LRF maintains genome integrity by regulating the non-homologous end joining pathway of DNA repair

Leukemia/lymphoma-related factor (LRF) is a POZ/BTB and Krüppel (POK) transcriptional repressor characterized by context-dependent key roles in cell fate decision and tumorigenesis. Here we demonstrate an unexpected transcription-independent function for LRF in the classical non-homologous end joining (cNHEJ) pathway of double-strand break (DSB) repair. We find that LRF loss in cell lines and mouse tissues results in defective cNHEJ, genomic instability and hypersensitivity to ionizing radiation. Mechanistically, we show that LRF binds and stabilizes DNA-PKcs on DSBs, in turn favouring DNA-PK activity. Importantly, LRF loss restores ionizing radiation sensitivity to p53 null cells, making LRF an attractive biomarker to direct p53-null LRF-deficient tumours towards therapeutic treatments based on genotoxic agents or PARP inhibitors following a synthetic lethal strategy.

T he ability to maintain a stable genome is crucial for normal cell function, and genomic instability may underlie many developmental disorders and human diseases, including cancer 1 . DNA double-strand breaks (DSBs) are perhaps the most deleterious threat to genomic stability. Cells use two main pathways to repair DSBs: non-homologous end joining (NHEJ) and homologous recombination (HR) 2 . These two pathways are largely distinct from one another. HR is particularly effective in S and G2 phases when the break is repaired using genetic information from a sister chromatid, whereas NHEJ can be effective at all times in the cell cycle, yet it is often error prone 3 . The DNA-dependent protein kinase (DNA-PK) complex, including catalytic subunit DNA-PKcs and DNA-binding subunits Ku70/80, is a key component of the classical non-homologous end joining (cNHEJ) apparatus. The physical interaction between DNA-bound Ku (Ku70/Ku80), in particular the C-terminal tail of Ku80, and DNA-PKcs at sites of DNA breaks defines a functional DNA-PK complex that concomitantly bridges the broken DNA ends and activates the DNA repair machinery through the phosphorylation of specific downstream targets 4,5 .
LRF (formerly known as POKEMON 6 , FBI-1 (ref. 7) or OCZF 8 ) is encoded by the ZBTB7A gene, and is a member of the POZ/BTB and Krüppel (POK) family of transcription factors. POK transcription factors can bind DNA through a Krüppellike-DNA-binding domain and repress transcription by recruiting co-repressor complexes through the POZ (Pox virus and Zinc finger) domain 9 . POK transcription factors have been recognized as critical developmental regulators and have been directly implicated in human cancer 10 . For example, BCL6 (B-Cell Lymphoma 6) and PLZF (Promyelocytic Leukemia Zinc Finger) are critical players in the pathogenesis of Non-Hodgkin's Lymphoma and acute promyelocytic leukemia, respectively 11,12 . LRF shares structural similarities with BCL6 and PLZF and plays critical context-dependent role in embryonic development, haematopoiesis and tumorigenesis 6,[13][14][15][16][17][18][19] .
In this work, we identify a novel and transcriptional independent function for LRF in the maintenance of genomic stability by regulation of cNHEJ. Mechanistically, we demonstrate that LRF is rapidly recruited on the sites of DNA damage where, by binding DNA-PKcs, it stabilizes the DNA-PK complex, in turn promoting DNA-PKcs kinase activity and efficient DSB repair. Importantly, LRF downregulation, a frequent hallmark of different types of human cancer, restores radiation sensitivity in p53 null cells, thus becoming a new potential biomarker of remarkable therapeutic relevance.

Results
LRF is required for maintenance of genomic integrity. LRF is a critical repressor of the tumour suppressor gene Arf, and cells such as mouse embryonic fibroblasts (MEFs), which lack Lrf become refractory to oncogenic transformation and undergo premature senescence 6 . In an effort to identify new functions of LRF unrelated to Arf regulation through a clean genetic approach, we compared the effects of acute Lrf deletion in Lrf flox/flox or Arf À / À Lrf flox/flox MEFs through infection with a Cre recombinase-containing retrovirus. Although Cre expression in both wild-type and Arf À / À MEFs had no effect on cell proliferation ( Supplementary Fig. 1a), and Cre-mediated deletion of Lrf in Lrf flox/flox MEFs triggered the expected growth suppression through Arf-dependent cellular senescence 6 ( Fig. 1a), surprisingly, loss of Lrf caused a profound growth suppression in the Arf À / À MEFs as well (Fig. 1a). The growth defect of Arf À / À Lrf deleted (Arf À / À Lrf f/f cre) MEFs was accompanied by evidence of chromosome breakage, as shown by Giemsa staining of metaphase chromosome spreads (Fig. 1b). Telomere Fish fluorescent in situ hybridization staining of chromosome spreads also indicated accumulation of chromosome breaks, aneuploidy, polyploidy and abnormal chromosomes in Arf À / À Lrf deleted MEFs ( Supplementary  Fig. 1b). Accordingly, neutral comet assay showed a significant accumulation of DNA DSBs in Lrf deleted MEFs (Fig. 1c), and immunofluorescence and western blot studies confirmed a marked increase in g-H2AX staining (Fig. 1d,e). To further characterize this phenotype, we assessed whether LRF conditional inactivation triggers unrepaired DNA damage in vivo. Villin-Cre and Mx1-Cre transgenes were used to delete floxed Lrf in the mouse intestine and hematopoietic systems, respectively 20,21 . Importantly, in LRF conditional knockout intestine and spleen the downregulation of LRF ( Supplementary Fig. 1c) was associated with a significant increase of g-H2AX levels (Fig. 1f), suggestive of persistent DNA damage in these cells 22 .
LRF deficiency sensitizes cells to ionizing radiation. Since LRF inactivation results in persistent DNA damage and genomic instability, we used clonogenic survival assays to assess the sensitivity of Arf À / À and Arf À / À Lrf deleted MEFs to different types of DNA-damaging agents. These included g-radiation, the radiomimetic drug phleomycin, the Topoisomerase II inhibitor ICRF-193, the Topoisomerase I inhibitor Camptothecin, and the DNA cross-linking agent, mitomycin C. Compared with Arf À / À control MEFs, Arf À / À Lrf deleted cells revealed hypersensitivity to g-radiation, phleomycin and ICRF-193 (Fig. 2a,b and Supplementary Fig. 2a), but no alteration in mitomycin C and Camptothecin sensitivity (Fig. 2c, and Supplementary Fig. 2b). Furthermore, upon treatment with phleomycin at various concentrations, Arf À / À Lrf deleted MEFs displayed a significant increase of comet tail DNA content and g-H2AX levels (Fig. 2d,e). We then further tested in vivo whether Lrf null mutants are hypersensitive to ionizing radiation (IR). Constitutive Lrf inactivation results in embryonic lethality 6 , while conditional Lrf inactivation in the adult hematopoietic system ('Lrf cKO'), using Mx1-Cre and pIpC induction, is compatible with a normal lifespan 14 . After a single dose of whole-body g-irradiation (7.5 Gy), all Lrf cKO mice died within 16 days, while all wild-type control mice remained healthy for 2 weeks after irradiation (Fig. 2f). After irradiation, Lrf cKO bone marrow cells accumulated much more unrepaired DNA damage (shown by g-H2AX staining) and became apoptotic (by cleaved caspase 3 staining). Lrf cKO mice were found to have died from acute bone marrow failure ( Supplementary Fig. 2c).
LRF participates in Xrcc4-dependent NHEJ. To directly test whether a specific DSB repair process requires LRF function, we next took advantage of selective DSB repair reporter assays (Fig. 3a, and Supplementary Fig. 3c) 23 . Notably, siRNA-mediated knockdown of LRF caused a significant decrease in NHEJ repair efficiency as shown by decreased I-SceI-induced GFP expression in NHEJ reporter cells (Fig. 3b). Classical NHEJ (cNHEJ) is a rapid and efficient process, requiring DNA Ligase IV and XRCC4. In cells lacking either of these genes, rejoining of DSBs occurs through a slower, highly error-prone process termed 'alternative end joining' (aEJ) 4,5,[24][25][26] . Using biallelically deleted Xrcc4 (here termed Xrcc4 D/D ) NHEJ reporter ES cells, we observed only a modest and statistically insignificant decrease of NHEJ efficiency after Lrf knockdown (Fig. 3c). Similarly, pharmacological inhibition of DNA-PKcs activity (NU7441) combined with siRNA-mediated knockdown of LRF showed only a mild and statistically insignificant reduction of NHEJ efficiency in Xrcc4 proficient ES cells when compared with DNA-PKcs inhibition alone ( Supplementary Fig. 3a) 23 . On the other hand, depletion of LRF had no impact on rejoining of I-SceI-induced DSBs in human U2OS cells carrying a reporter of micro-homology-mediated end joining 27 ( Supplementary  Fig. 3b), a frequent mediator of aEJ 27,28 . We found that LRF is also dispensable for HR, based on experiments performed using mouse ES and U2OS cells carrying an HR reporter 29 ( Fig. 3d and Supplementary Fig. 3c,d).
Collectively, these results clearly define LRF as a novel important player in Xrcc4/DNA-PK-dependent cNHEJ pathway of DSB repair. In future experiments, it will be instructive to define more fully the structure-function analysis of LRF-DNA-PKcs interaction as well as the epistatic relationships between LRF inactivation and the loss of other c-NHEJ genes.
LRF is known to act as a transcription factor. We therefore used microarray analysis in wild-type and Lrf conditional knockout MEFs to decrypt its activity in DSB repair (Supplementary Tables 1 and 2). Surprisingly, Lrf knockout cells did not display significant alterations in the expression of genes known to be essential for cNHEJ either at early (Supplementary Table 2) or at late passages ( Supplementary Fig 3e,f). Only the expression of MRE11, implicated in both aEJ and cNHEJ, resulted downregulated in Lrf null cells compared with wild-type (Supplementary Tables 1 and 2). Although MRE11 mild downregulation could explain the slight reduction in the efficiency of aEJ noted in Fig. 3c, the much more pronounced impact on the cNHEJ pathway in LRF-depleted cells (Fig. 3b) suggests a more fundamental role of LRF in this mechanism of DSB repair. Strongly supporting this hypothesis, chromatin immunoprecipitation (ChIP) experiments demonstrated the binding of LRF to site-specific DSBs generated by I-SceI (Fig. 3e), while an in vivo imaging approach proved the ability of LRF to localize to the vicinity of DSBs generated by laser damage (Fig. 3f, upper panel), with a kinetic closely comparable to other cNHEJ proteins, such that of Ku80 and DNA-PKcs 30 . Importantly, LRF recruitment to DSBs is not dependent on DNA-PKcs or Ku80 (Fig. 3f, lower panels-middle/right). Taken together, these results point to a transcriptional independent role for LRF in cNHEJ, a conclusion consistent with the observation that Lrf-deleted cells are hypersensitive to IR, phleomycin and ICRF-193, but not to mitomycin C and camptothecin.
LRF interacts with DNA-PKcs and regulates DNA-PK function.
To determine the transcriptional independent role of LRF in cNHEJ, we first assessed whether LRF could associate with DSB repair protein complexes. To this end, LRF-associated proteins were isolated through tandem affinity purification from HeLa cells stably expressing human LRF tagged with Flag and haemagglutinin (HA) epitopes and analysed by mass spectrometry. Importantly, we found LRF associated with the DNA-PK protein complex, including DNA-PKcs, Ku70 and Ku80 ( Supplementary Fig. 4a). Mass spectrometry data were then validated in pull-down experiments with overexpressed FLAG/ HA tagged LRF ( Supplementary Fig. 4b, and Supplementary  Table 3), as well as through the reciprocal co-immunoprecipitation of endogenous LRF with DNA-PKcs, Ku70 or Ku80 (Fig. 4a). The association between DNA-PKcs and the Ku70/Ku80 heterodimer is DNA dependent 31 . We therefore determined whether the association between LRF and DNA-PKcs or Ku requires DNA. Endogenous co-immunoprecipitations in the presence of ethidium bromide (50 mg ml À 1 ), which disrupts DNA-dependent interactions, indicated that the association between LRF and Ku is strictly dependent on the presence of DNA, while the interaction between LRF and DNA-PKcs, although favoured by DNA, persists in its absence (Fig. 4b).
Furthermore, in vitro binding with FLAG-tagged LRF and purified DNA-PK components also indicated that the binding of LRF to Ku70 and Ku80 requires DNA, while DNA bridging is not necessary for the interaction between DNA-PKcs and LRF (Fig. 4c). Interestingly, we observed that, in the absence of Lrf, the mobilization of DNA-PKcs to the chromatin fraction following DNA damage, is significantly decreased (Fig. 4d, Figure 2 | LRF-deficient cells are hypersensitive to ionizing radiation. (a-c) Clonogenic survival of Arf À / À and Arf À / À LRF-deleted MEFs treated with g-radiation (a), phleomycin (b) and mitomycin C (c). Data from n ¼ 4 independent experiments are presented as mean ± s.e.m. Associated P value calculated by Student's t-test analysis is indicated. (d) DNA damage levels in Arf À / À and Arf À / À LRF-deleted MEF treated with phleomycin assessed by comet assay. The percentage of DNA in comet tails is scored from 200 cells of three different experiments and presented as mean ± s.e.m. Associated P value calculated by Student's t-test analysis is indicated. (e) g-H2AX levels assessed by flow cytometric analysis of Arf À / À and Arf À / À LRF deleted MEFs, 1 h after 20 or 300 mM phleomycin treatment. Data from n ¼ 4 independent experiments are presented as mean ± s.e.m. Associated P value calculated by Student's t-test analysis is indicated. (f) Survival curve of Lrf hematopoietic system conditional knockout mice (Lrf cKO) (n ¼ 6) and sibling control mice (n ¼ 9) after single dose of whole-body g-radiation (7.5 Gy).
Supplementary Fig. 4c). Furthermore, significantly less DNA-PKcs was co-immunoprecipitated with Ku antibodies in LRF-depleted cells compared with controls ( Fig. 4e and Supplementary Fig. 4d Fig. 4f), we observed that the retention time, but not the recruitment time, of DNA-PKcs on the laser-induced breaks was significantly decreased in LRF knockdown compared with control cells (Fig. 4f). Endogenous DNA-PKcs autophosphorylation on serine 2056, a known correlate of DNA-PK activity 32-36 , was significantly reduced in LRF knockdown compared to control cells following treatment of cells with bleomycin (Fig. 4g). In keeping with these findings, in an in vitro assay using purified DNA-PKcs protein and extracts from wild-type and Lrf-deleted MEFs, we observed a substantially lower DNA-PKcs kinase activity in the absence of Lrf compared with controls (Fig. 5a).
LRF loss restores IR sensitivity in p53 null cells. A characteristic feature of p53 null cells is their resistance to IR 37 . This effect is reported to require normal DNA-PK function and loss of DNA-PKcs, Ku70 or Ku80 can restore the radiation sensitivity of p53 null cells 38 . We therefore tested whether LRF loss, which is observed in advanced cancers 15,17,18 , could restore IR sensitivity in p53-deficient cells. Indeed, Lrf loss restored IR-induced apoptosis of p53 null MEFs (Fig. 5b and Supplementary Fig. 4g).

Discussion
The human genome encodes B60 POK family proteins 16,19,39 , containing an amino-terminal POZ domain and several carboxy-terminal C2H2 Zinc finger domains. POK family proteins have been implicated in embryogenesis, the pathogenesis of cancer and other diseases primarily as transcriptional regulators of gene expression, although ZBTB1 has been recently shown to exert transcription-independent functions intriguingly associated with DNA repair 40 . Even though originally characterized as a proto-oncogene 6 , human ZBTB7A is located at 19p13.3, a chromosomal region that is frequently lost in different types of human cancer, including prostate cancer 18,41,42 . Interestingly, LRF has been recently characterized as a potent context-dependent tumour suppressor through the transcriptional repression of oncogenic pathways and glycolytic metabolism [15][16][17][18] . Here we identify LRF, a bonafide member of the POK family of proteins, as an important regulator of the DNA-PK complex required for the maintenance of genome integrity, which is a novel and unexpected function that LRF exerts independently of its transcriptional function. DNA-PKcs is the largest known protein kinase in the cell, which belongs to the phosphatidylinositol-3 (PI-3) kinase-related-kinase (PIKK) super-family based on primary structure. In current models, Ku association with DNA ends initiates a complex DNA-PKcs-dependent signalling pathway through phosphorylation of downstream effectors responsible for DSB repair 43 . Importantly, this study unravels a novel and unexpected transcriptional independent function of POK family of proteins into the critical cellular processes of DNA-PK function and cNHEJ. Notably, BCL6 (B-cell lymphoma-6), a further member of the POK family and a key oncogenic driver in B-cell lymphoma 44 , has been demonstrated to physically bind LRF 45  (a) DNA-PK kinase activity was assessed in total cell extracts from Arf À / À control (Arf À / À Lrf f/f con) and Arf À / À LRF-deleted MEF (Arf À / À Lrf f/f Cre). (b) Clonogenic survival of p53 À / À Lrf þ / þ and p53 À / À Lrf À / À MEF after different doses of g-radiation. (c) Schematic representation of LRF function in NHEJ DSBs repair pathway. Data are presented as mean±s.e.m. of three independent experiments. Associated P value calculated by Student's t-test analysis is indicated.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9325 ARTICLE has been recently reported to characterize specific subgroups of cancer patients 15,17,18,46,47 . As a novel component of the DNA-PK complex and regulator of DNA-PK stability and activity (Fig. 5c), LRF represents an attractive biomarker with important therapeutic implications since its downregulation might serve to identify those tumours that are particularly dependent on NHEJ activity, such as for instance a subset of p53-null cancers, towards therapeutic treatments based on genotoxic agents, radiation, or PARP inhibitors following the synthetic lethality paradigm. Retrovirus transduction of mouse embryonic fibroblast. All animal procedures have been approved by the Beth Israel Deaconess Medical Center and Harvard Medical School institutional review board. Lrf þ / À , Lrf flox/flox , Arf À / À and p53 À / À mice are previously described 6,14,48,49 . p53 À / À and p53 À / À Lrf À / À MEF were prepared from E13.5 mouse embryos obtained from the intercross of p53 À / À Lrf þ / À mice. To generate primary Lrf flox/flox and Arf À / À Lrf flox/flox MEF, Arf þ / À Lrf flox/flox mice were intercrossed. MEFs were transduced with MSCV-PIG-Cre or empty control vector retrovirus for 2 days at passage 2, and then selected with 2 mg ml À 1 puromycin for 2 days before use in subsequent experiments.

Methods
Microarray analysis. Lrf flox/flox MEFs were transduced with MSCV-PIG-Cre or empty control vector retroviruses for 2 days at passage 2. After selection with puromycin for 2 days, total RNAs were purified using the RNAeasy Mini Kit (Qiagen) and treated with RNase-free DNase set (Qiagen). RNAs from two independent experiments were labelled and hybridized using Affymetrix GeneChip HT Mouse Genome 430 arrays by the Beth Israel Deaconess Medical Center Genomics and Proteomics Center. Genes with normalized data values differing by a factor greater or less than 1.5-fold were selected and further evaluated statistically.
Cell growth assay. Cells were seeded in 12-well plates at a density of 10 4 /well, then left to grow for 4 days. Cells were fixed by paraformaldehyde at each time point, and the cell number determined by crystal violet staining as described 50 .
Comet assay. DNA lesions were assessed using a single-cell gel electrophoretic comet assay kit (Trevigen). Cells were combined with low melting point agarose and pipetted onto a slide. The cells were lysed, then subject to electrophoresis at 20 V for 30 min in TBE buffer. Following electrophoresis, slides were washed, dehydrated and stained with SYBR Green I. Images were taken with a fluorescent microscope and scored by CometScore software (TriTek Corporation).
G-banding and telomere FISH of metaphase chromosome. Metaphase chromosome spreads were prepared from exponentially growing cells after treatment with demecolcine. For G-banding, the metaphase chromosomes were then treated with trypsin and stained with Giemsa according to standard procedures. Telomere FISH was performed using a Cy3-labelled peptide nucleic acid probe (Cy3-(CCCTAA) 3 ) in metaphase chromosome spreads. Both the probe and the slides were heat denatured (80°C for 5 min) and hybridized at 37°C for 2 h. Slides were counterstained with DAPI. Images were captured using Zeiss microscope equipped with a CCD camera.
Protein complex purification and mass spectrometry. Procedures for LRFassociated protein complex purification have been described in detail previously 51 .
Briefly, FLAG-HA tandem tagged human LRF was stably expressed in HeLa cells, then nuclear extracts were sequentially immunoprecipitated with anti-FLAG and anti-HA beads. The LRF binding proteins were separated using SDSpolyacrylamide gel electrophoresis (SDS-PAGE), and protein bands were identified by mass spectrometry.
Immunoblotting and immunoprecipitation. Cells were lysed in buffer (50 mM Tris, pH8.0, 150 mM NaCl and 0.5% NP-40). Protein concentrations of the lysates were measured by Bradford assay. The lysates were then resolved by SDS-PAGE and immunoblotted with the indicated antibodies. For immunoprecipitation, 1 mg of cell lysate was incubated with the appropriate antibodies for 3-4 h at 4°C followed by 1-h incubation with protein A beads (Santa Cruz). Immuno-complexes were washed with buffer (20 mM Tris, pH8.0, 100 mM NaCl, 1 mM EDTA and 0.5% NP-40) before being resolved by SDS-PAGE and immunoblotted with the indicated antibodies. Uncropped scans of the most important blots are supplied as Supplementary Information.
Cell fractionation. p53 À / À Lrf þ / þ and p53 À / À Lrf À / À cells were incubated with 100 mM phleomycin for 1 hour at 37°C. The cell pellet was resuspended in buffer (150 mM NaCl, 50 mM Hepes PH7.5, 1 mM EDTA, 0.1% Triton X-100, protease and phosphatase inhibitor) for 10 min on ice. Lysates were pelleted, and detergent extractable supernatant collected. DNA-damaging agents clonogenic survival assay. Four hundred cells were seeded in six-well plates 24 h before treatment with the indicated drugs. g-radiation was supplied with a Cesium-137 source. After 10 days, colonies were stained with crystal violet and scored. A colony was defined as a cluster of more than B50 cells. Cells without drug treatment were used as control. Survival ratio ¼ sample/ control Â 100%. Results were reported as mean ± s.e.m. from three independent experiments.
DNA-PK kinase assay. DNA-PK kinase activity was measured using the SignaTECT DNA-Dependent Protein Kinase Assay System (Promega). Total cell lysates were extracted using a buffer containing 1%NP-40, 150 mM NaCl and 50 mM Tris (pH8.0). Endogenous DNA of cell lysates was removed using Sepharose fast flow (GE Healthcare). For each reaction, 2 mg of cell lysates were used, reactions were incubated at 30°C for 10 min, and then the supernatant was spotted onto a SAM membrane. DNA-PK protein kinase activity was calculated as the incorporation of 32 P into the peptide using a phosphoimager.
Immunofluorescence. Cells were seeded in 24-well plates containing round glass coverslips at the density of 2 Â 10 4 /well. 24 h after plating, cells were fixed with paraformaldehyde, permeabilized in 0.1% Triton-X-100/phosphate buffered sulphate (PBS). Coverslips were then incubated with primary antibody diluted in 1% BSA/PBS for 1 h. After washing, coverslips were incubated in secondary antibody diluted in 1% BSA/PBS for 1 h. Coverslips were washed, stained with DAPI, mounted and analysed by confocal microscopy (Zeiss).
Flow cytometry. Cells were trypsinised, fixed with 4% formaldehyde and permeabilised with 90% methanol. Cells were then incubated with Alexa Fluor 647 conjugated g-H2AX antibody (Cell Signaling Technology) in 0.5% BSA /PBS for 1 h at room temperature. Cells were washed with 0.5% BSA /PBS then analysed by flow cytometry (LSR II, BD Biosciences). Data was analysed with FCS Express V3 software.
RT-qPCR Primers. Total RNA was extracted with TRIzol Reagents (Invitrogen) according to the provided protocol. 1 mg total RNA was reversed transcribed with iScript cDNA Synthesis Kit (Bio-Rad). Real-time quantitative PCR was performed using diluted cDNA, SYBR Green JumpStart Taq ReadyMix (Sigma) and appropriate primers in StepOnePlus Real-Time PCR System (Applied Biosystems). Primers sequence is reported in Supplementary Table 4.
Statistical analysis. Results are expressed as mean±s.d or s.e.m as noted.
Comparisons between groups were assessed using Student's t-test analysis. Pr0.05 was considered significant.