SIRT6 Is a Target of Regulation by UBE3A That Contributes to Liver

Published OnlineFirst December 7, 2017; DOI: 10.1158/0008-5472.CAN-17-1692

Cancer Molecular Cell Biology Research

SIRT6 Is a Target of Regulation by UBE3A That Contributes to Liver Tumorigenesis in an ANXA2-Dependent Manner Saishruti Kohli, Abhishek Bhardwaj, Richa Kumari, and Sanjeev Das

Abstract

UBE3A is an E3 ubiquitin ligase well known for its role in the regulation and consequent induction of ANXA2 were critical for proteasomal degradation of p53 in human papillomavirus (HPV)- UBE3A-mediated tumorigenesis. Furthermore, in clinical speci- associated cancers. Here we report that UBE3A ubiquitylates and mens, increased UBE3A levels correlated with reduced SIRT6 levels triggers degradation of the tumor-suppressive sirtuin SIRT6 in and elevated ANXA2 levels in increasing tumor grades. Overall, our hepatocellular carcinoma. UBE3A ubiquitylated the highly con- findings show how the tumor suppressor SIRT6 is regulated in served Lys160 residue on SIRT6. FOXO1-mediated transcriptional hepatocellular carcinoma and establish the mechanism underlying repression of UBE3A was sufficient to stabilize SIRT6 and to UBE3A-mediated tumorigenesis in this disease. epigenetically repress ANXA2, a key mediator of UBE3A oncogenic Significance: These findings provide mechanistic insights function. Thus, UBE3A-mediated SIRT6 degradation promoted the into regulation of the tumor suppressive sirtuin SIRT6 and its proliferative capacity, migration potential, and invasiveness of implications for the development of hepatocellular carcinoma. cells. In mouse models of hepatocellular carcinoma, SIRT6 down- Cancer Res; 78(3); 645–58. 2017 AACR.

Introduction now suggest that SIRT6 functions as a tumor suppressor by modulating cellular transcription program (9–11). SIRT6 was UBE3A is the founding member of the HECT family of E3 also found to be downregulated in several cancers including ubiquitin ligases. UBE3A was first identified by its ability to hepatocellular carcinoma (10, 12). Although SIRT6 is implicated interact with p53 and target it for proteasomal degradation in in several metabolic pathways and is induced upon metabolic association with the human papillomavirus oncoprotein E6 stress (13), the mechanistic details of its regulation under star- (1). Besides cervical cancer, UBE3A has also been implicated in vation conditions are yet unclear. SIRT6 has been reported to be prostate and breast cancer (2–4). However, apart from E6- posttranslationally stabilized under glucose stress and serum dependent degradation of p53, the substrate repertoire of deprivation conditions (14). UBE3A was identified as a SIRT6- UBE3A involved in tumorigenesis is yet unclear. The only other interacting protein in a proteomics screen performed previously substrate linking UBE3A E3 ligase activity to tumorigenesis is (10). Our further studies now indicate that UBE3A ubiquitylates tumor suppressor PML. UBE3A promotes degradation of PML SIRT6, leading to its proteasomal degradation, under normal that facilitates Myc-driven B-cell lymphomagenesis (5). physiologic conditions. However, upon metabolic stress, UBE3A A recent study reported that UBE3A induces Annexin A2 (ANXA2) is transcriptionally repressed, resulting in SIRT6 stabilization. levels by an unidentified mechanism that promotes proliferation Furthermore, SIRT6 suppresses the expression of ANXA2 through and invasiveness of cancer cells (6). ANXA2 has been characterized an epigenetic mechanism, resulting in abrogation of UBE3A- as a calcium-dependent phospholipid-binding protein. Elevated mediated tumorigenesis. expression of ANXA2 has been reported in several cancers including liver cancer (7). Accumulating data suggest that ANXA2 plays a key role in diverse processes linked to tumorigenesis (8). þ Materials and Methods SIRT6 is a NAD -dependent histone deacetylase belonging to Cell lines and culture conditions the highly conserved sirtuin family of proteins. Accumulating data HepG2, H1299, HCT116, U2OS, and 293A cells were grown in DMEM with 10% FBS (Invitrogen) and 100 U/mL penicillin and 100 mg/mL streptomycin at 37C. The cell lines were Molecular Oncology Laboratory, National Institute of Immunology, New Delhi, India. obtained from ATCC between 2012 and 2016. The cell lines were authenticated by ATCC and tested for absence of myco- Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). plasma contamination (MycoAlert, Lonza). The cells used in experiments were within 10 passages from thawing. Amplifi- S. Kohli and A. Bhardwaj contributed equally to this article. cation and titration of recombinant adenoviruses was carried Corresponding Author: Sanjeev Das, Molecular Oncology Laboratory, National out as described previously (15). Cells were cultured to approx- Institute of Immunology, Aruna Asaf Ali Marg, New Delhi 110067, India. Phone: imately 50%–70% confluency followed by infection with 9111-2670-3702, Fax: 9111-2674-2125, E-mail: [email protected] recombinant adenovirus at a multiplicity of infection (MOI) doi: 10.1158/0008-5472.CAN-17-1692 of 10–20. For glucose starvation conditions, cells were cultured 2017 American Association for Cancer Research. to approximately 50% confluency and then grown in glucose-

www.aacrjournals.org 645

Kohli et al.

free DMEM (Invitrogen) with dialyzed FBS (Invitrogen) and Results 5 mmol/L glucose (Sigma). For inducing serum deprivation, SIRT6 levels are regulated by the ubiquitin proteasome pathway cellswereculturedtoapproximately50%confluency and then To investigate the molecular mechanisms underlying SIRT6 grown in serum-free DMEM (Invitrogen). Early-passage MEFs regulation, we examined SIRT6 levels in HepG2 cells under were cultured, and subjected to retroviral transduction as metabolic stress conditions including glucose and serum dep- described previously (16). rivation. We observed that there was a robust induction in SIRT6 protein levels while there was no significant change in its In vitro ubiquitylation assay mRNA levels over the time course of metabolic stress (Fig. 1A In vitro ubiquitylation assay was carried out using the E2- and B; Supplementary Fig. S1A and S1B). Similar observations Ubiquitin Conjugation Kit (Abcam) according to the manufac- were made in other cell types including H1299 and primary fl turer's instructions. Brie y, recombinant SIRT6 (Abcam) was MEFs (Supplementary Fig. S1C–S1F). These results indicate a C833A E. incubated along with His-UBE3A or His-UBE3A (from posttranscriptional mechanism of SIRT6 regulation. We next coli BL21) in a reaction buffer containing E1, E2 (UBCH7), and determined whether the increase in SIRT6 protein levels was ubiquitin, at 37 C for 2 hours. The reaction was terminated by due to increased translation or increased stability of SIRT6 adding the gel loading buffer. Immunoblotting was performed as protein. In the presence of cycloheximide, there was a gradual described above. decline in SIRT6 protein levels under unstressed conditions but under starvation conditions, sustained high levels of SIRT6 Proliferation assay protein were observed (Fig. 1C and D). However, there was Cells were cultured in triplicates in 12-well plates and subjected no significant difference in transcript levels between unstressed to trypsinization at the indicated time points to obtain cell and starvation conditions (Supplementary Fig. S1G). These suspensions. 0.4% Trypan blue (w/v) solution was mixed well results suggest that SIRT6 protein is stabilized upon metabolic with equal volume of cell suspension and hemocytometer was stress. As ubiquitin proteasomal machinery is frequently used to count the unstained healthy cells. involved in regulation of protein stability, we next investigated the effect of proteasome inhibitor MG132 on SIRT6 levels. We Plasmin generation assay observed that MG132 treatment led to sustained high levels of Cells were detached using an enzyme-free dissociation solu- SIRT6 protein throughout the time course of starvation without tion (Sigma). Cells were washed with PBS and tissue plasmin- discernable change in its transcript levels, while in absence of ogen activator (1 mg/mL; Sigma) was added. Cells were then MG132 treatment, there was a gradual increase in SIRT6 protein subjected to 30-minute incubation on ice, washed with PBS, levels upon metabolic stress with no significant change in and 0.4 mmol/L glu-plasminogen (Sekisui Diagnostics) was transcript levels (Fig. 1E and F). This was further corroborated added together with 10 mL of Spectrozyme (Sekisui Diagnos- by the observation that treatment with an inhibitor of the tics). Plasmin production was monitored by measuring absor- ubiquitin activating enzyme (E1) PYR41 resulted in similar bance at 405 nm. high levels of SIRT6 protein throughout the stress period without any discernable change in transcript levels, while in Matrix metalloproteinase activity assay the absence of PYR41 treatment, there was a gradual increase in Matrix metalloproteinase (MMP) activity was measured SIRT6 protein levels upon metabolic stress with no significant using MMP Activity Assay Kit (Abcam) following the manu- change in mRNA levels (Fig. 1G and H; Supplementary Fig. S1H facturer's instructions. Fluorescence intensity was determined at and S1I). Similar observations were made in H1299 and excitation/emission 490/520nm using a fluorogenic peptide primary MEFs (Supplementary Fig. S1J–S1M) Taken together, substrate. these results suggest that the ubiquitin proteasome pathway plays a key role in modulating SIRT6 levels in response to Xenograft studies metabolic stress. An Institutional Animal Care and Use Committee approved all the animal studies. The right flank of female nude mice (nu/nu) UBE3A downregulates SIRT6 protein levels 6 was subcutaneously injected with 4 10 cells/0.15 mL of Since UBE3A, an E3 ubiquitin ligase, was identified as a SIRT6- DMEM. The mice were randomly assigned to groups for injecting interacting protein in a proteomics screen described previously different cell lines. The tumor size was evaluated for 30 days. No (10), it raised the possibility that UBE3A may have a role in SIRT6 blinding was used. Tumor volume was measured with a caliper regulation. Thus, we examined SIRT6 protein and transcript levels and calculated using the formula: (widest diameter smallest in the presence or absence of UBE3A in different cell lines. Our 2 diameter )/2. After 30 days, the mice were euthanized using results indicated that SIRT6 protein levels increased markedly carbon dioxide, followed by tumor excision and weight determi- upon UBE3A depletion, while no significant change in its mRNA nation. Tumor extracts were then made and immunoblotting was levels was observed (Fig. 2A and B). We further observed that carried out as described previously. UBE3A coimmunoprecipitated with SIRT6 (Fig. 2C). To investigate SIRT6–UBE3A interaction under physiologic settings, we first Statistical analysis determined the effect of metabolic stress on UBE3A. Our results All experiments were conducted independently at least three indicated that UBE3A levels declined concomitant to induction in times. Results were expressed as means SD. Statistical analyses SIRT6 levels upon metabolic stress (Fig. 2D; Supplementary Fig. were performed by a standard two-tailed Student t test, one-way S2A). Similar observations were made in other cell types (Sup- ANOVA, and two-way ANOVA. P < 0.05 was considered plementary Fig. S2B and S2C). These results indicate that the significant. reduction in UBE3A levels upon metabolic stress could be

646 Cancer Res; 78(3) February 1, 2018 Cancer Research

UBE3A Downregulates SIRT6 to Promote Tumorigenesis

Figure 1. SIRT6 protein levels are upregulated upon metabolic stress. A and B, HepG2 cells were subjected to glucose starvation for the indicated time points.

Western blotting (A)andRT-qPCR(B) were then performed. DDCt method was used to analyze RT-qPCR data, with 18S rRNA as internal control, and relative mRNA levels were determined with respectto0-hourtimepoint.Errorbarsaremeans SD of three independent experiments with triplicate samples. C and D, HepG2 cells were starved for 24 hours. The cells were then subjected to cycloheximide chase (15 mg/mL) for the indicated time points under unstressed (25 mmol/L) or glucose starvation conditions (5 mmol/L). Western blotting (C) was then performed, and SIRT6 protein levels were quantitated (D) from three independent experiments. , P < 0.001. E and F, HepG2 cells were subjected to glucose starvation for indicated time points. The cells were also treated with MG132 (10 mmol/L) for the last 6 hours of every time point as indicated. Western blotting (E)andRT-qPCR(F)werethen performed. DDCt method was used to analyze RT-qPCR data, with 18S rRNA as internal control, and relative mRNA levels were determined with respect to control (untreated with MG132) cells at 0-hour time point. Error bars are means SD of three independent experiments with triplicate samples. G and H, HepG2 cells were subjected to glucose starvation for indicated time points. The cells were also treated with PYR41 (40 mmol/L) for the last 6 hours of every time point as indicated. Western blotting (G)andRT-qPCR(H) were then performed. DDCt method was used to analyze RT- qPCR data with 18S rRNA as internal control and relative mRNA levels were determined with respect to control (untreated with PYR41) cells at 0-hour time point. Error bars are means SD of three independent experiments with triplicate samples. n.s., nonsigniﬁcant.

www.aacrjournals.org Cancer Res; 78(3) February 1, 2018 647

Kohli et al.

Figure 2. UBE3A interacts with SIRT6 and suppresses its levels. A and B, HepG2, H1299, HCT116, U2OS, and HEK293 cells were transfected with control (scrambled) or UBE3A shRNA. Thirty-six hours posttransfection, Western blotting (A)andRT-qPCR(B) were performed. Error bars are means SD of three independent experiments with triplicate samples. C, HepG2 cells were infected with Ad-GFP or Ad-SIRT6 for 24 hours. Cells were treated with MG132 (10 mmol/L) for the last 6 hours. Immunoprecipitations were performed, followed by Western blotting. D, HepG2 cells were subjected to glucose starvation for the indicated time points. Western blotting was then performed. E, HepG2 cells were subjected to glucose starvation for the indicated time points and treated with MG132 (10 mmol/L) for the last 6 hours of every time point. Immunoprecipitations were performed, followed by Western blotting. F, HepG2 cells were stably transfected with control (scrambled) or UBE3A shRNA. HepG2 control (HepG2) and UBE3A knockdown (HepG2 UBE3Akd) cells were subjected to glucose starvation for the indicated time points. Western blotting was then performed. G, HepG2 cells were subjected to glucose starvation for the indicated time points and infected with indicated adenoviruses during the last 24 hours prior to the end of the 36-hour time point as indicated. Western blotting was then performed.

648 Cancer Res; 78(3) February 1, 2018 Cancer Research

UBE3A Downregulates SIRT6 to Promote Tumorigenesis

Figure 3. SIRT6 is a UBE3A E3 ligase substrate. A, Ubiquitylation assay was performed using recombinant SIRT6, wild-type UBE3A, and mutant UBE3A as indicated. B, HepG2 control (HepG2) and UBE3A knockdown (HepG2 UBE3Akd) cells were subjected to glucose starvation for the indicated time points and treated with MG132 (10 mmol/L) for the last 6 hours of each time point. Immunoprecipitations were performed, followed by Western blotting. C, HepG2 cells were subjected to glucose starvation for the indicated time points and treated with MG132 (10 mmol/L) for the last 6 hours of each time point. Immunoprecipitations were performed, followed by Western blotting. D, HepG2 cells were transfected with constructs expressing full-length or different domains of SIRT6 as GST fusion protein. Twenty-four hours posttransfection, the cells were subjected to glucose starvation for the indicated time points and treated with MG132 (10 mmol/L) for the last 6 hours of starvation period. GST pull-down was performed, followed by immunoblotting. E, HepG2 cells were transfected with the indicated GST-tagged SIRT6 subdomains. Twenty-four hours posttransfection, the cells were subjected to glucose starvation for the indicated time points and treated with MG132 (10mmol/L) for the last 6 hours of starvation period. GST pull-down was performed, followed by immunoblotting. F, HepG2 cells were transfected with the indicated HA-tagged SIRT6 mutants. Twenty-four hours posttransfection, the cells were subjected to glucose starvation for the indicated time points and treated with MG132 (10mmol/L) for the last 6 hours of starvation period. Immunoprecipitations were performed using anti-HA antibody, followed by Western blotting. G, Amino acid sequence comparison of SIRT6 region spanning residues 151–169 from different organisms.

www.aacrjournals.org Cancer Res; 78(3) February 1, 2018 649

Kohli et al.

Figure 4. FOXO1 is a transcriptional repressor of UBE3A. A, HepG2 cells were subjected to glucose starvation for the indicated time points. RT-qPCR was then performed. Error bars are means SD of three independent experiments with triplicate samples. , P < 0.01. B, HepG2 cells were subjected to serum starvation for the indicated time points. RT-qPCR was then performed. Error bars are means SD of three independent experiments with triplicate samples. , P < 0.01. C, Schematic representation of FOXO1-binding sites at the UBE3A promoter. (Continued on the following page.)

650 Cancer Res; 78(3) February 1, 2018 Cancer Research

UBE3A Downregulates SIRT6 to Promote Tumorigenesis

responsible for SIRT6 induction. We next investigated whether (Fig. 3E). To identify the specific site of polyubiquitylation, we endogenous SIRT6 and UBE3A interacted under metabolic mutated each of the five lysine residues within this region. Our stress conditions. We observed that SIRT6 and UBE3A interacted results indicated that the mutation of K160 residue abrogated under unstressed conditions and the interaction diminished over UBE3A-mediated K48-linked polyubiquitylation of SIRT6 under the time course of starvation (Fig. 2E). unstressed conditions even though similar to other mutants To further explore the role of UBE3A in regulating SIRT6, we SIRT6K160R could interact with UBE3A (Fig. 3F; Supplementary examined SIRT6 levels upon metabolic stress in presence or Fig. S3A). Moreover, protein sequence analysis of SIRT6 region absence of UBE3A. There was an induction of SIRT6 levels over encompassing amino acids 151–169 suggests that K160 residue is the time course of starvation period in HepG2 control cells. conserved across diverse organisms (Fig. 3G). Taken together, However, in HepG2 UBE3A knockdown cells, SIRT6 levels these results suggest that UBE3A is responsible for K48-linked were constitutively elevated over the time course of starvation polyubiquitylation of SIRT6 at the conserved K160 residue that period (Fig. 2F; Supplementary Fig. S2D). Similar observations leads to SIRT6 degradation. were made in H1299 and primary MEFs (Supplementary Fig. S2E and S2F). Furthermore, ectopic expression of UBE3A at FOXO1 downregulates UBE3A upon metabolic stress late phase of starvation (36 hours) led to a sharp reduction in To investigate the mechanistic basis of UBE3A downregulation, SIRT6 levels while in presence of E3-ligase–dead UBE3A we examined UBE3A transcript levels upon metabolic stress. We mutant (UBE3AC833A; ref. 17), there was no evident decline observed that there was a decrease in the transcript levels with in SIRT6 levels (Fig. 2G), even though both wild-type UBE3A increasing duration of metabolic stress (Fig. 4A and B). In silico and UBE3AC833A could interact with SIRT6 (Supplementary analysis of the UBE3A promoter sequence revealed two putative Fig. S2G). Thus, we concluded that UBE3A binds to SIRT6 and binding sites for FOXO1, an important member of the FOX regulates its levels. (Forkhead box) family of transcription factors, involved in metabolic stress response (Fig. 4C). To analyze the effect of FOXO1 on UBE3A ubiquitylates SIRT6 at K160 residue UBE3A expression, we examined UBE3A mRNA levels upon We next investigated whether UBE3A ubiquitylates SIRT6. Our metabolic stress, in presence or absence of FOXO1. We observed results indicated that wild-type UBE3A could polyubiquitylate that the metabolic stress–induced reduction in UBE3A transcript SIRT6 while the E3-ligase dead UBE3A mutant (UBE3AC833A) had levels was abrogated in HepG2 FOXO1 knockdown cells (Fig. 4D; no effect (Fig. 3A). We further investigated UBE3A-mediated Supplementary Fig. S3B). Similar observations were made in SIRT6 polyubiquitylation under metabolic stress conditions. Our other cell types (Supplementary Fig. S3C and S3D). We next results indicated that UBE3A polyubiquitylates SIRT6 under performed chromatin immunoprecipitation (ChIP) assay to unstressed conditions, which was abrogated upon metabolic examine the binding of FOXO1 to UBE3A promoter. Upon stress as UBE3A levels were downregulated (Fig. 3B). Moreover, metabolic stress, increasing levels of FOXO1 were detected at upon UBE3A knockdown, SIRT6 polyubiquitylation was abol- both the sites of the UBE3A promoter (Fig. 4E and F). As the ished. We also observed UBE3A-mediated polyubiquitylation to cellular distribution of FOXO1 and hence its activity is deter- be K48-linked (Fig. 3C). mined by its phosphorylation status (18), we analyzed the levels Next, we sought to map the specific SIRT6 domain that under- of phosphorylated FOXO1 and its cellular localization upon goes UBE3A-mediated polyubiquitylation. The GST-SIRT6 fusion metabolic stress. We observed that levels of phosphorylated protein comprising the amino acids 49–271 was polyubiquity- FOXO1, which is cytoplasmic, declined upon metabolic stress lated by UBE3A under unstressed conditions while upon meta- concomitant to an increase in nuclear FOXO1 levels (Fig. 4G). bolic stress this polyubiquitylation was abrogated as UBE3A levels As previous studies suggest that FOXO1 can also function as a were downregulated (Fig. 3D). To narrow down further, we transcriptional repressor (19), we next determined whether investigated UBE3A-mediated polyubiquitylation of GST-SIRT6 FOXO1 can repress UBE3A transcription. We observed that segments spanning amino acids 49–170 and 171–271. The GST- wild-type FOXO1 can progressively repress UBE3A promoter SIRT6 fusion protein comprising the amino acids 49–170 was more robustly over the time course of starvation period polyubiquitylated by UBE3A under unstressed conditions while (Fig. 4H). However, FOXO1 AAA mutant, which is exclusively upon metabolic stress this polyubiquitylation was abrogated nuclear (20), represses UBE3A promoter constitutively while

(Continued.) D, HepG2 cells were stably transfected with control (scrambled) or FOXO1 shRNA. HepG2 control (HepG2) and FOXO1 knockdown (HepG2 FOXO1kd) cells were subjected to glucose starvation for the indicated time points. RT-qPCR was performed (left panel). Error bars are means SD of three independent experiments with triplicate samples. FOXO1 knockdown was conﬁrmed by performing Western blotting (right). , P < 0.01. E, HepG2 cells were subjected to glucose starvation for indicated time points. ChIP assay was then performed. Error bars are means SD of three independent experiments with triplicate samples. , P < 0.01. F, HepG2 cells were subjected to glucose starvation for indicated time points. ChIP assay was then performed. Error bars are means SD of three independent experiments with triplicate samples. , P < 0.01. G, HepG2 cells were subjected to glucose starvation for the indicated time points. Western blotting was then performed from the nuclear and cytoplasmic fractions. H, HepG2 FOXO1 knockdown cells were cotransfected with UBE3A-Luc construct along with indicated FOXO1 constructs. Twenty-four hours posttransfection, cells were subjected to glucose starvation for indicated time points and luciferase assay was then performed (left). Error bars are means SD of three independent experiments with duplicate samples. Equal expression of different FOXO1 constructs was conﬁrmed by subjecting HepG2 cells transfected with the individual FOXO1 constructs to glucose starvation for 0 and 24 hours, followed by Western blotting (right). , P < 0.01. I, HepG2 control (HepG2), FOXO1 knockdown (HepG2 FOXO1kd), and FOXO1/UBE3A double knockdown (HepG2 FOXO1kd/UBE3Akd) cells were subjected to glucose starvation for indicated time points. Western blotting was then performed. J, HepG2 control (HepG2), FOXO1 knockdown (HepG2 FOXO1kd) cells, and FOXO1/UBE3A double knockdown (HepG2 FOXO1kd/UBE3Akd) cells were subjected to glucose starvation for indicated time points. Cells were treated with MG132 for the last 6 hours of each time point. Immunoprecipitations were performed, followed by Western blotting.

www.aacrjournals.org Cancer Res; 78(3) February 1, 2018 651

Kohli et al.

Figure 5. SIRT6 downregulates ANXA2 levels. A, HepG2 cells were subjected to glucose starvation for the indicated time points. Western blotting was then performed. B, HepG2 cells were subjected to glucose starvation for the indicated time points. RT-qPCR was performed. Error bars are means SD of three independent experiments with triplicate samples. (Continued on the following page.)

652 Cancer Res; 78(3) February 1, 2018 Cancer Research

UBE3A Downregulates SIRT6 to Promote Tumorigenesis

FOXO1 HRAAA mutant, which is exclusively nuclear, but lacks starvation, there was a sharp decline in ANXA2 levels upon ectopic DNA-binding ability (18), cannot repress UBE3A promoter at all. expression of wild-type SIRT6 while no such decline was observed Furthermore, deletion of any of the sites attenuates FOXO1- in case of deacetylase-dead SIRT6 mutant (SIRT6H133Y; Fig. 5D; mediated repression of UBE3A promoter, whereas deletion of Supplementary Fig. S4F). Thus, we concluded that SIRT6 could both the sites abrogates the repression (Supplementary Fig. S3E). function as a transcriptional corepressor of ANXA2. We further investigated the effect of FOXO1-mediated ANXA2 has been previously reported to be regulated by HIF1a UBE3A regulation on SIRT6 levels. Our results suggested that and c-JUN (21, 22), which are known SIRT6-interacting proteins in FOXO1 knockdown cells, UBE3A levels remained constitu- (23, 24). We observed that upon metabolic stress ANXA2 repres- tively elevated over the starvation period as compared with sion was abrogated in SIRT6 knockdown cells. However, upon control cells where there was a gradual decline in UBE3A levels simultaneous knockdown of HIF1a, there was no significant (Fig. 4I). Concomitantly, in FOXO1 knockdown cells, SIRT6 change in ANXA2 levels (Supplementary Fig. S5A and S5B). levels were constitutively repressed over the starvation period Similar results were observed for c-JUN (Supplementary Fig. while there was steady increase in SIRT6 levels in control cells. S5C and S5D). These results indicate that HIF1a or c-JUN were In addition, in FOXO1/UBE3A double knockdown cells, SIRT6 not involved with SIRT6 in repression of ANXA2. To identify the levels were constitutively elevated over the starvation period. transcription factor involved with SIRT6 in repressing ANXA2 Furthermore, in FOXO1 knockdown cells, K48-linked SIRT6 expression, we examined ANXA2 promoter sequence for potential polyubiquitylation was observed even upon prolonged meta- transcription factor–binding sites. Our analysis revealed one ELK1 bolic stress (36 hours), due to the presence of high levels of binding site at 1.546-kb position (Fig. 5E). Moreover, a recent UBE3A as compared with control cells. However, in FOXO1/ study demonstrated that SIRT6 interacts with ELK1 and function UBE3A double knockdown cells, SIRT6 was not polyubiquity- as a transcriptional corepressor (25). Thus, we investigated the lated even in unstressed conditions (Fig. 4J; Supplementary role of ELK1 in SIRT6-mediated regulation of ANXA2.We Fig. S3F). Similar observations were made in other cell types observed that in HepG2 SIRT6 knockdown cells, ANXA2 levels (Supplementary Fig. S3G and S3H). Taken together, these were constitutively elevated over the time period of starvation. results suggest that upon metabolic stress, FOXO1 represses However, upon simultaneous knockdown of ELK1, ANXA2 levels UBE3A, resulting in SIRT6 stabilization. were constitutively repressed (Fig. 5F; Supplementary Fig. S6A). Similar observations were made in H1299 and primary MEFs SIRT6-mediated H3K9 deacetylation results in ANXA2 (Supplementary Fig. S6B and S6C). We further observed that repression ectopic expression of SIRT6 failed to repress ANXA2 in the absence A previous report suggests that UBE3A regulates ANXA2 by an of ELK1. Moreover, ectopic expression of ELK1 led to significantly unidentified mechanism (6). Thus, we next examined the effect of higher ANXA2 levels in the absence of SIRT6 as compared with in metabolic stress on ANXA2 levels. We observed that upon met- presence of SIRT6 (Fig. 5G; Supplementary Fig. S6D). In addition, abolic stress, induction of SIRT6 levels was accompanied by a ELK1 was also able to trigger ANXA2 expression in the presence of reduction in ANXA2 levels (Fig. 5A). ANXA2 transcript levels were deacetylase-dead SIRT6 mutant (SIRT6H133Y). We also observed also repressed (Fig. 5B). Similar observations were made in other that there is no significant change in ELK1 levels during metabolic cell types (Supplementary Fig. S4A and S4B). We next examined stress (Supplementary Fig. S6E). Furthermore, SIRT6 interacts ANXA2 mRNA levels upon metabolic stress. In HepG2 UBE3A with ELK1 under metabolic stress conditions. knockdown cells, constitutive repression of ANXA2 levels was We next examined the presence of SIRT6 at ANXA2 promoter. observed over the starvation period unlike control cells where Increasing levels of SIRT6 were detected at the ANXA2 promoter there was a gradual decline (Fig. 5C; Supplementary Fig. S4C). In upon metabolic stress, while in the absence of ELK1, there was UBE3A/SIRT6 double knockdown cells, ANXA2 mRNA levels no discernable recruitment of SIRT6 at the ANXA2 promoter were constitutively elevated over the time period of metabolic (Fig. 5H). Similar observations were made in other cell types stress. Similar observations were made in other cell types (Sup- (Supplementary Fig. S6F and S6G). However, ELK1 binds to plementary Fig. S4D and S4E). We further investigated whether ANXA2 promoter irrespective of presence or absence of SIRT6 the deacetylase activity of SIRT6 is responsible for ANXA2 repres- (Supplementary Fig. S6H). As SIRT6 exhibits H3K9 and H3K56 sion. In UBE3A/SIRT6 double knockdown cells subjected to deacetylase activity, we investigated H3K9 and H3K56 acetylation

(Continued.) , P < 0.05. C, HepG2 control (HepG2), UBE3A knockdown (HepG2 UBE3kd), and UBE3A/SIRT6 double knockdown (HepG2 UBE3Akd/SIRT6kd) cells were subjected to glucose starvation for indicated time points. RT-qPCR was then performed. Error bars are means SD of three independent experiments with triplicate samples. , P < 0.05. D, HepG2 UBE3A/SIRT6 double knockdown cells (HepG2 UBE3Akd/SIRT6kd) were subjected to glucose starvation for 36 hours. The cells were infected with indicated adenoviruses during the last 12 hours of starvation. RT-qPCR was then performed. Error bars are means SD of three independent experiments with triplicate samples. , P < 0.001. E, Schematic representation of ELK1-binding sites at the Annexin A2 promoter. F, HepG2 control (HepG2), SIRT6 knockdown (HepG2 SIRT6kd), and SIRT6/ELK1 double knockdown (HepG2 SIRT6kd/ELK1kd) cells were subjected to glucose starvation for the indicated time points. RT-qPCR was then performed. Error bars are means SD of three independent experiments with triplicate samples. , P < 0.05. G, HepG2 SIRT6/ELK1 double knockdown cells were stably transfected with dual expression plasmid encoding wild-type SIRT6, ELK1, and ELK1 along with wild-type SIRT6 or SIRT6H133Y as indicated. Control represents HepG2 SIRT6/ELK1 double knockdown cells stably transfected with empty vector. Twenty-four hours posttransfection, RT-qPCR was then performed. Error bars are means SD of three independent experiments with duplicate samples. , P < 0.01; , P < 0.001. H, HepG2 control (HepG2) and ELK1 knockdown (HepG2 ELK1kd) cells were subjected to glucose starvation for indicated time points. ChIP assay was then performed. Error bars are means SD of three independent experiments with triplicate samples. , P < 0.05. I, HepG2 control (HepG2), SIRT6 knockdown (HepG2 SIRT6kd), and ELK1 knockdown (HepG2 ELK1kd) cells were subjected to glucose starvation for indicated time points. ChIP assay was then performed. Error bars are means SD of three independent experiments with triplicate samples. , P < 0.05. J, HepG2 control (HepG2), SIRT6 knockdown (HepG2 SIRT6kd), and ELK1 knockdown (HepG2 ELK1kd) cells were subjected to glucose starvation for indicated time points. ChIP assay was then performed. Error bars are means SD of three independent experiments with triplicate samples.

www.aacrjournals.org Cancer Res; 78(3) February 1, 2018 653

Kohli et al.

status at ANXA2 promoter. Significant decline in H3K9 acetyla- tumor size. Furthermore, smaller tumors were observed in the tion was observed over the metabolic stress period. However, in case of SIRT6/ANXA2 double knockdown cells as compared both HepG2 SIRT6 knockdown and ELK1 knockdown cells, there with UBE3A/SIRT6 double knockdown cells. These results was no change in H3K9 acetylation status (Fig. 5I). No significant establish the significance of SIRT6 downregulation in UBE3A- change in H3K56 acetylation status at the ANXA2 promoter was mediated tumorigenesis. observed over the time course of metabolic stress (Fig. 5J). Thus, our results suggest that SIRT6 acts as a corepressor of ELK1 to UBE3A-mediated derepression of ANXA2 promotes aggressive suppress ANXA2 expression. tumor phenotype We next used an orthotopic liver tumor model to examine the UBE3A-mediated SIRT6 downregulation is critical for its metastatic potential of HepG2 UBE3A knockdown, SIRT6 knock- oncogenic functions down, UBE3A/SIRT6 double knockdown, and SIRT6/ANXA2 As ANXA2 is a key mediator of UBE3A oncogenic functions, we double knockdown cells. We observed markedly larger orthotopic examined the effect of SIRT6-dependent ANXA2 repression on tumors in case of SIRT6 knockdown cells as compared with UBE3A-mediated neoplastic transformation. Our results indicat- UBE3A knockdown cells (Fig. 7A and B). Moreover, in case of ed that HepG2 UBE3A knockdown cells, expressing high levels of UBE3A/SIRT6 double knockdown cells, numerous metastatic SIRT6, exhibit significantly reduced proliferation capacity as com- nodules were also observed in lungs and pancreas of mice (Sup- pared with HepG2 SIRT6 knockdown cells (Fig. 6A and B). plementary Fig. S8B). This was further corroborated by the obser- Furthermore, HepG2 SIRT6/ANXA2 double knockdown cells vation that the primary orthotopic tumors exhibited downregula- have much reduced proliferation capacity as compared with tion of epithelial marker E-cadherin and upregulation of mesen- HepG2 UBE3A/SIRT6 double knockdown cells, which express chymal markers vimentin and fibronectin, which is suggestive of high levels of ANXA2. Similar observations were made in H1299 increased predisposition of these tumors to metastasize (Supple- cells (Supplementary Fig. S7A and S7B). We also determined the mentary Fig. S8C). Upon simultaneous knockdown of ANXA2 effect on malignancy of these cells. We observed that upon in SIRT6/ANXA2 double knockdown cells, significantly smaller abrogation of UBE3A expression, the invasiveness and migration tumors and no metastasis was observed as compared with potential of the cells was significantly reduced but when SIRT6 UBE3A/SIRT6 double knockdown cells. These results suggest expression was abrogated the invasiveness and migration poten- that UBE3A-mediated SIRT6 downregulation and consequent tial of the cells was notably increased (Fig. 6C and D; Supple- derepression of ANXA2 promotes tumor metastasis. mentary Fig. S7C). In addition, SIRT6/ANXA2 double knockdown As SIRT6 and ANXA2 have been linked to hepatocellular cells have much reduced invasiveness and migration potential as carcinoma, we investigated UBE3A, SIRT6, and ANXA2 levels in compared with UBE3A/SIRT6 double knockdown cells. Similar different grades of hepatocellular carcinoma. Elevated UBE3A and observations were made in H1299 cells (Supplementary Fig. ANXA2 levels were detected in hepatocellular carcinoma as com- S7D–S7F). pared with matched normal adjacent tissue (Fig. 7C). On the To further establish the role of ANXA2, we next investigated other hand, reduced SIRT6 levels were detected in hepatocellular plasmin levels, as ANXA2 facilitates plasmin generation, which carcinoma tissue as compared with matched normal adjacent promotes extracellular matrix degradation and cell invasion (26). tissue. Furthermore, UBE3A and ANXA2 levels increase with There was significantly more plasmin production in HepG2 SIRT6 increasing tumor grade while SIRT6 levels decline (Fig. 7D–F). knockdown and UBE3A/SIRT6 double knockdown cells as com- These observations indicate that UBE3A-mediated SIRT6 down- pared with HepG2 UBE3A knockdown cells. However, plasmin regulation and consequent ANXA2 upregulation is critical for production was much reduced in HepG2 SIRT6/ANXA2 double hepatocellular carcinoma progression. knockdown cells as compared with HepG2 UBE3A/SIRT6 double knockdown cells (Fig. 6E). Similar observations were made in H1299 cells (Supplementary Fig. S7G). As plasmin is a known Discussion activator of MMPs, we next investigated MMP activity in these Several E6-independent substrates of UBE3A have been iden- cells. Our results indicated that HepG2 SIRT6 knockdown and tified (27–29). However, the role of UBE3A E3 ligase activity in UBE3A/SIRT6 double knockdown cells exhibited enhanced MMP malignant transformation is poorly understood. Here we report activity as compared with HepG2 UBE3A knockdown cells. On that UBE3A ubiquitylates SIRT6 at the conserved K160 site, the other hand, MMP activity was lower in HepG2 SIRT6/ANXA2 leading to its proteasomal degradation. This results in abrogation double knockdown cells as compared with HepG2 UBE3A/SIRT6 of SIRT6-mediated epigenetic repression of ANXA2, a key deter- double knockdown cells (Fig. 6F). Similar observations were minant of UBE3A oncogenic functions. Thus, our findings expand made in H1299 cells (Supplementary Fig. S7H). Taken together the E6-independent substrate repertoire of UBE3A. Moreover, our our results suggest that ANXA2 repression by SIRT6 plays is critical results also indicate that even though ANXA2 levels decline in for suppressing UBE3A-mediated malignant transformation. tumors with UBE3A knockdown and SIRT6/ANXA2 double To examine the role of SIRT6 in suppressing UBE3A-medi- knockdown, the expression of epithelial marker E-cadherin is ated tumorigenesis, the tumorigenicity of HepG2 UBE3A lower in tumors with SIRT6/ANXA2 double knockdown as com- knockdown, SIRT6 knockdown, UBE3A/SIRT6 double knock- pared with tumors with UBE3A knockdown, and vice versa in case down, and SIRT6/ANXA2 double knockdown cells were exam- of mesenchymal markers vimentin and fibronectin. Thus, besides ined. Upon abrogation of UBE3A expression, significantly ANXA2, UBE3A could also be regulating other factors that play a smaller tumors were observed in comparison with control cells role in determining metastatic progression. As UBE3A targets p53 (Fig. 6G and H; Supplementary Fig. S8A). However, much larger for ubiquitin-mediated proteasomal degradation only in associ- tumors were observed upon SIRT6 knockdown, while upon ation with the HPV oncoprotein E6, our results also indicate that simultaneous knockdown of UBE3A, there was a reduction in p53 is maintained at low levels irrespective of presence or absence

654 Cancer Res; 78(3) February 1, 2018 Cancer Research

UBE3A Downregulates SIRT6 to Promote Tumorigenesis

Figure 6. SIRT6 downregulation facilitates UBE3A oncogenic functions. A, HepG2 control (HepG2), UBE3A knockdown (UBE3Akd), SIRT6 knockdown (SIRT6kd), UBE3A/SIRT6 double knockdown (UBE3Akd/SIRT6kd), and SIRT6/ANXA2 double knockdown (SIRT6kd/ANXA2kd) cells were harvested and immunoblotting was performed. B, HepG2 control (HepG2), UBE3A knockdown (UBE3Akd), SIRT6 knockdown (SIRT6kd), UBE3A/SIRT6 double knockdown (UBE3Akd/ SIRT6kd), and SIRT6/ANXA2 double knockdown (SIRT6kd/ANXA2kd) cells were cultured in complete medium and counted at the indicated time points. Error bars are means SD of three independent experiments with duplicate samples. , P < 0.01; , P < 0.001. C, In vitro invasion of HepG2 control (HepG2), UBE3A knockdown (UBE3Akd), SIRT6 knockdown (SIRT6kd), UBE3A/SIRT6 double knockdown (UBE3Akd/SIRT6kd), and SIRT6/ANXA2 double knockdown (SIRT6kd/ANXA2kd) cells was measured as percentage of cells migrating to chambers with 10% FBS. Error bars are means SD of three independent experiments with duplicate samples. , P < 0.01; , P < 0.001. D, Migration potential of HepG2 control (HepG2), UBE3A knockdown (UBE3Akd), SIRT6 knockdown (SIRT6kd), UBE3A/SIRT6 double knockdown (UBE3Akd/SIRT6kd), and SIRT6/ANXA2 double knockdown (SIRT6kd/ANXA2kd) cells was measured as percentage of cells migrating to chambers with 10% FBS. Error bars are means SD of three independent experiments with triplicate samples. , P < 0.01; , P < 0.001. E, Plasmin production in HepG2 control (HepG2), UBE3A knockdown (UBE3Akd), SIRT6 knockdown (SIRT6kd), UBE3A/ SIRT6 double knockdown (UBE3Akd/SIRT6kd), and SIRT6/ANXA2 double knockdown (SIRT6kd/ANXA2kd) cells was determined. Error bars are means SD of three independent experiments with duplicate samples. , P < 0.001. F, General MMP activity was determined for HepG2 control (HepG2), UBE3A knockdown (UBE3Akd), SIRT6 knockdown (SIRT6kd), UBE3A/SIRT6 double knockdown (UBE3Akd/SIRT6kd), and SIRT6/ANXA2 double knockdown (SIRT6kd/ANXA2kd) cells. Error bars are means SD of three independent experiments with triplicate samples. , P < 0.001. G, HepG2 control (HepG2), UBE3A knockdown (UBE3Akd), SIRT6 knockdown (SIRT6kd), UBE3A/SIRT6 double knockdown (UBE3Akd/SIRT6kd), and SIRT6/ANXA2 double knockdown (SIRT6kd/ANXA2kd) cells were subcutaneously injected into nude mice. Tumor volume was measured on indicated days. The data shown are representative of three independent experiments (n ¼ 5 mice per group) and error bars represent means SD (n ¼ 5 mice per group). H, At the end of 30 days, tumors were excised and weighed. The data shown are representative of three independent experiments (n ¼ 5 mice per group) and error bars represent means SD (n ¼ 5 mice per group).

www.aacrjournals.org Cancer Res; 78(3) February 1, 2018 655

Kohli et al.

Figure 7. Role of ANXA2 in UBE3A-mediated tumorigenesis. A, HepG2Luc2 cells were stably transfected with UBE3A shRNA, SIRT6 shRNA, and SIRT6 shRNA along with UBE3A shRNA or ANXA2 shRNA as indicated. HepG2Luc2 cells stably transfected with empty vector was used as control (Control). These cells were injected into the liver of nude mice. Bioluminescence imaging was performed weekly and representative images are shown. The data shown are representative of three independent experiments (n ¼ 5 mice per group). B, Bioluminescence quantiﬁcation (as in A)wasperformedattheindicatedtimepoints. The data shown are representative of three independent experiments (n ¼ 5 mice per group). Error bars represent means SD from ﬁve individual mice. , P < 0.01; , P < 0.001. C, Representative image of immunostaining of SIRT6, ANXA2, and UBE3A in different grades of human liver carcinoma (HCC) and matched normal adjacent tissue (NAT) sections. DAPI was used to counterstain nucleus. D, Quantitation of UBE3A levels in different grades of human liver carcinoma normalized with respect to matched normal adjacent tissue. The data shown are representative of three independent experiments. Error bars, means SD. E, Quantitation of ANXA2 levels in different grades of human liver carcinoma normalized with respect to matched normal adjacent tissue. The data shown are representative of three independent experiments. Error bars, means SD. F, Quantitation of SIRT6 levels in different grades of human liver carcinoma normalized with respect to matched normal adjacent tissue. The data shown are representative of three independent experiments. Error bars, means SD.

656 Cancer Res; 78(3) February 1, 2018 Cancer Research

UBE3A Downregulates SIRT6 to Promote Tumorigenesis

of UBE3A (Supplementary Fig. S8C). Thus, the difference in to exhibit a stepwise reduction from preneoplastic stages of tumor growth in absence of UBE3A is not due to p53 as there hepatocellular carcinoma (HCC; ref. 34), whereas ANXA2 levels is no significant change in its levels. It will be of interest to use are upregulated in HCC (6, 7). Our studies indicate that SIRT6 proteomics-based approaches to identify diverse substrates of represses ANXA2 expression in association with transcription UBE3A involved in malignant transformation. factor ELK1. SIRT6-mediated repression of ANXA2 is a key deter- SIRT6 has been reported to be targeted for proteasomal deg- minant of its tumor suppressor functions. In conclusion, our radation by E3 ligases including MDM2, ITCH, and CHIP. Using results provide insights on the significance of SIRT6 degradation exogenous overexpression-based studies, MDM2 has been for UBE3A-mediated tumorigenesis. reported to ubiquitylate SIRT6 and target it for proteasomal degradation (30). But the pertinence of MDM2-mediated ubiqui- Disclosure of Potential Conflicts of Interest tylation in SIRT6 regulation under stress conditions is yet unclear. No potential conflicts of interest were disclosed. ITCH is a HECT-type E3 ligase predominantly localized in late endosomal compartments and lysosomes (31). Loss of ITCH Authors' Contributions downregulates SIRT6 ubiquitylation and induces SIRT6 protein Conception and design: S. Kohli, A. Bhardwaj, S. Das levels (32). However, molecular mechanism of ITCH-mediated Development of methodology: S. Kohli, A. Bhardwaj, S. Das regulation of SIRT6 is yet unclear. Our studies indicate that there is Acquisition of data (provided animals, acquired and managed patients, no significant change in ITCH levels upon metabolic stress (Sup- provided facilities, etc.): S. Kohli, A. Bhardwaj, R. Kumari, S. Das plementary Fig. S9A). Neither its mutants nor wild-type SIRT6 Analysis and interpretation of data (e.g., statistical analysis, biostatistics, exhibited discernable interaction with ITCH under unstressed and computational analysis): S. Kohli, A. Bhardwaj, R. Kumari, S. Das Writing, review, and/or revision of the manuscript: S. Kohli, A. Bhardwaj, starvation conditions (Supplementary Fig. S3A and S9B). Pres- R. Kumari, S. Das ence or absence of ITCH had no effect on SIRT6 polyubiquityla- Administrative, technical, or material support (i.e., reporting or organizing tion under unstressed conditions, whereas under metabolic stress data, constructing databases): A. Bhardwaj, S. Das conditions, even in presence of ITCH, SIRT6 polyubiquitylation Study supervision: S. Das was abrogated as UBE3A levels were downregulated (Supplemen- tary Fig. S9C). Thus, our data suggest that UBE3A plays a key role Acknowledgments in determining SIRT6 levels independent of presence or absence We thank the members of the Molecular Oncology Laboratory for helpful of ITCH. Interestingly, CHIP noncanonically ubiquitylates SIRT6, discussions. We are also thankful to Dr. Vivek Rao, CSIR-Institute of Genomics & resulting in stabilization of SIRT6 protein (33). CHIP-dependent Integrative Biology (IGIB) for helping with live animal imaging. The authors fi SIRT6 stabilization promotes DNA repair. Our studies indicate would also like to acknowledge the nancial support received from NII Core Fund. S. Kohli is supported by a fellowship from the Council for Scientific and that UBE3A polyubiquitylates SIRT6 at K160 residue, which Industrial Research, Government of India and A. Bhardwaj was supported by a targets SIRT6 for proteasomal degradation. However, across fellowship from the Department of Biotechnology, Government of India. This diverse cell types, the induction in SIRT6 protein levels upon work was supported by a grant from the Department of Biotechnology, abrogation of UBE3A expression varies, which indicates that, in Government of India (BT/PR12875/BRB/10/1394/2015 to S. Das). addition to UBE3A, there could be other factors playing a role in determining SIRT6 levels. Thus, it would be interesting to use The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked proteomics-based approaches to identify other factors that play a advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate role in regulating SIRT6 levels. this fact. There is mounting evidence indicating that SIRT6 functions as a tumor suppressor while ANXA2 has been associated with Received June 8, 2017; revised September 29, 2017; accepted November 29, increased proliferation and invasiveness. SIRT6 levels are reported 2017; published OnlineFirst December 7, 2017.

References 1. Scheffner M, Huibregtse JM, Vierstra RD, Howley PM. The HPV-16 E6 and 7. Mohammad HS, Kurokohchi K, Yoneyama H, Tokuda M, Morishita E6-AP complex functions as a ubiquitin-protein ligase in the ubiquitina- A, Jian G, et al. Annexin A2 expression and phosphorylation are tion of p53. Cell 1993;75:495–505. up-regulated in hepatocellular carcinoma. Int J Oncol 2008;33: 2. Deng S, Zhou H, Xiong R, Lu Y, Yan D, Xing T, et al. Over-expression of 1157–63. genes and proteins of ubiquitin speciﬁc peptidases (USPs) and proteasome 8. Xu XH, Pan W, Kang LH, Feng H, Song YQ. Association of annexin A2 with subunits (PSs) in breast cancer tissue observed by the methods of RFDD- cancer development (Review). Oncol Rep 2015;33:2121–8. PCR and proteomics. Breast Cancer Res Treat 2007;104:21–30. 9. Sebastian C, Zwaans BM, Silberman DM, Gymrek M, Goren A, Zhong L, 3. Srinivasan S, Nawaz Z. E3 ubiquitin protein ligase, E6-associated protein et al. The histone deacetylase SIRT6 is a tumor suppressor that controls (E6-AP) regulates PI3K-Akt signaling and prostate cell growth. Biochim cancer metabolism. Cell 2012;151:1185–99. Biophys Acta 2011;1809:119–27. 10. Bhardwaj A, Das S. SIRT6 deacetylates PKM2 to suppress its nuclear 4. Ramamoorthy S, Tufail R, Hokayem JE, Jorda M, Zhao W, Reis Z, et al. localization and oncogenic functions. Proc Natl Acad Sci U S A 2016; Overexpression of ligase defective E6-associated protein, E6-AP, results in 113:E538–47. mammary tumorigenesis. Breast Cancer Res Treat 2012;132:97–108. 11. Kugel S, Sebastian C, Fitamant J, Ross KN, Saha SK, Jain E, et al. SIRT6 5. Louria-Hayon I, Alsheich-Bartok O, Levav-Cohen Y, Silberman I, Berger M, suppresses pancreatic cancer through control of Lin28b. Cell 2016;165: Grossman T, et al. E6AP promotes the degradation of the PML tumor 1401–15. suppressor. Cell Death Differ 2009;16:1156–66. 12. Min L, Ji Y, Bakiri L, Qiu Z, Cen J, Chen X, et al. Liver cancer initiation is 6. Zhou X, Deng S, Liu H, Liu Y, Yang Z, Xing T, et al. Knockdown of ubiquitin controlled by AP-1 through SIRT6-dependent inhibition of survivin. protein ligase E3A affects proliferation and invasion, and induces apoptosis Nat Cell Biol 2012;14:1203–11. of breast cancer cells through regulation of annexin A2. Mol Med Rep 13. Kugel S, Mostoslavsky R. Chromatin and beyond: the multitasking roles 2015;12:1107–13. for SIRT6. Trends Biochem Sci 2014;39:72–81.

www.aacrjournals.org Cancer Res; 78(3) February 1, 2018 657

Kohli et al.

14. Kanfi Y, Shalman R, Peshti V, Pilosof SN, Gozlan YM, Pearson KJ, et al. 25. Cea M, Cagnetta A, Adamia S, Acharya C, Tai YT, Fulciniti M, et al. Evidence Regulation of SIRT6 protein levels by nutrient availability. FEBS Lett for a role of the histone deacetylase SIRT6 in DNA damage response of 2008;582:543–8. multiple myeloma cells. Blood 2016;127:1138–50. 15. Satija YK, Das S. Tyr99 phosphorylation determines the regulatory milieu 26. Diaz VM, Hurtado M, Thomson TM, Reventos J, Paciucci R. Specific of tumor suppressor p73. Oncogene 2016;35:513–27. interaction of tissue-type plasminogen activator (t-PA) with annexin II on 16. Jones RG, Plas DR, Kubek S, Buzzai M, Mu J, Xu Y, et al. AMP-activated the membrane of pancreatic cancer cells activates plasminogen and pro- protein kinase induces a p53-dependent metabolic checkpoint. Mol Cell motes invasion in vitro. Gut 2004;53:993–1000. 2005;18:283–93. 27. Greer PL, Hanayama R, Bloodgood BL, Mardinly AR, Lipton DM, Flavell 17. Talis AL, Huibregtse JM, Howley PM. The role of E6AP in the regulation of SW, et al. The angelman syndrome protein Ube3A regulates synapse p53 protein levels in human papillomavirus (HPV)-positive and HPV- development by ubiquitinating arc. Cell 2010;140:704–16. negative cells. J Biol Chem 1998;273:6439–45. 28. Hattori T, Kishino T, Stephen S, Eberspaecher H, Maki S, Takigawa M, et al. 18. Brunet A, Bonni A, Zigmond MJ, Lin MZ, Juo P, Hu LS, et al. Akt promotes E6-AP/UBE3A protein acts as a ubiquitin ligase toward SOX9 protein. J Biol cell survival by phosphorylating and inhibiting a Forkhead transcription Chem 2013;288:35138–48. factor. Cell 1999;96:857–68. 29. Gossan NC, Zhang F, Guo B, Jin D, Yoshitane H, Yao A, et al. The E3 19. Ramaswamy S, Nakamura N, Sansal I, Bergeron L, Sellers WR. A novel ubiquitin ligase UBE3A is an integral component of the molecular circa- mechanism of gene regulation and tumor suppression by the transcription dian clock through regulating the BMAL1 transcription factor. Nucleic factor FKHR. Cancer Cell 2002;2:81–91. Acids Res 2014;42:5765–75. 20. Nakamura N, Ramaswamy S, Vazquez F, Signoretti S, Loda M, Sellers 30. ThirumurthiU,ShenJ,XiaW,LaBaffAM,WeiY,LiCW,etal.MDM2- WR. Forkhead transcription factors are critical effectors of cell death mediated degradation of SIRT6 phosphorylated by AKT1 promotes and cell cycle arrest downstream of PTEN. Mol Cell Biol 2000;20: tumorigenesis and trastuzumab resistance in breast cancer. Sci Signal 8969–82. 2014;7:ra71. 21. HuangB,DeoraAB,HeKL,ChenK,SuiG,JacovinaAT,etal.Hypoxia- 31. Melino G, Gallagher E, Aqeilan RI, Knight R, Peschiaroli A, Rossi M, et al. inducible factor-1 drives annexin A2 system-mediated perivascular Itch: a HECT-type E3 ligase regulating immunity, skin and cancer. Cell fibrin clearance in oxygen-induced retinopathy in mice. Blood 2011; Death Differ 2008;15:1103–12. 118:2918–29. 32. Stohr R, Mavilio M, Marino A, Casagrande V, Kappel B, Mollmann J, et al. 22. Moreau K, Ghislat G, Hochfeld W, Renna M, Zavodszky E, Runwal G, et al. ITCH modulates SIRT6 and SREBP2 to influence lipid metabolism and Transcriptional regulation of Annexin A2 promotes starvation-induced atherosclerosis in ApoE null mice. Sci Rep 2015;5:9023. autophagy. Nat Commun 2015;6:8045. 33. Ronnebaum SM, Wu Y, McDonough H, Patterson C. The ubiquitin ligase 23. Zhong L, D'Urso A, Toiber D, Sebastian C, Henry RE, Vadysirisack DD, et al. CHIP prevents SirT6 degradation through noncanonical ubiquitination. The histone deacetylase Sirt6 regulates glucose homeostasis via Hif1alpha. Mol Cell Biol 2013;33:4461–72. Cell 2010;140:280–93. 34. Marquardt JU, Fischer K, Baus K, Kashyap A, Ma S, Krupp M, et al. Sirtuin-6- 24. Sundaresan NR, Vasudevan P, Zhong L, Kim G, Samant S, Parekh V, et al. dependent genetic and epigenetic alterations are associated with poor The sirtuin SIRT6 blocks IGF-Akt signaling and development of cardiac clinical outcome in hepatocellular carcinoma patients. Hepatology 2013; hypertrophy by targeting c-Jun. Nat Med 2012;18:1643–50. 58:1054–64.

658 Cancer Res; 78(3) February 1, 2018 Cancer Research

SIRT6 Is a Target of Regulation by UBE3A That Contributes to Liver Tumorigenesis in an ANXA2-Dependent Manner

Saishruti Kohli, Abhishek Bhardwaj, Richa Kumari, et al.

Cancer Res 2018;78:645-658. Published OnlineFirst December 7, 2017.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-17-1692

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2017/12/07/0008-5472.CAN-17-1692.DC1

Cited articles This article cites 34 articles, 9 of which you can access for free at: http://cancerres.aacrjournals.org/content/78/3/645.full#ref-list-1

Citing articles This article has been cited by 1 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/78/3/645.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/78/3/645. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.