CTSL2 Is a Pro-Apoptotic Target of E2F1 and a Modulator of Histone Deacetylase Inhibitor and DNA Damage-Induced Apoptosis

ORIGINAL ARTICLE CTSL2 is a pro-apoptotic target of E2F1 and a modulator of histone deacetylase inhibitor and DNA damage-induced apoptosis

CH Wong1,5,ZWu2,5 and Q Yu1,3,4

Aberrant regulation of the pRB/E2F1 pathway has been invariably linked to inappropriate proliferation and/apoptosis in human cancers. Therefore, understanding the intricacies of the signaling pathway and identiﬁcation of novel E2F1 targets involved in apoptosis could pave way for new therapeutic manipulation. Here, we identiﬁed CTSL2 (cathepsin L2/cathepsin V) as a novel E2F1 target that participates in E2F1-dependent apoptosis. We showed that E2F1 directly binds to CTSL2 promoter and that CTSL2 is regulated by both exogenous and endogenous E2F1. RNAi-mediated depletion of CTSL2 effectively abrogated ectopic E2F1-induced apoptosis, coupled with reduced lysosomal membrane permeabilization (LMP) and mitochondrial membrane depolarization. CTSL2 knockdown also inhibited apoptosis mediated by the endogenous E2F1 activated by DNA damage. Furthermore, we showed that CTSL2 depletion in cancer cells resulted in inhibition of histone deacetylase inhibitor (HDACi)- induced apoptosis, and conversely ectopic overexpression of CTSL2-sensitized cancer cells to HDACi. This study uncovered a novel E2F1 target implicated in LMP and apoptosis activation, as well as in the modulation of HDACi and chemotherapeutic drugs response.

Oncogene (2014) 33, 1249–1257; doi:10.1038/onc.2013.72; published online 1 April 2013 Keywords: E2F1; apoptosis; CTSL2

INTRODUCTION Here, through a large-scale gene expression analysis using an E2F transcription factor family members have diverse roles in cellular E2F1-confined system, we identified CTSL2 gene, encoding physiology and homeostasis.1,2 Among them, E2F1 (and perhaps cathepsin like 2 (cathepsin L2/cathepsin V), as a novel E2F3) possesses a unique ability to induce apoptosis,3 which has E2F1 target. We showed that CTSL2 is not only required for been regarded as a fail-safe mechanism to protect cells from E2F1-induced apoptosis by engaging LMP but its level also aberrant oncogenic activation of Rb/E2F1 pathway.4,5 Therefore, modulates the cellular sensitivity to HDACi. These data provide a cancer cells with aberrant activity of E2F1 may preferentially undergo role for CTSL2 as a novel pro-apoptotic target of E2F1, which apoptosisincertaincontextorupondrugtreatment,thusproviding might be potentially important in determining the therapeutic a therapeutic window for tumor eradication. Although it is not response of HDACi. exactly clear how E2F1 possesses biphasic properties on cellular proliferation and death, the critical concentration of E2F1 has been proposed to be a critical regulatingpoint,withhighE2F1level RESULTS preferentially driving cellular proliferation, while an excessive E2F1 Exogenous E2F1-dependent transcriptional profiling identifies level tips the cellular balance to death.6,7 CTSL2 as a potential E2F1 target It is well known that E2F1-induced cell death relies on its ability Induction of E2F1 could lead to transactivation of its downstream to transactivate various target genes of which protein products targets, many of which are potent inducers of apoptosis. are prominent cell death mediators,5,8 in both p53-dependent To systematically identify E2F1 targets, we made use of a and p53-independent manner. For example, E2F1-mediated trans- p53-deficient Saos-2 osteosarcoma cell line that expresses an activation of p19ARF would result in p53 stabilization and exogenous E2F1 fused to the 4-hydroxytamoxifen (4-OHT)- eventual execution of p53-mediated apoptosis,9 whereas the responsive ligand-binding domain of the estrogen receptor (ER). activation of various cell death mediators, such as p73,10,11 In this system, addition of 4-OHT leads to translocation of ER-E2F1 APAF1,12 ASK1,13 caspases14 and pro-apoptotic Bcl-2 family to the nucleus, thereby allowing the induction of E2F1 target members,15,16 may lead to apoptosis in the absence of p53. genes.17 Using this E2F1-confined system, we performed Illumina However, relatively little is known about E2F1 apoptotic targets gene expression analysis and profiled E2F1-dependent gene that are functionally important for the therapeutic response of expression (Figure 1a). Using a twofold cutoff, we identified a anticancer drugs. cluster of genes that are differentially regulated in Saos-2 pBabe

1Cancer Biology and Pharmacology, Genome Institute of Singapore, Agency for Science, Technology and Research, Singapore, Singapore; 2State Key Laboratory of Animal Nutrition, College of Animal Science and Technology, China Agricultural University, Beijing, PRC; 3Department of Physiology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore and 4Cancer and Stem Cell Biology, DUKE-NUS Graduate Medical School of Singapore, Singapore, Singapore. Correspondence: Dr Q Yu, Cancer Biology and Pharmacology, Genome Institute of Singapore, Agency for Science, Technology and Research, 60 Biopolis Street, #02-01, Singapore 138672, Singapore. E-mail: [email protected] 5These authors contributed equally to this work. Received 26 September 2012; revised 3 December 2012; accepted 27 December 2012; published online 1 April 2013 E2F1 target CTSL2 modulates drug-induced apoptosis C Hooi Wong et al 1250 Saos-2

ER-E2F1 pBabe CTSL2: Saos-2 OHT0 8 24 48 0 8 24 48 (hr) 15 * DMSO Saos-2 ER-E2F1 APAF-1 OHT CASP3 CHX OHT - + CCND3 10 CHX+OHT CCNE1 Cyclin E1 Map3k5 EZH2 5 CTSL2 BCL2L11 CCNA1 β Relative mRNA level -Actin CTSA 0 CTSB ER-E2F1 ER-E2F1 E132 CTSK CTSL2 CTSF CTSL2: U2OS CTSE OHT 0 hr CTSL1 15 OHT 24 hr U2OS ER-E2F1 CTSS OHT 48 hr CTSZ * CTSC 10 4-OHT - + CTSW CTSG Cyclin E1 CTSL1 5 CTSO CTSL2 CTSC Relative mRNA level CTSD β-Actin 0 CTSH pBabe ER-E2F1 CTSK3 Figure 1. Ectopic E2F1 induces CTSL2 expression. (a) Gene expression heatmap of representative genes that are differentially regulated in Saos-2 cells expressing pBabe (vector control) or ER-E2F1. Saos-2 cells infected with a retrovirus expressing ER-E2F1 or the empty vector (pBabe) were treated with OHT (300 nM) for various time points as indicated. Red column represents genes with higher expression levels; green column represents genes with lower expression levels. (b) Quantitative real-time PCR (qRT-PCR) and western blot of CTSL2 mRNA and protein level in Saos-2 ER-E2F1 cells or ER-E2F1/E132 cells with or without OHT (300 nM) treatment for 24 h. Saos-2 ER-E2F1 and ER-E2F1/E132 cells were treated with or without cycloheximide (10 mg/ml) for 8 h before OHT treatment. (c) Left: qRT-PCR of CTSL2 mRNA expression in U2OS cells infected with an empty vector control (pBabe) or ER-E2F1 and treated with OHT (300 nM) for 24 h and 48 h. Total RNA was isolated and analyzed as described in Materials and methods. Right: western blot of CTSL2 protein level in U2OS ER-E2F1 cells with or without OHT (300 nM) treatment for 24 h. Results are representative of three independent experiments. *Po0.05.

and ER-E2F1 cells. In addition to many previously known E2F1 and protein expression, indicating a causal role for endogenous targets, such as CCNE1, APAF1 and caspase-3, we noticed that E2F1 in regulating CTSL2 (Figure 2b). CTSL2, a previously undescribed target, was robustly induced Endogenous E2F1 can also be activated via protein stabilization following the 4-OHT treatment over different time course by DNA damage.18,19 Consistently, we showed that the treatment (Figure 1a). Of further notice, CTSL2 seemed to be the most of DNA-damaging agents, adriamycin and etoposide (also known highly induced gene by E2F1, as compared with the rest of the as VP-16) in U2OS cells, resulted in marked induction of CTSL2 at cathepsin family member. We, thus, chose to focus on CTSL2, as both mRNA and protein levels accompanying E2F1 accumulation the role for CTSL2 in apoptosis regulation has not been previously (Figure 2c), which was abolished upon E2F1 knockdown studied. The array observation was further validated by quantita- (Figure 2d). Taken together, these data provide substantial tive reverse transcriptase-PCR analysis, which shows the induction evidence that CTSL2 is regulated by E2F1 expressed both of CTSL2 mRNA upon 4-OHT treatment in ER-E2F1-expressing exogenously and endogenously. Saos-2 cells, but not in Saos-2 cells expressing a DNA-binding- To further validate the relationship between E2F1 and CTSL2 deﬁcient ER-E2F1 mutant (E132) (Figure 1b). Moreover, induction expression in a clinically relevant setting, we interrogated a gene of CTSL2 mRNA could not be inhibited by pre-treatment of the expression data set that consists of a panel of 24 paired primary protein synthesis inhibitor, cyclohexamide, indicating that CTSL2 colorectal tumor and matched control samples.20 We found that could be a direct target of E2F1 (Figure 1b). Western blot analysis E2F1 and CTSL2 share a striking similarity in their expression in Saos-2 ER-E2F1 cells treated with 4-OHT also conﬁrmed the pattern in these paired tumor and normal samples (Figure 2e), upregulation of CTSL2 protein concomitant with the known E2F1 with a Pearson correlation of 0.692 (Po0.001). This supports a target Cyclin E (Figure 1b, right panel). A similar observation of direct correlation between E2F1 and CTSL2 expression level in CTSL2 mRNA and protein inductions by E2F1 were also found in patient samples. Taken together, the data support the endogen- U2OS ER-E2F1 cells (Figure 1c). ous regulation of CTSL2 by E2F1 and indicate that CTSL2 might be a clinically relevant E2F1 target in human cancer.

CTSL2 is regulated by endogenous E2F1 and DNA damage response E2F1 activates CTSL2 gene promoter activity We next investigated whether CTSL2 expression is also regulated Sequence analysis of the CTSL2 promoter identified a putative by endogenous E2F1. To this end, we began with the human lung E2F1-binding site (50-TTT(C/G)GGGC-30) in the proximal promoter fibroblast cells IMR90 and IMR90 cells transformed with oncopro- region (50-TTTCGCTC-30, at position À 1180/ À 1173) (Figure 3a). tein E1A (IMR90/E1A). IMR90/E1A is known to activate the To demonstrate the potential activity of E2F1 on CTSL2 promoter, endogenous E2F1 by inhibiting the tumor suppressor pRB.14 we cloned the CTSL2 promoter region pGL3-CTSL2/F1 ( À 1250/ Indeed, we detected a much higher level of both E2F1 and CTSL2 þ 200) and a region without the E2F1-binding site pGL3-CTSL2/F2 mRNA and protein level in IMR90/E1A cells compared with non- ( À 850/ þ 200) into the pGL3-luciferase reporter construct and transformed IMR90 cells (Figure 2a). Moreover, E2F1 knockdown in performed the reporter activity assay using U2OS cells. IMR90/E1A cells resulted in a significant reduction in CTSL2 mRNA Co-transfection of E2F1 and pGL3-CTSL2/F1 ( À 1250/ þ 200)

Oncogene (2014) 1249 – 1257 & 2014 Macmillan Publishers Limited E2F1 target CTSL2 modulates drug-induced apoptosis C Hooi Wong et al 1251

E2F1 IMR90/E1A: mRNA CTSL2 E2F1 IMR90/E1A 200 ** 1.5 CTSL2 siE2F1 - + 150 IMR90 IMR90/E1A 1.0 * * * E2F1 100 E2F1 level 0.5 CTSL2 50 CTSL2

Relative expression β-Actin Relative mRNA level 0.0 0 β-Actin IMR90 IMR90/E1A SiNC SiE2F1

CTSL2: U2OS U2OS 15 ADR - + +-- U2OS ETO ---++ SiNC siE2F1 SiNC siE2F1 10 M M M M ADR -+-+ETO - + - + μ μ μ μ 2 1 20

10 E2F1 5 E2F1 CTSL2 PARPc Relative mRNA level CTSL2 0 β-actin β-actin ETO ADR DMSO

E2F1 mRNA 800 700 600 500 400 300 200 100 Expression value 0 1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627282930313233343536373839404142434445464748 Correlation CTSL2 mRNA =0.692 3500 3000 2500 2000 1500 1000 500 Expression value 0 12345678910111213 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Normal (n=24) Tumor (n=24) Figure 2. Endogenous E2F1 regulates CTSL2 expression. (a) Quantitative real-time PCR (qRT-PCR) and western blot analysis of endogenous E2F1 and CTSL2 mRNA and protein level in IMR90 and IMR90/E1A cells. **Po0.01 (b) qRT-PCR and western blot analysis of E2F1 and CTSL2 level following transient knockdown of E2F1 in IMR90/E1A cells, as described previously.33 E2F1 and CTSL2 mRNA and protein expression were analyzed as in Figure 1b. ***Po0.001 (c) Left: qRT-PCR of CTSL2 level in U2OS cells treated with adriamycin (ADR; 1 mM) and etoposide (ETO; 10 mM) for 24 h. Right: western blot showing CTSL2 and E2F1 expression in U2OS cells following ADR and ETO treatment for 24 h at various doses as indicated. (d) Western blot of U2OS cells treated with ADR (1 mM) or ETO (10 mM) for 24 h upon transient siRNA-mediated silencing against E2F1, as described previously.33 (e) Raw expression values of E2F1 and CTSL2 mRNA levels in 24 paired colorectal normal and tumor samples extracted from our previously published microarray data (GEO accession number GSE10972). Correlation between E2F1 and CTSL2 mRNA level was measured as 0.692 by Pearson’s product moment correlation with a P-value o0.001. resulted in a robust induction of luciferase reporter activity Figure 4a. Endogenous E2F1 pull down was performed with a (Figure 3b), whereas the pGL3-CTSL2/F2 ( À 850/ þ 200) was specific E2F1 monoclonal antibody in U2OS cells treated with unresponsive to E2F1 (Figure 3b). A similar finding was also etoposide (1 mM) for 24 h, followed by genomic PCR analysis of the observed in HCT116 and HEK 293T cells (Figures 3c and d). above three regions. Of notice, PCR analysis of the proximal promoter region encompassing the putative E2F1-binding site ( À 1180/ À 1173, P3) showed the highest enrichment of E2F1 CTSL2 is a direct binding target of E2F1 following the drug treatment (Figure 4b), while there is no Next, we performed chromatin immunoprecipitation analysis to significant enrichment in more distant regions (P1 and P2). determine whether E2F1 can directly bind to the CTSL2 gene This finding supports that E2F1 is recruited to the CTSL2 promoter in vivo. We used three different pairs of primers (P1 to promoter following DNA damage and activates its transcription. P3) targeting different regions of CTSL2 promoter as shown in In addition, E2F1 pull down in IMR90/E1A cells at promoter region

& 2014 Macmillan Publishers Limited Oncogene (2014) 1249 – 1257 E2F1 target CTSL2 modulates drug-induced apoptosis C Hooi Wong et al 1252 CTSL2 promoter U2OS 10 TTTCGCTC E2F1 0 ng 8 E2F1 50 ng Exon 1 6

Luc 4 CTSL2/F1(-1250/+200) 2 Luc CTSL2/F2(-850/+200) Relative luciferase activity 0 Luc pGL3 basic pGL3 pGL3 CTSL2/F1 CTSL2/F2 E2F binding site

HCT116 HEK 293T E2F1 0 ng 4 15 E2F1 0 ng E2F1 50 ng E2F1 50 ng 3 10 2 5 1

0 Relative luciferase activity 0 Relative luciferase activity pGL3 basic pGL3 pGL3 pGL3 basic pGL3 pGL3 CTSL2/F1 CTSL2/F2 CTSL2/F1 CTSL2/F2 Figure 3. E2F1 activates CTSL2 gene promoter. (a) Schematic representation of the human CTSL2 promoter. Putative E2F1-binding sites were predicted using TFSEARCH program (www.cbrc.jp/research/db/TFSEARCH.html). The proximal promoter region (F1) and its deletion construct (F2) were cloned into pGL3 reporter plasmid. (b) Luciferase reporter activity in U2OS cells upon E2F1 activation. (c) Luciferase reporter activity in HCT116 cells upon E2F1 activation. (d) Luciferase reporter activity in HEK293T cells, cells were collected and analyzed as according to Figure 2b. All experiments were performed three times independently.

500bp +1 CTSL2

P1 P2 P3 (-1180/-1173)

U2OS IMR90/E1A ChIP-E2F1 * ChIP-E2F1 ChIP-IgG ChIP-E2F1 5 * DMSO 3 4 VP16 3 2 2 1

Fold enrichment 1 Fold enrichment 0 0 P1 P2 P3 P3 Figure 4. CTSL2 is a direct binding target of E2F1. (a) Schematic representation of CTSL2 gene promoter region. (b) Chromatin immunoprecipitation (ChIP) assay of E2F1 binding on CTSL2 promoter in U2OS cells treated with or without etoposide (ETO; 10 mM) for 24 h. (c) ChIP assay of E2F1 occupancy on CTSL2 promoter in vivo in IMR90/E1A cells. All experiments were performed three times independently. *Po0.05.

( À 1180/ À 1173, P3) also detected a significant enrichment of upon CTSL2 knockdown by two specific small interfering RNAs E2F1 compared with the IgG control (Figure 4c). Collectively, these (siRNAs; 14.4% and 19.7%, respectively) (Figure 5a). Western blot data clearly established CTSL2 as a direct binding target of E2F1. analysis indicated reduced PARP cleavage upon 4-OHT treatment in both CTSL2 siRNA and short-hairpin RNA-mediated gene silencing, suggesting that CTSL2 is involved in a caspase- CTSL2 is required for E2F1-induced LMP and apoptosis dependent apoptotic cell death (Figure 5b). To further ascertain We next assessed whether CTSL2 induction is involved in E2F1- a role of CTSL2 in E2F1-induced apoptosis, we performed FACS- induced apoptosis. In U2OS ER-E2F1 cells, exogenous E2F1 based analysis on caspase-3 activity (Figure 5c) and mitochondrial activation by 4-OHT treatment for 48 h resulted in marked cell membrane potential by JC-1 staining (Figure 5d). From the death (47.14%), as assessed by fluorescence-activated cell sorting results of both assays, we conclude that CTSL2 knockdown (FACS) analysis of cells in sub-G1, which was markedly reduced markedly inhibited E2F1-induced apoptotic events. Collectively,

Oncogene (2014) 1249 – 1257 & 2014 Macmillan Publishers Limited E2F1 target CTSL2 modulates drug-induced apoptosis C Hooi Wong et al 1253 U2OS ER-E2F1

SiNC siCTSL2#1 siCTSL2#2

Sub-G1: Sub-G1: Sub-G1: U2OS ER-E2F1 U2OS ER-E2F1 1200 800 3.91% 600 2.45% 2.37% 900 600 400 600 600 80 Ctrl Number * Number Number 200 - OHT 300 200 SiNC shCTSL2 shNC * siCTSL2#1 siCTSL2#2 60 0 0 0 0 306090120 150 0 50 100 150 0 50 100 150 4-OHT - +-+-+ -+-+ Chanels (FL2-A) Channels (FL2-A) Channels (FL2-A) 40 CTSL2

Sub-G1 (%) 20 PARPc 320 Sub-G1: 230 Sub-G1: 240 Sub-G1: 47.14% 14.59% 19.79% 240 210 130 β-Actin OHT 0 160 140 120

Number SiNC siCTSL2#1 siCTSL2#2 Number Number 80 70 60

0 0 0 0 50 100 150 0 50 100 150 0 50 100 150 Channels (FL2-A) Channels (FL2-A) Channels (FL2-A)

SiNC siCTSL2#1 siCTSL2#2 U2OS ER-E2F1 U2OS ER-E2F1 60 Ctrl 40 Ctrl OHT OHT - 40 30

20 20 JC-1+ (%) 10 Active Casp-3 (%) OHT 0 0 SiNC siCTSL2#1 siCTSL2#2 SiNC siCTSL2#1 siCTSL2#2

U2OS U2OS U2OS ER-E2F1 15 Ctrl Ctrl 80 * OHT ADR ** ETO 10 60 ** siCTSL2#1 siCTSL2#2 siCTSL2#1 siCTSL2#2 SiNC SiNC

40 ADR - + - + - + ETO - + - + - + 5 % Pale Cells

Sub-G1 (%) CTSL2 20 PARPc 0 SiNC siCTSL2#1 siCTSL2#2 0 β-Actin SiNC siCTSL2#1 siCTSL2#2

Figure 5. CTSL2 is required for E2F1-dependent apoptosis. (a) Left panel: cell-cycle profile of U2OS ER-E2F1 cells treated with OHT (300 nM) upon CTSL2 knockdown. The percentage of cells in the sub-G1 fraction is depicted in the upper right hand corner. Cells were collected 72 h post OHT treatment and fixed in 70% ethanol for 1 h. Fixed cells were stained with propidium iodide (50 mg/ml) for DNA content analysis by FACS in a FACS Calibur (Becton Dickinson Instrument) System. Cell-cycle fractions were quantified using the CellQuest software (Becton Dickinson). (a) Right panel: graphical representation of sub-G1 fraction in Figure 1a of U2OS ER-E2F1 cells, indicative of apoptotic cells. (b) Western blot of U2OS ER-E2F1 cells treated with OHT (300 nM) upon transient CTSL2 knockdown or stable knockdown by using short- hairpin RNA against CTSL2. (c) Cells were treated for 48 h following CTSL2 knockdown, and active caspase-3 staining was performed. (d) Cells were treated for 48 h following CTSL2 knockdown, and JC-1 staining was performed (e) Following CTSL2 knockdown, cells were reseeded into a coverglass slide chamber. The following day cells were treated with OHT (300 nM) for 24 h and then stain with AO before being analyzed by confocal microscopy. (f) Flow cytometric analysis of lysosomal membrane stability by measuring the decrease in red fluorescence, detected by the AO uptake method. (g) Percentage cell death of U2OS cells treated with adriamycin (ADR; 1 mM) or etoposide (ETO; 10 mM) for 72 h upon CTSL2 knockdown. (h) Western blot of U2OS cells treated with ADR (1 mM) or ETO (10 mM) upon transient siRNA-mediated silencing against CTSL2, as described in Figure 3a. All the data in the bar graphs represents mean±s.e.m., n ¼ 3. *Po0.05; **Po0.01. these results indicate that E2F1-induced CTSL2 activation partici- by using AO uptake method.22 The reduction of red fluorescence pates in apoptotic cell death. of AO resulted in the increase of ‘pale’ cells, which indicates a As CTSL2 is a lysosomal protease, we next investigated the reduced number of intact lysosomes. Again, addition of 4-OHT possibility that E2F1 induces LMP through induction of CTSL2. resulted in the increase of ‘pale’ cells in control cells, which was To do this we made use of a lysomotropic probe, acridine largely abrogated by CTSL2 knockdown (Figure 5f). Taken orange (AO), which fluoresces red in acidic organelles such as together, these data suggested that induction of CTSL2 by E2F1 lysosomes, while it remains in low concentration and fluoresces lead to LMP, mitochondrial membrane depolarization and green in the nucleus.21 Indeed, 4-OHT treatment of U2OS ER-E2F1 caspase-3 activation, which eventually culminated in apoptotic cells for 24 h resulted in a disappearance of red fluorescence cell death. of AO staining as well as a diffused red staining in the lysosomes, Next, we evaluated a role for CTSL2 in DNA damage-induced consistent with an LMP phenotype (Figure 5e). This phenotype apoptosis, as we showed above that DNA-damaging agents can was effectively rescued by CTSL2 knockdown (Figure 5e). We also also activate E2F1-dependent CTSL2 expression (Figure 1c). In observed enhancement of green fluorescence of the nucleus, U2OS cells, adriamycin or etoposide treatment for 48 h resulted in which may indicate nuclei acidification, which was consistent robust apoptosis, which was efficiently rescued upon CTSL2 with previous report21 (Figure 5e). To add credence to this knockdown, as examined by both FACS analysis and western blot observation, we performed lysosomal membrane stability assay analysis of PARP cleavage (Figures 5c and d). Collectively, these

& 2014 Macmillan Publishers Limited Oncogene (2014) 1249 – 1257 E2F1 target CTSL2 modulates drug-induced apoptosis C Hooi Wong et al 1254

CTSL2 mRNA HCT116 P53 WT: HCT116 P53 KO: CTSL2 mRNA CTSL2 mRNA 15 1.5 1.5

10 1.0 1.0

5 0.5 0.5

0 Relative expression level

Relative expression level 0.0 0.0

Normal SW620 RKO HT29 HCT116 Relative expression level Colon SiNC siCTSL2#1 siCTSL2#2 SiNC siCTSL2#1 siCTSL2#2

60 SW620 DMSO 100 DMSO 80 SAHA SAHA 3 μM DMSO SAHA 40 TSA 80 60 TSA 200 nM TSA * 60 40 * * * 20 * * Sub-G1 (%) 40 * Sub-G1 (%) 20 * Sub-G1 (%) * 0 20 0 SiNC siCTSL2#1 siCTSL2#2 SiNC siCTSL2#1 siCTSL2#2 0 PMN CTSL2 U2OS U2OS SW620 PMN CTSL2 PMN CTSL2 PMN CTSL2 PMN CTSL2 *

100 M M M M DMSO μ μ μ μ

M M M M SAHA 3 mM μ μ μ μ 80 TSA 200 nM DMSO DMSO DMSO * DMSO 60 SAHA 2 SAHA 3 SAHA 2 SAHA 3 TSA 200 nM TSA 300 nM TSA 300 nM TSA 300 nM DMSO DMSO DMSO DMSO SAHA 2 SAHA 3 SAHA 2 SAHA 3 TSA 200 nM TSA 300 nM TSA 300 nM TSA 300 nM 40 PARPc PARPc Sub-G1 (%) 20 HA HA 0 β-actin β-actin PMN CTSL2 Figure 6. CTSL2 is critical for HDACi-induced apoptosis. (a) Relative CTSL2 expression level of normal colon and colorectal cancer lines as assessed by quantitative real-time PCR (qRT-PCR) assay. (b) Upper panel: bar chart showing knockdown efficiency of CTSL2 in HCT116 P53 WT and P53 KO cells. Lower panel: percentage cell death of HCT116 P53 WT and P53 KO cells upon siCTSL2 transfection followed by suberolynic acid (SAHA; 3 mM) or trichostatin A (TSA; 200 nM) treatment for 72 h. Cell death is assessed by quantifying the sub-G1 fraction of the cell cycle following propidium iodide (PI) staining as described in Figure 3a. *Po0.05 of the respective knockdown treated sample as compared with NC-treated sample. (c) Upper panel: bar chart showing percentage cell death of SW620 cells stably transfected with CTSL2 conjugated with hemagglutinin (HA tag; CTSL2-HA) or PMN vector followed by SAHA (3 mM) or TSA (200 nM) treatment for 48 h. Following treatment, cells were fixed and stained by PI, as described previously in this paper. (c) Lower panel: western blot of SW620 cells stably transfected with CTSL2 or PMN vector followed by SAHA (3 mM) or TSA (200 nM) treatment for 48 h. HA level was used to determine CTSL2-HA transfection efficiency. (d) Left panel: percentage cell death of U2OS cells upon siCTSL2 transfection followed by SAHA (3 mM) or TSA (200 nM) treatment for 72 h. (d) Right panel: western blot showing PARP cleavage of U2OS cells stably transfected with CTSL2-HA or PMN vector followed by SAHA (3 mM) or TSA (200 nM) treatment for 48 h. All the data in the bar graphs represents mean þ s.e.m., n ¼ 3. *Po0.05.

data suggest a role of CTSL2 in E2F1-mediated apoptosis and Of notice, CTSL2 overexpression alone did not induce apoptosis participates in DNA damage-induced apoptotic response. but promoted HDACi-induced apoptosis, which indicates that CTSL2 is not a sufficient apoptosis inducer but is required for priming the cancer cells to undergo apoptosis in stress conditions. CTSL2 is critical for HDACi-induced apoptosis Again, the ability of CTSL2 in promoting apoptosis does not HDACi are novel classes of anticancer agents.23 One of the require normal p53 as SW620 is a p53-mutant cell line. Moreover, proposed mechanisms for HDACi anti-tumor effect involves the in the non-colorectal cancer cell line U2OS, CTSL2 overexpression apoptotic induction through various pathways, including E2F1- also sensitized suberolynic acid and trichostatin A-induced dependent apoptosis.13,16 We therefore sought to evaluate a apoptosis (Figure 6d). These findings lend credence to the CTSL2 possible role of CTSL2 in HDACi-induced apoptosis. We have knockdown data in HCT116 cells and further supported that CTSL2 shown above that CTSL2 expression exhibits various expression expression level might be an important determinant of HDACi- levels in colorectal cancer tumor samples (Figure 2e). We also induced apoptosis in human cancer cells. found that CTSL2 have a wide range of expression levels in a panel of colorectal cancer cell lines in which HCT116 and HT29 cells have markedly higher CTSL2 expression than SW620, RKO and normal DISCUSSION colon (Figure 6a). To determine a role of CTSL2 in HDACi response, Here, using multiple approaches we found and validated a novel we performed CTSL2 knockdown in HCT116 cells that shows and direct E2F1 target, CTSL2, that is implicated in cancer cell highest expression of CTSL2. CTSL2 knockdown in both p53 apoptotic induction. Importantly, we also demonstrated that wild-type and p53-deficient HCT116 cells markedly reduced CTSL2 could be an important molecular determinant of cell death the apoptosis response to two different HDACi, suberolynic acid response to cancer therapeutic treatments. This has advanced our (SAHA) and trichostatin A (TSA) (Figure 6b). This indicates that understanding of E2F1-mediated cell death pathway in cancer expression of CTSL2 is required for HDACi-induced cell death treatment, as till date very few E2F1 targets have been functionally and this effect is p53-independent. Conversely, enforced over- linked to therapeutic response. expression of CTSL2 in SW620 cells, which possessed the lowest CTSL2 was first cloned in 1998 and belongs to the human expression of CTSL2, enhanced HDACi-induced apoptosis, as papain family of cysteine proteinases.24 Because of its high examined by both FACS and PARP cleavage analyses (Figure 6c). sequence similarity with another member of the papain family,

Oncogene (2014) 1249 – 1257 & 2014 Macmillan Publishers Limited E2F1 target CTSL2 modulates drug-induced apoptosis C Hooi Wong et al 1255 cathepsin L, the novel gene is termed CTSL2 or cathepsin L2. MATERIALS AND METHODS Previously, cathepsin L2 was found to be highly expressed in the Cell culture and chemicals thymus, and was reported to be involved in positive selection of The human colorectal carcinoma HCT116 cells and its derivative isogenic 25 T-helper cells. In addition, CTSL2 was found to be differentially p53 À / À HCT116 cells were kindly provided by Dr Bert Vogelstein (John expressed in colorectal and breast tumor tissue as compared Hopkins University, MD, USA). IMR90-E1A cells were obtained from with normal cells,24 but its role in cancer has not been assigned Dr C. Brancolini (University of Udine, Udine, Italy). Other cell lines till date. In this study, we provide compelling evidence that used in this study, including human osteosarcoma U2OS and Saos-2, CTSL2 is a direct target of E2F1, and its induction is required human lung fibroblast IMR90, colon cancers SW620 and RKO, were for E2F1-induced apoptosis. CTSL2 was also upregulated upon obtained from the American Type Culture Collection (Manassas, VA, USA). Saos-2 and U2OS cells stably expressing ER-E2F1 were generated as chemotherapy-induced DNA damage, which activates described previously.16 All cell lines were maintained in Dulbecco’s endogenous E2F1. As such, knockdown of CTSL2 reduced the modified Eagle’s medium supplemented with 10% fetal bovine serum apoptotic response to chemotherapeutic drugs. This indicates that and antibiotics in a 37 1C humidified incubator containing 5% CO2. CTSL2 induction by E2F1 may be crucial for DNA-damaging 4-OHT, adriamycin and ETO were purchased from Sigma-Aldrich chemotherapeutic agents to induce efficient cell death in cancer (St Louis, MO, USA). Trichostatin A was purchased from Cell Signaling cells. Unlike other well-known E2F1 pro-apoptotic targets, such as Technology (Beverly, MA, USA) and suberolynic acid was from Alexis caspases and Apaf1, that induce apoptosis through mitochondrial Biochemicals (San Diego, CA, USA). Lysotracker Green DND and AO dyes death pathway, the mechanism by which CTSL2 induces apoptosis were purchased from Invitrogen (Carlsbad, CA, USA). Active Caspase-3 remains to be determined. However, other papain family staining kit and JC-1 staining kit were from Becton Dickinson (San Jose, CA, USA). members such as cathepsin D and cathepsin B are well known to induce apoptosis by LMP.26 Following LMP, cathepsins are also reported to translocate from the lysosomes to the mitochondria, Plasmids and stable cell lines and have been shown to proteolytically cleave downstream targets, such as Bid and caspases, leading to cell death pcDNA4-E2F1 was generated by real-time PCR using total RNA from 27,28 normal colon tissue. Retroviral-mediated gene transfer was performed execution. In agreement with our expectation, we demon- using pMN-GFP/IRES retrovirus vector-expressing CTSL2 tagged to strated a role of CTSL2 in LMP activation associated with hemagglutinin. Infected cells were sorted by green fluorescent protein E2F1-induced apoptosis. It is possible that the proteolytic signals and expanded for in vitro studies. All constructs were confirmed activity of CTSL2 contributes to the leakiness of lysosomal free of mutation by sequencing. membranes, resulting in the release of lysosomal contents into the cytosol. As LMP was shown to be an early event in apoptosis, it could be probable that LMP mediated by CTSL2 could lead to RNA interference mitochondrial membrane depolarization and eventual apoptosis. Specific siRNA oligos targeting E2F1 and CTSL2 mRNAs and non-targeting These results have provided a novel mechanistic insight into E2F1- control siRNA were from 1st Base Technology (Singapore). Cells were induced apoptosis and uncovered a novel role for E2F1 in the sequentially transfected two times at 48-h interval to secure efficient induction of LMP through transcriptional activation of CTSL2. gene silencing. Cells were transfected with 100 nM final concentration of E2F1-induced death pathways have been linked to HDACi- siRNA duplexes using Lipofectamine RNAiMax (Invitrogen) following the induced apoptosis.13,29 We have previously shown that Rb/E2F1 manufacturer’s instructions. SiCTSL2#1 targets the following sequence: 50-CGGCTTTGAAGGAGCAAAT-30. SiCTSL2#2 targets the following pathway is a target for HDACi-induced apoptosis, and that this sequence: 50-CTTATTGTGTAAGATAAAC-30. Primer used for non-targeting pathway is dependent on the activation of Bim, a pro-apoptotic control siRNA contains the following sequence: 50-TTCTCCGAACGTGT- Bcl-2 family member that has a critical role in mitochondrial- CACGTTT-30. To generate CTSL2 short-hairpin RNA stable cell lines, siRNA 29 dependent apoptosis. Moreover, we have also discovered that oligos targeting CTSL2 (target sequence: 50-CGGCTTTGAAGGAGCAAAT-30) ASK1 is a novel direct target of E2F1 in response to HDACi were cloned into the pSIREN-RetroQ ZS Green retroviral expression vector treatment, which regulates E2F1 activity through a positive (BD Biosciences, Rockville, MD, USA) according to the manufacturer’s feedback mechanism.13 We showed in this study that CTSL2 instruction. Virally infected cells were selected via green fluorescent expression as an E2F1 pro-apoptotic target is required for efficient protein Sorting with FacsAria. apoptosis induction by HDACi. In recent years, HDACi has emerged as a novel class of cancer therapeutic agents because of its tumor selectivity. Despite the success of HDACi in Dual luciferase reporter assays hematological malignancies, such as cutaneous T-cell Genomic DNA containing the human CTSL2 promoter elements were lymphoma,30 the use of HDACi for treatment of solid malig- cloned into pGL3-luciferase reporter construct (Promega, Madison, WI, nancies has been far from encouraging and resulted in various USA). HCT116 and HEK293T cells were transfected with either pcDNA4 empty vector (50 ng) or pcDNA4-E2F1 (50 ng) along with 10 ng of pRL-Null patients’ response to the treatment. Hence, there is great clinical Renilla Luciferase and pGL3 basic constructs as indicated in Figure 2a. interest in developing biomarker strategies to stratify solid tumor Transient transfection was performed using Fugene HD transfection patients for HDACi treatment. In line with this effort, we showed reagents (Promega). Transfected cells were collected 48 h upon transfec- that CTSL2 expression could be an important determinant of tion, and relative luciferase activity was measured by the Dual Luciferase HDACi sensitivity. Indeed, CTSL2 overexpression resulted in a System (Promega). Data are represented as fold change normalized with dramatic increase in cancer cell death response to HDACi the respective Renilla luciferase reading. treatment, whereas its knockdown reduced the sensitivity to HDACi. Given a wide range of CTSL2 expression levels as we have shown in primary colorectal tumor samples, we postulate Immunoblot analysis that CTSL2 could be a potential biomarker candidate for the Protein extracts were prepared by lysis in RIPA buffer (containing 50 mM prediction for HDACi response in colorectal tumors. Furthermore, Tris-HCl, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxy- CTSL2 is expressed in lower levels in normal colon tissues cholate, 0.1% SDS, 1 mM EDTA, 50 mM NaF, 0.1 mM Na3VO4) and protease compared with the tumor samples, which may, at least in inhibitor cocktail (Roche, Indianapolis, IN, USA). Lysates were resolved by SDS–polyacrylamide gel electrophoresis, transferred onto an immobilon part, provide the tumor selectivity of HDACis. Collectively, these membrane (Millipore, Billerica, MA, USA) and probed with indicated results uncovered a novel E2F1 pro-apototic target that may antibodies. Antibodies used in this study include: anti-E2F1 (Santa Cruz, CA, determine the sensitivity of cancer cells to DNA-damaging USA), anti-CTSL2 (Abcam Inc, Cambridge, MA, USA), anti-cleaved PARP agents and HDACi, which may have translational value for (Cell Signaling Technology), anti-HA mouse monoclonal (Santa Cruz) and cancer therapeutics. anti-actin mouse monoclonal antibody (Sigma-Aldrich).

& 2014 Macmillan Publishers Limited Oncogene (2014) 1249 – 1257 E2F1 target CTSL2 modulates drug-induced apoptosis C Hooi Wong et al 1256 Microarray gene expression analysis and quantitative real-time CONFLICT OF INTEREST PCR assay The authors declare no conflict of interest. Total RNA was isolated using Trizol (Invitrogen) and purified with the RNeasy Mini Kit (Qiagen, Hilden, Germany). Reverse transcription was performed using an RNA Amplification kit (Ambion, Carlsbad, CA, USA). The microarray hybridization was performed using the Illumina Gene Expres- ACKNOWLEDGEMENTS sion Sentrix BeadChip HumanRef-8_V2 (Illumina, San Diego, CA, USA), and We thank Dr Kristian Helin for the ER-E2F1 plasmids, Dr Joseph Nevins for the data analysis was performed using GeneSpring software from Agilent adenoviral E2F1, Dr Andrew Turnell for adenoviral E1A and Dr Claudio Brancolini for 31 Technologies (Santa Clara, CA, USA) as described. qRT-PCR was assessed IMR90-E1A cells. We are grateful to Li Juntao for providing assistance in data analysis. using the PRISM 7900 Sequence Detection System (Applied Biosystems, This work was supported by the Agency for Science, Technology and Research for Carlsbad, CA, USA) with specific probes from 1st Base Technology. Samples Singapore. were normalized to the levels of 18S mRNA. Primer sequences are available upon request.

REFERENCES Flow cytometry and cell-cycle analysis 1 Wu Z, Zheng S, Yu Q. The E2F family and the role of E2F1 in apoptosis. Int J Cells were collected and fixed in 70% ethanol. Fixed cells were stained with Biochem Cell Biol 2009; 41: 2389–2397. propidium iodide (50 mg/ml) after treatment with RNase A (100 mg/ml). 2 Iaquinta PJ, Lees JA. Life and death decisions by the E2F transcription factors. Curr The stained cells were analyzed for DNA content by FACS in a FACS Calibur Opin Cell Biol 2007; 19: 649–657. (Becton Dickinson Instrument). Cell-cycle fractions were quantified using 3 Polager S, Ginsberg D. E2F—at the crossroads of life and death. Trends Cell Biol the CellQuest software (Becton Dickinson). 2008; 18: 528–535. 4 Lowe SW, Cepero E, Evan G. Intrinsic tumour suppression. Nature 2004; 432: 307–315. Active caspase-3 staining 5 Ginsberg D. E2F1 pathways to apoptosis. FEBS Lett 2002; 529: 122–125. Cells were treated with 4-OHT for 48 h. Cells were then collected and 6 Hallstrom TC, Mori S, Nevins JR. An E2F1-dependent gene expression program stained with anti-caspase-3 antibody conjugated to FITC according to that determines the balance between proliferation and cell death. Cancer Cell manufacturer’s instructions (Becton Dickinson), and fluorescence was 2008; 13: 11–22. measured using CellQuest software (Becton Dickinson). 7 Dynlacht BD. Live or let die: E2F1 and PI3K pathways intersect to make life or death decisions. Cancer Cell 2008; 13: 1–2. 8 Putzer BM. E2F1 death pathways as targets for cancer therapy. J Cell Mol Med JC-1 staining 2007; 11: 239–251. Cells were treated with 4-OHT for 48 h. Cells were then collected and 9 Bates S, Phillips AC, Clark PA, Stott F, Peters G, Ludwig RL et al. p14ARF links the stained with JC-1 dye according to the manufacturer’s instructions (Becton tumour suppressors RB and p53. Nature 1998; 395: 124–125. Dickinson), and fluorescence were measured using CellQuest software 10 Stiewe T, Putzer BM. Role of the p53-homologue p73 in E2F1-induced apoptosis. (Becton Dickinson). Nat Genet 2000; 26: 464–469. 11 Irwin M, Marin MC, Phillips AC, Seelan RS, Smith DI, Liu W et al. Role for the p53 homologue p73 in E2F-1-induced apoptosis. Nature 2000; 407: 645–648. Chromatin immunoprecipitation assay 12 Moroni MC, Hickman ES, Lazzerini Denchi E, Caprara G, Colli E, Cecconi F et al. Chromatin immunoprecipitation assays were performed, as described Apaf-1 is a transcriptional target for E2F and p53. Nat Cell Biol 2001; 3: 29 previously. Briefly, sonicated extracts were precleared and incubated 552–558. with antibodies specific to either E2F1 (c-20, Santa Cruz) or IgG control (SC- 13 Tan J, Zhuang L, Jiang X, Yang KK, Karuturi KM, Yu Q. Apoptosis signal-regulating 2027, Santa Cruz) at 4 1C overnight on a 360-1C rotator. The immuno- kinase 1 is a direct target of E2F1 and contributes to histone deacetylase precipitated DNA was quantitated by quantitative real-time PCR using inhibitor-induced apoptosis through positive feedback regulation of E2F1 apop- Applied Biosystems 7900HT Fast Real-Time PCR System (Applied totic activity. J Biol Chem 2006; 281: 10508–10515. Biosystems). The enrichment of E2F1 binding at the examined regions 14 Nahle Z, Polakoff J, Davuluri RV, McCurrach ME, Jacobson MD, Narita M et al. was quantitated relative to the input DNA followed by normalization to the Direct coupling of the cell cycle and cell death machinery by E2F. Nat Cell Biol respective IgG control values. Primer sequences used are as follows: P1: 2002; 4: 859–864. 50-CACCCTCTTGCTTATTGGCTTT-30. P2: 50-TGACAGCTTTGGCCCCTAGT-30. 15 Hershko T, Ginsberg D. Up-regulation of Bcl-2 homology 3 (BH3)-only proteins by P3: 50-TGCATCACCTCAACCCTGATAT-30. P3 amplifies the region contain- E2F1 mediates apoptosis. J Biol Chem 2004; 279: 8627–8634. ing E2F1-binding site, À 1180/ À 1173. 16 Zhao Y, Tan J, Zhuang L, Jiang X, Liu ET, Yu Q. Inhibitors of histone deacetylases target the Rb-E2F1 pathway for apoptosis induction through activation of proapoptotic protein Bim. Proc Natl Acad Sci USA 2005; 102: 16090–16095. AO staining 17 Putzer BM, Stiewe T, Crespo F, Esche H. Improved safety through tamoxifen- AO is a lysomotropic probe that preferentially accumulates in the acidic regulated induction of cytotoxic genes delivered by Ad vectors for cancer gene 32 compartments in its charged form (AOH þ ). For AO staining and confocal therapy. Gene Ther 2000; 7: 1317–1325. microscopy, U2OS ER-E2F1 cells were counted and reseeded in a cover- 18 Lin WC, Lin FT, Nevins JR. Selective induction of E2F1 in response to DNA glass slide chamber (Becton Dickinson) following CTSL2 knockdown. Cells damage, mediated by ATM-dependent phosphorylation. Genes Dev 2001; 15: were treated with or without 4-OHT (300 nM) for 24 h before staining with 1833–1844. AO (5 mg/ml) for 15 min. Cells were then analyzed via confocal microscopy 19 Pediconi N, Ianari A, Costanzo A, Belloni L, Gallo R, Cimino L et al. Differential at a excitation wavelength of 488 nm. regulation of E2F1 apoptotic target genes in response to DNA damage. Nat Cell Biol 2003; 5: 552–558. 20 Jiang X, Tan J, Li J, Kivimae S, Yang X, Zhuang L et al. DACT3 is an epigenetic Lysosomal membrane stability assay by AO uptake regulator of Wnt/beta-catenin signaling in colorectal cancer and is a therapeutic U2OS ER-E2F1 cells were treated and stained as in AO staining as described target of histone modifications. Cancer Cell 2008; 13: 529–541. above. The cells were analyzed by flow cytometry by determining the 21 Zhang Y, Yang ND, Zhou F, Shen T, Duan T, Zhou J et al. (-)-Epigallocatechin-3- percentage of ‘pale cells’, or cells which have reduced number of intact Gallate Induces Non-Apoptotic Cell Death in Human Cancer Cells via ROS-Medi- lysosomes by measuring the decrease in red fluorescence following 4-OHT ated lysosomal membrane permeabilization. PLoS One 2012; 7: e46749. treatment. 22 Blomgran R, Zheng L, Stendahl O. Cathepsin-cleaved Bid promotes apoptosis in human neutrophils via oxidative stress-induced lysosomal membrane permeabilization. J Leukoc Biol 2007; 81: 1213–1223. Statistical analysis 23 Bolden JE, Peart MJ, Johnstone RW. Anticancer activities of histone deacetylase Statistical analysis was conducted using two-tailed Student’s t test. inhibitors. Nat Rev Drug Discov 2006; 5: 769–784. Correlation of E2F1 and CTSL2 mRNA expression level in Figure 2e was 24 Santamaria I, Velasco G, Cazorla M, Fueyo A, Campo E, Lopez-Otin C. Cathepsin L2, measured via Pearson’s product moment correlation. A P-value of o0.05 a novel human cysteine proteinase produced by breast and colorectal carcino- was defined as statistically significant. mas. Cancer Res 1998; 58: 1624–1630.

Oncogene (2014) 1249 – 1257 & 2014 Macmillan Publishers Limited E2F1 target CTSL2 modulates drug-induced apoptosis C Hooi Wong et al 1257 25 Sevenich L, Hagemann S, Stoeckle C, Tolosa E, Peters C, Reinheckel T. Expression 29 Wu ZL, Zheng SS, Li ZM, Qiao YY, Aau MY, Yu Q. Polycomb protein EZH2 regulates of human cathepsin L or human cathepsin V in mouse thymus mediates positive E2F1-dependent apoptosis through epigenetically modulating Bim expression. selection of T helper cells in cathepsin L knock-out mice. Biochimie 2010; 92: Cell Death Differ 2010; 17: 801–810. 1674–1680. 30 Johnstone RW. Histone-deacetylase inhibitors: novel drugs for the treatment of 26 Ivanova S, Repnik U, Bojic L, Petelin A, Turk V, Turk B. Lysosomes in apoptosis. cancer. Nat Rev Drug Discov 2002; 1: 287–299. Methods Enzymol 2008; 442: 183–199. 31 Tan J, Yang X, Zhuang L, Jiang X, Chen W, Lee PL et al. Pharmacologic disruption 27 Droga-Mazovec G, Bojic L, Petelin A, Ivanova S, Romih R, Repnik U et al. of polycomb-repressive complex 2-mediated gene repression selectively induces Cysteine cathepsins trigger caspase-dependent cell death through cleavage apoptosis in cancer cells. Genes Dev 2007; 21: 1050–1063. of bid and antiapoptotic Bcl-2 homologues. J Biol Chem 2008; 283: 32 Kreuzaler PA, Staniszewska AD, Li W, Omidvar N, Kedjouar B, Turkson J et al. Stat3 19140–19150. controls lysosomal-mediated cell death in vivo. Nat Cell Biol 2011; 13: 303–309. 28 Jaattela M, Cande C, Kroemer G. Lysosomes and mitochondria in the commitment 33 Wu ZL, Zheng SS, Li ZM, Qiao YY, Aau MY, Yu Q. Polycomb protein EZH2 regulates to apoptosis: a potential role for cathepsin D and AIF. Cell Death Differ 2004; 11: E2F1-dependent apoptosis through epigenetically modulating Bim expression. 135–136. Cell Death Differ 2010; 17: 801–810.

& 2014 Macmillan Publishers Limited Oncogene (2014) 1249 – 1257