Tumor Suppressor p14ARF Enhances IFN- −γ Activated Immune Response by Inhibiting PIAS1 via SUMOylation

This information is current as Jennifer Alagu, Yoko Itahana, Faizal Sim, Sheng-Hao Chao, of September 27, 2021. Xuezhi Bi and Koji Itahana J Immunol published online 30 May 2018 http://www.jimmunol.org/content/early/2018/05/29/jimmun ol.1800327 Downloaded from

Supplementary http://www.jimmunol.org/content/suppl/2018/05/29/jimmunol.180032 Material 7.DCSupplemental http://www.jimmunol.org/ Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

by guest on September 27, 2021 *average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2018 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published May 30, 2018, doi:10.4049/jimmunol.1800327 The Journal of Immunology

Tumor Suppressor p14ARF Enhances IFN-g–Activated Immune Response by Inhibiting PIAS1 via SUMOylation

Jennifer Alagu,* Yoko Itahana,* Faizal Sim,† Sheng-Hao Chao,‡,x Xuezhi Bi,‡ and Koji Itahana*

The ability of cells to induce the appropriate transcriptional response to inflammatory stimuli is crucial for the timely induction of host defense mechanisms. Although a role for tumor suppressor p14ARF (ARF) in the innate immune response was previously demonstrated, the underlying mechanism is still unclear. ARF is a potent upregulator of protein SUMOylation; however, no as- sociation of this function with the immune system has been made. In this study, we show the unique role of ARF in IFN-g–induced immune response using human cell lines. Through a systematic search of proteins SUMOylated by ARF, we identified PIAS1, an inhibitor of IFN-activated transcription factor STAT1, as a novel ARF-binding partner and SUMOylation target. In response to

IFN-g treatment, ARF promoted PIAS1 SUMOylation to inhibit the ability of PIAS1 to attenuate IFN-g response. Wild-type, but Downloaded from not ARF mutants unable to enhance PIAS1 SUMOylation, prevented the PIAS1-mediated inhibition of IFN-g response. Con- versely, the SUMO-deconjugase SENP1 deSUMOylated PIAS1 to reactivate PIAS1 that was inhibited by ARF. These findings suggest that PIAS1 function is negatively modulated by SUMO modification and that SUMOylation by ARF is required to inhibit PIAS1 activity and restore IFN-g–induced transcription. In the presence of ARF, in which case PIAS1 is inhibited, depletion of PIAS1 did not have an additive effect on IFN-g response, suggesting that ARF-mediated enhancement of IFN-g response is mainly

due to PIAS1 inhibition. Our findings reveal a novel function of ARF to inhibit PIAS1 by enhancing SUMOylation to promote the http://www.jimmunol.org/ robust induction of IFN-g response. The Journal of Immunology, 2018, 201: 000–000.

nflammation is a protective mechanism regulated by the inflammatory signaling is a critical factor in determining whether host’s immune system against potential or further harm. inflammation is resolved or prolonged. A chronic inflamma- I Detection of injury, infection, or transforming agents triggers tory microenvironment promotes tumorigenesis as well as other the release of chemokines from affected and surrounding cells for chronic illnesses (1). the acute mounting of immune cells. Activated immune cells se- The IFNs mediate antitumor and antiviral responses. Type I IFNs

crete proinflammatory such as IFNs and TNF-a and (IFN-a and IFN-b) are produced by all nucleated cells and bind by guest on September 27, 2021 chemokines to promote cytotoxicity toward infected cells and IFN-a/b receptor subunits to induce either the homodimerization microorganisms until clearance of threat is achieved. The ability of STAT1 or heterodimerization of STAT1 and STAT2. Type II of host cells to appropriately sense, respond to, and relay IFN (IFN-g) is secreted by immune cells and binds to the IFN-g receptor to induce STAT1 homodimerization. The activated

*Cancer and Stem Cell Biology Program, Duke–NUS Medical School, Singapore dimers transcriptionally initiate an inflammatory response from 169857, Singapore; †Temasek Polytechnic School of Applied Science, Singapore the promoters of IFN-stimulated (ISGs) that contain IFN- ‡ 529757, Singapore; Proteomics Groups, Bioprocessing Technology Institute, stimulated response elements (ISRE) and/or IFN-g–activated site Agency for Science, Technology and Research, Singapore 138668, Singapore; and xDepartment of Microbiology, National University of Singapore, Singapore 117597, (GAS) elements (1, 2). Singapore The tumor suppressor protein p14ARF (ARF) is expressed ORCIDs: 0000-0003-3560-4560 (Y.I.); 0000-0001-9475-8709 (S.-H.C.); 0000-0001- ubiquitously at very low levels and upregulated in response to 6702-0805 (X.B.); 0000-0002-7241-2894 (K.I.). oncogenic stress (3). The canonical function of ARF is to inhibit Received for publication March 5, 2018. Accepted for publication May 7, 2018. the E3 ligase activity of MDM2 toward , resulting in stabili- This work was supported by a Duke–NUS Medical School core grant, Singapore zation of p53 to induce cell cycle arrest, senescence, or apoptosis Ministry of Health’s National Medical Research Council Grant NMRC/OFIRG/ 15nov049/2016, National Research Foundation Competitive Research Program upon oncogenic stress. As an obstacle for transformation, ARF Grant NRF2012NRF-CRP001-056, and Singapore Ministry of Education Academic is lost in ∼40% of human cancers (4). Recently, several reports Research Fund Tier 2 Grants MOE2013-T2-2-123 and MOE2017-T2-1-081 (to K.I.). demonstrated that ARF has a noncanonical function to sense J.A., F.S., and X.B. performed experiments. J.A., Y.I., S.-H.C., and K.I. designed the inflammatory stimuli (5) and promote the robust induction of experiments and J.A. analyzed the results. J.A., Y.I., and K.I. interpreted the results and wrote the manuscript. proinflammatory genes independently of p53 (6). The underlying Address correspondence and reprint requests to Prof. Koji Itahana, Cancer and Stem mechanisms, however, are still unclear. Cell Biology Program, Duke–NUS Medical School, Room 07-18, Level 7, 8 College The promotion of small ubiquitin-like modifier (SUMO) con- Road, Singapore 169857, Singapore. E-mail address: [email protected] jugation on various proteins is another p53-independent function of The online version of this article contains supplemental material. ARF (7). SUMOylation is a posttranslational process by which Abbreviations used in this article: ARF, p14ARF; co-IP, coimmunoprecipitation; proteins are modified via the sequential actions of E1 (SAE1/2), GAS, IFN-g–activated site; ISRE, IFN-stimulated response element; LC-MS, liquid chromatography–mass spectrometry; MS, mass spectrometry; NF-kB-RE, NF-kB- E2 (Ubc9), and E3 enzymes in a manner analogous to ubiquitination. response element; PIAS, protein inhibitor of activated STAT; SENP, sentrin- SUMOylation serves to regulate protein activity, stability, and specific protease; siRNA, small interfering RNA; STUbL, SUMO-targeted ubiquitin localization. The conjugation of SUMO2 and SUMO3 is pre- ligase, SUMO, small ubiquitin-like modifier. dominantly induced as a protective response to cellular stresses Copyright Ó 2018 by The American Association of Immunologists, Inc. 0022-1767/18/$35.00 such as infection and inflammatory stimuli (8, 9). Although ARF

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1800327 2 ARF PROMOTES IFN-g RESPONSE BY INHIBITING PIAS1 is not a SUMO E3 ligase, ARF has been shown to promote the CTGCGGATCCAG-39 and 59-CTGGATCCGCAGTCAACCTCTCTT- SUMOylation of its interaction partners such as p53 (10, 11), TAATTAAAGCTC-39; K368R, 59-CATTCAGATGAATGAGAGAAA- MDM2 (12), NPM1 (13), and Werner’s helicase (14). We have ACCAACCTGGGTTTG-39 and 59-CAAACCCAGGTTGGTTTTCTC- TCATTCATCTGAATG-39; K408, 59-GATGAAATACAATTTAGGGA- previously reported that ARF enhances NPM1 SUMOylation to GGATGGCACTTGGG-39 and 59-CCCAAGTGCCATCCTCCCTAAA- repress centrosome amplification (13). However, none have been TTGTATTTCATC-39; and K456, 59-GTCCTCAAATAAAAACAAGA- characterized for immunological roles. GAGTAGAAGTGATTGACCTAAC-3 9 and 59-GTTAGGTCAATCAC- Protein inhibitor of activated STAT (PIAS) family proteins were TTCTACTCTCTTGTTTTTATTTGAGGAC-39. ARF constructs were previously described elsewhere (13). Adenoviral ARF was generated originally identified as inhibitors of STATactivity that also possess by ViraQuest. All transfections were carried out via calcium phosphate an E3 SUMO ligase property. Four members constitute the PIAS precipitation. family and share more than 40% sequence similarity and conserved functional domains (15, 16). During IFN signaling, PIAS1 is phos- Cell culture and reagents phorylated to suppress STAT1 activity (17, 18). PIAS1 inhibits H1299 lung carcinoma, U2OS osteosarcoma, HEK 293T kidney cells, and STAT1 activity by several mechanisms such as through binding the WI-38 fibroblast cell lines (ATCC) were cultured in DMEM supplemented transactivation domain of STAT1, enhancing the SUMOylation of with 10% FBS and penicillin/streptomycin antibiotics. H1299 cells stably expressing His6-SUMO3 were obtained by transfection of pcDNA3-His6- STAT1, or directly binding to the ISG promoters to block STAT1 SUMO3 construct and subsequent selection with G418. Cell cultures were recruitment (19–24). Although PIAS1 is known to be SUMOylated, maintained at 37˚C in a 5% CO2 incubator. IFN-g (SRP3058; Sigma- the regulation and function of PIAS1 SUMOylation is still poorly Aldrich), IFN-a (ab48750; Abcam), and TNF-a (T6674; Sigma-Aldrich) understood. were purchased commercially. In this study, we identified PIAS1 as a novel binding partner and RNA interference Downloaded from SUMOylation target of ARF. In response to IFN-g, ARF enhanced PIAS1 SUMOylation, and this served to inhibit PIAS1 activity Small interfering RNA (siRNA) targeted against PIAS1 and ARF or Si- lencer Negative Control #1 siRNA (Thermo Fisher Scientific) were and rescued PIAS1-mediated inhibition of IFN-g–activated re- transfected via calcium phosphate precipitation and used for analysis or porters. Our findings provide a novel SUMO-mediated mechanism experimentation 48 h after. The following siRNA sequences were used: by which ARF promotes IFN-g–induced immune response. siPIAS1, 59-CGAAUGAACUUGGCAGAAAdtdt-39; siARF (no. 1), 59-GAUGCUACUGAGGAGCCAGCGUCTAdtdt-39; and siARF (no. 2), http://www.jimmunol.org/ 59-GCGAGAACAUGGUGCGCAGUUdtdt-39. Materials and Methods Plasmids and adenovirus RNA isolation and expression analyses Full-length PIAS1 (aa 1–650) was amplified from WI-38 normal human fibro- Cells were treated with 10 ng/ml IFN-g for 4 h. Total RNA was isolated blast and cloned into vector pcDNA3.1(+) (Thermo Fisher Scientific) with FLAG using RNeasy Mini Kit (Qiagen), and cDNA was synthesized using iScript tagging. pcDNA3-GFP-PIAS1, pcDNA3-GFP-PIAS1 C351F, pcDNA3-FLAG- Reverse Transcription kit (Bio-Rad). Quantitative RT-PCR was performed SENP1,andpcDNA3-HA-SUMO3werekindlyprovidedbyDr.Y.Zhang with qPCRBIO SyGreen (PCR Biosystems) using the CFX96 Touch Real- (University of North Carolina at Chapel Hill). SUMO3 was extracted by re- Time PCR Detection System (Bio-Rad). TBP and HPRT1 were used as striction enzyme digestion and ligated into His tag–containing pcDNA3.1(+). internal controls for the calculation of relative gene expression using Bio- 6 Rad CFX Manager software. The following primers were used: CXCL9, ISRE-luciferase, GAS-luciferase, NF-kB-response element (NF-kB- by guest on September 27, 2021 RE)-luciferase, and pRL-TK reporter constructs were kindly provided by 59-TTGGGCATCATCTTGCTGGTTCT-39 and 59-TGGCTGACCTGTTT- Dr. M. Krishnan (Duke–NUS Medical School). All cloned constructs were CTCCCACTT-39; CXCL10,59-GAAGCAGTTAGCAAGGAAATGT-39 confirmed by direct DNA sequencing. Deletion mutants of PIAS1 were and 59-GACATATACTCCATGTAGGGAAGTGA-39; GBP1,59-TTCC- cloned using the following primers according to previously characterized AAAACTAAAACTCTTTCAGGA-39 and 59-GGTCAGCACCAGGCTC- regions (25) and inserted into pcDNA3-FLAG vectors: 1–650, 59-GCGGA- TCTA-39; TBP,59-CGCCGAATATAATCCCAAGC-39 and 59-TCCTG- CAGTGCGGAACTAAAGCAAAT-39 and 59-TCAGTCCAATGAAATAAT- TGCACACCATTTTCC-39; and HPRT1,59-TGACACTGGCAAAACAA- GTCTGGTATGATGC-39;1-415,59-GCGGACAGTGCGGAACTAAAGC- TGCA-39 and 59-GGTCCTTTTCACCAGCAAGCT-39. AAAT-39 and 59-TCACGGTGCCCAAGTGCCATCCT-39; 1–344, 59-GCG- Protein analysis GACAGTGCGGAACTAAAGCAAAT-39 and 59-TCAAATTGTCAGCCGC- ATTTTACCAAGT-39; 89–344, 59-TCTCCATCTACCATTCCACAACTCA- Abs against FLAG (M2; Sigma-Aldrich), GFP (Fitzgerald), PIAS1 (D33A7; CTTACGA-39 and 59-TCAAATTGTCAGCCGCATTTTACCAAGT; 89–415, Cell Signaling Technology), Myc (ab9106; Abcam), and b-actin (C4; 59-TCTCCATCTACCATTCCACAACTCACTTACGA-39 and 59-TCACGG- Millipore) were purchased commercially. Rabbit polyclonal ARF Ab was TGCCCAAGTGCCATCCT-39; 345–650, 59-TGTCGGGCCCTTACATGTT- custom produced by using human ARF C-terminal sequence as the epitope CTCATCTAC-39 and 59-TCAGTCCAATGAAATAATGTCTGGTATG- (26). For whole cell lysate, cells were lysed in 2% SDS lysis buffer ATGC-39; and 416–650, 59-ATGAGATCAAAAAAGGAAGTACAG- (50 mM Tris-HCl pH 6.8, 10% glycerol, and 2% SDS). Protein analysis GAAGTTTCTGC-39 and 59-TCAGTCCAATGAAATAATGTCTGGT- was conducted via Western blotting. HRP-conjugated secondary Abs were ATGATGC-39. PIAS1 point mutants were introduced by the PCR- used for chemiluminescent signal detection with reagents according to based HiFi Site-Directed Mutagenesis Kit (New England Biolabs) manufacturer’s instructions (Thermo Fisher Scientific). Odyssey Infrared using the following primers: K8R, 59-GACAGTGCGGAACTAAGG- Imaging System (LI-COR Biosciences) was used for the detection of CAAATGGTTATGAGC-39 and 59-GCTCATAACCATTTGCCTTAG- fluorescently labeled secondary Abs. TTCCGCACTGTC-39;K46R,59-GCCCTGCATTTGCTAAGGGCT- GGCTGTAGTCCTG-39 and 59-CAGGACTACAGCCAGCCCTTAGC- Coimmunoprecipitation AAATGCAGGGC-39;K58R,59-GTGCAAATGAAAATTAGGGAAC- TCTATAGGCGG-39 and 59-CCGCCTATAGAGTTCCCTAATTTTCA- Cells were lysed with 0.1% NP-40 lysis buffer (50 mM Tris-HCl pH 7.5, TTTGCAC-39; K117R, 59-GTTTCTCTTCTGGGACCTAGACATGA- 150 mM NaCl, 50 mM NaF, 1 mM Na3VO4, 1 mM PMSF, and protease ACTGGAACTCCC-39 and 59-GGGAGTTCCAGTTCATGTCTAGGT- inhibitor mixture), precleared with Sepharose CL-4B beads (Sigma- CCCAGAAGAGAAAC-39; K137R, 59-GTCCATCCGGATATAAGA- Aldrich), and incubated with Abs or Ab-conjugated beads for 16 h at m CTTCAAAAATTACC-39 and 59-GGTAATTTTTGAAGTCTTATATC- 4˚C with constant rotation. In the first case, 20 l Protein A/G Agarose CGGATGGAC-39; K140R, 59-CCGGATATAAAACTTCAAAGATT- beads (Thermo Fisher Scientific) was subsequently added and further incubated ACCATTTTATGATTTAC-39 and 59-GTAAATCATAAAATGGTAAT- for 2–4 h for the capture of Ab-protein complexes. Beads were washed CTTTGAAGTTTTATATCCGG-39; K152R, 59-CTGGATGAACTGA- three times with 0.1% NP-40 lysis buffer and resuspended in 2% SDS- TAAGACCCACCAGTCTAGC-39 and 59-GCTAGACTGGTGGGTC- containing loading buffer for elution and analysis by Western blotting. TTATCAGTTCATCCAG-39; K238R, 59-CCTTCCACCTACAAGAA- SUMOylation assay ATGGCGTGGAACC-39 and 59-GGTTCCACGCCATTTCTTGTAGG- TGGAAGG-39; K244, 59-AAATGGCGTGGAACCAAGGCGACCCA- Plasmid-encoding His6-SUMO3 was cotransfected with combinations of GCCGACCAA-39 and 59-TTGGTCGGCTGGGTCGCCTTGGTTCC- ARF-, PIAS1-, and SENP1-encoding plasmids. Forty-eight hours post- ACGCCATTT-39; K315R, 59-GAGCTTTAATTAAAGAGAGGTTGA- transfection, cells were trypsinized, washed, and divided into 20 and 80% The Journal of Immunology 3 aliquots. Twenty percent was lysed with 2% SDS lysis buffer for input either analyzed by silver staining or trypsin digested for LC-MS verification by Western blotting. Eighty percent was lysed with strong analysis. The eluates from the uninfected parental H1299 cells and denaturing buffer (6 M guanidinium-HCl, 0.1 M Na2HPO4-NaH2PO4 the cells expressing control vector served to identify nonspecifi- pH 8, 0.01 M Tris-HCl pH 8, 5 mM imidazole, and 10 mM 2-ME) and incubated with Ni-NTA beads (Thermo Fisher Scientific) for 4 h at room cally recovered polypeptides (Supplemental Fig. 1). Among the temperature with constant rotation. Beads were subsequently washed two unique polypeptides isolated from the ARF-infected cells, we times with lysis buffer and two times with wash buffer (8 M urea, 0.1 M focused our attention on PIAS1. Because PIAS1 is a potent in- Na2HPO4-NaH2PO4 pH 6.3, 0.01 M Tris-HCl pH 6.8, 10 mM 2-ME, and hibitor of the activated form of STAT1 (19, 23), which triggers 0.2% Triton X-100). His6-SUMO3–conjugated proteins were eluted from beads with loading buffer containing 200 mM imidazole and were ana- proinflammatory responses, we considered PIAS1 as a potential lyzed by Western blotting. target of ARF’s activity in the proinflammatory response. PIAS1 is known to undergo SUMOylation (28); however, the regulation Nano liquid chromatography–tandem mass spectrometry and and significance of SUMOylation on PIAS1 in the immune mass spectrometry data analysis context is unknown. Eluted protein from the SUMOylation assay was reduced, alkylated, ARF is known to enhance SUMOylation of proteins to which it trypsin-digested, acidified, and desalted before concentration by SpeedVac binds. Therefore, we first tested whether ARF could interact with vacuum as detailed in a filter-aided sample preparation method for liquid PIAS1 using H1299 cells. Coimmunoprecipitation (co-IP) of ec- chromatography–mass spectrometry (LC-MS) outlined by Wisniewski et al. (27). Mass spectrometry (MS) analysis was performed as previously topic GFP-tagged PIAS1 (Fig. 1A) or endogenous PIAS1 (Fig. 1B) described (13). Briefly, LTQ-Orbitrap Velos Pro ETD Mass Spectrometer with ectopic FLAG-tagged ARF revealed their binding. Recipro- (Thermo Fisher Scientific) was used for analysis, and tandem MS data cal co-IP of endogenous ARF with ectopic FLAG-PIAS1 also were analyzed with Sequest HT in Protein Discoverer 1.4 SP1 software confirmed the interaction (Fig. 1C). To define the region of ARF Downloaded from (Thermo Fisher Scientific). The human protein database from UniProtKB/ Swiss-Prot Release 2014_03 of Mar 19, 2014, was used for the database required for PIAS1 binding, we performed co-IP of GFP-tagged search, and decoy database search for target false detection rate was set as PIAS1 with ARF C-terminal (1–64) and N-terminal (65–132) q-value = 0.01. Multiconsensus report was generated from multiple deletion mutants using ARF-null U2OS cells. As shown in LC-MS results for different samples with both collision-induced dissoci- Fig. 1D, only the full-length and 1–64 deletion mutant of ARF ation and high-energy collision dissociation acquisition mode. could pull down PIAS1, indicating that only the N terminus half

Duolink in situ proximity ligation assay of ARF is required for PIAS1 interaction. We also attempted to http://www.jimmunol.org/ map the region of PIAS1 to which ARF binds and found that ARF Endogenous protein interaction was demonstrated using the Duolink II detection kit according to manufacturer’s guidelines (Sigma-Aldrich). Cells could form immunocomplexes with each of the PIAS1 deletion were fixed with 4% paraformaldehyde, permeabilized, blocked, and in- mutants, although with varying intensities (Supplemental Fig. 2). cubated with mouse anti-ARF and rabbit anti-PIAS1 primary Abs. Thus, as reported of other ARF-binding partners (11, 14, 29), ARF Species-matched secondary Abs conjugated with complementary oligo- interacts with multiple regions of PIAS1 (Supplemental Fig. 2). nucleotides were ligated and polymerized if bound to primary Abs within 40 nm distance. Addition of fluorescent probes indicated the location of Finally, to determine the intracellular location of ARF–PIAS1 interaction, which was detected with Olympus IX71 inverted microscope. association, proximity ligation assay of endogenous ARF and PIAS1 was performed to reveal their interaction in the nucleus

Dual luciferase reporter assay by guest on September 27, 2021 (Fig. 1E). ISRE, GAS, or NF-kB element–containing reporter plasmids were trans- fected together with pRL-TK plasmid into HEK 293T cells. Thirty hours PIAS1 is a novel SUMOylation target of ARF posttransfection, cells were stimulated with either 10 ng/ml IFN-g (unless To validate the ARF-mediated SUMOylation of PIAS1 in vivo, otherwise stated), 1 ng/ml IFN-a, or 20 ng/ml TNF-a for 16 h. Luciferase His -SUMO3 stably expressing H1299 (H1299-His -SUMO3) and renilla luciferase activities were subsequently measured using a Dual- 6 6 Luciferase Reporter Assay Kit (Promega). Transfection efficiency was cells were transfected with GFP-tagged PIAS1 alone or with in- normalized by renilla luciferase activity. creasing amounts of ARF. Electrophoretic analysis of whole cell lysates revealed the presence of slower-migrating GFP-tagged Statistical analysis PIAS1 bands, which suggested the induction of poly-SUMO Data are presented as mean 6 SD of three independent experiments. Two- modification, the intensity of which correlated with ARF dosage tailed unpaired Student t test was used to determine significant differences (Fig. 2A). To determine whether PIAS1 was indeed modified by when p , 0.05. All statistical analyses were conducted using GraphPad Prism 7.02 software. SUMO, PIAS1 was expressed without or with ARF in H1299- His6-SUMO3 cells, and the lysates were subject to Ni-NTA bead Results purification to pull down only the proteins conjugated with His6- SUMO3. We observed more intense and several higher m.w. forms PIAS1 is a novel nuclear interaction partner of ARF of GFP-PIAS1 with the addition of ARF than without, confirming ARF is induced by inflammatory stimuli and enhances proin- that ARF enhances PIAS1 poly-SUMOylation (Fig. 2B). These flammatory responses (5, 6). However, the underlying mechanisms results were recapitulated in U2OS (ARF-null) cells with ectopic are unclear, owing to the lack of known ARF binding partners that Myc-tagged ARF and FLAG-tagged PIAS1 (Fig. 2C) or endoge- could potentially mediate this noncanonical ARF function. Al- nous PIAS1 (Fig. 2D), demonstrating that ARF-mediated PIAS1 though ARF enhances SUMOylation of its binding partners (7), SUMOylation could be reconstituted in ARF-deficient cells. In the physiological role of ARF-mediated SUMOylation is also contrast, knocking down ARF in H1299 cells with two distinct elusive. To identify a possible link between the proinflammatory siRNAs showed around a 50% decrease of endogenous PIAS1 response and the SUMOylation mediated by ARF, we set out to SUMOylation (Fig. 2E). Together, these data indicate that ARF identify the proteins that undergo SUMO modification following enhances SUMOylation of PIAS1. ARF upregulation. SENP1 is a deSUMOylase of PIAS1 H1299 (p53-null) cells were engineered to stably express His6- SUMO3 and then were infected with an adenovirus encoding SUMOylation is a dynamic process that is reversed by the sentrin- human ARF or control vector. p53-null H1299 cells were chosen specific protease (SENP) family of SUMO-deconjugating enzymes to avoid the effect of p53 induction by ARF overexpression. His6- (8). Isoform SENP1 is resident in the nucleoplasm (30) and was SUMO3–tagged proteins were isolated by Ni-NTA beads and therefore tested as a potential deSUMOylase of PIAS1 to further 4 ARF PROMOTES IFN-g RESPONSE BY INHIBITING PIAS1 Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 1. PIAS1 is a novel binding partner of ARF. (A) Interaction of ectopic ARF with ectopic PIAS1. H1299 cells were transfected with GFP-PIAS1 alone or together with FLAG-ARF, and cell extracts were immunoprecipitated (IP) with anti-FLAG Ab and analyzed by Western blotting. (B) Interaction of ectopic ARF with endogenous PIAS1. H1299 cells were transfected with or without FLAG-ARF. Cell extracts were IP with anti-FLAG Ab and analyzed by Western blotting. (C) Interaction of ectopic PIAS1 with endogenous ARF. H1299 cells were transfected with or without FLAG-PIAS1. Cell extracts were IP with anti-FLAG Ab and analyzed by Western blotting. (D) Mapping of the ARF domains required for the interaction with PIAS1. U2OS cells were transfected with GFP-PIAS1 alone or together with FLAG-tagged wild-type or mutant ARF. Cell extracts were IP with anti-FLAG Ab and analyzed by Western blotting. Schematic of ARF deletion mutants and their binding results are summarized below the data. (E) Duolink proximity ligation assay (PLA). H1299 cells were analyzed by using anti-PIAS1 and anti-ARF Abs together (top panels) or individually (bottom panels) as negative controls. Nuclei were stained with DAPI. Green dots indicate positive ARF-PIAS1 interaction signals. Scale bar, 10 mm. Data are representative of three experiments. validate ARF-mediated PIAS1 SUMOylation. First, we coex- Thus, we identified SENP1 as a SUMO-deconjugase of PIAS1 and pressed GFP-tagged PIAS1 and Myc-tagged SUMO3 in U2OS reaffirmed the conclusion that the modification of PIAS1 incurred cells without or with SENP1 and analyzed the whole cell lysates by ARF was SUMOylation. A similar trend was observed in HEK by Western blotting (Fig. 2F). The slower migrating bands of 293T cells (Fig. 2H), indicating that ARF-mediated SUMOylation GFP-PIAS1 became almost undetectable with the addition of of PIAS1 is conserved in multiple cell lines. SENP1, and the amount of unconjugated Myc-SUMO3 detected increased coincidentally, suggesting that SENP1 deSUMOylated Full-length ARF is required for the promotion of SUMOylation PIAS1. To further test this, we performed a SUMOylation assay on the N terminus of PIAS1 and observed a dramatic reduction of all SUMO3-PIAS1 conju- To determine the region of ARF required for promoting SUMO gates when SENP1 was coexpressed with PIAS1 and SUMO3 modification of PIAS1, we overexpressed the ARF C- and N-terminal (Fig. 2G, lane 2 and 3). ARF-enhanced modification of PIAS1 deletion mutants with PIAS1 and SUMO3. Neither of the truncation (lane 4) was also reversed by the addition of SENP1 (lane 5). mutants was able to enhance PIAS1 SUMOylation (Fig. 3A), The Journal of Immunology 5 Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 2. PIAS1 is a novel target of ARF-promoted SUMOylation. (A) Western blotting analysis of PIAS1. H1299 cells stably expressing His6- SUMO3 (H1299-His6-SUMO3) were transfected with GFP-PIAS1 alone or with increasing amounts of ARF, and whole cell lysates were analyzed by Western blotting. Modified GFP-PIAS1 (Modi-GFP-PIAS1) indicates the slower-migrating species of PIAS1 due to posttranslational modifications. (B and C)

SUMOylation assay of exogenous PIAS1. H1299-His6-SUMO3 cells (B) and U2OS cells (C) were transfected as indicated. Cell extracts were precipitated with Ni-NTA beads, and eluates were analyzed by Western blotting. (D) SUMOylation assay of endogenous PIAS1. U2OS cells were transfected with

His6-SUMO3 alone or together with ARF. Cell extracts were precipitated with Ni-NTA beads, and eluates were analyzed by Western blotting. (E) SUMOylation assay of endogenous PIAS1. H1299 cells were transfected with His6-SUMO3 alone or together with two distinct siRNA against ARF. Cell extracts were precipitated with Ni-NTA beads, and eluates were analyzed by Western blotting. The relative band intensities of SUMOylated PIAS1 were calculated by normalizing to nonSUMOylated PIAS1 (SUMO-PIAS1/PIAS1), and the value for lane 2 was set to 1. (F) Western blotting analysis of PIAS1. U2OS cells were cotransfected with GFP-PIAS1 and Myc-SUMO3 with or without FLAG-SENP1. Whole cell lysates were analyzed by Western blotting. (G and H) SUMOylation assay of exogenous PIAS1. U2OS (G) or HEK 293T (H) cells were transfected with combinations of indicated plasmids. Cell extracts were precipitated with Ni-NTA beads, and eluates were analyzed by Western blotting. The value of SUMO-PIAS1/PIAS1 was set to 1 in lane 2 for (G) and (H), respectively. Data are representative of three experiments. 6 ARF PROMOTES IFN-g RESPONSE BY INHIBITING PIAS1 Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 3. Full-length ARF is required for the SUMO modification on the N terminus of PIAS1. (A) Western blotting analysis of PIAS1. U2OS cells were transfected with FLAG-PIAS1 alone, with His6-SUMO3, or both together with FLAG-tagged wild-type or deletion mutants of ARF. Whole cell lysates were analyzed by Western blotting to reveal modified FLAG-PIAS1 (Modi-FLAG-PIAS1). Immunoblots were cut from the same membrane with equal exposure. Schematic of ARF mutants and their effect on PIAS1 are summarized beside the data. Binding: the binding affinity for PIAS1; SUMO: the promotion of PIAS1 SUMOylation. (B) SUMOylation assay. U2OS cells were transfected with wild-type or deletion mutants of FLAG-PIAS1 alone or together with His6-SUMO3. Cell extracts were precipitated with Ni-NTA beads, and eluates were analyzed by Western blotting. Schematic of PIAS1 mutants and corresponding SUMOylation results are summarized below the data. Data are representative of three experiments. PINIT, Pro-Ile-Aso-Ile-Thr motif; RING, E3 ligase activity; SAP, scaffold attachment factor A/B/acinus/PIAS; S/T, Ser/Thr-rich region. The Journal of Immunology 7 revealing that although ARF N terminus was necessary to interact IFN-a–mediated activation of ISRE reporter could not be en- with PIAS1 (Fig. 1D), the full-length of ARF was required for the hanced by ARF overexpression (Supplemental Fig. 4D). TNF-a– promotion of PIAS1 SUMOylation (Fig. 3A, right panel). Next, mediated activation of NF-kB-RE reporter was not significantly we sought to locate the site(s) of PIAS1 SUMO modification. increased by ARF expression (Supplemental Fig. 4E). To deter- PIAS1 deletion mutants were constructed according to previously mine if the potentiation of IFN-g–mediated activation of ISRE and characterized domains (25) and transfected alone or with His6- GAS promoter by ARF was dependent on its ability to promote SUMO3 in U2OS cells. Ni-NTA bead–mediated purification of SUMOylation, ARF deletion mutants, which lost the ability to SUMO-conjugates was performed, and the precipitates were enhance SUMOylation, were tested. Only full-length ARF, but not subject to Western blotting analyses. As shown in Fig. 3B, the N- and C-terminal truncated mutants, was able to enhance IFN-g– three constructs (1–650, 1–415, and 89–415) out of seven that mediated activation of ISRE and GAS promoter (Fig. 4C, 4D). were modified by SUMO3 suggested that both the PINIT region Together, these data suggest that 1) ARF enhances IFN-g–stim- and RING domain are required for the SUMO ligation of PIAS1. ulated gene activation, 2) GAS elements appear to be more sen- These were in line with previous reports that showed the PINIT- sitive to ARF overexpression, and 3) both the N and C terminus of RING regions were minimally required for the E3 ligase activity ARF are required for IFN-g–stimulated gene activation. of PIAS1 (31, 32). PIAS1 may be SUMO-modified by its own E3 ligase activity or that of another E3 SUMO ligase. A single mu- ARF-mediated SUMOylation suppresses the gene-repressive tation of the cysteine residue at position 351 (C351F) in the RING activity of PIAS1 to enhance IFN-g–mediated transcriptional domain renders PIAS1 catalytically inactive (33). To test whether activation another E3 SUMO ligase is required for the SUMO modification Next, we began by validating the inhibitory role of PIAS1 in the Downloaded from of PIAS1, we examined the ability of this catalytic mutant to activation of ISRE reporter by IFN-g. We observed that ISRE undergo SUMOylation in the presence or absence of SUMO3. reporter activation by IFN-g was reduced by expression of PIAS1 Although multiple bands of higher m.w. were readily detectable in a dose-dependent manner (Supplemental Fig. 4F). Then, we for wild-type PIAS1, PIAS1 C351F mutant remained unmodi- tested the ability of wild-type and truncation mutants of ARF to fied (Supplemental Fig. 3A), indicating that PIAS1 undergoes inhibit the repressive activity of PIAS1. Ectopic PIAS1 suppressed

autoSUMOylation. Next, we wished to create a SUMOylation the activation of ISRE promoter by IFN-g to levels similar to that of http://www.jimmunol.org/ site-deficient mutant of PIAS1 that retained catalytic activity. Be- untreated with IFN-g (Fig. 5A, column 1 and 3). Addition of full- cause our data (Fig. 3B) indicated that SUMO modification possibly length ARF, however, was able to rescue the PIAS1-mediated sup- occurs within the SAP, PINIT, or RING domains, we carried out site- pression of reporter activity (column 4). The ARF C- and N-terminal directed mutagenesis of 13 candidate lysine residues reported to be deletion mutants, in contrast, could not rescue (column 5 and 6), SUMO-modified in a resource paper by Hendriks et al. (34) or consistent with their inability to enhance SUMOylation (Fig. 3A). predictedtobemodifiedbySUMOplot (http://www.abgent.com/ These data suggest that the SUMOylation-promoting function of sumoplot) within the consensus motif C-K-x-D/E (in which C is a ARF is essential for the inhibition of PIAS1 (Fig. 5A, right). hydrophobic residue and x is any amino acid). However, mutagenesis To assess whether the IFN-g–promoting activity of ARF is of any of these lysine residues could not abolish SUMOylation of dependent on PIAS1 inhibition by ARF, we studied the effect of by guest on September 27, 2021 PIAS1 (Supplemental Fig. 3B, 3C). Therefore, it is likely that SUMO depleting PIAS1 by siRNA on ARF-mediated enhancement of modification occurs simultaneously on multiple residues or is pro- GAS reporter activation induced by IFN-g. If ARF enhances IFN- miscuous or occurs within nonconsensus motifs. g response via inhibition of PIAS1, PIAS1 depletion should affect IFN-g response in the absence of ARF but not in the presence of ARF enhances IFN-g–mediated transcriptional activation ARF. As expected, PIAS1 depletion induced GAS reporter acti- SUMO modification often regulates the activity of the substrate vation in the absence of ARF (Fig. 5B, column 2 and 3) but not in (21, 28, 29, 35–42). Therefore, we hypothesized that changes to the presence of ARF (Fig. 5B, columns 4 and 5). The effect of the SUMOylation status of PIAS1 would affect its function, at PIAS1 depletion on GAS reporter activation was not equivalent to least within a particular physiological context. It is established that that of ARF overexpression (Fig. 5B, column 3 and 4). This was PIAS1 inhibits transcription factors STAT1 and NF-kB during IFN likely due to the incomplete depletion of PIAS1 by siRNA- and TNF-a signaling, respectively (24, 25). Because it has been mediated knockdown (Supplemental Fig. 4G). Taken together, reported that ARF has a proinflammatory role (6), we hypothe- these data suggest that ARF enhances IFN-g response via inhi- sized that ARF inhibits PIAS1 activity by SUMOylation to do so. bition of PIAS1. To test our hypothesis, we employed ISRE- and GAS-luciferase Because our data suggested that the SUMOylation of PIAS1 by reporters to monitor STAT1 activity in response to IFN-g or IFN-a ARF was important for promoting IFN-g response (Figs. 3A, 5A), stimulation and NF-kB-RE luciferase reporter to monitor NF-kB we wondered whether IFN-g could trigger the enhancement of activity in response to TNF-a stimulation. We first assessed the PIAS1 SUMOylation by ARF. U2OS cells were stimulated with effect of ARF on the activation of these reporters using HEK IFN-g in the presence or absence of ARF for various time periods, 293T cells. IFN-g–mediated ISRE reporter activation was signif- and purification of SUMO3-FLAG-PIAS1 conjugates was per- icantly increased upon addition of ARF in an ARF dose-dependent formed and subjected to Western blotting analysis. Although 2 h manner (Fig. 4A, Supplemental Fig. 4A). GAS reporter, which were required for IFN-g to enhance PIAS1 SUMOylation up to was only activated 1.2 times by IFN-g, was remarkably enhanced 1.4 times in the absence of ARF, IFN-g increased PIAS1 around 5 to 10 times by the expression of ectopic ARF (Fig. 4B, SUMOylation by almost three times within 30 min in the presence Supplemental Fig. 4B). Because IFN-g–induced GAS activation of ARF (Fig. 5C). These data indicated that ARF-mediated was low, we performed a dose titration of IFN-g for GAS acti- SUMOylation of PIAS1 is promoted by IFN-g stimulation. vation (Supplemental Fig. 4C). Our data revealed that GAS acti- vation was minimally improved by increasing the dose, indicating Suppression of IFN-g response by PIAS1 is enhanced by that in our cells, GAS activation is intrinsically limited. None- deSUMOylation theless, ectopic ARF expression was sufficient to activate GAS Because we could not generate a SUMOylation-deficient PIAS1 reporter with or without IFN-g stimulation (Fig. 4B). In contrast, mutant, we employed SENP1 as a tool to indirectly study whether 8 ARF PROMOTES IFN-g RESPONSE BY INHIBITING PIAS1 Downloaded from http://www.jimmunol.org/

FIGURE 4. ARF enhances IFN-g–mediated transcriptional activation. (A and B) Luciferase reporter assay. HEK 293T cells were transfected with ISRE- (A) or GAS-luciferase reporter (B) with or without Myc-ARF. After 24 h, cells were treated or not with 10 ng/ml IFN-g for 16 h, and reporter activity was analyzed. Bars represent mean 6 SDs of three independent experiments that were normalized by renilla luciferase and presented relative to the control (bar 1 for each experiment was set to 1). Two-tailed unpaired Student t test was performed to determine p value. **p , 0.001, ***p , 0.0001. (C and D)

Luciferase reporter assay. HEK 293T cells were transfected with ISRE- (C) or GAS-luciferase reporter (D) alone or with indicated lengths of Myc-tagged by guest on September 27, 2021 ARF. After 24 h, cells were treated or not with 10 ng/ml IFN-g for 16 h, and reporter activity was analyzed. Bars represent mean 6 SDs of three in- dependent experiments that were normalized by renilla luciferase and presented relative to the control (bar 1 for each experiment was set to 1).

PIAS1 deSUMOylation enhances its ability to attenuate IFN-g enhances the immunosuppressive activity of PIAS1, and the response. PIAS1 was transfected alone or together with SENP1, SUMOylation promoted by ARF inhibits the immunosuppressive and ISRE reporter activity was analyzed following IFN-g stimu- activity of PIAS1. lation. We adjusted the amount of PIAS1 plasmid for transfection ARF enhances transcription of IFN-g response genes to achieve around 50% reduction of IFN-g response by PIAS1 (Fig. 6A, column 2 and 3). Coexpression of SENP1 with PIAS1 Next, we overexpressed ARF in HEK 293T cells to assess by led to significant reduction of IFN-g response (Fig. 6A, column 4), quantitative RT-PCR the influence of ARF on IFN-g–inducible suggesting that deSUMOylation of PIAS1 may further enhance genes that are known to be suppressed by PIAS1 (24). We found the ability of PIAS1 to inhibit IFN-g response. Because SENP1 that addition of ARF could enhance the expression of CXCL9, deSUMOylates many nuclear substrates, it may itself influence CXCL10, and GBP1 by almost 2-fold upon IFN-g treatment ISRE reporter activation through the deSUMOylation of other (Fig. 7A). We knocked down PIAS1 by siRNA to test whether substrates. To exclude this possibility, we tested the effect of similar changes in gene expression would occur upon IFN-g SENP1 overexpression alone in two different doses. We did not treatment. As we expected, PIAS1 depletion enhanced the mRNA observe any confounding activity on IFN-g–activated ISRE re- levels of CXCL9, CXCL10, and GBP1 in the presence of IFN-g porter (Fig. 6B), which suggests that apart from its activity on (Fig. 7B). Finally, to study the immunological contribution of exogenous PIAS1, SENP1 overexpression had a neutral effect on ARF in a physiological, nontransformed context, we performed ISRE reporter activation. Because SENP1 was able to attenuate knockdown of ARF using two distinct siRNA sequences in a ARF-mediated SUMOylation of PIAS1 (Fig. 2G, 2H), we next normal human fibroblast cell line WI-38 and repeated RNA tested if SENP1 could inhibit the ARF-mediated suppression of analysis. We found that ARF depletion reduced the ability of cells PIAS1 on ISRE reporter. Addition of ARF efficiently reversed the to upregulate the IFN-g–inducible genes following IFN-g treat- inhibitory effect of PIAS1 on IFN-g response (Fig. 6C, column ment (Fig. 7C). Together, these data suggest that ARF physio- 2–4). The introduction of SENP1, however, reversed the rescue logically enhances transcription of several IFN-g response genes effect of ARF (Fig. 6C, column 4 and 5). These data correlated that are inhibited by PIAS1 upon IFN-g stimulation (Fig. 8). well with the SUMOylation status of PIAS1 as modulated by SENP1 and ARF in our earlier SUMO assays (Fig. 2G, 2H). Al- Discussion though by indirect means, together these data suggest that at least Protein SUMOylation is enhanced by ARF independently of p53 in the context of IFN-g stimulation, deSUMOylation of PIAS1 (7). However, associations between ARF-promoted SUMOylation The Journal of Immunology 9 Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 5. ARF-mediated SUMOylation suppresses the gene-repressive activity of PIAS1 to enhance IFN-g–mediated transcriptional activation. (A) Luciferase reporter assay. HEK 293T cells were transfected with ISRE-luciferase reporter alone, with GFP-PIAS1, or both together with indicated lengths of FLAG-tagged ARF. After 24 h, cells were treated or not with 10 ng/ml IFN-g for 16 h, and reporter activity was analyzed. Corresponding Western blotting of expressed plasmids are shown below the luciferase graph. Bars represent mean 6 SDs of three independent experiments that were normalized by renilla luciferase and presented relative to the control (bar 1 was set to 1). Schematic of ARF mutants and their respective activity on PIAS1 are summarized beside the data. Binding: the binding affinity for PIAS1; SUMO: the promotion of PIAS1 SUMOylation; Inhibition: inhibitory action on the gene-repressor activity of PIAS1. (B) Luciferase reporter assay. HEK 293T cells were transfected with GAS-luciferase reporter alone or with siRNA against PIAS1, Myc- ARF, or both. After 24 h, cells were treated or not with 10 ng/ml IFN-g for 16 h, and reporter activity was analyzed. Bars represent mean 6 SDs of three independent experiments that were normalized by renilla luciferase and presented relative to the control (bar 1 was set to 1). Two-tailed unpaired Student t test was performed to determine p value. (C) SUMOylation assay. U2OS cells were cotransfected with FLAG-PIAS1 and His6-SUMO3 with or without Myc-ARF and were exposed to 10 ng/ml IFN-g for the indicated time. Cell extracts were precipitated with Ni-NTA beads, and eluates were analyzed by Western blotting. The relative band intensities of SUMOylated PIAS1 were calculated by normalizing to nonSUMOylated PIAS1 (SUMO-PIAS1/PIAS1), and the values for lane 1 and 4 were set to 1, respectively. and the inflammatory response have not been reported. We revealed (44, 45). The significant p53-independent contribution of ARF to a SUMO-mediated mechanism by which ARF promotes the IFN-g– the innate immune response has since been highlighted by several activated immune response. We identified PIAS1 as a novel poly- studies (5, 6, 46, 47). Proof of concept for ARF as a sensor and SUMOylation target and interaction partner of ARF. We also responder to viral stress was provided when Garcı´a et al. (5) demonstrated that SENP1 is a PIAS1 SUMO-deconjugase. We demonstrated that ARF protects cells and mice from the infection found that ARF could inhibit the PIAS1-mediated repression of of IFN-sensitive viruses. Subsequent studies by the Hortelano IFN-g–activated transcription by promoting PIAS1 SUMOylation. group (46) demonstrated that Arf-deficient macrophages exposed ARF is usually expressed at scarce levels and upregulated upon to IFN-g or other inflammatory agents required ARF expression to oncogenic stress (43). More recently, ARF was also shown to be develop a proinflammatory phenotype. Arf-deficient macrophages transcriptionally induced in response to IFNs and viral stimulation had compromised ability to transcriptionally induce inflammatory 10 ARF PROMOTES IFN-g RESPONSE BY INHIBITING PIAS1 Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 6. PIAS1-mediated repression of IFN-g response is enhanced by deSUMOylation. (A) Luciferase reporter assay. HEK 293T cells were transfected with ISRE-luciferase reporter alone or with GFP-PIAS1 or with GFP-PIAS1 and FLAG-SENP1. After 24 h, cells were treated or not with 10 ng/ ml IFN-g for 16 h, and reporter activity was analyzed. Bars represent mean 6 SDs of three independent experiments that were normalized by renilla luciferase and presented relative to the control (bar 1 was set to 1). Two-tailed unpaired Student t test was performed to determine p value. Corresponding Western blotting of expressed plasmids are shown below the luciferase graph. ***p , 0.0001. (B) Luciferase reporter assay. HEK 293T cells were transfected with ISRE-luciferase reporter alone or with increasing amounts of FLAG-SENP1. After 24 h, cells were treated or not with 10 ng/ml IFN-g for 16 h, and reporter activity was analyzed. Bars represent mean 6 SDs of three independent experiments that were normalized by renilla luciferase and presented relative to the control (bar 1 was set to 1). Corresponding Western blotting of expressed plasmids are shown below the luciferase graph. (C) Luciferase reporter assay. HEK 293T cells were transfected with ISRE-luciferase reporter alone or with indicated combination of plasmids. After 24 h, cells were treated or not with 10 ng/ml IFN-g for 16 h, and reporter activity was analyzed. Bars represent mean 6 SDs of three independent experiments that were normalized by renilla luciferase and presented relative to the control (bar 1 was set to 1). Two-tailed unpaired Student t test was performed to determine p value. **p , 0.001, ***p , 0.0001. cytokines. Instead, they were polarized to produce the immuno- TNF-a receptor or TLRs (16). In the previous study, the insuffi- suppressive and proangiogenic cytokines commonly associated cient activation of NF-kB in response to inflammatory stimuli was with the maintenance of a protumor microenvironment. Arf-null observed in the Arf-null cells (6). Thus, NF-kB activation remains tumors preferably recruited such macrophages to increase angio- the proposed explanation underlying the proinflammatory role of genesis and promote their growth (46, 47). Importantly, Arf-null ARF. However, the induction of ARF following IFN to promote mice were remarkably less able to recruit neutrophils and mono- the production of inflammatory mediators also compellingly cytes to the site of drug-induced peritonitis (6). These data suggests the involvement of ARF in another major arm of the strongly indicate the requirement of ARF in both immune and immune response, namely the IFN receptor–activated JAK–STAT nonimmune cells to induce the appropriate inflammatory re- pathway. Nevertheless, the association of ARF with the JAK–STAT sponse. NF-kB is a transcription factor and the primary effector of pathway has not been reported. inflammatory signals transduced via stimuli such as virus-induced The IFN-mediated activation of the JAK–STAT pathway is both an protein kinase R (PKR) activation and binding of the antiviral and antitumoral immune response. Owing to the ability of The Journal of Immunology 11 Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021

FIGURE 7. ARF enhances transcription of IFN-g response genes. (A) Quantitative RT-PCR for the expression of the IFN-g response genes. HEK 293T cells were transfected with or without ARF. After 40 h, cells were treated or not with 10 ng/ml IFN-g for 4 h and harvested for RNA extractions. Gene expression was analyzed by quantitative RT-PCR. Bars represent the means 6 SDs of three independent experiments of gene expressions that were normalized by HPRT1 and presented relative to the control (bar 1 for each gene was set to 1). (B) Quantitative RT-PCR for the expression of the IFN-g response genes. HEK 293T cells were transfected with or without siRNA against PIAS1. After 40 h, cells were treated or not with 10 ng/ml IFN-g for 4 h, and gene expression was analyzed by quantitative RT-PCR. Bars represent the mean 6 SDs of three independent experiments of gene expressions that were normalized by HPRT1 and presented relative to the control (bar 1 for each gene was set to 1). (C) Quantitative RT-PCR for the expression of the IFN-g response genes. WI-38 cells were transfected with or without two distinct siRNA against ARF. After 40 h, cells were treated or not with 10 ng/ml IFN-g for 4 h, and gene expression was analyzed by quantitative RT-PCR. Bars represent the mean 6 SDs of three independent experiments of gene expressions that were normalized by TBP and presented relative to the control (bar 1 for each gene was set to 1).

IFN-g to eliminate cells undergoing transformation, IFN-g–secreting therefore, these cancers are resistant to IFN therapy (48, 49). For immune cells promote host antitumor immunity via paracrine and example, SOCS (suppressor of signaling) and PIAS1 are autocrine activation of the JAK–STAT pathway (1). Our finding that often upregulated in cancers. The SOCS family proteins inhibit ARF promotes IFN-g signal transduction is in line with the antitumor JAKs and thus prevent STAT1 phosphorylation and dimerization. functions of both ARF and IFN-g. Therefore, we reason that future PIAS1, in contrast, is a nuclear inhibitor of postactivated STAT1, studies for the role of ARF-mediated PIAS1 SUMOylation in tumor which serves to fine-tune the specificity and strength of the im- survival, as well as a regulatory mechanism for macrophage polari- mune transcriptional response without total abolishment (16, 50). zation and inflammatory potential, will likely prove profitable. Previous studies have shown that upon inflammatory signal input, Various cancers upregulate the negative regulatory mecha- PIAS1 is phosphorylated and discriminates between STAT1 target nisms against IFN-activated JAK–STAT signal transduction, and genes to preferentially inhibit the transcriptional induction of 12 ARF PROMOTES IFN-g RESPONSE BY INHIBITING PIAS1

genes regulated by NF-kB (17, 25). However, in our conditions, we did not see a significant enhancement of NF-kB promoter activation with ARF overexpression in HEK 293T cells. Whether this inability of ARF to activate NF-kB promoter is dependent on cell types remains to be investigated. To determine how SUMO modification of PIAS1 alters its ability to inhibit STAT1, future study will be required to assess the po- tential effect of SUMO modification of PIAS1 on its interaction with STAT1 and its cellular localization. In this study, we were unable to construct a SUMOylation-deficient PIAS1 mutant, al- though we tested various lysine residues of PIAS1 predicted or found to be endogenously SUMO-modified. This indicates that PIAS1 SUMO modification is likely to be promiscuous or occurs on multiple sites. Previous studies showed that mutation of three lysine residues was required to abolish SUMOylation (39, 51). Thus, further work is required for the elucidation of PIAS1 SUMOylation site(s). Nonetheless, we were able to identify and FIGURE 8. Model for ARF-mediation promotion of IFN-g response. (A employ the SUMO-deconjugase SENP1 as a means to eliminate and B) IFN-g stimulation activates STAT-induced transcription from the PIAS1 SUMOylation. Downloaded from promoters of genes containing ISRE or GAS response elements. PIAS1 is ARF interacts with the E2 SUMO ligase Ubc9 (7, 29, 52). activated as a negative regulator of STAT1 toward certain proinflammatory Therefore, it is possible that ARF promotes PIAS1 SUMOylation genes. (A) In cells expressing low levels or insufficient activity of ARF, by recruiting Ubc9. Because the elimination of the Ubc9-binding PIAS1 SUMO (S) modification is low, and thus activity of PIAS1 is high, resulting in decreased transcription of STAT1-response genes in response catalytic site of PIAS1 (33) rendered PIAS1 incapable of SUMO to IFN-g stimulation. (B) In cells expressing high levels of ARF, IFN-g modification, we conjecture that ARF may promote PIAS1 auto- stimulation enhances ARF-mediated PIAS1 SUMOylation, resulting in SUMOylation by increasing Ubc9 availability or promoting http://www.jimmunol.org/ decreased PIAS1 activity, thus allowing the enhanced induction of STAT1- PIAS1–Ubc9 interaction by forming a stabilizing complex. response genes. Poly-SUMOylation with SUMO2 and SUMO3 can lead to protein degradation by a SUMO-targeted (STUbL) proinflammatory cytokines and chemokines (17, 24). Thus, how (28, 36, 41). Indeed, it was recently shown that PIAS1 undergoes these genes can avoid repression by PIAS1 when cells need to STUbL-dependent degradation by RNF4 as a regulatory mecha- promote proinflammatory signaling is an important question. In nism of PIAS1 activity (28). However, no physiological context the current study, we identified a mechanism in which the has yet been determined. We show that ARF promotes SUMOy- SUMOylation-promoting function of ARF is induced by IFN-g to lation of PIAS1 under the condition of IFN-g stimulation as a selectively augment the transcription of proinflammatory genes. novel inhibitory mechanism on PIAS1 activity. In our conditions, by guest on September 27, 2021 Within 30 min of IFN-g stimulation, ARF enhanced the SUMO we did not observe consistent downregulation of PIAS1 protein conjugation of PIAS1 to inhibit the PIAS1-mediated suppression in response to ARF-mediated poly-SUMOylation. PIAS1 ubiq- of the proinflammatory STAT1-regulated genes. Sahin et al. (9) uitination and half-life assays in the presence of ARF and SUMO recently demonstrated that IFNs stimulate the dramatic increase of will be required to confirm if and what conditions during the IFN SUMO expression and conjugation that contributes to the antiviral response determine STUbL-mediated PIAS1 degradation. En- effects of IFN. This is consistent with the antiviral activity of dogenous levels of RNF4 might be a limiting factor in our system. ARF and its ability to promote SUMOylation of protein binding However, protein degradation may not be the only method of partners. PIAS1 functional control. IFN-g signal is rapidly transduced, and Although nonimmune cells were excluded from the above STAT1 is phosphorylated within 30 min of cell exposure to IFN-g studies of ARF-dependent response to IFN-g (6, 46), our study (1). PIAS1 protein degradation is unlikely to occur fast enough to shows that in human HEK 293T cells, ARF enhanced ISRE re- reduce its functional effect. Thus, the increase of PIAS1 SUMO porter activation in response to IFN-g but not to IFN-a. ARF modification, which we observed to occur within 30 min in the addition alone was sufficient to induce GAS reporter, and re- presence of ARF, could be a critical rapid fine-tuning mechanism markable potentiation by ARF was observed upon IFN-g treat- of PIAS1 activity. ment. Depletion of ARF in normal human fibroblast WI-38 cells Without IFN-g treatment, knockdown of ARF in H1299 cells caused reduced response to IFN-g. Our study demonstrates the did not dramatically reduce endogenous levels of PIAS1 importance of functional ARF in nonimmune cells to appropri- SUMOylation. This might be due to lack of stimulus because ately respond to IFN-g, which, physiologically, are released only SUMO modification is a stress response. ARF is also a nucleolar by immune cells. These data are consistent with the general tumor protein that translocates into the nucleoplasm under certain con- suppressor property of ARF, which is required in all cell types. ditions. It would be interesting to monitor changes in ARF lo- Future work is needed to address the interesting observation that calization upon inflammatory stress and subsequent alterations to ARF does not promote the signaling of Type I IFN-a, which are PIAS1 binding affinity and degree of SUMOylation. not produced in an immune cell-specific manner (1). We showed that when ARF is inhibiting PIAS1, depletion of NF-kB was previously proposed as an ARF-mediated activator PIAS1 did not affect the IFN-g response. This suggests that PIAS1 of the immune response (6). Considerable cross-talk between NF- inhibition is a downstream regulatory mechanism of a linear kB and STATs occurs physiologically, and it is therefore likely pathway regulated by IFN-g and ARF. However, we do not ex- that ARF can regulate both. For example, IFN-g primes macro- clude the possibility that ARF has a PIAS1-independent function phages for heightened responses to TLR activation, and ARF was to promote IFN-g response. It is also possible that other members shown to promote TLR expression and vice versa (6). PIAS1 is also of the PIAS proteins compensate for lack of PIAS1 because our known to selectively repress the transcription of proinflammatory siRNA was specifically targeted for PIAS1 only. The Shuai group The Journal of Immunology 13

(50) demonstrated that PIAS1 and PIAS4 cooperate to inhibit 14. Woods, Y. L., D. P. Xirodimas, A. R. Prescott, A. Sparks, D. P. Lane, and M. K. Saville. 2004. p14 Arf promotes small ubiquitin-like modifier conjugation STAT1 activity, and functional redundancy exists between the two of Werners helicase. J. Biol. Chem. 279: 50157–50166. members. We did not test any other PIAS members in our ex- 15. Rytinki, M. M., S. Kaikkonen, P. Pehkonen, T. Ja¨a¨skela¨inen, and J. J. Palvimo. periments because we only identified PIAS1 as an ARF binding 2009. PIAS proteins: pleiotropic interactors associated with SUMO. Cell. Mol. Life Sci. 66: 3029–3041. partner in our MS analysis. However, because we showed that 16. Shuai, K., and B. Liu. 2005. Regulation of gene-activation pathways by PIAS ARF could bind multiple regions of PIAS1, and the PIAS mem- proteins in the immune system. Nat. Rev. Immunol. 5: 593–605. bers exhibit large , ARF might similarly in- 17. Liu, B., Y. Yang, V. Chernishof, R. R. Loo, H. Jang, S. Tahk, R. Yang, S. Mink, D. Shultz, C. J. Bellone, et al. 2007. Proinflammatory stimuli induce IKKalpha- hibit at least some of the other members. This is a possible mediated phosphorylation of PIAS1 to restrict inflammation and immunity. Cell explanation as to why ARF overexpression showed increased ef- 129: 903–914. 18. Heo, K. S., E. Chang, Y. Takei, N. T. Le, C. H. Woo, M. A. Sullivan, C. Morrell, fect compared with knockdown of PIAS1 upon IFN-g stimulation. K. Fujiwara, and J. Abe. 2013. Phosphorylation of protein inhibitor of activated Investigation with PIAS1-knockout cells will clarify this potential. STAT1 (PIAS1) by MAPK-activated protein kinase-2 inhibits endothelial in- Interestingly, expression of PIAS1 is upregulated in several can- flammation via increasing both PIAS1 transrepression and SUMO E3 ligase activity. Arterioscler. Thromb. Vasc. Biol. 33: 321–329. cers (48), whereas ARF is commonly deleted (4). It would be 19. Liu, B., J. Liao, X. Rao, S. A. Kushner, C. D. Chung, D. D. Chang, and K. Shuai. interesting to determine whether mutual exclusivity exists between 1998. Inhibition of Stat1-mediated gene activation by PIAS1. Proc. Natl. Acad. the expression levels of these proteins in some types of cancers. Sci. USA 95: 10626–10631. 20. Liao, J., Y. Fu, and K. Shuai. 2000. Distinct roles of the NH2- and COOH- In summary, the immune system requires the robust induction of terminal domains of the protein inhibitor of activated signal transducer and inflammation to eradicate pathogenic threats before resolution (1). activator of transcription (STAT) 1 (PIAS1) in cytokine-induced PIAS1-Stat1 interaction. Proc. Natl. Acad. Sci. USA 97: 5267–5272. Our work (as summarized in Fig. 8) uncovers a noncanonical 21. Begitt, A., M. Droescher, K. P. Knobeloch, and U. Vinkemeier. 2011. SUMO mechanism for tumor suppressor ARF, in which ARF responds to conjugation of STAT1 protects cells from hyperresponsiveness to IFNg. Blood Downloaded from IFN-g stimulation to promote PIAS1 SUMOylation, allowing 118: 1002–1007. 22. Ungureanu, D., S. Vanhatupa, N. Kotaja, J. Yang, S. Aittomaki, O. A. Ja¨nne, the activation of proinflammatory IFN-g response in a p53- J. J. Palvimo, and O. Silvennoinen. 2003. PIAS proteins promote SUMO-1 independent manner. These indicate that tumor cells with loss of conjugation to STAT1. Blood 102: 3311–3313. ARF may have incapacitated ability to adequately detect and re- 23. Rogers, R. S., C. M. Horvath, and M. J. Matunis. 2003. SUMO modification of STAT1 and its role in PIAS-mediated inhibition of gene activation. J. Biol. spond to external inflammatory stimuli, promoting sustained Chem. 278: 30091–30097. inflammation that leads to tumor progression. 24. Liu, B., S. Mink, K. A. Wong, N. Stein, C. Getman, P. W. Dempsey, H. Wu, and http://www.jimmunol.org/ K. Shuai. 2004. PIAS1 selectively inhibits interferon-inducible genes and is important in innate immunity. Nat. Immunol. 5: 891–898. 25. Liu, B., R. Yang, K. A. Wong, C. Getman, N. Stein, M. A. Teitell, G. Cheng, Acknowledgments H. Wu, and K. Shuai. 2005. Negative regulation of NF-kappaB signaling by We thank Dr. Zhang Yanping for the plasmids containing PIAS1, SUMO3, PIAS1. Mol. Cell. Biol. 25: 1113–1123. and SENP1. We thank Dr. Manoj Krishnan for the luciferase reporters con- 26. Itahana, K., K. P. Bhat, A. Jin, Y. Itahana, D. Hawke, R. Kobayashi, and taining ISRE, GAS, and NF-kB response elements and for helpful discus- Y. Zhang. 2003. Tumor suppressor ARF degrades B23, a nucleolar protein in- volved in ribosome biogenesis and cell proliferation. Mol. Cell 12: 1151–1164. sions. We thank Younghwan Lee for helpful discussion of the data and 27. Wisniewski, J. R., A. Zougman, N. Nagaraj, and M. Mann. 2009. Universal manuscript. sample preparation method for proteome analysis. Nat. Methods 6: 359–362. 28. Kumar, R., R. Gonza´lez-Prieto, Z. Xiao, M. Verlaan-de Vries, and A. C.

O. Vertegaal. 2017. The STUbL RNF4 regulates protein group SUMOylation by by guest on September 27, 2021 Disclosures targeting the SUMO conjugation machinery. Nat. Commun. 8: 1809. The authors have no financial conflicts of interest. 29. Wang, S., S. Wang, L. Yang, H. Guo, X. Kong, L. Yuan, G. Xing, F. He, and L. Zhang. 2015. ARF-mediated SUMOylation of Apak antagonizes ubiq- uitylation and promotes its nucleolar accumulation to inhibit 47S pre-rRNA synthesis. J. Mol. Cell Biol. 7: 154–167. References 30. Bailey, D., and P. O’Hare. 2004. Characterization of the localization and pro- 1. Dunn, G. P., C. M. Koebel, and R. D. Schreiber. 2006. Interferons, immunity and teolytic activity of the SUMO-specific protease, SENP1. J. Biol. Chem. 279: cancer immunoediting. Nat. Rev. Immunol. 6: 836–848. 692–703. 2. Boehm, U., T. Klamp, M. Groot, and J. C. Howard. 1997. Cellular responses to 31. Takahashi, Y., and Y. Kikuchi. 2005. Yeast PIAS-type Ull1/Siz1 is composed of interferon-gamma. Annu. Rev. Immunol. 15: 749–795. SUMO ligase and regulatory domains. J. Biol. Chem. 280: 35822–35828. 3. Sharpless, N. E. 2005. INK4a/ARF: a multifunctional tumor suppressor . 32. Liang, M., F. Melchior, X. H. Feng, and X. Lin. 2004. Regulation of Smad4 Mutat. Res. 576: 22–38. sumoylation and transforming growth factor-beta signaling by protein inhibitor 4. Nobori, T., K. Miura, D. J. Wu, A. Lois, K. Takabayashi, and D. A. Carson. of activated STAT1. J. Biol. Chem. 279: 22857–22865. 1994. Deletions of the cyclin-dependent kinase-4 inhibitor gene in multiple 33. Brown, J. R., K. L. Conn, P. Wasson, M. Charman, L. Tong, K. Grant, human cancers. Nature 368: 753–756. S. McFarlane, and C. Boutell. 2016. SUMO ligase protein inhibitor of activated 5. Garcı´a, M. A., M. Collado, C. Mun˜oz-Fontela, A. Matheu, L. Marcos-Villar, STAT1 (PIAS1) is a constituent promyelocytic leukemia nuclear body protein J. Arroyo, M. Esteban, M. Serrano, and C. Rivas. 2006. Antiviral action of the that contributes to the intrinsic antiviral immune response to Herpes Simplex tumor suppressor ARF. EMBO J. 25: 4284–4292. virus 1. J. Virol. 90: 5939–5952. 6. Trave´s, P. G., R. Lo´pez-Fontal, A. Luque, and S. Hortelano. 2011. The tumor 34. Hendriks, I. A., R. C. D’Souza, B. Yang, M. Verlaan-de Vries, M. Mann, and suppressor ARF regulates innate immune responses in mice. J. Immunol. 187: A. C. Vertegaal. 2014. Uncovering global SUMOylation signaling networks in a 6527–6538. site-specific manner. Nat. Struct. Mol. Biol. 21: 927–936. 7. Tago, K., S. Chiocca, and C. J. Sherr. 2005. Sumoylation induced by the Arf 35. Cox, B., J. Briscoe, and F. Ulloa. 2010. SUMOylation by Pias1 regulates the tumor suppressor: a p53-independent function. Proc. Natl. Acad. Sci. USA 102: activity of the Hedgehog dependent Gli transcription factors. PLoS One 5: 7689–7694. e11996. 8. Hay, R. T. 2005. SUMO: a history of modification. Mol. Cell 18: 1–12. 36. Her, J., Y. Y. Jeong, and I. K. Chung. 2015. PIAS1-mediated sumoylation pro- 9. Sahin, U., O. Ferhi, X. Carnec, A. Zamborlini, L. Peres, F. Jollivet, A. Vitaliano- motes STUbL-dependent proteasomal degradation of the human telomeric pro- Prunier, H. de The´, and V. Lallemand-Breitenbach. 2014. Interferon controls tein TRF2. FEBS Lett. 589: 3277–3286. SUMO availability via the Lin28 and let-7 axis to impede virus replication. Nat. 37. Mascle, X. H., M. Lussier-Price, L. Cappadocia, P. Estephan, L. Raiola, Commun. 5: 4187. J. G. Omichinski, and M. Aubry. 2013. Identification of a non-covalent ternary 10. Ivanschitz, L., Y. Takahashi, F. Jollivet, O. Ayrault, M. Le Bras, and H. de The´. complex formed by PIAS1, SUMO1, and UBC9 proteins involved in tran- 2015. PML IV/ARF interaction enhances p53 SUMO-1 conjugation, activation, scriptional regulation. J. Biol. Chem. 288: 36312–36327. and senescence. Proc. Natl. Acad. Sci. USA 112: 14278–14283. 38. Zhang, D., Y. Liang, Q. Xie, G. Gao, J. Wei, H. Huang, J. Li, J. Gao, and 11. Chen, L., and J. Chen. 2003. MDM2-ARF complex regulates p53 sumoylation. C. Huang. 2015. A novel post-translational modification of nucleolin, Oncogene 22: 5348–5357. SUMOylation at Lys-294, mediates arsenite-induced cell death by regulating 12. Xirodimas, D. P., J. Chisholm, J. M. Desterro, D. P. Lane, and R. T. Hay. 2002. gadd45a mRNA stability. J. Biol. Chem. 290: 4784–4800. P14ARF promotes accumulation of SUMO-1 conjugated (H)Mdm2. FEBS Lett. 39. Rabellino, A., B. Carter, G. Konstantinidou, S. Y. Wu, A. Rimessi, L. A. Byers, 528: 207–211. J. V. Heymach, L. Girard, C. M. Chiang, J. Teruya-Feldstein, and P. P. Scaglioni. 13. Neo, S. H., Y. Itahana, J. Alagu, M. Kitagawa, A. K. Guo, S. H. Lee, K. Tang, 2012. The SUMO E3-ligase PIAS1 regulates the tumor suppressor PML and its and K. Itahana. 2015. TRIM28 is an E3 ligase for ARF-mediated NPM1/B23 oncogenic counterpart PML-RARA. Cancer Res. 72: 2275–2284. SUMOylation that represses centrosome amplification. Mol. Cell. Biol. 35: 40. Percherancier, Y., D. Germain-Desprez, F. Galisson, X. H. Mascle, L. Dianoux, 2851–2863. P. Estephan, M. K. Chelbi-Alix, and M. Aubry. 2009. Role of SUMO in 14 ARF PROMOTES IFN-g RESPONSE BY INHIBITING PIAS1

RNF4-mediated promyelocytic leukemia protein (PML) degradation: sumoyla- 46. Herranz, S., P. G. Trave´s, A. Luque, and S. Hortelano. 2012. Role of the tumor tion of PML and phospho-switch control of its SUMO binding domain dissected suppressor ARF in macrophage polarization: Enhancement of the M2 phenotype in living cells. J. Biol. Chem. 284: 16595–16608. in ARF-deficient mice. OncoImmunology 1: 1227–1238. 41. Tatham, M. H., M. C. Geoffroy, L. Shen, A. Plechanovova, N. Hattersley, 47. Jime´nez-Garcı´a, L., S. Herranz, M. A. Higueras, A. Luque, and S. Hortelano. E. G. Jaffray, J. J. Palvimo, and R. T. Hay. 2008. RNF4 is a poly-SUMO-specific 2016. Tumor suppressor ARF regulates tissue microenvironment and tumor E3 ubiquitin ligase required for arsenic-induced PML degradation. Nat. Cell growth through modulation of macrophage polarization. Oncotarget 7: 66835– Biol. 10: 538–546. 66850. 42. Ungureanu, D., S. Vanhatupa, J. Gro¨nholm, J. J. Palvimo, and O. Silvennoinen. 48. Rabellino, A., C. Andreani, and P. P. Scaglioni. 2017. The role of PIAS SUMO 2005. SUMO-1 conjugation selectively modulates STAT1-mediated gene re- E3-ligases in cancer. Cancer Res. 77: 1542–1547. sponses. Blood 106: 224–226. 49. Inagaki-Ohara, K., T. Kondo, M. Ito, and A. Yoshimura. 2013. SOCS, inflam- 43. Ozenne, P., B. Eymin, E. Brambilla, and S. Gazzeri. 2010. The ARF tumor mation, and cancer. JAK-STAT 2: e24053. suppressor: structure, functions and status in cancer. Int. J. Cancer 127: 2239– 50. Tahk, S., B. Liu, V. Chernishof, K. A. Wong, H. Wu, and K. Shuai. 2007. Control 2247. of specificity and magnitude of NF-kappa B and STAT1-mediated gene activa- 44. Sandoval, R., J. Xue, M. Pilkinton, D. Salvi, H. Kiyokawa, and tion through PIASy and PIAS1 cooperation. Proc. Natl. Acad. Sci. USA 104: O. R. Colamonici. 2004. Different requirements for the cytostatic and apoptotic 11643–11648. effects of type I interferons. Induction of apoptosis requires ARF but not p53 in 51. Impens, F., L. Radoshevich, P. Cossart, and D. Ribet. 2014. Mapping of SUMO osteosarcoma cell lines. J. Biol. Chem. 279: 32275–32280. sites and analysis of SUMOylation changes induced by external stimuli. Proc. 45. Ries, S. J., C. H. Brandts, A. S. Chung, C. H. Biederer, B. C. Hann, E. M. Lipner, Natl. Acad. Sci. USA 111: 12432–12437. F. McCormick, and W. M. Korn. 2000. Loss of p14ARF in tumor cells fa- 52. Rizos, H., S. Woodruff, and R. F. Kefford. 2005. p14ARF interacts with the cilitates replication of the adenovirus mutant dl1520 (ONYX-015). Nat. Med. SUMO-conjugating enzyme Ubc9 and promotes the sumoylation of its binding 6: 1128–1133. partners. Cell Cycle 4: 597–603. Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021 Supplemental Figure 1. Identification of proteins which undergo enhanced SUMOylation in the presence of ARF.

H1299 cells stably expressing His6-SUMO3 were infected with empty or ARF-expressing adenovirus and precipitated with Ni-NTA beads. Parental H1299 cells left uninfected and empty vector express- ing cells served as controls for non-specifically precipitated polypeptides. Bound proteins were eluted, separated by SDS-PAGE and visualized by silver staining. The eluates were trypsin-digested and analyzed by mass spectrometry. Supplemental Figure 2. ARF interacts with multiple domains of PIAS1. U2OS cells were transfected with FLAG-PIAS1 truncation mutants with or without Myc-ARF. Cell lysates were immunoprecipitated with anti-ARF antibody and analyzed by Western blotting with indicated antibodies. Schematic of PIAS1 deletion mutants and the corresponding ARF-binding results are summarized below the data. Supplementary Figure 3

A WT Cat Mut GFP-PIAS1 C351F

Myc-SUMO3 Modi- GFP-PIAS1 GFP-PIAS1 GFP-PIAS1 Myc-SUMO3 ACTIN

U2OS

B FLAG-PIAS1 lysine mutants WT K244R K315R K140R K152R K368R K137R

Myc-SUMO3

Modi- FLAG-PIAS1 FLAG-PIAS1

FLAG-PIAS1

Myc-SUMO3 Myc-SUMO3

ACTIN

U2OS

C FLAG-PIAS1 lysine mutants WT K408R, K46R K117R K58R K315R K8R K456R K238R K244R

Myc-SUMO3

Modi- FLAG-PIAS1 FLAG-PIAS1 FLAG-PIAS1

Myc-SUMO3 Myc-SUMO3

ACTIN

U2OS

Supplemental Figure 3. Analyses of PIAS1 SUMO-modification. (A) Western blotting of catalytic mutant PIAS1. U2OS cells were transfected with wild type or catalytic mutant (C351F Cat Mut) GFP-tagged PIAS1 alone or together with Myc-SUMO3. SUMO-modified PIAS1 (Modi-GFP-PIAS1) was detected by anti-GFP antibody following Western blotting of whole cell lysates. (B and C) Site-directed mutagenesis of potentially SUMO-modified lysine residues of PIAS1. U2OS cells were transfected with various lysine point mutants of FLAG-tagged PIAS1 alone or together with Myc-SUMO3. SUMO-modified PIAS1 (Modi-FLAG-PIAS1) was detected by anti-FLAG antibody following Western blotting of whole cell lysates. (B) Immunoblots were cut from different membranes of the same experiment and detected with equal exposure. (C) Immunoblots were cut from different membranes of the same experiment and detected with equal exposure. Note K244 and K315 are shown in both (B) and (C). C GAS 2.5 ty vi i 2.0 e act s

a 1.5 fer i

c 1.0 lu

0.5

Relative 0.0 IFNγ 0 10 50 100 (ng/mL)

G

Supplemental Figure 4. Activity of ARF and PIAS1 on IFNγ-promoted transcriptional activation. (A - F) Luciferase reporter assay. HEK 293T cells were transfected as indicated. After 24 hours, cells were treated or not with 10 ng/mL IFNγ (A, B and F) or indicated doses (C), 1 ng/mL IFNα (D), or 20 ng/mL TNFα (E) for 16 hours and reporter activity was analyzed. Bars represent means ± SDs of three independent experi- ments that were normalized by renilla luciferase and presented relative to the control (bar 1 for each experi- ment was set to 1.0). Two-tailed unpaired student t test was performed to determine p value. *p < 0.01, **p < 0.001, ***p < 0.0001. NS: non-significant. Corresponding protein expression was detected by Western blotting and shown below the bar graph. (G) Efficiency of siRNA-mediated depletion of endogenous PIAS1. HEK 293T cells were transfected with siRNA against PIAS1 and whole cell lysate was analyzed by Western blotting.