Limited Mitochondrial Permeabilization Causes DNA Damage

Preparing to load PDF file. please wait...

0 of 0
100%
Limited Mitochondrial Permeabilization Causes DNA Damage

Transcript Of Limited Mitochondrial Permeabilization Causes DNA Damage

Ichim, G. et al. (2015) Limited mitochondrial permeabilization causes DNA damage and genomic instability in the absence of cell death.Molecular Cell. Copyright © 2015 The Authors http://eprints.gla.ac.uk/102955/
Deposited on: 23 February 2015
Enlighten – Research publications by members of the University of Glasgow http://eprints.gla.ac.uk

Article

Limited Mitochondrial Permeabilization Causes DNA Damage and Genomic Instability in the Absence of Cell Death

Graphical Abstract

Authors
Gabriel Ichim, Jonathan Lopez, ..., Daniel J. Murphy, Stephen W.G. Tait

Correspondence
[email protected]

In Brief
During apoptosis, mitochondrial outer membrane permeabilization (MOMP) is widespread, leading to rapid cell death. Here, Ichim et al. demonstrate that MOMP can also be engaged in a minority of mitochondria without killing the cell. Instead, minority MOMP triggers caspase-dependent DNA damage and genomic instability, thereby promoting transformation and tumorigenesis.

Highlights
d MOMP can occur in a minority of mitochondria
d Minority MOMP triggers caspase activity but fails to kill cells
d Minority MOMP-induced caspase activity causes DNA damage and genomic instability
d Minority MOMP promotes cellular transformation and tumorigenesis

Ichim et al., 2015, Molecular Cell 57, 1–13 March 19, 2015 ª2015 The Authors http://dx.doi.org/10.1016/j.molcel.2015.01.018

Please cite this article in press as: Ichim et al., Limited Mitochondrial Permeabilization Causes DNA Damage and Genomic Instability in the Absence of Cell Death, Molecular Cell (2015), http://dx.doi.org/10.1016/j.molcel.2015.01.018
Molecular Cell
Article

Limited Mitochondrial Permeabilization Causes DNA Damage and Genomic Instability in the Absence of Cell Death
Gabriel Ichim,1,2 Jonathan Lopez,1,2 Shafiq U. Ahmed,2 Nathiya Muthalagu,1,2 Evangelos Giampazolias,1,2 M. Eugenia Delgado,3 Martina Haller,1,2 Joel S. Riley,1,2 Susan M. Mason,1 Dimitris Athineos,1 Melissa J. Parsons,4,5 Bert van de Kooij,6 Lisa Bouchier-Hayes,4,5 Anthony J. Chalmers,2 Rogier W. Rooswinkel,6 Andrew Oberst,7 Karen Blyth,1 Markus Rehm,3 Daniel J. Murphy,1,2 and Stephen W.G. Tait1,2,* 1Cancer Research UK Beatson Institute 2Institute of Cancer Sciences, University of Glasgow Garscube Estate, Switchback Road, Glasgow G61 1BD, UK 3Centre for Systems Medicine, Royal College of Surgeons in Ireland, Dublin 2, Ireland 4Center for Cell and Gene Therapy 5Department of Pediatrics-Hematology Baylor College of Medicine, Houston, TX 77030, USA 6Division of Immunology, The Netherlands Cancer Institute, Plesmanlaan 121, Amsterdam 1066 CX, the Netherlands 7Department of Immunology, University of Washington, 750 Republican Street, Seattle, WA 98109, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.molcel.2015.01.018 This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

SUMMARY
During apoptosis, the mitochondrial outer membrane is permeabilized, leading to the release of cytochrome c that activates downstream caspases. Mitochondrial outer membrane permeabilization (MOMP) has historically been thought to occur synchronously and completely throughout a cell, leading to rapid caspase activation and apoptosis. Using a new imaging approach, we demonstrate that MOMP is not an all-or-nothing event. Rather, we find that a minority of mitochondria can undergo MOMP in a stress-regulated manner, a phenomenon we term ‘‘minority MOMP.’’ Crucially, minority MOMP leads to limited caspase activation, which is insufficient to trigger cell death. Instead, this caspase activity leads to DNA damage that, in turn, promotes genomic instability, cellular transformation, and tumorigenesis. Our data demonstrate that, in contrast to its well-established tumor suppressor function, apoptosis also has oncogenic potential that is regulated by the extent of MOMP. These findings have important implications for oncogenesis following either physiological or therapeutic engagement of apoptosis.
INTRODUCTION
Following most apoptotic stimuli, the pro-apoptotic BCL-2 family members Bax and Bak permeabilize the outer membrane of the mitochondria, an event termed ‘‘mitochondrial outer membrane permeabilization’’ (MOMP). MOMP leads to rapid cell

death by releasing mitochondrial proteins including cytochrome c that activate caspases (Tait and Green, 2010). However, even in the absence of caspase activity, cells typically die once MOMP has occurred, most likely due to progressive mitochondrial dysfunction (Lartigue et al., 2009; Tait et al., 2014). Due to these catastrophic effects, MOMP is often considered the point of no return in the apoptotic program. Mitochondrial apoptosis plays numerous important pathophysiological roles. In cancer, inhibition of apoptosis both promotes tumorigenesis and impedes anti-cancer therapeutic efficacy (Delbridge et al., 2012). Apoptotic inhibition is often achieved by upregulation of anti-apoptotic BCL-2 family members that prevent MOMP. This has led to the development of new anticancer drugs, called BH3mimetics, which neutralize anti-apoptotic BCL-2 function (Ni Chonghaile and Letai, 2008).
Live-cell imaging has demonstrated that mitochondrial permeabilization is often an all-or-nothing event (Goldstein et al., 2000). Widespread mitochondrial permeabilization underpins the lethal effects of MOMP by ensuring robust caspase activity, or in its absence, massive mitochondrial dysfunction. In some limited circumstances, cells can survive MOMP. For example, growth factor-deprived neurons can survive MOMP due to a failure to properly engage caspase activity (Deshmukh and Johnson, 1998; Martinou et al., 1999; Wright et al., 2004). In proliferating cells, expression of the key glycolytic enzyme GAPDH can promote cell survival following MOMP provided caspase activity is inhibited (Colell et al., 2007). We have previously found that the ability of cells to survive MOMP depends on a few mitochondria that evade permeabilization and re-populate the cell (Tait et al., 2010).
Whereas earlier studies demonstrated that strong proapoptotic stimuli lead to rapid, synchronous, and complete MOMP, technical limitations have made it impossible to study the effects of sub-lethal stresses on individual mitochondria.

Molecular Cell 57, 1–13, March 19, 2015 ª2015 The Authors 1

Please cite this article in press as: Ichim et al., Limited Mitochondrial Permeabilization Causes DNA Damage and Genomic Instability in the Absence of Cell Death, Molecular Cell (2015), http://dx.doi.org/10.1016/j.molcel.2015.01.018

Here, we use newly developed imaging techniques to demonstrate that MOMP can occur in a limited subset of mitochondria following a sub-lethal stress. Crucially, this limited MOMP leads to caspase activation, which, while insufficient to trigger cell death, leads to limited cleavage of key caspase substrates. This in turn drives DNA-damage and genomic instability, promoting transformation and tumorigenesis. Importantly, our data argue that the mitochondrial apoptotic pathway may exert either a tumor suppressor or oncogenic function depending upon the extent of MOMP.
RESULTS
Limited Mitochondrial Permeabilization Occurs in the Absence of Cell Death Mitochondrial permeabilization during apoptosis is widespread such that most or all mitochondria within a cell undergo MOMP; this effectively commits a cell to die. However, the potential for sub-lethal apoptotic stresses to engage MOMP in a limited number of mitochondria has not been tested. To investigate this, we used ABT-737, the prototypic BH3-mimetic compound that sensitizes to apoptosis by antagonizing anti-apoptotic BCL-2 family proteins (Oltersdorf et al., 2005). HeLa or U2OS cells were treated with varying concentrations of ABT-737, enantiomer (less-active stereoisomer of ABT-737) or the apoptosis-inducer staurosporine (STS) and analyzed for apoptosis by Annexin V staining and flow cytometry. Importantly, whereas STS triggered apoptosis, treatment with ABT-737 at varying doses failed to induce detectable apoptosis (Figure 1A). Similarly, live-cell imaging using the cell impermeable dye Sytox green also failed to reveal a cytotoxic effect of ABT-737 treatment (Figure S1A). Finally, BH3-mimetic treatment at the indicated doses had no effect on long-term cell survival as determined by clonogenic assay (Figure S1B). We next asked if mitochondrial permeabilization occurred following this non-lethal BH3-mimetic treatment. HeLa cells were treated with ABT-737 or, as a positive control, Actinomycin D (Act D) and cytosolic fractions were probed for the presence of cytochrome c to detect MOMP. As expected, Act D treatment led to MOMP as demonstrated by the detection of cytochrome c in the cytosolic extract (Figure 1B). Surprisingly, treatment with a non-lethal dose of ABT-737 also led to low, but detectable levels of MOMP, implying that MOMP could be engaged without killing the cell (Figure 1B).
In caspase-proficient cells, complete MOMP invariably represents a point of no return; we therefore reasoned that MOMP could only be non-lethal if it occurred in a minority of mitochondria (called hereafter minority MOMP). To test this possibility, we developed a new approach to specifically visualize permeabilized mitochondria using fluorescent protein re-localization and chemically dimerizable FKBP/FRB domains (Belshaw et al., 1996). Two fluorescent probes were constructed: a cytosolic probe comprising of GFP fused to a FKBP domain (cytoGFP) and a mitochondrial targeted probe (mitoCherry) comprising of mCherry fused to an FRB domain and the mitochondrial anchoring sequence of Apoptosis Inducing Factor (AIF) (Otera et al., 2005). In the presence of chemical heterodimerizer (A/C heterodimerizer, AP21967), the two probes can only co-localize on mitochondria following MOMP when cytoGFP can gain ac-

cess to the mitochondrial inner membrane (Figure 1C). To validate this method, we treated U2OS cells with the apoptotic stimulus Act D. Importantly, Act D treatment led to robust mitochondrial re-localization of cytoGFP only in the presence of dimerizer that, as expected, was prevented by blocking MOMP by expression of BCL-xL (Figures 1D, S1C, and S1D). In line with this method marking MOMP as the key initiating apoptotic event, cytoGFP re-localization preceded caspase dependent apoptotic effects including cell shrinkage, rounding, and plasma membrane blebbing (Movie S1). Further verifying this technique, Act D-induced mitochondrial localization of GFP was only observed in cells in which MOMP had occurred, as demonstrated by the cytosolic release of Smac mCherry (a verified reporter of MOMP) (Figure S1E and Movie S2) (Tait et al., 2010). These fluorescent tools thus allow us to detect the permeabilization of individual mitochondria, and thereby assay for the presence of minority MOMP.
Using this approach, we investigated the extent of MOMP following sub-lethal apoptotic stimuli. Strikingly, following ABT737 treatment, we were able detect MOMP in a limited number of mitochondria in both HeLa and U2OS cells (Figure 1E). Confirming their permeabilization, mitochondria with relocalized CytoGFP had also released Smac-mCherry and cytochrome c (Figures S1F and S1G). The percentage of cells displaying minority MOMP increased in a dose-dependent and BCL-xL inhibitable manner following ABT-737 treatment (Figure 1F). Importantly, using live-cell imaging, cells displaying minority MOMP failed to undergo cell death during extended periods of analysis (Figure S1H). Collectively, these data demonstrate that MOMP can occur in a limited number of mitochondria in response to ABT-737 treatment without leading to cell death.
Minority MOMP Engages Sub-Lethal Caspase Activity We next sought to understand the consequences of minority MOMP, and in particular, whether it might lead to activation of caspases at sub-apoptotic levels. We first quantified the extent of minority MOMP in HeLa cells. This revealed an average of 2.5% of a cell’s mitochondria undergoing permeabilization following sub-lethal ABT-737 treatment (Figure 2A). Together with previously published criteria, this allowed us to adapt a mathematical HeLa cell model of the apoptosis execution phase (Rehm et al., 2006) to perform in silico simulations of the consequences of minority MOMP on the efficiency of caspase-3 processing and activation. Importantly, despite the presence of amplifying feedback loops which ensure rapid and full caspase activation in response to regular MOMP, simulations for minority MOMP conditions demonstrate that caspase-3 would be processed and activated sub-maximally and therefore, potentially, at sub-lethal levels (Figures 2B and S2A). To experimentally verify this, U2OS and HeLa cells were treated with the BH3mimetic ABT-737 in the presence or absence of caspase inhibitor quinolyl-valyl-O-methylaspartyl-[2,6-difluoro- phenoxy]methyl ketone (Q-VD-OPh). Treatment with a wide range of sub-lethal doses of ABT-737 triggered caspase activity as evidenced by pro-caspase-3 processing and PARP cleavage, effects that were blocked by caspase inhibition (Figures S2B and S2C). We next compared caspase activity between conditions that engage minority MOMP and apoptotic conditions. Cells

2 Molecular Cell 57, 1–13, March 19, 2015 ª2015 The Authors

Please cite this article in press as: Ichim et al., Limited Mitochondrial Permeabilization Causes DNA Damage and Genomic Instability in the Absence of Cell Death, Molecular Cell (2015), http://dx.doi.org/10.1016/j.molcel.2015.01.018

A

B

C

Annexin V positive cells (%)

30

HeLa

U2OS

* ABT-737 (μM): - 10 5 1 -

cytosol

DM cytosol MOMP DM

*

ActD: - - - - +

GFP DM DM

20

WCE

FKBP

Cyt c

DM DM

DM DM

OMM IMM

10

0

ENA

μM 0

μM .5

μM μM 5 .5

μM 1

1μM

7 1 37 7 -737 37 2 -737 STS

BT-73 BT-7 ABT BT-7 ABT

AA

A

D

CytoGFP MitoCherry Merge

HeLa

COX IV β-tubulin
Ponceau

OMM IMM

FRB

DM

IMS

E
CytoGFP MitoCherry Merge

AIF mC herry
AIF mC herry

FRB

DMFK

DM

BP

GFP

IMS

Ctrl ActD

ENA

U2OS LZRS

ABT -737

HeLa

Fluorescence intensity

5000 4000 3000 2000 1000
0 0

CytoGFP MitoCherry

100

200

300

Distance (pixels)

Ctrl

Act D

Cells with minority MOMP (%)

F 10
8 6 4 2 0

U2OS LZRS
* *

U2OS BCL-XL

Ctrl

-737

-737

BT μM 5

ABT .5 μM

A

2

ENA
U2OS BCL-XL
ABT737

U2OS

Fluorescence intensity

5000 4000 3000 2000 1000
0 0

CytoGFP MitoCherry

50

100

150

Distance (pixels)

Figure 1. Limited Mitochondrial Outer Membrane Permeabilization Occurs without Triggering Cell Death (A) HeLa and U2OS cells were treated for 3 hr with different concentrations of ABT-737 or enantiomer (ENA, 10 mM) or with staurosporine (STS, 1 mM) for 12 hr and analyzed by flow cytometry for Annexin V-positive cells. Data represent mean ± SEM of three independent experiments. (B) HeLa cells were treated for 3 hr with different concentrations of ABT-737 or actinomycin D for 6 hr (Act D, 1 mM), and cytosolic extracts were western blotted for cytochrome c, COX IV, and b-tubulin. WCE, whole-cell extract. (C) Schematic representation of GFP relocalization-based MOMP detection method. DM, chemical heterodimerizer; IMM, inner mitochondrial membrane; IMS, intermembrane space; OMM, outer mitochondrial membrane. (D) U2OS cells expressing vector or BCL-xL together with CytoGFP/MitoCherry were treated with Act D (1 mM) for 3 hr in the presence of heterodimerizer and imaged by confocal microscopy. (E) HeLa or U2OS cells expressing CytoGFP/MitoCherry were treated with vehicle or ABT-737 (5 mM) or enantiomer (5 mM, ENA) for 3 hr and imaged by confocal microscopy. Arrows denote permeabilized mitochondria. Line scans represent variation in red and green fluorescence intensity along the denoted line. (F) U2OS cells expressing CytoGFP/MitoCherry were treated for 3 hr with ABT-737, and minority MOMP was quantified. Data represent mean ± SEM of three independent experiments. *p < 0.05, compared versus control. See also Figure S1 and Movies S1 and S2.

were treated with ABT-737 to induce minority MOMP or with TNF/CHX or Act D to engage apoptosis (Figures 2C–2E and S2D). We determined executioner caspase-3 and -7 activity, by detection of their active, cleaved fragment (Figure 2C) or by their activity-dependent precipitation using biotin-Val-Ala-Asp-

Fluoromethyl Ketone (b-VAD) (Figure 2D). Both approaches demonstrated that caspase-3 and -7 activity was detectable but significantly less in ABT-737 treated cells undergoing minority MOMP compared with apoptotic cells (Figures 2C and 2D). Levels of active caspase-9, precipitated with b-VAD, were also

Molecular Cell 57, 1–13, March 19, 2015 ª2015 The Authors 3

Please cite this article in press as: Ichim et al., Limited Mitochondrial Permeabilization Causes DNA Damage and Genomic Instability in the Absence of Cell Death, Molecular Cell (2015), http://dx.doi.org/10.1016/j.molcel.2015.01.018

Volume of permeabilised mitochondria (%)
Procaspase-3 Processing (%)

A
100 90 80 70 60 50 5 4

U2OS

HeLa

B 100 MOMP versus minority MOMP C

HeLa

90 MOMP ABT-737: - + -

80

TNF/CHX: - -

+

70

full length

60

+50%

50

median minority pp1179

MOMP

40

-50%

full length

30

3 20 p20

2

1

10

0 Ctrl BT-737 Act D A

0

0

50

100 150 200

Time after onset of MOMP (min)

D

bVAD precipitated

Input

E

ABT-737: - + -

- -+ - -

TNF/CHX: - - + - - - + -

ActD: - - -

+ - - - +

full length

Caspase 3

25 kDa

p17

15 kDa

bVAD precipitated
ABT-737: - + - - TNF/CHX: - - + - -
ActD: - - - + -
full length

p17 full length
p20

long exposure
35 kDa
Caspase 7
15 kDa

p37 full length
p43

long

p43

p20

exposure

U2OS

- +

-

- -

+

25 kDa
Caspase 3

15 kDa

35 kDa
Caspase 7
15 kDa
HSP60

Input
+ - + - -

+
Caspase 9
40 kDa

Caspase 8
50 kDa

long exposure

Mean Fluorescence Intensity (fold change)

F
20 18 16 14 2.0 1.5 1.0 0.5 0.0
I

VC3AI ncVC3AI

*

*

Ctrl

-737

737/

ABT

ABT-QVD

G

VC3AI

Ctrl

ABT737
Act D ABT737 / QVD

SmacmCherry

Merge

H
10

*

*

VC3AI GFP positive cells (%)

5

0

Ctrl

-737

737 /

BT

BT- VD

A

AQ

ns 100

Act D

Colony number (%)

50

0 -737 -
ABT GFP

T-737 + AB GFP

FP+ tD G Ac

Figure 2. Minority MOMP Engages Sub-Lethal Caspase Activity (A) U2OS and HeLa cells expressing CytoGFP/MitoCherry were treated or not with ABT-737 (10 mM for 3 hr in the presence of dimerizer) to induce minority MOMP, and then z stack confocal imaging was performed. Act D (1 mM) was used to induce complete MOMP. Mitochondrial volume was measured using ImageJ. Data represent mean of permeabilized mitochondrial volume ± SEM from ten cells per condition. (B) Data from (A) were used as inputs into a mathematical HeLa cell model of apoptosis execution signaling to calculate the consequences of minority MOMP on the efficacy of procaspase-3 processing. (C) HeLa and U2OS cells were treated with ABT-737 (10 mM) or TNF/CHX (20 ng/ml TNF and 1 mg/ml CHX) for 3 hr, and cell extracts were western blotted for caspases-3 and -7. (D) Biotinylated-VAD-FMK (bVAD) was incubated with HeLa cells for 1 hr following 3 hr treatments with the indicated stimuli. Cell lysates and precipitated proteins were western blotted for caspases-3 and -7.
(legend continued on next page)

4 Molecular Cell 57, 1–13, March 19, 2015 ª2015 The Authors

Please cite this article in press as: Ichim et al., Limited Mitochondrial Permeabilization Causes DNA Damage and Genomic Instability in the Absence of Cell Death, Molecular Cell (2015), http://dx.doi.org/10.1016/j.molcel.2015.01.018

detectable but significantly less in ABT-737 treated cells undergoing minority MOMP in comparison to apoptotic cells (Figure 2E). In line with ABT-737 activating the mitochondrial caspase pathway, ABT-737 treatment led to caspase-9 but not caspase-8 activation (Figure 2E). Collectively, these data argue that minority MOMP can engage sub-lethal caspase activity. To corroborate these findings, we used a recently developed caspase reporter protein (VC3AI) that fluoresces following caspase-mediated cleavage (Zhang et al., 2013). As validation, apoptotic treatments led to an increase in fluorescence in VC3AI expressing HeLa cells in a caspase-dependent manner, whereas cells expressing the non-cleavable control (ncVC3AI) remained non-fluorescent (Figure S2E). Significantly, flow cytometry analysis demonstrated that ABT-737 treatment led to a detectable increase in caspase activity in viable cells that was inhibited by the caspase inhibitor Q-VD-OPh, further supporting the hypothesis that minority MOMP triggers sub-lethal caspase activation (Figure 2F). We then treated HeLa cells co-expressing Smac-mCherry together with either VC3AI or ncVC3AI with the BH3-mimetic ABT-737. In line with the flow cytometry data, treatment with ABT-737 specifically increased the percentage of cells displaying weak caspase-dependent fluorescence (Figures 2G, 2H, and S2F). Importantly, cells exhibiting caspase-activity failed to display apoptotic, widespread MOMP because Smac-mCherry remained localized in the mitochondria (Figure 2G). We next determined if cells displaying sub-lethal caspase activity could survive long-term. HeLa cells expressing the VC3AI reporter were treated either with ABT-737 to engage minority MOMP dependent caspase activity or Act D to trigger mitochondria-dependent apoptosis. Equal numbers of caspase active (GFP-positive cells) from both conditions were sorted and compared for clonogenic survival versus caspase inactive, GFP-negative cells (Figures 2I and S2G). Importantly, similar clonogenic survival was observed comparing ABT-737 treated caspase active versus inactive cells; in contrast, stimulation of mitochondrial dependent apoptosis by Act D prevented clonogenic outgrowth (Figure 2I). Taken together, these data demonstrate that minority MOMP can engage low level caspase activity under which cells can survive.
Minority MOMP Induces Caspase-Dependent DNA Damage DNA fragmentation is a classical apoptotic hallmark mediated by caspase-activated DNase (CAD) (Sakahira et al., 1998). We hy-

pothesized that limited caspase activity following minority MOMP might lead to low-level CAD activation, and, in turn, to induction of DNA damage in surviving cells. To test this, we treated HeLa and U2OS cells with ABT-737 in the presence or absence of Q-VD-OPh and analyzed for gH2A.X as readout for DNA damage. Importantly, in both cell lines, non-lethal treatment with BH3 mimetic led to caspase-dependent DNA damage as demonstrated by a caspase-dependent increase in gH2A.X (Figure 3A). Importantly, the extent of DNA-damage (measured by gH2A.X foci) correlated with minority MOMP, implicating a causal relationship between the two (Figures 3B and 3C). The ability of BH3 mimetics to engage DNA-damage in a caspase-dependent manner was also observed in other cell lines (Figure S3A). Further demonstrating caspase-dependent DNA damage, ABT-737 treatment also led to an increase in DNA breaks, measured by comet assay and Ser15 p53 phosphorylation dependent on caspase function, mirroring gH2A.X levels (Figures 3D, 3E, and S3B). Although p53 independent, induction of DNA damage depended on mitochondrial caspase activation, because overexpression of BCL-xL (HeLa), deletion of Bax (HCT-116) or Bax and Bak in murine embryonic fibroblasts (MEF), or knockdown of APAF-1 (HeLa) prevented BH3 mimetic-induced gH2A.X (Figures 3F– 3H and S3C–S3E). Supporting these findings, direct, non-lethal activation of caspase-9 by chemical dimerization also led to DNA damage (Figures 3I and S3F).
Because these data clearly demonstrate caspase-dependent activation of a DNA damage response, we next tested whether this effect is mediated by CAD activation. Caspase-3 activity is required for CAD activation through cleavage of its inhibitor ICAD. Consistent with caspase-3 and CAD-dependent DNA damage downstream of minority MOMP, gH2A.X was not observed following ABT-737 treatment of caspase-3 deficient MCF-7 cells in contrast to caspase-3 reconstituted MCF-7 (Figure S3G). To directly test the involvement of CAD in minority MOMP-induced DNA damage, we used CRISPR/Cas9 genome editing to generate CAD-deficient HeLa and U2OS cells (Figure 3J). In both cell lines, CAD deletion prevented BH3 mimetic induction of gH2A.X, demonstrating its requirement for minority MOMP-induced DNA damage (Figure 3J). Similar results were observed following CAD knockdown by siRNA (Figure S3H). In line with these findings, CAD deletion also effectively prevented ABT-737 induced DNA breaks, as determined by comet assay (Figure 3K). Furthermore, we also found that ICAD is cleaved following induction of minority MOMP, thereby supporting CAD

(E) As in (D), except HeLa cells were pre-incubated for 1 hr with bVAD and treated for 3 hr with ABT-737 (10 mM) or 16 hr Act D (16 hr, 1 mM) or 3 hr TNF/CHX (20 ng/ml TNF and 1 mg/ml CHX). Proteins were precipitated with neutravidin agarose resin. Cell lysates and precipitated proteins were western blotted for caspases-8 and -9. (F) HeLa cells stably expressing the caspase activity reporter (VC3AI) or non-cleavable control (ncVC3AI) were treated for 24 hr with 10 mM ABT-737 in presence or absence of Q-VD-OPh (10 mM). GFP mean fluorescence intensity of the viable cells was quantified by flow cytometry. Results represent the fold increase in fluorescence over control. Data represent mean ± SEM of four independent experiments. (G) HeLa cells stably expressing VC3AI together with Smac-mCherry were treated as in (F) and imaged for GFP. Arrows denote caspase reporter (GFP)-positive cells. (H) Quantification of percentage of GFP-positive cells following ABT-737 (10 mM) treatment in the presence or absence of Q-VD-OPh (10 mM). Data represent mean ± SEM of three independent experiments. (I) HeLa VC3AI cells were treated with ABT-737 (10 mM) for 24 hr or Act D (0.5 mM), and equal numbers of GFP-positive cells (ABT-737 and Act D treated sample) and GFP-negative cells (ABT-737 treated cells) were sorted by flow cytometry and assessed for clonogenic survival. Data represent mean ± SEM of three independent experiments. *p < 0.05, compared versus control. See also Figure S2.

Molecular Cell 57, 1–13, March 19, 2015 ª2015 The Authors 5

Please cite this article in press as: Ichim et al., Limited Mitochondrial Permeabilization Causes DNA Damage and Genomic Instability in the Absence of Cell Death, Molecular Cell (2015), http://dx.doi.org/10.1016/j.molcel.2015.01.018

A

B

CytoGFP MitoCherry ɣH2A.X Merge

ABT-737 (μM): ENA 10 7.5 5 2.5 1 ENA 10 7.5 5 2.5 1 QVD: - - - - - - + + + + + +
HeLa
U2OS

ɣH2A.X actin ɣH2A.X actin

Ctrl
ABT737
H2O2

Tail moment (pixels)

H2A.X foci / cell

C 40
30 20

* *

D

E

0.8

0.6

*

0.4

Ctrl

ABT-737

0.2

*

*

0.15

10

0.10

0.05

Ctrl ABT
-737 H 2O
2

0

0.00

ABT-737/QVD H O

Ctrl

-737

737 /

O2

22

ABT

ABT- QVD

H2

F

HeLa

ABT-737 (μM): - 10 7.5 5 2.5 1

HeLa BCL-XL - 10 7.5 5 2.5 1
ɣH2A.X

G MEF wt MEF Bax/Bak DKO
ABT-737 (μM): - 10 7.5 5 2.5 1 - 10 7.5 5 2.5 1 ɣH2A.X

actin

H

HeLa

HeLa APAF-1 KD

I

ABT-737 (μM): - 10 7.5 5 2.5 1 - 10 7.5 5 2.5 1

ɣH2A.X

actin

DM (pM): - 10 5 2.5 1 ɣH2A.X Caspase-3

actin

J

HeLa

HeLa CRISPR CAD

ABT-737 (μM): - 1 2.5 5 7.5 10 - 1 2.5 5 7.5 10
γH2A.X K

CAD

α-Tubulin

ABT-737 (μM): -

U2OS

U2OS CRISPR CAD

1 2.5 5 7.5 10 - 1 2.5 5 7.5 10

γH2A.X

CAD

α-Tubulin

actin A549 iCaspase-9

HeLa

0.4

*

0.3

**

0.2

0.1

HeLa CRISPR CAD
*

Tail moment (pixels)

L ɣH2A.X

0.0 Ctrl 37 VD O 2 T-7 / Q H 2
AB -737 ABT

Ctrl

-737

VD O 2 Q H2

BT 37 /

7

A BT-

A

Control Control

ABT-737

TUNEL Doxorubicin

Figure 3. Minority MOMP Induces Caspase-Dependent DNA Damage (A) HeLa and U2OS cells were treated for 3 hr with indicated sub-lethal doses of ABT-737 or enantiomer (10 mM, ENA) in presence or absence of caspase inhibitor Q-VD-OPh (10 mM). Cell lysates were for probed by western blot for gH2A.X and actin (as loading control).
(legend continued on next page)
6 Molecular Cell 57, 1–13, March 19, 2015 ª2015 The Authors

Please cite this article in press as: Ichim et al., Limited Mitochondrial Permeabilization Causes DNA Damage and Genomic Instability in the Absence of Cell Death, Molecular Cell (2015), http://dx.doi.org/10.1016/j.molcel.2015.01.018

activation (Figure S3I). Finally, we addressed if minority MOMP also promotes DNA damage in vivo. Mice were administered one dose of ABT-737 (either 75 mg/kg or 125 mg/kg) or administered daily with ABT-737 (75 mg/kg) over a 3-day period. Strikingly, a single dose of ABT-737 resulted in a significant increase of gH2A.X immunoreactivity in the small intestine of mice (Figure 3L). Repeated dosing did not result in an increase in gH2A.X immunoreactivity, potentially due to resolution of DNA damage in between doses (Figure S3J). Although not an absolute measure of apoptosis, because it detects cells only transiently during apoptosis, TUNEL staining failed to reveal any evidence of apoptosis following all ABT-737 treatments (Figures 3L and S3J). Collectively, these data show that sub-lethal stresses, causing minority MOMP, can trigger caspase-dependent DNA damage both in vitro and in vivo.
Sub-Lethal BH3-Only and Apoptotic Stress Induce Minority MOMP and DNA Damage Although our data demonstrate that the BH3 mimetic drug ABT737 triggers minority MOMP and DNA damage, we sought to demonstrate that these effects also occur following MOMP triggered through other means. BH3-only proteins are the endogenous inducers of MOMP through their ability to activate Bax and Bak. Therefore, we next asked whether BH3-only proteins themselves could also engage minority MOMP. For this purpose, we generated a MelJuSo cell line expressing the BH3-only protein tBID under a doxycycline inducible promoter. Whereas doxycycline addition at 1 mg/ml led to robust tBID expression and apoptotic cell death, we were able to titrate doxycycline down to induce non-lethal tBID expression (Figures 4A, 4B, and S4A– S4D). We assessed whether expression of tBID at non-lethal levels could also trigger minority MOMP. Importantly, sub-lethal amounts of tBID (induced by 2.5 and 1 ng/ml doxycycline) led to minority MOMP as detected by the presence of cytochrome c in the cytosol (Figure 4C). We aimed to validate these findings using our method to detect MOMP via GFP re-localization. Non-lethal levels of tBID expression led to a clear increase in cells displaying minority MOMP in a manner that could be prevented through co-expression of BCL-xL (Figures 4D and 4E). Importantly, as was observed for BH3-mimetic treatment, the caspase inhibitor Q-VD-OPh also prevented H2A.X phosphorylation upon induction of sub-lethal levels of tBID (Figure 4F). We next addressed if a physiological apoptotic stimulus could also trigger

minority MOMP and DNA-damage. Accordingly, sub-lethal treatment of U2OS cells with FAS ligand triggered minority MOMP and DNA-damage in a BCL-xL and caspase-inhibitable manner (Figures 4G, 4H, and S4E). Similarly, sub-lethal treatment of cells with the proteasome inhibitor MG132 also induced minority MOMP (Figure 4I). Collectively, these data demonstrate that similar to BH3-mimetics, apoptotic stimuli and pro-apoptotic BH3-only proteins also induce minority MOMP and DNA damage.
JNK Regulates the DNA Damage Response DNA damage-induced phosphorylation of H2A.X at S139 often occurs via the PI3K-related kinase family members ATM, ATR, and DNA-PK (Jackson and Bartek, 2009; Shiloh, 2003). However, we did not observe any significant increase in activated, phosphorylated ATM or ATR kinase following sub-lethal ABT-737 treatment (Figure 5A). Moreover, RNAi-mediated knockdown of ATM, ATR, or DNA-PK failed to affect BH3 mimetic-induced gH2A.X levels (Figures S5A and S5B). Besides ATM, ATR, and DNA-PK, c-Jun N-terminal kinase (JNK) has also been found to mediate H2A.X phosphorylation in some settings (Lu et al., 2006). Importantly, sub-lethal treatment with ABT-737 led to a caspase-dependent increase in JNK1/2 activation mirroring levels of gH2A.X (Figure 5B). JNK activation following ABT-737 administration was also detected in vivo in the small intestine (Figure 5C). To directly investigate the role of JNK in H2A.X phosphorylation, we used RNAi. Combined knockdown of JNK1/2 or selective knockdown of JNK2 effectively prevented ABT-737 induced gH2A.X implicating a direct role for JNK2 in H2A.X phosphorylation (Figures 5D and 5E). Accordingly, RNAi-mediated knockdown of CAD largely inhibited JNK1/2 phosphorylation (Figure 5F). These results identify JNK2 as a key player in the minority MOMP-induced DNA damage response.
Minority MOMP Promotes Genomic Instability Based on our results, we hypothesized that by causing DNA damage, minority MOMP may promote genomic instability and transformation. To test this possibility, we repeatedly treated HeLa and U2OS cells with sub-lethal doses of ABT-737 for five (P5) or ten passages (P10). Following blockage of cytokinesis, we then quantified the number of cells with micronuclei, a wellestablished marker for chromosomal damage (Figure S6A) (Fenech, 2007). Strikingly, U2OS and HeLa cells displayed a significant increase in micronuclei number following ABT-737

(B) U2OS cells transiently expressing CytoGFP and MitoCherry were treated with ABT-737 (5 mM) for 3 hr or H2O2 (25 mM) for 10 min and immunostained for gH2A.X. Representative images are shown. (C) Quantification of gH2A.X foci in cells displaying minority MOMP (ABT-737-treated cells), control, and H2O2-treated cells. Data represent mean ± SEM of three independent experiments. (D) HeLa cells were treated as in (B) and subject to comet assay. Representative images are shown. (E) Quantification of comet tail moment following ABT-737 treatment. Data represent mean ± SEM of three independent experiments. (F–H) HeLa and HeLa overexpressing BCL-xL (F), wild-type MEF and MEF double knockout for Bax and Bak (G), or HeLa versus HeLa knockdown for APAF-1 (H) were treated as in (A) and western blotted for gH2A.X and actin. (I) A549 cells expressing caspase-9 fused to a FKBP dimerization domain were treated with indicated sub-lethal concentrations of homodimerizer (DM) for 3 hr to induce caspase-9 dimerization and activation. Cleavage of caspase-3 and gH2A.X was assessed by western blot. (J) Wild-type HeLa and U2OS cells and their Cad-deleted counterparts were treated and immunoblotted as in (A). (K) Wild-type and Cad-deleted HeLa cells were treated as in (D) and used to perform comet assay. Graph represents quantification of comet tail moment. Data represent mean ± SEM of four independent experiments. (L) Representative images of gH2A.X and TUNEL immunohistochemical staining in small intestine of mice treated with ABT-737 (75 mg/kg) for 1 day (n = 3). *p < 0.05, compared versus control. See also Figure S3.
Molecular Cell 57, 1–13, March 19, 2015 ª2015 The Authors 7

Please cite this article in press as: Ichim et al., Limited Mitochondrial Permeabilization Causes DNA Damage and Genomic Instability in the Absence of Cell Death, Molecular Cell (2015), http://dx.doi.org/10.1016/j.molcel.2015.01.018

A
tBID lJuSotON lJuSo Me te Me wt

tBID

DOX 1 (μg/ml): - +

actin - +

D

B 30
20
10
0 DOX (ng/ml): -

CytoGFP MitoCherry

DAPI

Ctrl

DOX 2.5 ng
DOX 1 ng

DOX 1 μg

Annexin V positive cells (%) Cells with minority
MOMP (%)

0.1 0.5 Merge

*C
DOX (ng/ml): - 2.5 1 DOX (μg/ml): - - - 1 WCE
Cyt c COX IV β-tubulin
Ponceau

1 2.5 1 μg/ml
E 10
8

LZRS
* *

BCL-XL

6

4

2

F
DOX: QVD: -

0 Ctrl

DOX ng 2.5

DOXng 1

2.5 n1gng0.5 ng0.1-ng 2.5 n1gng0.5 ng0.1 ng1 μg 1 μg - - -- -++++ - +
ɣH2A.X
actin PARP tBID (short exp) tBID (long exp)

Cells with minority MOMP (%)
Cells with minority MOMP (%)

G
10 8 6 4 2

U2OS LZRS

*

*

H
U2OS BCL-XL

I
10

FAS (ng/ml): - 10 25 50 - 10 25 50

QVD: - - - - + + + +

γH2A.X

5

α-Tubulin

0

0

Ctrl / CHX CHX

FAS

PDAC LZRS
*

PDAC BCL-XL
*

Ctrl G132μM 5
M 2.

Figure 4. Sub-Lethal BH3-Only Protein and Apoptotic Stress Induces Minority MOMP and DNA Damage (A) MelJuSo tBID tetON or wild-type cells were treated for 12 hr with 1 mg/ml of doxycycline (DOX), and cell lysates were probed for tBID. (B) MelJuSo tBID tetON cells were treated for 6 hr with DOX and cell viability was assessed by Annexin V-based flow cytometry. Data represent mean ± SEM of three independent experiments. *p < 0.05, compared to control. (C) Cytosolic fractions from MelJuSo tetON tBID cells treated as in (B) were probed for cytochrome c. To induce apoptosis, 1 mg/ml DOX was used as a positive control. WCE, whole-cell extract. (D) MelJuSo tBID tetON expressing CytoGFP/MitoCherry were treated with DOX as in (A) and imaged by confocal microscopy. Arrows denote mitochondria undergoing permeabilization. (E) Quantification of cells undergoing minority MOMP. Data represent mean ± SEM of three independent experiments. (F) MelJuSo tBID tetON were treated with DOX as in (A) and cell lysates were probed by western blot for gH2A.X, PARP, and tBID. (G) U2OS cells stably expressing empty vector or BCL-xL were treated for 3 hr with FAS ligand (10 ng/ml) and CHX (1 mg/ml) and scored for the presence of minority MOMP. Data represent mean ± SEM of three independent experiments. (H) U2OS cells were treated for 3 hr with the indicated concentrations of FAS ligand and CHX (1 mg/ml), and western blot was performed for gH2A.X. (I) PDAC cells were treated with MG132 (2.5 mM) for 3 hr, and minority MOMP was quantified. Data represent mean ± SEM of three independent experiments. *p < 0.05, compared versus control. See also Figure S4.

treatment in a dose-dependent manner (Figures 6A and S6B). Ectopic BCL-xL expression inhibited micronuclei accumulation in U2OS cells, confirming that the observed genomic instability

required mitochondrial permeabilization (Figure 6B). In an analogous manner, induction of sub-lethal levels of the BH3only protein tBID in MelJuSo cells also promoted micronuclei

8 Molecular Cell 57, 1–13, March 19, 2015 ª2015 The Authors
MompMinority MompCellsDataAbsence