (Fig. S3C) and indicated genetic deletions (Fig. 5B) in the MEF panel. For comparison, endogenous PUMA␣ expression in wild-type MEFs following UV radiation, actinomycin D and VP16 was similar to the levels achieved by approximately 50–100 ng of Chipuk et al.
Over-exposures
bid-/-bim-/-
WB: Cytochrome c C8-BID (1 nM) PUMA (.25, .5, 1 & 5 µM) PUMA BH3 peptide (1, 5, 25 & 100 µM)
Fig. 5. PUMA reveals covert direct activator BH3-only protein expression within wild-type MEFs and primary liver, but not those derived from bid⫺/⫺bim⫺/⫺ animals. (A) SV40 immortalized MEFs were transiently transfected with 0 or 50 ng of pCMVneoBam-FLAG-PUMA␣ (Left) or PUMA (Right) (pCMV5 was used to keep the DNA mass equal), cultured for 24 h and analyzed for survival with AnnexinV-PE. GFP was cotransfected as a marker of efficiency; only GFP positive cells were scored for survival. White, black, dark gray, and light gray bars equal wild-type, bid⫺/⫺, bim⫺/⫺ and bid⫺/⫺bim⫺/⫺ MEFs, respectively. The error bars represent the standard deviation from triplicate data. (B) RIPA lysates from the SV40 immortalized MEFs were confirmed by Western blot analysis to be wild-type, bid⫺/⫺, bim⫺/⫺ or bid⫺/⫺bim⫺/⫺. Actin is shown for equal loading. (C and D) Isolated mitochondrial fractions from wild-type, bid⫺/⫺, bim⫺/⫺ or bid⫺/⫺bim⫺/⫺ livers were treated with indicated concentrations of C8-BID, PUMA, BID BH3 domain peptide, or PUMA BH3 domain peptide for 60 min at 37 °C and fractionated for the supernatant. (C) C8-BID and the BID BH3 domain peptide induced complete release at ⬍1 nM and 500 nM, respectively. (D) PUMA and PUMA BH3 domain peptide failed to induce substantial cytochrome c release at 5 M and 100 M, respectively, from the bid⫺/⫺bim⫺/⫺ mitochondria. Total cytochrome c was determined by a sample containing mitochondria solubilized in 1% CHAPS (C and D). (E) Isolated mitochondrial fractions were confirmed wild-type, bid⫺/⫺, bim⫺/⫺ or bid⫺/⫺bim⫺/⫺ by Western blot analysis.
PUMA␣ cDNA and did not elevate to levels produced by 1000 ng of PUMA␣ cDNA (Fig. S4 A and B). From these observations, we propose that stress-induced levels of PUMA cooperate with direct activator proteins (e.g., BID and BIM) to efficiently promote MOMP and apoptosis. Healthy cells also appear to harbor covert levels of direct activator BH3-only proteins that can be revealed by exogenous PUMA expression (Fig. 5A). In vitro, we examined the cooperation between PUMA and direct activator BH3-only proteins using primary murine liver heavy membrane fractions from untreated wild-type, bid⫺/⫺, bim⫺/⫺ and bid⫺/⫺bim⫺/⫺ animals. These heavy membrane fractions were first stimulated with C8-BID or the BID BH3 domain peptide to ensure that each can respond to direct activator stimulation (Fig. 5C). Both PNAS 兩 December 23, 2008 兩 vol. 105 兩 no. 51 兩 20331
CELL BIOLOGY
% F-dextran release
PUMA
(s)
bim-/-
C
BH3 Total
75
50
B
bim-/-
75
bid-/-bim-/-
bid-/-bim-/-
100
Wild-type
bim-/-
100
bid-/-
bid-/-
Wild-type
% Survival
Total
PUMA BH3
BAD BH3
DMSO
PUMA BH3
A
C8-BID MCL-1 C
BAD BH3
DMSO
PUMA
BAD BH3
PUMA
PUMA BH3
MCL-1 C
BCL-xL C
Buffer
C8-BID
C8-BID BCL-xL C
PUMA
A
C8-BID protein (0.1–1 nM) and BID BH3 domain peptide (100– 500 nM) efficiently induced complete cytochrome c release from all of the genotypes (Fig. 5C). The same concentrations of PUMA or the PUMA BH3 domain peptide failed to promote cytochrome c release using standard Western blot exposure times (Fig. S5A). Marked over-exposure of the PUMA BH3 domain peptide titration revealed minor cytochrome c release that was completely dependent on the presence of bid and bim (Fig. S5B) and unrelated to availability of cytochrome c for release (Fig. S6A) (23). Increasing the concentrations of PUMA and the PUMA BH3 domain peptide (up to 5 and 100 M, respectively) promoted more cytochrome c release but this was also mostly dependent upon endogenous bid and bim expression (Fig. 5D). The heavy membrane fractions from each genotype were confirmed by Western blot analysis (Fig. 5E). These data suggest that mitochondria derived from healthy primary tissues also harbor direct activator proteins, which are functional only when the entire anti-apoptotic BCL-2 is inhibited, in this case, by PUMA treatment. The BH3-only proteins function in cooperation with the BCL-2 effector proteins, BAX and BAK, to induce MOMP and subsequent apoptosis. It is hypothesized that one or more direct activator proteins are involved in many cellular stresses leading to the mitochondrial pathway of apoptosis, as these are required to engage BAX and/or BAK activation regardless of the apoptotic stimulus. The derepressor/sensitizer BH3-only proteins appear to act as sentinels for specific cellular stress pathways and liberate direct activator function at the OMM, as genetic deletion of individual members renders cells resistant to specific stress scenarios; for example, hrk- and puma-deficient animals exhibit defects in nerve growthfactorwithdrawalandcytokinedeprivation-dependentapoptosis, respectively (15, 24). Here, we were interested in understanding the cooperation between PUMA and preexisting direct activator proteins to engage the mitochondrial pathway of apoptosis. To explore this relationship, we used several derepression model systems where intact cells, mitochondria, or defined LUVs responded to direct activator proteins only upon inhibition of the anti-apoptotic BCL-2 repertoire by PUMA. Common to the cellular and mitochondrial derepression model systems was a direct activator:anti-apoptotic BCL-2 protein complex at the OMM (for example, C8-BID:BCL-2/BCL-xL/MCL-1 in Figs. 1 and 2); yet despite this association, mitochondria maintained their integrity (Figs. 1, 2, 3, 4 and 5) and cells remained viable (Figs. 1 and 5). These scenarios highlight the importance and fidelity of anti-apoptotic proteins on the OMM to inhibit unwarranted MOMP and 1. Green DR (2005) Apoptotic pathways: Ten minutes to dead. Cell 121:671– 674. 2. Chipuk JE, Green DR (2008) How do BCL-2 proteins induce mitochondrial outer membrane permeabilization? Trends Cell Biol 18:157–164. 3. Kluck RM, Bossy-Wetzel E, Green DR, Newmeyer DD (1997) The release of cytochrome c from mitochondria: A primary site for Bcl-2 regulation of apoptosis. Science 275:1132–1136. 4. Wei MC, et al. (2001) Proapoptotic BAX and BAK: A requisite gateway to mitochondrial dysfunction and death. Science 292:727–730. 5. Wei MC, et al. (2000) tBID, a membrane-targeted death ligand, oligomerizes BAK to release cytochrome c. Genes Dev 14:2060 –2071. 6. Letai A, et al. (2002) Distinct BH3 domains either sensitize or activate mitochondrial apoptosis, serving as prototype cancer therapeutics. Cancer Cell 2:183–192. 7. Kuwana T, et al. (2005) BH3 domains of BH3-only proteins differentially regulate Bax-mediated mitochondrial membrane permeabilization both directly and indirectly. Mol Cell 17:525–535. 8. Kuwana T, et al. (2002) Bid, Bax, and lipids cooperate to form supramolecular openings in the outer mitochondrial membrane. Cell 111:331–342. 9. Chipuk JE, et al. (2004) Direct activation of Bax by p53 mediates mitochondrial membrane permeabilization and apoptosis. Science 303:1010 –1014. 10. Chipuk JE, et al. (2005) PUMA couples the nuclear and cytoplasmic proapoptotic function of p53. Science 309:1732–1735. 11. Chen L, et al. (2005) Differential targeting of prosurvival Bcl-2 proteins by their BH3-only ligands allows complementary apoptotic function. Mol Cell 17:393– 403. 12. Oltersdorf T, et al. (2005) An inhibitor of Bcl-2 family proteins induces regression of solid tumours. Nature 435:677– 681. 13. Del Gaizo Moore V, et al. (2007) Chronic lymphocytic leukemia requires BCL2 to sequester prodeath BIM, explaining sensitivity to BCL2 antagonist ABT-737. J Clin Invest 117:112–121.
20332 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0808036105
apoptosis. The actively sequestered molecules may also be one mechanism for a cell to record its recent history of stressful encounters. In the event of sustained or irreparable stress, these sequestered molecules could provide a rapid means to induce apoptosis. When PUMA expression is a response stress (or ABT737 is present), our data suggest that the kinetics and efficiency of MOMP are greatly enhanced. Materials and Methods Reagents. All cell culture and transfection reagents were from Invitrogen; AnnexinV conjugates were from Caltag. Immortalized MEFS were produced by transfecting primary, unpassaged MEFs with SV40 genomic DNA and selected by colony formation and growth. pCMVneoBam-FLAG-PUMA␣/ were a gift from Karen Vousden (16). HeLa cells stably expressing cytochrome c-GFP were made as described (19). Antibodies: anti-PUMA (Cell Signaling), anti-BID (PharMingen, 550365), anti-BIM (Sigma), anti-actin (ICN, clone c4), anti-cytochrome c (for flow cytometry, clone 6H2.B4; for Western blot analysis, clone 7H8.2C12 PharMingen), anti-BAK (Upstate, clone NT), anti-BCL-2 (10C4), anti-BCL-xL (clone S-18), anti-A1 (FL-175), anti-MCL-1 (Rockland) and anti-p53 (Do7). Full-length human BID (FLBID), C8-BID, and human BCL-xL⌬C (except for NMR studies) were from R&D Systems. Human full-length MCL-1, full-length BCL-xL, MCL-1⌬C, p53UVIP— referred to as ‘‘cytosolic p53’’—N/C-BID, and full-length BAX were made as described (7, 9, 25–27). BH3 domain peptides: human BAD, BID, BIM, and PUMA (⬎98% purity, Anaspec) sequences as described (7). All peptides were resuspended in anhydrous DMSO in a N2 environment, stored at ⫺80 °C, and thawed only once. PCR primers for the bid⫺/⫺ animals were: 17B14, 5⬘-ccgaaatgtcccataagag-3⬘; JR23PGK-neo, 5⬘-tgctacttccatttgtcacgtcct-3⬘; 17B12, 5⬘-gagatggaccacaacatc-3⬘; wild-type (17B12 and 17B14) and knockout (17B12 and JR23PGK-neo) PCRs amplify a 123- and 350-base pair product, respectively. PCR primers for the bim⫺/⫺ animals were: PB20, 5⬘-cattctcgtaagtccgagtct-3⬘; PB65, 5⬘-ctcagtccattcatcaacag-3⬘; PB335, 5⬘-gtgctaactgaaaccagattag-3⬘; wild-type (PB20 and PB335) and knockout (PB20 and PB65) PCRs amplify a 380- and 540-base pair product, respectively. Combined bid⫺/⫺bim⫺/⫺ genomic samples were analyzed by both sets of PCR reactions. For detailed materials and methods, please refer to SI Materials. ACKNOWLEDGMENTS. We thank Andreas Strasser (Walter and Eliza Hall Institute, Melbourne, Australia), and the late Stanley Korsmeyer (Harvard Medical School, Boston) for the bim⫺/⫺ and bid⫺/⫺ animals, respectively; Simon Moshiach (St. Jude Children’s Research Hospital) for microinjection, Samual Connell (St. Jude Children’s Research Hospital) for the confocal images, Tudor Moldoveanu (St. Jude Children’s Research Hospital) for MCL-1⌬C, Jennifer Humberd and Blanca Schafer for technical assistance. ABT-737 was a gift from Dr. Stephen Fesik from Abbott Labs. Mammalian and prokaryotic PUMA expression vectors were kindly provided by Drs. Karen Vousden and Eric Eldering, respectively. This work was supported by NIH AI52735 and CA69381 (to D.R.G.), NIH R01CA082491 and R01CA092035 (to R.W.K.), NIH R21AG024478 (to T.K.), an NCI Cancer Center Core Grant P30CA21765 (at St. Jude) and the American Lebanese Syrian Associated Charities.
14. Certo M, et al. (2006) Mitochondria primed by death signals determine cellular addiction to antiapoptotic BCL-2 family members. Cancer Cell 9:351–365. 15. Jeffers JR, et al. (2003) Puma is an essential mediator of p53-dependent and -independent apoptotic pathways. Cancer Cell 4:321–328. 16. Nakano K, Vousden KH (2001) PUMA, a novel proapoptotic gene, is induced by p53. Mol Cell 7:683– 694. 17. Yu J, et al. (2003) PUMA mediates the apoptotic response to p53 in colorectal cancer cells. Proc Natl Acad Sci USA 100:1931–1936. 18. Kim H, et al. (2006) Hierarchical regulation of mitochondrion-dependent apoptosis by BCL-2 subfamilies. Nat Cell Biol 8:1348 –1358. 19. Goldstein JC, et al. (2000) The coordinate release of cytochrome c during apoptosis is rapid, complete and kinetically invariant. Nat Cell Biol 2:156 –162. 20. Rubin BY, et al. (1988) Correlation between the anticellular and DNA fragmenting activities of tumor necrosis factor. Cancer Res 48:6006 – 6010. 21. Willis SN, et al. (2005) Proapoptotic Bak is sequestered by Mcl-1 and Bcl-xL, but not Bcl-2, until displaced by BH3-only proteins. Genes Dev 19:1294 –1305. 22. von Ahsen O, et al. (2000) Preservation of mitochondrial structure and function after Bid- or Bax-mediated cytochrome c release. J Cell Biol 150:1027–1036. 23. Scorrano L, et al. (2002) A distinct pathway remodels mitochondrial cristae and mobilizes cytochrome c during apoptosis. Dev Cell 2:55– 67. 24. Coultas L, et al. (2007) Hrk/DP5 contributes to the apoptosis of select neuronal populations but is dispensable for haematopoietic cell apoptosis. J Cell Sci 120:2044 –2052. 25. Von Ahsen O, et al. (2000) The ‘‘harmless’’ release of cytochrome c. Cell Death Differ 7:1192– 1199. 26. Suzuki M, Youle RJ, Tjandra N (2000) Structure of Bax: coregulation of dimer formation and intracellular localization. Cell 103:645– 654. 27. Moldoveanu T, et al. (2006) The X-ray structure of a BAK homodimer reveals an inhibitory zinc binding site. Mol Cell 24:677– 688.
Chipuk et al.
