DNA damage and eIF4G1 in breast cancer cells reprogram for survival and DNA repair mRNAs

Michelle Baduraa,1,2, Steve Braunsteina,1,3, Jiri Zavadilb, and Robert J. Schneidera,b,c,4

aDepartment of Microbiology, bNYU Cancer Institute, and cDepartment of Radiation Oncology, New York University School of Medicine, New York, NY 10016

Edited by Stephen P. Goff, Columbia University College of Physicians and Surgeons, New York, NY, and approved September 26, 2012 (received for review March 6, 2012)

The cellular response to DNA damage is mediated through multiple metabolic demands. A small number of mRNAs that maintain pathways that regulate and coordinate DNA repair, cell cycle translation during cell stress responses contain in the 5′ un- arrest, and cell death. We show that the DNA damage response translated region (5′ UTR) both a cap and an internal (DDR) induced by ionizing radiation (IR) is coordinated in breast entry site (IRES), which permits eIF4E-independent translation cancer cells by selective mRNA translation mediated by high levels (6). Cellular IRES models of translation initiation generally in- of translation eIF4G1 (eukaryotic initiation factor volve the binding of eIF4G to secondary structures in the IRES 4γ1). Increased expression of eIF4G1, common in breast cancers, (7, 8). In mammalian cells increased levels of eIF4G facilitate was found to selectively increase translation of mRNAs involved in translation of strictly cap (eIF4E)-dependent mRNAs, mRNAs cell survival and the DDR, preventing autophagy and apoptosis containing multiple upstream (u)ORFs or IRES elements, and [Survivin, hypoxia inducible factor 1α (HIF1α), X-linked inhibitor those of low abundance (7, 9, 10). Furthermore, different mRNAs of apoptosis (XIAP)], promoting cell cycle arrest [growth arrest display a wide range of dependence on eIF4E levels and yet lack and DNA damage 45a (GADD45a), protein 53 (p53), ATR- an IRES (11). In yeast, higher expression levels of eIF4G promote translation of mRNAs with longer polyA tails (12) or promote interacting protein (ATRIP), Check point kinase 1 (Chk1)] and DNA direct ribosome–mRNA interactions (13), increasing translation repair [p53 binding protein 1 (53BP1), breast cancer associated pro- of mRNAs with intrinsically higher translation efficiencies. Thus, teins 1, 2 (BRCA1/2), Poly-ADP ribose polymerase (PARP), replication in both yeast and mammalian cells, reduced levels of eIF4G do factor c2–5(Rfc2-5), ataxia telangiectasia mutated 1 (ATM), MRE-11 not strongly impair translation of many mRNAs (9, 12, 13), meiotic recombination protein 11 ( ), and others]. Reduced suggesting that certain mRNAs are selectively increased in their expression of eIF4G1, but not its homolog eIF4G2, greatly sensi- translation with higher levels of eIF4G. tizes cells to DNA damage by IR, induces cell death by both apo- eIF4G1 is the most abundant member of the eIF4G family. ptosis and autophagy, and significantly delays resolution of DNA Preferential translation by the two eIF4G isoforms in flies, nem- damage foci with little reduction of overall protein synthesis. atodes, and mammals has been reported, where a preferential Although some mRNAs selectively translated by higher levels requirement for eIF4G2-like homologs was found in germline of eIF4G1 were found to use internal ribosome entry site (IRES)- cell developmental mRNA translation (14, 15). A poorly studied mediated alternate translation, most do not. The latter group related eIF4G family member, death associated protein 5 fi shows signi cantly reduced dependence on eIF4E for translation, (DAP5)/protein (p)97, lacks the NH2-terminal eIF4E and poly(A) facilitated by an enhanced requirement for eIF4G1. Increased ex- binding protein (PABP) interaction sites (16), and either stim- pression of eIF4G1 therefore promotes specialized translation of ulates (17) or inhibits (16) translation of certain IRES- survival, growth arrest, and DDR mRNAs that are important in containing mRNAs. cell survival and DNA repair following genotoxic DNA damage. eIF4G1 expression is strongly increased in breast cancers (10, 18, 19) and squamous cell lung cancers (20), associated with translational control | eIF metastatic progression and reduced survival. The mechanism of overexpression has not been well-studied. Given that eIF4G1, ells have developed complex, coordinated DNA repair like eIF4E, is also phosphorylated by, and is an effector of, Cresponses to protect and repair their genomes from geno- mTOR activity (9), that mTOR inhibition is associated with in- toxic damage, prevent mitotic entry, and promote survival (1). creased sensitivity to IR-mediated DNA damage (21), and that Paradoxically, although DNA damage induces the transcription eIF4G1 depletion does not strongly down-regulate overall pro- of a number of involved in the DNA damage response tein synthesis (9), we investigated whether the elevated expres-

(DDR), cells typically respond to DNA damage by downregu- sion of eIF4G1 associated with breast cancer progression is BIOCHEMISTRY lating protein synthesis (2, 3). Translational regulation is a poorly responsible for resistance to genotoxic DNA damage responses. explored but fundamental component of the cellular response to DNA damage (2). Down-regulation of protein synthesis after ionizing radiation (IR) involves inhibition of the cap-dependent Author contributions: M.B., S.B., and R.J.S. designed research; M.B. and S.B. performed mRNA translation machinery, controlled by the protein kinase research; M.B., S.B., J.Z., and R.J.S. analyzed data; and M.B. and R.J.S. wrote the paper. mammalian target of rapamycin (mTOR), a downstream effector The authors declare no conflict of interest. of the PI3 kinase–Akt signaling pathway (2, 4). This article is a PNAS Direct Submission. Translation in eukaryotes is mediated by multiple factors pri- marily at the step of initiation on mRNA, through assembly of Data deposition: The microarray data reported in this paper have been deposited in the the eukaryotic initiation factor (eIF)4F. eIF4F is composed of Gene Expression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. 7 GSE41627). eIF4E, which binds the m GTP cap, eIF4G (eukaryotic, initia- 1 tion factor 4γ), which is a scaffold protein upon which M.B and S.B contributed equally to this work. and eIFs assemble, and eIF4A, an ATP-dependent RNA heli- 2Present address: Department of Radiation Oncology, University of California, San case. eIF4E availability is regulated by eF4E binding protein 1 Francisco, CA 94127. (4E-BP1), a competitive eIF4E inhibitor, which is inactivated 3Present address: School of Medicine, Stanford University, Palo Alto, CA 94305. through hyperphosphorylation by mTOR (5). Translation of 4To whom correspondence should be addressed. E-mail: [email protected]. most mRNAs requires eIF4E interaction with the mRNA cap, This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. controlled by eIF4E/4E-BP regulation in response to growth and 1073/pnas.1203853109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1203853109 PNAS | November 13, 2012 | vol. 109 | no. 46 | 18767–18772 Downloaded by guest on October 1, 2021 Results which in eIF4G1-silenced cells, was strongly and persistently eIF4G1 Expression Promotes Resistance to Genotoxic DNA Damage increased even at 24 h. with Little Effect on Overall Protein Synthesis. Reduced eIF4G1 Silencing of eIF4G1 increased apoptosis even in the absence expression only moderately down-regulates overall protein syn- of IR, shown by a 20% increase in baseline Annexin V staining by FACS analysis, a marker of apoptosis (Fig. 2B). Post-IR, annexin thesis rates in mammalian and yeast cells (9, 12, 13). Accordingly, ∼ stable reduced expression of eIF4G1 by inducible shRNA silenc- staining increased by 40% at 48 h and 75% at 72 h. Cleaved ing in MCF10A immortalized/partially transformed breast epi- caspase 3, another marker of apoptosis, was not detected in NS thelial cells (Fig. 1A) or in MDA-MB-231, MCF-7, or HTB20 cells regardless of irradiation, whereas the two major cleavage A forms were increased by eIF4G1 silencing, and further increased breast cancer cells (Fig. S1 ), reduced protein synthesis rates by B only 15–20% irrespective of exposure to a high dose (8 Gy) of by IR (Fig. S3 ). Increased expression of eIF4G1 therefore IR and is, therefore, a common feature. Silencing of eIF4G1 facilitates resistance to both apoptosis and autophagy following resulted in a compensatory increased expression of eIF4G2 IR-mediated genotoxic DNA damage. regardless of IR treatment (Fig. 1B). Cosilencing of eIF4G2 and Prolonged DNA-damage is associated with terminal withdrawal eIF4G1 was shown previously not to further reduce protein from the cell cycle causing a senescence-like phenotype (SLP) (23) and cell death by both apoptosis and SLP [SLP demonstrated by synthesis (9). As described previously (2), exposure to IR does β β not result in eIF2α Ser51 phosphorylation and eIF2 inhibition acidic -galactosidase ( -gal) (23)]. Both were clearly evident in irradiated eIF4G1-silenced cells and not in controls (Fig. 2C). (Fig. S2), unlike UVB irradiation (3). β Cells reduced in eIF4G1 expression by stable shRNA silencing eIF4G1 silencing alone increased -gal activity to include 30% of cells, which was strongly increased in intensity and extent when demonstrated up to a 50-fold decline in viability with increasing > ARF IR dose compared with nonsilenced (NS) control cells (Fig. 1 C combined with IR ( 90% of cells). Increased p14 expression, D another marker of SLP (23), was also strongly enhanced (Fig. and ). Surprisingly, silencing of eIF4E or eIF4G2 only partially C increased sensitivity to DNA damage by IR compared with eIF4G1. S3 ). We, therefore, investigated the association of the DDR Silencing of overexpressed eIF4G1 in other breast cancer cell with eIF4G1 levels, as eIF4G1 links autophagy and apoptosis. lines also conferred significant sensitivity to IR-mediated DNA damage (Fig. S1C). We, therefore, pursued the mechanism for Reduced Expression of eIF4G1 Delays and Prolongs Assembly of DDR eIF4G1 promotion of cell survival. Complex at DNA Double-Strand Breaks and Impairs G2/M Cell Cycle Arrest. The abrogated expression of components of the DDR Increased Expression of eIF4G1 Prevents Autophagy, Apoptosis, and is reflected in overall delayed and mitigated signaling throughout Senescence Mediated by IR DNA Damage. Genotoxic DNA damage the DDR response. Formation of IR-induced foci (IRIF) in can promote cell death by both autophagy and apoptosis (22). eIF4G1-silenced cells at DNA double-strand break (DSB) sites Lysates of irradiated cells were first examined for formation of [marked by DNA repair proteins p53 binding protein 1 (53BP1) LC3-II, a marker of autophagy that decorates autophagosomes. and phosphorylated histone 2 AX protein (γH2AX)] was aber- As early as 6 h post-IR, eIF4G1-silenced cells undergo autophagy, rantly delayed and prolonged compared with NS controls (Fig. 3A), as evidenced by increased LC3-II (Fig. 2A), whereas NS cells with little expression of 53BP1 and reduced expression and delayed showed only slight autophagy at 24 h. LC3 fused to GFP also assembly of γH2AX. This is consistent with a failure to properly decorated autophagosomes with strong punctate staining only in assemble and/or resolve DSB repair complexes (24). Phos- IR-treated cells, consistent with induction of autophagy (Fig. S3A), phorylation of H2AX (γH2AX) generally occurs immediately following DNA damage (25). In eIF4G1-silenced and irradiated cells, H2AX phosphorylation was delayed and sustained without resolution, even at 24 h (Fig. 3 A and B), indicative of an inability A NS 4G1 shRNA B to resolve IRIFs. An IR-induced increase in 53BP1 levels was eIF4G1 NS 4G1 shRNA also impaired in eIF4G1-silenced cells. With eIF4GI silencing, 120 fi β-tubulin - + - + 8 Gy and consistent with de cient DSB repair, there was a sustained 100 increase in p21 protein levels, reduced signaling (activating 80 eIF4G1 60 eIF4G2 phosphorylation) of Check point kinase 1 (Chk1), Chk2, and breast cancer associated protein 1 (BRCA1) [mediated by ataxia 40 β-tubulin 20 telangiectasia mutated gene 1 (ATM), ataxia telangiectasia Rad 0 related kinase (ATR), and Chk2], and reduced BRCA1 and Chk1 synthesis activity Percent protein NS 4G NS 4G NS 4G NS 4G shRNA 0 h 6 h 12 h 24 h post-IR protein levels (Fig. 3B). There was also impaired G2/M cell cycle arrest (Fig. 3C). Collectively, these data indicate that there are C MCF10A Cell Viability D multiple disruptions within the DDR pathway linked to reduced 1 shRNA NS shRNA expression of eIF4G1 and enhanced radiosensitivity. eIF4G1 shRNA 0.1 eIF4E shRNA eIF4G1 Promotes Selectively Increased Translation of DDR and Survival eIF4G2 shRNA NS eIF4G1 eIF4G2 eIF4E 0.01 eIF4G1 mRNAs in Response to IR-Mediated DNA Damage. Studies previously eIF4G2 linked IR resistance to maintenance of protein synthesis (2). We, 0.001 therefore, determined whether the greatly increased radiosensi- eIF4E tivity and depletion of the DDR response in eIF4G1-silenced cells 0.0001 β-tubulin is attributable to selectively reduced mRNA translation. Trans- 0 2 4 6 8 Surviving fraction (log10) lation signature studies were carried out in NS and eIF4G1-si- Dose (Gy) lenced cells, with or without IR treatment, using gene expression Fig. 1. Effect of eIF4G1, eIF4G2 and eIF4E silencing on overall protein array analysis of mRNAs unassociated with ribosomes (non- synthesis and cell survival in IR-treated cells. (A) NS (non-silencing) and translated), associated with light polyribosomes (one to three eIF4G1-silenced MCF10A cells were treated with 8 Gy IR or mock-treated. ribosomes, poorly translated), or well-translated and associated 35 “ ” Protein synthesis rates determined by [ S]methionine incorporation, nor- with heavy polyribosomes (four or more ribosomes) (Table 1). malized to NS-shRNA nonirradiated control (0 h). SEMs are shown. Immu- Reduction of eIF4G1 levels by RNA silencing produced a ∼20% noblot is eIF4G1 in silenced and control cells. (B) Equal amounts of protein at reduction in the total level of polyribosomes (Fig. 4A), consistent 24 h post-IR were resolved by SDS/PAGE and immunoblotted with the in- with the reduction in overall protein synthesis. dicated antibodies. (C) Clonogenic survival assays performed on MCF10A In response to IR, eIF4G1-dependent mRNA translation was cells following IR, with stable shRNA to eIF4E, eIF4G1, eIF4G2, or an NS found to be reprogrammed. As shown by shRNA silencing of random sequence. Cells were plated overnight and irradiated, and colonies eIF4G1, heavy polysomes were selectively depleted largely of were scored after 2 wk (n = 3). (D) Immunoblot of equal protein amounts of mRNAs comprising (GO) categories for DNA lysates from C. damage response and repair (36% of mRNAs), cell cycle

18768 | www.pnas.org/cgi/doi/10.1073/pnas.1203853109 Badura et al. Downloaded by guest on October 1, 2021 A B 80 0 6 12 24 h post-IR NS shRNA 70 eIF4G1 shRNA NS 4G NS 4G NS 4G NS 4G shRNA 60 eIF4G1 50 β-tubulin 40 30 I LC3 II 20 Fig. 2. Depletion of eIF4G1 sensitizes cells to IR- mediated autophagy, apoptosis, and senescence. (A)

% Annexin V-PE % Annexin V-PE positive cells 10 0 Equal amounts of protein lysates from MCF10A cells 0 36 48 72 h post-IR stably silenced for eIF4G1 or NS, treated with 8 Gy C 0 Gy 4 Gy IR, resolved by SDS/PAGE, and immunoblotted with antibodies as shown. LC3-II is a marker for induction NS sh-eIF4G1 NS sh-eIF4G1 of autophagy. (B) FACS analysis conducted on 8 Gy– irradiated NS or eIF4G1-silenced MCF10A cells stained with PE-Annexin V, a marker of apoptosis. (C)Cells stably transfected with NS or eIF4G1 shRNA lentivi- ruses were subjected to 0 or 4 Gy irradiation, and β-gal assays for cellular senescence were performed. Representative phase-contrast photomicrographs of × fi β cells of triplicate studies shown at 100 magni ca- acidic -galactosidase staining tion. Arrows mark β-gal staining. Assays were n = 3.

inhibition/growth arrest and survival functions (51% of mRNAs), (GADD45a) (sixfold). GADD45a functions in cell cycle arrest and bioenergetics (5% of mRNAs) (Table 1 and Table S1), when and inhibition of DNA synthesis associated with DNA repair via normalized to wild-type control mRNA levels with a ≥twofold regulatory interactions with cell division control protein 2 (cdc2), enrichment requirement. Thus, ∼87% of mRNAs strongly de- p21, and proliferating cell nuclear antigen (PCNA) (27) and is pendent on increased expression of eIF4G1 in irradiated cells transcriptionally increased by p53 activation following DNA correspond to the central functions of survival, DNA damage and damage. A compelling list of DNA repair and response factor repair, and cell cycle control. Interestingly, there was a small mRNAs were also increased in heavy polysomes in an IR and group of mRNAs up-regulated by reduction of eIF4G1 levels eIF4G1-dependent manner, particularly the kinases ATM (five- with IR, corresponding primarily to functions involved in in- fold) and Chk1 (twofold), and other response factors such as creased cell adhesion, cell–matrix interactions and cell–cell meiotic recombination protein 11 (MRE-11), radiation induced interactions (70% of mRNAs), as well as immune responsiveness protein (Rad)50/51/54, Rfc (replication factor c) transcription (11%). Of the survival factor mRNAs, those strongly decreased elongation proteins 2/4/5, PARP, 53BP1, and BRCA1/2. Consis- in heavy polysomes with eIF4G1 silencing (Table 1), included tent with the multifactor nature of DDR and survival pathways, Survivin (eightfold), a key antiapoptotic protein, and hypoxia in- enforced overexpression of survivin or HIF1α in eIF4G1-si- ducible factor (HIF)1α(threefold), which activates the hypoxia lenced cells by translation of transfected cDNA from the hepa- response and promotes cell survival from IR-mediated DNA titis C virus (HCV) IRES conferred only a slight survival damage (26). Of growth arrest factors, two were prominent: p53 advantage to cells at lower doses of IR and none at higher levels (threefold) and growth arrest and DNA damage protein 45a (Fig. S4A).

A NS shRNA eIF4G1 shRNA B NS shRNA eIF4G1 shRNA min hours min hours actin γH2AX 53BP1 actin γH2AX 53BP1 0 10 30 1 8 24 0 10 30 1 8 24 0 min eIF4G1 53BP1 MRE11 10 min CHK2 CHK2 T68P 30 min CHK1

CHK1 S296P BIOCHEMISTRY p21 H2AX S139P 60 min BCRA1 BCRA1 S1524P Fig. 3. DNA-damage response signaling and G2/M 2 h β-tubulin arrest are impaired in eIF4G1-silenced cells. (A)Ki- netics of IRIF in MCF10A cells stably expressing NS or eIF4G1 shRNAs assessed by confocal microscopy. eIF4G1 reduction on cell cycle 8 h C Actin: direct immunofluorescence with FITC-phalloidin; Control 8 Gy 24 h γH2AX (H2AX-S139P): primary monoclonal anti- G2/M body and TRITC anti-mouse secondary antibody; 24 h 53BP1: rabbit primary antibody coupled to Alexa Fluor 488 anti-rabbit secondary antibody (n = 3). (B) Equal sh-NS protein amounts of total cell lysates analyzed by SDS/ PAGE and immunoblot with antibodies for protein and phospho-specificproteinsasshown.(C)Cellcycle G2/M analysis performed by propidium iodide staining of RNase A treated MCF10A NS and eIF4G1-silenced cells sh-eIF4GI following IR. DNA content (x axis) was determined by FACS analysis.

Badura et al. PNAS | November 13, 2012 | vol. 109 | no. 46 | 18769 Downloaded by guest on October 1, 2021 A 80S C D

NS 4E 4G1 shRNA 0 Gy 8 Gy IR eIF4E NS 4G1 NS 4G1 shRNA NS shRNA 35 eIF4G1 53BP1 S-met eIF4G1 shRNA 53BP1 35 60S 2 XIAP S-met GADD45a 40S 3 4 35

Abs 254 nm 5 6 7 GADD45a S-met Survivin XIAP light polys heavy polys actin 1-3 ribosomes ≥4 ribosomes

B MCF10A cells E NS 4G1 shRNA - + - + 8 Gy 1.2 eIF4G1 1.0 53BP1 Survivin 0.8 GADD45a 0.6 p53 p53 S15-P 0.4 XIAP 0.2

ATRIP Relative protein synthesis 0 PARP NS 4E 4G1 NS 4E 4G1 NS 4E 4G1 NS 4E 4G1 NS 4E 4G1 NS 4E 4G1 shRNA α-tubulin EMCV GADD45a Survivin 53BP1 XIAP actin

Fig. 4. eIF4E requirement of selectively translated DDR and survival mRNAs with eIF4G1 overexpression. (A) Cells silenced with NS or eIF4G1 shRNAs sub- jected to 8 Gy IR and fractionated by sucrose density gradient sedimentation into unbound mRNA, light (1–3) and heavy (≥4) ribosomes. (B) Immunoblot of NS and eIF4G1-silenced MCF10A cells at 12 h following 8 Gy IR from equal protein amounts of total cell lysates. (C) Relative rate of protein synthesis assessed by immunoprecipitation (IP) of de novo proteins shown following 2 h metabolic labeling with [35S]methionine with actinomycin D (20 μM) and MG132 (10 μM) at 2 h after 8 Gy IR. (D) Immunoblot of NS, eIF4E silenced or eIF4G1-silenced MCF10A cells following 8 Gy IR at 24 h, from equal protein amounts of total cell lysates. (E) Quantification of immunoblot studies shown in B. EMCV results derived from Renilla luciferase reporter studies (n = 3).

To verify these results, we performed quantitative (q)RT-PCR a small number of mRNAs that are selectively translated by analysis of mRNAs in polyribosomes, coupled with immunoblot pathologically high levels of eIF4G1 contain IRES elements. studies. Certain mRNAs such as 53BP1, HIF1α, BRCA-1, and An IRES is only one means for conferring selective mRNA GADD45a are strongly transcriptionally induced by DNA dam- translation under conditions of limiting eIF4E availability, as age, whereas Survivin and X-linked inhibitor of apoptosis (XIAP) found during stress conditions. Some mRNAs have been shown are increased by cellular transformation regardless of DNA to require only low levels of eIF4E for their translation (30) and damage (Fig. S4B). All mRNAs were strongly reduced in heavy yet do not use an IRES or internal initiation (11). Mechanistically, polyribosomes with eIF4G1 silencing (normalized to total mRNA these mRNAs share a requirement for higher levels of eIF4G1 levels), consistent with their reduced protein levels, as shown for but with little requirement for eIF4E (11, 31, 32). We, therefore, PARP, p53, XIAP, GADD45a, 53BP1, and ATR interacting pro- reduced the levels of eIF4E or eIF4G1 by shRNA silencing in tein (ATRIP) (Fig. 4B). As shown for 53BP1, XIAP and GADD45a MDA-MB-231 breast cancer cells in which eIF4G1 is strongly by metabolic labeling with [35S]methionine, the reduced abundance expressed and compared the expression of survival and DDR of protein is a result of diminished de novo protein synthesis as- proteins. Silencing of eIF4G1, as shown earlier, strongly reduced sociated with decreased eIF4G1 expression (Fig. 4C). the expression of 53BP1, GADD45a, survivin and XIAP proteins, regardless of the presence of IRES elements (Fig. 4 D and E), eIF4G1 and eIF4E Dependence in Translation of Survival and DDR which remained impaired despite IR treatment and mRNA in- Pathway mRNAs. Reduction of cap-dependent mRNA translation duction (Fig. S6). IRES-dependent mRNAs EMCV and GADD45a by 4E-BP1 sequestration of eIF4E, as facilitated by cellular stress were fully independent of eIF4E but strongly (fivefold) reduced (19), can promote IRES-mediated translation (6, 10, 18). Several by lowering eIF4G1 levels (Fig. 4 D and E). β-Actin mRNA has DDR and survival mRNAs have suspected IRES elements, in- a strong dependence on eIF4E-cap interaction for its translation cluding GADD45a and HIF1α, and their translation may be in- (11) and was strongly (>20-fold) reduced with eIF4E silencing. strumental in cell survival following stress (2, 6, 28). We, therefore, In contrast, 53BP1 and XIAP showed only a slight (∼2-fold) used bicistronic luciferase reporter mRNAs to test IRES activity. reduction with eIF4E silencing but ∼10 fold reduction with eIF4G1 Silencing eIF4G1 blocked IRES activity 24 h following 8 Gy IR silencing. Survivin showed a greater dependence on eIF4E but for the encephalomyocarditis virus (EMCV) (29), HIF1α, and still less than for eIF4G1, consistent with its reported stimulation GADD45a 5′ UTRs (Fig. S5). However, none of the other by elevated levels of eIF4E (33). These results are consistent mRNAs selectively translated by high levels of eIF4G1 following with a common strong requirement for high levels of eIF4G1 but IR DNA damage possessed IRES activity, as shown for XIAP, a reduced requirement for eIF4E in the IR-induced selective Survivin, and 53BP1 mRNA 5′ UTRs (Fig. S5). Thus, only translation of survival and DDR pathway mRNAs.

18770 | www.pnas.org/cgi/doi/10.1073/pnas.1203853109 Badura et al. Downloaded by guest on October 1, 2021 Table 1. mRNAs displaying strong dependence on elevated (38). Our studies are consistent with reduced p53 levels in in- levels of eIF4G1 for increased abundance in heavy (≥four) duction of both autophagy and apoptosis, in that p53 translation polyribosomes in response to IR requires higher levels of eIF4G1 following IR DNA damage. Moreover, reduced activity of mTOR by IR, a master regulator of Fold-increased heavy autophagy, can also induce autophagy (9, 39). GO group and gene polyribosome abundance* It was reasonable to assume that IRES-mediated translation DNA damage and repair would serve as the major form of specialized translation initia- tion during DNA damage mediated by IR. Few of the DDR and Mre11 3× survival mRNAs induced by IR-mediated DNA damage appear Rad50 3.5× XIAP × to use an IRES, including , reported previously to contain Rad51 3 an IRES element (40). However, XIAP shares a common feature Rad52 0.5× with all of the DDR/survival mRNAs that are selectively in- Rad17 2× creased in translation following IR-mediated DNA damage: it Rad54 4× has a relaxed requirement for eIF4E. Rfc2 3× eIF4G1 mediates a transcriptional–translational feed-forward Rfc3 1.3× loop for HIF1α and survivin expression that promotes cell survival, Rfc4 3× in that Survivin transcription is stimulated by HIF1α,themRNA Rfc5 3× of which shows strong eIF4GI-dependent IRES activity. Higher ATRIP 2× levels of eIF4G1 also establish a feed-forward pathway for in- BRCA1 2.5× creased cell survival through cell cycle arrest following DNA BRCA2 2.5× damage. GADD45a translation is strongly eIF4G1-dependent. BRCA1 internal protein 3.5× Activation of GADD45a promotes JNK-mediated activating phosphorylation of p53 at S15, stimulating the DDR pathway, PARP 3× × and simultaneously promoting strong G2/M cell cycle arrest (41); 53BP1 3.5 53BP1 also promotes phosphorylation and activation of p53, in × NBS1 1.5 addition to driving non-homologous end joining (NHEJ) and γH2AX 2× inhibiting homologous DNA recombination and repair (42). HUS1 1.5× DNA damage mediated by both UVB (3) and high-energy Effector kinases (X-ray) IR (this study) both reprogram the protein synthesis ma- Chk1 2× chinery to promote selective mRNA translation but of largely ATM 5× nonoverlapping mRNAs. UVB radiation is of lower energy and Growth arrest factors introduces primarily pyrimidine dimers, stimulating nucleotide GADD45a 6× excision repair (43). UVB selects for mRNAs with 5′ UTRs p53 3× containing uORFs, consistent with partial inactivation of eIF2 Survival factors and transiently delayed initiation further downstream (3). In † XIAP 1.5× contrast, high-energy X irradiation is ionizing, producing free radicals and reactive oxygen species that lead to single strand and Survivin 8× α† × double-strand DNA breaks (44). IR primarily inhibits mTOR HIF1 3 and down-regulates eIF4E availability (2). *Results show the average increase in mRNA abundance in heavy polyribo- The mechanisms for both stringent and relaxed requirements somes (≥four ribosomes) in nonsilencing vector-transformed cells at 12 h for eIF4E or eIF4G1 in mRNA translation initiation are still after 8 Gy IR compared with similarly treated cells silenced for eIF4G1. poorly understood, as are mechanisms that promote initiation by †Not detectable on chip; determined by qRT-PCR. NBS1, Nibrin; HUS1, hydroxy cellular IRES elements. eIF4G1 clearly has specific functions that urea sensitive protein 1. do not overlap that of eIF4G2, which remain to be understood mechanistically. With that said, eIF4G1 and eIF4G2 clearly trans- late many, if not most, of the same mRNAs, but this does not Examination of the non-IRES mRNAs for common compu- extend to the efficient translation of survival and DDR mRNAs tationally predicted features was not instructive. The average following IR-mediated DNA damage, consistent with results mammalian mRNA 5′ UTR is 100–200 nt long, ∼60% GC, with obtained in yeast with eIF4G1 silencing (12, 45). Our findings a composite stability of <−40 kcal/mol, with fewer than 10% of suggest an alternate means of translation initiation that might mRNAs containing a uORF (34). Although about half of the involve mRNA-specific recruitment of eIF4G1 that is associated eIF4G1-dependent mRNAs had extremely long 5′ UTRs with with eIF4E but possibly involves recycling of eIF4G1 on the extensive predicted stability, such as Chk1, RAD50, ATM, and mRNA to deliver eIF3, 40S ribosome subunits, and associated BRCA1, the others were within the mean range for all predicted factors and, therefore, is a requirement for the overexpression of

features (Table S2). Most of the mRNAs were predicted to con- eIF4G1 (Fig. S7). BIOCHEMISTRY tain one or more hairpin structures of somewhat higher than normal GC content (70 to >90%) that can impair translation Materials and Methods when located near the cap (35), which constituted a trend but not Irradiation. Cells were irradiated with a Varian Linac 2300 linear accelerator at a statistically significant correlation (P < 0.2) compared with room temperature for doses as shown. Mock treated cells were handled randomly selected mRNAs that did not show increased de- identically except for irradiation. pendence on high levels of eIF4G1. Retroviral and Lentivirus Expression Studies. Cloning of expression vectors into Discussion the pBABE and pLKO.1 vectors, production of virus, and transformation of The increased expression of eIF4G1, which occurs in breast and cells with vectors were described previously (46). Interfering shRNAs were some other human cancers, is a critical link between the response delivered by transduction of cells with lentivirus shRNA expression vectors. to DNA damage and expression of DNA repair and cell survival Double-stranded shRNAs for cloning into lentivirus vectors were directed to genes, coordinated at the level of translational control. The po- either the 5′ or 3′ UTRs of mRNAs targeted for gene silencing. tential role of eIF4G1 phosphorylation and regulation by MAP kinase interacting kinase 2 (Mnk2) (36) remains to be studied. Clonogenic Survival Assays. Cells were seeded in triplicate into 10-cm plates Although autophagy is primarily a cytoprotective process and at 102 to 105 cells/plate according to the test condition and different cell an inhibitor of apoptosis (37), in times of cellular stress such as lines. For IR experiments, a single dose of γ irradiation was applied once IR exposure, it can contribute to cell death (22). Dual autophagy cells were attached (24 h). Cells were cultured up to 14 d. Colonies were and apoptosis likely involve reduced expression/activation of p53 fixed in 70% (vol/vol) methanol and stained with crystal violet. All colonies

Badura et al. PNAS | November 13, 2012 | vol. 109 | no. 46 | 18771 Downloaded by guest on October 1, 2021 of 50 cells or greater were counted in quantitative assays. The survival (National Institute for Basic Biology, Okazaki, Japan). Other 5′-UTR test con- fraction (SF) was estimated according to the formula: SF = number of structs were prepared using the pRF vector. Details are available upon request. colonies formed in test condition / (number of cells seeded × plating effi- For transiently transfected constructs, reporter expression was assayed 24 and ciency of control group). 72 h posttransfection. Luciferase assays were performed using the Dual Luciferase Reporter Assay System (Promega). SA–β-Gal Activity. Cells were stained for β-gal activity using the Senescence β-Galactosidase Staining Kit (Cell Signaling). Briefly, cells were seeded in six- Statistical Analysis. Statistical analyses used the two-tailed Student’s t test, well plates containing coverslips. After the appropriate exposure, the cells with P < 0.05 taken as significance. were washed twice with PBS, fixed with 2% formaldehyde and 0.2% glu- taraldehyde in PBS, and washed twice in PBS. Cells were stained overnight in Gene Expression Data. Gene expression microarray data have been deposited X-gal staining solution [1 mg/mL X-gal, 40 mM citric acid/sodium phosphate in minimum information about a microarray experiment (MIAME)-compliant (pH 6.0), 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 150 mM format online in the Gene Expression Omnibus (GEO) database (www.ncbi. NaCl, 2 mM MgCl2]. Staining was visualized and digitally photographed on an nlm.nih.gov/geo). inverted Zeiss phase-contrast microscope. Additional Methods. Additional information can be found in SI Materials Reporter DNA Constructs and Assays. The pRemcvF reporter was generated by and Methods. insertion of the EMCV IRES from pBMN-IRES-EGFP (G. Nolan, Stanford Uni- versity, Stanford, CA) into the pRF plasmid (EcoRI-NcoI). The XIAP 5′-UTR β ACKNOWLEDGMENTS. We thank T. Venuto for completing certain studies. reporter construct, p gal/XIAP/CAT, was provided by M. Holcik (University of This work was supported by grants from the Breast Cancer Research Founda- ′ Ottawa, Canada) (47). The GADD45a 5 UTR reporter construct, pRLG45aFL tion, the Avon Foundation for Women, the Manhasset Women’s Coalition (5′UTR200-318), was provided by F. Chen (West Virginia University, Mor- Against Cancer, and the Department of Defense Breast Cancer Research Pro- gantown, VA) (28). Plasmid pEGFP-LC3 was provided by T. Yoshimori gram (to R.J.S.).

1. Hakem R (2008) DNA-damage repair; the good, the bad, and the ugly. EMBO J 27(4): 24. Lavin MF, et al. (2005) ATM signaling and genomic stability in response to DNA 589–605. damage. Mutat Res 569(1-2):123–132. 2. Braunstein S, Badura ML, Xi Q, Formenti SC, Schneider RJ (2009) Regulation of protein 25. Fillingham J, Keogh MC, Krogan NJ (2006) GammaH2AX and its role in DNA double- synthesis by ionizing radiation. Mol Cell Biol 29(21):5645–5656. strand break repair. Biochem Cell Biol 84(4):568–577. 3. Powley IR, et al. (2009) Translational reprogramming following UVB irradiation is 26. Moeller BJ, Dewhirst MW (2006) HIF-1 and tumour radiosensitivity. Br J Cancer 95(1):1–5. mediated by DNA-PKcs and allows selective recruitment to the polysomes of mRNAs 27. Zhan Q (2005) Gadd45a, a p53- and BRCA1-regulated stress protein, in cellular re- encoding DNA repair enzymes. Genes Dev 23(10):1207–1220. sponse to DNA damage. Mutat Res 569(1-2):133–143. 4. Budanov AV, Karin M (2008) p53 target genes sestrin1 and sestrin2 connect genotoxic 28. Chang Q, et al. (2007) Incorporation of an internal ribosome entry site- stress and mTOR signaling. Cell 134(3):451–460. dependent mechanism in arsenic-induced GADD45 alpha expression. Cancer Res 5. Hay N, Sonenberg N (2004) Upstream and downstream of mTOR. Genes Dev 18(16): 67(13):6146–6154. 1926–1945. 29. Kolupaeva VG, Pestova TV, Hellen CU, Shatsky IN (1998) Translation eukaryotic ini- 6. Holcik M, Sonenberg N, Korneluk RG (2000) Internal ribosome initiation of translation tiation factor 4G recognizes a specific structural element within the internal ribosome and the control of cell death. Trends Genet 16(10):469–473. entry site of encephalomyocarditis virus RNA. J Biol Chem 273(29):18599–18604. 7. de Breyne S, Yu Y, Unbehaun A, Pestova TV, Hellen CU (2009) Direct functional in- 30. Svitkin YV, et al. (2005) initiation factor 4E availability controls teraction of initiation factor eIF4G with type 1 internal ribosomal entry sites. Proc Natl the switch between cap-dependent and internal ribosomal entry site-mediated Acad Sci USA 106(23):9197–9202. translation. Mol Cell Biol 25(23):10556–10565. 8. Hundsdoerfer P, Thoma C, Hentze MW (2005) Eukaryotic translation initiation factor 31. Ali IK, McKendrick L, Morley SJ, Jackson RJ (2001) Truncated initiation factor eIF4G 4GI and p97 promote cellular internal ribosome entry sequence-driven translation. lacking an eIF4E binding site can support capped mRNA translation. EMBO J 20(15): Proc Natl Acad Sci USA 102(38):13421–13426. 4233–4242. 9. Ramírez-Valle F, Braunstein S, Zavadil J, Formenti SC, Schneider RJ (2008) eIF4GI links 32. De Gregorio E, Preiss T, Hentze MW (1999) Translation driven by an eIF4G core do- nutrient sensing by mTOR to cell proliferation and inhibition of autophagy. J Cell Biol main in vivo. EMBO J 18(17):4865–4874. 181(2):293–307. 33. Graff JR, et al. (2007) Therapeutic suppression of translation initiation factor eIF4E 10. Braunstein S, et al. (2007) A hypoxia-controlled cap-dependent to cap-independent expression reduces tumor growth without toxicity. J Clin Invest 117(9):2638–2648. translation switch in breast cancer. Mol Cell 28(3):501–512. 34. Pesole G, et al. (2001) Structural and functional features of eukaryotic mRNA un- 11. Shatsky IN, Dmitriev SE, Terenin IM, Andreev DE (2010) Cap- and IRES-independent translated regions. Gene 276(1-2):73–81. scanning mechanism of translation initiation as an alternative to the concept of 35. Babendure JR, Babendure JL, Ding JH, Tsien RY (2006) Control of mammalian trans- cellular IRESs. Mol Cells 30(4):285–293. lation by mRNA structure near caps. RNA 12(5):851–861. 12. Clarkson BK, Gilbert WV, Doudna JA (2010) Functional overlap between eIF4G iso- 36. Hu SI, et al. (2012) MNK2 inhibits eIF4G activation through a pathway involving forms in Saccharomyces cerevisiae. PLoS ONE 5(2):e9114. serine-arginine-rich protein kinase in skeletal muscle. Sci Signal 5(211):ra14. 13. Park EH, Zhang F, Warringer J, Sunnerhagen P, Hinnebusch AG (2011) Depletion of 37. Klionsky DJ, Emr SD (2000) Autophagy as a regulated pathway of cellular degrada- eIF4G from yeast cells narrows the range of translational efficiencies genome-wide. tion. Science 290(5497):1717–1721. BMC Genomics 12:68. 38. Lee SB, Tong SY, Kim JJ, Um SJ, Park JS (2007) Caspase-independent autophagic cy- 14. Franklin-Dumont TM, Chatterjee C, Wasserman SA, Dinardo S (2007) A novel eIF4G totoxicity in etoposide-treated CaSki cervical carcinoma cells. DNA Cell Biol 26(10): homolog, Off-schedule, couples translational control to meiosis and differentiation in 713–720. Drosophila spermatocytes. Development 134(15):2851–2861. 39. Cao C, et al. (2006) Inhibition of mammalian target of rapamycin or apoptotic 15. Baker CC, Fuller MT (2007) Translational control of meiotic cell cycle progression and pathway induces autophagy and radiosensitizes PTEN null prostate cancer cells. spermatid differentiation in male germ cells by a novel eIF4G homolog. Development Cancer Res 66(20):10040–10047. 134(15):2863–2869. 40. Holcik M, Yeh C, Korneluk RG, Chow T (2000) Translational upregulation of X-linked 16. Imataka H, Olsen HS, Sonenberg N (1997) A new translational regulator with ho- inhibitor of apoptosis (XIAP) increases resistance to radiation induced cell death. mology to eukaryotic translation initiation factor 4G. EMBO J 16(4):817–825. Oncogene 19(36):4174–4177. 17. Marash L, et al. (2008) DAP5 promotes cap-independent translation of Bcl-2 and CDK1 41. Hollander MC, Fornace AJ, Jr. (2002) Genomic instability, centrosome amplification, to facilitate cell survival during mitosis. Mol Cell 30(4):447–459. cell cycle checkpoints and Gadd45a. Oncogene 21(40):6228–6233. 18. Silvera D, et al. (2009) Inflammatory breast cancer pathogenesis mediated by trans- 42. Wang B, Matsuoka S, Carpenter PB, Elledge SJ (2002) 53BP1, a mediator of the DNA lation initiation factor eIF4G overexpression and unorthodox protein synthesis. Nat damage checkpoint. Science 298(5597):1435–1438. Cell Biol 11(7):903–910. 43. Shuck SC, Short EA, Turchi JJ (2008) Eukaryotic nucleotide excision repair: From un- 19. Silvera D, Formenti SC, Schneider RJ (2010) Translational control in cancer. Nat Rev derstanding mechanisms to influencing biology. Cell Res 18(1):64–72. Cancer 10(4):254–266. 44. Cann KL, Hicks GG (2007) Regulation of the cellular DNA double-strand break re- 20. Comtesse N, et al. (2007) Frequent overexpression of the genes FXR1, CLAPM1 and sponse. Biochem Cell Biol 85(6):663–674. EIF4G located on amplicon 3q26-27 in squamous cell carcinoma of the lung. Int J 45. Gilbert WV, Zhou K, Butler TK, Doudna JA (2007) Cap-independent translation Cancer 120(12):2538–2544. is required for starvation-induced differentiation in yeast. Science 317(5842): 21. Albert JM, Kim KW, Cao C, Lu B (2006) Targeting the Akt/mammalian target of rapa- 1224–1227. mycin pathway for radiosensitization of breast cancer. Mol Cancer Ther 5(5):1183–1189. 46. Connolly E, Braunstein S, Formenti S, Schneider RJ (2006) Hypoxia inhibits protein 22. Eisenberg-Lerner A, Kimchi A (2009) The paradox of autophagy and its implication in synthesis through a 4E-BP1 and 2 kinase pathway controlled by cancer etiology and therapy. Apoptosis 14(4):376–391. mTOR and uncoupled in breast cancer cells. Mol Cell Biol 26(10):3955–3965. 23. Chen JH, Hales CN, Ozanne SE (2007) DNA damage, cellular senescence and organ- 47. Holcik M, Lefebvre C, Yeh C, Chow T, Korneluk RG (1999) A new internal-ribosome- ismal ageing: Causal or correlative? Nucleic Acids Res 35(22):7417–7428. entry-site motif potentiates XIAP-mediated cytoprotection. Nat Cell Biol 1(3):190–192.

18772 | www.pnas.org/cgi/doi/10.1073/pnas.1203853109 Badura et al. Downloaded by guest on October 1, 2021