An ATM/TRIM37/NEMO Axis Counteracts Genotoxicity By

Published OnlineFirst September 25, 2018; DOI: 10.1158/0008-5472.CAN-18-2063

Cancer Molecular Cell Biology Research

An ATM/TRIM37/NEMO Axis Counteracts Genotoxicity by Activating Nuclear-to- Cytoplasmic NF-kB Signaling Geyan Wu1,2, Libing Song3, Jinrong Zhu1,2, Yameng Hu1,2, Lixue Cao1,2, Zhanyao Tan1,2, Shuxia Zhang1,4, Ziwen Li1,2, and Jun Li1,2

Abstract

Blocking genotoxic stress-induced NF-kB activation damaging anticancer drug cisplatin in vitro and in vivo would substantially enhance the anticancer efficiency of through activation of the NF-kB pathway. Genotoxic genotoxic chemotherapy. Unlike the well-established classical stress-activated ATM kinase directly interacted with and NF-kB pathway, the genotoxic agents-induced "nuclear-to- phosphorylated TRIM37 in the cytoplasm, which induced cytoplasmic" NF-kB pathway is initiated from the nucleus translocation of TRIM37 into the nucleus, where it formed a and transferred to the cytoplasm. However, the mechanism complex with NEMO and TRAF6 via a TRAF6-binding motif linking nuclear DNA damage signaling to cytoplasmic IKK (TBM). Importantly, blocking the ATM/TRIM37/NEMO axis activation remains unclear. Here, we report that TRIM37, a via cell-penetrating TAT-TBM peptide abrogated genotoxic novel E3 ligase, plays a vital role in genotoxic activation of agent-induced NEMO monoubiquitination and NF-kB NF-kB via monoubiquitination of NEMO at K309 in the activity, resulting in hypersensitivity of cancer cells to gen- nucleus, consequently resulting in nuclear export of NEMO otoxic drugs. Collectively, our results unveil a pivotal role and IKK/NF-kB activation. Clinically, TRIM37 levels correlated for TRIM37 in genotoxic stress and shed light on mechan- positively with levels of activated NF-kB and expression of isms of inducible chemotherapy resistance in cancer. Bcl-xl and XIAP in esophageal cancer specimens, which also associated positively with clinical stage and tumor-node- Significance: In response to genotoxic stress, TRIM37 acti- metastasis classification and associated inversely with overall vates NF-kB signaling via monoubiquitination of NEMO, and relapse-free survival in patients with esophageal cancer. which subsequently promotes cisplatin chemoresistance and Overexpression of TRIM37 conferred resistance to the DNA- tumor relapse in cancer. Cancer Res; 78(22); 6399–412. 2018 AACR.

Introduction Meanwhile, cells have evolved a precisely controlled network of DNA damage response (DDR) to respond to genotoxic stresses, Chromosomal integrity of all living organisms is endlessly includes sensing of damaged DNA, activation of cell cycle check- jeopardized by genotoxic stress generated from endogenous met- points, assembly of DNA repair machineries, and transactivation abolic sources, such as reactive oxygen species (ROS), and envi- of DNA damage-responsive gene expression. For instance, ATM ronmental resources, such as ionizing radiation and genotoxic kinase and PARP-1 act as sensor proteins to detect DNA lesions chemicals (1–4). Unrepaired or inappropriately repaired DNA and modify a variety of proteins, which initiate DNA repair and damage consequently attribute to genetic variation, apoptosis, cell-cycle checkpoint control (6, 7). aging, degenerative diseases, inflammation, and cancer (2–5). On the other hand, the NF-kB signaling pathway was emerged as a vital mediator for cellular responses to genotoxic threats via inducing survival genes that allow cells to repair damaged DNA 1Key Laboratory of Liver Disease of Guangdong Province, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, P.R. China. 2Department of bio- and promote survival (8, 9). However, unlike the well-established chemistry, Zhongshan school of medicine, Sun Yat-sen University, Guangzhou, classical NF-kB pathway in which signals are initiated from P.R. China. 3State Key Laboratory of Oncology in South China, Collaborative cell surface receptors and transduced from the cytoplasm to Innovation Center for Cancer Medicine, Sun Yat-sen University Cancer Center, the nucleus (10, 11), genotoxic agents triggered the "nuclear-to- 4 Guangzhou, P.R. China. Key Laboratory of Protein Modification and Degrada- cytoplasmic" NF-kB pathway that is initiated from the nucleus fi tion, School of Basic Medical Sciences, Af liated Cancer Hospital &Institute of and transferred to the cytoplasm (8, 9). In response to DNA Guangzhou Medical University, Guangzhou Medical University, Guangzhou, P.R. China. damage signals, PARP-1 is rapidly recruited to DNA damage sites and induces auto-poly(ADP-ribosyl)ation (PARylation), which Note: Supplementary data for this article are available at Cancer Research assembles NEMO, PIASy, ATM, PIDD, and LRP16 into a nucle- Online (http://cancerres.aacrjournals.org/). oplasmic signalosome (12–16). Furthermore, signalosome for- G. Wu, L. Song, and J. Zhu contributed equally to this article. mation induces sumoylation of NEMO at K277/K309 in a PIDD/ Corresponding Author: Jun Li, Sun Yat-sen University, 74 Zhongshan Road II, PARP1/PIASy-dependent manner and ATM-dependent phos- Guangzhou, Guangdong 510080, P.R. China; Phone: 86-20-87335828; phorylation of NEMO at S85 (12–17). Then phosphorylated- Fax: 86-20-87335828; E-mail: [email protected] NEMO is monoubiquitinated and exported from the nucleus doi: 10.1158/0008-5472.CAN-18-2063 together with ATM and ELKS, which form a complex with IKK 2018 American Association for Cancer Research. catalytic subunits in the cytoplasm, consequently resulting in

www.aacrjournals.org 6399

Wu et al.

activation of IKK/NF-kB signaling (18). Therefore, nuclear mono- Plasmids, virus constructs, and retroviral infection of target ubiquitin of NEMO, which is essential for its nuclear export, is the cells key step in genotoxic agent-triggered "nuclear-to-cytoplasmic" Human TRIM37, NEMO, and TRAF6 were amplified by PCR NF-kB signaling. However, the E3 ligase responsible for NEMO andclonedintothepSin-EF2vector.Fragmentsofthehuman nuclear monoubiquitination remains unknown. TRIM37 and TRAF6-coding sequence were amplified by PCR Tripartite motif containing 37 (TRIM37) is a newly identified andclonedintothepSin-EF2vector.Theindicatedmutants E3 ubiquitin ligase that comprises a tripartite motif (TRIM, were created using primers and a Stratagene Mutagenesis Kit RING-B-box-coiled-coil) domain, TRAF domain (TD), and according to the protocol recommended by the manufacturer. polyacidic domain (19–21). It has been recently reported that pNF-kB-luc and control plasmids (Clontech) was used to TRIM37 plays vital roles in various biological processes quantitatively examine NF-kB activity. Transfection of siRNAs depending on TRIM domain-dependent E3 ligase activity, such or plasmids was performed using the Lipofectamine 3000 as promotion of peroxisomal matrix protein import via direct reagent (Invitrogen) according to the manufacturer's instruc- monoubiquitination of PEX5 at K464 and silencing of gene tion. Retroviral production and infection were performed as expression through monoubiquitination of histone H2A (22, described previously (25). Stable cell lines expressing indicated 23). In this study, we unveiled a novel function of TRIM37 in genes were selected for 10 days with 0.5 mg/mL puromycin regulating nuclear-to-cytoplasmic NF-kB signaling. We found 48 hours after infection. The primers used were listed in that, upon genotoxic stimulation, TRIM37 rapidly translocated Supplementary Table S4. into the nucleus where it interacted directly with TRAF6 to catalyze monoubiquitination of NEMO at K309. Blocking Immunohistochemistry TRIM37/TRAF6 interaction using a cell-penetrating TAT-TD IHC analysis was performed to study altered protein expression peptide abrogated NEMO monoubiquitination-dependent in 441 human esophageal cancer tissues according previous report NF-kB signaling, resulting in hypersensitivity of esophageal (26). Paraffin-embedded tissues were analyzed using IHC with cancer cells to DNA-damaging chemotherapeutics. Hence our anti-TRIM37 antibody (Abcam; 1:200), anti-NF-kB p65 antibody study reveals a crucial role of TRIM37 in genotoxic stress- (Abcam; 1:200), anti-Bcl-XL antibody (Cell Signaling; 1:100), induced NF-kB activation and sheds light on mechanisms of anti-XIAP (1007-1008) antibody (Proteintech; 1:100). All the inducible chemotherapy resistance in esophageal cancer. antibodies used in this study has been listed in Supplementary Table S5. The degree of immunostaining of formalin-fixed, par- fi Materials and Methods af n-embedded sections were reviewed and scored separately by two independent pathologists uninformed of the histopathologic Ethics statement features and patient data of the samples. The scores were deter- Informed consent was signed by all patients, and the investi- mined by combining the proportion of positively-stained tumor gation has been conducted in accordance with the ethical stan- cells and the intensity of staining. The scores given by the two dards according to the Declaration of Helsinki, national and independent pathologists were combined into a mean score for international guidelines, which has also been approved by the further comparative evaluation. Tumor cell proportions were authors' Institutional Review Board. scored as follows: 0, no positive tumor cells; 1, <10% positive tumor cells; 2, 10% to 35% positive tumor cells; 3, 35% to 75% Tissue specimens and patient information positive tumor cells; 4, >75% positive tumor cells. Staining All of the patients received standardized platinum-based che- intensity was graded according to the following standard: 1, no motherapy. Informed consent was obtained from all patients and staining; 2, weak staining (light yellow); 3, moderate staining approvals from Sun Yat-sen University Cancer Center Institution- (yellow brown); 4, strong staining (brown). The staining index al Research Ethics Committee were obtained for this study. A total (SI) was calculated as the product of the staining intensity score fi of 441 paraf n-embedded, archived esophageal cancer specimens and the proportion of positive tumor cells. Using this method of and freshly collected 1 esophageal cancer-adjacent normal tissue assessment, we evaluated protein expression in benign esophage- and 9 esophageal cancer tissues were histopathologically and al epithelia and malignant lesions by determining the SI, with clinically diagnosed at Sun Yat-sen University Cancer Center possible scores of 0, 2, 3, 4, 6, 8, 9, 12, and 16. Samples with an SI (Guangdong, China) between 2005 and 2016. The clinical infor- 8 were determined as high expression and samples with an SI < mation of the samples is shown in Supplementary Tables S1 to S3. 8 were determined as low expression. Cutoff values were deter- Detailed description provided in Supplementary methods. mined on the basis of a measure of heterogeneity using the log- rank test with respect to OS. Cells Primary cultures of normal esophageal epithelial cells were Coimmunoprecipitation assay established from fresh specimens of the adjacent noncancer- Cells grown in 100-mm culture dishes were lysed using 500 mL ous esophageal tissue, which was over 5 cm from the cancerous of lysis buffer [25 mmol/L HEPES (pH 7.4), 150 mmol/L NaCl, tissue, according previously report (24). The Eca109 cells were 1% NP-40, 1 mmol/L EDTA, 2% glycerol, 1 mmol/L phenyl- kindly provided by Professors Tsao SW (The University of methylsulfonyl fluoride (PMSF)]. After being maintained on ice Hong Kong) and grown in DMEM medium (Invitrogen) sup- for 30 minutes, the lysates were clarified by microcentrifugation at plemented with 10% FBS (HyClone). All the cell lines been 12,000 rpm for 10 minutes. To preclear the supernatants, the tested for Mycoplasma contamination. All cell lines were lysates were incubated with 20 mL of agarose beads (Calbiochem) authenticated by short tandem repeat (STR) fingerprinting at for 1 hours with rotation at 4C. After centrifugation at 2,000 rpm Medicine Lab of Forensic Medicine Department of Sun Yat-Sen for 1 minutes, the supernatants were incubated with 20 mLof University. antibody-cross-linked protein G-agarose beads overnight at 4C.

6400 Cancer Res; 78(22) November 15, 2018 Cancer Research

TRIM37 Activates NF-kB Signaling via NEMO Monoubiquitination

The agarose beads were then washed six times with wash buffer phalloidin (Alexa Fluor 488; Invitrogen) and DAPI (Sigma- [25 mmol/L HEPES (pH 7.4), 150 mmol/L NaCl, 0.5% NP-40, Aldrich) according to manufacturer's protocols. The images were 1 mmol/L EDTA, 2% glycerol, 1 mmol/L PMSF]. After removing captured using the AxioVision Rel.4.6 computerized image anal- all the liquid, the pelleted beads were resuspended in 30 mLof1M ysis system (Carl Zeiss). The mice used in this study were sacrificed glycine (pH 3), after which, 10 mLof4 sample buffer was added, when the volume of control tumors reached to 1.5-cm diameter or the samples were denatured, and the sample components were the mice become moribund. All of the animal procedures were electrophoretically separated on SDS-PAGE for immunoblot approved by the Sun Yat-sen University Animal Care Committee. analysis. Statistical analysis Stochastic optical reconstruction microscopy All statistical analyses were carried out using SPSS130.0 statis- Eca109 cells seeded on No. 118 mm round coverslips were tical software. A chi-squared test was used to analyze the rela- washed twice with PBS, fixed with 4% PFA in PBS for 1 hour, tionship between TRIM37 expression and the clinicopathologic permeabilized with the PBST for 10 minutes, and blocked with characteristics. Survival curves were plotted using Kaplan–Meier 4% BSA in PBS for 1 hour. The cells were then incubated with method and compared by log-rank test. Survival data were eval- primary antibodies for 3 hours and with Alexa Fluor 647 conju- uated by univariate and multivariate Cox regression analyses. P < gated goat anti-rabbit IgG (1:200 dilution) and pre-adsorbed 0.05 was considered statistically significant. Alexa Fluor 568 conjugated goat anti-mouse IgG (1:200 dilution; Abcam, ab175733) antibodies for 1.5 hours. The sample was kept in PBS until imaging. Results TRIM37 promotes genotoxic stress-induced NF-kB activation Generation and preparation of TAT-37/TBM peptides To identify potential nuclear E3 enzyme for NEMO monou- Peptides were synthesized by the ChinaPeptides using standard biquitination, affinity purification and mass spectrometry (MS) HOBt/Fmoc chemistry and purified by reverse phase HPLC to was conducted using nuclear extracts from etoposide-treated >95% purity. The final amino acid compositions were verified Eca-109 esophageal cancer cells. In addition to previously utilizing amino acid analysis and MALDI TOF mass spectrometry. reported interacting proteins, such as TRAF6, ATM, and PIASy The cell permeation sequence used was previously demonstrated (12–17), we found that E3 ligase TRIM37 may also be a potent to have 10-fold increased intracellular concentration compared nuclear NEMO-interacting protein (Fig. 1A). Prominently, com- with the native TAT sequence from HIV-1 (27). We designed a pared with control cells, etoposide-induced NF-kB activation was negative control peptide by changing DFEVGE to AFAVGA. The rapidly elevated in TRIM37-overexpressing cells but decreased in amino acid sequences of these peptides are: TAT-Ctrl: TRIM37 / cells (Fig. 1B and C), suggesting that TRIM37 played a RKKRRORNRRRAFAVGA; TAT-TBM: RKKRRORNRRRDFEVGE. vital role in etoposide-induced NF-kB activation. This hypothesis was further confirmed by multiple assays, in which overexpres- Xenografted tumor models, IHC, and H&E staining sion of TRIM37 significantly increased, but knockout or knock- In the subcutaneous patient-derived xenografts (PDX) tumor down of TRIM37 decreased the IKK activity, the expression of NF- model, freshly isolated clinical esophageal cancer patient kB-regulated antiapoptotic genes Bcl-XL and XIAP, and the NF-kB tissues were subdivided into 2 to 3 mm3 pieces, coated with transcriptional activity in the etoposide-treated cells (Fig. 1C; Matrigel (BD Biosciences) and media in a 1:1 ratio, and Supplementary Fig. S1A and S1B). Importantly, the elevated embedded within the subcutaneous space underneath the skin NF-kB activity induced by irradiation, camptothecin, or CDDP of female NOD/Shi-scid/IL-2Rgnull (NOG) mouse (6–8 weeks was also drastically decreased in TRIM37 / cells (Supplementary old; CREA Japan Inc.). Tumor growth was monitored by mea- Fig. S1C and S1D), demonstrating that TRIM37 contributed to suring the tumor diameters, and the tumor volume was calcu- genotoxic stress-induced NF-kB activation. lated using the equation (L W2)/2. When the tumor became palpable, mice were intratumoral treated with cisplatin (CDDP; Clinical relevance of TRIM37/NF-kB signaling in human 5 mg/kg) and intratumoral injection of TAT-control peptide or esophageal cancer TAT-TRIM37/TBM peptide three times per week (as per cycle) Statistical analyses revealed that TRIM37 levels were positively for up to 6 weeks. At the end of treatment, the mice were correlated with level of activated NF-kB and expression of Bcl-xl sacrificed and the tumors were excised and weighed, and and XIAP in 441 paraffin-embedded and nine freshly collected confirmed by histology. esophageal cancer specimens (Fig. 1D and E). IHC staining In the subcutaneous tumor model, the indicated luciferase showed that TRIM37 protein was slightly expressed in normal expressing cells (1 106) were injected subcutaneously into nude esophageal tissues but showed markedly higher expression in mice. When the luminescence signal reached 2 107 p/sec/cm2/ esophageal cancer and was further elevated in relapsed esoph- sr, mice were intratumoral treated with CDDP (5 mg/kg) three ageal cancer (Fig. 1F). Furthermore, we found that TRIM37 times per week (as per cycle) for up to 6 weeks. Mice were expression was positively associated with the clinical stage (P ¼ sacrificed when moribund as determined by an observer 0.001), tumor-node-metastasis classification (P ¼ 0.026; P ¼ blinded to the treatment, and tumors were excised and paraffin- 0.004; P ¼ 0.039) and inversely associated with overall and embedded. Serial 4.0 mm sections were cut and subjected to IHC relapse-free survival in patients with esophageal cancer (Supple- and hemotoxylin and eosin (H&E) staining. After deparaffiniza- mentary Table S1–S3 and Fig. 1G). Importantly, Kaplan–Meier tion, sections were H&E-stained with Mayer's hematoxylin solu- plotter analysis revealed that expression of TRIM37 were signif- tion, or IHC-stained using antibodies of NF-kB p65 (1:100; icant correlated with shorter overall and relapse-free survival in Abcam), or stained with TUNEL (In Situ Cell Death Detection patients with breast cancer, ovarian cancer, or liver cancer (Sup- Kit, TMR red; Roche Applied Science), and counterstained with plementary Fig. S1E). These results indicate that TRIM37

www.aacrjournals.org Cancer Res; 78(22) November 15, 2018 6401

Wu et al.

Figure 1. TRIM37 promotes genotoxic stress-induced NF-kB activation. A, Immunoprecipitation assay was performed in nuclear extracts from etoposide (Etop; 10 mmol/L, 2 hours)-treated NE1/Flag-NEMO cells using anti-Flag affinity agarose, followed by mass-spectrometric peptide sequencing. E3 ligase TRIM37 was identified as one of the proteins present in the precipitate. B and C, NF-kB DNA-binding activity by EMSA and IKK kinase activity by in vitro kinase activity assay were examined in NE1 cells transfected with 0, 0.5, 1.5, and 5.0 mg of a Flag-tagged TRIM37 plasmid (B) or in Eca-109/TRIM37/ cells, followed by the treatment with etoposide (10 mmol/L, 2 hours) or TNFa (10 ng/mL, 15 minutes; C). D, Analysis of expression (left) and correlation (right) of TRIM37 with Bcl-XL and XIAP mRNA expression, as well as NF-kB DNA-binding activity in one freshly collected esophageal cancer-adjacent sample and nine esophageal cancer samples. Each bar represents the mean SD of three independent experiments. E, TRIM37 levels associated with nuclear NF-kB p65, Bcl-XL, or XIAP expression in 441 primary human esophageal cancer specimens. Left, two representative specimens with low and high levels of TRIM37 expression are shown. Original magnification, 200. Right, percentages of specimens showing low or high TRIM37 expression relative to the level of nuclear NF-kB p65, Bcl-XL, or XIAP. Scale bars, 50 mm. F, IHC staining of TRIM37 in normal esophageal tissue, nonrelapsed, and relapsed esophageal cancer tissues. Scale bars, 50 mm. G, Kaplan–Meier curves of overall survival (left) and relapse-free survival (right) of patients with esophageal cancer with low versus high expression of TRIM37 (n ¼ 441; P < 0.001, log-rank test).

6402 Cancer Res; 78(22) November 15, 2018 Cancer Research

TRIM37 Activates NF-kB Signaling via NEMO Monoubiquitination

Figure 2. TRIM37 overexpression confers esophageal cancer cells resistance to CDDP in vitro and in vivo. A, FACS analysis of Annexin V staining (left) and quantiﬁcation (right) of indicated cells treated with vehicle or CDDP (5 mmol/L) at 24 hours. B, Representative images (left) and quantiﬁcation (right) of colony number of the indicated cells. C, Xenograft model in nude mice. Representative images of tumor-bearing mice (left) and tumor volumes were examined on the indicated days (right; n ¼ 6/group). D, EMSA assay of NF-kB activity in the indicated tumors (n ¼ 6/group). OCT-1/DNA-binding complex served as a control. a-Tubulin was used as a loading control. E, H&E and IHC staining of nuclear NF-kB p65 and TUNEL-positive cells in the indicated tumors (n ¼ 6/group), Scale bars, 50 mm. F, Kaplan–Meier survival of the indicated mice (n ¼ 6/group). Each bar in A, B, and E represents the mean SD of three independent experiments. , P < 0.05; , P < 0.01; , P < 0.001.

overexpression may contribute to development and progression exhibited remarkable resistance to CDDP therapy, as indicated by of multiple types of human cancer. rapid tumor progression, lower proportion of apoptotic cells, and increased NF-kB activation, consequently resulting in shorter TRIM37 confers resistance to CDDP in vitro and in vivo survival of tumor-bearing mice (Fig. 2C–F). In contrast, CDDP NF-kB activation usually induces various antiapoptotic genes treatment led to signiﬁcant remission of TRIM37 / /tumors that allow cells to survive in the presence of DNA-damaging drugs (Fig. 2C–F). These results demonstrate that TRIM37 overexpres- (8, 9). We therefore examined the effect of TRIM37 on resistance sion confers resistance to CDDP via NF-kB activation. to CDDP, a DNA-damaging drug that is commonly used in anticancer therapies. As shown in Fig. 2A and B, overexpressing TRIM37 induces NEMO monoubiquitination at residue K309 TRIM37 decreased the CDDP-induced apoptotic death and Although we observed no alterations of etoposide-induced increased the colony formation. The same results were also sumoylation and phosphorylation of NEMO between TRIM37- obtained using an in vivo tumor model in which TRIM37/tumors dysregulated and control cells, the expression of etoposide-induced

www.aacrjournals.org Cancer Res; 78(22) November 15, 2018 6403

Wu et al.

Figure 3. TRIM37 promotes NEMO monoubiquitination at residue K309. A, The whole cell extracts prepared from the indicated cells treated with etoposide (Etop; 10 mmol/L, 2 hours), then immunoblot analysis of expression of p-ATM, total ATM, sumo-NEMO, p-NEMO (S85), immunoprecipitated NEMO, and ubiquitinated NEMO. a-Tubulin was used as a loading control. B, Immunoblot analysis of expression of immunoprecipitated NEMO, IKKb, cIAP1, TAK1, ATM, and PIASy in vector- and myc-TRIM37–transfected cells, followed by treatment with etoposide (10 mmol/L, 2 hours). (Continued on the following page.)

6404 Cancer Res; 78(22) November 15, 2018 Cancer Research

TRIM37 Activates NF-kB Signaling via NEMO Monoubiquitination

monoubiquitinated NEMO was rapidly increased in TRIM37- These results demonstrated that TRIM37- mediated monoubiqui- transduced cells but decreased in TRIM37 / cells (Fig. 3A; tination of NEMO promotes nuclear export of NEMO. Supplementary Fig. S2A). The promotive effect of TRIM37 on etoposide-induced NEMO monoubiquitination was drastically Genotoxic stress induces nuclear translocation of TRIM37 abrogated by silencing of ATM or PIASy but remained in IKKb-, Coimmunoprecipitation (Co-IP) assays revealed the genotoxic cIAP1-, or TAK1-silenced cells (Fig. 3B). These results suggest that stress-induced TRIM37/NEMO interaction only occurred in the TRIM37 may participate in NEMO monoubiquitination. nucleus but not in the cytoplasm (Fig. 4A–C; Supplementary Fig. MS analysis revealed that Lys309 was the TRIM37-monoubi- S3A and S3B). We then asked whether genotoxic stress could quitinated residue (Fig. 3C). Accordingly, overexpressing TRIM37 induce translocation of TRIM37 into nuclear whereas TRIM37 in the etoposide-treated NE1/NEMO / cells had no impact on interacted with and monoubiquitinatd NEMO. Interestingly, the the monoubiquitinated level of Lys309 NEMO mutant (K309A) nuclear expression of TRIM37 dramatically elevated at 30 min but significantly increased the monoubiquitinated level of after genotoxic stress (Fig. 4D and E; Supplementary Fig. S3C). NEMO/WT and other NEMO lysine mutants (K277A, K285A, However, TRIM37 containing a mutant nuclear localization signal and K399A; Fig. 3D). Meanwhile, TRIM37 overexpression in NE1/ prevented genotoxic stress-induced nuclear translocation of NEMO / cells dramatically enhanced the recovery effect of TRIM37 and interaction with NEMO, as well as TRIM37-induced NEMO/WT, but not NEMO/K309A, on the genotoxic stress- genotoxic NF-kB activation and NEMO monoubiquitination induced DNA-binding and transcriptional activities of NF-kB and (Supplementary Fig. S3D–S3G). IKK activity (Fig. 3E and F; Supplementary Fig. S2B–S2D). These results demonstrate that TRIM37-mediated NEMO monoubiqui- TRAF6 is required for TRIM37-mediated NEMO tination at K309 is vital for genotoxic NF-kB activation. monoubiquitination Far-Western blot analysis that TRIM37 could not interact TRIM37 promotes nuclear export of NEMO directly with NEMO, suggesting that TRIM37-mediated genotoxic In agreement with previous findings that monoubiquitination NEMO monoubiquitination required other protein(s). By indi- of NEMO is crucial for its nuclear export (15), we found that vidually silencing all identified nuclear NEMO-interacting pro- overexpressing TRIM37 dramatically decreased the nuclear teins in Figure 1A, we found that silencing TRAF6 in the etoposide- expression of NEMO/WT but had no impact on nuclear level of treated cells almost entirely abrogated genotoxic stress-induced NEMO/K309A in etoposide-treated cells (Fig 3G), suggesting that TRIM37/NEMO interaction but knocking down NEMO did not TRIM37 promoted nuclear export of NEMO. This hypothesis reduce the binding affinity of TRIM37 for TRAF6, and that ablating was further supported by the observation that the duration of TRIM37 had no obvious impact on NEMO/TRAF6 association, etoposide-induced nuclear NEMO in TRIM37-transduced cells indicating that TRAF6 is required for formation of a TRIM37/ was much shorter than that in control cells, while TRIM37 / cells TRAF6/NEMO complex (Fig. 4F). Consistently, the nuclear level showed sustained nuclear NEMO signal (Fig. 3H; Supplementary of TRAF6 was also drastically elevated at 30 min after genotoxic Fig. S2E). However, the effect of TRIM37 on nuclear export of stress (Supplementary Fig. S4A). The direct nuclear interaction of NEMO was abolished by NEMO/K309A (Fig. 3H). Consistent TRIM37 and TRAF6 was confirmed by far-western blot and with previous report (28), calcium chelation prevented the cyto- STORM analyses (Fig. 4G and H). Interestingly, although the plasmic expression of monoubiquitinated NEMO in the TRIM37- zoom-in STORM image showed TRAF6 (red) stained in close transduced cells (Fig. 3I and J), which provided further evidence proximity with TRIM7(green) both in the cytoplasm (middle, that TRIM37 mediated monoubiqutination of NEMO in the top) and nucleus (middle, bottom), 3D render revealed that nucleus. Remarkably, the etoposide-induced association of TRAF6 interacted directly with TRIM37 in the nucleus (right, NEMO with Ran, an export receptor and the cargo protein for bottom) but not in the cytoplasm (right, top; Fig. 4H). These nuclear export of NEMO (29), was increased in TRIM37-trans- results provided further evidence that TRIM37 forms complex duced cells but was nearly undetectable in TRIM37 / cells, and with TRAF6 in the nucleus upon genotoxic treatment. Consis- no interaction of Ran and NEMO/K309A was observed (Fig. 3K). tently, silencing TRAF6 abolished the effect of TRIM37 on

(Continued.) Total cell lysates were probed with anti-Flag antibody and a-tubulin was used as a loading ycontrol. C, MS spectrum of an ubiquitinated NEMO peptide shown in the myc-TRIM37/Flag-NEMO cotransduced cells, followed by treatment with etoposide (10 mmol/L, 2 hours). Fragment ions are indicated. D, Immunoblot analysis of expression of immunoprecipitated NEMO in etoposide-treated control or myc-TRIM37-overexpressing in NE1/NEMO/ cells transfected with plasmids encoding Flag-NEMO/WT or the indicated Flag-NEMO/mutant. Total cell lysates were probed with anti-Flag antibody and a-tubulin was used as a loading control. E, NF-kB DNA-binding and IKK kinase activities were examined in vector-, Flag-NEMO/WT-, or Flag-NEMO/K309A–transduced NEMO/ cells cotransfected with vector or myc-TRIM37. Thirty-six hours after transfection, cells were stimulated with etoposide (10 mmol/L, 2 hours) and analyzed for anti-Flag and anti-NEMO antibodies. a-Tubulin was used as a loading control. F, NF-kB DNA-binding and IKK kinase activities were examined in the indicated cells 2 hours after treated with ionizing radiation (IR; 10 Gy), camptothecin (CPT; 10 mmol/L), or CDDP (5 mmol/L). Total cell lysates were probed with anti-Flag antibody and a-tubulin was used as a loading control. G, Immunoblot analysis of expression of TRIM37/wt and TRIM37/mutant in nuclear extracts from indicated etoposide- treated NEMO / cells (10 mmol/L, 2 hours). b-Actin was used as a cytoplasmic loading control and lamin B1 was used as a nuclear loading control. H, Representative pictures (left) and quantiﬁcation (right) of nuclear Flag-NEMO/WT or Flag-NEMO/K309A in the indicated cells treated with etoposide for the indicated times. Each bar represents the mean SD of three independent experiments. , P < 0.05; , P < 0.01; , P < 0.001. Scale bars, 10 mm. I, Immunoblot analysis of expression of monoubiqutinated-NEMO in the cytoplasm (CE) and nucleus (NE) of indicated cells, preincubated either with DMSO or BAPTA (20 mmol/L) for 30 minutes and further treated with etoposide (10 mmol/L, 2 hours). b-Actin was used as a cytoplasmic loading control and lamin B1 was used as a nuclear loading control. J and K, Immunoprecipitation assay was performed using anti-NEMO antibody in nuclear extracts of NE-1/V and NE-1/TRIM37 cells or Eca-109/Ctrl and Eca-109/TRIM37/ cells (J), or using anti-NEMO antibody in nuclear extracts of Flag-NEMO/WT- or Flag-NEMO/K309A-transduced NEMO/ cells (K), which were preincubated with BAPTA (20 mmol/L, 30 minutes) and then treated with etoposide (10 mmol/L, 2 hours), and then analyzed by immunoblot with anti-Ran antibody.

www.aacrjournals.org Cancer Res; 78(22) November 15, 2018 6405

Wu et al.

Figure 4. TRAF6 is required for TRIM37-mediated NEMO monoubiquitination. A, Immunoprecipitation assays were performed in cells transfected with vector- or myc-TRIM37 prior to the etoposide (Etop) exposure (10 mmol/L, 2 hours) or TNFa treatment (10 ng/mL, 15 minutes) using anti-Flag antibody and immunoblot was analyzed using anti- NEMO and anti-IKKb antibodies. B, Immunoprecipitation/immunoblot analyses were performed in cells treated with etoposide (10 mmol/L, 2 hours) or TNFa (10 ng/mL, 15 minutes) using anti-TRIM37 or anti-NEMO antibodies. C, Immunoprecipitation/immunoblot analyses were performedinfractionatedcytoplasmic(C)andnuclear(N) extracts from etoposide-treated cells (10 mmol/L, 2 hours) using anti-TRIM37 and anti-NEMO antibodies. D, Immunoblotanalysis of TRIM37 expressionin the nucleus extract from etoposide-treated cells at the indicated time. Lamin B1 was used as a nuclear loading control. E, Representative images of TRIM37 immunostained with anti-TRIM37 antibody (left) and the expression of TRIM37 in subcellular fractions (right) of the Eca-109 cells treated with or without etoposide (10 mmol/L, 30 minutes). Scale bars, 10 mm. F, Co-IP assays were performed in the indicated cells using anti-TRIM37 or anti-NEMO antibodies. Left, depletion of TRAF6 impaired the interaction between TRIM37 and NEMO. Middle, depletion of TRIM37 did not affect the interaction between NEMO and TRAF6. Right, depletion of NEMO did not alter the interaction between TRIM37 and TRAF6. G, Far-Western blotting analysis was performed using IgG or TRIM37 antibody-immunoprecipitated proteins and detected using anti-TRAF6 antibody and then reblotted with anti-TRIM37 antibody. Recombinant GST-TRAF6 was used as a control. H, The interaction of TRAF6 and TRIM37 was examined using STORM captured in wide shot (left; scale bars, 8 mm), further zoomed-in (middle; scale bars, 100 nm) and 3D-rendered (right). I, NF-kB DNA-binding activity by EMSA and expression of monoubiqutinated NEMO by immunoblot analysis were examined in vector/cells or TRIM37/cells pretransfected with control or TRAF6-siRNA and then analyzed 2 hours after treatment with ionizing radiation (10 Gy), camptothecin (CPT; 10 mmol/L), or CDDP (5 mmol/L). J, Immunoprecipitation assays were performed using anti-TRIM37 antibody in the etoposide-treated cells at the indicated times and analyzed by immunoblot with anti-NEMO and anti-TRAF6antibodies. K, Immunoprecipitation assay was performed using anti-TRIM37 antibody in etoposide-treated cells (10 mmol/L, 2 hours) transfected with the indicated doses of TRAF6 and analyzed by immunoblot with anti-NEMO antibody. L, NF-kB DNA-binding activity by EMSA and immunoblot analysis was performed with indicated cells treated with etoposide (10 mmol/L, 2 hours). M, In vitro ubiquitination assay was performed with puriﬁed recombinant GST-NEMO, GST-NEMO(85A), or GST-NEMO/K309A that was incubated with His-TRIM37 (10, 50, or 100 ng) in a reaction mixture containing 2 mmol/L ATP, 1 mg His-ubiquitin, 50 ng E1 (UBE1), and 100 ng E2 (His-UBCH5B) with or without 1 mgTRAF6or20mLp-ATMfor60minutesat37C. Blots were probed with anti-NEMO, anti-ub, anti-TRAF6, anti-p-ATM, and anti-TRIM37 antibodies.

6406 Cancer Res; 78(22) November 15, 2018 Cancer Research

TRIM37 Activates NF-kB Signaling via NEMO Monoubiquitination

genotoxic stress-induced NEMO monoubiquitination and NF-kB showed that TRIM37/DTBM and TRIM37/RF-mu abrogated the activity (Fig. 4I; Supplementary Fig. S4B). Moreover, co-IP assays interaction between NEMO and Ran (Fig. 6G). revealed that with increasing degrees of genotoxic stress, more TRIM37 was found complexed with TRAF6 and NEMO, and that TRIM domain was essential for TRIM37-mediated NEMO TRAF6 induced-interaction between TRIM37 and NEMO was in a monoubiquitination TRAF6 dose-dependent manner (Fig. 4J and K; Supplementary To further determine whether the E3 ligase activity of TRIM37 Fig. S4C). Importantly, overexpressing TRAF6-C70A mutant or was essential for NEMO monoubiquitination, a TRIM37 deriva- S13A/T330A failed to recover the level of NEMO monoubiqui- tive (TRIM37/RF-mu) bearing a point mutation in a conserved tination in TRAF6-silencing cells treated with etoposide, suggest- cysteine residue in the RING finger motif (C18R), which interferes ing that nuclear translocation and E3 activity of TRAF6 contributes with catalytic activity, was constructed. Although TRIM37/RF-mu to TRIM37-mediated NEMO ubiquitination (Fig. 4L). In vitro could still form a complex with TRAF6/NEMO upon etoposide protein binding and ubiquitination assays further demonstrated treatment, overexpressing TRIM37/RF-mu in the TRIM37 / cells that TRAF6 was essential for TRIM37-mediated NEMO mono- could not recover the genotoxic stress-induced monoubiquiti- ubiquitination via direct interaction with TRIM37 (Fig. 4M). nation and nuclear export of NEMO and NF-kB activation Notably, NEMO S85A mutant also dramatically abrogated the (Fig. 6C–G; Supplementary Fig. S5A and S5B), demonstrating effect of TRIM37 on NEMO monoubiquitination, indicating that that TRIM37-mediated NEMO monoubiquitination is TRIM ATM-mediated S85 phosphorylation of NEMO was early event of domain-dependent. its monoubiquitination (Fig. 4M). TAT-TRIM37/TBM peptide promotes genotoxic agent-induced ATM-mediated TRIM37 phosphorylation and nuclear tumor regression translocation The cell-penetrating HIV-1 transactivator of transcription TRIM37 was previously reported to be localized in peroxi- (TAT)-peptide has been widely used as an anticancer molecular somal membranes (19–23) and ATM kinase was also found to delivery system due to its high solubility and penetrability (33, be recruited by PEX5 to peroxisomal membranes (30, 31). 34). We then examined the inhibitory effect of TAT-conjugated Consistently, immunofluorescence and subcellular fraction- TRIM37/TBM (TAT-TBM) peptide on genotoxic NF-kB activation ation assays revealed that TRIM37 and ATM were colocalized and cancer progression. As shown in Fig. 7A and B, treatment with in peroxisomal membranes (Fig. 5A and B). Interestingly, TAT-TBM peptide dramatically reduced genotoxic stress-induced etoposide-induced TRIM37 nuclear translocation was drastical- TRIM37/TRAF6/NEMO complex formation, IKK/NF-kB activa- ly prevented by an inhibitor of ATM (Fig. 5C), suggesting that tion, and NEMO monoubiquitination in the cells with higher ATM was involved in TRIM37 nuclear translocation upon TRIM37 expression, consequently resulting in enhanced effect of genotoxic stress. Meanwhile, we found that genotoxic stress CDDP on cell death as indicated by increased caspase 3 positive induced the phosphorylation of TRIM37 at 196TQ/801SQ cells and decreased colony formation (Fig. 7C and D). Further- motifs (Fig. 5D) and mutating TRIM37 TQ/SQ sites T196 and more, a PDX model, using two freshly collected clinical esoph- S801 to alanine reduced genotoxic stress-induced phosphory- ageal cancer tissues, showed that cotreatment with CDDP and fi lation and nuclear translocation of TRIM37 (Fig. 5E). More- TAT-TBM peptide resulted in signi cant remission of esophageal over, we observed that that the genotoxic stress-induced ATM/ cancer tumor volume and mass (T#6: 34 mg vs. 657 mg; T#9: 65.8 mg vs. 983 mg) compared with control tumors (Fig. 7E), as well as TRIM37 complex only formed in the cytoplasm but not in the þ nucleus (Fig. 5F and G). Co-IP assays using ATM and serially higher percentage of TUNEL cells and reduced genotoxic NF-kB truncated TRIM37 fragments demonstrated that ATM interacted activity (Fig. 7F and G), in comparison with CDDP treatment with the TD of TRIM37 (Fig. 5H). Importantly, treatment alone. Taken together, these results further support the notion that with antioxidants not only dramatically prevented genotoxic TRIM37 overexpression enhanced genotoxic stress-induced NF- stress-induced TRIM37 nuclear translocation but also reduced kB activation, consequently resulting in chemoresistance and genotoxic stress-induced ATM/TRIM37 interaction and NEMO poorer clinical outcomes in human cancer. monoubiquitination (Fig. 5I–L). Therefore, oxidative stress may be also involved in ATM-mediated phosphorylation and Discussion nuclear translocation of TRIM37 and NEMO monoubiquitination upon genotoxic stress. Unlike activation of the classical NF-kB signaling pathway that displays a fast response to typical signals initiated from cell surface TRAF6-binding motif in TRIM37 is required for the TRAF6 receptors, "atypical" NF-kB activators, such as DNA damage or interaction oxygen stress, trigger a slow NF-kB signal (with peak activities Co-IP assays using serially truncated TRIM37 and TRAF6 frag- reached after 2–4 hours; refs. 8, 35). These delayed kinetics were ments demonstrated that the TD was the interaction region of previously thought to be at least partially attributable to the time TRIM37 and TRAF6 (Fig. 6A and B). The TD of TRIM37 contains a required to transfer the nuclear damage signal to the cytoplasmic sequence (DFEVGE; residues 366-371) with homology to a con- IKK complex. Multiple recent studies have provided insights sensus TRAF6-binding motif (TBM), PXEXX (aromatic/acidic that genotoxic threats triggering NF-kB activation require residue; ref. 32). Interestingly, a TRIM37 mutant containing a molecular trafficking between the nucleus and cytoplasm. For TBM deletion (TRIM37/DTBM) lost the capacity to directly bind to instance, genotoxic stress-induced nuclear translocation of IKK TRAF6, resulting in resistance to genotoxic stress-induced mono- unbound-NEMO is the key event for DNA damage-dependent ubiquitination and nuclear export of NEMO, as well as NF-kB IKK/NF-kB signaling (36). Meanwhile, genotoxic stress-induced activation (Fig. 6C–F; Supplementary Fig. S5A and B). Consis- nuclear translocation of PIDD resulted in augmentation of tently, immunoprecipitation assays using anti-Ran antibody sumoylation and ubiquitination of NEMO (14). However,

www.aacrjournals.org Cancer Res; 78(22) November 15, 2018 6407

Wu et al.

Figure 5. ATM-mediated TRIM37 phosphorylation and nuclear translocation. A, Colocalization of ATM and TRIM37 on peroxisomes in Eca-109 cells as analyzed by immunofluorescence. Scale bars, 10 mm (left), 2.5 mm (right). B, Peroxisomal fractionation/immunoblot analysis of expression of ATM, TRIM37, catalase, and PMP70. WCE, whole cell extracts; PO, peroxisome. C, Indicated cells were exposed to ATM inhibitor KU-55933 (10 mmol/L) or DMSO for 1 hour before addition of etoposide (Etop; 10 mmol/L, 2 hours) and then analyzed for subcellular fractionation/immunoblot analysis of TRIM37 expression. b-Actin was used as a cytoplasmic loading control and lamin B1 was used as a nuclear loading control. D, Immunoprecipitation assays using anti-Flag antibody were performed in TRIM37/wt- and TRIM37/ mutant-transfected cells treated with etoposide (10 mmol/L, 2 hours) with or without ATM inhibitor KU-55933 (10 mmol/L, 1 hour) pretreatment, and analyzed by immunoblot with anti-pTQ/SQ antibody. E, Immunoblot analysis of expression of TRIM37/wt- and TRIM37/mutant in nuclear/cytoplasmic extracts form the cells treated with or without etoposide (10 mmol/L, 2 hours). b-Actin was used as a cytoplasmic loading control and lamin B1 was used as a nuclear loading control. F, Immunoprecipitation/immunoblot analyses were performed in fractionated cytoplasmic (C) and nuclear (N) extracts from etoposide-treated cells (10 mmol/L, 2 hours) using anti-TRIM37 and anti-ATM antibodies. Scale bars, 8 mm (left), 100 nm (middle). G, Immunofluorescence analysis revealed that TRIM37 and ATM interacted in the cytoplasm. H, Left, schematic illustration of wild-type and truncated TRIM37. Right, immunoprecipitation assays were performed using anti-Flag antibody in etoposide-treated cells transfected with Flag-tagged TRIM37 or the indicated Flag-tagged TRIM37 mutants, and immunoblot analyzed with anti-Flag,ATM,andp-ATM antibodies. I–J, Immunoblot (I)andimmunofluorescence (J) analyses of TRIM37 expression in cells treated with etoposide (10 mmol/L, 2 hours) with or without pyrrolidinedithiocarbamate pretreatment (PDTC; 20mmol/L, 1 hour). b-Actin was used asa cytoplasmic loading controland lamin B1 wasused asa nuclear loading control. Scale bars, 10 mm. K, Immunoblot analysis of expression of ATM-immunoprecipitated TRIM37 in cells treated with the indicated agents. L, Immunoprecipitation/ immunoblot analyses of monoubiquitinated NEMO expression and EMSA of NF-kB DNA-binding activity in cells treated with the indicated agents.

6408 Cancer Res; 78(22) November 15, 2018 Cancer Research

TRIM37 Activates NF-kB Signaling via NEMO Monoubiquitination

Figure 6. TRAF6-binding motif of TRIM37 is required for TRIM37/TRAF6 interaction. A, Left, schematic illustration of wild-type and truncated TRAF6. Right, co-IP assays were performed using anti-Flag antibody in etoposide-treated HEK293T cells (10 mmol/L, 2 hours) after transfection with Flag-tagged TRIM37 or the indicated myc-tagged TRAF6 mutants and analyzed by immunoblot with anti-Flag and anti-myc antibodies. B, Co-IP assays were performed using anti-myc antibody in etoposide-treated HEK293T cells (10 mmol/L, 2 hours) pretransfected with myc-tagged TRAF6 or the indicated Flag-tagged TRIM37 mutants for 36 hours and analyzed by immunoblot with anti-Flag and anti-myc antibodies. C, Left, schematic illustration of TRIM37 mutants TRIM37/DTBM and TRIM37/RF-mu. Bottom, co-IP assays were performed in etoposide-treated HEK293T cells (10 mmol/L, 2 hours) pretransfected with myc-TRIM37/wt, myc-TRIM37/DTBM, or myc-TRIM37/RF-mu, and analyzed by immunoblot with anti-Flag, anti-NEMO, and anti-TRAF6 antibodies. D, Far-Western analysis was performed using immunoprecipitated myc-TRIM37/wt, myc-TRIM37/DTBM, or myc-TRIM37/RF-mu, which were gel-puriﬁed, transferred to a membrane, and incubated with recombinant TRAF6, then detected using anti-TRAF6 antibody and then reblotted with anti-TRIM37 antibody. Recombinant GST-TRAF6 was used as a control. E, NF-kB DNA-binding and IKK activities (left) and expression of monoubiquitinated NEMO were examined in etoposide-treated cells (10 mmol/L, 2 hours) pretransfected with myc-TRIM37, myc-TRIM37/DTBM, or myc-TRIM37/RF-mu. BCL-XL and XIAP expression was analyzed at 6 hours in the indicated cells treated with etoposide (Etop; 10 mmol/L). GST-IkBa or a-tubulin was used as loading control. F, Representative pictures (top) and quantiﬁcation (bottom) of nuclear Flag-NEMO/WT in the indicated cells treated with etoposide for the indicated times. Each bar represents the mean SD of three independent experiments. , P < 0.05, , P < 0.01, , P < 0.001. Scale bars, 10 mm. G, Immunoprecipitation assays using anti-Ran antibody were performed in TRIM37/wt-, TRIM37/DTBM-, or TRIM37/RF-mu-transfected cells, which were preincubated with BAPTA (20 mmol/L, 30 minutes), and then treated with etoposide (10 mmol/L, 2 hours) and analyzed by immunoblot with the indicated antibody.

www.aacrjournals.org Cancer Res; 78(22) November 15, 2018 6409

Wu et al.

Figure 7. TAT-TRIM37/TBM peptide augments genotoxic stress-induced tumor regression. A, Co-IP assays using anti-TRIM37 antibody were performed in NE1/TRIM37 (left) and Eca-109 (right) cells preincubated with TAT-Ctrl or TAT-37/TBM peptide for 2 h then further treated with CDDP (5 mmol/L, 2 hours), and analyzed by immunoblot with anti-NEMO and anti-TRAF6 antibodies. B, NF-kB DNA-binding and IKK activities and expression of monoubiquitinated NEMO in the indicated cells preincubated with TAT-Ctrl or TAT-37/TBM peptide for 2 hours then further treated with etoposide (Etop; 10 mmol/L, 2 hours) or CDDP (5 mmol/L, 2 hours). The expression of BCL-XL and XIAP was examined at 6 hours after indicated treatment. OCT-1 DNA-binding complex served as a DNA-binding control and GST-IkBa or þ a-tubulin was used as loading control. C, Representative pictures (left) and quantiﬁcation (right) of activated caspase-3 -cells in the indicated cells preincubated with TAT-Ctrl or TAT-37/TBM peptide for 2 hours, then further treated with or without CDDP (5 mmol/L, 24 hours). Scale bars, 50 mm. D, Representative pictures (left) and quantiﬁcation (right) of colony numbers of indicated cells as determined by an anchorage-independent growth assay. E, Representative images of tumors from the PDX model cotreated with CDDP (5 mg/kg, three times per week for up to 6 weeks) plus TAT-Ctrl or TAT-37/TBM peptide (top); tumor volumes were examined on the indicated days (bottom). Detailed information for T#6 and T#9 is shown in Fig. 1D. F, NF-kB DNA-binding activity and expression of monoubiquitinated NEMO and activated caspase-3 were examined in the indicated tumors. OCT-1 DNA-binding complex served as a DNA-binding control and a-tubulin was used as a loading control. G, IHC staining of nuclear NF-kB p65 and TUNEL-positive cells in the indicated tumors. Each bar in D and G represents the mean SD of three independent experiments. , P < 0.05; , P < 0.01; , P < 0.001. Scale bars, 50 mm.

6410 Cancer Res; 78(22) November 15, 2018 Cancer Research

TRIM37 Activates NF-kB Signaling via NEMO Monoubiquitination

monoubiquitinated NEMO, which complexes with ATM and apoptotic transcription. Treatment with a cell-penetrating TAT- ELKS, was exported from the nucleus and was shown to be TBM peptide, which blocked interaction of TRIM37/TRAF6, abro- essential for cytoplasmic IKK activation (18). Herein, we found gated genotoxic stress-induced NEMO monoubiquitination and that in response to genotoxic stress, it took nearly 30 minutes for NF-kB activation, resulting in hypersensitivity of cancer cells to peroxisomal E3 ligase TRIM37 to translocate into the nucleus genotoxic chemotherapy. Therefore, our findings not only reveal a where it associated TRAF6 and NEMO, which resulted in NEMO crucial role of TRIM37 in genotoxic stress-induced NF-kB activa- monoubiquitination at K309. Therefore, the DNA damage- tion, but also have important translational implications for the triggered slow NF-kB signal may be caused by TRIM37 nuclear mechanistic understanding of therapeutic TRIM37 inhibitors that translocation. Importantly, we demonstrated that ATM kinase, a can potentate the effect of chemotherapeutic drugs or ionizing sensor of DNA damage, played an important role in genotoxic radiation in cancer therapy. stress-induced nuclear translocation of TRIM37 via direct physi- In conclusion, preventing genotoxic stress-induced NF-kB acti- cal interaction with and phosphorylation of TRIM37. Interesting- vation, which results in development of chemotherapy resistance, ly, two ATM-mediated phosphorylation sites are present in will be beneficial for a large group of patients with cancer. Herein, TRIM37, in which T196Q is within the peroxisomal targeting we demonstrated that genotoxic stress-induced E3 ligase TRIM37 signal of TRIM37, and S801Q is in proximity to the nuclear contributed to NEMO monoubiquitination and genotoxic IKK/ localization signal of TRIM37. These results suggested that NF-kB activation, consequently leading to genotoxic stress resis- ATM-mediated phosphorylation may lead to dissociation of tance of esophageal cancer. Therefore, further investigation into TRIM37 from the peroxisome and nuclear transition via exposure the role of TRIM37 in resistance of chemotherapy-induced geno- of the TRIM37 nuclear localization signal. Although it has been toxic stress will not only provide valuable insights to better reported that genotoxic stress-activated nuclear ATM translocated understand imitation and progression of cancers but also may into cytosolic and membrane fractions within 10 minutes (36), eventually lead to the development of novel therapeutic strategies we found that antioxidants prevented genotoxic stress-induced for treatment of human cancers. ATM/TRIM37 interaction and TRIM37 nuclear translocation, suggesting that oxidative stress-activated ATM, instead of DNA Disclosure of Potential Conflicts of Interest damage-activated ATM, contributed to phosphorylation of No potential conflicts of interest were disclosed. TRIM37. This was consistent with previous reports that oxidative stress could induce activation of peroxisomal ATM kinase (30, Authors' Contributions 31). Therefore, our study provided a novel mechanism and role Conception and design: L. Song, J. Li for ATM in genotoxic stress-induced NF-kB activation. Development of methodology: G. Wu, J. Zhu, The TRIM37 gene on chromosome 17q22–23 was originally Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Wu, J. Zhu, Y. Hu, L. Cao, Z. Tan, S. Zhang, Z. Li, found to be frequently mutated in patients with mulibrey nanism, Writing, review, and/or revision of the manuscript: L. Song, J. Li a disease with dramatic growth impairment in several organs (19, Study supervision: J. Li 20). Further studies demonstrated that the roles of TRIM37 in various biological processes depend on TRIM domain-dependent Acknowledgments E3 ligase activity. For instance, TRIM37 is involved in peroxisomal This work was supported by Natural Science Foundation of China matrix protein import via monoubiquitination of PEX5 (22). [No. 81830082, 91740119, 91529301, and 81621004 (all to J. Li); Enforced expression of a TRIM37 mutant that lacks E3 ligase 91740118, 81773106, and 81530082 (all to L. Song)]; Guangzhou Science activity could not prevent the TRIM37 depletion-resulted super- and Technology Plan Projects (201803010098 to J. Li); Guangdong Natural Science Foundation (2018B030311009 to J. Li; 2016A030308002 to numerary centrosomal-component foci (37). Although multiple L. Song); The Fundamental Research Funds for the Central Universities studies reported that TRIM37 is mainly distributed in the cyto- (No. 17ykjc02 to J. Li). plasm, such as in peroxisomes (20–22), Bhatnagar and colleagues found that TRIM37 associated with PRC complex in the nucleus to The costs of publication of this article were defrayed in part by the establish a repressive chromatin structure (23), suggesting that payment of page charges. This article must therefore be hereby marked TRIM37 could translocate into the nucleus. Herein, we found that advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate in response to DNA damage, ATM-mediated phosphorylation of this fact. TRIM37 led to its rapid translocation into the nucleus, where it forms a TRIM37/TRAF6/NEMO complex that catalyzes NEMO Received July 5, 2018; revised August 28, 2018; accepted September 21, 2018; monoubiquitination, ultimately leading to NF-kB-mediated anti- published first September 25, 2018.

References 1. Lindahl T, Barnes DE. Repair of endogenous DNA damage. Cold Spring 6. Ciccia A, Elledge SJ. The DNA damage response: making it safe to play with Harb Symp Quant Biol 2000;65:127–33. knives. Mol Cell 2010;40:179–204. 2. Harper JW, Elledge SJ. The DNA damage response: ten years after. Mol Cell 7. Shiloh Y. ATM and related protein kinases: safeguarding genome integrity. 2007;28:739–45. Nat Rev Cancer 2003;3:155–68. 3. Jackson SP, Bartek J. The DNA-damage response in human biology and 8. McCool KW, Miyamoto S. DNA damage-dependent NF-kappaB activation: disease. Nature 2009;461:1071–8. NEMO turns nuclear signaling inside out. Immunol Rev 2012;246:311–26. 4. Ward JF. DNA damage produced by ionizing radiation in mammalian cells: 9. Miyamoto S. Nuclear initiated NF-kappaB signaling: NEMO and ATM take identities, mechanisms of formation, and reparability. Prog Nucleic Acid center stage. Cell Res 2011;21:116–30. Res Mol Biol 1988;35:95–125. 10. Karin M. Nuclear factor-kappaB in cancer development and progression. 5. Kawanishi S, Hiraku Y, Pinlaor S, Ma N. Oxidative and nitrative DNA Nature 2006;441:431–6. damage in animals and patients with inﬂammatory diseases in relation to 11. Hayden MS, Ghosh S. Shared principles in NF-kappaB signaling. Cell inﬂammation-related carcinogenesis. Biol Chem 2006;387:365–72. 2008;132:344–62.

www.aacrjournals.org Cancer Res; 78(22) November 15, 2018 6411

Wu et al.

12. Huang TT, Wuerzberger-Davis SM, Wu ZH, Miyamoto S. Sequential 25. Hahn WC, Dessain SK, Brooks MW, King JE, Elenbaas B, Sabatini DM, et al. modification of NEMO/IKKgamma by SUMO-1 and ubiquitin med- Enumeration of the simian virus 40 early region elements necessary for iates NF-kappaB activation by genotoxic stress. Cell 2003;115: human cell transformation. Mol Cell Biol 2002;22:2111–23. 565–76. 26. Song L, Gong H, Lin C, Wang C, Liu L, Wu J, et al. Flotillin-1 promotes 13. Stilmann M, Hinz M, Arslan SC, Zimmer A, Schreiber V, Scheidereit C. A tumor necrosis factor-alpha receptor signaling and activation of NF- nuclear poly(ADP-ribose)-dependent signalosome confers DNA damage- kappaB in esophageal squamous cell carcinoma cells. Gastroenterology induced IkappaB kinase activation. Mol Cell 2009;36:365–78. 2012;143:995–1005.e12. 14. Janssens S, Tinel A, Lippens S, Tschopp J. PIDD mediates NF-kappaB 27. Gammon ST, Villalobos VM, Prior JL, Sharma V, Piwnica-Worms D. activation in response to DNA damage. Cell 2005;123:1079–92. Quantitative analysis of permeation peptide complexes labeled with 15. Wu ZH, Shi Y, Tibbetts RS, Miyamoto S. Molecular linkage between the Technetium-99m: chiral and sequence-specific effects on net cell uptake. kinase ATM and NF-kappaB signaling in response to genotoxic stimuli. Bioconjug Chem 2003;14:368–76. Science 2006;311:1141–6. 28. Berchtold CM, Wu ZH, Huang TT, Miyamoto S. Calcium-dependent 16. Wu Z, Wang C, Bai M, Li X, Mei Q, Li X, et al. An LRP16-containing regulation of NEMO nuclear export in response to genotoxic stimuli. preassembly complex contributes to NF-kappaB activation induced by Mol Cell Biol 2007;27:497–509. DNA double-strand breaks. Nucleic Acids Res 2015;43:3167–79. 29. Hwang B, McCool K, Wan J, Wuerzberger-Davis SM, Young EW, Choi EY, 17. Mabb AM, Wuerzberger-Davis SM, Miyamoto S. PIASy mediates NEMO et al. IPO3-mediated nonclassical nuclear import of NF-kB essential sumoylation and NF-kappaB activation in response to genotoxic stress. modulator (NEMO) drives DNA damage-dependent NF-kB Activation. Nat Cell Biol 2006;8:986–93. J Biol Chem 2015;290:17967–84. 18. Wu ZH, Wong ET, Shi Y, Niu J, Chen Z, Miyamoto S, et al. ATM- and NEMO- 30. Zhang J, Tripathi DN, Jing J, Alexander A, Kim J, Powell RT, et al. ATM dependent ELKS ubiquitination coordinates TAK1-mediated IKK activa- functions at the peroxisome to induce pexophagy in response to ROS. tion in response to genotoxic stress. Mol Cell 2010;40:75–86. Nat Cell Biol 2015;17:1259–69. 19. Avela K, Lipsanen-Nyman M, Idanheimo N, Seemanova E, Rosengren S, 31. Tripathi DN, Zhang J, Jing J, Dere R, Walker CL. A new role for ATM in Makela TP, et al. Gene encoding a new RING-B-box-Coiled-coil protein is selective autophagy of peroxisomes (pexophagy). Autophagy 2016;12: mutated in mulibrey nanism. Nat Genet 2000;25:298–301. 711–2. 20. Kallijarvi J, Avela K, Lipsanen-Nyman M, Ulmanen I, Lehesjoki AE. The 32. Ye H, Arron JR, Lamothe B, Cirilli M, Kobayashi T, Shevde NK, et al. Distinct TRIM37 gene encodes a peroxisomal RING-B-box-coiled-coil protein: molecular mechanism for initiating TRAF6 signalling. Nature 2002;418: classification of mulibrey nanism as a new peroxisomal disorder. 443–7. Am J Hum Genet 2002;70:1215–28. 33. Koren E, Apte A, Sawant RR, Grunwald J, Torchilin VP. Cell-penetrating TAT 21. Zapata JM, Pawlowski K, Haas E, Ware CF, Godzik A, Reed JC. A diverse peptide in drug delivery systems: proteolytic stability requirements. family of proteins containing tumor necrosis factor receptor-associated Drug Deliv 2011;18:377–84. factor domains. J Biol Chem 2001;276:24242–52. 34. Rizzuti M, Nizzardo M, Zanetta C, Ramirez A, Corti S. Therapeutic applica- 22. Wang W, Xia ZJ, Farre JC, Subramani S. TRIM37, a novel E3 ligase for tions of the cell-penetrating HIV-1 Tat peptide. Drug Discov Today PEX5-mediated peroxisomal matrix protein import. J Cell Biol 2017; 2015;20:76–85. 216:2843–58. 35. Wang W, Mani AM, Wu ZH. DNA damage-induced nuclear factor-kappa B 23. Bhatnagar S, Gazin C, Chamberlain L, Ou J, Zhu X, Tushir JS, et al. TRIM37 activation and its roles in cancer progression. J Cancer Metast Treat is a new histone H2A ubiquitin ligase and breast cancer oncoprotein. 2017;3:45–59. Nature 2014;516:116–20. 36. Jin HS, Lee DH, Kim DH, Chung JH, Lee SJ, Lee TH. cIAP1, cIAP2, and XIAP 24. Deng W, Tsao SW, Guan XY, Lucas JN, Si HX, Leung CS, et al. Distinct act cooperatively via nonredundant pathways to regulate genotoxic stress- profiles of critically short telomeres are a key determinant of different induced nuclear factor-kappaB activation. Cancer Res 2009;69:1782–91. chromosome aberrations in immortalized human cells: whole- 37. Kallijarvi J, Lahtinen U, Hamalainen R, Lipsanen-Nyman M, Palvimo JJ, genome evidence from multiple cell lines. Oncogene 2004;23: Lehesjoki AE. TRIM37 defective in mulibrey nanism is a novel RING finger 9090–101. ubiquitin E3 ligase. Exp Cell Res 2005;308:146–55.

6412 Cancer Res; 78(22) November 15, 2018 Cancer Research

An ATM/TRIM37/NEMO Axis Counteracts Genotoxicity by Activating Nuclear-to-Cytoplasmic NF- κB Signaling

Geyan Wu, Libing Song, Jinrong Zhu, et al.

Cancer Res 2018;78:6399-6412. Published OnlineFirst September 25, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-18-2063

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2018/09/25/0008-5472.CAN-18-2063.DC1

Cited articles This article cites 37 articles, 8 of which you can access for free at: http://cancerres.aacrjournals.org/content/78/22/6399.full#ref-list-1

Citing articles This article has been cited by 4 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/78/22/6399.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/78/22/6399. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.