Oncogene https://doi.org/10.1038/s41388-018-0621-5

ARTICLE

Targeting TRIM3 deletion-induced tumor-associated lymphangiogenesis prohibits lymphatic metastasis in esophageal squamous cell carcinoma

1,2 1,3 4 1 2 1 1 1 Jinrong Zhu ● Geyan Wu ● Zunfu Ke ● Lixue Cao ● Miaoling Tang ● Ziwen Li ● Qiaojia Li ● Junhao Zhou ● 1 3 1,2 Zhanyao Tan ● Libing Song ● Jun Li

Received: 23 July 2018 / Accepted: 23 November 2018 © Springer Nature Limited 2018

Abstract Tumor-associated lymphangiogenesis has attracted increasing attention because of its potential contribution to lymph node metastasis. However, the molecular mechanisms underlying lymphangiogenesis in cancer remains elusive. In the current study, we demonstrate that tripartite motif-containing 3 (TRIM3) directly interacts with and induces E3 ligase-dependent proteasomal turnover of importin α3 and α-Actinin-4 (ACTN4), which controls nuclear factor kappa B (NF-κB) activity at a well-ordered level. Heterozygous deletion-mediated TRIM3 downregulation led to NF-κB constitutive activation through 1234567890();,: 1234567890();,: disruption of the NF-κB-IκB-α negative feedback loop and enhancement of the p65 DNA-binding affinity and transcriptional activity via promoting symmetrical dimethylarginine modification of NF-κB/p65 at Arg30 and Arg35, which consequently promoted lymphatic metastasis of esophageal squamous cell carcinoma (ESCC) cells. Treatment with Tecfidera, a medication used to treat multiple sclerosis, restored the negative feedback inhibition of NF-κB by reducing the NF-κB/ ACTN4 interaction and decreasing symmetrically dimethylated NF-κB levels, resulting in inhibition of ESCC lymphatic metastasis both in vitro and in vivo. Taken together, our results uncover a novel mechanism for constitutive NF-κB activation in cancer and may represent an attractive strategy to treat ESCC lymphatic metastasis.

Introduction

The significant contribution of tumor-associated lym- phangiogenesis to lymph node (LN) metastasis and distant metastasis, which results in poor prognosis of cancer These authors contributed equally: Jinrong Zhu, Geyan Wu, Zunfu Ke patients with lymphatic metastasis, has attracted increasing – Supplementary information The online version of this article (https:// attention over the last few decades [1 3]. For instance, the doi.org/10.1038/s41388-018-0621-5) contains supplementary 5-year overall survival rate inpatients with esophageal material, which is available to authorized users. squamous cell carcinoma (ESCC) with lymphatic metastasis is only 10–15%, and the median survival time is less than 1 * Jun Li [email protected] year. In addition, patients with ESCC develop lymphatic metastasis at an early stage. LN metastasis has been iden- 1 Key Laboratory of Liver Disease of Guangdong Province, The tified as the most important predictor of prognosis in ESCC fi Third Af liated Hospital, Sun Yat-sen University, Guangzhou, [4–6]. Many studies have demonstrated that tumor- Guangdong, China associated lymphangiogenesis and lymphatic metastasis in 2 Department of biochemistry, Zhongshan School of medicine, Sun a variety of malignancies were largely attributed to upre- Yat-sen University, Guangzhou, Guangdong, China gulated vascular endothelial growth factor C (VEGF-C) [7– 3 State Key Laboratory of Oncology in South China, Cancer Center, 10]. A VEGF-C monoclonal antibody (VGX-100) was Collaborative Innovation Center for Cancer Medicine, Guangzhou, Guangdong, China examined in a clinical trial for malignant tumors that were 4 fi refractory to standard therapy. However, treatment with the Department of Pathology, The First Af liated Hospital, Sun Yat- VEGFC sen University, Guangzhou, Guangzhou, Guangdong 510060, VEGF-C monoclonal antibody or silencing of only China elicited a 60–70% reduction in lymphatic metastasis [11, J. Zhu et al.

12]. Therefore, the identification of novel therapeutic target fluorescence in situ hybridization (FISH) analysis of our (s) to inhibit tumor-associated lymphangiogenesis might collected ESCC specimens (n = 308) and by analysis of The represent a more effective strategy for the clinical inter- Cancer Genome Atlas (TCGA)-ESCC data set (n = 77) vention in cancer lymphatic metastasis. using GISTIC 2.0 (Genomic Identification of Significant Previous studies suggested that nuclear factor kappa B Targets in Cancer). Similar to the TCGA data set analysis (NF-κB) is abnormally activated in ESCC, and plays vital (38.59%), the FISH analysis revealed 36.04% heterozygous roles in lymphangiogenesis and lymphatic metastasis [13, TRIM3 deletions in our ESCC specimens (Fig. 1a and 14]. Approaches that terminate the hyperactivation of NF- Supplementary Figure S1a). Interestingly, an immunohis- κB have long been investigated as potential treatments for tochemical assay revealed that the TRIM3 was cancer [15, 16]. However, long-term treatment with most strongly expressed in normal esophageal tissues; however, NF-κB inhibitors leads to broad suppression of innate its level was noticeably lower in TRIM3-nondeleted ESCC, immunity [15, 16]. The anti-inflammatory drug, Tecfidera, and was further substantially reduced in TRIM3-deleted approved for the treatment of multiple sclerosis in the ESCC tissues (Fig. 1a, b). Concordantly, analyses of 15 United States and the European Union, exhibits immune- freshly collected clinical ESCC tissues and the TCGA- modulatory properties [17]. A Tecfidera-mediated NF-κB ESCC data set revealed that TRIM3 mRNA expression in inhibitory effect was attributed to reduction of the DNA- TRIM3-deleted ESCC was lower than that in TRIM3-non- binding activity and nuclear translocation of NF-κB[18– deleted ESCC (Supplementary Figure S1b and c). These 20]. However, the inhibitory effect of Tecfidera on cancer results suggested that genetic deletion represents a major lymphatic metastasis is unknown. mechanism of TRIM3 protein reduction in ESCC. Tripartite motif-containing 3 (TRIM3), also named Consistent with the results obtained from both our col- brain-expressed RING finger protein (BERP) or E3 ligase lected ESCC specimens and the TCGA-ESCC data set, RNF22, belongs to the RING finger-B-box-coiled coil patients with TRIM3-deleted ESCC had a significantly (RBCC) subgroup of RING finger [21–23]. TRIM3 shorter overall and relapse-free survival compared with was originally identified as a protein from the cytoskeleton- those without TRIM3-deletion. We found that the TRIM3 associated recycling or transport complex that is involved in protein-lower cohort had shorter overall and relapse-free receptor recycling [22, 23]. Furthermore, loss of TRIM3 survival compared with the TRIM3 protein-higher cohort contributes to hippocampal plasticity and brain tumorigen- (Fig. 1c and Supplementary Figure S1d and e). Statistical esis, which is dependent on its E3 ligase activity [23, 24]. analysis showed that TRIM3 levels inversely correlated However, the precise molecular mechanism and clinical with clinical stage and tumor-node-metastasis (TNM) clas- significance of TRIM3 remains unclear. sification (Supplementary Table S2). Multiple clin- In the present study, we reported that loss of TRIM3 in icopathological variables, such as gender, age, T ESCC prolonged NF-κB activation via promotion of classification, clinical stage, N classification, M classifica- nuclear accumulation and transcriptional activity of NF-κB, tion, and copy number and expression of TRIM3, were which promoted ESCC lymphatic metastasis. Importantly, included for univariate and multivariate Cox regression treatment with Tecfidera inhibited the TRIM3 deletion- analysis, which showed that TRIM3 expression was an induced NF-κB constitutive activation, resulting in con- independent prognostic factor for survival outcome of siderable inhibition of lymphangiogenesis and LN patients with ESCC (P < 0.001; Supplementary Table S3). metastasis in ESCC. These findings demonstrated that het- The above data demonstrated that heterozygous deletion- erozygous deletion of TRIM3-sustained NF-κB signaling is mediated TRIM3 reduction was associated with ESCC a candidate driver of lymphatic metastasis in ESCC and progression. suggested that Tecfidera has therapeutic value to treat ESCC lymphatic metastasis. TRIM3 reduction correlates with lymphatic metastasis and microlymphatic vessel density

Results Lymphatic metastasis and lymphangiogenesis have been identified as predictors of prognosis in ESCC [4–6]. Strik- Heterozygous deletion-mediated TRIM3 reduction ingly, we observed that TRIM3 reduction was not only correlates with poor prognosis in ESCC significantly associated with lymphatic metastasis (P < 0.001; Fig. 1d), but also correlated with microlymphatic The TRIM3 is located on 11p15, which vessel density (MLD) in both intratumoral and peritumoral frequently undergoes a loss of heterozygosity in human tissues (P < 0.01; P < 0.001) (Fig. 1e and Supplementary tumors, including ESCC [25, 26]. We examined the dele- Table S2), which suggested that TRIM3 reduction probably tion frequency of the TRIM3 gene by performing results in lymphangiogenesis and LN metastasis in ESCC. Targeting TRIM3 deletion-induced tumor-associated lymphangiogenesis prohibits lymphatic metastasis in. . .

Fig. 1 Heterozygous deletion of TRIM3 correlates with poor prognosis TRIM3 (n = 308). d Left panel: two representative cases are shown. in human ESCC. a, b Representative FISH and IHC images (a) and Right panel: percentages of ESCC specimens showing low- or high- quantification of heterozygous TRIM3 deletions and TRIM3 expres- TRIM3 expression relative to lymphatic metastasis, *** P < 0.001. e sion in clinically collected ESCC samples (n = 308) (b). Significant Representative micrographs and the percentages of tissue specimens correlation between TRIM3 protein expression and TRIM3 deletion in with high and low levels Lyve-l+- microvessels in ESCC with low or ESCC. ***P < 0.001. c Kaplan–Meier relapse-free survival curves for high expression of TRIM3. left panel: intratumoral; right panel: peri- ESCC patients with heterozygous deletion versus non-deletion of tumoral; **P < 0.01, ***P < 0.001 TRIM3 and for ESCC patients with low- versus high-expression of

Deletion of TRIM3 promotes lymphatic metastasis of compared with control tumors (Fig. 2b). Importantly, het- ESCC in vivo erozygous deletion of TRIM3 significantly enhanced, whereas upregulation of TRIM3 reduced, the lymph meta- To further determine the effect of TRIM3 heterozygous static capability of KYSE30 cells compared with that of deletion on lymphangiogenesis as well as LN metastasis, control cells, as determined by the volume and Luc+-tumor TRIM3 heterozygous knockout cells, established in KYSE30 cells in the lymph nodes of mice (Fig. 2c). These results cancer cells using clustered regularly interspaced short indicated that TRIM3-deletion promotes lymphangiogenesis palindromic repeat (CRISPR)-Caspase-9 (Cas-9) (Supple- and lymphatic metastasis in ESCC in vivo. mentary Figure S2a-d), was investigated using the popliteal lymph node metastasis model (Fig. 2a). We observed that Loss of TRIM3 induces lymphangiogenesis and TRIM3+/− tumors displayed a greater number of lymphatic invasiveness of ESCC cells in vitro vessel endothelial hyaluronan receptor 1+ (Lyve-1+) -lym- phatic vessels within both the tumor and tumor-adjacent Furthermore, we observed that TRIM3 deletion strongly tissues, whereas TRIM3/tumors exhibited decreased MLD, enhanced, but overexpressing TRIM3 reduced, the J. Zhu et al.

capability of tube formation, as well as migration, in human cells displayed reduced invasiveness (Fig. 2f, g). These lymphatic endothelial cells (HLECs) in KYSE30 and results indicated that loss of TRIM3 is involved in regula- NE1 cells (Fig. 2d, e), suggesting that TRIM3 is involved in tion of the pro-lymphangiogenic capability and epithelial– lymphangiogenesis under both physiological and patholo- mesenchymal transition (EMT) process of ESCC cells gical conditions. Meanwhile, TRIM3+/−-cells also displayed in vitro. a stronger invasive capacity, whereas TRIM3-transduced Targeting TRIM3 deletion-induced tumor-associated lymphangiogenesis prohibits lymphatic metastasis in. . .

Fig. 2 Heterozygous deletion of TRIM3 promotes lymph node and Supplementary Figure S3c), suggesting that importin metastasis in vivo. a A popliteal lymph node metastasis model was α3 and ACTN4 contribute to TRIM3 deletion-induced NF- established in nude mice by inoculating the foot pads with the indi- κB activation. cated cells. b Representative micrographs of tumor sections immu- nostained with the anti- Lyve-l antibody (left) and quantification (right). c Left, representative micrographs of the popliteal lymph nodes Loss of TRIM3-induced importin α3 upregulation immunostained with anti-luciferase antibody. Right: the ratios of promotes NF-κB nuclear transition metastatic to total dissected popliteal lymph nodes from mice inocu- lated with the indicated cells. d Representative images (left panel) and quantification (right panel) of HLECs cultured on Matrigel coated Reciprocal co-immunoprecipitation (co-IP) and far-western plates with conditioned media from the indicated KYSE30 cells. e blotting analyses demonstrated that TRIM3 directly inter- Representative images (left panel) and quantification (right panel) of acts with importin α3 (Supplementary Figure S3d and e). migrated HLECs cultured with conditioned media from indicated Consistent with TRIM3 E3 ubiquitin ligase function [23, KYSE30 cells. f Representative images (left panel) and quantification (right panel) of invaded KYSE30 and NE1 cells. g Representative 24], loss of TRIM3 increased the expression and half-life of micrographs of indicated cells cultured in a 3D spheroid invasion importin α3 and decreased level of K48-polyubiquitinated assay. Error bar presented in experiments b, d–f represents the mean ± importin α3. And overexpression of TRIM3 showed *P **P SD of three independent experiments. < 0.05, < 0.01 opposite effects. However, dysregulation of TRIM3 had no effect on mRNA expression of importin α3 (Supplementary Heterozygous deletion of TRIM3 sustains NF-κB Figure S3f-h). As expected, TRIM3-induced importin α3 activity reduction was abrogated in proteasome inhibitor-treated cells and in TRIM3 Ring-domain mutant (C22S/C25S) Cignal finder reporter arrays revealed that the heterozygous overexpressing cells (Supplementary Figure S3i and j). deletion of TRIM3 in both NE1 and KYSE30 cells resulted These results indicated that TRIM3 stimulates proteasome- in significant NF-κB activation (Fig. 3a), suggesting that dependent degradation of importin α3. TRIM3 might contribute to modulating NF-κB signaling. Consistent with the effect of importin α3 on NF-κB p65/ This hypothesis was further confirmed by multiple assays, p50 nuclear import [27], TRIM3+/−-cells showed increased in which the heterozygous deletion of TRIM3 significantly levels of the p65/importinβ1/α3 complex, elevated levels of increased, but overexpression of TRIM3 reduced, the DNA- NF-κB p65 in the nuclear membrane, and more rapid TNF- binding activity of NF-κB, the expression of numerous α-induced NF-κB nuclear transition compared with those in well-known NF-κB targets, NF-κB nuclear expression, and the control cells (Supplementary Figure S4a-c). Moreover, NF-κB-driven luciferase activity (Fig. 3b–e and Supple- we found that the nuclear expression of NF-κB upon TNF-α mentary Figure S3a). The inverse correlation between treatment was dramatically increased in the TRIM3-deleted TRIM3 expression and NF-κB activation was further cells compared with that in the control cells, whereas the determined in clinical ESCC specimens and using a gene set promotive effect of TRIM3 deletion on NF-κB nuclear enrichment analysis (GSEA) (Fig. 3f, g). We also observed transition was abolished by silencing the expression of that NF-κB inhibitor alpha (IκB-α) mRNA expression importin α3, resulting in the suppression of NF-κB trans- induced by tumor necrosis factor alpha (TNF-α) or inter- activation (Supplementary Figure S4c and d). These results leukin (IL)-1β was significantly prolonged in TRIM3 suggested that TRIM3 deletion-induced importin α3 upre- +/−-cells, but rapidly reduced in TRIM3-transduced cells gulation promotes the nuclear transition of NF-κB. (Supplementary Figure S3b), demonstrating that the loss of TRIM3 sustains NF-κB activity. ACTN4 abolishes the negative feedback of IκB-α on NF-κB ACTN4 and importin α3 are TRIM3-interacting proteins Strikingly, we found that TNF-α pretreatment-stimulated nuclear accumulation of NF-κB persisted much longer in However, we did not observe any obvious alterations in TRIM3+/−-cells, but deceased rapidly in TRIM3-transduced phosphorylated-IκB-α and phosphorylated inhibitor of cells after the removal of TNF-α (Fig. 4c and Supplemen- nuclear factor kappa B kinase subunit beta (IKKβ) between tary Figure S5a). However, silencing ACTN4, but not TRIM3+/−-cells and control cells (Fig. 3e), indicating that importin α3, significantly abrogated TRIM3 deletion- TRIM3 dysregulation has no impact on upstream IKK induced NF-κB nuclear accumulation (Fig. 4c), suggesting activation. Furthermore, we found that among the multiple that ACTN4 is involved in NF-κB nuclear export. potent TRIM3-binding proteins identified by affinity pur- NF-κB termination is achieved by IκBα by removing ification/mass spectrometry (IP/MS), only individual silen- NF-κB from DNA and exporting NF-κB back to the cyto- cing of importin α3orα-Actinin-4 (ACTN4) significantly plasm [27–29]. Interestingly, overexpression of ACTN4 decreased NF-κB activity in TRIM3-deleted cells (Fig. 4a, b reduced the interaction of p65 with IκB-α in a dose- J. Zhu et al.

Fig. 3 Heterozygous deletion of TRIM3 sustains NF-κB activity in by a log2 transformation. d Subcellular localization of NF-κB p65 in normal esophageal epithelial and ESCC cells. a Cignal finder reporter the indicated cells as analyzed by an immunofluorescence staining arrays showing that heterozygous deletion of TRIM3 significantly assay. e Western blotting analysis of level of cytoplasmic-p65, activated NF-κB signaling in the KYDE30 and NE1 cells. Error bars nuclear-p65, p- IKK-β, total IKK-β, p-IκB-α, and total IκB-α in the represent the mean ± SD from three independent experiments, **P < indicated cells treated with TNF-α (2 ng/ml), β-actin served as a 0.01, *P < 0.05. b EMSA of the endogenous NF-κB activity in the loading control. f TRIM3 expression was inversely associated with indicated TRIM3+/−, or TRIM3-transduced and control cells. Oct-1/ NF-κB activity, as determined by an EMSA analysis, in nine collected DNA-binding complex was used as a control, cells treated with TNF-α samples. Lower panel, the quantification of the relative NF-κB activity (2 ng/ml). c Real-time PCR analysis indicating an apparent overlap and the relative expression of TRIM3 (r = -0.842, P<0.001). g GSEA between NF-κB-dependent gene expression and TRIM3-regulated analysis showing that TRIM3 expression was inversely correlated with gene expression. The pseudocolor represents the intensity scale of NF-κB target gene signatures in two published ESCC data sets TRIM3 versus the vector, or TRIM3+/− versus the control, generated dependent manner; however, ACTN4 silencing and TRIM3- further confirmed using a chromatin fraction assay, in which deletion dramatically decreased the p65/IκB-α association, the level of DNA-bound NF-κB was increased in TRIM3- resulting in decreased nuclear expression and transactiva- deleted and ACTN4-overexpressing cells, but decreased in tion of NF-κB (Fig. 4d and Supplementary Figure S5b-d), TRIM3-transduced and ACTN4-silenced cells (Fig. 4e). which suggested that ACTN4 abolished the negative feed- Consistent with previous studies [30, 31], a co-IP assay back of IκBα on NF-κB activity. This hypothesis was showed that ACTN4 also forms a complex with NF-κB/p65 Targeting TRIM3 deletion-induced tumor-associated lymphangiogenesis prohibits lymphatic metastasis in. . .

Fig. 4 TRIM3 interacts with and induces proteasome-dependent overexpressing ACTN4 reduced the p65/IκB-α interaction in a dose- degradation of ACTN4. a Lysates from KYSE30/ Flag-TRIM3 cells dependent manner (left) whereas ACTN4 silencing dramatically were immunoprecipitated using anti-Flag affinity agarose, followed by increased p65/IκB-α association in TRIM3+/−-cells. e Chromatin mass-spectrometric peptide sequencing. b NF-κB-driven luciferase fraction and western blotting analysis of DNA-bound NF-κB. activity following individual silencing potent TRIM3-interacting pro- H3 served as a loading control. f Co-IP assay showing that endogenous teins in KYSE30 and NE1 cells treated with TNF-α (2 ng/ml). Error p65 interacted with endogenous ACTN4 in the nucleus. g Schematic bars represent the mean ± SD from three independent experiments, **P illustration of wild-type and truncated NF-κB (upper). Co-IP assay < 0.01, *P < 0.05. c Immunofluorescence staining of NF-κB in the showing that ACTN4 interacted with the DNA-binding domain of p65 indicated cells treated with TNF-α (10 ng/ml) for 20 min and then (lower) replaced by medium without TNF-α. d Co-IP assay showing that in ESCC cells. However, the ACTN4/NF-κB interaction TRIM3 induces proteasome-dependent degradation only occurred in the nucleus and not in the cytoplasm of of ACTN4 TNF-α-treated cells. ACTN4 interacts with the N-terminal DNA-binding domain of p65, which is also the IκB-α- Consistently, we found that TRIM3 interacted with ACTN4 binding region (Fig. 4f, g), further supporting the notion in vitro and in vivo, which resulted in reduced expression that ACTN4 competitively inhibits the negative feedback of and half-life of ACTN4, but increased levels of K48- IκB-α on NF-κB. polyubiquitinated ACTN4 (Supplementary Figure S5e-i). J. Zhu et al.

However, treatment with proteasome inhibitors recovered suggested that the TRIM3/ACTN4/PRMT5/NF-κB axis the ACTN4 level in TRIM3+/−-cells, and overexpressing the plays an important role in VEGF-C upregulation. The TRIM3 mutant (C22S/C25S) had no effect on ACTN4 relevance of this observation was further confirmed in expression (Supplementary Figure S5g, 5j and S1k). These clinical human ESCC specimens, in which the protein results suggested that TRIM3 induces proteasome- expression of TRIM3 in 308 ESCC tissues was inversely dependent degradation of ACTN4. correlated with importin α3(P = 0.023), ACTN4 (P = 0.011), nuclear-p65 (P < 0.001), and VEGF-C levels (P < ACTN4 promotes TRIM3 deletion-induced NF-κB 0.001) (Fig. 5g). dimethylation Tecfidera inhibits TRIM3 deletion-induced lymphatic The DNA-binding activity of NF-κB is controlled by pro- metastasis tein arginine methyltransferase 5 (PRMT5)-mediated sym- metric arginine dimethylation (SDMA) [32]. Interestingly, Tecfidera was previously reported to exert an inhibitory the SDMA level of NF-κB was drastically increased in effect on NF-κB signaling via decreasing the NF-κB/DNA- TRIM3+/−-cells and ACTN4-overexpressing cells, but binding affinity [18–20]. Therefore, whether Tecfidera has decreased in ACTN4-silenced TRIM3+/−- ESCC cells (Fig. therapeutic value in treatment of lymphatic metastasis was 5a), suggesting that ACTN4 is involved in PRMT5-induced examined. Strikingly, we found that Tecfidera treatment NF-κB SDMA. This hypothesis was supported by a co-IP drastically inhibited the ACTN4/NF-κB interaction, reduced assay in which overexpressing ACTN4 promoted, but dimethylated-NF-κB levels, decreased the DNA-binding silencing ACTN4 reduced, the NF-κB/PRMT5 interaction; and transcriptional activity of NF-κB, and reduced VEGF-C however, ablating p65 decreased the ACTN4/PRMT5 and matrix metalloproteinase 9 (MMP9) expression in interaction (Fig. 5b, c). As expected, silencing of either TRIM3+/−-deleted cells (Fig. 6a–d), resulting in a remark- PRMT5 or p65 reduced the TRIM3 deletion-induced NF-κB able inhibitory effect on TRIM3 deletion-induced pro-lym- activation (Fig. 5d). Importantly, forced expression of p65/ phangiogenesis and invasiveness of ESCC cells (Fig. 6e–g). R174A completely recovered the reduced NF-κB transac- Furthermore, in vivo experiments revealed that Tecfidera- tivation, whereas p65/R30A or p65/R35A only partially treated TRIM3+/−/tumors exhibited dramatically reduced recovered it, and p65/R30A/R35A completely failed to expression of nuclear NF-κB and VEGF-C, and decreased recover it, which indicated that SDMA of NF-κB R30/35 is small Lyve-1+-lymphatic vessels within both the tumor and essential for TRIM3 deletion-induced transactivation of NF- tumor-adjacent tissues compared with those in the control κB (Fig. 5d). Consistently, the promoting effects of TRIM3- tumors (Fig. 7a). Consistently, the proportion of metastatic deletion on lymphangiogenesis and invasiveness were lymph nodes in vehicle-treated TRIM3+/− mice (93.7% and drastically reduced by silencing PRMT5 or ACTN4 (Fig. 5e, 87.5%) were significantly reduced to 6.25% and 0% in f), suggesting that NF-κB dimethylation is required for Tecfidera -treated TRIM3+/− mice, when the foot-pad pri- TRIM3 deletion-induced pro-lymphangiogenesis and inva- mary tumors grew to the same size in the Tecfidera and siveness of ESCC cells. control-treated mice (Fig. 7b). Furthermore, Tecfidera- treated-TRIM3+/- tumors exhibited dramatically reduced VEGF-C contributes to TRIM3 deletion-induced expression of nuclear NF-κB and VEGF-C compared with lymphangiogenesis and LN metastasis that in the control tumors (Fig. 7c). These results demon- strated that Tecfidera has potential therapeutic value to treat Previous reports demonstrated that VEGF-C play vital role lymphatic metastasis in ESCC (Fig. 7d). in lymphangiogenesis and LN metastatic [4–6]. NF-κBisa key transcriptional regulator of VEGFC and the mRNA expression of VEGFC was significantly increased in TRIM3 Discussion +/−-cells, but decreased in TRIM3-transduced cells (Fig. 3c); therefore, we examined whether NF-κB contributes to Tumor-associated lymphangiogenesis is the rate-limiting TRIM3 deletion-induced VEGF-C upregulation. As shown step for lymphatic metastasis, and plays an important role in in Supplementary Figure S6a-f, the VEGF-C protein level the development of lymphatic vessels within and adjacent to and enrichment of ACTN4, NF-κB, dimethylated-NF-κB, tumors, thus promoting tumor cell metastasis to lymph and VEGF-C promoter H3-acylation levels were sig- nodes [33, 34]. Herein, we reported that treatment with nificantly increased in TRIM3+/−-ESCC cells and decreased Tecfidera, a medication used clinically to treat multiple in TRIM3-transduced cells. In addition, symmetric arginine sclerosis, showed more inhibitory effects on ESCC lym- dimethylated NF-κB was essential for the inductive effect of phangiogenesis and lymph node metastasis, demonstrating ACTN4 and PRMT5 on VEGF-C upregulation, which that Tecfidera may have therapeutic value to treat ESCC Targeting TRIM3 deletion-induced tumor-associated lymphangiogenesis prohibits lymphatic metastasis in. . .

Fig. 5 Symmetrical dimethylarginine modification of NF-κBis transfected with p65/R30A/R35A, or p65/R30A, or p65/R35A, or p65/ required for TRIM3 deletion-induced VEGF-C upregulation. a R174A. e Representative images (left panel) and quantification (right Immunoprecipitated endogenous p65 was probed by western blotting panel) of HLECs cultured on Matrigel coated plates with conditioned for the SDMA of p65 in the indicated cells treated with TNF-α (2 ng/ media from the indicated KYSE30 and NE1 cells. f Representative ml). b Co-IP assay showing that overexpressing ACTN4 increased the images (left panel) and quantification (right panel) of invaded KYSE30 PRMT5/p65 interaction in a dose-dependent manner (left) whereas and NE1 cells. g TRIM3 levels were negatively associated with silencing ACTN4 decreased the PRMT5/p65 association in TRIM3 importin α3, ACTN4, nuclear p65, or VEGF-C expression in 308 +/−-cells. c Co-IP assay of the interaction of PRMT5 with ACTN4 in primary ESCC specimens. Error bar presented in experiments d–g the control and p65-silenced cells. d Luciferase-reported NF-κB represents the mean ± SD of three independent experiments. *P < 0.05, activity in PRMT5-orp65-silenced cells or in p65-silenced cells **P < 0.01. ***P < 0.001 J. Zhu et al.

lymphatic metastasis. Meanwhile, consistent with the con- dual role in the inhibition of lymphatic metastasis via sup- cept that LN metastasis is a complex multistep process, pression of tumor-induced lymphangiogenesis and the which involves in lymphangiogenesis, dissemination of migration and invasiveness of tumor cells. Interestingly, we tumor cells to lymphatic vessels, transport of tumor cells to also found that the lymph nodules in Tecfidera-treated LNs, and settlement and colonization expansion of tumor immunodeficient nude mice were much smaller compared cells in LNs [1–3], we found that Tecfidera treatment also with those in vehicle-treated mice, suggesting that dramatically reduced the invasive capability of ESCC cells. Tecfidera-mediated inhibition of lymphatic metastasis might Therefore, our results suggested that Tecfidera might play be independent of its immune-modulatory properties. Targeting TRIM3 deletion-induced tumor-associated lymphangiogenesis prohibits lymphatic metastasis in. . .

Fig. 6 Tecfidera inhibits TRIM3 deletion -induced prolymphangio- Previously, it was reported that ACTN4 was predominantly genic and invasive capabilities of ESCC cells in vtrio. a Co-IP assay expressed in cellular protrusions and is essential for their fi μ showing that Tec dera (20 M) treatment reduced p65/ACTN4 inter- formation [37, 38]. Interestingly, when expressed in cell action and decreased dimethylated-p65 expression but increased p65/ IκB-α interaction in the KYSE30/TRIM3+/−- and NE1/TRIM3 nuclei, ACTN4 functions as a co-activator of multiple +/−-cells. b EMSA of the endogenous NF-κB activity in the control transcriptional factors, including NF-κB, which is indepen- + − + − and KYSE30/TRIM3 / - and NE1/TRIM3 / -cells treated with either dent of the cytoplasmic actin-binding activity of ACTN4 vehicle or Tecfidera (20 μM). Oct-1/DNA-binding complex was used – κ [30, 31, 39 43]. However, the precise mechanism by which as a control. c Luciferase-reporter NF- B activity in the indicated cells κ treated with vehicle plus TNF-α (2 ng/ml) or Tecfidera plus TNF-α (2 ACTN4 mediates NF- B transactivation remains unclear. ng/ml). d Real-time PCR analysis of VEGF-C mRNA (left panel) and We demonstrated that ACTN4 competitively interacts with MMP-9 mRNA (right panel) expression in the indicated cells treated the N-terminal DNA-binding domain of p65 that binds to α fi α with the vehicle plus TNF- (2 ng/ml) or Tec dera plus TNF- (2 ng/ IκB-α, and promoted PRMT5-mediated symmetric dime- e fi ml). Representative images (left panel) and quanti cation (right κ panel) of HLECs cultured on Matrigel coated plates with conditioned thylation of p65 at R30/R35, which disrupted the NF- B- media from the indicated cells treated with the vehicle plus TNF-α (2 IκB-α negative feedback loop and enhanced the affinity of ng/ml) or Tecfidera plus TNF-α (2 ng/ml). f Representative images p65 for DNA via increasing Van der Waals contacts forces fi (left panel) and quanti cation (right panel) of migrated HLECs cul- [32, 44]. Meanwhile, we found that the heterozygous dele- tured with conditioned media from the indicated cells treated with the TRIM3 vehicle plus TNF-α (2 ng/ml) or Tecfidera plus TNF-α (2 ng/ml). g tion of in ESCC induced ACTN4 upregulation via Representative images (left panel) and quantification (right panel) of proteasome-dependent regulation; therefore, genomics- invaded KYSE30 and NE1 cells treated with the vehicle plus TNF-α guided patient selection and patient monitoring would pro- fi α (2 ng/ml) or Tec dera plus TNF- (2 ng/ml). Error bar presented in vide the maximum benefit for clinical ESCC lymphatic experiments c-g represents the mean ± SD of three independent fi experiments. *P < 0.05, **P < 0.01 metastasis achieved from Tec dera treatment. The loss of TRIM3 heterozygosity has been reported in However, it has been recently reported that after 12 months malignant gliomas, and a neuroblast-specific TRIM3-RNAi of Tecfidera treatment, multiple sclerosis patients showed can induce an adult brain tumor phenotype [24, 45]. We increased monocytes, natural killer cells and naive B cells demonstrated that heterozygous deletion of TRIM3 pro- but decreased (non) class-switched memory B cells and moted ESCC lymphatic metastasis, further supporting the double negative B cells [35, 36]. Since nude mice could tumor-suppressive function of TRIM3. Therefore, addi- also exhibit lower levels of NK cell-mediated cytotoxicity, tional investigations into the role of TRIM3 in inhibition of it is thus worthy to further examine whether Tecfidera human cancers will provide valuable evidence to better treatment could change the percentages of monocytes, nat- understand cancer development and progression, which ural killer cells and other lymphocytes in the nude mice. might eventually lead to the development of novel ther- The negative feedback mechanism by which IκBα-med- apeutic strategies to treat cancer lymphatic metastasis. iates NF-κB termination, achieved by dissociating DNA- bound NF-κB and exporting nuclear NF-κB to the cyto- plasm, has been extensively demonstrated [27–29]. How- Materials and methods ever, mounting evidence indicates that NF-κB signaling is constitutively activated in various human cancer types [15, Cell lines 16], suggesting that the inhibitory effect of IκBα on NF-κB activation is abrogated in cancer. In the present study, we The ESCC cell line KYSE30 was obtained from DSMZ, the reported that TRIM3 directly induces K48-polyubiquitina- German Resource Centre for Biological Material. The normal tion/degradation of ACTN4, which competitively interacts esophageal epithelial cell line, NE1, which was immortalized with p65’s binding to IκB-α and increased the symmetrical using human papillomavirus E6/E7/telomerase, was kindly dimethylation of NF-κB and importin α3, which resulted in provided by Professors S.W. Tsao and G. Srivastava (The NF-κB import into the nucleus [27, 30, 31]. Furthermore, we University of Hong Kong, Hong Kong) [46]. HLECs were demonstrated that heterozygous of TRIM3-mediated TRIM3 purchased from ScienCell Research Laboratories (Carlsbad, protein reduction in ESCC cells sustained NF-κB activation CA, USA) and cultured with indicated medium (ECM, Cat. by promoting its nuclear accumulation and enhancing the #1001). All cell lines used in this study were authenticated DNA-binding and transcriptional activity of NF-κB. There- using short tandem repeat (STR) profiling. fore, our results revealed a novel mechanism that abrogates the auto-feedback loop of NF-κB in cancer cells, which Clinical ESCC tissue samples might constitutively maintain the NF-κB activation induced by extracellular stimuli in cancer cells. ESCC tissue samples (n = 308) were collected from ESCC Another key finding in the current study is the novel patients (diagnosed from 2003 to 2012, at the Sun Yat-sen mechanism by which ACTN4 induces NF-κB activation. University cancer center). Fifteen ESCC tissues, which J. Zhu et al.

Fig. 7 Tecfidera inhibits TRIM3 deletion-induced lymph node metas- dissected popliteal lymph nodes from mice inoculated with the indi- tasis in vivo. a Representative micrographs of the indicated tumor cated KYSE30 cells. c Representative micrographs of the indicated sections immunostained with anti-Lyve-l antibody (left) and quantifi- tumor sections immunostained with anti-nuclear p65, anti-VEGF-C cation (right) as indicated by MLD. Error bars represent the mean ± SD and anti-MMP9 antibodies. d Schematic diagram illustrating that loss from three independent experiments, ***P < 0.001. b Left: Repre- of TRIM3 disrupts NF-κB negative feedback and promotes NF-κB sentative micrographs of the popliteal lymph nodes immunostained dimethylation in esophageal squamous cell carcinoma with anti-luciferase antibody. Right: The ratios of metastatic to total were freshly collected, were frozen in liquid nitrogen until approved by the Institutional Research Ethics Committee of required. The stages and classification of ESCC tissue Sun Yat-sen University. The approval of acquisition or use samples were verified histologically by professional of ESCC specimens was obtain by Institutional Research pathologists (Supplementary Table S1). All patients gave Ethics Committee ethical. informed consent prior to the study and this study was Targeting TRIM3 deletion-induced tumor-associated lymphangiogenesis prohibits lymphatic metastasis in. . .

Western blotting analysis variables were evaluated using Student’s t test. The rela- tionship between TRIM3 expression and the clin- Western blotting (WB) was performed using anti-TRIM3 icopathological characteristics were carried out using the (Novus), anti-ACTN4 (Abcam), anti-importin α3 (Abcam), chi-squared test, log-rank analysis, the Kaplan–Meier anti-PRMT5 (Abcam), anti-dimethyl-arginine, symmetric method, and the Cox regression model. P < 0.05 was con- (SYM10, Millipore), anti-Flag, anti-myc, anti-HA (Sigma); sidered statistically significant. anti- IκBα, anti-p-IκBα, anti-IKK-β and anti-p-IKK-β anti- bodies (Cell Signaling). Blotting membranes were stripped Acknowledgements This work was supported by Natural Science and re-probed with anti-GAPDH, anti-p84, and anti-H3 Foundation of China (No. 81830082, 91740119, 91529301, α 81621004, 91740118, 81773106 and 81530082); Guangzhou Science (abcam, 1:1000), anti-Lamin B (Cell Signaling) and anti- - and Technology Plan Projects (201803010098); Natural Science Tubulin antibodies and β-actin(Sigma, Saint Louis, MO) as Foundation of Guangdong Province (2018B030311009 and a loading control. 2016A030308002); The Fundamental Research Funds for the Central Universities [No. 17ykjc02]. Immunohistochemistry Compliance with ethical standards The immunohistochemistry assay was carried out using a Conflict of interest The authors declare that they have no conflict of previously published standard protocol [47]. The antibodies interest. used in IHC assay were: anti-TRIM3 (Novus), anti- Lyve-1 (Abcam), anti-importin α3 (Abcam), anti-ACTN4 (Abcam), References anti-VEGFC (Abcam), Anti- NF-κB p65 (Abcam), and Anti-MMP9 (Abcam). The images were captured and ana- 1. Alitalo K, Tammela T, Petrova TV. Lymphangiogenesis in lyzed by using an Axio Imager.Z2 system (Carl Zeiss Co. development and human disease. Nature. 2005;438:946–53. Ltd., Jena, Germany). 2. Dieterich LC, Detmar M. Tumor lymphangiogenesis and new drug development. Adv Drug Deliv Rev. 2016;99:148–60. 3. Hirakawa S. Regulation of pathological lymphangiogenesis Popliteal lymph node metastasis model requires factors distinct from those governing physiological lym- phangiogenesis. J Dermatol Sci. 2011;61:85–93. Five- or 6-week-old BALB/c-nu mice were purchased from 4. Schoppmann SF, Jesch B, Zacherl J, Riegler MF, Friedrich J, Birner P. Lymphangiogenesis and lymphovascular invasion diminishes the Center of Experimental Animal of Guangdong Province prognosis in esophageal cancer. Surgery. 2013;153:526–34. (Guangdong, China), and approval for the experiments was 5. Saad RS, Lindner JL, Liu Y, Silverman JF. Lymphatic vessel obtained from the Institutional Animal Care and Use density as prognostic marker in esophageal adenocarcinoma. Am J – Committee. Briefly, the BALB/c nude mice were randomly Clin Pathol. 2009;131:92 98. n = 6. Kumagai Y, Tachikawa T, Higashi M, Sobajima J, Takahashi A, divided into many groups ( 16/group) and were injected Amano K. et al. Vascular endothelial growth factors C and D and 5 with 5 × 10 indicated cells in their foot-pads. The mice lymphangiogenesis at the early stage of esophageal squamous cell were treated with vehicle or Tecfidera (20 mg/kg via oral carcinoma progression. Dis Esophagus. 2018;31:1–7. gavage) once daily for 21 days, 1 week after inoculation of 7. Liu P, Zhou J, Zhu H, Xie L, Wang F, Liu B, et al. VEGF-C promotes the development of esophageal cancer via regulating the cells. After 4 weeks, the vehicle-treated mice were CNTN-1 expression. Cytokine. 2011;55:8–17. executed and the Tecfidera-treated mice were sacrificed 8. Karpanen T, Egeblad M, Karkkainen MJ, Kubo H, Yla-Herttuala when the foot-pad primary tumors grown to the same size as S, Jaattela M, et al. Vascular endothelial growth factor C promotes control mice. Immunohistochemistry was carried out to tumor lymphangiogenesis and intralymphatic tumor growth. Cancer Res. 2001;61:1786–90. detect the protein levels of Lyve-1, VEGFC, and MMP9 in 9. Hirakawa S, Brown LF, Kodama S, Paavonen K, Alitalo K, the primary tumors and in popliteal lymph nodes. Detmar M. VEGF-C-induced lymphangiogenesis in sentinel lymph nodes promotes tumor metastasis to distant sites. Blood. – Chromatin fractionation 2007;109:1010 7. 10. Mandriota SJ, Jussila L, Jeltsch M, Compagni A, Baetens D, Prevo R, et al. Vascular endothelial growth factor-C-mediated lymphangio- Chromatin fractionation extract analysis was performed genesis promotes tumour metastasis. EMBO J. 2001;20:672–82. according to previously published standard methods [48]. 11. He W, Zhong G, Jiang N, Wang B, Fan X, Chen C, et al. Long Western blotting analysis was performed using anti-p65 noncoding RNA BLACAT2 promotes bladder cancer-associated lymphangiogenesis and lymphatic metastasis. J Clin Invest. antibodies and Lamin B1 was used as the loading control. 2018;128:861–75. 12. Feng Y, Hu J, Ma J, Feng K, Zhang X, Yang S, et al. RNAi-mediated Statistical analysis silencing of VEGF-C inhibits non-small cell lung cancer progression by simultaneously down-regulating the CXCR4, CCR7, VEGFR-2 and VEGFR-3-dependent axes-induced ERK, p38 and AKT signal- SPSS 21.0 statistical software (IBM, Armonk, NY, USA) ling pathways. Eur J Cancer. 2011;47:2353–63. was used to analyze the data. Differences between two J. Zhu et al.

13. Lin C, Song L, Liu A, Gong H, Lin X, Wu J, et al. Overexpression (ACTN4) is a transcriptional co-activator of RelA/p65 sub-unit of of AKIP1 promotes angiogenesis and lymphangiogenesis in NF-kB. Oncotarget. 2013;4:362–72. human esophageal squamous cell carcinoma. Oncogene. 31. Zhao X, Hsu KS, Lim JH, Bruggeman LA, Kao HY. alpha- 2015;34:384–93. Actinin 4 potentiates nuclear factor kappa-light-chain-enhancer of 14. Su C, Chen Z, Luo H, Su Y, Liu W, Cai L, et al. Different patterns activated B-cell (NF-kappaB) activity in podocytes independent of of NF-kappaB and Notch1 signaling contribute to tumor-induced its cytoplasmic actin binding function. J Biol Chem. lymphangiogenesis of esophageal squamous cell carcinoma. J Exp 2015;290:338–49. Clin Cancer Res. 2011;30:85. 32. Wei H, Wang B, Miyagi M, She Y, Gopalan B, Huang DB, et al. 15. Li F, Sethi G. Targeting transcription factor NF-kappaB to over- PRMT5 dimethylates R30 of the p65 subunit to activate NF- come chemoresistance and radioresistance in cancer therapy. kappaB. Proc Natl Acad Sci USA. 2013;110:13516–21. Biochim Biophys Acta. 2010;1805:167–80. 33. Stacker SA, Williams SP, Karnezis T, Shayan R, Fox SB, Achen 16. Baud V, Karin M. Is NF-kappaB a good target for cancer therapy? MG. Lymphangiogenesis and lymphatic vessel remodelling in Hopes and pitfalls. Nat Rev Drug Discov. 2009;8:33–40. cancer. Nat Rev Cancer. 2014;14:159–72. 17. Trial watch: Phase III success for Biogen’s oral multiple sclerosis 34. Ribatti D. Historical overview of lymphangiogenesis. Curr Opin therapy. Nat Rev Drug Discov. 2011;10:404–404. Immunol. 2018;53:161–6. 18. Peng H, Guerau-de-Arellano M, Mehta VB, Yang Y, Huss DJ, 35. Montes Diaz G, Fraussen J, Van Wijmeersch B, Hupperts R, Papenfuss TL, et al. Dimethyl fumarate inhibits dendritic cell Somers V. Dimethyl fumarate induces a persistent change in the maturation via nuclear factor kappaB (NF-kappaB) and extra- composition of the innate and adaptive immune system in multiple cellular signal-regulated kinase 1 and 2 (ERK1/2) and mitogen sclerosis patients. Sci Rep. 2018;8:8194. stress-activated kinase 1 (MSK1) signaling. J Biol Chem. 36. Smith MD, Calabresi PA, Bhargava P. Dimethyl fumarate treat- 2012;287:28017–26. ment alters NK cell function in multiple sclerosis. Eur J Immunol. 19. Nicolay JP, Muller-Decker K, Schroeder A, Brechmann M, Mobs 2018;48:380–3. M, Geraud C, et al. Dimethyl fumarate restores apoptosis sensi- 37. Honda K, Yamada T, Endo R, Ino Y, Gotoh M, Tsuda H, et al. tivity and inhibits tumor growth and metastasis in CTCL by tar- Actinin-4, a novel actin-bundling protein associated with cell geting NF-kappaB. Blood. 2016;128:805–15. motility and cancer invasion. J Cell Biol. 1998;140:1383–93. 20. Kastrati I, Siklos MI, Calderon-Gierszal EL, El-Shennawy L, 38. Agarwal N, Adhikari AS, Iyer SV, Hekmatdoost K, Welch DR, Georgieva G, Thayer EN, et al. Dimethyl fumarate inhibits the Iwakuma T. MTBP suppresses cell migration and filopodia for- nuclear factor kappab pathway in breast cancer cells by covalent mation by inhibiting ACTN4. Oncogene. 2013;32:462–70. modification of p65 protein. J Biol Chem. 2016;291:3639–47. 39. Chakraborty S, Reineke EL, Lam M, Li X, Liu Y, Gao C, et al. 21. El-Husseini AE, Fretier P, Vincent SR. Cloning and character- Alpha-actinin 4 potentiates myocyte enhancer factor-2 transcrip- ization of a gene (RNF22) encoding a novel brain expressed ring tion activity by antagonizing histone deacetylase 7. J Biol Chem. finger protein (BERP) that maps to human chromosome 11p15.5. 2006;281:35070–80. Genomics. 2001;71:363 –7. 40. Khurana S, Chakraborty S, Cheng X, Su YT, Kao HY. The actin- 22. Yan Q, Sun W, Kujala P, Lotfi Y, Vida TA, Bean AJ. CART: an binding protein, (ACTN4), is a nuclear receptor Hrs/actinin-4/BERP/myosin V protein complex required for effi- coactivator that promotes proliferation of MCF-7 breast cancer cient receptor recycling. Mol Biol Cell. 2005;16:2470–82. cells. J Biol Chem. 2011;286:1850–9. 23. Schreiber J, Vegh MJ, Dawitz J, Kroon T, Loos M, Labonte D, 41. Khurana S, Chakraborty S, Zhao X, Liu Y, Guan D, Lam M, et al. et al. Ubiquitin ligase TRIM3 controls hippocampal plasticity and Identification of a novel LXXLL motif in alpha-actinin 4-spliced learning by regulating synaptic gamma-actin levels. J Cell Biol. isoform that is critical for its interaction with estrogen receptor 2015;211:569–86. alpha and co-activators. J Biol Chem. 2012;287:35418–29. 24. Mukherjee S, Tucker-Burden C, Zhang C, Moberg K, Read R, 42. An HT, Kim J, Yoo S, Ko J. Small leucine zipper protein (sLZIP) Hadjipanayis C, et al. Drosophila Brat and human ortholog negatively regulates skeletal muscle differentiation via interaction TRIM3 maintain stem cell equilibrium and suppress brain with alpha-actinin-4. J Biol Chem. 2014;289:4969–79. tumorigenesis by attenuating notch nuclear transport. Cancer Res. 43. Zhao X, Khurana S, Charkraborty S, Tian Y, Sedor JR, Bruggman 2016;76:2443–52. LA, et al. alpha actinin 4 (ACTN4) regulates glucocorticoid 25. Moskaluk CA, Rumpel CA. Allelic deletion in 11p15 is a com- receptor-mediated transactivation and transrepression in podo- mon occurrence in esophageal and gastric adenocarcinoma. Can- cytes. J Biol Chem. 2017;292:1637–47. cer. 1998;83:232–9. 44. Lu T, Stark GR. NF-kappaB: regulation by methylation. Cancer 26. Lam CT, Tang CM, Lau KW, Lung ML. Loss of heterozygosity Res. 2015;75:3692–5. on in esophageal squamous cell carcinomas. 45. Boulay JL, Stiefel U, Taylor E, Dolder B, Merlo A, Hirth F. Loss Cancer Lett. 2002;178:75–81. of heterozygosity of TRIM3 in malignant gliomas. BMC Cancer. 27. Carlotti F, Dower SK, Qwarnstrom EE. Dynamic shuttling of 2009;9:71. nuclear factor kappa B between the nucleus and cytoplasm as a 46. Deng W, Tsao SW, Guan XY, Lucas JN, Si HX, Leung CS, et al. consequence of inhibitor dissociation. J Biol Chem. Distinct profiles of critically short telomeres are a key determinant 2000;275:41028–34. of different chromosome aberrations in immortalized human cells: 28. Chiao PJ, Miyamoto S, Verma IM. Autoregulation of I kappa B whole-genome evidence from multiple cell lines. Oncogene. alpha activity. Proc Natl Acad Sci USA. 1994;91:28–32. 2004;23:9090–101. 29. Arenzana-Seisdedos F, Thompson J, Rodriguez MS, Bachelerie F, 47. Li J, Zhang N, Song LB, Liao WT, Jiang LL, Gong LY, et al. Thomas D, Hay RT. Inducible nuclear expression of newly syn- Astrocyte elevated gene-1 is a novel prognostic marker for breast thesized I kappa B alpha negatively regulates DNA-binding and cancer progression and overall patient survival. Clinical cancer transcriptional activities of NF-kappa B. Mol Cell Biol. research: an official journal of the American Association for. 1995;15:2689–96. Cancer Res. 2008;14:3319–26. 30. Aksenova V, Turoverova L, Khotin M, Magnusson KE, Tul- 48. Girard C, Will CL, Peng J, Makarov EM, Kastner B, Lemm I, chinsky E, Melino G, et al. Actin-binding protein alpha-actinin 4 et al. Post-transcriptional spliceosomes are retained in nuclear speckles until splicing completion. Nat Commun. 2012;3:994.