Radiation activates the NORAD expression to promote ESCC radioimmunotherapy resistance via EEPD1/ATR/Chk1 signaling by inhibiting the pri-miR-199 processing and exosomal transfer of miR-199a-5p


 Background Radioresistance, a poorly understood phenomenon, results in the failure of radiotherapy and consequent local recurrence, threatening a large proportion of ESCC patients. To date, lncRNAs have been found to be involved in diverse biological processes, including radioresistance.Methods ELISA was used to evaluated the H3 modifications in radio-resistant ESCC cells. FISH and qRT-PCR were adopted to examine the expression and localization of lncRNA-NORAD, pri-miR-199a and miR-199a. Electron microscopy and Nanoparticle tracking analysis (NTA) was conducted to observe and identify exosomes. High-throughput RNA sequencing and TMT mass spectrometry were performed to identify the functional lncRNAs and proteins involved in ESCC radioresistance. A series of in vitro and in vivo experiments were performed to investigate the biological effect of NORAD. CHIP, qPCR-RIP, co-IP and dual-luciferase reporter assays were used to explore the interaction of related RNAs and proteins. Results We show here that a DNA damage activated non-coding RNA-NORAD, which is critical for ESCC radio-resistance. NORAD was highly expressed in radio-resistant ESCC cells and tissues. Irradiation treatment promotes NORAD expression via enhancing H3K4me2 enrichment on its region. NORAD knockdown cells exhibit significantly hypersensitivity to irradiation in vivo and in vitro. NORAD is required for initiating repair and restart of stalled forks, G2 cycle arrest and homologous recombination repair upon irradiation treatment. Mechanistically, NORAD inhibits miR-199a expression by competitively binding PUM1 from pri-miR-199a, inhibiting the process of pri-miR-199a. Mature miR-199a in NORAD-knockdown cells can be packaged into exosomes; miR-199a restores the radiosensitivity of radioresistant cells by targeting EEPD1, then inhibiting ATR/Chk1 signaling pathway. Simultaneously, NORAD knockdown blocks the ubiquitination of PD-L1, leads to the better response for radiation and anti-PD-1 treatment in mouse model.Conclusion This study raises the possibility that LncRNA-NORAD could be a potential treatment target for improving the efficiency of immunotherapy in combination with radiation in ESCC.

Radiation activates the NORAD expression to promote ESCC radioimmunotherapy resistance via EEPD1/ATR/Chk1 signaling by inhibiting the pri-miR-199 processing and exosomal transfer of miR-199a-5p  [4,7], partially due to radiation or chemotherapy resistance. Although terri c progress has been made in identifying novel biomarkers and therapeutic targets for improving radiation sensitivity, the molecular mechanism underlying radioresistance remains ambiguous and complex; it is thought to involve cell cycle checkpoints that prevent cancer cells from sustaining radiation-induced DNA damage, activation of the DNA damage response, the self-renewal of cancer stem cells, epithelial-mesenchymal transition(EMT), etc. [7,8]. In this manuscript, we aimed to identify key genes that participate in sensitizing ESCC cells to radiation therapy and to lay the foundation for drug development.
In recent years, long noncoding RNAs (lncRNAs) have attracted great attention as a result of their diverse biological functions in the development and progression of multiple cancers [9]. To explore key lncRNAs that are involved in ESCC radioresistance, we performed RNA-sequencing analyses to identify differentially expressed lncRNAs between radioresistant ESCC cells and control cells. Combined with our previous studies, the DNA damage-activated noncoding RNA-NORAD (also named linc00657) garnered our attention. NORAD is abundantly expressed in multiple eukaryotic cells, is conserved across mammalian species and has been shown to play oncogenic roles in numerous cancers. NORAD maintains genomic stability by sequestering and negatively regulating PUMILIO proteins through its 18 conserved PUMILIO response elements [10]. Another study reported that NORAD interacted with DNA damage-repair-and DNA replication-related proteins, assembling a topoisomerase complex to sustain genomic stability in response to DNA damage [11]. In addition, NORAD binds to a group of microRNAs, modulating their abundance to exert oncogenic functions [12] [13]. In our previous study, we identi ed that NORAD knockdown signi cantly sensitized ESCC cells to radiation. However, how NORAD regulates radiobiological processes in ESCC is unclear.
The main anticancer mechanism of radiation is the generation of a cluster of lethal lesions, which in turn induce DNA damage in cells and tissues [14] Damaged DNA can be repaired by the activation of homologous recombination (HR) and nonhomologous end joining (NHEJ) pathways in vitro [15]. The crosstalk between DNA damage repair and radiosensitivity is obvious [16]; our previous studies showed that NORAD expression can be induced by radiation treatment in ESCC cells. Furthermore, some studies have found that inhibiting DNA damage repair enhances the e ciency of immune checkpoint therapy, especially when combined with radiotherapy. Frank P et al reported that the ATR kinase inhibitor AZD6738 reduced the exhaustion of CD8+ T cells induced by radiation. In addition, AZD6738 combined with radiation can generate immunologic memory for tumors in mice [17]. The latest study found that DNA double-strand breaks promoted PD-L1 expression by activating STAT1 and STAT3 signaling [18].
TMT-based proteomics analyses also suggest the potential function of NORAD in regulating the immune response. However, how suppressing DNA damage repair responses can amplify immune checkpoint therapy e ciency in combination with radiotherapy remains elusive. In this context, we hypothesize that NORAD modulates ESCC radiosensitivity by regulating the DNA damage repair process. We assumed that NORAD is involved in the synergy between radiotherapy and immune checkpoint therapy.

Cell culture, Lentivirus infection and transfection
The and TE-1 cells were seeded into 48-well plates and grown to 50% con uence. Lentivirus (1×10 8 TU/ml) and infection reagents were mixed and added to the cells. The medium was changed after 48 hours of infection, and puromycin (3 µg/ml) was added to the medium to select cells that were successfully infected. Transient knockdown or overexpression of candidate genes was achieved by transfection of relative siRNAs or gene overexpression plasmid. Transient transfections were performed by using Lipofectamine 3000(Invitrogen/Thermo Fisher Scienti c) according to standard protocol.

qRT-PCR
Total RNA was extracted from KYSE-150 and TE-1 cells using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and then reverse transcribed into cDNA using a PrimeScript™ RT reagent kit (TaKaRa, Dalian, China). MicroRNA-speci c cDNAs were generated by using an EvoM-MLV RT kit. qRT-PCR was carried out using SYBR Premix Ex TaqTM II (TaKaTa, Dalian, China). GAPDH and U6 were used as reference genes for mRNA and microRNA, respectively.

Western blot
Total protein from KYSE-150 and TE-1 cells was extracted by RIPA buffer (Sigma Aldrich, Cambridge, MA) and quanti ed by BCA (Sigma Aldrich, Cambridge, MA). Then, proteins were subjected to SDS-PAGE through 10% gels and transferred onto PVDF membranes, which were then incubated with primary antibodies at 4℃ overnight. Secondary antibodies were then incubated with the membranes for 1 hour at room temperature before the protein bands were visualized using an ECL kit.

Extraction and identi cation of exosomes
Electron microscopy was used to observe and identify the extracellular vesicle-like structure of exosomes.
Nanoparticle tracking analysis (NTA) was conducted to test the particle size-concentration distribution on a ZetaView PMX 110 (Particel Metrix, Germany). CD63 was considered an exosome marker, and protein expression was evaluated by western blot. For the exosome tracer experiment, 2000 cells/well were seeded in a 96-well plate; the exosome suspension was incubated with 2 µM PKH26 for 5 minutes at room temperature and added to the target cells in a volume of 20 µl/well. The red uorescence indicates the process of exosomes entering target cells.

Fluorescence in situ hybridization
We used KYSE-150 and TE-1 to produce cell slides, which were pretreated with HCl and then xed with neutral formalin. A probe targeting NORAD was used for hybridization with cell slides after treatment with diluted prehybridization solution. DAPI was used to stain nuclei, and cell slides were observed by confocal microscopy.
6. Immuno uorescence KYSE-150 and TE-1 cells were seeded on glass sides in 48-well plates and cultured overnight. Cells on the slides were xed with 4% paraformaldehyde before they were treated with Triton X-100 (5%) to permeate cells and BSA solution to block nonspeci c binding. Cells were treated with primary antibodies and incubated at 4°C overnight. After unbound primary antibody was removed, the cells were then incubated with PE-conjugated secondary antibody for 1 hour at 37 ℃. DAPI was added to the slides for nuclear staining, and glycerin was used to block the slides. Images of the slides under a confocal microscope were obtained. The primary antibodies and PE-conjugated secondary antibody were purchased from Abcam.

co-IP
Cells were lysed at 4°C for 5 min in RIPA buffer, containing protease inhibitors. Whole cell lysates were then precleared with protein A/G beads. PD-L1 antibody was then added overnight at 4 ℃ . The antibody/antigen complex was pulled out of the lysates by using protein A/G-coupled agarose beads. Beads were washed by RIPA buffer for 3 times and resuspended by 2loading buffer. The Western blot and assays were described in previous.

RNA immunoprecipitation (RIP)
We purchased a Magna RIP™ RNA-binding Protein Immunoprecipitation Kit (Millipore, Billerica, MA, USA) to analyze the interaction between RNA and proteins. KYSE-150 cells were lysed in RIP lysis buffer for further experiments. We used Ago2 antibody for immunoprecipitation and IgG antibody as the negative control. The expression levels of NORAD and pri-miR-199a were evaluated by qRT-PCR. 9. Homologous recombination (HR) reporter assay HR reporter and I-SceI expression plasmids were purchased from Genechem (Shanghai, China). For the HR reporter assay, KYSE-150-sh-NC and KYSE-150-sh-NORAD cells were transfected with the HR reporter.
G418 (1 mg/ml) was used to select successfully transfected cells. Then, these cells were transfected with the I-Scel plasmid to induce double-strand breaks on the HR reporter plasmid. Cells were harvested for analysis after 48 hours. Cells that underwent HR repair are positive for green uorescence positive. Lentiviruses containing sh-NORAD and sh-NC used here were not uorescently labeled.

Apoptosis and cell cycle analyses
We used an Annexin V-APC 7-AAD Apoptosis Detection Kit I (BD Pharmingen TM, New Jersey, USA) for the apoptosis assay. NORAD knockdown and NC cells were harvested and resuspended (5×10 5 cells per sample) before 5 μl of Annexin V-APC and 7-AAD were added to the suspensions and incubated for 15 minutes at room temperature in the dark. Flow cytometry was used to evaluate the luciferase intensity of Annexin V-APC and 7-AAD. A Cell Cycle Staining Kit (BD Pharmingen TM, New Jersey, USA) was used for evaluating the cell cycle distribution. Cells were xed with 70% ethanol for 2 hours at 4°C. Target cells were treated with DNA staining solution and permeabilization solution in darkness for 15 minutes at room temperature. Flow cytometry was used to evaluate the luciferase intensity at 24 hours after treatment.

Replication fork recovery
Immuno uorescence assays of BrdU foci (BrdU in double-stranded DNA) were used to evaluate replication fork recovery after exposing ESCC cells to radiation [19,20]. BrdU was incorporated into double-stranded DNA, and the immuno uorescence intensity of BrdU foci represented activation of replication fork recovery. Cells were exposed to 8 Gy of radiation before BrdU was added to the culture medium at a nal concentration of 10 µM and incubated with the cells for 30 minutes. Cells with active replication forks were indicated by the cells with positive BrdU foci under a uorescence microscope.
Cells that did not receive radiation treatment were used as the control group.

Colony formation assay
Cells (500, 1000, 2000, 4000, or 8000 per well) were seeded into 6-well plates and cultured overnight in a 5% CO2 incubator before they were exposed to 0, 2, 4, 6, or 8 Gy of X-ray radiation. Colonies containing more than 50 cells were counted after 2 weeks of incubation. The survival curve after radiation was tted based on the single-hit multitarget model: SF = 1 − (1 − e-D/D0)n.

Tandem mass tag proteomic-based quantitative proteome analysis
Cells successfully infected with lentivirus containing sh-NORAD or sh-NC were sonicated in lysis buffer (8 M urea, 1% Protease Inhibitor Cocktail) to extract total protein. The protein sample was pretreated with dithiothreitol, alkylated with iodoacetamide and diluted with TEAB. Then, the protein mixture was incubated with trypsin, desalted on a Strata X C18 SPE column (Phenomenex) and vacuum-dried. A tandem mass tag (TMT) kit was used for further labeling. The peptides were thereafter evaluated by tandem mass spectrometry (MS/MS) on a Q ExactiveTM Plus (Thermo) coupled to the UHPLC system.
The data-dependent procedure alternated between one MS scan followed by 20 MS/MS scans with 15.0 s dynamic exclusion and an automatic gain control (AGC) of 5E4. The xed rst mass was set as 100 m/z.
After the tumor volume reached 50-100 mm 3 in the KYSE-150 model and 150 mm 3 in the AKR model, local radiation (2 Gy/day for 4 consecutive days) was administered to the tumor site. Mice receiving concurrent treatment were also injected with anti-PD-1 antibody at 0, 7, and 14 days after the rst radiation treatment. The tumor volume was measured with calipers and calculated using the following formula: tumor volume=0.5´ width 2 × length.

Whole exome sequencing
Agilent V6 Exon + UTR region probe was used. Primitive paired-end reads were screened to generate clean reads. Then BWA was used to compare the clean reads to H19 and rearrange the clean reads according to Karyotype by Picard. Then Samtools and Picard were used for redundancy removal. After that, GATK was used to generate the SNPs/INDELs variation information, which was annotated by VEP. Tumor mutation burden (TMB) was then calculated (TMB= total somatic mutations/total length of target region).

Statistical analysis
GraphPad Prism 8.2.1 and R 3.3.1 were used for statistical analysis and data visualization. Student's t test was performed to test differences between two groups, and one-way ANOVA was used to test differences among multiple groups. P 0.05 was considered statistically signi cant.

High NORAD expression indicates ESCC radioresistance.
To explore the NORAD's mechanism in ESCC radio-resistance, we established the acquired radioresistant ESCC cells (termed as KYSE-150R and TE-1R) as described in method 2. Colony formation assay con rmed that radioresistant ESCC cells were constructed successfully ( Fig.1A-D). The qRT-PCR results showed that NORAD expression was signi cantly upregulated in KYSE-150R and TE-1R cells compared with KYSE-150 and TE-1 cells ( Fig. 2A). Following irradiation and other genotoxic agents (doxorubicin and cisplatin), NORAD expression was approximately 2~3-fold higher in ESCC cells (Fig. 2B). qRT-PCR and Immuno uorescence (IF) results con rmed that cytoplasmic NORAD but not nuclear NORAD were signi cantly increased in response to irradiation (Fig. 2C-E). FISH results from 77 ESCC patients who had been treated with de nitive radiation indicated that NORAD was mainly expressed in the cytoplasm (Fig.  2F); thus, NORAD expression was subsequently examined in 41 ESCC tissues by qRT-PCR. High NORAD expression was associated with local recurrence after radiotherapy in ESCC patients (Fig. 2G-H). ROC curves indicated that NORAD can strongly predict local recurrence for ESCC patients who were treated with radiotherapy (Fig. 2I).

Irradiation activates NORAD expression via enhancing H3K4me2 enrichment
Histone H3 modi cations of radioresistant ESCC cells and their parental cells were measured by using the EpiQuik™ Histone H3 Modi cation Multiplex Assay Kit. We observed that H3K4me3, H3K27me3 and H3k9ac were decreased while H3K4me2 showed increased expression in radioresistant ESCC cells ( Fig.  3A). By using genome bioinformatics analysis (http://genome.ucsc.edu/), we found that NORAD had high enrichment of H3K4me2 in MCF-7, HCT-116, SK-N-SH and KMS-11 cancer cells ( Supplementary Fig.  1A). Western-blot analyses results showed a dynamic change of H3K4me following irradiation treatment, reaching a peak at 24 h (Fig. 3B). We next performed the ChIP assay in radioresistant ESCC cells and found that H3K4me2 was enriched at NORAD region. Notably, the enriched intensity of H3K4me2 was enhanced in radio-resistant cells compared with the parental cells (Fig. 3C). Based on TCGA datasets, we found the H3K4 methyltransferases, including Ash1, KMT2A, KMT2B, KMT2C, KMT2D, KMT2E, Set1A, Set1B, SETD7, SMYD2, SMYD3, WDR5 were positively correlated with NORAD expression ( Supplementary   Fig.1B). ESCC patients with KMT2B, KMT2C or KMT2D muted showed slightly lower NORAD expression but there was no statistic difference (Fig.3D). Knockdown ASHL2 (H3K4 methyltransferases) resulted in dereased H3K4me2 and NORAD expression (Fig.3E); and impairing the up-regulation of NORAD upon irradiation (Fig.3F). Together, these results showed that irradiation induced NORAD high expression were due to enhancing H3K4me2 enrichment at the NORAD region.

NORAD knockdown sensitizes ESCC to irradiation in vitro and in vivo by inhibiting homologous recombination repair
We next knockdown NORAD in KYSE-150R and TE-1R cells by lentivirus. Flow cytometry-based apoptosis assays demonstrated that NORAD knockdown alone increased the apoptosis rates of KYSE-150 and TE-1 cells and markedly increased the apoptosis rates in irradiated cells (Fig.4A-B). Xenograft tumor from NORAD knockdown and control cells showed that NORAD knockdown sensitized ESCC to radiation signi cantly in vivo ( Fig.4C-D). BrdU immuno uorescence assays were adopted to measure replication fork restart after stalling upon DNA damage. Nascent DNA after DNA damage could be labeled with BrdU. Fewer BrdU-positive cells were observed in the NORAD knockdown group, suggesting that NORAD knockdown delayed the recovery of the stalled replication fork after radiation treatment ( Fig. 4E-F).
Regarding cell cycle distribution, NORAD knockdown arrested ESCC cells at G1 phase. We exposed ESCC cells to 8 Gy of radiation and found a signi cant increase in the number of cells arrested in G2 phase, whereas NORAD knockdown in combination with radiation attenuated G2 phase arrest (Fig. 4G-H). To determine if NORAD regulates DNA repair, recombination DR-GFP plasmids were transfected into cells with and without NORAD knockdown. DR-GFP plasmid-based HR assay showed that NORAD knockdown led to a 3~4-fold reduction in homologous recombination repair upon DSBs (Fig4.I-J).

NORAD knockdown-derived exosomes sensitizes ESCC cells to irradiation
We co-cultured NORAD knockdown cells with radioresistant ESCC cells, colony formation assay upon 2 Gy treatment results showed that NORAD knockdown not only sensitized cells themselves to radiation, but also sensitized KYSE-150R and TE-1R which were co-cultured with NORAD knockdown cells to radiation. It is notable that this inhibitory effect could be reversed by GW4869 (Fig.5A-B). We speculated that NORAD knockdown cells might affect co-cultured cells via exosomes. We then extracted exosomes from NORAD-knockdown cells, which were termed sh-NORAD exosomes. The extracellular vesicle-like structure of exosomes derived from sh-NORAD cells was con rmed by electron microscopy (Fig.5C); NTA (Nano Track Analysis) was also used to assess the particle size and relative number of particles in combination with CD63 expression to con rm their identity as exosomes (Fig. 5D-F). Exosome tracer experiments showed that sh-NORAD exosomes began to enter KYSE-150R and TE-1R cells after 6 hours of incubation and were enriched in the cytoplasm after 24 hours (Fig.5F). Decreased colony counts were observed as well for KYSE-150R and TE-1R cells treated with sh-NORAD exosomes in response to 2 Gy radiation ( Fig.5G-H).

NORAD knockdown impairs homologous recombination repair by down-regulating EEPD1.
To gain further insight into the potential proteins regulated by NORAD, we performed TMT mass spectrometry to quantify the protein levels in KYSE-150-sh-nc and KYSE-150-sh-NORAD cells (Supplementary Fig.2A-B). KEGG analysis of the downregulated protein in NORAD knockdown cells, demonstrating these proteins were enriched for nervous system disease such as huntington disease, parkinson disease, rheumatoid arthritis, cell cycle process, oxidative phosphorylation, immune system process and so on ( Supplementary Fig.2C-D). Simutanously, starbase dataset was adopted to predicred the potential targets of miR-199a-5p. We found that EEPD1, the gatekeeper for repair of stressed replication forks, which was up-regulated in radio-resistant cells and signi cantly down-regulated in NORAD knockdown cells ( Supplementary Fig.2E-F Fig.8A). IHC staining of xenograft tumors showed that tumors derived from NORAD-knockdown cells exhibited less EEPD1 staining (Fig.8B). EEPD1 w as downregulated after transducing radioresistant cells with miR-199a-5p mimics and upregulated after transducing normal control cells with miR-199a-5p inhibitors (Fig. 8C). miR-199a-5p mimics signi cantly decreased the relative luciferase activity of EEPD1-Wt but failed to in uence the relative luciferase activity of EEPD1-Mut, indicating that miR-199a-5p interacts with the 3′UTR of EEPD1 mRNA (Fig. 8D). EEPD1 expression in 77 ESCC patients was further evaluated by IHC. EEPD1 is mainly expressed in the nucleus (Fig.8E). Among patients with radioresistant ESCC, 20 (20/29) showed high EEPD1 expression. The c 2 examination results showed that the radioresistance characteristic in ESCC was signi cantly associated with high EEPD1 expression (χ 2 =8.151; p=0.004**; Table 1). Based on TCGA datasets, we next identi ed radioresistance-related genes in ESCC by using random forest algorithm; the IncMSC and IncNodePurity values of EEPD1 were obviously higher than those of the other genes (Fig.8F). Furtherly, based on EEPD1 expression in TCGA datasets, the values of the area under the ROC curve (AUC) for segregating radioresistant patients from radio-sensitive patients were 0.889 (95%CI 0.729-1.000) (Fig.8G). Xenograft tumor from EEPD1 knockdown and control cells showed that EEPD1 knockdown sensitized ESCC to radiation signi cantly in vitro and in vivo (Fig.8H). The HR reporter assay showed that EEPD1-knockdown cells presented lower HR rates after DNA double-strand breaks as well (Fig.8I). We found that EEPD1 knockdown inhibits the ATR and Chk1 phosphorylation upon irradiation treatment, which were consistent with NORAD knockdown cells' phenotype( Fig.8J-K). And EEPD1 overexpression restored the phosphorylation of ATR-Chk1, which was reduced in NORAD-knockdown cells after 8 Gy MV X-rays treatment (Fig. 8L).

High NORAD expression predicts radio-immunotherapy failure in vivo.
Considering that irradiation leaded to the severe DNA damage in cells and NORAD knockdown increased the genomic instability under DNA damage, we reasoned that NORAD knockdown increased the tumor mutation burden (TMB) load in ESCC upon irradiation. Based on xenografts tumor model, we found that radiation or NORAD knockdown alone did not change the TMB. But NORAD knockdown combined radiotherapy increased the TMB load following radiation treatment (Fig.9A-C). Previous studies suggested that tumor mutational burden (TMB) may predict clinical response to immune checkpoint inhibitor, we investigated if NORAD knockdown combined with radiation enhanced the anti-PD-1 e ciency on C57BL/6 mice. AKR-radioresistant cells were construced by exposing cells to 2 Gy irradiation every 2 days(termed as AKR-R).We found that Anti-PD-1 monotherapy was not enough for controlling the tumor growth. In combination therapy group, we found that radiotherapy combined with anti-PD-1 therapy suppressed the tumor growth e ciently only in NORAD knockdown group but not control group. (Fig.9D-E). We next examined PD-L1 expression in NORAD-knockdown cells. NORADknockdown cells displayed elevated PD-L1 protein expression (Fig. 9F), and IHC staining of xenograft tumors showed that tumors derived from NORAD-knockdown cells exhibited stronger PD-L1 staining (Fig.  9G) than did tumors derived from control cells. We treated cells with the proteasome inhibitor MG132 and subsequently Co-IP experiments indicated that the levels of PD-L1 ubiquitination was elevated in radioresistant ESCC cells. NORAD knockdown impaired PD-L1 ubiquitination (Fig9.H). These results demonstrated that NORAD inhibits ubiquitination of PD-L1. By subsequently evaluating the tumor in ltrating lymphocyte, however, we did not nd signi cantly change for tumor in ltrating lymphocyte between NORAD knockdown mice group and control group( Supplementary Fig.3A). Based on CIBERSORT, we found that NORAD expression did not affect the immune cell in ltration either( Supplementary Fig.3B).
It is demonstrated that NORAD in uenced the immunogenic function of radiation but not the tumor immune microenvironment. Taken together, these results uncover that NORAD is a potential treatment target for improving the e cacy of immunotherapy in ESCC patients.

Discussion
Previously, our group identi ed that lncRNA-NORAD, whose expression is induced by DNA damage, plays key roles in mediating radiation resistance. NORAD is overexpressed in ESCC both in vivo and in vitro; furthermore, NORAD knockdown sensitizes ESCC cells to radiation treatment. In this manuscript, we further investigated the corresponding molecular mechanisms. NORAD delays pri-miR-199a maturation by inhibiting the interaction of PUM1 with pri-miR-199a and subsequently releasing EEPD1, which can be recruited to damaged DNA and promote HR in response to radiation. Concurrently, NORAD knockdown inhibits PD-L1 ubiquitination and enhances anti-PD-1 treatment e cacy, especially in combination with radiotherapy. Our results indicate that NORAD acts as a novel target for improving the e cacy of radiation and anti-PD1 therapy by regulating the DNA damage response.
The HR-reported assay results con rm that NORAD knockdown sensitizes ESCC to radiation by inhibiting HR e ciency. We subsequently con rmed that NORAD knockdown delays the HR process by inhibiting ATR/Chk1 signaling activation in response to radiation. Evidence has suggested that inhibiting DNA repair components, including ATM, ATR, Chk1 and Chk2, markedly sensitizes cancer cells to radiation [23][24][25]. For example, Teng et al found that inhibiting either ATM or ATR signi cantly enhanced the radiation response in gynecological cancer cells (ovarian, endometrial, and cervical cancer cells) [26]. Notably, the generation of double-strand breaks is the central mechanism of action of radiotherapy. In response to radiation-induced DNA damage, ATM, ATR and DNA-PK S , three important DNA damage sensors, are immediately activated and arget a variety of overlapping substrates that promote DNA repair, cell cycle arrest and apoptosis [27,28]; these biological processes help cancer cells escape radiation damage. We found that under radiation stress, NORAD knockdown signi cantly increased ESCC apoptosis rates and impaired arrest at G2 phase, leading to genomic instabilities in ESCC cells. Based on these results, we inferred that NORAD is an effecitve target for enhancing the cytotoxic effects of radiation in ESCC.
Two major mechanisms have been reported with regard to NORAD-mediated regulation of genomic stability. A recent study revealed that NORAD acts as an RNA-binding protein; interacts with RBMX, TOP1 and other proteins; and facilitates the formation of the topoisomerase complex [11]. An earlier study con rmed that NORAD maintained genomic stability by binding to and negatively regulating PUMILIO (PUM1 and PUM2) in the cytoplasm [10,29]. To explore how NORAD regulates ATR signaling activation in response to radiation, we performed TMT mass spectrometry to identify the NORAD-regulated proteins.
The KEGG and GO analyses of differentially expressed proteins revealed that the biological processes mainly focused on DNA metabolic process, cell cycle, immune stimulation and so on. Consistently, we con rmed the interaction between NORAD and PUM1 and found that NORAD knockdown signi cantly upregulated PUM1 expression.
As an RNA-binding protein, PUM1 binds to RNAs containing the PUMILIO response element (UGUANAUA or UGUANAUN), thus mediating RNA deadenylation, decapping, and degradation [29]. In this study, we found that PUM1 can bind pri-miR-199a and facilitate the Drosha processing of pri-miR-199a. miRNAs are transcribed initially as pri-miRNAs and recognized by RNase III DROSHA and RNA-binding protein DGCR8 in the nucleus. The anking single-stranded sequences are cleaved by DROSHA and generate pre-miRNAs, which are then cleaved by the RNase III Dicer to generate double-stranded RNA ~21 nucleotides (nt) in length. Recently, numerous studies have reported that the expression level of mature miRNA is driven by both transcription and processing rates during the Drosha and Dicer processes [30]. For example, Guil et al stated that hnRNP A1, an RNA-binding protein, speci cally bound to pri-miR-18a before the Drosha process [31]. Another stem cell factor, LIN28, interacted with pre-miRNAs of the let-7 family and recruited uridyltransferases to these pre-miRNAs, essentially blocking their expression [32]. Here, we found that PUM1 can bind to pri-miR-199a and function as an auxiliary factor for the pri-miR-199a Drosha process. This interaction could be inhibited by NORAD, as upregulation of NORAD signi cantly delays the Drosha process of pri-miR-199a by sequestering PUM1 in the cytoplasm. Notably, NORAD knockdown not only facilitates the processing of miR-199a-5p by inhibiting PUM1 but also restores the radiosensitivity of cocultured radioresistant ESCC cells. NORAD knockdown drives the transfer of exosomal miR-199a-5p from NORAD-knockdown cells to radioresistant cells. In addition, EEPD1 was downregulated in radioresistant ESCC cells. Taken together, these results expanded the potential clinical importance of NORAD in ESCC radiotherapy.
With the assistance of WGCNA and randomForest, we identi ed the target genes of miR-199a-5p and EEPD1. Dual-luciferase reporter assays validated the direct interaction of EEPD1 and miR-199a-5p. Based on the IHC results, we observed that EEPD1 is a powerful biomarker for predicting the radioresistance of ESCC. Wu et al previously reported that EEPD1 can be recruited to stalled forks in response to DNA damage, where it promotes the restarting of these stalled forks by resecting the 5′ DNA end near the fork junction, thus permitting the invasion of the 3′ single strand and the initiation of HR [33]. Hyun-Suk Kim et al stated that EEPD1 acts as a gatekeeper for HR; it cleaves replication forks and creates a binding site for Exo1 on the free 5′ DNA end [19]. Furthermore, Changzoon Chun et al reported that the biological function of EEPD1 is similar to that of BRCA1. Depletion of EEPD1 in stressed zebra sh embryos results in chromosomal abnormalities, including anaphase bridges and micronuclei [34]. These results con rm that EEPD1 functions at the initiation of HR and that deleting EEPD1 sensitizes cells to DNA replication stressors.
Interestingly, we found that NORAD knockdown also enhances the e cacy of immune checkpoint inhibitors in combination with radiotherapy in tumor treatment. NORAD knockdown upregulates PD-L1 expression by inhibiting its ubiquitination. These phenomena may suggest the poor response of radioresistant ESCC patients to immune checkpoint inhibitors. One study found that PD-L1 protein expression uctuated during the cell cycle and peaked at G1 phase. Researchers have reported that PD-L1 is regulated by the G1 cycle checkpoint Cyclin D-CDK4 via the Cullin3 SPOP -E3 ligase pathway [35]. Here, we found that NORAD knockdown induces ESCC cell arrest in the G1 cell cycle, which might be responsible for the observed PD-L1 upregulation. In ESCC patients, only a limited number of patients could bene t from anti-PD1 therapy. In the last prospective clinical trial, pembrolizumab did not improve overall survival compared with paclitaxel as a second-line therapy for advanced esophageal cancer [36]. Radiation both alone and in combination with anti-PD-1 treatment suppressed tumor growth, but the combination did not signi cantly improve the tumor response compared to that of radiation monotherapy. For mice bearing AKR-sh-NORAD tumors, the use of combination therapy signi cantly inhibited tumor growth compared with that in response to radiation monotherapy. It is worth noting that NORAD knockdown signi cantly facilitates the antitumor e ciency of anti-PD-1 and radiation in ESCC. These results uncover that NORAD is a potential treatment target for improving immunotherapy e cacy         A. PUM1 CLIP data plotted across pri-miR-199a RNA in K562 cells based on ENCODE datasets. B. qPCR assay on KYSE-150 after RIP performed by using anti-PUM1 or control IgG for detection of NORAD expression. Values were normalized to levels of immunoprecipitated pri-miR-199a-5p by using normal control IgG. C. qPCR assay on KYSE-150 with or without NORAD knockdown after RIP performed by using anti-PUM1 or control IgG for detection of NORAD expression. Values were normalized to levels of immunoprecipitated pri-miR-199a-5p by using normal control IgG. D. PUM1 expression analyzed by western-blot in KY-SE150 and TE-1 with or without 8 Gy MV X-rays treatments and cells with or without