Negative Regulation of Host Innate Antiviral Signaling

Transcription

1 University of Miami Scholarly Repository Open Access Dissertations Electronic Theses and Dissertations Negative Regulation of Host Innate Antiviral Signaling Qinjie Zhou University of Miami, Follow this and additional works at: Recommended Citation Zhou, Qinjie, "Negative Regulation of Host Innate Antiviral Signaling" (2015). Open Access Dissertations This Embargoed is brought to you for free and open access by the Electronic Theses and Dissertations at Scholarly Repository. It has been accepted for inclusion in Open Access Dissertations by an authorized administrator of Scholarly Repository. For more information, please contact

2 UNIVERSITY OF MIAMI NEGATIVE REGULATION OF HOST INNATE ANTIVIRAL SIGNALING By Qinjie Zhou A DISSERTATION Submitted to the Faculty of the University of Miami in partial fulfillment of the requirements for the degree of Doctor of Philosophy Coral Gables, Florida May 2015

3 2015 Qinjie Zhou All Rights Reserved

4 UNIVERSITY OF MIAMI A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy NEGATIVE REGULATION OF HOST INNATE ANTIVIRAL SIGNALING Qinjie Zhou Approved: Edward W. Harhaj, Ph.D. Associate Professor of Oncology Johns Hopkins School of Medicine Zhibin Chen, Ph.D. Associate Professor of Microbiology and Immunology Wasif N. Khan, Ph.D. Professor of Microbiology and Immunology Enrique A. Mesri, Ph.D. Associate Professor of Microbiology and Immunology Joyce M. Slingerland, Ph.D. Professor of Medicine M. Brian Blake, Ph.D. Dean of the Graduate School Fanxiu Zhu, Ph.D. Associate Professor of Biological Science Florida State University

5 ZHOU, QINJIE (Ph.D., Microbiology and Immunology) Negative Regulation of Host (May 2015) Innate Antiviral Signaling Abstract of a dissertation at the University of Miami. Dissertation supervised by Professor Edward W. Harhaj No. of pages in text (111) The human T cell leukemia virus type I (HTLV-I) inhibits host antiviral signaling pathways although the underlying mechanisms are unclear. Here we found that the HTLV-I Tax oncoprotein induced the expression of SOCS1, an inhibitor of interferon signaling. Tax required NF-κB, but not CREB, to induce the expression of SOCS1 in T cells. Furthermore, Tax interacted with SOCS1 in both transfected cells and in HTLV-I transformed cell lines. Although SOCS1 is normally a short-lived protein, in the presence of Tax, the stability of SOCS1 was greatly increased. Accordingly, Tax enhanced the replication of a heterologous virus, vesicular stomatitis virus (VSV), in a SOCS1- dependent manner. Surprisingly, Tax required SOCS1 to inhibit RIG-I-dependent antiviral signaling, but not the interferon-induced JAK/STAT pathway. Inhibition of SOCS1 by RNA-mediated interference in the HTLV-I transformed cell line MT-2 resulted in increased IFN-β expression accompanied by reduced HTLV-I replication and p19 Gag levels. Taken together, our results reveal that Tax inhibits RIG-I-dependent antiviral signaling, in part, by hijacking an interferon regulatory protein. The transcription factor interferon regulatory factor 7 (IRF7) is a key regulator of type I interferon and plays essential roles in restricting virus infection and spread. IRF7

6 activation is tightly regulated to prevent excessive inflammation and autoimmunity; however, how IRF7 is suppressed by negative regulators remains poorly understood. Here, we have identified aryl hydrocarbon receptor interacting protein (AIP) as a new binding partner of IRF7. AIP inducibly interacts with IRF7 upon virus infection and potently inhibits IRF7-induced type I IFN (IFNα/β) production. Overexpression of AIP blocks virus-induced activation of IFN, whereas knockdown of AIP by sirna potentiates viral-activated IFN production. Consistently, AIP-deficient murine embryonic fibroblasts are highly resistant to virus infection due to increased production of IFNα/β. AIP inhibits IRF7 function by antagonizing the nuclear localization of IRF7. Together, our study identifies AIP as a novel inhibitor of IRF7 and negative regulator of innate RLR signaling.

7 DEDICATION To my parents who support me unconditionally iii

8 ACKNOWLEDGEMENT I am deeply appreciative and grateful for having the opportunity to pursue my graduate training at the University of Miami Miller Medical School in the Microbiology and Immunology program, as well as Johns Hopkins University School of Medicine in the Viral Oncology program. During the last five-plus years, I was surrounded by outstanding scientists, staff members, energetic students, and superior post-doctoral fellows. First and foremost, I want to thank with deepest appreciation my advisor and mentor, Dr. Edward W. Harhaj, for leading me into the world of academic research with tremendous patience, encouragement, and mentoring. I m grateful and feel fortunate to have received his continuous guidance and the opportunity to work on so many important projects with fruitful outcomes. I also express my sincere gratitude to Dr. Alfonso Lavorgna, who is not just my closest colleague and mentor, but also my best friend and dear brother in the lab. His patience, generous help, deep understanding, and unconditional support in both of my professional and personal life have been invaluable. A special thanks to all my other lab members: Dr. Young Choi, Melissa Bowman; prior lab members: Dr. Linlin Gao, Dr. Soratree Charoenthongtrakul, Dr. Noula Shembade, Dr. Kislay Parvatiyar; and Noula s wife, Raj, for all the immediate help and cover they provided when I needed it. Moreover, I thank all my current and prior committee members who always provided constructive suggestions and guidance along the way: Dr. Zhibin Chen, Dr. Wasif Khan, Dr. Joyce Slingerland, Dr. Enrique Mesri, and Dr. Shunbin Ning. I also extend a special thanks to my defense outside-examiner, Dr. Fanxiu Zhu, from Florida State University. iv

9 Last but not least, I thank my dear families, friends in Miami as well as at Hopkins. Without their support, I would not have made it. Looking back, it was a journey worth travelling. Thanks to all of you. v

10 TABLE OF CONTENTS Page LIST OF FIGURES... viii Chapter 1 Introduction Type I IFN signaling pathway and regulation Overview PRRs sensing and recognition RLR signaling cascade Type I IFN production IRF3 and IRF7: key IFN regulatory factors Regulation of RLR signaling IFNAR signaling pathway and regulation Overview IFNAR binding and recognition JAK-STAT signaling cascade Interferon stimulated genes (ISGs) Suppressor of cytokine signaling (SOCS) proteins HTLV-1 and RLR signaling HTLV-1 infection and associated diseases HTLV-1 transmission and viral replication Viral evasion of antiviral pathways HTLV-1, Tax and antiviral signaling pathways Aryl hydrocarbon receptor (AhR) and AhR interacting protein (AIP) AhR overview AhR and virus infection AIP overview AIP and the regulation of the host antiviral response Chapter 2 HTLV-1 Tax and innate RLR signaling Background Results Tax induces the expression of SOCS Tax requires NF-κB, but not CREB, to up-regulate SOCS vi

11 2.2.3 Tax interacts with and stabilizes SOCS Tax requires SOCS1 to inhibit RLR signaling SOCS1 enhances replication of HTLV-1 in MT-2 cells Chapter 3 AIP and RLR signaling Background Results AIP interacts with IRF AIP inhibits RLR signaling AIP-deficient cells overproduce type-i IFN and are resistant to virus infection AIP specifically targets IRF7 for inhibition by interfering with its nuclear localization Discussion Chapter 4 Concluding remarks and future perspectives SOCS1 and Tax AIP and IRF AhR in RLR signaling Environmental pollutants and virus infection Chapter 5 Materials and Methods References vii

12 LIST OF FIGURES Fig 1. 1 A schematic view of antiviral signaling pathways triggered by virus infection... 1 Fig 1. 2 A schematic view of TLR, RLR and NLR signaling pathways triggered by RNA virus infection Fig 1. 3 TLR signaling in conventional DCs and plasmacytoid DCs Fig 1. 4 RLR signaling cascade triggered by RNA virus infection Fig 1. 5 A schematic view of RLR and IFN signaling pathways focusing on IRF3 and IRF Fig 1. 6 Cytosolic nucleic acid pattern recognition and activation of ISGs Fig 1. 7 A schematic view of AIP protein structure Fig 2. 1 Tax induces the expression of SOCS Fig 2. 2 Tax requires NF-κB to induce SOCS1 expression Fig 2. 3 Tax interacts with SOCS1 and promotes SOCS1 stability Fig 2. 4 Tax requires socs1 to inhibit RLR antiviral signaling Fig 2. 5 SOCS1 is a positive regulator of HTLV-1 replication Fig 3. 1 AIP interacts with IRF7 in a virus-inducible manner Fig 3. 2 AIP inhibits RLR antiviral signaling and promotes virus infection Fig 3. 3 Aip -/- MEFs produce elevated type I IFNs and are resistant to virus infection Fig 3. 4 AIP does not inhibit IRF3 and IRF7 phosphorylation Fig 3. 5 AIP inhibits IRF7 nuclear localization and transactivation Fig 3. 6 A model for AIP inhibition of IRF Fig 4. 1 TBK1-induced AIP phosphorylation sites identified by liquid chromatography and tandem mass spectrometry Fig 4. 2 TBK1 phosphorylation consensus sites within AIP predicted by Nepos 2.0 server (Technical University of Denmark) Fig 4. 3 The AIP triple phospho-mimetic strongly interacts with IRF7 and promotes virus replication Fig 4. 4 The AIP triple phospho-mutant and phospho-mimetic mutants both inhibit the induction of type I IFN, but not IL Fig 4. 5 Endogenous AhR protein, but not mrna, is increased by virus infection and enhances virus replication Fig 4. 6 A model depicting the potential involvement of AhR in inhibiting RLR antiviral signaling viii

13 Chapter 1 Introduction 1.1 Type I IFN signaling pathway and regulation Overview Innate immunity plays an essential role in host defense against invading pathogens, such as virus infections. Serving as the first line of defense, it plays a vital role in initiating and activating subsequent immune effector functions, adaptive immunity and other defense mechanisms. Fig 1. 1 A schematic view of antiviral signaling pathways triggered by virus infection. 1

14 2 The host response to virus infection and triggering of antiviral signaling cascades involve three stages (Fig 1.1): I. sensing (e.g. recognition of pathogenassociated molecular patterns (PAMPs) by host pattern recognition receptors (PRRs)), II. signal transduction and signaling cascade involving multiple factors (e.g. IRF3/7 phosphorylation and nuclear translocation), and III. activation of type I interferon (IFN) and pro-inflammatory cytokine genes. Subsequently, type I IFNs engage the interferon α/β receptor (IFNAR), which activates the JAK/STAT pathway to induce the expression of hundreds of interferon stimulated genes (ISGs) that serve to restrict virus infection PRRs sensing and recognition There are four major classes of PRRs including the Toll-like receptors (TLRs), the retinoic acid-inducible gene I-like receptors (RLRs) [1], the nucleotide oligomerization domain-like receptors (NLRs) [2] and cytosolic DNA sensors [3, 4] (Fig 1.2, [5]). PRRs are strategically localized to a variety of locations, either on the cell surface, in the cytoplasm or in endosomes, to sense a wide array of pathogens [6].

15 3 Fig 1. 2 A schematic view of TLR, RLR and NLR signaling pathways triggered by RNA virus infection TLRs TLRs are type I transmembrane proteins that share a common architecture consisting of extracellular leucine-rich repeats (LRRs) and a cytoplasmic Toll- Interleukin-1 Receptor (TIR) domain [7]. TLRs can traffic between the plasma membrane and endosomal vesicles. Of the 9 types of TLRs shared by human and mice, TLR1/2/4/5/6 are localized on the cell surface and can sense extracellular PAMPs such as bacterial lipids and proteins; TLR3/7/8/9 are localized in endosome membranes and are the major PRRs that recognize distinct types of virally-derived nucleic acids. TLR4 is established to sense bacterial lipopolysaccharide (LPS) [8]. TLR3 recognizes double stranded RNA (dsrna) in conventional dendritic cells (cdcs) and possibly epithelial cells [9], while TLR7 [10, 11] and TLR9 [12] recognize single stranded RNA (ssrna)

16 4 and unmethylated CpG dinucleotides respectively, and are highly expressed in plasmacytoid DCs (pdcs) [13], a cell type that is well known to produce high levels of type I IFNs in response to virus infection (Fig 1.3). Fig 1. 3 TLR signaling in conventional DCs and plasmacytoid DCs. TLRs signal as homo/heterodimers by recruiting adaptor proteins, such as Tollinterleukin 1 receptor (TIR) domain-containing adaptor protein (TIRAP) [14], Myeloid differentiation primary response gene 88 (MyD88) [15, 16], TIR-domain-containing adaptor inducing interferon-β (TRIF) [17] and TRIF-related adaptor molecule (TRAM) [18]. MyD88 serves as a critical adaptor for all TLRs, except for TLR3. TRIF functions as an adaptor molecule in the TLR3 and TLR4 pathways. TIRAP facilitates recruitment of MyD88 to TLR2 and TLR4. TRAM is involved in MyD88-independent TRIF-TBK1- IRF3 activation downstream of TLR4 ligation. Upon TLR engagement by specific PAMPs, signal cascades are initiated that activate key transcription factors including nuclear factor kappa B (NF-κB), and interferon regulatory factors 1/3/5/7 (IRF1/3/5/7), which further drive expression of IFNs, cytokines, and chemokines [19].

17 RLRs The RLR family is composed of three members: retinoic acid-inducible gene (RIG- I) [20], melanoma differentiation-associated gene 5 (MDA5) [21] and laboratory of genetics and physiology-2 (LGP2) [22]. RIG-I and MDA5 are both comprised of tandem N-terminal caspase activation and recruiting domains (CARDs) and DExD/H box RNA helicase domains, the latter containing a C-terminal repressor domain (RD) with AIPase activity. In comparison, LGP2 contains only the RNA helicase domain, suggesting that it may function as a negative regulator of RLR signaling pathways, which was confirmed by in vitro assays [23-26]. Upon RNA virus infection, RIG-I and MDA5 can both detect viral RNA products produced during virus infection and replication. Specifically, RIG-I recognizes 5 - triphosphorylated, uncapped ssrna [27, 28] and short dsrna. Conversely, MDA5 recognizes polyinosinic acid (poly (I:C)), a synthetic dsrna mimetic, as well as long dsrna, which is associated with the replication of certain RNA viruses [29, 30]. Genetic studies using RIG-I and MDA5-deficient mice have revealed specificity for each of these receptors in viral sensing. For example, RIG-I has been reported to detect viruses such as Sendai virus, vesicular stomatitis virus, influenza A/B, Ebola, Lassa, Rift Valley fever virus, Hepatitis C virus, murine hepatitis virus, murine norovirus-1, and Epstein-Barr virus EBER (DNA virus). MDA5 can recognize Mengo virus, Theiler s virus, encephalomyocarditis virus, and vaccinia virus (DNA virus). Both receptors can detect flaviviruses, such as dengue virus, West Nile virus, and reovirus, etc [31]. DDX3, another member of the DExD/H box RNA helicase family, has recently been reported to regulate the RLR signaling pathway, by binding to cytosolic poly (I:C) and

18 6 virus RNA, and associating with downstream signaling factors of RIG-I and MDA5 to induce type I IFNs [32, 33] NLRs While TLRs and RLRs are important for the production of type I IFNs and cytokines, most NLRs can regulate interleukin-1β (IL-1β) maturation through activation of caspase-1. NOD1 and NOD2 are closely related NLRs that recognize peptidoglycan motifs from the bacterial cell wall and activate NF-κB and proinflammatory cytokines [34]. NOD1 can sense the dipeptide γ-d-glu-mdap, present in all Gram-negative and certain Gram-positive bacteria. NOD2 can sense muramyl dipeptide (MDP), found in both Gram-positive and Gram-negative bacteria. Recent studies have shown that cytoplasmic NOD-LRR receptors, Nacht domain-, leucine-rich repeat- and PYD-containing protein 3 (NALP3) [35] and ICE-protease activating factor (IPAF), together with the adapter protein, apoptosis-associated specklike protein containing a CARD (ASC), form a large protein complex with caspase-1 termed the inflammasome, which processes pro-il-1β cleavage into mature IL-1β. Although it is unclear whether NALP3 directly recognizes dsrna, it was reported that dsrna and poly (I:C) activated the inflammasome via a NALP3-dependent pathway [36] Cytosolic DNA sensors Besides TLR9, which is well established to recognize viral DNA with CpG motifs in the endosome, several other proteins have been identified in cytosolic DNA sensing. For example, DNA-dependent activator of IFN-regulatory factors (DAI) [37], RNA polymerase III [38], Leucine-rich repeat flightless-interacting protein (LRRFIP1) [39],

19 7 IFI16 [40] and DHX9/36 [41] have been reported as putative DNA sensors. A major DNA sensor known as cgas (cgamp synthase) has been identified that can bind to DNA in the cytoplasm and synthesize cyclic guanosine monophosphate-adenosine monophosphate (cyclic GMP-AMP or cgamp), which is sensed by the ER resident protein STING [42, 43]. STING then activates TBK1 and IRF3 to induce type I IFN [44]. The cgas-sting pathway is essential for type I IFN production in response to infection with HSV-1, adenovirus and other pathogens containing DNA genomes. This DNA sensing pathway has also recently been shown to be critical for innate immune sensing of immunogenic tumors [45] RLR signaling cascade RIG-I contains tandem N-terminal CARD domains, a central RNA helicase and a C- terminal regulatory domain which directly binds to 5 -triphosphate RNA [46, 47]. Under homeostatic conditions, RIG-I is autoinhibited by intramolecular interactions of the CARD and helicase domains. Upon viral RNA binding, it triggers a conformational change that exposes the RIG-I CARDs and promotes lysine 63 (K63)-linked polyubiquitination of RIG-I by the E3 ligase tripartite motif-containing 25 (TRIM25). RIG-I K63-linked polyubiquitination facilitates multimerization and CARD-CARD interactions with the outer mitochondrial membrane adaptor MAVS (also known as IPS- 1, CARDIF or VISA) [48-51]. MAVS forms prion-like aggregates and nucleates a signaling complex containing tumor necrosis factor (TNF) receptor-associated factor (TRAF) E3 ligases (TRAF2 [52], TRAF3 [53], TRAF5, TRAF6 [54]), the adaptor molecule NF-κB essential modulator (NEMO; also known as IKKγ) [55] and the noncanonical IκB kinases TBK1 and IKKε [56]. TBK1 and IKKε phosphorylate the

20 8 interferon regulatory factors (IRFs) IRF3 and IRF7 that trigger their dimerization, nuclear translocation and activation of type I IFN gene expression [57]. In addition to IRF activation, the NF-κB transcription factor is also activated downstream of RLRs and contributes to the induction of type I IFN. Within the MAVS signalosome, TRAF2/6 signal to IKK and NF-κB whereas TRAF3 activates TBK1 and IRF transcription factors. The canonical IKK complex, comprising IKKα, IKKβ and NEMO, phosphorylates IκBα, which leads to its degradation and release of NF-κB dimers into the nuclear compartment (Fig 1.4). Fig 1. 4 RLR signaling cascade triggered by RNA virus infection. Fas-associated death domain (FADD) and receptor interacting protein (RIP1) have also been shown to interact with the MAVS signaling complex [58]. FADD and RIP1 interact with the C-terminus of MAVS and activate NF-κB by interacting with and activating caspase-8 and caspase-10 [59]. Furthermore, recent studies have demonstrated

21 9 that TNFR-associated death domain (TRADD) forms a complex with MAVS, TRAF3, TANK, FADD and RIP1, which leads to the activation of IRF3 and NF-κB [60]. In this context, STING was also identified as a critical scaffold protein for IRF3 activation [42, 61], highlighting an alternative signaling pathway in the context of cytosolic DNA sensing Type I IFN production Type I IFNs include several IFN-α subtypes and a single IFN-β subtype [62, 63]. In the early phase of the host innate response to RNA virus infection, IFN-β is the predominant type I IFN that is produced. Virus-induced IFN-β transcription requires contributions from several transcription factors including IRF3/7, NF-κB, activating transcription factor 2 (ATF2)/c-Jun, which translocate to the nuclear compartment and bind to the positive regulatory domain (PRD) of the IFN-β promoter at prd II, prd IV, and prd III-1 respectively[64], assembling a large transcriptional complex termed the enhanceosome [65]. Together, these transcription factors promote chromatin remodeling and robust transcription through interaction with the RNA polymerase complex. In the late phase of the host innate response to RNA virus infection, cells produce copious amounts of IFN-α in response to IRF7 activation IRF3 and IRF7: key IFN regulatory factors IRFs represent a family of transcription factors comprised of nine members (IRF1-9) in mammalian cells. Structurally, they share significant homology within the conserved N-terminal DNA-binding domain (DBD), whereas diverse C-terminal domains confer distinct functions for each of the IRFs. In general, IRFs contain an IRF-association

22 10 domain, a nuclear export sequence, an autoinhibitory domain, and a signal-response domain that contains key serine residues targeted for phosphorylation events upon infection with viruses. In particular, IRF3 and IRF7 are the key players in RLR-induced antiviral signaling pathways [66]. IRF3 and IRF7 exhibit key differences in their expression and inducibility patterns. IRF3 is constitutively expressed in a wide variety of cell types as a latent transcription factor in the cytoplasm. Conversely, constitutive IRF7 expression is restricted to immune cells such as B lymphocytes, pdcs and monocytes, and can be strongly induced in other cell types by type I IFNs, virus infection and other stimuli such as bacterial LPS. This positive regulatory feedback loop between IRF7 and type I IFNs represents the major source of IRF7 expression, compared to its low endogenous basal expression in most cell types [67]. The first wave of antiviral signaling is triggered by RIG-I/MDA5-mediated RNA binding and results in strong IRF3 and NF-κB activation. These two transcription factors are then recruited to Ifna and Ifnb promoters and upregulate type I IFN transcription, mainly IFN-β, which is then synthesized and secreted and binds to IFNA receptors in an autocrine or paracrine manner. IFNAR signaling activates the JAK/STAT pathway to induce a large number of ISGs, including IRF7. Upon transcription and translation, IRF7 becomes highly expressed and orchestrates the second wave of type I IFN induction, including IFN-β and various subtypes of IFN-α, while IRF3 is downregulated by proteasomal degradation during this later phase (Fig 1.5). Studies of IRF7-deficient mice have revealed that IRF7 also plays unanticipated roles in the initial wave of IFN production upon virus infection. Indeed, Irf7-/- mice are

23 11 more susceptible than Irf3-/- mice to virus infections, correlating with a marked decrease in serum IFN levels [68]. Studies with IRF7-deficient mice have reinforced the notion that IRF7 is the master regulator of type I IFN. Although IRF7 is expressed at low levels in unstimulated cells, it is absolutely critical in IFN-α/β induction, in the form of IRF7 homodimers or IRF7/IRF3 heterodimers. Upon the initial activation of IFN genes, a positive feedback loop becomes established and IRF7 is induced in a secondary but more robust wave of IFN production [62]. Fig 1. 5 A schematic view of RLR and IFN signaling pathways focusing on IRF4 and IRF Regulation of RLR signaling RLR signaling and type I IFN production must be tightly regulated since aberrant IFN and inflammatory cytokine production downstream of RLRs has been reported in many immune disorders and cancers. Although an insufficient amount of IFN production may promote persistent virus infection, an overproduction of IFN may conversely trigger

24 12 an overactive inflammatory response that can inflict nonspecific tissue damage. In order to fine-tune and balance RLR signaling in a timely manner, a variety of host proteins can positively or negatively regulate RLR signaling by many mechanisms. For example, RIG-I, MDA5 and IRF7 are all ISGs that are induced in a positive feedback loop upon initial IFN production. Also, a number of host proteins have been reported to target key signaling proteins within this pathway, including but not limited to RIG-I, MDA5, IPS-1, TRAF3, TBK1, IKKε, IRF3 and IRF7, and regulate them either positively or negatively by the following mechanisms: (1) influence protein stability, for example, RNF125 targets RIG- I/MDA5/MAVS for degradation [69]; (2) disrupt protein complexes, for example, ABIN1 disrupts and inhibits the interactions between TRAF3 and TBK1/IKKε [70]; and LGP2 sequesters and inhibits dsrna-rig-i interaction [26]; (3) modulate protein subcellular localization, for example inhibition of IRF3/7 nuclear translocation; (4) inhibit transcription factor binding to DNA, for example, Krüppel-like factor 4 (KLF4) targets and prevents IRF3 from binding to the Ifnb gene promoter [71]. In addition, RLR regulatory proteins can also target signaling effectors indirectly by manipulating their post-translational modifications (PTMs), such as phosphorylation, ubiquitination and sumoylation, to achieve the effects described above. For example, the adaptor protein TAX1BP1 [72] and the ubiquitin-editing enzyme A20 (also known as TNFAIP3) [73] block TBK1/IKKε K63-linked polyubiquitination, and the deubiquitinase CYLD inhibits RIG-I K63-linked polyubiquitination [74, 75], both of which play essential roles in RLR signaling. Finally, a number of viral-encoded proteins have been found to suppress RLR signaling and type I IFN by either directly targeting and inhibiting RLR signaling

25 13 components, or indirectly by hijacking host negative regulators. Additional details and relevant examples are forthcoming in section 1.3 and Chapter IFNAR signaling pathway and regulation Overview Upon induction and secretion, IFN molecules bind to IFNAR in an autocrine or paracrine manner and trigger signaling through the Janus kinase signal transducer and activator of transcription (JAK-STAT) signaling pathway, leading to the transcriptional induction of hundreds of ISGs [62]. Collectively, ISGs serve to restrict virus replication and maintain a potent anti-viral state in infected as well as surrounding cells. Some ISGs function to limit RLR signaling in negative feedback loops. Type I IFN also plays an important role in activating and shaping the adaptive immune response at later stages after virus infection (Fig 1.6). Fig 1. 6 Cytosolic nucleic acid pattern recognition and activation of ISGs.

26 IFNAR binding and recognition There are three main classes of IFNs [76]: (1) type I IFNs, the largest IFN class, comprising IFN-α, IFN-β, IFN-ε, IFN-κ and IFN-ω in humans [76, 77], and all of these bind to the type I IFN heterodimeric receptor complex comprising IFN-α receptor 1 (IFNAR1) and IFNAR2 subunits at the cell surface; (2) type II IFN or IFN-γ which mainly signals through the IFN-γ receptor complex (IFNGR) [78]; (3) type III IFNs, comprising IFN-λ1/2/3/4 which utilize a broadly distributed low affinity receptor subunit (IL-10R2) as well as the high-affinity type III IFN receptor subunits which are expressed in epithelial cells [76]. Moreover, type I and type III IFN share similar JAK-STAT signaling pathways as explained in more detail below JAK-STAT signaling cascade In unstimulated conditions, Janus Kinase (JAK) 1, JAK2 and tyrosine kinase 2 (TYK2) are ubiquitously expressed and bind to the cytoplasmic domain of each IFN receptor chain as inactive forms [79]. Once IFN binds to the IFN receptor at the cell surface, receptor chains are brought into close proximity, and two JAK kinase domains juxtapose and undergo trans-phosphorylation and activation. Once activated, JAKs phosphorylate IFN receptor chains on highly conserved tyrosine residues, and STAT proteins are recruited and bind to the receptor via Src homology 2 (SH2) domain interactions [80, 81]. Furthermore, STATs are phosphorylated on conserved tyrosine residues and then released from the receptor complex. After undergoing conformational changes to form homo- (type II IFN) or heterodimers (type I & III IFN), the nuclear localization signal

27 15 (NLS) of STATs becomes exposed which leads to its nuclear translocation, where it functions as a transcription factor and activates the transcription of ISGs [57, 82, 83]. STAT1 homodimers and STAT1-2 heterodimers bind to IFNγ activation site (GAS) enhancer elements in gene promoters [84]. In addition, STAT1 and STAT2 can associate with IRF9 to form a heterotrimeric transcription factor complex known as ISGF3 [85], which binds to IFN-stimulated response elements (ISREs) in promoters [86] Interferon stimulated genes (ISGs) ISGs have a wide variety of functions and can fine-tune RLR signaling and maintain an antiviral state after virus-induced IFN production. Classically, they can be grouped according to their functions: (1) positive regulators, for example, JAK2, STAT1/2, and IRF9 in the IFNAR pathway, or RIG-I and IRF7 in the RLR pathway. Some of these genes exhibit a low level of expression that can be induced as ISGs, and once induced will reinforce the IFN response in a positive feedback loop; (2) negative regulators, for example, suppressor of cytokine signaling (SOCS) [87] proteins and ubiquitin-specific peptidase 18 (USP18) [88, 89], which can desensitize the IFN response by targeting the JAK-STAT signaling pathway; (3) antiviral effectors, such as the tripartite motif family member TRIM5α [90, 91], a potent inhibitor of HIV-1 that binds to the capsid lattice and causes premature uncoating of the virion core. Another well known ISG is ISG15 [92], which is a ubiquitin-like protein that covalently attaches to both pathogen and host derived proteins, a process termed ISGylation [93], that can be considered a type of posttranslational modification (PTM) that regulates protein stability and/or binding to other proteins. Thus far only a few examples of protein ISGylation have been described, however, with the increasing number of target proteins found to be ISGylated, it is

28 16 possible that ISGylation may function as a general, nonspecific mechanism of host defense [94] Suppressor of cytokine signaling (SOCS) proteins SOCS proteins and cytokine-inducible SRC homology 2 (SH2)-domain containing proteins (CISs) comprise a family of intracellular proteins with eight members (CIS and SOCS1-7) that regulate responses of immune cells to cytokines [95-97]. Each member of the family has a central SH2 domain, which determines the protein binding partner; an N- terminal domain of various length and sequences among different members [96]; and a C-terminal 40-amino-acid module known as the SOCS box, which mediates interactions with elongin B/C, cullin-5 and RING-Box-2 (RBX2), and may have intrinsic E3 ubiquitin ligase activity to mediate degradation of SOCS binding proteins [98]. As mentioned above, SOCS proteins represent a class of ISGs that inhibit JAK- STAT signaling by binding to phosphorylated tyrosine residues on IFN receptors or the JAK proteins. Ultimately, SOCS proteins attenuate JAK activation, which results in decreased STAT binding and activation [99]. Also, the SOCS box domain further inhibits IFNAR signaling by recruiting proteins involved in receptor ubiquitination and proteasomal degradation [100, 101]. Taken together, SOCS proteins are ISGs and are induced early in the IFN response and function as IFN desensitizers in a negative feedback loop. Moreover, some cell types have elevated endogenous levels of SOCS proteins, leading to a lower sensitivity of these cells to an IFN response, for example human pluripotent stem cells, which may be physiologically important for host development [102].

29 HTLV-1 and RLR signaling HTLV-1 infection and associated diseases Human T-cell leukemia virus type 1 (HTLV-1) is the first identified human retrovirus, discovered in the early 1980s from a cell line derived from a patient with adult T-cell leukemia/lymphoma (ATLL) [ ]. Notably, HTLV-1 is the only known retrovirus which is directly linked to a human cancer (ATLL) in a causal relationship [107]. HTLV-1 is also associated with various inflammatory diseases such as HTLV-1- associated myelopathy (HAM)/tropical spastic paraparesis (TSP) [109]. Worldwide, there are approximately 20 million people infected with HTLV-1 [109], with the highest prevalence in endemic regions in southern Japan, the Caribbean islands, parts of South America and Central Africa, and foci in the Middle East and Australia [110, 111]. Around 2-5% of HTLV-1 carriers will develop ATLL after a latency period of years [112], while the incidence of HAM/TSP development ranges from % [110]. The mortality of ATLL patients varies among the different clinical subtypes of this disease, ranging from 4-6 months in the highly aggressive acute or lymphoma subtypes, to more than 5 years for the more indolent chronic or smoldering subtypes [113] HTLV-1 transmission and viral replication Cell-free HTLV-1 is poorly infectious. HTLV-1 is transmitted predominantly by cell-to-cell contact and the formation of virological synapses elicited by cytoskeletal reorganization and microtubule organizing center (MTOC) polarization at the site of cell contact between the infected cell and the target T cell [114]. Interactions between the integrin LFA-1 and its ligand ICAM1 are necessary for viral spread through the

30 18 virological synapse and also ensure that T cells are targeted for infection by infected cells. The viral load in HTLV-1 infected individuals is maintained by clonal expansion of HTLV-1 infected cells [115, 116], primarily in CD4+ lymphocytes in vivo [117]. HTLV-1 is a complex retrovirus that encodes requisite structural, enzymatic and regulatory genes that facilitate the viral life cycle. As is typical of other retroviruses, the viral gag, pro, pol and env genes are flanked by long terminal repeats (LTRs) at the 5 and 3 ends that contain cis-acting elements that regulate viral gene transcription. In addition, the px region located between env and the 3 LTR, contains four partially overlapping open reading frames (ORFs) that encode several regulatory proteins made through alternative splicing and internal initiation codons [118]. One of the key HTLV-1 regulatory proteins is Tax, which functions as a potent viral oncogene sufficient to immortalize T cells [119], transform rodent cells [120], and induce tumorigenesis in multiple Tax transgenic mouse models [121]. Tax functions as an oncogene in part by: (I) promoting cellular survival and proliferation, (II) enhancing reactive oxygen species and subsequent DNA damage and (III) evading the immune system. Tax utilizes a variety of strategies to promote cell survival, proliferation and oncogenesis. Tax is a trans-activating protein that can recruit host transcription factors such as CREB and co-activators CREB binding protein (CBP) and p300 to the viral LTR to drive viral gene expression. Tax also modulates signaling pathways to influence cellular gene expression patterns that favor cell proliferation and survival. Tax constitutively activates both canonical and noncanonical pathways of NF-κB [ ], a key transcription factor family that regulates expression of anti-apoptotic and

31 19 inflammatory genes. NF-κB is constitutively activated in most tumor cell types [126], including HTLV-1-transformed T cells [127, 128]. Tax activates NF-κB persistently by interacting with the NEMO subunit of the IKK complex and promoting IKK activation [124]. Tax also induces the expression of IL-17RB to create a positive feedback loop that sustains persistent NF-κB signaling in T cells [129]. As seen with other viral oncogenes, Tax also targets tumor suppressor proteins p53 and Rb to promote virus replication and tumorigenesis. Tax inhibits the function of p53, although the precise mechanism remains unclear [ ]. Also, the majority of ATLL tumors retain wild-type p53, suggesting an active mechanism by HTLV-1 to suppress p53 function. Tax also deregulates cell cycle progression and inhibits the G1/S checkpoint to allow cell cycle progression even in the presence of DNA damage [133], partly through inhibition of p53 [134]. Tax promotes the activation of CDK4/cyclin D complexes by binding to and inactivating the CDK inhibitor p16 INK4A [135]. This event leads to Rb hyperphosphorylation and inactivation and E2F activation and cell cycle progression. Tax can also target Rb for inactivation by proteasomal degradation [136]. Finally, Tax also disrupts cellular DNA repair pathways and chromosomal segregation, leading to accumulation of DNA damage and aneuploidy, key factors for cells to undergo transformation [137]. Although Tax is required for cellular transformation in HTLV-1 infected cells, it also serves as the main target of cytotoxic T lymphocytes (CTLs) and the host immune response [138, 139]. Therefore, it is likely that a balance is reached between cells expressing Tax with higher proliferation rates and susceptibility to CTL killing. In ATLL tumors, Tax expression is silenced in ~60% of patients by genetic or epigenetic

32 20 mechanisms (e.g. mutations or deletions in the provirus or methylation of the 5 LTR). Therefore, it is likely that Tax is necessary for the initial stages of tumorigenesis, and is not essential for maintenance of the transformed phenotype [140, 141]. Another viral protein, HBZ that is encoded by the antisense strand of the HTLV-1 genome, appears to be the only viral protein expressed in all ATLL tumors [142], and therefore may function to maintain the transformed phenotype of ATLL cells [143] Viral evasion of antiviral pathways There has been a long, constant battle between viruses and their hosts, that in many cases it has evolved into a Yin-Yang balance that allows both to survive in the presence of the other. Although host mammalian immune systems have evolved powerful innate and adaptive antiviral systems to combat viruses, it is also clear that viruses have evolved to evade immunity either by actively suppressing critical antiviral signaling proteins, or indirectly by hijacking host-derived regulatory proteins. An increasing number of viral proteins have been identified that can actively suppress host innate antiviral signaling pathways. For example, the E6 viral protein of human papilloma virus (HPV) [144], NS1 encoded by Influenza A [145], and Npro viral protein of bovine diarrhea virus (BDV) [146] and many more have been found to target and inhibit IRF3, and/or NF-κB and AP-1 signaling pathways, to block type I IFN production. In addition, several viral proteins, such as NS5 of Japanese encephalitis virus (JEV) [147] and the C protein encoded by Sendai virus [148], target and inhibit factors in JAK-STAT pathways to suppress the induction of ISGs. Viruses can also disrupt host antiviral signaling by hijacking host regulatory proteins. For example, hepatitis C virus (HCV) and herpes simplex virus type 1 (HSV-1)

33 21 can induce expression of SOCS3, a negative regulator of the IFN pathway [149, 150]. Also, in respiratory epithelial cells infected with the Influenza A virus, both SOCS1 and SOCS3 expression were induced as a mechanism to enhance virus replication [151] HTLV-1, Tax and antiviral signaling pathways Although multiple mechanisms have been elucidated as to how viruses antagonize host antiviral signaling, it remains poorly understood how HTLV-1 evades host innate immunity. HTLV-1-infected T cell clones derived from patients with HAM/TSP by single-cell cloning, were resistant to the anti-proliferative action of IFN-β [152] suggesting a viral mechanism to disable the IFN pathway. Also, an infectious molecular clone of HTLV-1 suppressed IFN-α-dependent ISRE activation by inhibiting the phosphorylation of tyrosine kinase 2 (TYK2) and STAT2 [153]. Moreover, HTLV-1 Tax can compete with STAT2 for binding to the co-activators CBP/p300 binding and thereby inhibits IFN signaling [154]. However, whether and/or how Tax influences RLR signaling pathways remains unknown. 1.4 Aryl hydrocarbon receptor (AhR) and AhR interacting protein (AIP) AhR overview The aryl hydrocarbon receptor (AhR) has been well established to mediate the toxicity of xenobiotics such as 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), formed as a side product in the industrial organic synthesis of herbicides. TCDD was found to be the chemical agent responsible for choloracne outbreaks in chemical workers after exposure, and besides the skin conditions, it may also result in progressive liver failure, myocardial degeneration, and a variety of other medical conditions [155].

34 22 AhR is a host transcription factor that is widely expressed in all tissues [156] and is evolutionarily conserved among vertebrates [157]. In the absence of a ligand, such as dioxins, polychlorinated biphenyls (PCBs) and polyaromatic hydrocarbons (PAHs), AhR shuttles between the nucleus and cytoplasm [158], but mostly is retained as an inactive complex in the cytoplasm. AhR forms a complex with heat shock protein 90 (HSP90) [159, 160], AhR-interacting protein (AIP) [161, 162], and p23 [163, 164] under homeostatic conditions. Upon ligand binding, AhR undergoes a conformational change, exposes its nuclear localization signal (NLS) [165], and subsequently interacts with AhR nuclear translocator (ARNT). AhR and ARNT shuttle into the nucleus, are released from the HSP90 complex, and AhR binds to genomic regions containing dioxin response elements (DREs) and regulates expression of genes in the xenobiotic response, such as drug-metabolizing enzymes CYP1A1, CYP1A2, CYP1B1 [166, 167]. In addition to its function as a regulator of the xenobiotic response, AhR can also regulate and cross-talk with other signal transduction pathways such as NF-κB [168, 169], protein tyrosine kinases and epidermal growth factor receptor (EGFR) [170, 171], p53 and retinoblastoma (prb) [172, 173], transforming growth factor β (TGF-β) [174], mitogen-activated protein kinases (MAPKs) [175, 176], estrogen receptor (ER) [177, 178] and others [ ] AhR and virus infection Recent studies have revealed new and surprising functions of AhR in the immune system [183]. AhR has been implicated in autoimmunity [184, 185], allergic inflammatory diseases [186], and the host response to infection [187, 188]. Most strikingly, AhR has been shown to play a role in the development of specific subsets of T

35 23 lymphocytes. For example, activation of AhR induces regulatory T cell (Tregs) differentiation [189, 190]. AhR also plays an essential role in the development of IL-17- producing T helper cells (Th17) cells, a subset of CD4 + T cells important for microbial (bacteria and fungi) immunity at epithelial/mucosal barriers. When dysregulated, Th17 cells promote persistent inflammation and tissue injury in autoimmune diseases such as multiple sclerosis, Crohn s disease and rheumatoid arthritis [ ]. Furthermore, AhR can negatively regulate the TLR4 signaling pathway, as AhR-deficient macrophages over produce interleukin (IL)-6 and tumor necrosis factor alpha (TNF)-α upon LPS stimulation [194]. The role of AhR in virus infections has been mostly studied using mice infected with human influenza A viruses. For example, treatment with a single oral dose of TCDD increases morbidity and mortality in mice infected with influenza A virus [195] by suppressing both humoral and cell-mediated immune responses [196]. AhR activation also impairs the priming but not the recall of virus specific CD8+ T cells in the lung [197]. Furthermore, AhR activation may influence the host immune response to other viruses as well [198, 199]. However, the underlying mechanisms and specific gene targets of AhR remain to be determined AIP overview Aryl hydrocarbon receptor interacting protein (AIP) [200], also known as X- associated protein-2 (XAP2) [201], Ah receptor-activated 9 (ARA9) [161], or FK506- binding protein 37 (FKBP37) [202], was first identified as an interacting protein of AhR in a yeast 2-hybrid (Y2H) screen using AhR as bait. The human AIP gene encodes a 37 kda protein of 330 amino acids, comprising an N-terminal immunophilin-like domain

36 24 [203] and three C-terminal tetratricopeptide repeat (TPR) domains, the later of which is necessary and sufficient for AhR interaction [204, 205] (Fig 1.7). The Aip genome sequence is highly evolutionarily conserved among many species [206]. Furthermore, AIP is mainly a cytoplasmic protein [200], with ubiquitous expression in human and murine tissues at both mrna and protein levels [162, 200, 201, 205, 207, 208]. Fig 1. 7 A schematic view of AIP protein structure. Under homeostatic conditions, AIP forms a protein complex with AhR and then dimerizes with HSP90 in the cytoplasm. Although the precise role of AIP in the regulation of the AhR pathway is unclear, it has been reported that AIP may either enhance or inhibit the transcriptional activity and expression level of AhR [200, 209]. Nevertheless, there is agreement that AIP protects AhR from ubiquitin-dependent degradation [210, 211]. Moreover, it has been demonstrated that AIP retains AhR in the cytoplasm by altering the recognition of the bipartite NLS of AhR by importin-β [212]. Germ line mutations in the Aip gene have been recently identified and found to be associated with the development of pituitary adenomas [213]. About 30% of familial isolated pituitary adenomas (FIPA) and approximately 80% of patients with AIP-related FIPA exhibit acromegaly or gigantism due to excessive growth hormone secretion [214], suggesting that AIP functions as a tumor suppressor gene in the pituitary gland [215, 216]. In addition, AIP may play an essential role in cardiac development and maintaining

37 25 productive erythropoiesis in mice, as AIP deletion leads to embryonic lethality accompanied by a range of heart deformations in an Aip knockout mouse model [217] AIP and the regulation of the host antiviral response In addition to binding to AhR, it is also noteworthy that AIP was also identified in unbiased screens as a binding partner of select viral proteins. AIP was found to interact with the X protein of the hepatitis B virus (HBV) by Y2H screening and confirmed by in vitro binding assays and immunocytochemistry [201]. Overexpression of AIP suppressed X protein transcriptional activity, suggesting that AIP may function as a negative regulator of the X protein [201]. Furthermore, EBNA-3, one of the six nuclear antigens expressed by Epstein-Barr virus (EBV)-immortalized lymphoblastic cell lines also interacts with AIP as revealed by Y2H and in vitro GST pull-down assays [218]. Further, AIP was found to translocate to the nucleus upon expression of EBNA-3, resulting in enhanced transcription of AhR-responsive genes [219]. A previous study using mass spectrometry to map the innate immune interactome upon virus infection identified a potential interaction between AIP and IRF7, one of the key regulators in innate RLR signaling [220]. Moreover, AIP was shown to interact with the CARMA1 adaptor molecule in T lymphocytes and could enhance CARMA1 binding to MALT1 and BCL10 to promote T cell receptor-mediated NF-κB activation and IL-2 production [221]. Taken together, in addition to its role in AhR signaling in the xenobiotic response, AIP may also regulate innate RLR signaling pathways.

38 Chapter 2 HTLV-1 Tax and innate RLR signaling 2.1 Background Human T cell leukemia virus type 1 (HTLV-1) is etiologically linked to the development of adult T-cell leukemia/lymphoma (ATLL) and the demyelinating disease HTLV-1-associated myelopathy/tropical spastic paraparesis (HAM/TSP) [222]. The HTLV-1-encoded viral protein Tax plays an essential role in HTLV-1-mediated pathogenesis. Tax is a trans-activating protein that deregulates cellular gene expression by modulating the activity of signaling pathways, including NF-κB [223]. Tax is thought to be required for HTLV-1-mediated transformation of T lymphocytes, but is clearly not required for the maintenance of the transformed phenotype in ATLL [118]. Tax is also the main viral target of host cytotoxic T cells (CTLs) and is highly expressed in HAM/TSP patients with high pro-viral loads [224]. Viruses are first detected by the innate immune system via cytoplasmic pattern recognition receptors (PRRs), including RIG-I and MDA5, which recognize specific features of viral nucleic acids [20, 31]. RIG-I, upon detection of viral RNAs bearing 5 - triphosphate, initiates a signaling cascade that ultimately leads to the activation of the interferon regulatory factor 3 (IRF3) transcription factor [225, 226]. RIG-I interacts with the mitochondrial protein IPS-1 (also known as MAVS and Cardif), which leads to the formation of a protein complex anchored to the mitochondrial membrane consisting of TRAF3, IKKγ (also known as NEMO), TBK1, and IKKε [49, 55, 56, 227, 228]. TBK1 phosphorylates several C-terminal serine residues in IRF-3, triggering its dimerization and nuclear translocation where it activates expression of IFN-β [57]. IFN-α is a member of the type I interferon family of genes that binds to the IFN α/βreceptor 1 (IFNAR1) 26

39 27 triggering the JAK/STAT pathway which in turn activates the expression of hundreds of interferon stimulated genes (ISGs) that together coordinate the host antiviral response [229]. Suppressor of cytokine signaling (SOCS) proteins are comprised of a family of eight cytokine inhibitors (CIS and SOCS1 to 7) [87]. The SOCS proteins are induced in a negative-feedback loop and inhibit activated JAK kinases or cytokine receptors to ensure transient activation of cytokine signaling pathways [95]. SOCS proteins all contain an amino-terminal kinase inhibitory domain of variable length, a central Src homology 2 (SH2) domain that binds to phosphorylated tyrosines and a carboxy-terminal 40-amino acid domain known as the SOCS box which promotes the degradation of protein substrates [96, 97]. The SOCS box recruits several regulatory proteins, including elongin B, elongin C, cullin-5 and RING-box-2 (RBX2) to couple SOCS proteins to the ubiquitin-proteasome pathway [230]. Gene targeting studies in mice have confirmed an essential role for SOCS1 as a negative regulator of IFN-γ and IL-2 pathways [231, 232]. SOCS1 also inhibits IFN-α/β signaling and the induction of antiviral genes suggesting that SOCS1 may mediate IFN resistance during antiviral treatment [233]. Many viruses have devised multiple mechanisms to inhibit host antiviral pathways, a necessary step to avoid immune detection and promote viral replication. With regard to HTLV-1, little is known regarding viral mechanisms of counteracting host antiviral signaling. T cell clones derived from HAM/TSP patients are resistant to the antiproliferative effects of IFN-β through an undefined mechanism [152]. A subsequent study demonstrated that HTLV-1, when expressed from an infectious molecular clone, inhibited an interferon stimulated response element (ISRE) reporter [153]. HTLV-

40 28 1blocked IFN-α mediated induction of ISGs by antagonizing the phosphorylation of TYK2 and STAT2 [153]; however, the viral protein responsible for the IFN inhibitory effect was not identified. Another study implicated HTLV-1 Tax as an inhibitor of IFN signaling via competition of Tax with STAT2 for the co-activators CBP/p300 [154]. Taken together, it appears that HTLV-1, and possibly Tax, inhibits IFN signaling. 2.2 Results In this chapter, we demonstrate that HTLV-1-transformed cell lines all exhibit elevated levels of SOCS1 mrna. Tax up-regulates the expression of SOCS1 via the NFκB pathway. Furthermore, Tax interacts with SOCS1 and promotes its stabilization. Tax enhances the replication of the heterologous rhabdovirus vesicular stomatitis virus (VSV) in a SOCS1-dependent manner. Tax does not require SOCS1 to inhibit IFN-α/β signaling, but rather uses SOCS1 to inhibit innate RLR signaling pathways Tax induces the expression of SOCS1 Since it was previously demonstrated that HTLV-1 antagonizes type I IFN signaling [153], we examined the expression of SOCS1, a potent inhibitor of IFN signaling, in a panel of HTLV-1-transformed cell lines by qrt-pcr. SOCS1 mrna was significantly up-regulated in the HTLV-1-transformed cell lines MT-2, MT-4, SLB-1 and C8166 compared to HTLV-1-negative T cell lines Jurkat and SUP-T1 (Fig 2.1A). As expected, Tax mrna was detected selectively in the HTLV-1-transformed cell lines (Fig 2.1A, right panel). Because Tax is a trans-activating viral protein and has been implicated in the inhibition of IFN signaling previously [153], we next determined if Tax was responsible for the increased expression of SOCS1. For this purpose, we used a Jurkat Tax Tet-On cell line that expresses Tax upon doxycycline (Dox) treatment [234, 235]. As shown in

41 29 Fig 2.1B, treatment of Jurkat Tax Tet-On cells with Dox resulted in an enhancement of SOCS1 expression as detected by qrt-pcr. Dox treatment was confirmed to activate Tax expression (Fig 2.1B, right panel). Jurkat Tax Tet-On cells treated with Dox also exhibited an increase in SOCS1 protein levels (Fig 2.1C). Therefore, Tax clearly upregulates the expression of SOCS1, and HTLV-1-transformed cell lines all overexpress SOCS1. We next transfected a Tax plasmid into primary CD4 + T cells to determine if Tax up-regulates SOCS1 in untransformed cells. Tax expression was confirmed by western blotting, which correlated with up-regulated SOCS1 protein (Fig 2.1D). Thus, Tax induces SOCS1 expression in primary CD4 + T cells Tax requires NF-κB, but not CREB, to up-regulate SOCS1. Tax modulates the expression of host genes by deregulating cellular signaling pathways-two of the best-characterized targets of Tax are the NF-κB and CREB pathways. Tax point mutants selectively defective in either NF-κB or CREB are useful to delineate the requirement of each of these pathways in a particular function of Tax [236]. The Tax mutant M22 is defective for NF-κB, but is active for CREB activation. Conversely, Tax mutant M47 is defective for CREB, but is competent for NF-κB activation. Tax was expressed in Jurkat T cells using murine retroviral vectors expressing wild-type Tax, Tax M22 or Tax M47 [237]. Although wild-type Tax and M47 induced SOCS1 expression, M22 was largely defective in the induction of SOCS1, despite similar expression levels of Tax and the two mutants (Fig 2.2A).

42 30 Fig 2. 1 Tax induces the expression of SOCS1. (A) SOCS1 mrna is overexpressed in HTLV-1-transformed cell lines. qrt-pcr measuring SOCS1 and β-actin was performed using mrna from control Jurkat and SUP-T1 cell lines and HTLV-1-transformed cell lines MT-2, MT-4, SLB-1 and C8166. The expression of Tax and GAPDH was examined by RT-PCR (right panel). **, P < for Jurkat versus HTLV-1-transformed cell lines. (B, C) Tax induces the expression of SOCS1 mrna and protein. Jurkat Tax Tet-On cells were treated with Dox (1 µg/ml) for 24 h to turn on Tax expression. qrt-pcr measuring SOCS1 and β -actin was performed using mrna from untreated and Dox-treated cells (B). RT-PCR was also performed for Tax and GAPDH expression (B-left panel). Immunoblotting was conducted with whole-cell lysates from untreated and Doxtreated cells (C). Immunoblotting was performed with anti-socs1 and anti-β-actin. **, P < for untreated versus Dox-treated samples in panel B. (D) Tax upregulates SOCS1 protein in primary CD4+ T cells. Empty vector or Tax plasmid were transfected into primary CD4+ T cells. Immunoblotting was performed with anti-tax, anti-socs1 and anti-β-actin.

43 31 Tax interacts with the IKKγ subunit of the IKK complex as an essential step in activation of NF-κB [238]. Accordingly, Tax is unable to activate NF-κB in a Jurkat cell line variant lacking IKKγ expression (SVT 2C) [239]. Therefore, wild-type Tax was expressed using a retroviral vector in the SVT 2C-IKKγ-deficient cells and the SVT parental cell line. As expected, Tax up-regulated the expression of SOCS1 in the parental SVT cells, but was significantly attenuated in the IKKγ-deficient Jurkat cells, despite similar levels of Tax expression in both cell types (Fig 2.2B). To further demonstrate the role of NF-κB in Tax-mediated induction of SOCS1, we used a cell permeable small molecule inhibitor with selectivity for IKKβ (IKK2 inhibitor IV) [240]. We first validated the inhibitor by demonstrating a potent block in TNF-induced activation of an NF-κB luciferase reporter gene (Fig 2.2C, left panel). Treatment of the HTLV-1 transformed cell line MT-2 with inhibitor IV significantly reduced SOCS1 mrna levels as detected by qrt-pcr (Fig 2.2C, right panel). These data provide further support that NF-κB plays an important role in the regulation of SOCS1 expression in the context of HTLV-1 transformed cell lines Tax interacts with and stabilizes SOCS1 Since Tax exerts many of its effects on cellular signaling pathways by directly interacting with proteins [241], we next examined if Tax interacts with SOCS1. Indeed, transfected Tax interacted with ectopic HA-tagged SOCS1, but not Flag-SOCS3, in 293- T cells (Fig 2.3A). Furthermore, endogenous Tax and SOCS1 formed a stable complex in the HTLV-1 transformed cell lines C8166 and MT-4 that was detectable by coimmunoprecipitation (Fig 2.3B). Next, we examined the binding of Tax and Tax mutants

44 32 Fig 2. 2 Tax requires NF-κB to induce SOCS1 expression. (A) Tax requires NF-κB to upregulate SOCS1. qrt-pcr measuring SOCS1 and β- actin was performed using mrna from Jurkat cells infected with retroviral vectors expressing Tax, Tax M22 or Tax M47. The expression of Tax and GAPDH was examined by RT-PCR (right panel). **, P < for mock versus Tax and Tax M47 samples. (B) Tax induction of SOCS1 is impaired in IKKγ-deficient Jurkat cells. qrt-pcr measuring SOCS1 and β-actin was performed using mrna from parental Jurkat SVT cells or SVT 2C IKKγ-deficient cells infected with a Tax retroviral vector. The expression of Tax and GAPDH was examined by RT-PCR (right panel). **, P < for Tax samples in Jurkat SVT WT versus those in Jurkat SVT 2C. (C) Inhibition of IKKβ reduces SOCS1 expression in HTLV-1-transformed cells. (Left) A luciferase assay was performed with lysates from 293-T cells transfected with an NFκB-Luc reporter and treated with TNF for 8 h. Where indicated, cells were also pretreated with inhibitor IV (40 nm) for 30 min prior to TNF stimulation. (Right) qrt-pcr measuring SOCS1 and β -actin was performed using mrna from the HTLV-1-transformed cell line MT-2 treated with vehicle or inhibitor IV (40 nm) for 2 h. **, P < for vehicle versus inhibitor IV treatment.

45 33 with SOCS1 via a yeast two-hybrid (Y2H) assay. pgbkt7-socs1 and pgadt7-tax were co-transformed in yeast and selected on minimal medium lacking either leucine (Leu - ), tryptophan (Trp - ), histidine (His - ) or adenine (Ade - ). As shown in Fig 2.3C, only when yeast were co-transformed with SOCS1 and Tax (or Tax mutants), colonies formed under highly stringent conditions on plates lacking histidine, leucine, tryptophan and adenine (His - Leu - Trp - Ade - ). These results indicate interactions between wild-type Tax, Tax M22 and Tax M47 with SOCS1 in the yeast two-hybrid assay. Thus, although Tax M22 is unable to induce the expression of SOCS1, it is able to interact with SOCS1. Collectively, these data indicate that SOCS1 is a novel Tax interacting protein. SOCS1 is a relatively unstable protein with a half-life of approximately 1 to 2 h [242]. We next examined if Tax had any effect on the stability of the SOCS1 protein. Therefore, we conducted cycloheximide (CHX) chase assays to examine SOCS1 stability in the absence or presence of Tax. As expected, SOCS1 was a labile protein, which was mostly lost by 6 h of the CHX chase (Fig 2.3D, left panel). However, Tax enhanced the stability of SOCS1, and no loss of SOCS1 protein was observed during the CHX chase (Fig 2.3D, left panel). In contrast, Tax had no effect on the stability of another SOCS family member SOCS3 (Fig 2.3D, right panel). Therefore, it appears that Tax specifically interacts with SOCS1 as a mechanism to antagonize its degradation and promote its stability.

46 34 Fig 2. 3 Tax interacts with SOCS1 and promotes SOCS1 stability. (A) Ectopic SOCS1, but not SOCS3, interacts with Tax. A co-ip assay was performed with lysates from 293-T cells expressing Tax, HA-SOCS1 and Flag-SOCS3 as indicated. Immunoprecipitations were conducted with anti-tax and immunoblotting was performed with anti-ha, anti-flag, and anti-tax as a control for the IPs. Immunoblotting was also performed with lysates using anti-tax, anti-ha and anti- Flag. (B) Endogenous SOCS1 interacts with Tax in HTLV-1-transformed cell lines. A co-ip assay was performed with lysates from C8166 and MT-4 cells as indicated. Immunoprecipitations were conducted with an isotype control immunoglobulin (αigg) or anti-socs1 followed by immunoblotting with anti-tax. Immunoblotting was also performed with lysates using anti-tax and anti-socs1. (C) Tax and SOCS1 interact in a yeast two-hybrid assay. Various combinations of pgbkt7-socs1 and pgadt7-tax (also Tax M22 or Tax M47) or empty vectors were cotransformed in the yeast strain AH109 and plated on minimal medium lacking Leu, Trp, His or Ade as indicated. (D) Tax enhances the stability of SOCS1, but not SOCS3. A cycloheximide chase assay was performed to examine the stability of SOCS1 and SOCS T cells expressing HA-SOCS1 (left panel) or Flag-SOCS3 (right panel), in the absence or presence of Tax, were treated with CHX for the indicated times and lysates were subjected to immunoblotting. Immunoblotting was performed with anti- HA, anti-flag, and anti-tax.

47 Tax requires SOCS1 to inhibit RLR signaling Tax has previously been shown to inhibit type I interferon signaling by competition with STAT2 for the co-activators CBP/p300. Since other viruses such as herpes simplex virus type 1 (HSV-1) [243] and Influenza A [244] virus enhance SOCS1 expression to inhibit IFN signaling, we next investigated if Tax was dependent on SOCS1 to inhibit IFN signaling. Consistent with previous studies [153], Tax expression significantly reduced the activation of an ISRE reporter by IFN-α or IFN-β stimulation (Fig 2.4A). To determine the contribution of SOCS1 to the inhibition of IFN signaling by Tax, we used RNA-mediated interference (RNAi) to inhibit SOCS1 expression. A pool of four different SOCS1 sirnas effectively down-regulated SOCS1 protein levels in 293-T cells (Fig 2.4B). Knockdown of SOCS1 by RNAi enhanced IFN signaling, although Tax was still able to inhibit IFN signaling (Fig 2.4C), suggesting that Tax inhibits IFN signaling independently of SOCS1. We have found that Tax also functions as a potent inhibitor of the RIG-I/TLR3 antiviral pathways (Hyun et al., submitted). Thus, we next examined if Tax inhibition of the RIG-I/MDA5 antiviral pathway was dependent on SOCS1. Indeed, Tax blocked the activation of an IFN-β reporter in response to the double-stranded RNA mimetic poly (I:C) that was largely abolished by sirna knock-down of SOCS1 (Fig 2.4D). Therefore, Tax requires SOCS1 to effectively block poly (I:C)-mediated activation of the IFN-β promoter.

48 36 Fig 2. 4 Tax requires SOCS1 to inhibit RLR antiviral signaling. (A) Tax blocks type I interferon signaling. A luciferase assay was performed with lysates from 293-T cells transfected with an ISRE-Luc reporter and Tax where indicated. Cells were also stimulated with IFN-α, IFN-β, or IFN-γ. **, P < for empty vector versus Tax samples for IFN-α, IFN-β, and IFN-γ treatments. (B) SOCS1 sirna inhibits SOCS1 expression. Immunoblotting was performed with lysates of 293-T cells transfected with control sirna or SOCS1 sirna. Immunoblotting was performed with anti-socs1 and anti-β-actin. (C) Tax does not require SOCS1 to inhibit IFN-α signaling. A luciferase assay was performed with lysates from 293-T cells transfected with an ISRE-Luc reporter, Tax, control sirna or SOCS1 sirna as indicated. Cells were also treated with IFN-α. **, P < for empty vector versus Tax samples (for both cont. sirna and SOCS1 sirna).

49 37 Figure 2.4 Continued (D) Tax requires SOCS1 to inhibit antiviral signaling. A luciferase assay was performed with lysates from 293-T cells transfected with an IFN-β-Luc reporter, poly (I:C), Tax, control sirna or SOCS1 sirna as indicated. *, P < 0.05 for empty vector versus Tax samples [+cont. sirna and poly(i:c)]. (E) Tax requires SOCS1 to enhance the replication of VSV-GFP. Micrographs of 293-T cells transfected with Tax, control sirna or SOCS1 sirna as indicated. VSV-GFP, at an MOI of 0.1, was used to infect cells 24 h after transfections. Pictures were taken 24 h postinfection. The expression of Tax and GAPDH was examined by RT-PCR (right panel). Next, the effect of Tax and SOCS1 on the replication of VSV expressing the green fluorescent protein (GFP) was observed, using fluorescence microscopy. Remarkably, Tax significantly enhanced the replication of VSV-GFP in the presence of a control scrambled sirna. However, RNAi-mediated inhibition of SOCS1 resulted in only low levels of VSV-GFP replication in the presence of Tax (Fig 2.4E). Tax expression was similar in the presence of either control or SOCS1 sirna (Fig 2.4E). Thus, Tax clearly requires SOCS1 as part of its mechanism to enhance viral replication SOCS1 enhances replication of HTLV-1 in MT-2 cells Since SOCS1 is overexpressed in HTLV-1 transformed cell lines (Fig 2.5A), we next asked if SOCS1 modulated HTLV-1 replication. We conducted our studies with MT-2 cells, a T-cell line chronically infected with HTLV-1 [245]. MT-2 cells produce low levels of infectious virus that can be increased by 1 to 2 logs with TNF-α stimulation [246]. MT-2 cells were transfected with control or SOCS1 sirna, and treated with TNFα for 2 h. Viral replication was measured by the detection of the viral structural protein p19 Gag in both the supernatants and cell lysates using western blotting and ELISA. Transfection efficiency was typically between 15 to 30% as measured by flow cytometry of MT-2 cells transfected with FITC-labeled control sirna (data not shown). An ELISA

50 38 for p19 Gag was conducted with supernatants from transfected MT-2 cells, and p19 Gag was significantly reduced in the presence of SOCS1 sirna (Fig 2.5A). Knockdown of SOCS1 also resulted in a decrease in p19 Gag in cell lysates, suggesting that SOCS1 regulates viral replication (Fig 2.5B). Furthermore, the decrease in p19 Gag levels was accompanied by an increase in IFN-β mrna as measured by qrt-pcr (Fig 2.5C). Taken together, these results reveal that SOCS1 is an important host factor induced by HTLV-1 that enhances viral replication by inhibiting RLR signaling and IFN-β production. Fig 2. 5 SOCS1 is a positive regulator of HTLV-1 replication. (A) Loss of SOCS1 decreases extracellular HTLV-1 levels. p19gag ELISA was conducted using supernatants from MT-2 cells transfected with control sirna or SOCS1 sirna. *, P < 0.05 for control sirna versus SOCS1 sirna samples.

51 39 Figure 2.5 Continued (B) Loss of SOCS1 decreases cellular HTLV-1 p19gag. Immunoblotting was performed with lysates from MT-2 cells transfected with control sirna or SOCS1 sirna. Immunoblotting was performed with anti-p19gag and anti-β-actin. p19gag proteins were quantified with ImageJ software and relative levels are represented as a percentile compared to control sirna (arbitrarily set at 100). (C) SOCS1 negatively regulates IFN-β in MT-2 cells. qrt-pcr measuring IFN-β and β -actin was performed using mrna from the HTLV-1-transformed cell line MT-2 transfected with control scrambled sirna or SOCS1 sirna. *, P < 0.05 for control sirna versus SOCS1 sirna samples. 2.3 Discussion SOCS1 is induced by cytokines and functions in a classical negative-feedback loop to restrain activation of cytokine signaling pathways [97]. SOCS1 inhibits JAK/STAT signaling by directly binding to JAK proteins and inhibiting their kinase activity[95]. Due to the importance of SOCS1 in limiting IFN signaling, viruses, including HSV-1 and Influenza A induce the expression of SOCS1 as a mechanism to evade innate immune detection [243, 244]. Our study indicates that SOCS1 gene expression is up-regulated in HTLV-1 infected cell lines. The HTLV-1-encoded Tax protein induces SOCS1 expression in an NF-κB-dependent manner. Tax interacts with SOCS1, but not SOCS3, to mitigate its degradation and inhibit RLR signaling, ultimately leading to the enhancement of viral replication. Prior studies examining the regulatory elements in the SOCS1 promoter have identified IRF1 and IRF2 and Egr-1 binding sites [247]. To our knowledge, NF-κB has not been previously implicated in the regulation of SOCS1 expression. TNFα induces the expression of SOCS1 in mouse hepatocytes, although it is uncertain if NFκB is important for SOCS1 expression in this regard [248]. A search of the

52 40 SABiosciences (Qiagen) regulatory transcription factor search portal ( reveals several putative NF-κB binding sites in the SOCS1 promoter. Whether Tax promotes direct binding of NF-κB to any of these sites or NF-κB indirectly activates SOCS1 via other transcription factors will be important to address in future studies. It is surprising that SOCS1 is dispensable for Tax to inhibit IFN signaling, but is required for Tax to inhibit RLR signaling that regulates IFN-β transcription. Although SOCS1 is well known as an inhibitor of JAK/STAT signaling, its role in RLR signaling is poorly understood. Overexpression of SOCS1, but not SOCS3, inhibits Influenza A virus-mediated activation of an IRF-3 reporter suggesting that SOCS1 may directly inhibit host RLR signaling [244]. Indeed, a recent study demonstrated that HTLV-1 infection results in the induction of SOCS1, which promotes the degradation of IRF-3 [249]. Our data confirmed and extend the results of Oliere et al. since we have shown that Tax induces SOCS1 in an NF-κB-dependent manner and Tax subsequently interacts with and stabilizes SOCS1 protein. Tax relies on SOCS1 to effectively thwart RIG-I/MDA5-mediated antiviral signaling. Finally, our preliminary results indicate that SOCS1 is a potent inhibitor of the RIG-I antiviral pathway, although it appears to function upstream of IRF-3 (data not shown). Human immunodeficiency virus type 1 (HIV-1) also induces the expression of SOCS1 which regulates the intracellular trafficking and maturation of HIV-1 Gag [250]. With regard to HTLV-1, IFN-α inhibits viral assembly by preventing Gag interaction with lipid rafts, although the mechanism is not clear [251]. Knockdown of SOCS1 with sirna reduced the levels of p19 Gag in the supernatant and lysates likely due to enhanced

53 41 expression of IFN-β (Fig. 2.5). Thus, the expression of viral proteins and viral replication is likely reduced when SOCS1 is silenced because of a more robust antiviral response that is typically restrained by SOCS1. The specific antiviral pathways elicited by HTLV- 1, and retroviruses in general, are poorly understood. A recent study has indicated that HIV-1 DNA is recognized by an unknown cellular DNA sensor, that signals to the adaptor molecule STING, TBK-1 and IRF-3 which activates IFN-β expression [252]. HIV-1 inhibits the host innate immune response via the exonuclease TREX1, which digests HIV-1 nucleic acid[252]. Additional studies are necessary to determine if TREX1 is also used by HTLV-1 to counteract innate immune responses. Furthermore, it will be necessary to determine if SOCS1 negatively regulates the host DNA antiretroviral pathway. In summary, we have found that SOCS1 is induced by HTLV-1 as a mechanism to evade innate immunity and promote virus replication. Tax not only induces SOCS1 expression, but also interacts with SOCS1 to enhance its stability and prevent its proteasomal degradation. These finding have implications for HTLV-1-associated diseases, particularly HAM/TSP which is associated with high Tax expression and a high proviral load. Small molecule antagonists of SOCS1 have been described [243] and may potentially be useful to reduce the proviral load in asymptomatic individuals who are at high risk for HAM/TSP. Since IFN-α is used in the clinic as a therapeutic for ATL and HAM/TSP patients [253], elevated SOCS1 may also confer resistance to antiviral drugs.

54 Chapter 3 AIP and RLR signaling 3.1 Background The innate immune system represents the first line of defense against virus infection and is critical for initiating and amplifying adaptive immunity. Virus replication generates multiple pathogen associated molecular patterns (PAMPs), such as doublestranded RNA or single-stranded RNA containing a 5 -triphosphate, which are sensed by host pattern recognition receptors (PRRs) [31, 225, 226]. Activation of PRRs trigger the production of type I interferons (IFNs) including IFN-α/β, which activate the transcription of interferon-stimulated genes (ISGs) that serve to restrict virus replication [254]. There are two classes of PRRs that recognize and respond to viral RNAs: cytoplasmic retinoic acid inducible gene-i (RIG-I)-like helicase receptors (RLRs) and membrane-bound Toll-like receptors (TLRs). The cytoplasmic helicases RIGI and melanoma differentiation associated gene 5 (MDA5) are key RLRs that detect 5 - triphosphate RNA and dsrna (or the dsrna mimetic Poly (I:C)) respectively (2). RIG-I contains tandem amino (N)-terminal CARD domains, a central DECH-box RNA helicase and a carboxyl (C)-terminal regulatory domain. RIG-I is autoinhibited due to intramolecular interactions of the CARD and helicase domains. RIG-I binding to viral RNA triggers a conformational change that exposes the CARDs and promotes lysine 63 (K63)- linked polyubiquitination of RIG-I by the E3 ligase tripartite motif-containing 25 (TRIM25) [255]. RIG-I K63-linked polyubiquitination facilitates multimerization and CARD-CARD interactions with the outer mitochondrial membrane adaptor MAVS (also known as IPS-1, CARDIF or VISA) [228]. MAVS forms prion-like aggregates and 42

55 43 nucleates a signaling complex containing TRAF E3 ligases (TRAF2, TRAF3, TRAF5, TRAF6), the adaptor molecule NEMO and the noncanonical IκB kinases TBK1 and IKKε [55, 256, 257]. TBK1 and IKKε phosphorylate IRF transcription factors (IRF3 and IRF7) that trigger their dimerization, nuclear translocation and activation of type I IFN gene expression [57]. In addition to IRF activation, the NF-κB transcription factor is also activated downstream of RLRs and contributes to the induction of type I IFN [258]. IRF3 and IRF7 are key transcription factors that together orchestrate the production of type I IFN. IRF3 is expressed constitutively in a variety of tissues and is responsible for the early phase of virus-mediated type I IFN induction [259]. Conversely, IRF7 is expressed at very low levels in most cell types, except for lymphoid cells and plasmacytoid dendritic cells (pdcs), and is inducible by interferon [260]. Although IRF3 plays important roles in the initial phase of type I IFN induction, IRF7 is the major player in establishing a secondary and larger wave of IFN productions [66]. Indeed, knockout studies in mice have revealed that IRF7 is essential for virus-mediated MyD88- independent production of type-i IFN [68], and hence IRF7 is known as the master regulator of type I IFN. RLR signaling requires a precise balance between activation and termination to restore homeostasis after viral clearance since deregulated and overproduction of type I IFN may promote inflammatory or autoimmune diseases. As such, a growing number of inhibitors of RIG-I, MAVS, TRAF3 and TBK1 in RLR signaling have been described, including RNF125 [261], USP21 [262], Itch [263], Triad3A [264], NLRP4 [265], A20 [266], TAX1BP1 [72] and ABIN1 [70]. In addition, several inhibitors have been found to

56 44 target either IRF3 [267, 268] and/or IRF7 [ ] by distinct mechanisms to restrict RLR signaling. Aryl hydrocarbon receptor interacting protein (AIP; also known as XAP2, ARA9 and FKBP37) was first identified as an interacting protein of the aryl hydrocarbon receptor (AhR) in a yeast two-hybrid (Y2H) screen using AhR as bait [161, 200, 201]. The human AIP gene encodes a 37-kDa protein of 330 amino acids. AIP contains an N- terminal immunophilin-like domain, three C-terminal tetratricopeptide repeat (TPR) domains and an alpha-7 helix [161, 204]. Previous studies have established that AIP forms a tetrameric complex together with AhR and a dimer of heat shock protein 90 (HSP90) in the cytoplasm [205, 273]. Hsp90 and AIP retain AhR in the cytoplasm and AIP prevents the ubiquitination and degradation of AhR [210]. Upon AhR ligand binding, AIP is dissociated from the complex allowing AhR to enter the nucleus, dimerize with AhR nuclear translocator (ARNT) and activate genes involved in the xenobiotic response [212, 274]. AIP was also identified as a protein associated with several viral proteins, such as the X protein of the hepatitis B virus (HBV; [201]) and EBNA-3 of the Epstein- Barr virus (EBV; [218] ). Emerging studies indicate that AIP has roles outside of its well-known regulation of AhR and the xenobiotic response. AIP can interact with the CARMA1 adaptor molecule in T lymphocytes and enhances CARMA binding to MALT1 and BCL-10 to promote T cell receptor-mediated NF-κB activation and IL-2 production [221]. In this study, we describe a novel unction of AIP as a negative regulator of IRF7 and RLR signaling.

57 Results In this study, we have identified AIP as an inhibitor of IRF7 and virus-induced type I IFN. AIP expression and/or function may potentially be modulated to control IRF7 activation and the production of type I IFN during virus infections or in the setting of autoimmunity AIP interacts with IRF7 A previous study using mass spectrometry to map the innate immune interactome in response to virus infection found that IRF7 may serve as a potential binding partner of AIP [220]. To investigate this possibility we conducted co-ip assays in 293T cells transfected with Myc-AIP and Flag-IRF7 plasmids. IRF7 was detected upon immunoprecipitation of AIP, and AIP was found in IRF7 immunoprecipitates (Fig 3.1A), indicating that ectopic AIP can interact with IRF7. We next examined the interaction of endogenous AIP with Flag-IRF7, in the absence or presence of Sendai virus (SeV) infection. Interestingly, endogenous AIP interacted with IRF7 and the interaction was further enhanced after virus infection (Fig 3.1B). Increased virus induced AIP-IRF7 interaction was observed despite similar expression levels of AIP and IRF7 in the lysates (Fig 3.1B). The enhanced virus triggered interaction between AIP and IRF7 appeared to be specific since AIP did not interact with overexpressed MDA5, with or without virus infection (Fig 3.1B). AIP contains an N-terminal PPIase (peptidylprolyl cis trans isomerase)-like domain, three C-terminal tetratricopeptide repeat (TPR) domains (TPR1, TPR2, TPR3) and an alpha-7 helix [161]. To identify the domain(s) within AIP required for its interaction with IRF7, we generated a series of N- or C-terminal deletion mutants of AIP. AIPΔN lacked

58 46 Fig 3. 1 AIP interacts with IRF7 in a virus-inducible manner. (A) 293T cells were transfected with Myc-AIP (200 ng) and/or Flag-IRF7 (500 ng) as indicated. After 24 hours, cells were lysed and subjected to co-ips and western blotting using the indicated antibodies. (B) 293T cells were transfected with 500 ng of Flag-IRF7 or Flag-MDA5. After 24 hours, cells were infected with Sendai virus (20 HA/ml), and lysates subjected to co-ips and western blotting using the indicated antibodies. the PPIase domain; AIPΔC1 lacked the alpha-7 helix; AIPΔC2 lacked the alpha-7 helix and TPR3; AIPΔC3 lacked the alpha-7 helix, TPR2 and TPR3; AIPΔC4 lacked the alpha- 7 helix, TPR1, TPR2 and TPR3 (Fig 3.1C). As expected, full length (FL) AIP interacted with IRF7 and the interaction was enhanced by SeV infection (Fig 3.1E). The basal interactions between AIPΔC1 and ΔC2 and IRF7 appeared to be diminished, however SeV-induced binding between these mutants and IRF7 remained intact (Fig 3.1E). AIPΔC3 and ΔC4 mutants were completely impaired in binding to AIP, both basally and virus-induced (Fig 3.1E), indicating that AIP TPR2 is important for the observed binding with IRF7. Although AIPΔN could interact with IRF7, enhanced interaction upon SeV

59 47 Figure 3. 1 Continued (C) Schematic of AIP domains and deletion mutants. (D) Schematic of IRF7 domains and deletion mutants. (E) 293T cells were transfected with 200 ng of wild-type AIP or AIP deletion mutants together with Flag-IRF7 (500 ng). After 24 hours, cells were lysed and subjected to co-ips and western blotting using the indicated antibodies. (F) 293T cells were transfected with 500 ng of wild-type IRF7 or IRF7 deletion mutants. After 24 hours, cells were lysed and subjected to co-ips and western blotting using the indicated antibodies. IgH=immunoglobulin heavy chain; IgL=immunoglobulin light chain. infection did not occur (Fig 3.1E). Therefore, it appears that both PPIase-like and TPR2 domains of AIP mediate interactions with IRF7 triggered by SeV infection. IRF7 contains an N-terminal DNA binding domain (DBD), constitutive activation domain (CAD) between amino acids (aa) 151 and 246, a virus-activated domain (VAD) between aa 278 and 305, inhibitory domain (ID) between aa 372 and 467 and signal

60 48 response domain (SRD) containing key virus-induced phosphorylation sites between aa 468 and 491 [275]. Concurrently, we utilized a series of IRF7 C-terminal deletion mutants to identify the essential domain(s) involved in its interaction with AIP (Fig 3.1D). IRF7Δ1 lacks the SRD; IRF7Δ2 lacks part of the ID and SRD; IRF7Δ3 lacks the entire ID and SRD; IRF7Δ4 lacks the VAD, ID and SRD; IRF7Δ5 lacks part of the CAD, VAD, ID and SRD. Full-length IRF7, and IRF7Δ1 and Δ2 mutants interacted with AIP and the binding was further enhanced by SeV infection (Fig 3.1F). The binding between IRF7Δ3 and AIP was considerably weaker, and IRF7Δ4 and Δ5 no longer interacted with AIP (Fig 3.1F). Thus, it appears that the VAD and part of the ID comprise the AIP interaction motifs within IRF AIP inhibits RLR signaling IRF7 is known as the master regulator of type I IFN and is critical for virus-induced type I IFN in a variety of cell types [68]. Therefore, we next investigated if AIP played a role in virus-induced type I IFN production. First, we conducted a luciferase reporter assay using an interferon-β (IFN-β) luciferase reporter plasmid. Interestingly, AIP inhibited SeV-induced IFN-β promoter activation in a dose-dependent manner (Fig 3.2A). AIP inhibition of the IFN-β promoter appeared to be specific since AIP did not inhibit TNF-induced activation of an NF-κB luciferase reporter (Fig 3.2B). Because IRF7 is essential for virus-mediated induction of IFN-α production [68], we next examined if AIP could modulate IRF7 activation of an IFN-α4 luciferase reporter. We transfected small amounts of IRF7 to activate the reporter since 293T cells express miniscule amounts of IRF7. Indeed, robust activation of the IFN-α4 promoter was observed with the combination of SeV infection and ectopic IRF7 (Fig 3.2C). However, overexpression

61 49 of AIP significantly diminished IFN-α4 activation by SeV and IRF7 (Fig 3.2C). Together, these results indicate that AIP functions as a negative regulator of RLR signaling pathways that control virus-induced production of type I IFN. Fig 3. 2 AIP inhibits RLR antiviral signaling and promotes virus infection. (A-C) 293T cells were transfected with Myc-AIP (200 ng) and/or Flag-IRF7 (500 ng) together with 100 ng of IFN-β-luc (A), NF-κB-luc (B) or IFN-α4-luc (C) and prl-tk renilla plasmid (10 ng). After 24 hours, cells were infected with Sendai virus (20 HA/ml) or treated with TNF (20 ng/ml) as indicated. Cell lysates were subjected to dual-luciferase assays and western blotting using the indicated antibodies. (D) 293T cells were transfected with either non-targeting control or AIP sirnas. After 24 hours, Flag-IRF7 (500 ng), IFN-α4-luc (100 ng) and prl-tk renilla (10 ng) plasmids were transfected followed by Sendai virus infection (20 HA/ml). The next day, cell lysates were harvested and subjected to dual-luciferase assays and western blotting using the indicated antibodies.

62 50 We next performed loss-of-function studies to determine if loss of AIP enhanced virus-mediated IFN-α4 promoter activation. Expression of endogenous AIP was suppressed by transfection of a pool of four distinct sirnas into 293T cells, and confirmed by western blotting (Fig 3.2D, bottom). Knockdown of endogenous AIP did not affect basal levels or the induction of IFN-α4 promoter activation by IRF7 alone (Fig. 3.2D). However, sirna-mediated knockdown of AIP significantly enhanced IFN-α4 promoter activation by SeV infection together with IRF7 transfection (Fig. 3.2D). To determine if AIP regulation of type I IFN production played any role in modulating virus replication, we next investigated if either overexpression or knockdown of AIP influenced the replication of vesicular stomatitis virus expressing green fluorescence protein (VSV-GFP). Cells were transfected with Myc-AIP, infected with VSV-GFP and the subjected to immunofluorescence microscopy and western blotting to examine GFP expression as a surrogate for virus infection. Overexpression of AIP strongly enhanced VSV-GFP replication as observed by microscopy and western blotting for GFP (Fig 3.2E). Conversely, sirna-mediated knockdown of AIP suppressed replication of VSV-GFP (Fig 3.2F). Taken together, these data provide strong evidence that AIP inhibits RLR signaling pathways and type I IFN activation that leads to increased virus replication.

63 51 Figure 3. 2 Continued (E, F) 293T cells were transfected with either 200 ng empty vector or Myc-AIP (E), or 20 pmol control non-targeting or AIP sirnas (F), followed by infection with VSVGFP (MOI 0.001). After 18 hours, cells were examined by fluorescence microscopy (images taken at 10X magnification). Cell lysates were subjected to western blotting using the indicated antibodies AIP-deficient cells overproduce type-i IFN and are resistant to virus infection Genetic ablation of AIP in mice leads to embryonic lethality due to cardiac malformation [217, 276]. To provide genetic evidence for a role of AIP in the regulation of RLR signaling we used Aip / mouse embryonic fibroblasts (MEFs). Whereas wildtype (WT) MEFs were readily infected by VSV-GFP, Aip / MEFs were profoundly resistant to virus infection at an identical multiplicity of infection (MOI) (Fip 3.3A). To demonstrate that these effects were due solely to loss of AIP, we reconstituted Aip / MEFs with an AIP plasmid. Reconstitution of the AIP-deficient MEFs with AIP restored virus replication as determined by immunofluorescence microscopy and western blotting for GFP (Fig 3.3B).

64 52 Based on our earlier results, we hypothesized that the resistance of Aip / MEFs to virus infection was imparted by deregulated RLR signaling and increased levels of virusmediated type I IFN production. To examine this notion, we conducted ELISAs using Aip / MEFs to quantify the magnitude of virus-triggered I IFN production. Initially, Aip / MEFs were transfected with poly (I:C), a synthetic analog of double-stranded RNA, to activate RLR signaling. IFN-β in the cell supernatants was quantified by ELISA. In wildtype MEFs, transfection with poly (I:C) induced approximately 50 pg/ml of IFN-β (Fig 3.3C). However, poly (I:C) induced over 400 pg/ml of IFN-β in supernatants from Aip / MEFs (Fig 3.3C). Wild-type and Aip / MEFs were also infected with SeV and IFN-α4 in cell supernatants was quantified by ELISA. SeV infection induced about 10 pg/ml of IFN-α4 from wild-type MEFs, whereas Aip / MEFs produced over 200 pg/ml after SeV infection (Fig 3.3D). Therefore, RLR signaling is clearly deregulated in Aip / MEFs and both IFN-β and IFN-α4 are overproduced upon virus infection or poly (I:C) transfection. These data further confirm that AIP is an essential negative regulator of RLR signaling.

65 53 Fig 3. 3 Aip -/- MEFs produce elevated type I IFNs and are resistant to virus infection. (A) Aip / and wild-type MEFs were infected with VSV-GFP (MOI 1.0). After 18 hours, cells were examined by fluorescence microscopy (images taken at 10X magnification), and lysates were subjected to western blotting using the indicated antibodies. (B) Aip / MEFs were transfected with 200 ng of empty vector or Myc-AIP, followed by VSV-GFP virus infection (MOI 1.0). The next day cells were subjected to fluorescence microscopy (images taken at 20X magnification), and lysates used for western blotting using the indicated antibodies. (C, D) Aip / and wild-type MEFs were transfected with poly (I:C) (C; 20 µg) or infected with Sendai virus (D; 20 HA) for the indicated times. Cell supernatants were harvested and subjected to ELISA for the detection of IFN-β and IFN-α4 as indicated.

66 54 We next conducted quantitative real-time PCR (qrt-pcr) assays to examine the effect of either AIP overexpression or deficiency on type I IFN mrnas. In wild-type MEFs, transfection of IRF7 induced the expression of IFN-α4 mrna which was further potentiated by SeV infection (Fig. 3.3E). However, AIP overexpression significantly diminished SeV/IRF7-induced expression of IFN-α4 mrna (Fig 3.3E). We next reconstituted Aip / MEFs by transient transfection with an AIP plasmid, infected with SeV and examined the expression of IFN-β, IFN-α4 and IL-6 mrnas by qrt-pcr. As expected, SeV infection up-regulated the expression of IFN-β, IFN-α4 and IL-6 mrnas in Aip / MEFs; however, reconstitution with AIP suppressed virusmediated induction of type I IFN (IFN-β, IFN-α4), but not IL-6 (Fig 3.3F). Given that IL- 6 is predominantly regulated by the NF-κB pathway downstream of RIG-I, these data indicate that AIP specifically suppresses the IRF pathway that controls type I IFN.

67 55 Figure 3. 3 Continued (E) Wild-type MEFs were transfected with Flag-IRF7 (500 ng) and/or Myc-AIP (200 ng) plasmids, followed by Sendai virus infection (20 HA/ml). Total mrna was extracted and subjected to qrt-pcr for IFN-α4 expression (upper panel), and RT- PCR for IRF7 and AIP expression (lower panels). 18S rrna was used as the internal control in both cases. (F) Aip / MEFs were transfected with 200 ng of empty vector or Myc-AIP, followed by Sendai virus infection (20 HA/ml). The next day, total mrna was extracted and subjected to qrt-pcr for IFN-β, IFN-α4 and IL-6 expression. Cell lysates were subjected to western blotting using the indicated antibodies.

68 AIP specifically targets IRF7 for inhibition by interfering with its nuclear localization Thus far we have demonstrated that AIP interacts with IRF7 in a virus-inducible manner and AIP suppresses virus-induced type I IFN production. IRF3 is a key transcription factor activated by TBK1 downstream of RIG-I and MAVS that induces the expression of type I IFN [56]. The next series of studies were undertaken to determine if AIP inhibited IRF7 and/or IRF3 function. First, we conducted an IFN-β luciferase reporter assay to determine if sirna-mediated knockdown of AIP enhanced activation by a super-active form of IRF3 (IRF3SA), a phosphomimetic containing five C-terminal Ser/Thr to Asp mutations [277]. Although knockdown of AIP augmented SeV-mediated IFN-β activation, it had no effect on IRF3SA-induced activation of IFN-β (Fig 3.4A). IRF3 and IRF7 are activated upon phosphorylation by the IKK-related kinases TBK1 and IKKε [57]. Western blotting was performed to determine if AIP blocked the phosphorylation of IRF3 and/or IRF7. IRF3 phosphorylation was detected by a phosphorylation specific antibody that recognizes phosphorylated Ser396. As expected, overexpression of a constitutively active form of RIG-I (ΔRIG-I) comprising the N- terminal CARD domains, elicited phosphorylation of endogenous IRF3 (Fig 3.4B). Overexpression of AIP had no effect on IRF3 phosphorylation; however, the ubiquitin editing enzyme A20 blocked IRF3 phosphorylation as previously reported (Fig 3.4B) [266]. To investigate IRF7 phosphorylation, we transfected IRF7 in 293T cells together with a TBK1 expression plasmid. Due to a lack of functional phospho-irf7 antibodies, we monitored IRF7 phosphorylation by a band-shift due to a slower migration pattern on SDS-PAGE gels. Indeed, TBK1 triggered the phosphorylation of IRF7, which was not

69 57 inhibited by AIP overexpression (Fig 3.4C). Therefore, AIP does not block either IRF3 or IRF7 phosphorylation. Fig 3. 4 AIP does not inhibit IRF3 and IRF7 phosphorylation. (A) 293T cells were transfected with either non-targeting control or AIP sirnas. After 24 hours, cells were transfected with either empty vector (500 ng) or IRF3SA (500 ng) plasmids, together with IFN-β-luc (100 ng) and prl-tk renilla (10 ng), followed by Sendai virus infection (20 HA/ml). The next day, cell lysates were harvested and subjected to dual-luciferase assays. (B, C) 293T cells were transfected with GFP-ΔRIG-I (500 ng), Myc-AIP (200 ng), Flag-A20 (200 ng), Flag-IRF7 (500 ng) and GFP-TBK1 (500 ng) as indicated. The next day cell lysates were harvested and subjected to western blotting using the indicated antibodies.

70 58 Upon TBK1-induced phosphorylation, IRF3 and IRF7 form homodimers and translocate to the nucleus to activate type I IFN genes [277, 278]. Therefore we next examined if AIP inhibited virus-triggered nuclear localization of IRF7 and IRF3. First, we conducted biochemical fractionation experiments to examine TBK1- induced IRF7 nuclear translocation in the absence or presence of AIP. As expected, overexpressed TBK1 promoted increased IRF7 expression in the nuclear compartment (Fig 3.5A). However, exogenous AIP blocked TBK1-induced IRF7 nuclear localization (Fig 3.5A). AIP expression was detected in both cytoplasmic and nuclear compartments (Fig 3.5A), consistent with previous studies [207]. Therefore, AIP may inhibit IRF7 by antagonizing its nuclear localization. Fig 3. 5 AIP inhibits IRF7 nuclear localization and transactivation. (A) 293T cells were transfected with Flag-IRF7 (500 ng), GFP-TBK1 (500 ng) and Myc-AIP (200 ng). After 24 hours, cells were lysed and subjected to subcellular fractionation, followed by western blotting using the indicated antibodies. Anti-LDH1 was used as a marker for the cytoplasmic compartment. Anti-PARP was used as a marker for the nuclear compartment.

71 59 To further examine this notion, we conducted immunofluorescence microscopy experiments in Aip / MEFs reconstituted with an AIP plasmid and subsequently infected with SeV. We also transfected an IRF7 plasmid since MEFs expressed extremely low levels of endogenous IRF7 (data not shown). In uninfected Aip / MEFs (in the absence or presence of AIP transfection), IRF7 was mainly localized in the cytoplasm and perinuclear region (Fig 3.5B). After SeV infection, IRF7 was mostly found in the nucleus; however, reconstitution of Aip / MEFs with AIP blocked SeV-mediated IRF7 nuclear localization (Fig 3.5B). Figure 3. 5 Continued (B, C) Aip / MEFs were seeded on glass coverslips overnight and transfected with Flag-IRF7 (500 ng), Flag-IRF3 (500 ng) and Myc-AIP (200 ng) as indicated, followed by Sendai virus infection (20 HA/ml). After 18 hours, cells were subjected to immunofluorescence microscopy. Flag-IRF7 staining is green; Myc-AIP staining is red; DAPI (nuclear staining) is blue.

72 60 This effect was specific to IRF7 since AIP did not inhibit SeV-induced nuclear translocation of IRF3 (Fig 3.5C). Therefore, AIP specifically blocks virus-induced IRF7, but not IRF3, nuclear localization. Finally, we examined the effect of AIP on IRF7 transactivation using a Gal4-IRF7 fusion comprising the DNA binding domain (DBD) of Gal4 fused to the coding sequence of IRF7 [279]. Luciferase assays were conducted using a reporter plasmid containing the luciferase coding gene driven by Gal4 binding sites. Expression of Gal4-IRF7 strongly activated the Gal4-Luc reporter; however, AIP suppressed Gal4-IRF7 activation in a dose-dependent manner (Fig 3.5D) indicating that AIP blocks IRF7 transactivation. Figure 3. 5 Continued (D) HEK293T cells were transfected with Gal4-IRF7 (500 ng), Myc-AIP (100/200/500 ng) and Gal4-luc (100 ng). After 24 hours, cell lysates were subjected to dual-luciferase assays and western blotting assay using the indicated antibodies.

73 Discussion RLR signaling and the production of type I IFN are tightly regulated to ensure effective viral clearance with minimal collateral tissue damage and without excessive inflammation. However, chronic virus infection or genetic factors may perturb the delicate balance between activation and termination of RLR signaling pathways. In this regard, uncontrolled production of type I IFNs has been implicated in a number of cancers and immune disorders such as psoriasis, systemic lupus erythematosus (SLE) and multiple sclerosis [280]. In this study, we have identified AIP as a novel inhibitor of the RIG-I antiviral pathway and virus-induced type I IFN production. In addition to its wellstudied role as a chaperone and regulator of AhR subcellular localization, AIP also interacts with and limits the activation of IRF7 during virus infection (Fig 3.6). Our results indicate that AIP interacts with IRF7 and this interaction is enhanced by virus infection. We have also found that AIP inhibits virus-induced nuclear localization of IRF7. We have mapped the AIP binding domains in IRF7 to the VAD and part of the ID. The IRF7 VAD domain between aa 278 and 305 is required for optimal activation of IRF7 and is necessary for virus-induced IRF7 nuclear translocation [275]. The ID domain is thought to maintain IRF7 in an inactive state by intra-molecular interactions with the N-terminal DBD and transactivation domains and C-terminal SRD containing key virusinducible phosphorylation sites [275]. Therefore, it is tempting to speculate that AIP blocks IRF7 nuclear translocation by interacting with and masking the VAD domain. Since AIP interacts with part of the IRF7 ID domain, AIP may also keep IRF7 in an inactive closed form similar to what has been proposed for Kaposi s sarcoma herpes

74 62 virus (KSHV) ORF45 inhibition of IRF7 [281]. The following mechanisms may also contribute to the inactivation of IRF7: (1) AIP may induce a conformational change in IRF7 that impairs NLS recognition recognized or binding to import machinery; (2) AIP may enhance nuclear export of IRF7; and (3) AIP may promote the proteasomal turnover of nuclear IRF7. Interestingly, the inhibition of IRF7 nuclear localization by AIP is strikingly similar to the regulation of AhR by AIP. AIP binds to AhR and inhibits AhR nuclear translocation by triggering a conformational change that prevents binding of its bipartite NLS to importin-β [212]. Therefore, AIP may antagonize the nuclear localization of AhR and IRF7 transcription factors by similar mechanisms, although further studies are needed to address this possibility. AIP is ubiquitously expressed in human tissues and cell lines, and it appears that AIP expression is not induced by virus infection (data not shown). However, our results indicate that the AIP-IRF7 interaction is induced by virus infection and thus raises the possibility that AIP functions as a negative feedback inhibitor of RLR signaling. It is likely that AIP undergoes specific posttranslational modifications (PTMs) in response to virus infection that facilitates enhanced interactions with IRF7. Indeed, our preliminary studies suggest that overexpression of TBK1 alters the mobility of AIP on SDS-PAGE gels and promotes a band-shift that could represent phosphorylation (Fig 3.4C). Future studies will identify potential virus-triggered PTMs in AIP and how these regulate AIP inhibition of IRF7 and RLR signaling.

75 63 Fig 3. 6 A model for AIP inhibition of IRF7. Since IRF7 functions as the master regulator of type I IFN production, not surprisingly it is targeted for inactivation by a number of viruses. For example, Ebola Zaire virus VP35 [282], Epstein-Barr virus BZLF1/LF2 [283, 284] and KSHV RTA/ORF45/vIRF3 [ ] all directly target IRF7 for inhibition. Another potential mechanism by which viruses may inactivate IRF7 is by hijacking host-encoded inhibitors of RLR signaling. In this regard, we have reported that human T-cell leukemia virus type 1 (HTLV-1) Tax hijacks a host factor, suppressor of cytokine signaling 1 (SOCS1), to inhibit RLR signaling and dampen type I IFN production [288]. In light of our findings that AIP functions as a negative regulator of IRF7 and RLR signaling, it is interesting that AIP has been found to interact with several viral proteins such as hepatitis B virus X protein [201] and EBNA-3 [218]. It is plausible that certain viruses may hijack AIP to restrict RLR signaling and promote virus replication. AhR has been long known as a regulator of the host xenobiotic response to aromatic hydrocarbons such as 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) [289]. More recently, AhR has been implicated as an important regulator of adaptive immunity and the

76 64 development of regulatory T cells and Th17 T cells [192, 290]. AhR also regulates the host response to virus infection. Treatment of mice with a low dose of the AhR ligand TCDD increased morbidity and mortality in mice infected with influenza A virus [291], by impairing the priming of CD8+ T cells in the lung [292]. Given the broad roles of AhR in the regulation of multiple facets of immune cell development, activation and differentiation, it will be interesting to determine if AhR inhibits innate RLR signaling and type I IFN together with AIP. Gene targeting studies in mice have revealed that AIP plays essential roles in cardiac development and maintaining productive erythropoiesis [217]. In humans, germline mutations in AIP have been identified and linked to the development of pituitary adenomas [213]. AIP mutations have been found in 30% of familial isolated pituitary adenomas (FIPA) and approximately 80% of patients with AIP-related FIPA exhibit acromegaly or gigantism due to excessive growth hormone secretion [214]. Many of the reported AIP mutations lead to loss-of-function and result in truncated or no protein [273], suggesting that AIP functions as a tumor suppressor gene in the pituitary gland. Whether FIPA patients with loss of function mutations in AIP are more resistant to virus infection and/or exhibit heightened inflammatory responses owing to increased IRF7 activation remains to be seen.

77 Chapter 4 Concluding remarks and future perspectives 4.1 SOCS1 and Tax Innate RLR signaling constitutes an important part of the host response to virus infection by activating type I IFN transcription to restrict virus infection. Upon the production of type I IFN, inflammatory cytokines, chemokines and other proinflammatory molecules, host cells become armed in an antiviral state. However, it is vital to mount a precise and targeted response that will eliminate the pathogens without triggering excessive inflammation and tissue damage to the host. Therefore, a number of checks and balances exist that exert tight control over the innate immune response to pathogens. In this thesis, we have identified and characterized two host proteins functioning as negative regulators in innate RLR signaling. The first is SOCS1, an established inhibitor of IFN signaling, and an ISG induced in a negative feedback loop by JAK/STAT proteins [95]. SOCS1 is a potent inhibitor of interferon IFN-γ signaling, as SOCS1-deficient mice die prematurely due to severe hepatotoxicity, multi-organ inflammation and aberrant hematopoiesis [293, 294]. Treatment of SOCS1-deficient mice with IFN-γ neutralizing antibody can rescue the phenotype[295]. Recently, the role of SOCS1 in the regulation of IFNAR signaling has been explored in greater detail. SOCS1-deficient neonatal mice exhibit enhanced viral resistance and lower viral loads upon Semliki Forest virus (SFV) infection [296]. Similarly, our studies in the HTLV-1 transformed cell line MT-2 [245] have revealed that sirna-mediated knockdown of SOCS1 diminished HTLV-1 viral load as indicated by p19 Gag expression [288]. Recently, an inhibitory role of SOCS1 in type I IFN signaling has been described in the context of 65

78 66 infection with other viruses including HSV-1 [297], Influenza A virus [151], HIV-1 [298], RSV [299], Japanese encephalitis virus (JEV) [300] and hepatitis C virus (HCV) [301]. Together, these studies indicate a tight correlation between SOCS1 expression and viral load. It is worth noting that a decreased viral load due to lower SOCS1 expression could be due to either: (1) increased type I IFN transcription, as we have observed with IFN-β mrna and HTLV-1 viral load in MT-2 cells; or (2) a more robust IFNAR/JAK/STAT pathway activation that results in increased induction of ISGs- for example, IFN-α/β expression was not elevated but actually diminished in SOCS1- deficient mice infected with SFV [296]. While the molecular basis of these intriguing findings remain unknown, only a handful of studies have addressed the role of SOCS1 in the RLR antiviral pathway [151, 249, 288]. Nevertheless, the direct target of SOCS1 in the RLR pathway remains to be determined. In our studies, we have demonstrated that the HTLV-1 Tax oncoprotein targeted SOCS1 by multiple mechanisms. Tax not only promoted SOCS1 mrna expression via the NF-κB pathway, but also interacted with and stabilized SOCS1 protein. Our data provide strong evidence that Tax hijacks SOCS1 as a mechanism to inhibit virus-induced IFN-β expression. Since Tax does not require SOCS1 to dampen IFN signaling (Fig 2.4), it appears that the main function of Tax hijacking of SOCS1 is to block RLR signaling. Given that Tax also inhibits IFN signaling, which plays an important role in the induction of SOCS1 transcription, it is interesting that Tax instead utilizes NF-κB to upregulate SOCS1 expression. Tax also regulates SOCS1 post-translationally by increasing the half life of SOCS1 protein. The precise mechanism is unknown although it is possible that Tax may disrupt or target an E3 ligase complex that promotes SOCS1 degradation. It is

79 67 also possible that Tax modulates SOCS1 post-translational modifications, such as phosphorylation and/or K48-linked polyubiquitination to enhance its stability. 4.2 AIP and IRF7 In addition to SOCS1, we have also identified AIP as an inhibitor of RLR signaling. Our work on AIP arose due to efforts to understand how SOCS1 inhibited the RIG-I pathway. In order to identify potential binding partners of SOCS1, we conducted a yeast two-hybrid screen using a mouse macrophage cdna library and SOCS1 as bait. One of the top candidates identified was AIP (data not shown). Although we were unable to confirm the interactions between AIP and SOCS1 biochemically, it is plausible that this putative interaction could be transient or perhaps additional stimuli such as virus infection are necessary. Nevertheless, both AIP and SOCS1 are negative regulators of innate RLR signaling. Therefore, it is tempting to speculate that AIP may function cooperatively with SOCS1 to inhibit RLR signaling. Future studies should examine if SOCS1 can target IRF7 for inhibition, as we have observed with AIP. Since Tax hijacks SOCS1 to dampen IFN production, we have also conducted preliminary studies to determine if Tax similarly has usurped AIP function. Indeed, our preliminary data indicates that Tax also interacts with and stabilizes AIP protein (data not shown), strikingly similar to what was observed with SOCS1. However, it does not appear that Tax up-regulates AIP expression at the transcriptional level. A panel of HTLV-1 infected T cell lines express different levels of endogenous AIP (data not shown). We also did not observe increased AIP expression in response to Sendai virus or VSV infection (Fig 4.5A). Therefore, it appears that AIP is not induced by type I IFN and is not an ISG. Given its ubiquitous expression among different tissues and cell types and

80 68 established roles in different pathways (e.g. AhR), it is likely that AIP inhibition of RLR signaling is triggered by post-translational modifications and/or inducible interactions with RLR signaling proteins. It will be of interest in future studies to determine if Tax requires AIP to inhibit RLR signaling, and also if Tax can modulate the subcellular localization and/or post-translational modifications of AIP (and IRF7). An interesting question to consider is whether the role of AIP is to fine-tune RLR signaling or inhibit the pathway by a negative feedback mechanism. During the course of our studies, we have found that TBK1 can phosphorylate AIP, leading to a slower migrating form of AIP on western blots, which disappeared after calf intestinal alkaline phosphatase (CIP) treatment (Fig 4.1A). Indeed, bioinformatics analysis of the human AIP protein sequence for consensus TBK1 phosphorylation sites yielded two potential TBK1 sites: S104 and S132 (Fig 4.2). We also conducted liquid chromatography and tandem mass spectrometry to identify AIP phosphorylation sites induced by recombinant TBK1 protein (Fig 4.1B). This effort identified T40, S131 and S132 as potential phosphorylation sites (Fig 4.1C). Therefore, combined bioinformatics analysis and mass spectrometry have revealed four putative AIP phosphorylation sites: T40, S104, S131 and S132. We have initiated studies to determine the functional role of AIP phosphorylation in its regulation of IRF7. In this regard, we have generated phospho-mutant serine-toalanine mutations and phospho-mimetic serine-to-glutamic acid mutants. As expected, the phospho-mutant does not undergo a band shift upon TBK1 overexpression. Conversely, the phospho-mimetic undergoes a band shift even in the absence of TBK1 (Fig 4.3A).

81 69 Fig 4. 1 TBK1-induced AIP phosphorylation sites identified by liquid chromatography and tandem mass spectrometry. (A) HEK293T cells were transfected with Flag-TBK1 and Myc-AIP plasmids. After 24 hours, cells were lysed and incubated with CIP as indicated for 1 h at 37 C, followed by western blotting using the indicated antibodies. (B) HEK293T cells were transfected with Flag-TBK1 and Myc-AIP plasmids followed by SDS-PAGE and Commassie blue staining. (C) Bands shown in 4.1B were removed, digested with trypsin and subjected to liquid chromatography and tandem mass spectrometry. The identified phosphorylation sites within AIP are indicated.

82 70 Fig 4. 2 TBK1 phosphorylation consensus sites within AIP predicted by Nepos 2.0 Server (Technical University of Denmark). Interestingly, the triple-phospho-mimetic AIP mutant (T40E/S131E/S132E) binds more strongly with IRF7, which is not further inducible by virus infection (Fig 4.3B). The AIP triple phospho-mimetic also co-localizes with and sequesters IRF7 more efficiently than wild-type AIP in the cytoplasm, regardless of virus infection (Fig 4.3D). Moreover, this mutant potently promotes virus replication (Fig 4.3C) and inhibits IFN-β/α4 induction at 24 and 48 hours post-infection with Sendai virus (Fig 4.4 A-B) but not IL-6 (Fig 4.4C). These data suggest that AIP phosphorylation by TBK1 may promote its IRF7 inhibitory function; therefore TBK1 may orchestrate negative feedback of RLR signaling by activating AIP. Conversely, the corresponding AIP phospho-mutant (T40A/S131A/S132A) could still bind to IRF7 (Fig 4.3B) and inhibit IFN-β/α4 induction (Fig 4.4 A-B) but not IL-6 (Fig 4.4C), therefore paradoxically loss of function was not observed with this particular mutant. It is likely that there is redundancy between multiple phosphorylation sites and additional sites should be mutated in future studies. An obvious candidate is S104 and we have recently generated an AIP quadruple phospho-mutant (T40A/S104A/S131A/S132A). Functional studies with this new mutant are currently underway in the lab.

83 71 Fig 4. 3 The AIP triple phospho-mimetic strongly interacts with IRF7 and promotes virus replication. (A) HEK293T cells were transfected with Myc-AIP WT, Myc-AIP phospho-mimetic (3E), Myc-AIP phospho-mutant (3A) with or without GFP-TBK1. After 24 hours, cells were lysed and lysates were subjected to western blotting using antibodies as indicated. (B) HEK293T cells were transfected with Myc-AIP WT, Myc-AIP phospho-mimetic (3E), Myc-AIP phospho-mutant (3A) and Flag-IRF7. After 24 hours, cells were infected with SeV at 20 HA/ml for 1 hour. After an overnight incubation, cells were lysed and subjected to co-immunoprecipitation with Flag antibody, and western blotting was performed using antibodies as indicated. (C) Aip / MEFs were reconstituted with Myc-AIP WT or Myc-AIP phospho-mimetic (3E). After 24 hours, cells were infected with VSV-GFP for 1 hour. After an overnight incubation, cells were analyzed by fluorescence microscopy. Magnification: 20X.

84 72 Fig 4. 3 Continued (D) Aip / MEFs were seeded on glass coverslips and the following day were transfected with Flag-IRF7, Myc-AIP WT and/or Myc-AIP phospho-mimetic (3E). After 24 hours, cells were infected with SeV at 20 HA/ml for 1 hour. After an overnight incubation, cells were fixed and analyzed by immunofluorescence microscopy using anti-flag (rabbit, IRF7, green channel), anti-myc (mouse, AIP, red channel), followed with DAPI (blue channel) staining for nuclear compartment. Magnification: 60X oil.

85 73 Fig 4. 4 The AIP triple phospho-mutant and phospho-mimetic mutants both inhibit the induction of type I IFN, but not IL-6. (A-C) Aip / MEFs were reconstituted with Myc-AIP WT and/or Myc-AIP phosphomimetic (3E). After 24 hours, cells were infected with SeV at 20 HA/ml. Cells were harvested at the indicated time points, and mrnas were extracted and subjected to qrt-pcr quantification using specific primers as indicated.

86 AhR in RLR signaling Since AIP was initially identified and characterized as a chaperone for AhR, it raises an interesting question whether AhR plays any role in the regulation of innate RLR signaling. In this regard, we have some preliminary data that point to the potential involvement of AhR in RLR signaling. Firstly, we have shown by western blotting that AhR protein is up-regulated as early as 1 hours post-infection with Sendai virus, and is further accumulated up to 48 hours post-infection (Fig 4.5A). Since AhR is not induced by virus infection at the transcriptional level, this suggests either a stabilization mechanism by modulation of post-translational modifications and/or enhanced translation of AhR protein upon virus infection. Interestingly, treatment of cells with L-Kynurenine (L-Kyn), a known endogenous ligand for AhR, enhances VSV replication compared to vehicle control (Fig 4.5B-C). Since L-Kyn activates endogenous AhR, these data indicate a possible negative regulatory function of AhR in the RLR pathway. Interestingly, indoleamine 2,3-dioxygenase (IDO), which catalyzes L-tryptophan to yield L-Kyn, can be inducible by type I and II IFN and virus infections [302]. Therefore, IFN itself may contribute to a negative feedback mechanism by inducing IDO and hence L-Kyn to activate AhR (Fig 4.6). Future studies will be needed to determine if AhR and AIP cooperate and are dependent on each other to inhibit virus-mediated IFN induction. It will also be necessary to determine if AhR targets IRF7 for inactivation, similar to AIP.

87 75 Fig 4. 5 Endogenous AhR protein, but not mrna, is increased by virus infection and enhances virus replication. (A) MCF-7 cells were infected with Sendai Virus at an MOI of 20 HA/ml at various time points as indicated, followed by RT-PCR and western blotting to assess mrna and protein levels of AhR and AIP. 18S and β-actin served as internal controls. (B-C) MCF-7 cells were treated with L-Kynurenine (L-Kyn) at a concentration of 50 µm, followed by infection with VSV-GFP at MOIs of 0.01, 0.05 and 0.1. After 24 hours, cells were analyzed by fluorescence microscopy and western blotting.

88 Fig 4. 6 A model depicting the potential involvement of AhR in inhibiting RLR antiviral signaling. 76