A long noncoding RNA sensitizes genotoxic treatment by attenuating ATM activation and homologous recombination repair in cancers

Ataxia-telangiectasia mutated (ATM) is an apical kinase of the DNA damage response following DNA double-strand breaks (DSBs); however, the mechanisms of ATM activation are not completely understood. Long noncoding RNAs (lncRNAs) are a class of regulatory molecules whose significant roles in DNA damage response have started to emerge. However, how lncRNA regulates ATM activity remains unknown. Here, we identify an inhibitor of ATM activation, lncRNA HITT (HIF-1α inhibitor at translation level). Mechanistically, HITT directly interacts with ATM at the HEAT repeat domain, blocking MRE11-RAD50-NBS1 complex–dependent ATM recruitment, leading to restrained homologous recombination repair and enhanced chemosensitization. Following DSBs, HITT is elevated mainly by the activation of Early Growth Response 1 (EGR1), resulting in retarded and restricted ATM activation. A reverse association between HITT and ATM activity was also detected in human colon cancer tissues. Furthermore, HITTs sensitize DNA damaging agent–induced cell death both in vitro and in vivo. These findings connect lncRNA directly to ATM activity regulation and reveal potential roles for HITT in sensitizing cancers to genotoxic treatment.


Introduction
Cells are inevitably challenged by endogenous or exogenous sources of DNA damage [1]. To preserve genetic integrity, organisms have evolved elegant mechanisms to cope with various forms of DNA damage, collectively known as the DNA damage response (DDR) [2]. Deficiency in the DDR is associated with genomic instability, predisposition to cancer, or cell death in cases when damage is irreparable [3]. DNA damage also represents the backbone of cancer treatment. In this context, activation of DNA damage repair pathways promotes genotoxic resistance, which remains a major obstacle in successful cancer treatment [4,5]. Thus, unveiling the mechanisms underlying the DDR may not only inform our knowledge of tumorigenesis but also provide predictive markers for the patients' responses to therapeutic DNA damage and offer new opportunities for the improvement of treatment efficiency. Double-strand breaks (DSBs) are the most toxic DNA lesion that can arise following ionizing radiation (IR) or DNA-based chemotherapy [4]. Cells utilize two prominent pathways to repair DSBs: homologous recombination (HR) and nonhomologous end joining (NHEJ) [6]. Ataxia-telangiectasia (A-T) mutated (ATM) is an apical kinase of phosphorylation cascades in the HR pathway, at the heart of the cellular response to DSBs [7]. One of the most striking features of patients with A-T, a disorder caused by somatic ATM mutation, is their sensitivity to IR and DNA-damaging agents that induce DSBs [8]. Conversely, activation of ATM is linked to the survival of cancer cells following therapy [9,10]. ATM is an attractive target for cancer treatment. Several highly selective small molecule inhibitors of ATM have entered clinical trials in patients with advanced cancers [11,12].
Given its physiological and pathological significance, the regulation of ATM activity remains a major focus of the field. The MRE11-RAD50-NBS1 (MRN) complex is vital for the ATM-dependent DDR [13,14]. MRN is the first complex to be recruited to DSB sites, where it provides a platform for ATM recruitment and facilitates the autophosphorylation of ATM at serine (S) 1981, leading to optimal ATM activation that subsequently triggers the phosphorylation of a variety of ATM effectors essential for the DDR, such as checkpoint kinase 2 (Chk2) and Breast Cancer gene 1 (BRCA1) [15]. Interestingly, mutations in MRN component genes Meiotic Recombination 11 (Mre11) or Nijmegen Breakage Syndrome 1 (NBS1), which are linked to A-T-like disorder and Nijmegen breakage syndrome, respectively, are accompanied by defective ATM activity with symptoms resembling those of A-T patients [16,17], suggesting that MRN is required for the activation of ATM in vivo.
MRN is clearly involved in ATM activation at DSBs; however, how it activates this is still not fully understood. In addition, it has recently become apparent that multiple layers of regulation exist to ensure that the ATM signal is accurate and restricted to appropriate cellular contexts. For example, posttranslational modifications of MRN components, such as NBS1 ubiquitination [18] or MRE11 UFMylation [19], promote MRN-mediated ATM recruitment to DSB sites and thus play important roles in initiating or amplifying the ATM signal. It has also been noted that the rise and fall of ATM activity are equally important for the proper execution of DSB repair [6,20,21]. However, compared with the mechanisms of ATM activation, much less is known about the mechanisms of ATM inhibition.
To complicate the matter, evidence is steadily accumulating that long noncoding RNAs (lncRNAs), a class of transcribed RNA molecules greater than 200 nucleotides (nt) in length with extremely limited protein-coding capabilities, play essential roles in a wide range of physiological and pathological processes [22,23]. Interestingly, lncRNA integrates contextual and environmental cues not only during development but also under multiple stresses [24]. A variety of lncRNAs are DNA damage responsive and able to modulate the DNA repair program in turn, which highlights the essential roles of lncRNAs in the DDR. For example, lncRNAs DINO (Damage Induced Noncoding) [25], PANDA (P21 associated ncRNA DNA damage activated) [26], and linc-p21 [27] are induced upon DNA damage and are engaged in a feedback loop to regulate apoptosis or DNA repair by modulating p53 protein stability or p53's transcriptional activity. However, despite being a master regulator of the DDR, whether and how ATM activation is directly regulated by lncRNAs remains to be determined.
We recently identified an lncRNA, linc00637, which is a 2,056-nt intergenic lncRNA containing three exons mapped at 14q32, a chromosome region that has been associated with the early onset and metastatic recurrence of colon cancer and many other types of cancers [28][29][30]. We found that linc00637 is down-regulated in multiple types of cancers. Further functional studies reveal that it is a hypoxia-responsive lncRNA, whose expression is reduced in the face of hypoxic stress and which plays essential roles in inhibiting hypoxia-inducible factor 1α (HIF-1α), mainly by repressing its translation; thus, we named it HITT (HIF-1α inhibitor at translation level) [31]. Besides hypoxia, cancer cells are also constantly insulted by DNA damage, particularly when receiving conventional genotoxic treatments. Thus, we wondered whether HITT is a stress responsive gene that plays roles in regulating the DDR. To this end, we explored the expression, function, and mechanisms of HITT upon DSBs. Interestingly, we identify HITT as a physiological brake of ATM activation, which is induced and maintained at high levels after DSBs and contributes to the promoted chemosensitization. Mechanistically, HITT is physiologically associated with ATM protein, thereby blocking MRN-mediated ATM recruitment. Thus, HITT represents the first RNA molecule that is directly involved in ATM regulation and may be a potential treatment for anticancer chemosensitization.

DSB-induced HITT expression represses HR repair
To understand whether HITT is a DNA damage-responsive lncRNA, expression levels of HITT were measured after exposing HCT116 cells to a panel of cytotoxic agents. HITT was significantly increased by drugs that have been reported to induce DSBs, such as doxorubicin (Dox), etoposide (Eto), the radiomimetic compound bleomycin (Bleo), and calicheamicin (CLA) [32][33][34], but not DSB-independent pro-death treatments, such as tumor necrosis factor-α (TNF-α)/cycloheximide (CHX) and taxol, although similar death rates were induced ( Fig 1A). In addition, the induction of HITT was dose-dependent with an approximately 5-to 7-fold increase at 4 μg/ml Dox in both HeLa and HCT116 cells (Fig 1B). Time-course analysis revealed that HITT was increased with prolonged Dox treatment and reached a plateau after 2 h (Fig 1C). Similar time-or dose-dependent HITT expression patterns were detected after CLA, Eto, and Bleo exposure (S1A and S1B Fig). Furthermore, DSB-induced HITT expression occurred in a panel of cancer cells with different tissue origins, regardless of p53 status (S1C and S1D Fig). Therefore, DSB-induced HITT upregulation is likely a common phenomenon that is independent of p53.
This inspired us to investigate whether HITT plays roles in regulating DSB repair progress. In a DNA comet assay, significant increases in comet tail scores were detected after Dox treatment, and this damage was rapidly repaired in a time-dependent manner after washing out of Dox (Figs 1D, 1E and S1E). Interestingly, stable HITT transfectants markedly attenuated the DDR process (Figs 1D and S1E). Two independent small interfering RNAs (siRNAs) specifically targeting HITT reduced HITT expression by approximately 50%. Accordingly, DNA repair progress was significantly accelerated (Figs 1E and S1E).
HR and NHEJ are two prominent pathways in regulating DSB repair. Thus, the efficiencies of HR and NHEJ were further compared after HITT overexpression or knockdown (KD) using cell lines containing the indicated integrated green fluorescent protein (GFP) reporters upon I-SceI-induced breaks [35]. RAD51 [36] and X-Ray Repair Cross Complementing 4 (XRCC4) [37] are key regulators of HR and NHEJ pathways, respectively. siRNA-mediated RAD51 and XRCC4 KD, respectively, reduced HR and NHEJ as expected (S1F Fig). Under such conditions, we found that HITT overexpression markedly inhibited HR, as indicated by Direct Repeat (DR)-GFP, but not NHEJ, as indicated by EJ2-and EJ5-GFP (Fig 1F and 1G). HITT KD by two independent siRNAs constantly and significantly increased HR but not NHEJ (Fig 1F and 1G). These data demonstrate that HITT is up-regulated upon DSBs independently of p53 and plays important roles in restraining HR repair.

HITT reduces HR repair by repressing ATM activation
We next sought to understand the mechanisms by which HITT impairs HR. ATM activation is needed for the initiation of DSB repair by HR [38]. In line with this widely accepted notion, two independent ATM inhibitors (ATMi), KU-60019 (ATMi-1) and KU-55933 (ATMi-2), significantly reduced DNA repair in general (Fig 2A) and HR ( Fig 1F). Interestingly, HITT completely lost its ability to regulate HR repair in the presence of ATMi (Figs 1F and 2A). ATMi had no obvious impacts on HITT expression levels (S2A Fig). In addition, it has been recently proposed that ATM inhibits HR by repressing DNA resection [39]. In line with this idea, we found that HITT inhibits DNA resection in an ATM activity-dependent manner by assessing the accumulation of RPA2 into nuclear foci, a well-established resection marker [40] (S2B Fig). Therefore, we reasoned that HITT inhibits DSB repair by interfering with HR through blocking ATM-dependent pathways.
To this end, the activity of ATM, as indicated by the autophosphorylation of ATM at S1981 and the phosphorylation of the well-documented ATM target Chk2, was first compared in . DNA damage was monitored by comet assay after DMSO or Dox (1 μg/ml) treatment, or at different periods of time following Dox washout (middle). Tail moment per cell is analyzed as described in the Materials and methods and presented in the bar graph (right), scale bar, 10 μm. (F, G) HR or NHEJ efficiencies of ISce-I-induced DSBs in U2OS cells containing DR-GFP (HR, F) or EJ2-or EJ5-GFP reporter (NHEJ, G) were determined by measuring GFP-positive cells by flow cytometry (FACS) after overexpression or KD of HITT in the presence or absence of 10 μM ATMi-1 or ATMi-2. Data are derived from three independent experiments and presented as mean ± SEM in the bar graphs. Values of controls were normalized to 1. � P < 0.05; �� P < 0.01; relative to the untreated control ("Ctl.") (A, B, C, F, and G); #P < 0.05; ##P < 0.01; relative to vector ("vect.") or si-Ctl. with the indicated treatment (D and E). See also S1 Fig Neither HITT expression nor KD affected total protein levels of ATM and Chk2 (Fig 2B and 2C). In contrast to ATM, the activation of another phosphatidylinositol 3-kinase-related protein kinase family member, Ataxia Telangiectasia And Rad3-Related Protein (ATR), did not change with HITT expression (Fig 2B and  2C). Therefore, HITT specifically inhibits ATM activity.
A time-course experiment was further performed. A mild induction of p-ATM occurred as early as 5 min after Dox treatment and kept rising in a time-dependent manner with the prolonged treatment, reaching plateau 2 h after treatment (Fig 2D), whereas HITT was increased relative late, 1 h after Dox treatment (S3A Fig). Dynamic induction of p-ATM was significantly delayed and attenuated by HITT overexpression (Fig 2D) and accelerated and elevated by HITT KD (Fig 2E).
It is also noteworthy that inhibition of HITT-mediated ATM activation was not specific to cell type or p53 status, with similar phenomena observed in p53 null H1299 and SW620 cells ( It is known that HR is more prevalent after DNA replication [6]; however, HITT expression was not related to cell cycle ( Thus, we deduced that the retarded S and G2/M phase entrance in HITT overexpression cells was due to decreased ATM activation, which is consistent with previous reports that ATM regulates cell-cycle progression [41].
These data collectively suggest that HITT is an important inhibitor of ATM activation and therefore plays essential roles in attenuating HR repair upon DSBs.

HITT inhibits MRN-mediated ATM recruitment to the DSB sites
We next investigated the underlying mechanisms by which HITT inhibits ATM activity. MRN is a sensor of DSBs and integral to ATM-dependent DNA repair signaling [42]. NBS1 is an MRN component that plays key roles to recruit ATM to the DSB sites. As expected, NBS1 KD reduced ATM activity. Interestingly, despite no obvious impacts on HITT levels (S4A Fig), NBS1 KD completely abolished the effect of HITT on ATM activity ( Fig 3A). These data suggest that HITT inhibition of ATM is dependent on MRN action. However, neither HITT overexpression nor KD produced a significant impact on MRN proteins (Fig 2B and 2C). In addition, cell-staining and chromatin fractionation assays revealed that HITT does not influence the formation of nuclear foci by NBS1 ( Fig 3B) and RAD50 (S4B Fig) or the association of MRN components (NBS1 and RAD50) with chromatin (S4C Fig). In contrast, DNA damage-induced p-ATM foci were barely detectable in HITT overexpression HeLa cells (Figs 3B and S4B). In agreement, the chromatin-associated ATM was decreased with HITT overexpression and increased with HITT KD, whereas the association of chromatin with ATR was not affected by either HITT expression or KD (S4C Fig). These data suggest that HITT may inhibit ATM activation by regulating MRN-mediated ATM recruitment to sites of DNA damage. Indeed, a coimmunoprecipitation (co-IP) assay revealed that the association of p-ATM or ATM with NBS1, a key component of MRN that has been reported to be directly associated with ATM [43], was detected and was dramatically decreased upon ectopic HITT expression; however, the association of NBS1 with another MRN component, RAD50, remained unaffected ( Fig 3C). Therefore, HITT appears to specifically interfere with the association of ATM with MRN upon DSBs.

HITT is physically associated with ATM at an essential NBS1-binding site
We next asked how HITT inhibits the association between ATM and NBS1. lncRNA may act by interacting with proteins. We have previously shown that HITT is distributed both in the cytoplasm and in the nucleus [31] and therefore wondered whether HITT binds with ATM or MRN. To this end, an in vitro RNA-binding assay was first applied. As shown, the assay revealed that ATM coprecipitated with Biotin-HITT but not Biotin-antisense-HITT (Fig 3D), whereas none of the MRN components bound with HITT or antisense-HITT ( Fig 3D). UV cross-linking and immunoprecipitation (CLIP) is standard in RNA research used to determine the direct interactions between proteins and nucleic acids. As revealed by CLIP assay, HITT was present in ATM, but not immunoglobulin G (IgG) or RAD50, immunoprecipitates ( Fig  3E). The association between ATM and HITT was not only significantly enhanced with ectopic HITT expression but also increased with Dox treatment (Fig 3E and 3F). These data suggest that HITT is physically and specifically associated with ATM, which may be biologically significant in regulating the DDR.
We further investigated what the molecular mechanism by which HITT binds with ATM is. HITT does not have a homolog in mice, which thus provides a convenient tool to map the HITT sequence that contributes to its association with ATM, without interference from endogenous HITT. Different HITT fragments (F) were generated and similar levels of HITT were introduced into the mouse cell line 4T1. CLIP of endogenous ATM revealed that comparable rates of FL and F3 were present in ATM immunoprecipitates, whereas F1 and F2 were unlikely to contribute to the interaction ( Fig 3G). The binding sites within F3 were further narrowed down to F3-1 (1,030-1,247 bp) ( Fig 3G). In line with this result, we found that F3 (F3-1), but not the other fragments, inhibited ATM activity ( Fig 3H). Full-length HITT and F3 elicits similar effect in inhibiting HR efficiency ( Fig 3I).
We also examined which domain in ATM contributes to the interaction with HITT. To answer this question, we generated different ATM truncations as indicated in Fig 3I. ATM mutant types (MTs) that lost the MT3 sequence within the HEAT repeat domain failed to bind with HITT ( Fig 3J). The direct interaction between HITT and ATM-MT3 was further verified by in vitro RNA pull-down assay using in vitro-translated ATM-MT3 protein and in vitrosynthesized sense-HITT ( Fig 3K). Interestingly, MT3 has been identified to be an essential NBS1-binding site [43], suggesting HITT may mask the NBS1-binding sites, leading to the repressed association between ATM and NBS1. In agreement, we found that NBS1 bound with MT2 and MT3, which contain NBS1-binding domain, but not with MT4. Such interaction was dramatically reduced in cells overexpressing HITT compared with control cells (Fig 3L).
It is thus logical to propose that HITT binds the NBS1-binding site in ATM (MT3) via F3-1 (1,030-1,247 bp) and prevents NBS1-mediated ATM recruitment to DSBs.

HITT is transcriptionally activated by the induction of EGR1 upon DSBs
Considering the functional significance of HITT in regulating the response to DSBs, we next explored the underlying mechanisms of HITT up-regulation in response to chemotherapeutic treatment. Actinomycin D, an RNA synthesis inhibitor, completely abolished Dox-induced HITT expression, suggesting that HITT up-regulation is not due to RNA stabilization (S5A Fig). Interestingly, an approximately 2-to 3-fold increase of HITT luciferase activity was detected after Dox treatment in a dose-dependent manner (Fig 4A), which was similar to the extent of HITT induction, suggesting that the up-regulation of HITT is predominately due to the promoted RNA synthesis.
By searching potential binding motifs of transcription factors located at promoter regions of HITT in the UCSC Genome Browser chromatin immunoprecipitation (ChIP) sequencing (ChIP-seq) database, the most potent transcription factors were Early Growth Response 1 (EGR1) and TATA-box binding protein associated factor 1 (TAF1) (S5B Fig), both of which have been implicated in the DDR previously [44,45]. However, the expression of EGR1, but not TAF1, was found to be elevated by Dox (Figs 4B and S5C); thus, EGR1 was subjected to   (Fig 4B). EGR1 by specific siRNAs abrogated Dox-induced HITT luciferase reporter activity and HITT expression ( Fig 4C). These data support the idea that EGR1 is required for Dox-induced HITT transcription.
We further investigated whether HITT is a direct target of EGR1. First of all, we found that transfection with increasing doses of EGR1-expressing plasmids led to increasing HITT expression (S5D Fig). Two independent siRNAs specifically targeting EGR1 diminished the expression of the target and also reduced HITT expression (S5E Fig). Additionally, consistent with bioinformatics prediction, the MT1 reporter, which did not have the predicted EGR1binding sites, failed to respond to EGR1 expression ( Fig 4D). In contrast, the activity of the MT2 mutant reporter was as effective as the wild-type reporter in response to EGR1 (Fig 4D). A ChIP assay further revealed that EGR1 was associated with the HITT promoter at the corresponding sites, and such interaction was significantly promoted by Dox (Fig 4E). Therefore, HITT is a direct target of EGR1, which is critical for maintaining high levels of HITT after exposure to DNA damage. In line with this notion, siRNA-mediated EGR1 KD abolished Dox-induced HITT expression and facilitated ATM activation (Figs 4F and S5F).
Collectively, EGR1 is required for HITT induction upon DSBs.

HITT down-regulation may contribute to ATM activation in vivo in human colon cancers
Given the significance of HITT in regulating ATM's activity, we further investigated their association in vivo in human colon cancers. HITT levels were commonly decreased in colon cancer tissues to approximately 32% of their paired adjacent normal controls (n = 40 pairs, S6A

HITT sensitizes Dox-induced apoptosis by inhibiting ATM activation both in vitro and in vivo
Many chemotherapeutic drugs remove cancer cells by inducing severe DNA damage. The activation of ATM and the subsequent DNA damage repair can limit the effectiveness of DNA-reporter assay after transfection with the indicated reporter plasmids together with EGR1 constructs. (E) Interaction of EGR1 with the HITT promoter was determined by ChIP assay after treating HeLa cells with or without Dox (1 μg/ml, 6 h). PCR band intensities were quantified using ImageJ and presented in the bar graph (right damaging reagents [32]. Considering the important function of HITT in regulating ATM activity and the DDR, its effects on chemotherapy-induced cell death were explored using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Our results show that cell viability was reduced by Dox in a much more dramatic manner in HITT stable lines than in corresponding vector controls (Fig 5A), whereas HITT KD dramatically alleviated Dox's inhibitory effect on cell viability (Fig 5B). ATMi-1/2 reduced cell viability in the presence of Dox and also completely abolished the impact of HITT (Fig 5A and 5B). Supportively, HITT F3, which is shown to contribute to ATM activity and HR inhibition, produced similar effect to increase Dox-induced cell death (S7A Fig). Under unstressed basal conditions, HITT had no obvious effect on clonal number and size when cells were grown on soft agar, whereas Dox treatment produced a much more dramatic effect on the inhibition of clonal growth in HITT-overexpressing cells than in controls (Fig 5C). Furthermore, HITT significantly elevated apoptosis rates, as indicated by Annexin V-positive staining and cleaved caspase 3/7 after Dox treatment in both HeLa and HCT116 cells (Fig 5D and 5E). The effects of HITT on clonal formation and apoptosis were completely abolished by ATMi (Fig 5C and 5E). KD HITT regulators, EGR1, inhibited Dox-induced apoptosis (Fig 5F and 5G). Similar results were obtained in Bleo-and Eto-treated HCT116 and HeLa cells (S7B- S7E Fig) and Dox-treated p53-null H1299 cells (S7F and S7G Fig). These data collectively suggest that HITT facilitates DNA damaging agent-induced apoptosis by inhibiting ATM in a p53-independent manner. The impact of HITT on the effect of Dox was also evaluated in vivo using a nude mice xenograft model. To avoid big initial tumor volume differences on HITT overexpression and control xenografts before Dox treatment, more HITT-overexpressing HCT116 cells than vector controls were inoculated into nude mice. The mice with similar tumor volumes (about 200 mm 3 ) were subjected to the indicated treatments (n = 6 for each group). Even more HCT116/ HITT cells were injected, with the growth rates of HITT xenografts still slightly lower than those of vector controls (Fig 6A-6C). Dox treatment only slightly reduced tumor growth in the vector control xenografts, and no statistical significance was obtained in comparison with the untreated vector controls (Fig 6A-6C). Remarkably, Dox reduced tumor weight by approximately 4-fold in HITT overexpression xenografts compared with vector xenografts (Fig 6A and 6B). Furthermore, ATMi-1, which has been tested in preclinical [46], produced a mild inhibitory effect on tumor growth (Fig 6A-6C). Consistent with previous reports, the effect of ATMi-1 was more evident when combined with the DNA-damaging agent Dox. Interestingly, no synergistic effect was observed with the combination of HITT overexpression and ATMi-1 (Fig 6A-6C), which further supports the idea that HITT inhibits tumor growth by preventing the activation of ATM. Dox treatment led to a mild decrease of body weight in mice (Fig 6D).
In agreement with the in vitro data, HITT was elevated by Dox in vivo (Fig 6E). Western blot (WB) assays revealed that ATM activity was increased by Dox and was diminished by HITT ( Fig 6F). ATMi-1 produced a similar effect with ectopic HITT expression on ATM activity. Neither HITT expression nor ATMi-1 had an obvious effect on the expression of total ATM and MRN components.

Discussion
DNA repair inhibitors have been developed as clinical agents to reverse radiation or genotoxic drug resistance [32]. Basic research into understanding the regulatory mechanisms underlying DNA damage repair is crucial for the identification of tumor markers, to allow for more-effective targeted cancer treatment.  Here, we identify a novel intrinsic inhibitor of ATM activation that is a key and apical event in initiating DDR cascades. Unlike previous reported ATM regulators, this molecule brake is an RNA, namely HITT, that specifically and physiologically binds with ATM. Upon DSBs, HITT expression is elevated and its association with ATM is increased accordingly, thereby retraining MRN-mediated recruitment of ATM to the DNA damage sites and compromising DNA repair mediated by HR (Fig 7). Two independent ATMi completely abrogates HITT's effect on the DDR program and HR. HITT does not further chemosensitize cells treated with ATMi, supporting the idea that HITT mainly interferes with the same prosurvival signals as ATMi.
Of note, no less important than the activation of the ATM-mediated DDR network is to limit its activation after DNA damage repair. This process may also be complex and highly structured. The mechanisms underlying ATM inhibition have emerged recently. It has been shown that H1.2 and Bridging Integrator 1 (BIN1) attenuate ATM activation by directly interacting with ATM [47] or by indirect mechanisms involving E2F1 [48]. Notably, HITT expression is elevated 1 h after Dox treatment, which is later than ATM activation is detected (5 min). It is reasonable to propose that cells limit ATM from overactivation by enhancing HITT after DNA damage repair. HITT may represent an important brake in controlling the ATMmediated DDR network, which is a key cell fate determinant, particularly upon prolonged exposure to DNA-damaging agents in cancer treatment. We also made an interesting observation that HITT binds with ATM under basal conditions (Fig 7), suggesting that it may provide Under unstressed conditions, EGR1 contributes to the basal expression levels of HITT, which can be directly associated with ATM. In response to DSBs-e.g., following Dox treatment-EGR1 is elevated and then play roles in promoting HITT expression by binding at particular regions of the HITT promoter. Elevated HITT may transiently bind ATM at the HEAT repeats domain at the typical NBS1-binding sites, via bases 1,030-1,247. Such activity of HITT changes the ability of ATM in binding with MRN complex, leading to reduced HR and the promotion of apoptosis. si-HITT-mediated HITT inhibition or HITT down-regulation facilitates ATM recruitment by NBS1 to the sites of DSBs, leading to ATM autophosphorylation and optimal activation followed by effective HR, which prevents server DNA damage and apoptosis. ATM, Ataxia-telangiectasia mutated; Chk2, checkpoint kinase 2; Dox, doxorubicin; DSB, double-strand break; EGR1, Early Growth Response 1; HITT, HIF-1α inhibitor at translation level; HR, homologous recombination; Mre11, Meiotic Recombination 11; MRN, MRE11-RAD50-NBS1; N.S., not significant; NBS1, Nijmegen Breakage Syndrome 1; si-, small interfering.
https://doi.org/10.1371/journal.pbio.3000666.g007 quality control to counteract toxic HR. Indeed, a deficiency in ATM inhibition has been associated with carcinogenesis, resistance to DNA damage-induced cancer treatment, and loss of mitotic checkpoint [49][50][51]. It is reasonable to propose that the expression of HITT may alleviate and loss of HITT expression may reinforce such effects. In addition, it has become apparent that ATM is active in response to multiple conditions involved in the maintenance of cellular homeostasis [7]. It would be interesting to investigate the physiological significance of the newly identified HITT-ATM interaction in a DNA damage-independent context. MRN is a sensor for cells to recognize DSBs [52]. Recruitment and optimal activation of ATM is mainly dependent on its interaction with the MRN component NBS1 [43]. Consistent with this notion, the mutation of ATM-binding sites in NBS1 completely abrogates ATM accumulation at DSB sites [43]. Interestingly, the regulation of ATM by HITT is MRN(NBS1)dependent. We further defined the ATM site that facilitates HITT binding to a previously identified NBS1-binding site (the HEAT repeats domain), raising the possibility that HITT competes with NBS1 in binding with ATM. Indeed, ectopic HITT expression led to increased association of HITT/ATM and reduced association of NBS1/ATM, without influencing total ATM or NBS1 protein levels. Also notably, NBS1 is essential to induce ATM monomerization and activation [13,53]. HITT and NBS1 bind with ATM at the same sites, although eliciting opposite effects on ATM activity. It remains to be investigated whether HITT plays roles in influencing ATM monomerization and, if so, whether such activity of HITT is dependent on its inhibitory effect on NBS1/ATM interaction.
Notably, lncRNA is normally expressed at lower levels than protein-coding genes [54]. One potential explanation for the robust effect of HITT on ATM activity is that HITT may transiently interact with ATM to modulate ATM modification to render it inactive. Indeed, it has been reported that ATM modification may influence its activation [20,55]. In addition, although HITT itself may have no catalytic activity, its effect on ATM downstream signaling can be amplified and is not necessarily linearly associated with its expression levels. It will be interesting to test the above models in future studies.
We also suggest that HITT's inhibition of ATM may be of clinical significance. A small peptide containing the evolutionarily conserved region of NBS1, which binds with ATM, disrupts the interaction between NBS1 and ATM. This peptide also prevents DNA damage signaling and radiosensitization of cells, underscoring the notion that blocking the NBS1/ATM interaction is a potential target for anticancer treatment [56]. Indeed, we found that HITT-mediated ATM inhibition leads to increased cell death when cells are treated with Dox both in vitro and in vivo. Because of its significant effects regarding improving chemosensitivity, our data encourage the future development of lncRNA-based cancer therapies for patients resistant to genotoxic treatments.
Furthermore, p53 is an important gene in regulating genome stability and the DDR [57]. However, we made an interesting observation that the induction of HITT upon DSB-and HITT-dependent DDR activity is p53-independent. This is important, as p53 is the most commonly mutated tumor suppressor gene in human cancers [58]. The discovery of a p53-independent function of HITT implies that HITT could be utilized to sensitize cells to cytotoxic drug treatments that are applicable to a wide range of human cancers, regardless of their p53 status.
Moreover, HITT expression is commonly decreased, whereas p-ATM levels are increased, in colon cancer tissues. Despite being elevated in a subset of cancer samples, NBS1 overexpression is not likely to be important for ATM activation, which suggests that additional layers of regulation exist for NBS1-mediated ATM activation. Importantly, the extent of the reduction of HITT expression is significantly associated with increasing p-ATM, which suggests that HITT down-regulation may contribute to the activation of p-ATM in vivo. This is important, because ATM activation has been associated with treatment failure in multiple cancers. Detection of ATM variance or its protein expression has been suggested to be a potential predictive marker for drug response [5], and our findings provide an effective marker, HITT, for the prediction of ATM activity. In addition, BRCA1-and BRCA2-mutated cells are defective in HR repair and are thus highly sensitive to poly ADP-ribose polymerase (PARP) inhibitor treatment [12]. In addition, ATM deficiency has been shown to increase sensitivity to PARP and ATR inhibitors [59]. HITT specifically inhibits the ATM-dependent HR pathway. The synergistic effects of HITT and PARP inhibitors warrant future investigation. ATMi has also been shown to be a promising single agent for the treatment of human cancers with defects in particular oncogene signals, such as a loss of PTEN (phosphate and tension homology deleted on chromosome ten) [60]. Further work is thus necessary to decipher whether HITT could be applied alone in subtypes of cancers.
We have also found that Dox-induced HITT up-regulation is predominately controlled by EGR1. EGR1 can be rapidly up-regulated by a wide variety of extracellular stimuli, including the activation of growth and differentiation signals, tissue injury, and apoptotic signals, such as IR. Despite its transient activation, the expression of EGR1 is likely required for HITT up-regulation, since EGR1 KD completely abolishes Dox-induced HITT. These data are consistent with the fact that HITT is involved in regulating the initiation steps of the DDR. In addition, EGR1 KD is accompanied by impaired DDR and p53 activation [61]. Here, we provide the first evidence that EGR1 exerts its effect on the DDR by modulating lncRNA expression. However, HITT is maintained at high levels, even when EGR1 expression drops back down to unstressed levels, suggesting that additional transcriptional regulation of HITT also exists. Indeed, a number of additional transcription factors have been indicated to bind with HITT promoter as shown in UCSC Genome Browser ChIP-seq database. It will be interesting to investigate the impacts of these candidate transcription factors on HITT expression in the context of DSB.
Altogether, our results demonstrate that HITT plays essential roles in inhibiting the HR pathway by restraining ATM activity and thus may be an effective adjuvant therapy for genotoxic or radiomimetic compounds or a predictive marker for treatment efficiency in cancer.

Colorectal cancer tissues samples
The specimens were collected and stored in liquid nitrogen immediately after surgery. The total proteins and RNAs were extracted and then subjected to WB and real-time RT-PCR analysis.

In vivo xenograft mouse study
HITT expression was stably restored in HCT116 cells, by using empty vector as a control. The female nude mice between 4 and 5 wk old were purchased from Beijing HFK Bioscience. To avoid the initial big volume difference between the HITT overexpression and control groups, the 1 × 10 7 vector cells and 1.5 × 10 7 HITT stable HCT116 cells were inoculated subcutaneously. The tumor volumes were measured every week and calculated as length × width 2 × 0.5. When the tumor volume reached about 200 mm 3 , the mice were randomly divided into four groups and subjected to the treatment of DMSO and Dox alone (10 mg/kg, intraperitoneal) once per week for 2 wk or in combination with ATMi-1 (60 mg/kg, intraperitoneal) daily for 5 d (n = 6 for each group). The tumor size and the body weights of the mice were measured every 2 d. After treatment for 2 wk, the mice were anesthetized and culled. The tumor was carefully removed, photographed, and weighed.

IP
Cells were lysed in NETN buffer (50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 1% NP-40, 1 mM EDTA), with Proteinase Inhibitor Cocktail (MedChemExpress, #HY-K0010) added freshly before use. After being precleaned by protein G sepharose beads 4 Fast Flow (GE Healthcare, #17061802), specific antibodies or control IgG was added to the supernatant, which was incubated with FBS blocked beads for at least 20 h at 4˚C. Beads with the bound immunoprecipitates were collected following four washes with cold NETN. The subsequent immunoprecipitates were extracted for WB assay.

Luciferase reporter assay
After the indicated transfection, the luciferase activities were detected with the luciferase assay system (Promega, #E1910) according to the manufacture's introduction. The relative luciferase activities were normalized with the Renilla luciferase activities.

Apoptosis assay
After the indicated treatments, both suspended and attached cells were collected. Cell suspension in binding buffer were incubated with 5 μL Annexin V/FITC for 10 min and then with 5 μL propidium iodide (PI) (Sungene) for 5 min at room temperature in the dark. The rate of apoptosis was determined by flow cytometry.

Caspase3/7 activity assay
Following the indicated treatments, cells were subjected to the caspase 3/7 activity assay by Caspase-Glo_3/7 Assay Systems (Promega, #G8091) according to the manufacturer's instructions. The assay was conducted in triplicates and repeated independently for three times, which was represented as a fold increase of fluorescence calculated by comparing cells with untreated control cells.

Comet assays
Following the indicated treatments, comet assays with EB (Sigma) staining were performed as reported previously [62]. The quantification of comet rate and tail moment were performed with CASP software (http://www.casp.of.pl).

NHEJ assay and HR assay
DR-U2OS, EJ2-U2OS, and EJ5-U2OS cells were transfected with HITT expression plasmids or two independent siRNA oligos specifically targeted HITT, together with ISce-I plasmid.
Forty-eight to 72 h after transfection, cells were harvested and resuspended in 0.5 ml of PBS (pH 7.4). GFP signal was analyzed by flow cytometry (FACS).

ChIP
Cells were incubated with formaldehyde to yield protein-DNA cross-link complexes, which were purified and sheared by sonication. The chromatin was divided equally into two groups for further IP reaction with anti-EGR1 antibody or IgG control. The immunoprecipitates were pelleted by centrifugation and then incubated at 65˚C to reverse the protein-DNA cross-linking. The DNA was extracted by the Axygen product purification kit and subjected to PCR analysis.

CLIP
UV-irradiated cells were collected in lysis buffer (5 mM PIPES [pH 8.0], 85 mM KCl, 0.5% NP40 and 1% SDS, 10 mM EDTA, 50 mM Tris-HCl [pH 8.1]) and supplemented with Protease Inhibitor Cocktail and RNase inhibitor (Thermo Fisher). The cell lysates were precleaned with protein G sepharose beads and then incubated with the indicated antibodies or IgG control, rotating at 4˚C overnight. The antibody-RNA complexes were collected. The immunoprecipitated RNA was eluted and extracted for further real-time RT-PCR analysis.

In vitro RNA pull-down assay
Biotin-labeled HITT and its antisense were in vitro synthesized by Biotin RNA Labeling Mix (Roche, 11685597910). After treatment with RNase-free DNase I, secondary structure of Biotin-labeled RNA was recovered and incubated with streptavidin agarose beads (Invitrogen) overnight. The fresh cell lysates or in vitro-translated protein (TNT T7 Quick coupled transcription/translation system, Promega) were collected and incubated with RNA-captured beads at 4˚C for 1 h. The associated proteins were detected by WB.

Clonogenic survival assay
HITT stable HeLa cells were treated with Dox (1 μg/ml) with or without Ku60019 (10 μM) or Ku55933 (10 μM) for 24 h. The untreated and the treated cells were trypsinized and subjected to a typical clonogenic survival assay. A total of 5 × 10 4 cells were mixed with 2×DMEM medium with 20% FBS containing 0.7% agar and then spread on the top of a bottom agar layer (1% agar in drug-free DMEM full growth medium) in a 6-well plate. DMEM medium (2 mL) was added and refreshed every 3-4 d. Cells were grown for 1 mo. Colonies were counted and photographed after stained with 1% Crystal Violet. The colony numbers and relative colony size were analyzed by ImageJ.

Chromatin fraction
Cells were fractionated as previously described [63]. As brief, cells were resuspended in buffer A (10 mM HEPES [pH 7.9], 10 mM KCl, 1.5 mM MgCl 2 , 0.34 M sucrose, 10% glycerol, 1 mM DTT). Triton X-100 (0.1%) was added and incubated on ice for 5 min. Cell lysates were centrifuged at 1,300g for 4 min, and the remaining pellet (enriched with nuclei) was washed with buffer A (10 mM HEPES [pH 7.9], 10 mM KCl, 1.5 mM MgCl 2 , 0.34 M sucrose, 10% glycerol, 1 mM DTT) once and then lysed in buffer B (3 mM EDTA, 0.2 mM EGTA, 1 mM DTT) on ice for 5 min. The nuclei lysate was then centrifuged at 1,700g for 4 min and the supernatant (soluble nuclear fraction) was removed. The final pellet is the chromatin fraction that is ready for further analysis.

MTT assay
Cell viabilities after the treatment were assessed by the colorimetric MTT (Sigma-Aldrich) assay. The absorbance was measured with a spectrometer at 490 nm. Each experiment was conducted in triplicates and repeated independently for three times.

Cell-cycle synchronization assay
To synchronize the cell cultures, HeLa cells were exposed to 2.5 mM TdR (Sigma) for 16 h (first block) and then placed on fresh medium supplemented with 10% FBS for 12 h (first release). After that, cells were subjected to another round of TdR treatment (second block) before incubation with fresh medium for an additional 0, 6, 8, and 10.5 h, respectively. The cells were collected at different time points for cell-cycle analysis by flow cytometer, and HITT expression was measured by real-time RT-PCR.

Immunofluorescence staining and BrdU incorporation assay
Cells grown on cover slips in a 24-well plate were fixed in 4% paraformaldehyde for 20 min and then treated with 0.1% Triton X-100 solution on ice for 4 min. Cells were then blocked by 3% BSA for 1 h followed by the antibody incubation at 4˚C overnight. After that, cells were washed 3 × 5 min in PBS and then incubated with the fluorescently labeled secondary antibody for 1 h. DAPI was stained for 3 min for visualizing the nucleus. The slices were mounted by 90% glycerinum and images were captured by a Zeiss Axio Observer confocal microscope.

Trypan blue staining
After different drug treatment for 24 h, both survival and dead cells were collected and subjected to trypan blue staining and counted by hemocytometer. The percentage of dead cells (stained)/total cells was determined by counting an average of 200-500 total cells, three times for each sample. The experiments were repeated independently three times.