REGULATION OF dsRNA-INDUCED TRANSCRIPTION BY NFκB AND IRF-3 THROUGH TLR3 AND RIG-I

by

CHRISTOPHER POPE ELCO

Submitted in partial fulfillment of the requirements

For the degree of Doctor of Philosophy

Thesis Advisor: Dr. Ganes C. Sen

MSTP / Molecular Virology Program

CASE WESTERN RESERVE UNIVERSITY

August, 2007 CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the dissertation of

______

candidate for the Ph.D. degree *.

(signed)______(chair of the committee)

______

______

______

______

______

(date) ______

*We also certify that written approval has been obtained for any proprietary material contained therein. TABLE OF CONTENTS

LIST OF TABLES……………………………………………………………. 3

LIST OF FIGURES…………………………………………………………... 4

ACKNOWLEDGEMENTS…………………………………………………… 6

LIST OF ABBREVIATIONS………………………………………………… 8

ABSTRACT………………………………………………………………….. 10

CHAPTER PAGE

1. INTRODUCTION……………………………………………………. 12

Innate immunity as regulators of immunity PAMPs and PRRs Toll-like 3 CARD-helicase Transcription factors IRF-3 NFκB Signaling by viral dsRNA

2. MATERIALS AND METHODS……………………………………. 35

Cell culture and generation of cell lines…………………………….. 35 Cell treatment protocols…………………………………………….. 36 inhibitors dsRNA, IFN and TNFα treatment viral infection siRNA treatment lentiviral preparation and infection macrophage preparation RNA techniques…………………………………………………….. 40 Ribonuclease protection assay (RPA) cDNA microarray analysis Quantitative real-time PCR Northern blot analysis Techniques………………………………………………….. 44 Western blot analysis Electrophoretic mobility shift assay (EMSA)

1 Nuclear Fractionation Two-dimensional gel analysis Native PAGE Immunofluorescence

3. ROLE FOR STAT1 IN -INDUCIBLE, NOT………. 49 CONSTITUTIVE RESPONSIVENESS TO ADDED dsRNA

Abstract……………………………………………………………… 49 Introduction………………………………………… ………………. 50 Results……………………………………………… ………………. 53 Discussion…………………………………………………………… 88

4. DIFFERENTIAL REGULATION OF INDUCED BY…… 94 dsRNA SIGNALING THROUGH TLR3

Abstract……………………………………………………………… 94 Introduction………………………………………… ………………. 95 Results………………………………………………………………. 96 Discussion…………………………………………………………… 117

5. ANALYSIS OF ROLES OF TLR3, IFN, IRF-3 AND NFκB IN…. 123 INDUCTION BY SENDAI INFECTION

Abstract………………………………………………..…………… 123 Introduction………………………………………… ……………… 124 Results……………………………………………… ……………… 126 Discussion………………………………………………………….. 143

6. INHIBITION OF INDUCED BY IRF-3…... 153

Abstract…………………………………………………………….. 153 Introduction………………………………………… ……………… 154 Results……………………………………………… ……………… 156 Discussion………………………………………………………….. 183

7. DISCUSSION AND PERSPECTIVE……………………………… 189

BIBLIOGRAPHY………………………………………………….. 196

2 LIST OF TABLES

Table 2-1: Description of cell lines used in this dissertation. 37

Table 2-2: Sequences of real-time PCR primers used. 42

Table 2-3: Description of antibodies used in this dissertation. 45

Table 5-1: Genes induced or repressed by SeV infection in various cell lines. 150

3 LIST OF FIGURES

Figure 1-1: through TLR3 and RIG-I. 17

Figure 3-1: Differential induction of ISG56 by extracellular dsRNA and 54

transfected dsRNA.

Figure 3-2: Cellular gene induction by extracellular dsRNA and SeV in 57

U3A cells.

Figure 3-3: STAT1 cannot restore dsRNA responsiveness in U3A cells. 60

Figure 3-4: STAT1 not required for extracellular dsRNA signaling. 62

Figure 3-5: Impaired activation of IRF-3 and NFκB in U3A cells. 65

Figure 3-6: RIG-I and PKR not necessary for extracellular dsRNA signaling. 68

Figure 3-7: TLR3 and , but not MyD88 required for extracellular 71

dsRNA signaling.

Figure 3-8: Defective TLR3 mRNA expression in U3A cells. 74

Figure 3-9: Induction of dsRNA responsiveness by interferon pretreatment. 77

Figure 3-10: Effect of TLR3 on EMCV replication. 80

Figure 3-11: TLR3-independent ISG56 induction by influenza B. 83

Figure 3-12: Differential induction of genes by influenza B, SeV and dsRNA. 86

Figure 4-1: Differential responsiveness of endogenous and exogenous TLR3. 98

Figure 4-2: Direct and indirect gene induction by dsRNA 101

Figure 4-3: IRF-3 mediated gene induction by dsRNA. 104

Figure 4-4: NFκB dependent gene induction. 107

Figure 4-5: Role of PI3K in gene induction by dsRNA. 111

4 Figure 4-6: Differential induction of ISG56 family of genes. 115

Figure 5-1: Effect of TLR3 on the regulation of cellular genes by SeV 127

and dsRNA.

Figure 5-2: Differential induction of genes by SeV and dsRNA. 130

Figure 5-3: IFN dependent regulation of SeV-induced genes. 133

Figure 5-4: Requirement of NFκB for gene induction by SeV. 136

Figure 5-5: IRF-3 dependent gene induction. 139

Figure 5-6: Regulation of SeV-induced genes by IRF-3. 141

Figure 5-7: Negative regulation of SeV-induced A20 mRNA expression 144

by IRF-3 not in Wt cells.

Figure 6-1: Modulation of A20 mRNA and ISG56 mRNA expression by 157

cellular levels of IRF-3.

Figure 6-2: Effect of IRF-3 on expression of A20 protein. 159

Figure 6-3: A20 induction by dsRNA and TNFα. 162

Figure 6-4: NFκB activation in IRF-3 overexpressing cells. 165

Figure 6-5: Dependence on NFκB of IRF-3 inhibited genes. 168

Figure 6-6: Effect of IRF-3 on expression of anti-apoptotic genes. 171

Figure 6-7: A20 repression by IRF-3 is cell type specific. 174

Figure 6-8: Cell-type specific effects of IRF-3 expression. 176

Figure 6-9: Differential activation of IRF-3 by SeV and added dsRNA. 179

Figure 6-10: Expression of IRF-3 varies between tissues and cell types. 181

5 ACKNOWLEDGEMENTS

I first would like to thank my advisor Dr. Ganes Sen for his guidance throughout this work. I have only just begun to realize how much I have grown as a scientist during my time in his lab. It is my heartfelt wish that the corresponding increase in Ganes’s blood pressure, while he has struggled with keeping me focused all these years, was not nearly as great.

I would also like to thank my graduate committee of Clark Distelhorst, Steven

Emancipator, George Stark, and Bryan Williams. Without their advice and perspective on not only science, but life in general, this work would not have been possible.

My thanks also extend to all the members of the Sen lab who I have had the privilege of working with over the past five years. It has been an honor to work with everyone, but special thanks must go to my fellow investigators on the mysteries of dsRNA signaling: Lenette Lu, Kristi Peters, Saumen Sarkar, and Heather Smith. In addition, I must give credit to Theresa Rowe and more credit to Kristi Peters for looking after me so well and assuring all my laboratory needs were always taken care of. Finally,

I must acknowledge the other members my ‘graduate school support group’ not already mentioned, John Fuller, Jason Hill, Danny Hui, Shoudong Li and Aatur Singhi.

There are many more collaborators and colleagues to which I am deep grateful that are not listed here for both the sake of brevity, and because I’m running out of different ways to write thank you.

Most importantly, I want to acknowledge, thank, get down on my knees and worship, my loving father and mother who have been my most invaluable and constant source of support throughout not only graduate school, but life.

6 Finally, science as we practice it is not possible without funding, and for that I am deeply grateful to MERCK and the UNCF for the fellowship which I was privileged enough to receive from them. I am also extremely grateful to the NIH for their support in funding my research, as well as the Medical Science Training Program at Case

Western Reserve University.

7 LIST OF ABBREVIATIONS

CARD caspase activated recruitment domain

DC dendritic cell

DExD/H box RNA helicases motif dsRNA double-stranded RNA

EMCV encephalomyocarditis virus

Est expressed sequence tag

GAS gamma-activated site

IFN interferon

IRF interferon regulatory factor

ISG interferon stimulate gene

ISRE interferon stimulated response element

LRR leucine-rich repeat

Mda-5 melanoma differentiation-associated gene 5 moi multiplicity of infection

NFκB nuclear factor kappa B

PAMP pathogen-associated molecular pattern pfu plaque forming unit

PI3K phosphatidylinositol 3-kinase

PKR protein kinase RNA regulated

PRR pattern recognition receptor

RIG-I retinoic acid inducible gene

8 SeV Sendai virus

siRNA small interfering RNA

ssRNA single-stranded RNA

STAT signal transducer and activator of transcription

TIR toll/IL-1R homology domain

TLR toll-like receptor

TNFα tumor necrosis factor alpha wt wild-type

9 Regulation of dsRNA-Induced Transcription by NFκB and IRF-3 Through TLR3 and RIG-I

Abstract

by

CHRISTOPER POPE ELCO

dsRNA is an important regulator of gene induction. The cellular response to dsRNA is mediated through either toll-like receptor 3 (TLR3) or the RNA helicases,

RIG-I and . While TLR3 and RIG-I/mda5 recognize dsRNA in different cellular compartments, both can signal to activate the transcription factors IRF-3 and NFκB.

Both transcription factors, in turn, regulate the expression of hundreds of genes. This dissertation contains the following contributions to the field of dsRNA signaling:

1. Sendai virus, a negative-strand RNA virus, does not induce cellular genes through TLR3, rather RIG-I is required for the induction of ISG56 by SeV. In contrast, adding dsRNA to the culture media leads to only TLR3-mediated gene induction.

2. STAT1-null, U3A cells do not respond to dsRNA added to the media as the result of low TLR3 expression. Expression of STAT1 in U3A cells allows them to respond to interferon, but not dsRNA. However, pre-treatment of STAT1-restored U3A cells with interferon-γ or interferon-β induces TLR3 expression, and makes the cells responsive to dsRNA. This result shows that stimulation with interferons, by inducing

TLR3, can make cells, otherwise unresponsive to dsRNA, responsive.

10 3. By microarray analysis, which dsRNA and Sendai virus-induced genes are

regulated by IRF-3 or by NFκB were determined. Genes induced indirectly by dsRNA signaling, either dependent upon or independent of autocrine IFN signaling were also

identified. Finally, the PI3K inhibitor LY blocks TLR3-mediated induction of IRF-3 but

not NFκB-dependent genes, thus showing that unlike IRF-3 dependent gene induction,

NFκB-dependent induction through TLR3 does not require PI3K.

4. IRF-3 inhibits the expression of a subset of genes normally induced by SeV and dsRNA. The degree to which SeV-induced gene expression is inhibited is inversely proportional to the expression level of IRF-3. The inhibitory effects of IRF-3 are independent of its transcriptional activation, as gene induction by TNF-α, which does not activate IRF-3, is also inhibited. Although many of the genes negatively regulated by

IRF-3, including A20, c-IAP1, c-IAP2, and TRAF1, are NFκB-dependent, IRF-3 does not inhibit NFκB activation.

In total these results add to the understanding of the complexities of dsRNA- mediated gene induction.

11 CHAPTER 1

INTRODUCTION

All life, down to the smallest bacteria, is susceptible to attack by microbial organisms. As a result, systems of defense or immunity have evolved. More complex organisms have developed more complex immune systems. In vertebrates the immune system has developed two branches. These are the adaptive (or acquired) immune response and the innate immune response.

There are a number of distinctions that define adaptive and innate immunity, but the three most important can be summed up as follows: 1. The adaptive response against a given microbe is highly specific, while innate immunity at its most specific is based on the recognition of a limited number of evolutionarily conserved microbial components, often termed pathogen associated molecular patterns, or PAMPs. 2. While the innate response is immediate, the adaptive response is delayed, and can take days to become fully effective. 3. Once established, the adaptive response is ‘remembered’ and can be quickly and specifically activated again. In contrast multiple infections by the same microbe are not remembered as such by the innate immune response.

Adaptive and innate immunity can both be further sub-divided into humoral and cellular components. Humoral immunity can be defined by its archaic roots as being mediated by molecules in the blood. Prime examples of adaptive and innate humoral immune effectors would be antibodies and the complement system, respectively.

Cellular immunity as its name implies, is mediated by cells, with common examples to represent adaptive and innate branches being T cells and NK cells.

12 Interferons as regulators of immunity.

Cytokines are central effectors of the innate immune response. In response to

viral infections, chief among these are the type I interferons (IFNs). Type I IFNs have

long been appreciated for their ability to prime both the innate as well as the adaptive immune response. Pre-treatment of cells with IFN can dramatically improve the outcome of viral infection both in cultured cells and in vivo. Conversely, inhibition of IFN production or preventing its function increases susceptibility to infection. In and of themselves IFNs do not have any anti-viral activity, rather they mediate their effect by signaling the transcription and translation of hundreds of cellular genes, collectively termed ISGs for IFN stimulated genes.

There are many members of the type I IFN family. The most well known are

IFN-α and IFN-β, which share little structural homology with each other. IFN-β was first isolated from fibroblasts, while IFN-α was discovered in lymphoid cells. Within the

IFN-α family alone, there are over 13 subtypes encoded by different genes. The proteins they synthesize can vary in size from 15 to 21 kD, and contain different amounts of post- translational modifications. Despite the apparent diversity in type I IFNs, the genes encoding all the IFNα subtypes along with IFNβ can be found on the same locus of 9 in humans. In addition all type I IFNs signal through the same hetero- dimeric cell-surface receptor composed of IFNAR1 and IFNAR2c. There is, however, variation in the type and degree of response generated.

Signaling occurs principally through the Jak/STAT pathway, but involves different members of the two families of proteins. Independent of ligand binding, Tyk2 a member of the Jak family associates with IFNAR1, while Jak1, STAT1 and STAT2

13 associate with IFNAR2c. The binding of IFN brings both receptor subunits into close

proximity with each other. This allows Jak1 to phosphorylate Tyk2, which in turn

phosphorylates Jak1 along with Tyr466 of IFNAR1. The phosphorylation of Tyr466 on

IFNAR1 allows STAT2 to bind to it instead of IFNAR2c. IFNAR1-bound STAT2 is

then phosphorylated on Tyr690 leading to the STAT2-dependent phosphorylation of

STAT1 on Tyr701. The two tyrosine-phosphorylated STAT proteins preferentially

dissociate from the receptor and associate with each to form a heterodimer. The STAT1-

STAT2 dimer associates with a member of the interferon regulatory factor family, IRF-9,

to form the trimeric transcription complex ISGF3 (interferon stimulated gene factor 3).

ISGF3 induces transcription by translocating from the cytoplasm to the nucleus, where it binds to ISRE (interferon stimulated regulatory element) sequences present in the promoters of all ISGs. An ISRE is defined by the loose consensus binding site of

GAAANNGAAA. Within the ISGF3 complex, IRF-9 provides the specificity for binding to the ISRE element, while the COOH-terminus transactivation domain of

STAT2 appears to be critical for transcription. However, in addition to either of these components, deletion of STAT1 or any other protein involved in the IFN signaling cascade will severely impair all ISRE-mediated gene induction by type I IFNs.

Role of PAMPs and Pattern Recognition Receptors in IFN induction.

While IFNs may be key mediators of innate immunity, they are expressed only upon stimulation. One of the longstanding questions of innate immunity is how do IFNs get induced in the first place. It was not until the mid-1990s that answers to this question really began coming, but since then the field has exploded. Interferons were originally

14 discovered in relation to virus infection, but it was unknown how virus induced them. In the search for ways to induce IFN in the absence of virus infection, it was discovered that dsRNA was a good inducer of IFN. Since many contain RNA genomes, or produce dsRNA during their replication, it was thought that viral dsRNA might be the signal which cells were recognizing for the induction of type I IFNs.

Further study of dsRNA as an inducer of IFN revealed a number of things. First, pre-treating cells in culture with dsRNA could help inhibit viral replication. Second, in the absence of the IFN signaling pathway, dsRNA could still induce many genes on its own. Third, dsRNA could induce interferon induced genes without activating the IFN signaling pathway. Taken together these findings suggested that, like IFN, dsRNA alone was a potent antiviral signal capable of inducing an anti-viral state by stimulating the induction of many genes.

At the same time, the importance of the components of other microbes in stimulating an immune response was beginning to become appreciated. Endotoxic shock, as a result of the bacterial cell wall component LPS, was long known to be mediated by a strong cellular response. The term pathogen associated molecular pattern, or PAMP, was coined to loosely describe any of the fairly generic, microbial components that were able to induce an immune response.

The cellular proteins vital to innate immunity which were able to recognize specific PAMPs were termed Pattern Recognition Receptors (PRRs). Many proteins meeting the definition of a PRR were previously known for some time, such as PKR and

OAS in the context of dsRNA, or mannose binding lectins in the context of bacteria.

15 However, the key mediators of signaling remained elusive until the discovery of

Toll-like receptors, and more recently cytosolic NOD/CARD receptors.

Pattern Recognition Receptors for dsRNA.

Overview. Multiple cellular proteins have the ability to bind to dsRNA. PKR can bind dsRNA, and in turn inhibit translation through the phosphorylation of eIF2α. In addition

PKR has been reported to activate NFκB. PACT can bind dsRNA and activate PKR.

The recognition of dsRNA by 2’-5’ oligoadenylate synthetase ultimately leads to the activation of RNAse L and the degradation of RNA within the cell (Torrence, Maitra et al. 1993). The siRNA system provides a specific response to short dsRNAs, by targeting the degradation of specific cognate transcripts in the cell. However gene induction, and

IFN induction in particular, is mediated by two distinct PRRs, the toll-like receptor family member, TLR3 (Alexopoulou, Holt et al. 2001) and a class of CARD-helicase proteins, of which two members, RIG-I (Yoneyama, Kikuchi et al. 2004) and Mda5

(Andrejeva, Childs et al. 2004) are known. Both types of PRRs have distinct dsRNA recognition domains and signaling domains. Through their signaling domains both have been shown to induce IFN through the activation of transcription factors IRF-3, NFκB and AP-1. The largest difference between the two is that they recognize dsRNA in different cellular compartments. CARD-helicases are thought to see dsRNA within the cytoplasm, while TLR3 sees endocytosed dsRNA, as well as dsRNA on the cell surface.

Figure 1-1 gives an overview of the signaling pathways by which these three transcription factors are activated by dsRNA signaling through both TLR3 and RIG-I.

16 Virus

CD14 endosome/lysosome TLR3 (fibroblasts) CD14

TLR3 RIG-I TRIF (CARD-helicase) PI3K

TRAF3 IPS-1/MAVS TRAF6 RIP1

TAB2 TBK1/ TAB1 TAK1 NEMO IKKi IKKα IKKβ

IRF-3 p38 JNK IκB p65 p50 ATF2 c-Jun

AP-1 kB ISRE GGRNNYYCC GAAANNGAAA

17 Toll-like receptor 3 as a mediator of dsRNA signaling. Toll-like receptors are so named due to their similarity to the original Drosophila protein, Toll (Rock, Hardiman et al.

1998). Although originally identified for its role in development, Drosophila Toll was shown to have anti-fungal immune effects. Currently there are twelve known mammalian members of the Toll-like receptor family, most of which are extremely well conserved between species (Akira, Uematsu et al. 2006). In addition to TLR3 which

mediates dsRNA signaling, three other TLRs are known to recognize nucleic acid. TLRs

7 and 8 have been shown to recognize ssRNA and TLR9 recognizes DNA containing

CpG motifs. All toll-like receptors share three features in common, a cytoplasmic

Toll/IL-1R homology domain (TIR), a transmembrane domain, and an ectodomain containing multiple Leucine-rich repeats (LRRs).

The ectodomain, which forms the amino-terminal end of the protein, is important for recognition of PAMPs specific to the receptor. Each TLR is composed of between 19 and 25 LRRs, repeated in tandem. TLR3 has one of the larger amino-termini, which contains 23 Leucine rich repeats. Each repeat is 24 to 29 amino acids in length and contains the conserved motif, XLXXLXLXX (Choe, Kelker et al. 2005; Akira, Uematsu et al. 2006). Analysis of the crystal structure of the ectodomain indicates it is a horseshoe-shaped solenoid. The surface of the ectodomain is heavily glycosylated, with

35% of the total weight of TLR3 being accounted for by carbohydrate (Sun, Duffy et al.

2006). Glycosylation is critical to proper function of TLR3 and mutation of either of two asparagine residues in LRR 8 or 15 will abolish signaling. A single surface of the ectodomain of TLR3 does lack glycosylation, and two charged regions on this face are likely to mediate the homo-dimerization, as well as facilitate RNA binding.

18 The cytoplasmic portion of TLR3 is comprised of the 179 amino acids of the

COOH-terminus. Of these, the last 125 amino acids, 756-904 comprise the Toll/IL-1R homology domain (TIR). This region is so named because of the homology shared between the cytoplasmic regions of TLRs and the IL-1 receptor. The TIR domains are required for both IL-1R and TLR mediated signaling and deletion or mutation can result in a dominant negative protein (Poltorak, He et al. 1998). In essence the TIR domain acts as a scaffold for the recruitment of downstream adaptor molecules also containing TIR domains. There is a degree of specificity among TIRs as evidenced by the fact that different TLRs signal through different adaptor proteins.

TLR3 is not a ubiquitously expressed protein; rather its expression is normally restricted to select cell types. Immune cells, including dendritic cells, macrophages, B- cells, and some T-cells, along with fibroblasts, epithelial cells and, in the brain, astrocytes and glial cells are all known to express TLR3 (Town, Jeng et al. 2006). There is also variation between where in the cell TLRs are expressed. TLRs 1, 2, 4, 5 and 6 have all been shown to reside on the cell surface, while TLRs 7, 8, and 9 are endosomal. TLR3 localization has been differentially reported in either place. It is thought that in fibroblasts, TLR3 is a cell surface protein, while in Dendritic cells (DCs) it resides inside the cell (Matsumoto, Funami et al. 2003). Localization may depend on a linker region in the cytoplasmic domain, as Funami et al have shown that mutation of the 30 amino acids between the TIR domain and the transmembrane domain will cause TLR3 to go to the cell surface instead of localizing to vesicles (Funami, Matsumoto et al. 2004).

Activation of TLR3 requires its recognition of and direct binding to dsRNA (Bell,

Askins et al. 2006). In order to be recognized by TLR3, dsRNA must first get to it.

19 CD14, a originally identified as critical for LPS-induced signaling through

TLR4, has been reported to bind dsRNA and mediate its uptake into the cell (Lee,

Dunzendorfer et al. 2006). In macrophages from CD14-deficient mice, cytokine induction by TLR3 is reduced. Both dsRNA and TLR3 co-localize to lysosomes via endosomal trafficking and reports suggest that pH is an important factor for an interaction between the two (de Bouteiller, Merck et al. 2005). It is not clear however, if signaling through TLR3 expressed on the cell surface of fibroblasts is, likewise, affected by either

CD14 or the pH of the environment. Upon activation by dsRNA a number of changes occur. TLR3 dimerizes and at least two Tyrosine residues within its TIR domain, Y759 and one other, get phosphorylated. Downstream signaling components are recruited and bind to the TIR domain (Sarkar, Peters et al. 2004). At the moment, it is not clear in what order these three events occur. Phosphorylation of TLR3 is required for the binding of downstream signaling components TRIF and PI3K. However since TLR3 has no kinase activity, a protein tyrosine kinase needed for its phosphorylation may or may not be recruited along with the other downstream components.

Activated TLR3 activates the transcription factors Interferon Regulatory Factor 3

(IRF-3), Nuclear Factor kappa B (NFκB), and activator protein 1 (AP-1). While the mechanisms of activation for each factor are different, the adaptor protein TRIF (TIR domain-containing adaptor inducing IFNβ) is required for the induction of them all

(Hoebe, Du et al. 2003; Oshiumi, Matsumoto et al. 2003). TRIF belongs to a family of

TLR adaptor proteins of which there are three other members. MyD88 was initially found to mediate the signaling effects of LPS through TLR4, but has been shown to be involved in cytokine induction by most other TLRs (Medzhitov, Preston-Hurlburt et al.

20 1998). TRAM mediates the activation of IRF-3 from TLR4 (Fitzgerald, Rowe et al.

2003; Yamamoto, Sato et al. 2003), and TIRAP has also been shown to play a role in signaling through TLR4 (Fitzgerald, Palsson-McDermott et al. 2001; Horng, Barton et al.

2001). All four proteins contain a COOH-terminal TIR domain of their own, by which they interact with the TIR domains of the various TLRs. It appears that the role of the adaptor proteins is more of a scaffolding one, mediating the activation of downstream factors. The NFκB/AP-1 activation pathways diverge from IRF-3 activation pathway beyond TRIF (Sato, Sugiyama et al. 2003; Jiang, Mak et al. 2004).

The activation of IRF-3 by TLR3 requires both TRIF-dependent and TRIF- independent components. Either of the kinases TBK1 or IKKε (also known as IKK-i) is required for the direct phosphorylation and partial activation of IRF-3, and in their absence TLR3 signaling fails to induce IFN production (Sharma, tenOever et al. 2003;

Hemmi, Takeuchi et al. 2004). TRIF recruitment and activation of TBK1 has been shown to require TRAF3(Hacker, Redecke et al. 2006; Oganesyan, Saha et al. 2006). For complete transactivation of IRF-3 a second, TRIF independent, signal is also needed.

This is mediated by phosphatidylinositol 3-kinase (PI3K), which is recruited to Tyr759 in the TIR domain of TLR3. PI3K indirectly causes additional phosphorylation of IRF-3, which is required for its complete transcriptional activation.

There is conflicting data as to which proteins immediately downstream of TRIF are needed for NFκB activation. Either RIP1 or TRAF6 may be involved in signaling.

RIP1, previously shown to regulate NFκB activation by TNFα, was reported to interact with the COOH-terminal of TRIF and play a role in the activation of NFκB, but not IRF-

3 or MAP kinases (Meylan, Burns et al. 2004). TRAF6, alternatively, is reported to bind

21 to residues 250 to 255 of TRIF (Jiang, Mak et al. 2004). One report indicates TRAF6 deficiency impairs MyD88-dependent, but not TRIF-dependent activation of NFκB

(Gohda, Matsumura et al. 2004) in macrophages, although multiple other papers report

TRAF6 is needed for NFκB activation. At this point there is too much conflicting data to say for certain whether either or both RIP1 and TRAF6 are required for NFκB activation through TLR3.

Downstream of RIP1/TRAF6, signaling through TLR3 activates the protein complex of TAK1/TAB1/TAB2 (Jiang, Zamanian-Daryoush et al. 2003). Within this complex, TAK1 (TGFβ-activated kinase 1), a MAPKKK family member, was shown to be critical for both NFκB and JNK activation (Shim, Xiao et al. 2005). Activation of

TAK1 leads to its direct phosphorylation of IKKβ (IκB kinase β) within the IKK complex, thereby activating NFκB as discussed further below.

The AP-1 transcription complex is made up of the transcription factors ATF2 and c-Jun, which are activated by phosphorylation by the MAP kinases: p38 and JNK (c-jun

NH2-terminal kinase) (Karin 1995). In turn, TRAF6 was reported to be required for p38 and JNK activation through TLR3. Downstream of TRAF6, TAK1 appears to be the kinase which activates at least JNK. It is therefore likely that AP-1 pathway diverges from the NFκB pathway at the level of TRAF6/TAK1, and not upstream at TRIF, like

IRF-3.

In this whole process it is unclear if PI3K plays any role in NFκB or AP-1 induction through TLR3. It has been reported in the case of TLR2 that blocking PI3K signaling inhibits the phosphorylation of the p65 subunit of NFκB on Ser536, as well as blocking p38 and JNK activation, and cytokine release (Strassheim, Asehnoune et al.

22 2004). However, phosphorylation of Ser536 of p65 has yet to be linked to any transcriptional effect on gene induction.

CARD-helicases as mediators of dsRNA signaling. Interferon induction by cytoplasmic

dsRNA is mediated by members of the CARD-helicase family of proteins. The two

currently identified members known to effect dsRNA signaling are retinoic acid inducible

gene (RIG-I) and melanoma differentiation-associated gene 5 (mda5). Both of these

proteins share conserved domains for signaling and RNA binding.

The COOH terminus of both RIG-I and Mda5 share a common DExD/H-box

RNA helicase motif. The RNA helicase domain of RIG-I is the critical regulatory

domain of the protein, and its deletion results in a constitutively active mutant. RNA

helicases containing the DEAD box and DExD/H motifs are fairly common and well-

conserved between species. The name DEAD box was originally coined based on a

conserved motif (Linder, Lasko et al. 1989). DExD/H box proteins are likewise named,

with the x representing any . In addition to RIG-I and Mda5 other proteins that contain this domain include the translation initiation factor, eIF4α and the hepatitis C virus protein NS3 (Tanner and Linder 2001). However, genome wide analysis found only one protein, LGP2 which had significant overall homology to RIG-I and Mda5

(Yoneyama, Kikuchi et al. 2005). LGP2 lacks the common amino-terminus domain of the other two proteins, so is unable to signal, but can act as an inhibitor of their functions.

The amino terminals of RIG-I and Mda5 contain CARD, or caspase activated recruitment domains, which are critical for downstream signaling. CARD domains were initially characterized for their role in the recruitment of caspases, and have since been

23 shown to be important in gene induction by intracellular bacteria through the NOD

proteins (Hofmann, Bucher et al. 1997; Kobayashi, Inohara et al. 2002). Like TIR

domains, CARDs mediate homotypic interactions between proteins(Levy and Marie

2004). Deletion of the CARD domain of RIG-I results in a mutant that, similar to LGP2,

acts as a dominant negative.

The adaptor protein IPS-1 (interferon β promoter stimulator) is required for

downstream signaling by both RIG-I and Mda5 (Kawai, Takahashi et al. 2005). IPS-1,

which is also known as MAVS, VISA, and CARDIF, binds to RIG-I and Mda5 through a

CARD within its amino terminus (Seth, Sun et al. 2005). Downstream activation of IRF-

3 through IPS-1 has been shown to be TBK1/IKKε-dependent. Conversely, NFκB

activation though IPS-1 is not well understood. While NFκB activation is IPS-1-

dependent, it may or may not also require RIP1 (Cusson-Hermance, Khurana et al. 2005).

Overexpression of IPS-1 leads to phosphorylation of ATF2 by JNK, indicating a role for

the adaptor in MAP kinase activation as well. Interestingly, the COOH terminus of IPS-1

is important to anchor the protein on the outer mitochondrial membrane. Deletion of the

region required for anchoring to the mitochondria also destroys the ability of IPS-1 to

signal (Seth, Sun et al. 2005).

Activation of NFκB.

The NFκB family of transcription factors is made up of five members, p65 (-A), rel-B,

c-rel, NFκB1 (p50/p105) and NFκB2 (p52/p100). All share conserved 300 amino acid

regions involved in DNA binding known as Rel homology domains (RHDs) (Hayden and

Ghosh 2004). However, only the three rel proteins have transactivation domains and can

24 signal as transcription factors. Classic NFκB signaling is mediated by a heterodimer of p65/p50, although p65 homodimers and other combinations can signal as well. The heterodimer is usually bound by an Inhibitor of kappa B (IκB), which prevents signaling.

Activation of NFκB is mediated by the IκB kinase (IKK) complex, which is composed of

IKKα, IKKβ along with the regulatory subunit IKKγ. Phosphorylation of IκB by IKKβ leads to its subsequent degradation and the release of p65/p50(DiDonato, Hayakawa et al.

1997).

Mechanisms regulating NFκB signaling following its release from IκB are many and not as well understood. After its release from IκB, NFκB translocates to the nucleus, where it recognizes a consensus binding site of GG(A/G)NN(C/T)(C/T)CC. In order for

NFκB to induce transcription, the p65 subunit must undergo further phosphorylation.

There is evidence of at least four different sites on p65 that can be phosphorylated.

While the importance of the individual phosphorylation sites on the transcriptional activation of p65 are controversial, it is safe to say that phosphorylation is important for gene induction (Chen and Greene 2004; Hayden and Ghosh 2004).

In addition, p65/p50 heterodimers and p65 homodimers have been shown to differentially induce transcription. Different IκB isoforms, alpha, beta and epsilon, along with IKK family members are likely responsible for regulating the type of NFκB complex activated. IκBα preferentially binds p65/p50 heterodimers, and IKKβ preferentially phosphorylates IκBα. Consequently differential activation of IKKs could alter the profile of genes induced by NFκB. Using computational analysis, such a model has been predicted for NFκB mediated gene induction by LPS and TNFα (Hoffmann,

Levchenko et al. 2002; Werner, Barken et al. 2005).

25 One of the most confusing aspects of NFκB signaling during viral infection is its

role in apoptosis or its prevention. A p65-/- mutation in mice is normally embryonic lethal. However generating such a mouse is possible in either a TNFα-/- or TNFα receptor (TNFR)-/- background, thus revealing p65 is required to prevent TNFα mediated apoptosis (Alcamo, Mizgerd et al. 2001). This would suggest that the activation of p65 would be favorable to viral replication as it would prevent apoptosis of the infected cell. However, p65-/-, TNFR-/- mice are more susceptible to virus infection than

TNFR-/- mice indicating this is not the case.

Unlike p65, a p50-/- mutation is not lethal and has no effect on apoptosis, but does lead to decreased responsiveness to LPS (Sha, Liou et al. 1995). If such an effect holds true for dsRNA signaling remains to be seen.

Activation of IRF-3.

The interferon regulatory factor family is comprised of ten different members. Of these, three including IRF-3, IRF-1, and IRF-9 have been shown thus far to function solely as inducers of transcription. IRFs 2, 4, 5, 7 and 10 have all have been shown to function as both inducers and repressors of transcription to some degree (Nehyba,

Hrdlickova et al. 2002; Barnes, Richards et al. 2004). Finally, IRF-8 has been reported to function only as a repressor (Barnes, Richards et al. 2004). Along with IRF-3, IRFs 5 and 7 have been implicated mediating gene induction by TLRs and RIG-I and Mda5.

However while both of these latter two IRFs are expressed at low levels in non-lymphoid cells and can be induced through IFN signaling, IRF-3 is expressed in all cell types, and is not inducible (Lowther, Moore et al. 1999).

26 Common among all IRFs is an amino terminus DNA binding domain which is

distinguished by five tryptophan repeats. It is this region that is critical for the

recognition of the DNA binding sequences of GAAA and AANNNGAA (Escalante, Yie

et al. 1998). In addition to the amino terminus binding domain, IRF-3 also has a COOH

terminus domain known as the IRF association domain (IAD). This region is thought to

be responsible to the interaction with other IRF proteins, as well as containing the

transactivation domain need for interaction with CBP/p300 and other factors required for transcriptional induction. The very end of IRF-3 (a.a. 382-414) is an auto-inhibitory domain, which prevents the constitutive activation of IRF-3 (Akira, Uematsu et al. 2006).

Inactive IRF-3 is present in the cell as a monomer residing mainly in the cytoplasm. However IRF-3 contains both a nuclear export signal (NES) at amino acids

141 to 147 and a nuclear localization sequence (NLS) around amino acids 77-78 of the amino terminus (Reich 2002). Inhibition of nuclear export by leptomycin B causes IRF-3 to accumulate in the nucleus. This indicates that while it continuously shuttles between the nucleus and cytosol, the dominant NES keeps the vast majority inactive IRF-3 localized principally in the cytoplasm.

Activation of IRF-3 involves phosphorylation within its COOH terminal auto- inhibitory domain. TBK1/IKKε is required for this phosphorylation to occur in response to signaling through both TLR3 and RIG-I/Mda5 (Fitzgerald, Rowe et al. 2003; Sharma, tenOever et al. 2003; McWhirter, Fitzgerald et al. 2004). There is some controversy over the specific phosphorylation sites required for transactivation of IRF-3.

Alternatively, serines 385/386 and serines 396/398 have been reported to be needed for

IRF-3 activation, and TBK1 appears to phosphorylate at least one residue in each group

27 by in vitro studies. In addition, serines 402 and 405 along with threonine 404 have also been implicated as potential phosphorylation sites of IRF-3. The crystal structure of the

COOH-terminus of IRF-3, including the IAD, predicts IRF-3 to have a structural fold which would be altered by phosphorylation, allowing IRF-3 assume a more open and presumably active configuration (Qin, Liu et al. 2003; Takahasi, Suzuki et al. 2003).

Following phosphorylation by TBK1, IRF-3 is able to dimerize and translocate to the nucleus. However phosphorylation through the TLR3-TRIF-TRAF3-TBK1 axis alone is not enough for gene induction through IRF-3. Additional phosphorylation of

IRF-3 is required along the TLR3-PI3K-AKT axis (Sarkar, Peters et al. 2004).

Unfortunately, it is not clear at present if the PI3K-dependent axis is also required for transactivation of IRF-3 by RIG-I or Mda5, or how and where IRF-3 is phosphorylated through this pathway.

Other IRFs mediating signaling.

In addition to IRF-3, IRF-7 and IRF-5 appear to be important in determining the profile of genes induced by dsRNA or during viral infection. Although all three IRFs can induce many of the same genes, they appear to mediate very distinct overall mRNA expression profiles (Barnes, Field et al. 2003; Barnes, Richards et al. 2004;

Schoenemeyer, Barnes et al. 2005). It is reported that IRF-5 signaling strongly induces genes involved in the immune response and adhesion, while IRF-7 signaling induces select mitochondrial genes as well as many involved in DNA repair and structure

(Barnes, Richards et al. 2004).

28 Distinctions between the gene induction profiles of different IRFs were first appreciated in the context of type I IFNs. Activation of IRF-3 by dsRNA leads to the production of IFNβ and IFNα4, while other isoforms of IFNα fail to get induced directly through IRF-3. However, autocrine IFNβ signaling induces the expression of IRF-7 which can in turn induce other IFNα isoforms (Marie, Durbin et al. 1998).

Adding further to the complexity of dsRNA signaling is the differential expression of signaling factors between cell types. For example, in mouse embryonic fibroblasts, IFNβ treatment is needed for the induction of IFNα by SeV. However, in leukocytes SeV can induce IFNα genes independent of IFNβ treatment (Erlandsson,

Blumenthal et al. 1998). The difference is that lymphocytes naturally express IRF-7 while MEFs do not. Such variable expression of IRFs has led to considerable debate over whether IRF-3 or IRF-7 is the more important immune effector (Honda, Yanai et al.

2005).

29 Role of dsRNA PRRs in viral infection.

It is unwise to assign too much faith to any one observation about dsRNA

signaling pathways at present time since so much of the field is in flux. There are a

couple of conclusions safe to make. 1. Signaling through the RIG-I, Mda5 and TLR3 signaling pathways elicit distinct responses despite activating many of the same

transcription factors. 2. dsRNA in different cellular compartments activates different

signaling pathways. 3. Secondary autocrine signaling plays a substantial, but indirect role in the profile of genes induced through either NFκB or IRF-3. 4. Genes induced by

dsRNA signaling have diverse roles, many beneficial to limiting viral infection. 5.

Viruses have evolved different methods of inhibiting dsRNA signaling.

Recognition of viral dsRNA. TLR3 plays a much less important role in the recognition of viral dsRNA than originally thought. No effect has been seen in the absence of TLR3 in murine cytomegalovirus (MCMV), vesicular stomatitis virus (VSV), lymphocytic choriomeningitis virus (LCMV), or reovirus infections (Edelmann, Richardson-Burns et al. 2004). As discussed within this dissertation, the same is true of Sendai virus (SeV) a negative single-stranded RNA paramyxovirus. While all work in cell culture indicates that adding dsRNA to the media induces a TLR3 dependent response, it has now been indicated that Mda5-deficient mice do not respond when injected with intravenously with

poly(I:C), raising questions as to the importance of TLR3 in the recognition of even

extracellular dsRNA.

Where TLR3 does seem important is in regulating the response of immune cells,

particularly in modulating the adaptive response. Dendritic cells deficient in TLR3

30 cannot efficiently cross prime a cytotoxic T cell response when presented with encephalomyocarditis virus (EMCV)-infected cells to phagocytose (Schulz, Diebold et al.

2005). Also in dendritic cells, TLR3 could work synergistically with TLRs 7, 8 or 9 to facilitate Th1 T-cell response (Napolitani, Rinaldi et al. 2005). Genetic silencing of

TLR3 by siRNA led to decreased cytokine induction in response to respiratory syncytial virus (RSV) in human lung fibroblasts and epithelial cells (Rudd, Burstein et al. 2005).

Finally, infection of TLR3-/- mice with RSV led to pathologic increases in mucus production within airways due to an overly Th2-polarized immune response (Rudd, Smit et al. 2006).

Other TLRs play roles in viral signaling. TLR9 recognizes the DNA viruses,

MCMV, HSV1 and HSV2. TLR2 mediates a cellular response to HSV as well, although it appears to detrimentally lead to lethal encephalitis (Kurt-Jones, Chan et al. 2004).

Despite these findings, the majority of RNA viruses that have been studied appear to signal through either RIG-I or Mda5 instead of TLRs. RIG-I is the critical cellular receptor in mice for signaling in response to infection with SeV, Newcastle Disease

Virus, and Japanese Encephalitis virus (Kato, Takeuchi et al. 2006). In the case of SeV this may be because the viral V proteins interact with and inhibit Mda5, but not RIG-I

(Andrejeva, Childs et al. 2004). EMCV infection conversely required Mda5 and not

RIG-I for cytokine induction.

It is unclear how important the method of viral PAMP recognition is to the overall efficacy of the innate immune response, since there is a high degree of crosstalk between the different pathways at multiple levels of signaling. Aside from TLR3, many TLRs utilize the same adaptor protein, MyD88, to signal. As mentioned previously, the same is

31 true for RIG-I and Mda5, which both signal via IPS-1. At the level, both TLR and CARD proteins induce an NFκB signaling response. RIG-I, Mda5 and

TLR3 all activate IRF-3 in a TBK1/IKKε dependent manner. Additional crosstalk occurs at the promoter level, as discussed previously, where IRF-3, IRF-7, and ISGF3 can all induce common genes through ISREs.

The functions of many of the genes induced through both TLR3 and RIG-I have been found to directly or indirectly assist in the anti-viral immune response. including IFNβ, TNFα, IL-8, and IL-6 among others can modulate both the innate and adaptive immune responses. Signaling by dsRNA also up-regulates many genes directly involved in the immune response. These include class I MHC molecules, which are involved in antigen presentation and the adaptive immune response, and mannose binding lectin 2, which is involved in opsinization and the activation of complement, as part of the innate immune response (Eisen and Minchinton 2003).

Many genes with the potential to limit replication within infected cells are also induced by dsRNA and type I IFNs. These include the well-studied Mx proteins, PKR, and 2’-5’ Oligoadenylate synthetase (Silverman 1994; Williams 2001; Haller and Kochs

2002). In recent years the anti-viral properties of many other genes have come to .

ISG56 encodes a protein which binds to the eIF3 complex and thereby prevents both viral and host protein translation (Hui, Bhasker et al. 2003). Cig5, also known as Viperin, is an endoplasmic reticulum-associated protein which has been shown to block human cytomegalovirus (HCMV) replication (Chin and Cresswell 2001). Up-regulation of

NOXA in virus-infected cells leads to apoptosis, thus potentially limiting infection (Sun and Leaman 2005).

32 Given the bevy of anti-viral proteins induced in response to viral dsRNA, it is not

surprising that many viruses have developed means of inhibiting dsRNA signaling.

Inhibition, which can occur at all levels of the dsRNA signaling cascade, is often

mediated by viral non-structural proteins (Haller, Kochs et al. 2006). At the level of the

dsRNA receptor, some paramyxovirus V-proteins can bind to and block Mda5 signaling

(Andrejeva, Childs et al. 2004). Moving downstream, the Hepatitis C viral protease,

NS3/4A, blocks RIG-I and TLR3-mediated signaling, by cleaving the adaptor proteins

IPS-1 and TRIF, respectively (Foy, Li et al. 2005; Li, Foy et al. 2005; Li, Sun et al.

2005). Many viral proteins target gene induction specifically by IRF-3. The P-proteins

of Rabies virus and Borna Disease virus have both been shown to inhibit the

phosphorylation of IRF-3 by TBK-1 (Brzozka, Finke et al. 2005; Unterstab, Ludwig et al.

2005). Another method to block IRF-3 signaling is employed by Human Herpes virus

8, which expresses a viral homolog of IRF-1 that effectively serves as an IRF-3 dominant

negative (Lin, Genin et al. 2001). Viral modulation of NFκB signaling also occurs,

however it is not exclusively inhibitory. It appears that, in the case of some latent

viruses, NFκB signaling is beneficial to prolonged survival and infection. In addition to

inhibiting dsRNA signaling, many viral inhibitors of IFN signaling have also been identified. These serve to limit the secondary IFN-induced response.

Questions addressed in this thesis. A number of different questions about the nature of gene induction by dsRNA are addressed in this dissertation.

Chapter 3 concerns the long-standing question of the nature of the dsRNA signaling defect in the STAT1-null cell line, U3A. In particular the relationship between

33 cellular responsiveness to dsRNA and STAT1 was addressed. STAT1 was found to be required for IFN-induced TLR3 expression and dsRNA-responsiveness, but not for basal

TLR3 expression or dsRNA-responsiveness. This work has already been published (Elco and Sen 2007).

While many dsRNA-induced genes have previously been identified, their method of induction is by no means uniform. Chapters 4 and 5 concern gene induction profiling of the different transcription factors and signaling pathways involved in dsRNA signaling. Chapter 4 details the importance of PI3K, IRF-3, and NFκB in TLR3 signaling, while chapter 5 concerns the role of NFκB, IRF-3 and secondary IFN signaling in SeV infection. The work in chapter 5 has already been published (Elco, Guenther et al. 2005).

One of the most interesting findings described in chapter 5 was that IRF-3 could impair the expression of select viral-induced genes. Chapter 6 addresses the mechanism behind how IRF-3 repressed the induction of these genes.

34 CHAPTER 2

MATERIALS AND METHODS

Cell culture and generation of cell lines. Cell lines used in this dissertation were cultured in DMEM with 10% FBS, 2mM L-Glutamine, 50 units/ml penicillin, and

50µg/ml streptomycin. Table 2-1 lists the origins and characteristics of individual human cell lines. Cell lines are listed in sub-groupings underneath their parental line, as appropriate. The generation of 2fTGH, 2C4 and lines derived from them by mutagenesis have been described previously (Pellegrini, John et al. 1989; Watling, Guschin et al.

1993). Creation of P2.1 cells was described by Leaman, et al (Leaman, Salvekar et al.

1998). The generation of the IRF-3 overexpressing cell lines U4C.2, p2.1.17, p2.1.6, along with 2F-SR cells, which express the dominant negative IκBα super repressor was described by Peters, et al (Peters, Smith et al. 2002). U3A cells expressing STAT1 mutants Y701F and S727A were originally obtained from Jim Darnell (Bandyopadhyay,

Leonard et al. 1995). Wt11 (293/TLR3) cells were derived from 293 cells transfected with a TLR3 expression vector as described previously and cultured in 400 µg/ml G418

(Sarkar, Smith et al. 2003).

To make the STAT1-silenced GRE line, GRE s.5, GRE cells were transfected with a siRNA construct containing U6 promoters driving the expression of two hairpin siRNAs targeting STAT1 in the pcDNA3 backbone. Clones were selected by culture in

800 µg/ml of G418, then screened by western blot for decreased STAT1 expression.

1080.10 cells are a clonal line derived from HT1080 cells co-transfected with the IRF-3 expression vector, pCDNA3/hIRF-3 and the selection vector pBABE/Puro. Following selection with puromycin, 1080.10 and other clones were picked from the stably

35 transfected cells and screened by Western blotting for increased IRF-3 expression. The

IRF-3 overexpressing cell lines 2F-SR.3, z3P and 333p were made similarly by Dr. Kristi

Peters. U4CRIG-IC cells were cloned from U4C cells co-transfected with an expression plasmid for RIG-IC (Yoneyama, Kikuchi et al. 2004) along with pBABE Puro.

Following selection with puromycin, individual clones were screened for loss of p56 induction by SeV. HT-siPKR cells were cloned from HT1080 cells transfected with a siRNA vector against PKR (from Maryam Zamanian-Daryoush) and pcDNA6 followed by selection with blastomycin. U3AR+ and U3ATLR3 cells were derived as pools from

U3A cells infected with lentiviral expression vectors pLV-STAT1-puro and pLV-TLR3- puro then selected with puromycin.

Cell treatment protocols.

Inhibitors. To assess the role of PI3K, cells were pre-treated with LY 294002 at 50µM for 0.5 hours before treatment with dsRNA. The inhibitor was left on throughout treatment. Inhibition of translation by cycloheximide was begun 0.5 hours before dsRNA treatment at a concentration of 50µg/ml. To look at nuclear accumulation of proteins, cells were treated with leptomycin B at a dilution of 1:1000 for 30 minutes

dsRNA, IFN and TNFα treatment: Cell treatment with dsRNA was done using poly(I)•poly(C) from Amersham prepared as previously described (Peters, Smith et al.

2002). For extracellular dsRNA treatment, poly(I)•poly(C) was added directly to the

culture media to a final concentration of 100 µg/ml for six hours unless otherwise noted.

To study stimulation by intracellular dsRNA, poly(I)•poly(C) was transfected using Fu-

36 TABLE 2-1. Description of cell lines used in this dissertation Method of Affected signaling Cell Line Altered protein(s) alteration pathway(s)

HT1080 (origin: fibrosarcoma) None wild type HT-siPKR PKR siRNA silencing PKR 1080.10 IRF-3 increased expression IRF-3

2fTGH None selection system wild type 2F-SR IkB dominant negative NFκB 2F-SR.3 IRF-3 increased expression NFκB, IRF-3

U3A STAT1, TLR3 mutagenesis IFN, TLR3 U3AR+ STAT1 restored expression TLR3 U3ATLR3 TLR3 restored expression IFN 701 Y701F STAT1 expression of mutant IFN, TLR3 727 S727A STAT1 expression of mutant IFN, TLR3

U3B STAT1 mutagenesis IFN U5A IFNAR2 mutagenesis type I IFN R2C wt IFNAR2 restored expression wild type

2C4 None selection system wild type U3C STAT1 mutagenesis IFN U4C Jak1 mutagenesis IFN U4CRIGIC RIG-I dominant negative IFN, RIG-I U4C.2 IRF-3 increased expression IFN, IRF-3 P2.1 IRF-3 mutagenesis IFN, IRF-3 P2.1.6 IRF-3 increased expression IFN, IRF-3 P2.1.17 IRF-3 increased expression IFN, IRF-3

293 (origin: embyonic kidneys cells) None wild type z3p IRF-3 increased expression IRF-3 wt11 (293/TLR3) TLR3 increased expression TLR3 333p IRF-3 increased expression IRF-3, TLR3

GRE (origin: glioma) IFN deletion mutation wild type* GRE s.5 STAT1 siRNA silencing IFN

MRC5 (origin: lung fibroblasts) None wild type

*GRE cells can respond to, but not produce IFN

37 gene at a final concentration of 3 µg/ml of media. For comparison of poly(I)•poly(C)

and poly(I)•poly(C)-LC, the final concentration of dsRNA added to the media was

10µg/ml. Human interferon β was added directly to media at a final concentration of

1000 U/ml. Human interferon-γ (Peprotech) was used likewise at a final concentration of

1 ng/ml. TNFα treatment was done at a final concentration of 20ng/ml.

Viral infections and plaque assay. Sendai virus infection was carried out as previously described by Heylbroeck, et al (Heylbroeck, Balachandran et al. 2000). Culture media was removed from cells and replaced with serum-free DMEM. Sendai virus (Charles

River, SPAFAS, Cantell strain) was added directly to the media to a final concentration of 1HAU/4.0x103 cells (~10-15 pfu/cell) and incubated for 1 hour with gentle shaking every 10 minutes. Following viral adsorption, the media was removed and replaced with

DMEM containing 10% FBS for the remainder of the experiment. Infection of cells with

Influenza B virus (provided by Robert Krug), was carried out as described for Sendai virus, with the exception that cells were incubated at 35 degrees C during and after viral adsorption.

Cells were infected with EMCV at an MOI of 1.0 or 0.01 in serum-free media and incubated for two hours with gentle shaking every fifteen minutes. Virus media was then replaced with media containing 10% serum. Following incubation, plates were flash frozen once to lyse the cells, then supernatant was then collected and used to infect a confluent monolayer of L929 cells. Following viral adsorption, L929 cells were overlayed with 1X MEM containing 10% FBS, 100 units/ml Pen/Strep and 1% low melting point agarose, this was allowed to solidify a room temperature, then incubated

38 twenty-four hours at 37 degrees C. At this point an additional agarose layer, containing neutral red, was placed on the cells and they were incubated until plaques could be visualized and counted.

siRNA treatment. Gene expression was silenced by the transfection of 100nM siRNA using Oligofectamine (Invitrogen). Cells were used for experiments 48 hours after transfection of siRNA. Short interfering for TRIF and MyD88 have been described previously and were ordered as custom siRNAs (Ambion) (Oshiumi,

Matsumoto et al. 2003). TLR3 was silenced using siGENOME SMARTpool reagent from Dharmacon. Finally, negative control siRNA #1 (Ambion), was used as a control in all siRNA experiments.

Lentiviral preparation and infection. To generate the TLR3 lentiviral construct, pLV-

TLR3-puro, human TLR3 with an N-terminal flag tag (Sarkar, Smith et al. 2003) was cloned into the lentiviral vector pLV-Puro-LoxP-TetO, which was generously provided by A.D. Singhi and A.V. Gudkov. Similarly TLR3 with the TIR domain deleted was used to generate the TLR3 dominant negative construct, pLV-TLR3∆TIR-puro.

STAT1α was sub-cloned from STAT1 alpha pRc/CMV (from Jim Darnell) into lentivirus to make pLV-STAT1-puro.

Recombinant virus was made similar to as described previously (Gurova, Hill et al. 2004). Briefly, 293T cells were co-transfected with lentiviral vector and packaging plasmids encoding structural proteins (pCMV∆8.2) and the G-protein of vesicular

39 stomatitis virus (pVSV-G) using fu-GENE. Media containing virus was collected after

48 hr and used in a 1:1 ratio with complete DMEM to infect U3A cells in the presence of

10µg/ml Polybrene for twelve hours.

Macrophage preparation. Primary bone marrow derived macrophages were isolated

similar to as described previously(Whitmore, DeVeer et al. 2004). Briefly, bone marrow

was collected from the femurs of 129, type I IFN- /ßR-/-, and STAT1-/- mice.

Macrophages were allowed to differentiate in culture for seven or ten days by incubating in RPMI media supplemented with 20% serum, 20% L-cell conditioned media, and 50 units/ml Pen/Strep. L-cell conditioned media was prepared from L929 cells and filtered prior to use. It served as the source of the macrophage colony-stimulating factor, needed to stimulate macrophage differentiation.

RNA expression analysis. For all experiments requiring RNA, total cellular RNA was first isolated from using the RNA-Bee (Tel-Test, Friendswood, TX).

RNAse protection assays. All RNAse protection assays were performed using the

RPA III kit (Ambion, Austin, TX). The actin, A20 and ISG56 RPA probes have been described previously (Elco, Guenther et al. 2005). The RPA probe for ISG54 corresponds to the final 270 base pairs of the coding sequence and was made by cutting cDNA with PvuII. A 200 probe was likewise made for ISG60 using AseI to cut the cDNA. The A20, Follistatin and NOXA RPA probes were made from their respective cDNA microarray constructs and correspond to the final 220, 164, or 158

40 nucleotides of each mRNA. The measurement of anti-apoptotic genes was done using

the hAPO-5 Multi-Probe Template set from BD PharMingen. Gene induction was

measured using a Storm Imager and quantified with Image Quant 5.2 (Molecular

Dynamics).

Northern Blotting. Northern blotting was performed as described previously(Geiss, Jin et al. 2001). The full length cDNA was used as a probe for OAS p69.

Reverse Transcription and quantitative real-time PCR. For real-time PCR, all RNA samples were first purified further using the DNA-free kit (Ambion). cDNA was made using 1µg of purified total RNA by the Superscript III First Strand Synthesis System

(Invitrogen). For amplification the primer sets were used can be found in table 2-2.

Real-time quantitative PCR data was generated using SYBR Green PCR core reagents

(Applied Biosystems) and an ABI Prism 7700 Sequence Detector. Relative TLR3 and

ISG56 mRNA expression was determined by normalizing to RPL32 by the Comparative

CT Method.

cDNA microarray analysis cDNA Array construction. The array used in this study comprised a subset of sequence verified cDNA clones from the Research Genetics 40K set. The clones included 950 genes containing adenylate/uridylate rich elements and 18 genes potentially involved in

AU-directed mRNA decay as previously described (Frevel, Bakheet et al. 2003), 855

interferon stimulated genes representing an expansion of a previously described clone set

41 TABLE 2-2: Sequences of real-time PCR primers used Gene Name Sequence

TLR3 (503-720) SN 5’-GTCAAGCAGAAGAATTTAATCAC-3’ AS 5’-CACCCTGGAGAAAACTCTTT-3’

TLR7 SN 5-ACGACAATGACATCTCTTCCTCC-3' AS 5'-GTTTGTGGCTGAGGTCC-3'

TLR8 SN 5'-TGCAGCCTGGGAAAGGAGAC-3' AS 5'-TGCTGTACATTGGGGTTGTGG-3'

TLR9 SN 5'-CAGCATGGGTTTCTGCCGC-3' AS 5'-TCCACTTGAGGTTGAGATGCC-3'

TRIF SN 5'-ACCATCCTCGGCTTTGC-3' AS 5'-TCATCTCCTCAGGCTCC-3'

RIG-I SN 5'-CTGGACCCTACCTACATCC-3' AS 5'-AATCCCAACTTTCAATGGCTTC-3' mda5 SN 5'-ATGGGAAGTGATTCAG-3' AS 5'-TCAGTTGGGTATCACC-3' cig5 SN 5'-TGCCCAGCGTGAGCATCGTG-3' AS 5'-CCAGCGGACAGGGTTTAGTGC-3'

IL-8 (107-293) SN 5'-TTCCAAGCTGGCCGTGGCTC-3' AS 5'-TGTGTTGGCGCAGTGTGGTCC-3'

ISG56 (6-354) SN 5’-TCTCAGAGGAGCCTGGCTAAG-3’ AS 5’-GTCACCAGACTCCTCACATTTGC-3’ hRPL32 SN 5’-GCCAGATCTTGATGCCCAAC-3’ AS 5’-CGTGCACATGAGCTGCCTAC-3’.

42 (de Veer, Holko et al. 2001), 288 genes responsive to the viral analog poly (I).poly(C),

representing an expansion of the clone set described by Geiss et.al.(Geiss, Jin et al. 2001) and 85 housekeeping genes. DNA preparation and slide printing were as previously described (Frevel, Bakheet et al. 2003) except for the use of 50% DMSO as printing solution.

RNA labeling and array hybridization. Cy3 or Cy5 labeled cDNA was prepared by direct incorporation. 100 µg of total RNA, 2 µg dT12-18 primer (Invitrogen), and 1 µl anti-

RNAse (Ambion) were combined in a reaction volume of 30 µl and incubated for 10 min

at 70οC. Reverse transcription was for 2hr at 42οC in a 50 µl reaction containing annealed

RNA template, 1X 1st strand buffer (Invitrogen), 500 µM each dATP, dCTP, dGTP,

300µM dTTP, 250 µM dUTP (Sigma), 3nmol Cy3-dUTP or Cy5-dUTP, 10mM DTT,

and 6U/µl Superscript II. For template hydrolysis, 10 µl 0.1M NaOH was added to the reverse transcription reaction and the mixture was incubated for 10 mins at 70οC, allowed

to cool at room temperature for 5 min and neutralized by addition of 10 µl 0.1M HCl.

The cDNA was purified with GFX columns following the manufacturer’s instructions

(Amersham Pharmacia Biotech), dried down, and resuspended in hybridization buffer

containing 2X SSC, 0.1% SDS, 4 µg of poly(dA)40-60, and 4 µg of yeast tRNA. The

Cy3 and Cy5-labeled cDNAs were pooled and after 2 min denaturation at 90οC the hybridization mixture was applied to the microarray slide under a coverslip.

Hybridization proceeded overnight in a sealed moist chamber in a 55οC waterbath. Post- hybridization, slides were washed successively for 5 min each in 2 X SSC plus 0.1% SDS

43 at 55οC, 2 X SSC at 55οC, and 0.2 X SSC at room temperature, spun dry, and scanned on

a GenePix 4000A scanner (Axon).

Data acquisition and normalization. Data were acquired with a GenePix 4000A laser scanner and GenePix Pro 5.0 software as previously described(Frevel, Bakheet et al.

2003). Raw data were imported into Genespring 6.0 software (Silicon Genetics) and normalized based on the distribution of all values with locally weighted linear regression

(LOWESS) before further analysis.

For the microarray data shown in chapters three and four, a single microarray experiment was run for each treatment in a given cell line. To assure reproducibility, the exceptions were dsRNA treatment of p2.1 and p2.1.17 cells, where two array experiments were run for each cell line, and 2fTGH cells where four separate microarray experiments were run. For the data shown in chapter five, three microarray experiments, using different biological repeats and one dye swap, were performed for every cell line infected

with SeV. For assessing the effects of dsRNA, two microarray experiments, using

different biological samples were conducted in both the 293 and 293/TLR3 cell lines.

Spots representing genes with fluorescent intensities of less than 200 units in the SeV- or

dsRNA-treated samples were excluded when surveying for changes in expression. A

gene was considered to be induced by SeV or dsRNA in a particular cell type if its

expression in treated cells was at least two-fold greater than its expression in uninfected

cells in at least two microarray experiments.

44 TABLE 2-3: Description of antibodies used in this dissertation Antibody Host Type Dilution Reference

A20 Mouse Monoclonal 1:1000 Imgenex, IMG-161 actin Rabbit Polyclonal 1:1000

IRF-1 Rabbit Polyclonal Cell Signaling, #4966

IRF-3 Rabbit Polyclonal 1:10,000 Michael David

NFκB p65, CT Rabbit Polyclonal 1:5000 Upstate, #06-418 Phospho-NF-κB p65 (Ser536) Rabbit Polyclonal 1:2000 Cell Signaling, #3031 Phospho-NF-kappaB p65 (Ser276) Rabbit Polyclonal 1:2000 Cell Signaling, #3037 p56 Rabbit Polyclonal 1:5000 p76 Rabbit Polyclonal 1:10,000

PKR Rabbit Polyclonal 1:5000 Santa Cruz, K-17

STAT1, CT Rabbit Polyclonal 1:5000 Upstate, #06-501 STAT1, phospho-specific (Tyr701) Rabbit Polyclonal 1:5000 Upstate, #07-307 STAT1, phospho-specific (Ser727) Rabbit Polyclonal 1:5000 Upstate, #07-714

TLR3 Mouse Monoclonal 1:1000 Imgenex

45 Protein Techniques.

Western Analysis. The lysis buffer described by Leaman et al (Leaman, Salvekar et al.

1998) was used to prepare protein lysates from both nuclei and whole cells. Lysates were electrophoresed on a 10% SDS-PAGE gel. For immunoblotting analysis, a list of the primary antibodies used in this dissertation, along with their source may be found in table

2-3. HRP conjugated goat-anti rabbit and goat-anti-mouse secondary antibodies were used and visualized by ECL.

Electrophoretic Mobility Shift Assay (EMSA). EMSAs were performed using whole cell lysates prepared in lysis buffer formulated as stated previously and analyzed according to the method described by Sizemore et al (Sizemore, Leung et al. 1999). As a probe, the

κB site from the promoter of the IP-10 gene was used (5’-

GAGCAGAGGGAAATTCCGTAACTT-3’).

Nuclear Fractionation. IRF-3 activation was monitored by measuring IRF-3 levels in the nuclear fractions of cells. For this, cells were washed and collected in PBS. Cells were then resuspended and allowed to incubate in hypotonic buffer (20 mM Tris-HCl, pH 8.0, containing 4 mM MgCl2, 6 mM, CaCl2, and 0.5 mM DTT) for five minutes, as described

by Guo et al (Guo, Stacey et al. 2002). An equal volume of lysis buffer (0.6 M sucrose,

0.2% Nondiet p-40, and 0.5 mM DTT) was added and nuclei were released by 20 strokes with a dounce homogenizer. Nuclei were then pelleted at 1,500g for 8 minutes by centrifugation at 4 °C. and washed once in glycerol wash buffer (50 mM Tris-Cl, pH 8.3,

5 mM MgCl2, 0.1 mM EDTA and 40% (v/v) glycerol), once with PBS, and then resuspended in western lysis buffer for analysis by western blot.

46 Two-dimensional gel analysis. To analyze IRF-3 by two-dimensional gel electrophoresis,

either whole cell lysate (untreated) or nuclear fractions (SeV infected and dsRNA treated)

were used. Nuclear fractions were isolated in the same way described above, except

following the glycerol wash buffer, the nuclei pellet was washed once with water and

then lysed in 0.75 ml of rehydration buffer (8 M urea, 2% (w/v) CHAPS, trace

bromophenol blue). Lysates from untreated cells were made by washing twice with PBS

and directly lysing the cells in 1−2 ml rehydration buffer on the plate. Both nuclear and whole cell lysates were sonicated for 5 seconds, and spun down at 16,000g for 10 min to remove debris. Lysates were then separated in the first dimension by isoelectric focusing

on Immobiline dry strips, pH 4−7 (Amersham). Focused proteins were separated in the second dimension by SDS-PAGE and western blotted for IRF-3.

Native PAGE. IRF-3 dimerization was determined by native PAGE (Sarkar, Peters et al.

2004). One hour after treatment with dsRNA, cells were lysed in native lysis buffer (75

mM NaCl, 50 mM Tris-Cl, pH 7.5, 1 mM EDTA, 1% (v/v) NP40, 12.5 mM β-

glycerophosphate, and 1x protease inhibitor (complete EDTA-free protease inhibitor

tablets, Roche Molecular Biochemicals). Samples were mixed with loading buffer (final

concentration: 0.125 mM Tris-Cl, pH 6.8, 20% (v/v) glycerol and 0.1 mg ml-1

bromophenol blue) and electrophoresed by native PAGE.

Immunofluorescence. IRF-3 was visualized by immunofluorescence exactly as described

previously by Peters et al. (Peters, Smith et al. 2002). Cells plated on a glass coverslips

were washed twice with PBS then fixed in 4% paraformaldehyde for 30 minutes at room

47 temperature. After three additional PBS washes, cells were permeabilized in PBS containing 0.2% triton X-100 for 15 minutes at room temperature, then blocked (PBS with 0.02% tween 20, 3% BSA, 3% Goat serum, and 4% glycine) overnight at 4 deg C.

Coverslips were probed for 1 hour at room temperature with IRF-3 antibody diluted

1:10,000 and Alexa 488 conjugated-goat anti-rabbit secondary antibody diluted 1:1500 in

PBS containing 0.02% tween and 3% BSA. Finally coverslips were mounted on slides with vectashield plus DAPI and viewed by fluorescent microscopy.

48 CHAPTER 3

ROLE FOR STAT1 IN INTERFERON-INDUCIBLE, NOT CONSTITUTIVE

RESPONSIVENESS TO ADDED dsRNA

ABSTRACT

Distinct, but partially overlapping signaling pathways mediate the response to

extracellular versus intracellular sources of dsRNA, by TLR3 and RIG-I, respectively.

Different cell types signal through these pathways to widely varying degrees. It was

previously observed that exposure to extracellular dsRNA, delivered by its addition to the

culture media, could induce ISG56 in HT1080 fibroblasts, but not the HT1080-derived

cell line, U3A, which lacks functional STAT1. In this study, the nature of the dsRNA

signaling defect in U3A cells is further investigated. A defect affecting basal TLR3

mRNA expression prevents U3A cells from responding to extracellular dsRNA. This

defect does not impair dsRNA signaling in response to viral infection or transfected

dsRNA. Although U3A cells are deficient in STAT1, STAT1 was not required for basal

TLR3 expression because other cell lines lacking STAT1 expressed TLR3. Moreover,

restoration of STAT1 expression failed to restore TLR3 mRNA expression in U3A cells.

However, treatment of STAT1-restored U3A cells with either interferon beta or gamma

induced TLR3 expression and restored responsiveness to extracellular dsRNA.

Interestingly, despite indications that EMCV and Influenza B viruses might activate

TLR3, neither basal nor induced responsiveness to extracellular dsRNA impaired the replication of EMCV or significantly altered the profile of genes induced by Influenza B

infection. Together the results presented here demonstrate that STAT1 is critical for

interferon-induced, not basal responsiveness to extracellular dsRNA.

49 INTRODUCTION

Double-stranded (ds)RNA is an important signal for the activation of many

signaling pathways and subsequent gene induction (Elco, Guenther et al. 2005). In virus- infected cells, recognition of viral dsRNA is critical for the induction of many genes involved directly in the inhibition of viral replication as well as the initiation of the immune response (Yoneyama, Kikuchi et al. 2004). Recognition of dsRNA is not limited to within the cytoplasm of virus infected cells. Many cell types can also recognize extracellular dsRNA. The ability to respond to phagocytosed dsRNA is critical for the cross-priming of dendritic cells by virus infected cells (Schulz, Diebold et al. 2005).

However the responsiveness to extracellular dsRNA is not a universal trait, rather there is

considerable variation between cell types, with phagocytes such as dendritic cells and macrophages, being the most responsive (Matsumoto, Funami et al. 2003). Little is known at present about the mechanisms controlling responsiveness to extracellular

dsRNA.

Distinct pathways exist for the detection of and response to both intracellular and

extracellular sources of dsRNA. Gene induction by intracellular dsRNA signaling is

mediated by cytoplasmic RNA-helicases RIG-I (Yoneyama, Kikuchi et al. 2004) and

Mda5 (Andrejeva, Childs et al. 2004), while the only known mediator of extracellular

dsRNA signaling is Toll-like receptor 3 (TLR3) (Alexopoulou, Holt et al. 2001). Both

signaling pathways converge to activate a common set of transcription factors, IRF-3 and

NFκB. As a result intracellular and extracellular dsRNA induce many of the same genes

50 despite large differences in upstream signaling pathway components (Elco, Guenther et

al. 2005).

IRF-3 normally resides as an inactive monomer in the cytoplasm, however upon

activation by phosphorylation, it dimerizes and translocates to the nucleus where it

induces gene transcription through an IFN-stimulated response element (ISRE) present

within the promoter. Type I IFNs also induce genes through the same core ISRE

sequence (GAAANNGAAA), although signaling is mediated through ISGF3, a

transcriptional complex composed of STAT1, STAT2, and IRF-9, a family member of

IRF-3. As a consequence of variations in sequence flanking the ISRE, not all ISGF3

inducible genes are equally induced through IRF-3 (Bandyopadhyay, Leonard et al.

1995). There is still large overlap in the number of ISRE-containing genes induced by

dsRNA and IFN. As a result both signals evoke many of the same anti-viral responses

during infection.

In the context of the overall immune response dsRNA and type I IFNs are not

redundant in their roles, despite inducing many of the same genes. Gene induction by

dsRNA occurs immediately but in a localized population. Meanwhile, interferon signals in paracrine, autocrine, and endocrine fashions to upregulate local as well as distal tissue responses, but has to first be induced itself. A remarkable amount of interplay has developed between the two pathways. dsRNA induces IFN-β, thereby indirectly activating IFN signaling, while type I IFNs can augment or even confer responsiveness to dsRNA by up-regulating TLR3, RIG-I and Mda5 expression.

Since IFNs and dsRNA are so intimately linked, both in inducing a common set of genes and in regulating each other, we sought to determine if type I IFNs or any of the

51 components involved in its signaling were also important for dsRNA signaling.

Previously the Sen lab examined the ability of extracellular dsRNA to induce ISG56 mRNA in five mutant cell lines defective in different components of type I IFN signaling, along with their parental line, 2fTGH (Bandyopadhyay, Leonard et al. 1995). ISG56 encodes one of the messages most strongly induced by both dsRNA and type I interferons. Regulation of human ISG56 is mediated through an ISRE responsive to both

ISGF3 and IRF-3. In the mutant cell lines, dsRNA induced ISG56 effectively in U1A,

U2A, U4A, and U6A cells, which lack Tyk2, IRF-9, Jak1, or STAT2, respectively.

However in U3A cells, which lack STAT1, extracellular dsRNA was unable to induce

ISG56. This finding suggested STAT1, but not any other components of the IFN signaling pathway might be required for the induction of at least a subset of genes by dsRNA. In this paper the role of STAT1 in gene induction by dsRNA as well as the nature of the dsRNA signaling defect in U3A cells is further characterized.

U3A cells are unable to respond to extracellular dsRNA due to a defect in basal

TLR3 expression. This defect affects extracellular dsRNA signaling only, and is independent of the presence or absence of functional STAT1 seen in U3A cells, but is due to another unknown mutation. Conversely, STAT1 is critical for not only type I but also type II interferon-induced responsiveness to extracellular dsRNA. Further, extracellular dsRNA responsiveness induced by IFN correlates with increased TLR3 expression. We suggest that distinct transcriptional mechanisms control basal (STAT1- independent) and IFN-induced (STAT1-dependent) responsiveness to extracellular dsRNA through the regulation of TLR3 expression.

52 RESULTS

Differential induction of ISG56 mRNA in U3A cells by extracellular dsRNA or transfected dsRNA. ISG56 is highly induced by dsRNA through an interferon-stimulated response element (ISRE) within its promoter. Previously, we had observed that dsRNA added to the cell culture media poorly induced ISG56 in the chemically mutagenized cell line, U3A (Bandyopadhyay, Leonard et al. 1995). Conversely in 2fTGH cells, the parental line of U3A, extracellular delivery of dsRNA strongly induced ISG56. Because intracellular and extracellular dsRNA signal through partially overlapping pathways, we have now examined whether transfected dsRNA was also unable to induce ISG56 in U3A cells. As shown in Fig. 3-1, when dsRNA was transfected into U3A cells it was able to induce ISG56 at levels equivalent to those seen in 2fTGH cells, whereas dsRNA added to the culture medium was effective only in 2fTGH cells. Treatment of cells with ssRNA or its transfection failed to induce ISG56 in either cell line, demonstrating that the observed response in U3A cells was specific to dsRNA. These results showed that U3A cells retained normal responsiveness to transfected dsRNA.

The repertoire of genes induced defectively in response to extracellular dsRNA. We next sought to determine the breadth of the defect in gene induction by extracellular dsRNA in

U3A cells. Three cell lines were used for this purpose, the parental 2fTGH cells and the derived mutant lines U3A and U4C both of which are unresponsive to interferons because the U3A cells are missing functional STAT1 and the U4C cells are missing functional Jak1. In the first experiment, we used a quantitative RPA to measure the levels

53

Figure 3-1. Differential induction of ISG56 by extracellular dsRNA and transfected dsRNA. ISG56 mRNA expression in U3A and 2fTGH cells left untreated (lanes 1 and

7), treated with either dsRNA (lanes 2 and 8) or ssRNA (lanes 3 and 9) by addition directly to the media, or transfection with dsRNA (lanes 5 and 10) or ssRNA (lanes 6 and

12). ISG56 and actin (control) mRNA levels were measured by RNAse protection assay

(RPA) six hours after treatment.

54 added transfected added transfected 1723456 8 910 1112

ISG56 mRNA

Actin mRNA

U3A 2fTGH

55 of mRNA expression of ISG56 and two related genes ISG54 and ISG60 in the three cell

lines in response to dsRNA added to the culture medium or produced intracellularly by

infection with Sendai virus. As shown in Fig. 3-2A, like ISG56, neither ISG54 nor

ISG60 was induced by adding dsRNA to the media in U3A cells but all three genes were

induced in 2fTGH and U4C cells. In contrast, intracellular dsRNA produced by SeV

infection could induce ISG54, ISG56 and ISG60 mRNAs in all three lines, including

U3A. Transcription of the above ISGs is mediated by IRF-3 activated by dsRNA-

treatment of cells. To determine whether dsRNA could induce genes through

transcription factors other than IRF-3 in U3A cells, we looked at the NFκB-regulated

gene A20. Adding dsRNA to the media also failed to induce the A20 mRNA in U3A

cells (Fig. 3-2B).

The inability of added dsRNA to induce either IRF-3 or NFκB regulated genes in

U3A cells suggested a global defect in extracellular dsRNA signaling. This suspicion was validated by transcription profiling using a custom cDNA microarray including hundreds of known dsRNA-induced genes (Geiss, Jin et al. 2001). Using this cDNA microarray, we measured the changes in mRNA expression induced by adding dsRNA to the media and SeV-infection in 2fTGH and U3A cells. There was hardly any gene whose transcription was induced or repressed by dsRNA in U3A cells, although many were affected in the parental line (Fig. 2C). SeV, as expected, induced cellular gene expression equally well in U3A cells and 2fTGH cells. These results established that

U3A cells have a global defect in extracellular dsRNA signaling.

56

Figure 3-2. Cellular gene induction by extracellular dsRNA and SeV in U3A cells.

Changes in mRNA expression were measured in 2fTGH, U4C and U3A cells six hours after Sendai virus (SeV) infection or treatment with ex-dsRNA. RNase protection assays

(RPAs) showing (A) ISG54, ISG56 and ISG60 mRNA expression or (B) A20 mRNA expression, following SeV infection or ex-dsRNA treatment. (C) Global changes in mRNA levels following ex-dsRNA treatment (upper panels) or SeV infection (lower panels) in 2ftGH cells (left) or U3A cells (right) as determined by cDNA microarray experiment. Points on the scatter plot represent the relative expression of individual mRNA messages in untreated cells (x-axis) plotted against their expression six hours after SeV infection or ex-dsRNA treatment (y-axis). The central diagonal line denotes equal expression in treated and untreated samples, while two-fold differences in expression are indicated by the two flanking diagonal lines.

57 (A)

Unprotected +SeV +dsRNA +SeV +dsRNA +SeV +dsRNA (B) ISG54 mRNA - + - + dsRNA

A20 mRNA

ISG60 mRNA

ISG56 mRNA Actin mRNA 2fTGH U3A

Actin mRNA

2fTGHU4C U3A

(C)

10000 10000

dsRNA

1000 1000

100 100

100 1000 10000 100 1000 10000

10000 10000

SeV

1000 1000

100 100

100 1000 10000 100 1000 10000 2fTGH U3A

58 STAT1 is not required for dsRNA signaling. U3A cells have a defect in STAT1

expression and consequently in IFN signaling (McKendry, John et al. 1991). Type I and type II IFN responses can be restored in these cells by complementing them with STAT1.

STAT1 might also play a role in extracellular dsRNA signaling. Using phospho-specific antibodies against both major STAT1 phosphorylation sites we found that extracellular

dsRNA treatment did not cause Tyr701 phosphorylation nor increase basal Ser727

phosphorylation (Fig. 3-3A and B) in IFNAR2 knockout U5A cells, or in IFN receptor

knockout mouse macrophages. However in U5A cells with IFNAR2 restored (R2C) or

wild-type macrophages, phosphorylation of both sites was seen following dsRNA

treatment, probably because of signaling by IFNs produced as intermediates. This

suggested that these sites were not important for dsRNA signaling. Consistent with this,

U3A cells containing STAT1 with either a Y701F or S727A mutation failed to restore

ISG56 induction by dsRNA (Fig. 3-3C). This was not specific to ISG56, as shown in

figure 3-3C, expression of the Y701F mutant in U3A cells failed to restore any dsRNA

signaling by cDNA microarray analysis. Since it was possible that both Ser 727 and

Tyr701 of STAT1 might have some structural importance independent of their

phosphorylation, we looked at the ability of wild-type STAT1 to restore dsRNA

signaling. Figure 3-3E shows that the protein product of ISG56, p56, was induced by

added dsRNA in 2fTGH cells, but not w.t. STAT1-restored U3AR+ cells, although IFN-

β could induce the protein in both cell lines. To further verify that the U3AR+ cells

could respond normally, we looked at IRF-1 induction by IFNγ. As can be seen in figure

3-3F, IRF-1 was induced even better in U3AR+ cells, than in wild-type cells. The

inability of STAT1 to restore dsRNA-induced p56 expression in U3A cells, coupled with

59

Figure 3-3 STAT1 cannot restore dsRNA responsiveness in U3A cells. (A) Western

blot of STAT1 phosphorylation on Tyr 701 (upper panel) or Ser 727 (lower panel)

following dsRNA treatment. IFNAR2-/- cells (U5A) and cells with IFNAR2 expression

restored (IFNAR2+, R2C) were treated with dsRNA added to the culture media for the

number of hours indicated. (B) STAT1 phosphorylation by added dsRNA was also

assessed in bone marrow derived macrophages from wild-type (129) and IFN receptor

knockout mice (IFNAR KO) (C) RPA analysis of dsRNA induced ISG56 mRNA

expression in STAT1 mutants. RNA isolated from wild-type cells (lane 1), STAT1

deficient U3A cells (lane 2), or U3A cells expressing the STAT1 mutants, Y701F and

S727A (lanes 3 and 4) six hours after addition of dsRNA was assessed for ISG56

expression. (D) Gene induction by added dsRNA in U3A cells expressing Y701F

mutant STAT1 was assessed by microarray analysis as described in fig. 3-2C. (E)

Expression of p56 (upper panel) and STAT1 (lower panel) in 2fTGH and U3AR+ (w.t.

STAT1-restored U3A) cells was analyzed by western blot either without treatment (lanes

1 and 4) or following addition of dsRNA (lanes 2 and 5) or IFNβ (lanes 3 and 6). (F)

IRF-1 expression was similarly measured after treatment with IFNγ.

60 (A) (B)

IFNAR2- IFNAR2+ 129 IFNAR KO 036036 0 1350135 pY-701 pY-701

pS-727 pS-727 Actin

(C) (D) 12 34

10000 ISG56 mRNA

dsRNA 1000

actin mRNA

100

10 10 100 1000 10000 untreated Y701F

(E) (F) 1 234 56 _ _ IFNγ + + p56 IRF-1

STAT1 Actin

2fTGH U3AR+ U3AR+ 2fTGH

61

Figure 3-4. STAT1 not required for extracellular dsRNA signaling. (A) STAT1

protein expression in U4C, U3A, U3B, U3C and STAT1 restored U3A cells (U3AR+)

was assessed by immunoblot with anti-STAT1 Ab. (B) ISG56 mRNA expression

following the addition of dsRNA to the culture media was determined by RPA. (C)

STAT1 protein expression was determined by western blot in GRE cells and a clone with

STAT1 expression stably knocked down by siRNA (GRE s.5). (D) ISG56 mRNA expression in GRE and GRE s.5 cells left untreated (lanes 1 and 4), following the addition of dsRNA (lanes 2 and 5) or IFN β (lanes 3 and 6).

62 (A) U4CU3A U3B U3C U3AR+ STAT1

Actin

(B) _ _ _ _ dsRNA ISG56 mRNA

Actin mRNA

U3AU3AR+ U3B U3C

(C) GRE GRE s.5 STAT1 actin

(D) 123456

ISG56 mRNA

Actin mRNA

GRE GRE s.5

63 the lack of direct STAT1 phosphorylation by dsRNA suggested that the observed defect

in gene induction by extracellular dsRNA in U3A cells was not due to the absence of

STAT1.

To further verify that STAT1 was not required for dsRNA signaling, we looked at

other cell types without STAT1. U3B and U3C cells, like U3A cells, do not express

STAT1 (Fig. 3-4A), however when treated with added dsRNA, ISG56 mRNA was

induced in both cell lines. By contrast no ISG56 mRNA expression was seen in dsRNA treated U3A or U3AR+ cells. Similarly, stable knock-down of STAT1 by siRNA in GRE cells (Fig. 3-4C) impaired ISG56 induction by IFNβ, but failed to alter its induction by added dsRNA (Fig. 3-4D).

Impaired activation of IRF-3 and NFκB. Activation of the two major transcription factors, IRF-3 and NFκB in dsRNA-treated cells, occurs in multiple steps. Therefore, the observed block in gene induction by extracellular dsRNA, in U3A cells, could be caused by either a total or a partial defect in transcription factor activation pathways.

To distinguish between these possibilities, we first examined the IRF-3 activation pathway. Little IRF-3 was retained in the nuclei of extracellular dsRNA-treated U3A cells, as measured by immunofluorescence (Fig. 3-5A) or Western blotting (Fig. 3-5B).

As expected, in contrast, ample IRF-3 was present in the nuclei of 2fTGH cells after

adding dsRNA to the media, or in the nuclei of both U3A and 2fTGH cells following

SeV infection. An activation step that precedes IRF-3 translocation to the nucleus is its

dimerization in the cytoplasm. No dimeric IRF-3 was present in the cytoplasmic extract

of dsRNA-treated U3A cells (Fig. 3-5C). Basal phosphorylation of IRF-3 was likewise

64

Figure 3-5. Impaired activation of IRF-3 and NFκB in U3A cells. (A) Sub-cellular

localization of IRF-3 was determined by immunofluorescence in untreated 2fTGH and

U3A cells and following the addition of dsRNA or infection with SeV. (B) Levels of

IRF-3 in nuclear lysates made from 2fGH and U3A cells were assessed by western blot.

Expression of the nuclear protein p76 is shown as a loading control. (C) IRF-3

dimerization in untreated 2fTGH cells (lane 1) and 2fTGH and U3A cells after adding

dsRNA (lanes 2 and 3) was assessed by native gel. The lower band is unactivated,

monomeric IRF-3 while the higher band is active dimerized protein. (D) Basal

phosphorylation of IRF-3 in HT1080 and U3A cells was assessed by two-dimensional gel

electrophoresis followed by blotting for IRF-3. (E) IRF-3 localization is shown in U3A

cells by immunofluorescence following leptomycin B treatment. (F) NFκB activation

was assayed by EMSA using lysates from ex-dsRNA treated and untreated 2fTGH and

U3A cells.

65 (A) Untreated +dsRNA 1.5 hr +SeV 1.5 hr

2fTGH

U3A

(B)2fTGH U3A (C) -+---+ 1 23 IRF-3 dimer p76 monomer 2fTGH U3A

(D) (E)

HT1080

U3A

U3A+leptomycin B

(F) 2fTGH U3A - + - + dsRNA

NFκB

66 unaltered in U3A cells as shown by 2-dimension gel-electrophoresis in figure 3-5D.

Finally, inactive IRF-3 still shuttles in and out of the nucleus without being retained.

IRF-3 accumulated in the nucleus of U3A cells treated with leptomycin B, an inhibitor of nuclear export, indicating that nuclear import and export of IRF-3 can occur normally

(Fig 3-5E). Together our results indicated an upstream defect in the pathway leading to

IRF-3 activation, not a defect in IRF-3 itself.

A similar early defect in the NFκB pathway was detected by the electrophoretic mobility shift assay that measures the release of NFκB from the IκB inhibitory complex

(Fig. 3-5E). Thus, the signaling defect in the U3A cells was probably at a step that preceded the bifurcation of the IRF-3 and the NFκB pathways.

Proteins needed for dsRNA-signaling. To understand further the nature of the defect in the U3A cells, we investigated the previously uncharacterized pathways of dsRNA- signaling in the HT1080 cells and the derived U series of cell lines. In the next series of experiments, we explored the possible involvement of a number of proteins that are known to participate in various dsRNA-signaling pathways. In cells, these proteins were functionally ablated by expressing the cognate siRNAs or dominant-negative mutants and the ability of dsRNA added to the media to induce ISG56 mRNA was monitored. As expected, expression of a dominant-negative mutant of the intracellular dsRNA receptor,

RIG-I completely blocked ISG56 induction by SeV infection, but its induction in response to extracellular dsRNA was unaffected (Fig. 3-6A) indicating that RIG-I is not a component of the pathway of our interest. The same was true for PKR, whose ablation by siRNA, did not diminish ISG56 induction by dsRNA in these cells, and actually

67

Figure 3-6. RIG-I and PKR not necessary for extracellular dsRNA signaling. . (A)

The ability of a dominant negative RIG-I mutant to block expression of ISG56 mRNA in response to added dsRNA or SeV was determined by RPA. U4C cells (odd lanes) and

U4C cells expressing RIG-IC, a RIG-I dominant negative mutant (even lanes), were stimulated by adding dsRNA or infecting with virus as indicated. (B) PKR expression in

HT1080 cells and HT-siPKR cells was compared by western blotting with anti-PKR Ab.

HT-siPKR cells were generated from HT1080 cells by the stable transfection of an siRNA construct against PKR. (C ISG56 mRNA expression in HT1080 and HT-siPKR cells was compared by RPA. (D) Nuclear lysates both from untreated cells and following dsRNA treatment were western blotted for IRF-3 expression. (E) A20 mRNA expression was similarly assessed by RPA.

68 (A) 123456

ISG56 mRNA

actin mRNA

Untreated dsRNA SeV

(B) (C) __ HT1080 HT-siPKR ddsRNA ISG56 mRNA STAT1

PKR actin mRNA

HT1080 HT-siPKR

(D)_ _ (E) _ _ + + + + dsRNAd IRF-3 A20 mRNA

p76

actin mRNA HT1080 HT-siPKR HT1080 HT-siPKR

69 augmented IRF-3 nuclear translocation as well as A20 mRNA induction (Fig. 3-6B-E). In contrast, ablation of the adaptor protein TRIF, but not MyD88, severely affected dsRNA- signaling (Fig. 3-7A). Because the upstream partner of the TRIF-mediated dsRNA- signaling pathway is TLR3, we assayed for its presence by Western blotting, but could not detect the protein, probably because of its low level. However, the corresponding mRNA was detectable and its ablation by a specific siRNA (Fig. 3-7B, left panel) caused a severe inhibition of induction of ISG56 mRNA in U3B cells, as measured by quantitative RT-PCR (Fig. 3-7B, right panel) or RPA (Fig. 3-7C). These results clearly indicated that, in these cells, the TLR3-TRIF pathway was responsible for mediating the signals generated by exogenous dsRNA.

Expression of TLR3 in U3A cells. Knowing that extracellular dsRNA signaling was mediated by the TLR3 pathway in U3B cells, we wanted to identify if TLR3 or TRIF, as well as other nucleic acid receptor proteins were defective in U3A cells. Because the levels of poorly expressed proteins, like TLR3 could not be accurately measured by western blot, mRNA expression was determined utilizing quantitative real-time PCR

(Fig. 3-8A). Expression of mRNA for toll-like receptors TLR7, TLR8, and TLR9, along with RNA helicases, RIG-I and mda5 was always comparable to or greater in U3A cells than it was in U4C cells. Equivalent levels of TRIF mRNA were also observed in U3A,

U4C and HT1080 cells. In contrast, U3A cells expressed only one-thirtieth the amount of

TLR3 mRNA expressed in U4C cells and one-fiftieth of that seen in HT1080 cells. As expected U3B and U3C cells expressed TLR3 mRNA at levels within two-fold of those seen in U4C cells, while U3AR+ cells expressed little more TLR3 mRNA than what was

70

Figure 3-7. TLR3 and TRIF, but not MyD88 required for extracellular dsRNA signaling. (A) ISG56 mRNA expression in U3B cells with TRIF and MyD88 expression silenced by siRNA was measured by RPA. U3B cells were transfected with non-specific siRNA (siCtrl) or siRNA specific to MyD88 (siMyD88) or TRIF (siTRIF) were treated by adding dsRNA to the media for six hours where indicated. (B) The effect of silencing

TLR3 expression by siRNA on added dsRNA signaling in U3B cells was examined by real-time quantitative PCR. U3B cells were transfected as described above, but with siRNA specific to TLR3 (siTLR3). Relative expression of TLR3 mRNA (left) and

ISG56 mRNA (right) in individual samples was determined by normalizing to RPL32 expression. Data shown represent the average of three experiments. (C) ISG56 mRNA expression in U3B cells with TLR3 expression silenced by siRNA was also assessed with and without ex-dsRNA treatment by RPA

71 (A) _ _ _ + + + dsRNA ISG56 mRNA

actin mRNA

siCtrl siMyD88 siTRIF

(B) 1.0 25

20 0.75

15

0.5

10

0.25 5 fold ISG56 mRNA induction fold ISG56 mRNA relative TLR3 mRNA expression TLR3 mRNA relative

0.0 0 siCtrl siTLR3 siCtrl siCtrl+ds siTLR3 siTLR3+ds

(C) __

ISG56 mRNA

siCtrl siTLR3

72 seen in U3A cells (Fig. 3-8B). Thus, the ability of these cell lines to respond to dsRNA corresponded perfectly with their TLR3 mRNA expression. As further proof we found

TLR3 mRNA expressed in both w.t. and STAT1 knock-out bone-marrow derived mouse macrophages.

These results suggested that U3A cells could not respond to dsRNA due to a lack of expression of TLR3. To address this possibility, we created a pool of U3A cells expressing exogenous TLR3 (Fig. 3-8C, right panel). TLR3 expression alone was sufficient to confer responsiveness in U3A cells to added dsRNA (Fig. 3-7C, left panel), indicating that TRIF and other proteins required for signaling downstream ofTLR3 were functional in U3A cells. The induction of ISG56 mRNA in U3ATLR3 cells was stronger than that in U4C cells, probably because of its greater level of TLR3 expression.

Restoration of dsRNA-response by IFN-treatment. The lack of response of U3A cells to dsRNA was reminiscent of earlier observations with HeLa-M cells. Tiwari et al. reported that these cells were unresponsive to added dsRNA, but they could respond well if primed by pretreatment with IFNβ or IFNγ (Τιωαρι, Κυσαρι ετ αλ. 1987). Tiwari hypothesized that IFNs induced the synthesis of a protein needed for signaling by dsRNA. After its discovery, TLR3 became an attractive candidate because it is an IFN- inducible protein. Indeed, TLR3 mRNA level was very low in HeLa-M cells; but the level increased highly after treatment with IFNγ (Fig. 3-9A), thus conferring sensitivity to added dsRNA.

To examine whether the same was true for U3A cells, they were treated with

IFNβ or IFNγ. As anticipated, these treatments caused the induction of TLR3 mRNA in

73

Figure 3-8. Defective TLR3 mRNA expression in U3A cells. (A) Real time

quantitative PCR was run using RNA isolated in duplicate from HT1080, U3A and U4C

cells. Relative expression of RIG-I, mda5, TRIF and TLRs 7-9 as well as TLR3 was

determined by normalizing to RPL32 expression. (B) Relative expression of TLR3

mRNA in unstimulated U4C, U3A, U3AR+, U3B and U3C cells was compared in

triplicate by real-time quantitative RT-PCR and normalized to RPL32 mRNA expression.

(C) RT-PCR for mouse TLR3 was run on RNA isolated from bone marrow-derived

macrophages from both wt 129 mice and STAT1-/- mice. (D) Expression of exogenous

TLR3 in U3A cells restores ex-dsRNA signaling. Left panel: TLR3 protein levels in

U3A cells with and without exogenous TLR3 expressed by lentivirus (U3ATLR3) were measured by immunoblotting with anti-TLR3 Ab, with anti-Actin Ab as a control. Right panel: RPA analysis of ISG56 mRNA levels in both untreated and ex-dsRNA treated

U3ATLR3 and U4C cells.

74 (A) TLR3 TLR7 TLR8 TLR9 RIG-I Mda-5 TRIF 60 5 6 2.4 4 6 4

50 2.0 5 4 5 3 3 40 4 1.6 4 3

30 3 1.2 2 3 2 2 20 2 0.8 2 1 1 1 10 1 0.4 1 relative mRNA expression mRNA relative

0 0 0 0.0 0 0 0

U3A U4C U3A U4C U3A U4C U3A U4C U3A U4C U3A U4C U3A U4C HT1080 HT1080 HT1080 HT1080 HT1080 HT1080 HT1080

(B) 1.4 (C) 1.2 wt 129 1.0 STAT1-/- mTLR3 0.8 0.6 0.4 0.2 relative TLR3 mRNA expression mRNA TLR3 relative 0 U4C U3A U3AR+ U3B U3C

(D) U4C U3ATLR3 U3A U3ATLR3 -+-+dsRNA TLR3 ISG56 mRNA

Actin Actin mRNA

75 U3AR+ cells but not U3A cells (Fig. 3-9A). The functional consequence of IFNγ

pretreatment was measured by measuring ISG56 mRNA induction (Fig. 3-9B). Unlike

naïve U3AR+ cells, the pretreated cells responded strongly to dsRNA. To monitor the

effects of IFNβ pretreatment, we needed a different assay because ISG56 is directly induced by IFNβ. Instead, we examined nuclear translocation of IRF-3 as an index for effective dsRNA-signaling. After IFNβ treatment of U3AR+ cells, but not of U3A cells, dsRNA caused nuclear translocation of IRF-3 as measured by Western blotting (Fig. 3-

9C) or immunofluorescence (Fig. 3-9D). These results indicate that both U3A and HeLa-

M cells are unresponsive to dsRNA because they cannot express TLR3 constitutively; but their pretreatment with IFN induces TLR3 expression and confers responsiveness to dsRNA added to the media.

Since IFN-primed TLR3 is functional in U3AR+ cells, the mutation that prevents the unprimed cells from responding to extracellular dsRNA lies in the expression of

TLR3. To determine why TLR3 is not constitutively expressed in U3A cells, we began by analyzing the promoter. It has previously been reported that the promoter of TLR3 contains an ISRE, as well as a putative GAS site (Heinz, Haehnel et al. 2003). Both sites

in the TLR3 promoter were unmutated in U3A cells. Additional sequencing of the TLR3

promoter to 1.5 kb upstream of the transcription start site revealed absolutely no

differences between the promoters of the TLR3 gene in U3A and 2fTGH cells or the

consensus sequence from NCBI. This indicated that U3A cells either had a defect in the

basal transcriptional machinery for TLR3 or the mRNA message was rapidly degraded

following transcription.

76

Figure 3-9. Induction of dsRNA responsiveness by interferon pretreatment. (A)

Induction of TLR3 mRNA by both IFN-β and IFN-γ was determined by real-time

quantitative PCR. U3A, U3AR+, and HelaM cells were left untreated, or treated with

either IFN-β or IFN-γ for 24 hr, at which time fold-increase in TLR3 mRNA expression

was determined. (B) ISG56 mRNA levels were assessed by RPA in U3A and U3AR+ cells pre-treated with IFN-γ for 16 hours. Cells were then left untreated or treated with ex-dsRNA for an additional six hours. (C) Nuclear translocation of IRF-3 in IFN-β pre- treated cells following stimulation with ex-dsRNA was examined by immunoblot.

Nuclear fractions from U3A and U3AR+ cells pre-treated with IFN-β, as well as untreated U3B cells were prepared both with and without ex-dsRNA stimulation and immunoblotted with anti-IRF-3 Ab and anti-P76 Ab as a loading control. (D) Activation of IRF-3 by nuclear translocation was assessed by immunofluorescence with anti-IRF-3

Ab. U3AR+ cells pre-treated with IFN-β were either left untreated or treated with ex-

dsRNA treatment for 1.5 hr.

77 (A)

20

16

12

8

4 fold TLR3 mRNA induction TLR3 mRNA fold 0 ctrlβγγ ctrl β ctrl γ U3A U3AR+ Hela M

(B) +IFN-γ pre-treat (C) _ dsRNA _ IFN-β pre-treat__ dsRNA ISG56 mRNA IRF-3

p76 actin mRNA U3A U3AR+ U3B U3A U3AR+

(D) no dsRNA +dsRNA, 1.5 hr

78 Effect of basal ex-dsRNA responsiveness on viral infection. To determine the effect of basal responsiveness to extracellular dsRNA on encephalomyocarditis virus (EMCV) replication, we compared U3A and U3B cells. By comparing U3A and U3B cells we were able to study the importance of basal TLR3 expression independently from induced expression. Further, since both cell lines lack STAT1, the effects of secondary IFN signaling are absent. U3A and U3B cells were infected with EMCV either at an MOI of

1.0 for 8 hours, or an MOI of 0.01 for 24 hours after which viral yield was measured by plaque assay. The short infection at high MOI was meant to study the role of the extracellular dsRNA signaling in preventing viral replication within infected cells. In contrast, prolonged infection at a low MOI was meant to determine if ex-dsRNA signaling could impair viral replication by mediating a paracrine immune response within a population where IFN signaling was neutralized. As can be seen in Fig.3-10, EMCV replicated over 50% more efficiently in U3A cells than U3B cells at both high and low

MOIs. However, EMCV replicated as efficiently in U3ATLR3 cells as it did in U3A cells, which suggested that differences in U3A and U3B cells, other than basal TLR3 expression levels, accounted for the variation in EMCV replication seen between these lines.

Influenza A virus has been reported to signal through TLR3 (Guillot, Le Goffic et al. 2005). To see if the same was true for Influenza B virus, we examined the induction of ISG56 mRNA in both wt11 cells, which express high levels of exogenous TLR3, and their parental line, HEK293, six hours after infection with Influenza B. Unlike extracellular dsRNA which failed to induce ISG56 in the absence of TLR3, Influenza B infection induced ISG56 in both 293 and wt11 cells (Fig. 3-11A). Quantification

79

Figure 3-10. Effect of TLR3 on EMCV replication. Replication of EMCV in U3A,

U3B and U3A cells expressing exogenous TLR3 (U3ATLR3) was assessed by plaque assay. Cells were either infected at an MOI of 1.0 for 8 hours or an MOI of 0.01 for 24 hours, after which time supernatants were collected. Plaque assays were performed to determine viral titer using mouse L929 cells infected with the supernatants.

80 8.E+07

6.E+07

4.E+07 PFU/ml

2.E+07

0.E+00 U3A U3B U3ATLR3 U3A U3B U3ATLR3 MOI 1.0, 8 hr MOI 0.01, 24 hr

81 revealed that ISG56 expression following influenza infection was 1.6-fold as strong in

wt11 compared with 293 cells, while ISG56 induction by dsRNA increased 128-fold.

This suggested that if Influenza B could induce ISG56 through TLR3 it was not very

efficient.

In contrast to TLR3 signaling, IFN signaling appeared to be important for strong

induction of ISG56 by influenza. We measured ISG56 mRNA expression at various

times after influenza B infection in U5A and R2C cells. Like U3A cells, U5A cells were

derived from 2fTGH cells and cannot respond to type I IFNs. However, this is because

U5A cells lack IFNAR2, not STAT1. R2C cells are U5A cells which have had their IFN responsiveness restored by expression of exogenous IFNAR2. As can be seen in figure

3-11B, although influenza B induces ISG56 expression in the absence of IFN signaling, its expression at later time points is much higher when the IFN signaling pathway is restored.

As would be expected from our results with 293 and wt11 cells, when compared with U5A cells, U3A cells expressed equivalent levels of ISG56 mRNA following influenza infection (Fig. 3-11C). This further supports our finding that TLR3 is not required for ISG56 induction by influenza.

To determine if Influenza B could likewise induce other genes independently of

TLR3, we compared gene induction profiles between U3A and 2fTGH cells following

Influenza B infection. Genes were considered to be induced by Influenza B if their expression increased over 2.5-fold relative to untreated cells. By this criteria a total of

199 genes were identified as Influenza B induced in either 2fTGH or U3A cells.

Influenza B was able to induce 193 of these genes in 2fTGH cells. In U3A cells,

82

Figure 3-11. TLR3-independent ISG56 induction by influenza B. (A) ISG56 mRNA expression in 293 and wt11 cells as determined by RPA. Cells were either left untreated

(lanes 1 and 4) or treated with added dsRNA (lanes 2 and 5) or influenza B (lanes 3 and

6). (B) ISG56mRNA expression in U5A cells (IFNAR2-/-) and R2C cells

(U5A+IFNAR2) was measured by RPA at the indicated times following infection with influenza B. Expression of ISG56 is shown here normalized to actin. (C) RPA of

ISG56mRNA expression in U5A and U3A cells after infection with influenza B.

83 (A) 123456

ISG56 mRNA

actin mRNA

293 Wt11

(B) 200 U5A R2C

100 ISG56 expression

0 02468 hours

(C)

ISG56 mRNA

actin mRNA

U5A U 3A

84 Influenza B induced 53 genes of which 47 were also induced by Influenza B in 2fTGH

cells (Fig. 3-12A). To determine which of these genes might be getting induced through

TLR3 or RIG-I, we next looked at the induction of the 199 Influenza B induced genes by dsRNA signaling via TLR3 as well as SeV. Surprisingly, the majority of Influenza B- induced genes failed to get induced by either extracellular dsRNA or SeV at 6 hours.

None of the six genes we considered induced by Influenza B in U3A cells but not 2fTGH cells were strongly up-regulated by infection with SeV. Only two were also induced by

SeV in U3A cells, while the remaining four genes were not induced by SeV or dsRNA in either U3A or 2fTGH cells. Of the 146 genes induced by Influenza B only in 2fTGH, only 18 were induced by other stimuli, with 8 being induced by dsRNA and 16 being induced by SeV. An example of some of these genes, including NOXA and cytochrome c-1 are shown in figure 3-12B. The microarrays identified only two genes, Ifi 6-16 and an est similar to KIAA0638, induced in 2fTGH cells by dsRNA and Influenza B, but not by SeV nor by any stimuli in U3A cells. However, it is already known that Ifi 6-16 induction by dsRNA is indirect and mediated by IFN, making it likely that these genes were not induced in U3A cells due to the absence of STAT1 and not TLR3.

Of the 46 TLR3/STAT1-independent genes induced by influenza B in both U3A and 2fTGH cells only 13 were induced by either dsRNA or SeV as well. Examples of these 46 genes are shown in figure 3-11C, grouped according to their induction by dsRNA and SeV. The lack of any obvious similarities between the profiles of gene induced by influenza B and either dsRNA or SeV suggests influenza B might signal through entirely different pathways than RIG-I or TLR3.

85

Figure 3-12. Differential induction of genes by influenza B, SeV and dsRNA. (A)

Venn diagram showing the overlap of mRNA transcripts expressed over 2-fold more following influenza B infection in both 2fTGH cells and U3A cells. (B) Regulation of specific genes differentially induced by influenza B in U3A cells (3A) and 2fTGH cells

(2f). Each tile shows the fold-increase in mRNA expression for a specific gene following infection with influenza B or SeV, or the addition of dsRNA relative to that seen in untreated cells as a function of color. Green tiles indicate expression in unchanged, while

yellow transiting to red indicate expression was induced from 2 to above 5-fold by

stimulus. Examples of genes induced by influenza B in 2fTGH cells but not U3A cells

are labeled as 2fTGH-specific, while the opposite is true for genes induced only in U3A cells. (C) Regulation of gene induced by influenza B in both U3A and 2fTGH cells.

Influenza B induced genes are grouped according to their induction by SeV or dsRNA as well.

86 (A)

2fTGH+Influenza B U3A+Influenza B

146 46 6

(C) FLU SeV dsRNA integrin, alpha l CREG1 interferon regulatory factor 1 leukemia associated gene 2 neurotensin fibroblast growth factor 2 (B) cig5 FLU SeV dsRNA ISG54 inhibin, beta a ISG56 induced caspase 7 CIAP2 SeV + dsRNASeV + CIAP1 ADAM9 2'-5'-OAS 2 eIF3-theta ISG15 mannose-binding lectin 1 dsRNA dsRNA NOXA induced ALCAM (CD6 ligand) cytochrome c-1 ran, ras-related nuclear protein ests, similar to kiaa0638 STK25, serine/threonine kinase ifi-6-16 membrane metallo-endopeptidase

2fTGH specific RIG-I replication factor c1 TRIM22 topoisomerase 2 nuclear antigen sp100 retinoblastoma 1 integrin, alpha 2 immediate early response 3 hypoxia-inducible factor 1-alpha ptpl1-associated rhogap 1 ifi9-27 megakaryocyte stimulating factor kiaa1046 protein integrin, beta 1 Influenza B only SeV induced ubiquitin-activating enzyme e1-like integrin, alpha v vascular cell adhesion molecule 1 acyl-CoA synthetase 4 ARFD1 dead/h box 18, RNA helicase

U3A specific U3A SAT glia maturation factor, beta uroporphyrinogen decarboxylase guanine nucleotide binding protein 10 2f 3A2f 3A 2f 3A 2f 3A2f 3A 2f 3A

1.0 2.0 5.0

87 DISCUSSION

We had begun this study knowing that extracellular dsRNA failed to induce

ISG56 in the STAT1 deficient cell line U3A (McKendry, John et al. 1991;

Bandyopadhyay, Leonard et al. 1995). Initially our suspicions were that STAT1 might be involved in dsRNA signaling. Instead we provide evidence here that U3A cells are completely unable to respond to extracellular dsRNA because their basal expression of

TLR3 is impaired. Expressing STAT1 in U3A cells failed to increase basal TLR3 mRNA expression or restore ex-dsRNA signaling (Fig. 3-8B and 3-4). Further, two other related STAT1 null cell lines, U3B and U3C were able to respond to ex-dsRNA and expressed substantially higher levels of TLR3 mRNA. While our results do not rule out the possibility that a single mutation affects expression of both STAT1 and TLR3, we conclude that the defect in basal TLR3 expression in U3A cells is STAT1-independent, and that STAT1 is not required for the basal cellular response to ex-dsRNA.

The work we present here highlights the danger of using chemically mutagenized cell lines, like the U series, as representative of individual mutations. In addition to mutations affecting the phenotype which the cells were screened to select for, it is impossible to know what additional mutations they may have acquired during mutagenesis. Numerous studies have been conducted using U3A cells to represent

STAT1 null background. The fact that U3A cells also have a defect in ex-dsRNA signaling as well may call into question some of these findings. Since dsRNA in its role as a PAMP and IFNs have closely related functions in the immune response, it is conceivable that effects mediated by ex-dsRNA signaling have been misinterpreted as

88 being the result of STAT1. While the U3 series of cells still offer one of the best methods of studying a truly STAT1 null background in human cells, more care must be taken in the conclusions drawn from any single U3 cell line.

Unstimulated U3A cells do not respond to extracellular dsRNA, because they do not express sufficient levels of TLR3. Expression of TLR3 alone in U3A cells is sufficient to make the cells responsive to ex-dsRNA, indicating that all other cellular proteins required for gene induction by ex-dsRNA are present and functional (Fig. 3-8C).

When compared with U3A cells, the other STAT1 null lines, U3B and U3C, or the Jak 1 null line U4C expressed over fifty-fold more TLR3 (Fig. 3-8B). The ultimate cause of the defect in TLR3 expression is unclear. Sequencing of the TLR3 promoter in U3A cells revealed no mutations, and when induced the gene could still produce fully functional protein. These facts suggest a problem with the protein factors required for the basal transcription of TLR3. Previous work by Heinz et al. has shown that the promoter of human TLR3 contains an ISRE and a putative GAS site (Heinz, Haehnel et al. 2003).

Deletion of the GAS site decreased basal expression of a TLR3 promoter reporter construct 50%, but did not abolish it. Similarly, we found in STAT1-null U3B and U3C cells that basal TLR3 mRNA expression was not abolished. However, expression of

STAT1 in U3A cells (Fig. 3-8B and 3-9A) or U3B cells (data not shown), failed to significantly increase basal TLR3 expression. This indicates that if the putative GAS site is involved in basal TLR3 expression its function is not mediated by STAT1.

Basal expression of TLR9, like TLR3, has been shown to require an ISRE within its promoter (Guo, Garg et al. 2005). IRF-2 also constitutively binds to the ISREs in both the TLR9 and the TLR3 promoters. Furthermore, deletion of the ISRE has been shown

89 to completely abolish basal TLR3 reporter expression (Heinz, Haehnel et al. 2003). It is

therefore a possibility that IRF-2 is also required for basal expression of TLR3. A

mutation affecting IRF-2 signaling could account for the lack of basal TLR3 expression

seen in U3A cells, but it remains to be seen whether this is the case.

The mechanisms regulating basal and induced expression of TLR3 mRNA are

distinct. U3A cells can be made responsive to extracellular dsRNA by pretreatment with

either type I or type II interferons, but only if STAT1 is first expressed in them (Fig. 3-

9B-D). Our findings agree with and add to the previous observation by Heinz et al. that

in STAT1-deficient mouse macrophages IFNβ could not induce TLR3 mRNA (Heinz,

Haehnel et al. 2003). We have shown that the same is true in human cells in response to

both IFNβ and IFNγ; furthermore, induction of TLR3 mRNA expression by IFNs also

correlates with the development of a functional response to extracellular dsRNA. STAT1

is required for both the type I and type II IFN responses, so its requirement for TLR3

induction by both types of IFN is not unexpected.

Here we show that IFN-γ induces TLR3 mRNA expression in HelaM cells (Fig.

7A). This answers a longstanding question from our previous research. Originally, we

had observed that pre-treatment with IFN-γ was necessary in order for extracellular dsRNA to induce ISG56 in HelaM cells, but we were unable to determine the mechanism responsible. It is likely that, as with U3A cells, TLR3 induction by IFN is critical in making Hela-M cells responsive to ex-dsRNA.

Tisari et al. have previously reported IFNs could enhance both TLR3 expression and cytokine induction by ex-dsRNA signaling in epithelial cells (Tissari, Siren et al.

2005). Our work adds to this by showing that induction of TLR3 by IFN not only

90 enhances ex-dsRNA signaling, but also facilitates it in cells devoid of any basal

responsiveness to ex-dsRNA. U3AR+ and HelaM cells are two cell lines that lack basal

responsiveness to ex-dsRNA, but can be induced to respond through IFN treatment. IFN

probably partially enhances the dsRNA signaling response by its up-regulation of other

proteins involved in dsRNA signaling, including TRIF and IKK epsilon (Siren, Pirhonen et al. 2005). However, expression of TLR3 without the induction of any other genes by

IFN is sufficient to make the U3A cells responsive to ex-dsRNA (Fig. 3-8C). It therefore seems that regulation of TLR3 expression is the critical determinant of whether or not a response to ex-dsRNA can be evoked.

Although extracellular and intracellular dsRNA signaling activates many of the same transcription factors, activation of the two pathways is very different. U3A cells are able to respond to transfected dsRNA, yet treatment with ex-dsRNA fails to induce a signaling response (Fig.3-1). Even at the extremely high concentrations used to treat cells in culture, ex-dsRNA cannot induce an intracellular dsRNA signaling response in

U3A cells (Fig. 3-1). In U4C cells, conversely, SeV, which signals through intracellular dsRNA signaling pathways, is completely unable to signal through TLR3 when intracellular signaling is blocked by a RIG-I dominant negative mutant (Fig. 5D).

Together these results reveal that, in the absence of RIG-I, SeV dsRNA does not induce

genes through TLR3, while in the absence of TLR3, dsRNA added to the media does not

get into the cell in a form in which it can be recognized by RIG-I and potentially other

dsRNA binding proteins. While the former could be the result of inhibition of TLR3

signaling by SeV, the inability of RIG-I to mediate a response to extracellular dsRNA in

91 U3A cells suggests distinct roles for the different dsRNA receptors which are dependent

upon localization of dsRNA within the cell.

Over the past few years it has become increasingly evident that intracellular and

extracellular responses to pathogen-associated molecular patterns (PAMPs), such as

dsRNA, serve distinct functions. Our work here characterizing the nature of the ex-

dsRNA signaling defect in U3A cells has shown that even the mechanisms governing

regulation of basal responsiveness to intracellular and extracellular dsRNA are different.

TLR3 expression is absent in U3A cells due to a defect in the transcriptional machinery

needed for basal transcription, yet both transfected dsRNA and SeV could induce genes

normally through intracellular dsRNA signaling pathways. Unlike TLR3 mRNA

expression, RIG-I mRNA expression in U3A cells was normal (Fig 3-8A). Since basal

TLR3 expression can be lost without affecting intracellular dsRNA signaling, we conclude that distinct mechanisms must control basal responsiveness to extracellular and intracellular dsRNA.

It is still unclear if basal responsiveness to ex-dsRNA plays any role over the course of a viral infection. With a majority of viruses, TLR3 appears not to inhibit viral

replication or improve the outcome of infection. However in some cases, TLR3 has been

shown to mediate chemokine induction both directly within infected cells and in

neighboring uninfected immune cells. One example of the latter phenomenon is seen

with EMCV (Schulz, Diebold et al. 2005). Phagocytosis of EMCV-infected VERO cells

by dendritic cells induced a cytokine response through TLR3. However, the ability of

EMCV to replicate in U3A cells at both low and high MOIs was unaffected by the

expression of TLR3. This indicated that while TLR3 might initially be important in

92 modulating the humoral innate immune response, it does not play a significant role in

EMCV replication within individual cells.

Influenza A is another virus for which it has been reported that TLR3 signaling is important. Expression of a TLR3 dominant negative, lacking the intracellular TIR domain, prevented gene induction within lung airway epithelial cells (Guillot, Le Goffic et al. 2005). However conflicting results were obtained, suggesting that RIG-I mediates signaling in response to Influenza B (Matikainen, Siren et al. 2006). Our attempts to recreate either observation with the more pathologically relevant influenza B met with mixed success. Influenza B was able to induce ISG56 in both U3A and 293 cells, both of which fail to respond to added dsRNA. In 293 cells, ISG56 induction was barely augmented by high levels of TLR3 expression, while induction of ISG56 in U3A cells was higher than was seen in the naturally dsRNA responsive U5A line. Analysis of the genes induced by influenza B by microarray revealed a dramatically different profile than we saw for either dsRNA signaling through TLR3 or SeV signaling through RIG-I. This strongly suggests that influenza may be primarily inducing cellular genes through a different pathway than TLR3 or RIG-I. If viral nucleic acid is the PAMP recognized during influenza infection, then TLR7, TLR8, or Mda5 would be the most likely candidates for the PRR. Since all previous inconclusive work on RIG-I and TLR3 was done using dominant negatives, the importance of these three candidates, along with

TLR3 and RIG-I, in influenza signaling should be determined in either null cell lines or by siRNA knockdown.

93 CHAPTER 4

DIFFERENTIAL REGULATION OF GENES INDUCED BY dsRNA SIGNALING THROUGH TLR3

ABSTRACT

TLR3 mediates the induction of many cellular genes in response to dsRNA.

Transcriptional activation of individual genes has been shown to require transcription factors IRF-3 and NFκB, in addition to phosphoinositide 3-kinase. Previously we showed that adding poly(I:C), a synthetic dsRNA, to the culture media of HT1080-derived cells resulted in TLR3-dependent gene induction. Using a cDNA microarray and mutant cell lines derived from HT1080 cells, we identified genes induced through TLR3 by IRF-3 and NFκB. Inhibition of phosphoinositide 3-kinase by LY impaired the induction of

IRF-3 dependent genes, but not NFκB dependent genes by TLR3 in HT1080 cells. We

also report variation between either species or cell type in gene induction by IRF-3.

ISG54 and ISG56, which are induced directly by IRF-3 in human fibroblasts, required

secondary IFN signaling for their induction through TLR3 in bone marrow-derived

mouse macrophages. In GRE cells, which lack the IFNα and β loci, cycloheximide

blocked the induction of one third of the dsRNA up-regulated genes, showing that a large

number of genes are induced indirectly by dsRNA through means other type I IFN

signaling. Finally, we observed that the synthetic dsRNA stabilized with a L-lysine and

carboxymethylcellulose backbone, poly(I:C)-LC, was a far less efficient activator of

TLR3 than poly(I:C).

94 INTRODUCTION

The cellular response to double-stranded RNA (dsRNA) involves the regulation of hundreds of genes. Depending on where dsRNA is initially recognized, different signaling pathways can mediate this response with similar but distinct outcomes. In the case of dsRNA initially encountered outside of the plasma membrane, the response is mediated by toll-like receptor 3 (TLR3). TLR3 is a member of a family of genes which recognize pathogen associated molecular patterns.

Signaling through TLR3 activates a number of downstream signaling pathways.

Through interaction with the adaptor protein TRIF via its C-terminal TIR domain, TLR3 activates the transcription factors IRF-3 and NFκB, in addition to the stress kinases JNK and p38. TRIF-independent signaling through PI3K is also required for gene induction by IRF-3, and possibly other factors. In addition many of the genes induced directly by TLR3 signaling, such as IFN β and TNFα, can in turn regulate gene induction themselves. As a consequence, many additional cellular signaling pathways and transcription factors are activated indirectly by TLR3. With so many different signaling pathways active, it is difficult to determine through which one(s) TLR3 mediates the induction of individual genes.

In order to address this question we looked at gene induction by dsRNA on cDNA microarrays of 2196 genes, including 288 dsRNA induced genes and 855 interferon induced genes, using mutant cell lines or inhibitors designed to look separately at specific components of TLR3 signaling. In this manner we were able to differentiate between genes induced directly and indirectly through TLR3, as well as genes dependent on NFκB, IRF-3 or

PI3K for their induction. In GRE cells, which lack the IFN locus, induction of almost a third

95 of the genes up-regulated by dsRNA were blocked by cycloheximide, indicating a significant contribution of indirect signaling pathways other than type I IFN to the cellular response to dsRNA. Based on our microarray data, we report that the PI3K inhibitor, LY294002 blocks the induction of IRF-3 dependent genes, but not NFκB dependent genes. Among the genes dependent on IRF-3 for their activation were the ISG56 family members, ISG54 and ISG60, but not ISG58. Unlike human fibroblasts where both genes are induced in an IRF-3- dependent manner, neither ISG54 nor ISG56 was induced by dsRNA in bone-marrow derived mouse macrophages without a functional IFN signaling response. Finally we show that gene induction by the synthetic dsRNA, poly(I:C), but not the L-lysine and carboxymethylcellulose stabilized analog, poly(I:C)LC, occurs through TLR3 within six hours of treatment in cells naturally responsive to dsRNA. Overexpression of TLR3 was required for gene induction by poly(I:C)LC within the same time frame.

RESULTS

Endogenous versus exogenous TLR3 signaling. Conjugation of a L-lysine and

carboxymethylcellulose backbone to synthetic dsRNA (poly(I:C)LC) prolongs its half-

life by protecting against degradation. To determine whether the backbone had any effect

on the ability of dsRNA to signal, we compared the ability of poly(I:C) and poly(I:C)LC

to induce ISG56 in cells naturally responsive to dsRNA added to the culture media

(endogenous) or cells expressing transfected TLR3 (exogenous). As can be seen in figure

4-1A, both poly(I:C) and poly(I:C)LC induced ISG56 equally well in 293 cells

96 expressing high levels of exogenous N-terminal FLAG-tagged TLR3. However while

poly(I:C) was able to induce ISG56 well in MRC-5 cells and GRE cells, two lines

naturally responsive to added dsRNA, no induction of ISG56 mRNA was seen by six

hours after addition of poly(I:C)LC to the culture media. This suggested a difference

between signaling seen through exogenous and endogenous TLR3.

Since exogenous TLR3 behaved differently than endogenous TLR3, we wanted

to determine the reason for the difference in functionality between the two besides level

of expression. Unlike endogenous TLR3 our expression construct lacks the 49 amino- terminal amino acids of TLR3 and instead has a FLAG-epitope tag. To test the possibility that this difference could alter the functionality of our TLR3 construct, we expressed the N-terminal FLAG-tagged TLR3 without its TIR domain (TLR3∆TIR) in

U4C cells along with U3A cells expressing only full length exogenous TLR3. TLR3

lacking the TIR domain has been shown previously to function as a dominant negative

for dsRNA signal. If there was no difference between the abilities of the TLR3 construct

and endogenous TLR3 to signal, then the TLR3∆TIR should block both equally as well.

Surprisingly TLR3∆TIR had no effect on ISG56 induction by dsRNA in U4C cells,

which did not even express detectable levels of TLR3 (figure 4-1B). In contrast,

expression of equivalent amounts of the dominant negative in U3A cells expressing

detectable levels of exogenous TLR3 resulted in an almost complete loss of ISG

induction by dsRNA. The ability of TLR3∆TIR to block dsRNA in the U3A cells

expressing much more exogenous TLR3, but not U4C cells suggests that dsRNA signals

preferentially through endogenous TLR3.

97

Figure 4-1. Differential responsiveness of endogenous and exogenous TLR3. (A)

ISG56 mRNA expression following poly(I:C) or poly(I:C)LC treatment was measured by

RPA in wt11, GRE and MRC5 cells. (B) Western blot showing expression of TLR3 in

U3A cells, U3A cells expressing exogenous TLR3 (U3ATLR3) and U4C cells along with cells expressing TLR3∆TIR (+∆TIR). Full-length (fl) TLR3 and TLR3∆TIR (∆TIR) migrate separately and are labeled. (C) RPA of ISG56 mRNA expression in following the addition of dsRNA (poly(I:C)) to the culture media (upper panel). RNA was isolated from the last four cell lines described above in B. Relative expression of ISG56 mRNA in the different cell lines was quantified and normalized to actin mRNA (lower panel).

98 (A) pICLC pIC pICLC pIC pICLC pIC

ISG56 mRNA

actin mRNA

GRE MRC5 WT11

(B) ∆TIR: + + TLR3-fl

∆TIR Actin U3AU3ATLR3 U4C

(C) U3ATLR3 U4C -+-+-+-+dsRNA (pIC) ISG56 mRNA

actin mRNA

∆TIR ∆TIR

40 35 30 25 20 15

fold-induction 10 5 0 -+-+ -+-+dsRNA ∆TIR ∆TIR

99 Direct versus indirect gene induction by added dsRNA. Many of the genes induced by dsRNA proteins, such as IFNβ and TNFα, which can signal to modulate gene induction themselves. Because of this many of the genes seen up-regulated after exposure to dsRNA could be induced by secondary autocrine or paracrine signaling and not directly by dsRNA. To determine how many dsRNA stimulated genes were directly regulated, we therefore compared expression profiles of GRE cells, with and without cycloheximide treatment by cDNA microarray. Since at six hours after dsRNA treatment, the directly induced gene ISG56 is expressed at its highest level, we used this time point in all our microarray experiments so as to see the greatest difference in alternatively regulated genes. In a single cDNA microarray experiment, RNA from untreated GRE cells was compared with RNA from dsRNA-treated GRE cells to determine changes in expression brought on by dsRNA as fold-induction over untreated.

This was done either in the presence or absence of cycloheximide in both treated and untreated samples, so as to control for any gene induction caused by cycloheximide alone.

Since results are from only a single microarray experiment, genes had to meet two criteria to be considered to be induced by dsRNA. First, their expression had to be at least 2.5-fold higher in dsRNA-treated cells than in untreated cells. Second, regardless of fold-induction over untreated, individual spot had to meet a minimum relative fluorescent intensity of 200 in the dsRNA treated samples to be considered induced. Using these criteria, we determined that dsRNA treatment for 6 hours induced 76 unique genes, some represented by multiple spots upon the array. We decided that a difference of greater than 2-fold between the fold-induction values of individual genes with or without

100

Figure 4-2. Direct and indirect gene induction by dsRNA. Regulation of specific

genes by added dsRNA in GRE cells with or with cycloheximide treatment was determined by cDNA microarray. RNA from untreated GRE cells as a baseline control

was compared to RNA from cells six hours after the addition of dsRNA to the culture

media to assess the magnitude of gene induction. Both the untreated baseline control and

dsRNA treated cells received cycloheximide pretreatment when determining the effect of

cycloheximide (+CHX) on gene induction. Tiles show the fold-increase in mRNA

expression over baseline for specific genes following dsRNA treatment as a function of

color. (A) Genes induced over 2-fold more in the presence or absence of cycloheximide

were considered differentially regulated. Examples of genes which failed to get induced

in CHX-treated cells are shown above (indirectly induced by dsRNA), while genes

induced more strongly in CHX-treated cells are shown below. (B) Examples of genes

induced equally well in the presence or absence of CHX.

101 (A) (B) chemokine, CX3C motif, ligand 1 ISG56 monocyte chemotactic protein 1 fibroblast growth factor 2 cytochrome c IP-10 amyloid beta a4 precursor protein Cig5 ATRX ISG54 janus kinase 2 IRF-2 natural killer cell transcript 4 gtp cyclohydrolase 1 butyrophilin, subfamily 3, member a3 TNFAIP6 megakaryocyte stimulating factor IRF-1 transcription factor iif, alpha interleukin 6 membrane metallo-endopeptidase FAS TRAF1 IkB, alpha PAM ectodermal-neural cortex 1

indirectly induced by dsRNA by induced indirectly c-IAP1 kruppel-like factor 4 FXR1 pathway inhibitor 2 CDC2 acyl-CoA synthetase 3 mrna for caldesmon, 3' utr mannose-binding lectin 2 proteoglycan 1, secretory granule receptor-interacting protein 2 (RIP2) integrin, beta 1 syndecan 4 HNRPA2B1 MARCH6 activin a receptor, type I A20 MIP1-beta interleukin 1 receptor, type i lysine hydroxylase 2 cartilage linking protein 1 ALCAM polymerase (dna), gamma MHC, HLA-G interleukin 1, alpha b-factor, properdin bcl-2 binding component 3 UTX syntaphilin mito. translational initiation factor 2 induced more in in more induced CHX-treated cells CHX-treated activating transcription factor 3 MIP1-alpha WNT5A _ nucleoside phosphorylase CHX: + putative cyclin g1 interacting protein vascular endothelial growth factor c MHC I, HLA-F _ integrin, alpha 2 CHX: +

1.0 2.0 10

102 cycloheximide treatment represented a significant difference in induction. By this definition, 24 genes originally found to be induced by dsRNA were inhibited by cycloheximide (Fig. 4-2). Of the remaining 52, we identified 17 genes that were induced

better by dsRNA in the presence of cycloheximide.

IRF-3 and NFκB-dependent gene induction. Treatment with dsRNA directly activates a number of different transcription factors, the two most studied of which are NFκB and

IRF-3. We therefore wanted to assess by microarray analysis the importance of both of

these factors in the induction of genes by dsRNA.

Overexpression of IRF-3 has been shown to dramatically increase the

expression of IRF-3-responsive genes upon dsRNA treatment without augmenting their

basal expression. To identify dsRNA-induced genes regulated by IRF-3, we therefore

compared p2.1 cells with p2.1.17, a clone strongly overexpressing IRF-3. Both cell lines

are unable to respond to IFN; in addition, p2.1 cells express unusually low levels of IRF-

3 due to an unknown mutation. As shown in figure 4-3A, SeV or added dsRNA induced

little to no ISG56, respectively, in p2.1 cells, yet both strongly induced ISG56 in the IRF-

3 overexpressing P2.1.17 line. Identifying genes induced in p2.1.17, but not in p2.1

provides an ideal way of separating IRF-3-dependent from independent genes.

We identified 31 different genes that were induced consistently by dsRNA in

p2.1.17 cells in two different microarray experiments. Of these 31 genes, six, including

IL-8, NOXA and A20 were also induced in p2.1 cells, suggesting that they are regulated

independently of IRF-3. The remaining 25 genes, which included ISG56, along with

follistatin, ISG54, FUT2, and cig5, were induced in p2.1.17 cells, but not in p2.1 cells

103

Figure 4-3. IRF-3 mediated gene induction by dsRNA. (A) ISG56 expression in p2.1

and p2.1.17 cells was compared between untreated cells (lanes 1 and 5), and following

SeV infection (lanes 2 and 6), dsRNA treatment (lanes 3 and 7) or IFNβ treatment (lanes

4 and 8). (B) Examples of genes induced by dsRNA signaling through TLR3 in p2.1

(IRF-3-)and p2.1.17(IRF-3+) cells as determined by microarray. Genes are grouped for

IRF03 dependence according to whether or not they were induced in either or both cell line. Individual genes were considered to be induced in a cell line if they were induced over two fold (yellow to red on the color bar).

104 (A) 123 45 6 78

ISG56 mRNA

actin mRNA

P2.1 P2.1.17

(B) cig5 10 interleukin 6 WNT11 plasminogen activator inhibitor 1 ISG15 IP10 ISG56

5.0 ISG54 IRF-3 IRF-3

enhanced follistatin MyD88 FUT2 activating transcription factor 3 NF kappa B (p105) short stature 2.0 interferon, gamma-inducible protein 16 MARCH6 mannose-binding lectin 2 A20 NOXA interleukin 8 IRF-3 IRF-3

neutral neutral jun B 1.5 IkB alpha syndecan 4 peroxisome biogenesis factor 1 thymine-DNA glycosylase inhibitor of DNA binding 2

IRF-3 IRF-3 karyopherin (importin) beta 1 repressed repressed 1.0 tight junction protein 2 _ c-IAP 2 IRF-3: + 105 (Fig. 4-3B). They are likely to be induced by dsRNA in an IRF-3-dependent manner.

Finally, we identified six additional genes that were induced in p2.1 cells but not in

p2.1.17 cells, suggesting that induction of their expression by dsRNA was actually

impaired by IRF-3 overexpression.

To determine genes dependent on NFκB but not IRF-3, we used a somewhat different approach. The IκB super-repressor can inhibit the release of NFκB and consequently prevent NFκB-dependent gene induction. By assessing which genes failed to get induced by dsRNA in a cell line expressing the super repressor but were induced in their parental line, we were able to determine which genes were dependent upon NFκB for their induction by dsRNA. Since many NFκB-regulated genes have complex promoters which can respond to multiple different transcription factors, including IRF-3, we wanted to identify those genes which were NFκB-dependent and IRF-3-independent.

We therefore compared gene induction by dsRNA in a cell line, 2F-SR.3, which not only expressed the IκB super repressor but also overexpressed IRF-3, with induction in the parental line, 2fTGH.

By cDNA microarray analysis we identified 32 different genes induced over 2- fold in 2fTGH in at least two out of five separate experiments. Seventeen of these 32 genes were not induced in 2F-SR.3 cells by dsRNA, indicating that they were NFκB- dependent (Fig 4-4). As expected, many of the 17 genes, such as A20, which failed to get induced in the 2F-SR.3 cells, have previously been identified as being NFκB-driven in other signaling pathways. Many of the remaining 15 genes induced in both 2F-SR.3 and 2fTGH cells genes we determined to be IRF-3 dependent, such as ISG54 and ISG56.

106

Figure 4-4. NFκB dependent gene induction. Examples of genes determined to be differentially regulated in wild-type 2fTGH cells (NFκB+) and 2F-SR.3 cells where

NFκB signaling is inhibited (NFκB-). Genes were considered to be NFκB dependent if they were induced over two-fold in 2fTGH cells but not 2F-SR.3 cells

107 follistatin cytochrome c 10 MARCH6 protein phosphatase 2, catalytic subunit, alpha heterogeneous nuclear ribonucleoprotein d-like jun activating transcription factor 3 caspase 7 endothelin receptor type b sex hormone-binding globulin cycloxygenase 2, COX2 5.0 NOXA FUT2 NFkappaB NFkappaB independent ISG54 ISG56 ISG15 cig5 2'-5'-OAS 2 (p69) eIF3-theta interferon, gamma-inducible protein 16 2.0 RIG-I interleukin 6 nucleosome assembly protein 1-like 4 fragile x mental retardation 1 myosin ic A20 cytochrome c-1 interleukin 1, beta AKT1 1.5 tight junction protein 2 janus kinase 1 CRSP2 c-IAP2 jun b macrophage inflammatory protein 1-alpha-P ests, weakly similar to kiaa0638 protein NFkappaB dependent NFkappaB interleukin 1, alpha cytokine-inducible kinase 1.0 h1 histone family, member 0 NFkappaB 2 (p52/p100)) IkB alpha IRF-1 NFκB: +-

108 Additional genes were identified that were induced by dsRNA in 2F-SR.3 cells,

but not 2fTGH, indicating that they are IRF-3-dependent and NFκB-independent. Eleven of them were found to be induced in p2.1.17 cells in at least one experiment as well.

Role of PI3-kinase in gene induction by dsRNA. We have previously shown that PI3K activation is required for TLR3-mediated gene induction by dsRNA. In the presence of the PI3K inhibitor LY, IRF-3 is partially activated by dsRNA treatment, but does not appear able to mediate gene transcription. As a result, pre-treatment of cells with LY prevents the induction of the IRF-3-dependent gene ISG56 by dsRNA signaling through

TLR3. It was still unclear how universal the PI3K requirement was in gene induction by dsRNA. In work done with Dr. Kristi Peters we looked at the effect of LY on gene induction by dsRNA added to the culture media of HT1080 cells. Added dsRNA signaling in HT1080 cells is mediated by TLR3 and ISG56 induction is known to be blocked by LY. Similarly to the results of experiments done with cycloheximide, the effect of PI3K on gene induction by dsRNA was assessed by comparing the fold- induction in RNA expression in dsRNA treated versus untreated cells, determined separately either in the presence or absence of LY. Genes were again considered to be induced by dsRNA if their expression changed by 2.5-fold compared to untreated cells.

By using this method we identified 84 genes that were induced at six hours after treatment with dsRNA, either in samples receiving or not pre-treated with LY. Of these

84 genes, 12 were induced over twice as strongly in cells not treated with LY (Fig4-5A).

Another 61 of the genes were induced within two-fold of each other in the presence or absence of LY, and were therefore considered to by PI3K-independent. Finally, the

109 remaining 11 genes were actually induced over twice as strongly in the presence of LY,

suggesting that PI3K may actually impair their expression following dsRNA treatment.

To verify that LY did not inhibit the induction of all genes, we looked at A20

mRNA expression following dsRNA treatment in the presence or absence of LY.

Furthermore, to show that this effect was universal and not limited to HT1080 cells, we

used 293 cells expressing exogenous TLR3. Like HT1080, the response to added dsRNA

is generated through TLR3. As is shown in figure 4-5B, LY failed to inhibit A20 expression following dsRNA treatment, verifying our microarray results.

It was of note that many of the genes found to be PI3K dependent were also IRF-3 regulated, for example IP-10 and ISG54, while a number of the genes identified as being induced equally well or better in the presence of LY were NFκB regulated, like A20, IL-

1β and c-IAP2. To look at this more closely, we analyzed the IRF-3-regulated gene cig5, and the NFκB-regulated gene IL-8 by RT-PCR for 20 cycles. While this assay is not quantitative, it can clearly be seen that cig5 expression, but not IL-8 expression, is inhibited by LY (Fig. 4-5C). These results add further support the theory that IRF-3- mediated, but not NFκB-mediated, gene induction by dsRNA is PI3K dependent.

Adding dsRNA to the media leads to gene induction through TLR3, while transfecting dsRNA into cells or infecting with viruses such as SeV induces genes through the RIG-I signaling pathway. Since signaling through both TLR3 and RIG-I lead to gene induction through IRF-3, we asked whether PI3K was needed for gene induction by transfected dsRNA as well. To test this, we looked at the ability of LY to block

ISG56 induction by either added or transfected dsRNA in U4C cells. As can be seen in figure 4-5D, LY was able to block ISG56 induction by transfected dsRNA to the same

110

Figure 4-5. Role of PI3K in gene induction by dsRNA (A) Microarray data showing

examples of the genes induced through TLR3 in HT1080 in the presence (+) or absence

(-) of the PI3K inhibitor, LY. As before, fold-induction over baseline is shown for

individual genes as a function of color. Genes are grouped into one of three categories,

based on whether they were (1) induced over two-fold more in the absence of LY (PI3K-

dependent); (2) induced within two-fold of each other with and without LY (not regulated

by PI3K); or (3) induced over 2-fold more in the presence of LY (PI3K inhibited). (B)

A20 mRNA expression was measured by RPA in wt11 cells left untreated (1), or with

dsRNA alone added to the culture media (2), as well as in the presence of cycloheximide

(3) or LY (4). (C) Expression of IL-8 and cig5 mRNA was determined by RT-PCR in

LY-treated wt11 cells (1) or following dsRNA treatment alone (2), or in addition to LY

(3) (D) RPA of ISG56 mRNA expression in HT1080 cells in the presence or absence of

LY following addition of dsRNA to the culture media (added) or by transfection

(transfected).

111 (A) LY: _ + ISG56) 10 2'-5'-OAS2, p69 IP-10 (B) vacuolar protein sorting 36 1 23 4 apolipoprotein b (APOB) A20 mRNA peptidase-beta, mitochondrial kinase-inducible ras-like protein

PI3K dependent neuralized, drosophila, homolog-like IFITM1 (9-27) actin mRNA 5.0 ISG54 IFN-gamma-inducible protein 16 ISG15 IFN-gamma-inducible protein 30 RIG-I NOXA TRAIL (C) 12 3 TRIM22 IL-8 2.0 monocyte chemotactic protein 1 Cig 5 c-IAP 2 HRPSL amyloid beta a4 precursor protein IkappaB, alpha

not regulated by PI3K by not regulated syndecan 4 A20 c-IAP 1 (D) _ _ + + : LY interleukin 6 ISG56 mRNA NFkappaB2 (p52/p100) 1.5 jun b NFkappaB1 (p105)

endothelial cell-specific molecule 1 actin mRNA formyl peptide receptor-like 1 added transfected inhibin, beta a complement component 2 interleukin 1, beta

PI3K inhibited traf1 1.0 mannose-binding lectin 2

112 efficiency it blocked dsRNA added to the culture media. This suggests that PI3K is needed for RIG-I signaling as well as TLR3 signaling.

ISG56 family of genes. ISG54 and ISG56 belong to a family of genes encoding proteins containing tetratricopeptide repeats (TPRs). Also included in this family are two other interferon stimulated genes, ISG58 and ISG60. In contrast to ISG54 and ISG56 which were consistently two of the most strongly induced genes in our microarray experiments,

ISG60 was only weakly and inconsistently induced while ISG58 was not induced in any experiments. We therefore examined the expression of ISG58 and ISG60 by RNAse protection assay to determine more accurately if they behaved the same as their other two family members.

We looked at the effect of IRF-3 overexpression on the induction of ISG60 by comparing its induction by dsRNA in HT1080 and p2.1.17 cells (Fig. 4-6A). Previously we showed that ISG60 could be induced independently of IFN signaling by both added dsRNA and SeV (Fig 3-2A). However unlike ISG54 and ISG56, we failed to see increased ISG60 induction in IRF-3 overexpressing cells in our microarray experiments.

Normalizing ISG expression to actin revealed that, following dsRNA treatment, ISG60 expression in p2.1.17 cells was only 3.2-fold higher than it was in HT1080 cells. In contrast, ISG54 and ISG56 mRNAs were expressed 28.4-fold and 18.7-fold more in the

IRF-3 overexpressing cells, respectively. However, when we looked at ISG60 induction in p2.1 cells, we found that, like ISG54 and ISG56, it was not induced by dsRNA and was induced only poorly by SeV. Together these results suggest that, although ISG60 is

113 induced through IRF-3 like ISG54 and ISG56, its expression is not as dependent on cellular IRF-3 levels.

Unlike its three family members, ISG58 appeared not to be induced directly or indirectly at six hours by either dsRNA or SeV in p2.1 cells or p2.1.17 cells (Fig. 4-6B).

However a low level of ISG58 mRNA was constitutively detected in all cells tested, suggesting that its constitutive expression is not IRF-3 dependent.

In addition to differences in induction seen between ISG56 family members, we also sought to determine if there were any differences in induction between species. We therefore looked at the ability of dsRNA to induce the homologs of ISG54 and ISG56 in wild-type mouse macrophages as well as IFN receptor and STAT1 knock-out macrophages. Unlike many mouse cell types, macrophages express TLR3 and consequently can respond to added dsRNA. As shown in figure 4-6C, when wild-type macrophages were treated with dsRNA, both ISG54 and ISG56 were strongly induced.

By contrast, both ISGs were only weakly induced in STAT1 knockout macrophages while neither was induced in IFN-receptor knockout macrophages. These results indicate a species- or cell type-dependent difference in ISG54 and ISG56 induction by dsRNA. In contrast to human fibroblasts, where their human homologs can be directly induced by dsRNA through IRF-3, mouse ISG54 and ISG56 are only indirectly induced through activation of the IFN signaling pathway in murine macrophages.

114

Figure 4-6. Differential induction of ISG56 family of genes. (A) Expression of

ISG54, 56 and 60 by dsRNA signaling through TLR3 was compared in HT1080 and p2.1.17 cells by RPA. (B) RPA showing the expression of ISG58 in addition to ISGs 54,

56 and 60 in untreated p2.1 and p2.17 cells, or following SeV infection or TLR3 activation by added dsRNA. (C) Murine ISG54 and ISG56 expression in bone marrow- derived macrophages. Macrophages were isolated from w.t. 129 mice as well as

IFNAR2-/- and STAT1-/- mice. At seven or ten days after isolation dsRNA or IFNβ was added to the culture media as indicated and RNA was isolated. IFNAR2-/- macrophages were challenged 10 days after isolation.

115 _ (A) ++:dsRNA

ISG54 mRNA

ISG60 mRNA

ISG56 mRNA

actin mRNA

HT1080

p2.1.17 (B) + + : dsRNA + + : SeV ISG54 mRNA

ISG60 mRNA

ISG58 mRNA

ISG56 mRNA

actin mRNA

p2.1 p2.1.17

dsRNA (C) -+ + -+ + dsRNA IFN 7 7 10 7 7 10 days after isolation

moISG56 mRNA

moISG4 mRNA

Cyclophillin mRNA

IFNAR2-/-STAT1-/- wt 129

116 DISCUSSION

Here we have shown that poly(I:C) and poly(I:C)LC differentially induce ISG56

through TLR3. In wt11 cells, which express high levels of exogenous TLR3, both

poly(I:C)LC and poly(I:C) are able to induce ISG56 equally well. However in MRC4

and GRE cells, two lines naturally able to respond to dsRNA when it is added to the culture media, only poly(I:C) induces ISG56 mRNA expression. The reason behind this difference is unclear. In trials poly(I:C)LC is employed as a therapeutic agent in preference to poly(I:C) because it remains stable longer due to its L-lysine and carboxymethylcellulose-conjugated backbone, and it has been shown to be less toxic. It is possible that stabilized the backbone prevents efficient uptake of poly(I:C)LC or its recognition by TLR3, and it is only with prolonged treatment or elevated TLR3 expression, such as that seen in wt11 cells, that signaling can occur. In vivo, this may result in a more cell-type selective signaling response, accounting for the reduced toxicity of poly(I:C)LC.

Another possibility is that endogenous TLR3 and the TLR3 construct expressed exogenously in wt11 cells, which has a FLAG-tag on its amino-terminus, respond differently to dsRNA. Deletion of the TIR domain from our TLR3 construct results in a dominant negative which was able to block ISG56 induction by dsRNA in U3A cells expressing exogenous TLR3. In contrast, expression of the same amount of dominant negative in U4C cells completely failed to block ISG56 induction through substantially lower levels of endogenous TLR3. This suggested that there was a difference between signaling through endogenous and exogenous TLR3. The TLR3 construct we used is

117 lacking the 49 N-terminal amino acids and has an artificial signal peptide and flag tag in

their place. It is possible that these modifications alter the localization of exogenous

TLR3 or impair the efficiency with which it recognizes dsRNA. These results show regardless that modification of the N-terminal of TLR3 can alter its ability to mediate gene induction in response to dsRNA.

Many genes are induced by dsRNA through TLR3 once signaling occurs. We have identified which of these genes are directly and indirectly induced by dsRNA. Over two thirds of the genes induced six hours after treatment with dsRNA in GRE cells were still induced in the presence of cycloheximide. This suggested that despite inducing the expression of a large number of signaling transcription factors, the majority of genes up- regulated after dsRNA treatment are still directly induced within six hours. As shown in figure 4-2, dsRNA directly induced genes encoding proteins with direct antiviral functions, such as cig5 as well as cytokines like MIP1-α and MIP1-β.

Cycloheximide did block the induction of a third of the genes normally induced by dsRNA in GRE cells. GRE cells lack the IFN locus and are therefore unable to signal indirectly through the IFN pathway (Miyakoshi, Dobler et al. 1990). This means that the indirect gene induction we see in GRE cells must be mediated by other signaling pathways and transcription factors than IFN pathway and ISGF3. One good possibility is

TNFα which, when induced through IRF-3 can alter the NFκB signaling response

(Covert, Leung et al. 2005). Unfortunately, the cohort of genes, with their induction repressed by cycloheximide, were not obviously dependent on NFκB or any particular transcription factor.

118 A few genes were also identified as being better expressed in cycloheximide-

treated cells. This was not unexpected as there are many instances of feedback regulation in signaling. Since cycloheximide treatment can alone induce many genes, as well as stabilizing unstable mRNAs, it was not possible to directly compare mRNA expression in dsRNA treated cells in the presence or absence of cycloheximide. Instead of directly comparing expression, we compared fold induction of genes by dsRNA relative to untreated cells. To determine fold induction in the presence of cycloheximide, mRNA expression in dsRNA- and cycloheximide-treated cells was compared directly with

mRNA expression in cycloheximide-treated cells without dsRNA treatment. In this way,

any cycloheximide-specific mRNA regulation is accounted for when we finally compared

fold-induction by dsRNA between cycloheximide-treated and untreated samples.

We also found roles in TLR3-mediated gene induction by dsRNA for PI3K, IRF-

3 and NFκB. In p2.1 cells that have low levels of IRF-3, dsRNA failed to induce the

expression of the IRF-3 dependent gene, ISG56. However expression of exogenous IRF-

3 in p2.1.17 cells resulted in strong induction (Fig. 4-3A lanes 3 and 7). Since p2.1 and

p2.1.17 cells cannot respond to IFN, induction occurs directly by IRF-3 and not indirectly

by ISGF3 (lanes 4 and 8). Comparison of the genes induced by dsRNA in p2.1 and

p2.1.17 cells by cDNA microarray revealed 24 genes in addition to ISG56 which required

IRF-3 expression for their induction by dsRNA. The list of IRF-3 dependent genes

included many genes previously well documented to be regulated by an ISRE within their

promoters, such as IP-10 and ISG15. Our results also confirmed the dependence on IRF-

3 for induction of many more genes, like ISG54 and MyD88, known to have putative

119 ISRE sites in their promoters. In addition a number of genes were newly identified as

IRF-3 dependent, including FUT2 and follistatin.

Our microarray work also revealed six genes which did not require IRF-3 for their

induction, including A20, IL-8, and the pro-apoptotic gene NOXA. Surprisingly, six

additional genes were found to be induced in p2.1 cells but not p2.1.17 cells suggesting

IRF-3 expression actually impaired their induction. It was unclear what mechanism lay

behind this, in particular whether it was a direct effect, or mediated indirectly by an IRF-

3 induced gene.

The second major transcription factor activated through TLR3 is NFκB. Out of

32 genes induced by dsRNA through TLR3 in 2fTGH cells, seventeen were found to be

dependent on NFκB for their induction. We identified NFκB-dependent genes by

comparing the induction profile of 2fTGH cells with that of 2F-SR.3 cells following

dsRNA treatment. 2F-SR.3 cells not only express the IκB super-repressor, which

prevents NFκB signaling, but also overexpress IRF-3. The promoters of many dsRNA

induced genes contain both κB binding sites and ISREs. By augmenting IRF-3 signaling

in an NFκB null background we are able to clearly distinguish genes absolutely requiring

NFκB for their induction from those where signaling through IRF-3 could compensate

for its absence. We therefore conclude that the genes which we have identified as

NFκB dependent, either do not contain an ISRE within their promoters, or require both

IRF-3 and NFκB for their induction.

In addition to IRF-3 and NFκB, we also looked at the ability of PI3K to regulate gene induction by dsRNA. Our lab has previously shown that PI3K signaling is required for the phosphorylation of IRF-3 and subsequent gene induction. Not surprisingly, our

120 microarray data reflected this finding. When the gene induction profiles from LY treated

and untreated HT1080 cells stimulated by dsRNA were compared, many of the genes we

determined previously to be IRF-3 dependent were impaired by the inhibition of PI3K

signaling. In contrast, induction of most NFκB-dependent genes was not impaired by

LY. These results were confirmed by RPA and RT-PCR for the NFκB dependent genes

IL-8 and A20, as well as the IRF-3-dependent gene, cig5. Together these results suggest that although PI3K is required for IRF-3 mediated gene induction, it is not required for signaling by NFκB.

Two of the genes most strongly induced by dsRNA through IRF-3 are ISG56 and

ISG54. Both genes, along with ISG58 and ISG60, are members of the same family of genes, grouped together because of their tetratricopeptide repeats. We had previously shown that ISG60 induction was could be induced by SeV and dsRNA similar to ISG54 and ISG56. However, ISG58 did not appear to be likewise regulated. Unlike its family members, ISG58 was constitutively expressed at a low level in HT1080-derived cells and failed to be induced by either dsRNA or SeV, even in IRF-3 overexpressing cells. While

ISG58 has been shown to be induced by IFNα in the acute promyelocytic leukemia cell line, NB4, we did not test for its induction by type I IFNs in HT1080 cells (Niikura,

Hirata et al. 1997). It is therefore possible that the ISG58 locus is mutated in the HT1080 background. Another possible reason for the difference in the regulation of ISG58 is due

to the sequences of the ISREs found in all four genes. ISG54 has only a single ISRE

element, while ISG60 and ISG56 both have two. The extended sequence of one ISRE in

all three genes is identical, GGAAAGTGAAACT. In contrast the promoter of ISG58 has

121 three, putative ISREs (GAAANNGAAA), but the closest match to the conserved ISRE of

the other three genes is AGAAACTGAAACT.

Mouse ISG56 has an identical ISRE to the conserved ISRE of human ISGs 54, 56, and 60, yet induction of neither its mRNA nor mouse ISG54 mRNA by dsRNA is directly mediated through dsRNA in bone marrow derived macrophages. Instead both genes are induced through autocrine feedback through the IFN signaling pathway. The fact that dsRNA fails to directly induce mouse ISG56, while still being able to activate signaling through the IFN pathway, suggests a difference in dsRNA-mediated gene induction exists between human fibroblasts and murine macrophages. Whether this difference is species or cell-type specific remains to be determined.

122 CHAPTER 5

ANALYSIS OF ROLES OF TLR3, IFN, IRF-3 AND NFκB IN GENE INDUCTION BY SENDAI VIRUS INFECTION

ABSTRACT

Sendai virus (SeV) infection causes the transcriptional induction of many cellular genes that are also induced by interferon (IFN) or double-stranded (ds) RNA. We took advantage of various mutant cell lines to investigate the putative roles of the components of the IFN-and dsRNA-signaling pathways in the induction of those genes by SeV.

Profiling the patterns of gene expression in SeV-infected cells demonstrated that Toll-

Like Receptor 3, while essential for gene induction by extracellular dsRNA, was dispensable for gene induction by SeV. In contrast, Jak1, which mediates IFN signaling, was required for the induction of a small subset of genes by SeV. NFκB and IRF-3, the two major transcription factors activated by virus infection, were essential for the induction of two sets of genes by SeV. As expected, some of the IRF-3-dependent genes, such as ISG56, were more strongly induced by SeV in IRF-3-overexpressing cells.

Surprisingly, in those cells, a number of SeV-induced genes, such as A20, were expressed poorly following viral infection. Thus, in addition to mediating the induction of many genes, IRF-3 can also suppress the expression of select SeV-induced cells.

123 INTRODUCTION

Many cellular genes, encoding proteins of diverse functions, are transcriptionally induced during viral infections (Zhu, Cong et al. 1998; Chang and Laimins 2000;

Mossman, Macgregor et al. 2001; Geiss, Salvatore et al. 2002; Strahle, Garcin et al.

2003). Because host-virus co-evolution has tolerated this host response to virus infection, it is thought to be conducive to maintaining proper homeostasis between viral proliferation and host survival. A large number of virally induced proteins have direct or indirect anti-viral effects. They may inhibit protein synthesis in the infected cells (Guo,

Hui et al. 2000), impair viral assembly (Chin and Cresswell 2001), initiate the innate immune response, or prime the adaptive immune response (Malmgaard 2004). Diverse components of the infecting viruses or viral intermediates, produced during its replication, can be the responsible agents that trigger the signaling pathways leading to cellular gene induction. Depending on the specific virus, proteins (Boyle,

Pietropaolo et al. 1999), viral ribonucleoproteins (tenOever, Servant et al. 2002), or viral single-stranded (ss) (Lund, Alexopoulou et al. 2004) or double-stranded (ds) RNAs

(Alexopoulou, Holt et al. 2001; Diebold, Montoya et al. 2003; Yoneyama, Kikuchi et al.

2004) have been shown to be the critical agents. Among these, dsRNA has been viewed as the most important agent because it is produced in many virus-infected cells.

Moreover, synthetic dsRNA, when added to cells, can induce transcription of some of the same cellular genes that are induced by virus infection (Geiss, Jin et al. 2001). The major transducer of signaling, generated by extracellular dsRNA, is a member of the Toll-like receptor (TLR) family, TLR3 (Alexopoulou, Holt et al. 2001).

124 One set of genes induced in common by many viruses and dsRNA encode type I

interferons (IFNs) (Malmgaard 2004). These secreted cytokines have strong anti-viral

effects. Surprisingly, many dsRNA-induced and virally induced cellular genes are also

induced by IFNs (Sarkar and Sen), thus creating a positive feedback loop that reinforces

induction of the same genes in infected cells.

To better understand the repertoire of genes induced by Sendai virus (SeV)

infection of cells in culture, we have classified them into groups that are induced by

IFNs, via TLR3, or independently of either. For identifying SeV-induced genes, we used

cDNA microarrays customized for this purpose, and based on genes we previously

identified whose transcription is induced by IFNs (Der, Zhou et al. 1998; de Veer, Holko et al. 2001) or dsRNA via TLR3 (Geiss, Jin et al. 2001). In addition to many virally induced genes, dsRNA- and IFN-inducible genes were represented in the microarray used here. We had previously used mutant cell lines, which could not respond to IFN or extracellular dsRNA to delineate the signaling pathways activated by the different inducers (Tiwari, Kusari et al. 1987; Bandyopadhyay, Leonard et al. 1995; Leaman,

Salvekar et al. 1998). To assess the relative contributions of the IFN- and the TLR3 mediated dsRNA-signaling pathways in gene induction by SeV, we again took advantage of some of those mutant cell lines. As we had done previously in chapter 4 for dsRNA signaling via TLR3, we used other mutant cell lines to investigate the relative contributions of NFκB and IRF-3, the major transcription factors activated by SeV infection, to cellular gene induction (Peters, Smith et al. 2002; Hiscott, Grandvaux et al.

2003).

125 Our study revealed that, among the genes that are induced immediately after SeV

infection, only a few were dependent on IFN signaling and none were dependent on

dsRNA signaling through TLR3. As expected, one class of genes required NFκB,

whereas another class required IRF-3 for induction in SeV-infected cells. Surprisingly,

viral induction of a subset of NFκB-dependent genes was negatively regulated by IRF-3, thus revealing a new aspect of cross-talk between the two transcription factors.

RESULTS

Role of TLR3 in gene induction by Sendai virus. To determine the importance of the

TLR3-mediated dsRNA- signaling pathway for gene induction by SeV we used HEK293- derived cell lines. The parental cell line does not express any TLR3 whereas a derived line, 293/TLR3, does. In the absence of TLR3, 293 cells do not respond to dsRNA added to the media, however upon the expression of TLR3, the cells are able to respond robustly (Alexopoulou, Holt et al. 2001). Thus, a comparison of the genes induced by

SeV in this pair of cell lines enabled us to assess the importance of TLR3-mediated viral dsRNA-signaling. To focus on the primary cellular response to virus in the absence of paracrine or autocrine signaling, cells were infected with SeV at a high m.o.i. and gene

induction was measured six hours after infection. From our previous experience, six

hours was the optimum time to observe the strong direct induction of genes by SeV, with

minimal secondary induction (Guo, Peters et al. 2000).

For transcript profiling, a cDNA microarray of 2196 genes including 288

previously identified dsRNA-regulated genes(Geiss, Jin et al. 2001) and 855 interferon

126

Figure 5-1. Effect of TLR3 on the regulation of cellular genes by SeV and dsRNA.

Global changes in mRNA levels in 293 cells (A, C) or 293 cells expressing TLR3 (B, D)

six hours after SeV infection at 1HAU/4.0x103 cells (A, B) or dsRNA treatment at

100µg/ml (C, D) as determined by cDNA microarray experiment. Each point on the scatter plots represents the expression of an individual mRNA, as determined by units of fluorescent intensity, in untreated cells (x-axis) plotted against its expression six hours after SeV infection or dsRNA treatment (y-axis). The central diagonal line (black) represents equal expression in treated and untreated samples, while two-fold differences in expression are indicated by the two flanking blue lines. (E) Regulation of specific genes by dsRNA and SeV in 293(-) and 293/TLR3 (+) cells. The tiles show the fold- increase in mRNA expression for specific genes in SeV- or dsRNA-treated cells relative to untreated cells as a function of color. Green, expression was unchanged.

YellowÆRed, expression was induced to increasing degrees. (F) RNase protection assay

(RPA) of ISG56 induction in untreated (1, 4), dsRNA treated (2, 5), or SeV-infected (3,

6) 293 cells (lanes 1-3) or 293/TLR3 cells (lanes 4-6).

127 (A) (B) 293 293/TLR3

105 105

104 104 +SeV

103 103

102 102 102 103 104 105 102 103 104 105

(C) (D)

105 105

104 104 +dsRNA 103 103

102 102 102 103 104 105 102 103 104 105

(E) 10 TLR3: -+ -+ (F) A20 123456 5.0 ISG56 ISG56 mRNA ISG54 2.0

ISG15 Actin 1.5 mRNA IL-8 293 293/TLR3 1.0 SeV dsRNA

128 induced genes(de Veer, Holko et al. 2001) was used. Because dsRNA and IFN induce many of the same genes as viral infection does, this array was well suited to study differential gene induction. Total RNAs from infected cells and their uninfected counterparts was used to make fluorescently-labeled cDNA and compared by microarray analysis. Individual gene induction or repression by SeV infection was determined from the ratio of relative fluorescent intensities between the infected and uninfected samples.

SeV infection was able to alter the expression of multiple genes in 293 cells; expression of 36 genes was elevated whereas that of many others was reduced in virus- infected cells (Fig. 5-1A). In 293/TLR3 cells, the patterns were not markedly different

(Fig. 5-1B), indicating that TLR3 plays an insignificant role in gene induction by SeV. In contrast, TLR3 was required for the responsiveness of the cells to dsRNA treatment (Fig.

5-1C and D).

We next examined induction of specific genes to verify that SeV did not induce any genes through TLR3. All of the genes induced over 2-fold in at least two of three microarray experiments by SeV in 293/TLR3 cells were also induced in 293 cells, proving that this was the case. Interferon stimulated genes (ISGs) 15, 54 and 56, along with the A20 (TNFAIP3) and interleukin 8 genes, were all strongly induced by SeV- infection and dsRNA-treatment; induction of all five genes by dsRNA required TLR3, but SeV was able to induce them all without TLR3 (Fig. 5-1E). To confirm the microarray data, we used RNase-protection assays to quantitate the levels of

ISG56mRNA (Fig. 5-1F). As expected, SeV induction of ISG56 was TLR3-independent, while its induction by dsRNA was TLR3-dependent. Moreover, the levels of ISG56 mRNA in SeV-infected 293 and 293/TLR3 cells were similar (Fig. 5-1F, lanes 3 and 6),

129

Figure 5-2. Differential induction of genes by SeV and dsRNA. Select genes differentially regulated by SeV and dsRNA in 293 cells expressing (+) or not expressing

(-) TLR3. Colors represent fold-induction by SeV (left column) or dsRNA (right column) as described in Fig. 1.

130 10 TLR3: - + - + Protein-tyrosine phosphatase, N11 CD44 antigen 5.0 Mannose-binding lectin 2 Activating transcription factor 3

2.0 genes SeV specific Tissue factor pathway inhibitor 2

Alpha-1 antiproteinase

1.5 Nuclear factor erythriod 2-like 3 Nuclear Domain 10 Protein 52 dsRNA dsRNA

specific genes specific Vascular Endothelial G.F. C 1.0 SeV dsRNA

131 indicating that even for the maximal induction of this mRNA, the viral signaling pathway did not require TLR3.

Although there was a differential need of TLR3, because both agents induced many common genes, we wondered whether the profiles of genes induced by dsRNA and

SeV were completely overlapping. This was not the case. Of the 36 genes were induced by SeV in 293 cells, and the 35 genes induced by dsRNA in 293/TLR3 cells, only nine were in common. In addition to the five genes shown in figure 5-1E, FUT2, syndecan4,

IκBα, and NOXA also belonged to this class. A few examples of differentially induced genes are given in figure 5-2. These results suggest that the signaling pathways used by the two agents are different, although some components may be shared.

Involvement of IFN-signaling in gene induction by SeV. SeV-infection is known to induce IFN synthesis and many of the genes induced by SeV are also induced by IFNs.

Thus, it is possible that some genes are induced in SeV-infected cells by the autocrine action of newly synthesized IFN. To test this possibility, we compared gene induction in

SeV-infected U4C and 2fTGH cells (Fig. 5-3). U4C cells, unlike their parental line

2fTGH, lack functional Jak1, which is required for both type I and type II IFN signaling.

Consequently, IFNs cannot induce any genes in U4C cells (McKendry, John et al. 1991).

As in 293 cells, many genes were strongly induced in 2fTGH cells after 6h of SeV- infection (Fig. 5-3A). Using the criterion of 2-fold induction as the cut-off, we observed that 49 genes were induced in 2fTGH cells. Only, a small subset of genes, 14 of the 49 were not induced over 2-fold by SeV in U4C cells as well. Hence the majority of genes induced in 2fTGH cells were also induced in U4C cells indicating that Jak1, and hence

132

Figure 5-3. IFN dependent regulation of SeV-induced genes. Select SeV-regulated

genes displaying differential induction patterns in Wt (2fTGH) and Jak1-/- (U4C) cells.

(A) Microarray data showing average fold-increase in mRNA expression six hours after

SeV infection in 2fTGH and U4C cells. Genes are grouped based on their independence, dependence, or impairment by Jak1 for induction. (B) RPA (A20) and Northern (OAS, p69) analysis of mRNA expression in 2fTGH and U4C cells, before or six hours after

SeV infection.

133 (A) 10

2fTGHU4C FUT2

NOXA ISG54

5.0 ISG15 IL-8

IKK alpha

JAK I - independent JAK I - A20

syndecan 4

2.0 NFκB p105

2 -5 OAS, p69

MHC, class I, F MHC, class I, A MHC, class I, C 1.5 JAK I - dependent JAK I - MCP1

PI-3K, gamma

complement component 2 Kallmann Syndrome I JAK I impaired 1.0 +-JAK I

(B) 2fTGH U4C -+-+ A20

OAS, p69

134 IFN-signaling, were not involved in their induction. A20 and OAS p69 were selected as representatives of IFN-independent and -dependent genes and induction of the corresponding mRNAs was monitored by RPA and Northern blotting, respectively.

Those quantitative assays supported the conclusions drawn from the microarray analyses:

A20 mRNA was induced by SeV-infection of either cell line but OAS mRNA was not induced in U4C cells (Fig. 5-3B). Surprisingly, another small subset of six genes was induced in U4C cells but not in 2fTGH cells, suggesting a suppressive function of Jak1

(Fig. 3A).

Role of NFκB in SeV signaling. One of the major transcription factors activated by viral infections is NFκB. To assess its role in gene induction by SeV, we took advantage of another cell line, 2F-SR. This line was derived from the 2fTGH line by expressing a non- phosphorylatable mutant of IκB, which acts as a super-repressor of NFκB. Cytokines, dsRNA or SeV cannot activate NFκB in the 2F-SR cells (Van Antwerp, Martin et al.

1996; Algarte, Nguyen et al. 1999; Peters, Smith et al. 2002). Of the 49 genes induced by

SeV infection in 2fTGH cells, 28 of these were not induced in 2F-SR cells. Thus, almost

60% of the induced genes were NFκB-dependent while just over 40% were not.

Examples of both classes of genes are given in figure 5-4A. Quantitative assays again

confirmed our conclusions: NOXA and ISG56 mRNAs were induced equally well in both

cell lines whereas A20 and mannose binding lectin 2 mRNAs were not induced in 2F-SR

cells (Fig. 5-4B). These results confirm the notion that NFκB is a major transcription factor used by SeV to induce cellular genes.

135

Figure 5-4. Requirement of NFκB for gene induction by SeV. Genes regulated by

SeV in Wt (2fTGH) and NFκB null (2F-SR) cells. (A) Average fold-induction of SeV- regulated genes, grouped by dependence on NFκB, as determined by microarray. (B)

Quantitative analysis of NOXA, ISG56 and A20 mRNA induction by RPA.

136 (A)

2fTGH2F-SR 10 nexin, PAI type 1

fucosyltransferase 2

NOXA ISG54 5.0 ISG15 B - independent independent B -

κ ISG56 NF nucleoside phosphorylase

2.0 syndecan 4

NFκB p105 IKK alpha

jun B 1.5

B - dependent B - IL-8 κ

NF mannose binding lectin 2

MHC, class I, F A20 1.0 +-NFκB (B) 2fTGH 2F-SR -+-+ NOXA

ISG56 mRNA

A20

MBL2

137 Role of IRF-3 in gene induction by SeV. In addition to NFκB, another major transcription factor activated by SeV is IRF-3. To identify genes requiring IRF-3 for induction by SeV, we used P2.1 cells (Leaman, Salvekar et al. 1998). These cells were derived from U4C cells and hence lack Jak1. In addition, their IRF-3 level is very low and consequently, the IRF-3 signaling pathway is defective in these cells (Peters, Smith et al. 2002). As expected, we observed that one set of genes, 30 of the 42 genes induced by SeV in U4C cells, was induced well in P2.1 cells and another set, the remaining 12 genes, was not. Examples of the two sets of genes are given in figure 5-5. These results demonstrated that induction of some genes by SeV-infection required IRF-3.

To further characterize this requirement, we used U4C.2 cells in which IRF-3 was overexpressed. Forty-five genes were induced in U4C.2 cells, including a group of 19 genes that were induced much better than in U4C cells (Fig. 5-6A). An extreme example of this group is follistatin; quantitative assays confirmed that its mRNA was barely induced in U4C cells but strongly induced in U4C.2 cells (Fig. 5-6C). Induction of another group of mRNAs was hardly affected by the IRF-3 level (Fig. 5-6B and D).

These results indicate that among the IRF-3-dependent genes, some, but not all, can be induced efficiently only when the cellular IRF-3 level is high.

Role of IRF-3 in suppressing gene induction by SeV-infection. In addition to the results

discussed above, the array analysis shown in figure 5-6 produced an unexpected result.

We observed that at least eight genes, or about 20% of those strongly induced in U4C, were poorly induced in U4C.2 cells, indicating that IRF-3 was negatively affecting their

138

Figure 5-5. IRF-3 dependent gene induction. Examples of genes regulated by SeV in

U4C cells and p2.1 cells in which IRF-3 expression is impaired. Genes are grouped by their dependence on IRF-3 for induction.

139 U4C p2.1 10 A20 IL-8 5.0

IRF-3 IRF-3 IKK alpha

2.0 independent NOXA ISG15 1.5 ISG54 IRF-3 IRF-3 ISG56 1.0 dependent

140

Figure 5-6. Regulation of SeV-induced genes by IRF-3. Altered expression of SeV- regulated genes in U4C cells and U4C.2 cells, expressing high levels of IRF-3. (A, B)

Microarray data for the fold-induction of select SeV-regulated genes in U4C and U4C.2 cells. Genes are grouped as follows: IRF-3 enhanced, genes with augmented expression in IRF-3 overexpressing cells (induced more strongly in U4C.2); IRF-3 neutral, genes unaffected by cellular IRF-3 levels (induced at equivalent levels in both cell lines); IRF-3 repressed, SeV-induced genes negatively regulated by IRF-3 (induced in U4C, but not

U4C.2 cells). Quantitative RPA analysis of (C) Follistatin, (D) NOXA, and (E) A20 mRNA induction in U4C and U4C.2 cells before or after SeV infection.

141 (A) U4C U4C.2 30 nucleoside phosphorylase 15 MHC, class I, F (C) -+-+SeV toll-like receptor 4 5.0 Follistatin parathymosin mRNA 2.0 follistatin IRF-3 IRF-3 IP-10 enhanced Actin mRNA 1.5 ISG56 U4C U4C.2 FUT2 ISG54 (D) 1.0 (B) -+-+SeV 10 IKK alpha NOXA mRNA mannose-binding lectin 2

IRF-3 IRF-3 Actin 5.0 neutral NOXA mRNA U4C U4C.2 jun B

A20 -+-+ 2.0 Kallmann Syndrome I (E) SeV

PI3K, gamma A20 mRNA

IRF-3 IRF-3 apoptosis inhibitor I

1.5 repressed AP-2 gamma Actin apoptosis inhibitor 2 mRNA U4C U4C.2 similar to S. cerevisiae SSM4 1.0

142 expression (Fig. 5-6B). RPA for A20 mRNA confirmed the conclusions from the microarray analysis (Fig. 5-6E).

Although the RPA validated our original observation, we were concerned that the phenomenon might be restricted to the U4C cells from which all the above lines were derived. Because the U4C cells lacked Jak1 and were obtained by extensive mutagenesis of 2fTGH cells, it remained possible that the observed suppressing function of IRF-3 required either the absence of Jak1 or the presence of an unknown mutation in U4C cells.

If that were the case, our observation would have restricted significance. To test the generality of the observation, a new cell line was derived from the Wt HT1080 cells.

This line, 1080.10, expressed a much higher level of IRF-3 as compared to the parental cells (Fig. 5-7A). Confirming the results seen with the U4C derivative lines, A20 mRNA was induced strongly upon SeV infection in HT1080 cells, but only weakly in 1080.10 cells (Fig. 5-7B). The above series of experiments allowed us to conclude that the cellular abundance of IRF-3 can both positively and negatively influence the extent of induction of specific cellular genes in response to infection by SeV.

DISCUSSION

In this study, we focused our attention on identifying the signaling pathways responsible for inducing specific sets of cellular genes early after SeV infection. The cDNA microarray screening used was convenient for profiling the expression of a large number of genes simultaneously, but to ensure reliability we had to perform multiple repeats of the same screen and accept an arbitrary cut-off of at least two-fold change as

143

Figure 5-7. Negative regulation of SeV-induced A20 mRNA expression by IRF-3 in

Wt cells. (A) Western analysis of IRF-3 expression in HT1080 and 1080.10 cells. (B)

RPA comparing A20 mRNA induction in untreated (-) or SeV-infected (+) HT1080 and

1080.10 cells.

144 (A) IRF-3 Actin

HT1080 1080.10 (B) -+-+SeV

A20 mRNA

Actin mRNA HT1080 1080.10

145 significant. Consequently, our calculations of the number of genes induced in different

cell lines are likely to be underestimates. The overall gene induction patterns in different

SeV-infected cell lines are summarized in Table 5-1. This Table includes only the genes

whose expression was induced by at least 2-fold in wt 2fTGH cells. Following

classification of different induced genes into distinct groups, representative members

were selected for further analysis using more Northern blotting or RNase-protection

assays. The choice of these sentinel genes was dictated by their degree of inducibility and their abundance, because we could make much more reliable conclusions for genes that were highly induced and the corresponding mRNA levels were easily measurable.

Other groups have previously sought to profile some of the genes induced by

SeV. Strahle et al. employed SeV mutants to profile the effects of viral proteins on the induction of cellular genes in virus infected cells (Strahle, Garcin et al. 2003), and

Matikainen et al. looked at chemokine induction by SeV (Matikainen, Pirhonen et al.

2000). In the current study we see induction by SeV of many of the same genes previously identified, including IL-8, IP-10 and class I MHC genes. Further, we have been able to characterize the signaling pathways involved in the induction of not only these genes, but numerous other genes not formerly known to be induced by SeV.

Given the parameters of our experiment, there did not appear to be a role for

TLR3 in gene induction by SeV. The paramyxovirus replication cycle takes upwards of

10 hours, so it remains possible that at later times after virus infection TLR3-signaling is used by viral dsRNA for induction of the dsRNA-inducible genes that were not induced by SeV at 6h (Fig. 5-2) (Lamb and Kolakofsky 2001).

146 Our data showed that TLR3 was not required for gene induction early after SeV

infection, however viral dsRNA signaling through cytoplasmic sensors of dsRNA is

likely to be responsible for gene induction. PKR, Mda5, and RIG-I are three such

intracellular dsRNA-receptors that mediate dsRNA-signaling, and SeV has been shown

to require RIG-I for gene induction (Diebold, Montoya et al. 2003; Yoneyama, Kikuchi et

al. 2004). The apparent irrelevance of TLR3 in gene induction by SeV is supported by

our own findings in chapter 3 that SeV failed to induce ISG56 via TLR3 when RIG-I

signaling blocked, and by the fact that TLR3 -/- mice are not deficient in clearing

infections with other ssRNA viruses in vivo (Edelmann, Richardson-Burns et al. 2004).

Further, vesicular stomatitis virus (VSV), another single-stranded RNA virus has also

been shown to activate downstream signaling factors independent of TLR3 in TLR3-

deficent MEFs (tenOever, Sharma et al. 2004).

Our investigation of the role of IFN and the corresponding signaling pathway in

gene induction by SeV produced both anticipated and unanticipated results. In U4C cells, lacking functional Jak1, we observed that some genes were not induced, as compared to Wt cells. This result was expected because some genes might be induced by the autocrine action of IFN induced by SeV-infection. Many genes of this class, such as

OAS2 and MHC class I genes, are known IFN-stimulated genes (Fig. 5-3). Further, our findings with SeV correspond with those observed in Newcastle Disease virus-infected cells, where type I IFN signaling was required for OAS induction (Nakaya, Sato et al.

2001). However, other genes of the same family, such as ISG56 and ISG54, were induced even in U4C cells, confirming previous observations by us and others that these genes are directly induced upon infection with many viruses (Navarro, Mowen et al.

147 1998; Guo, Peters et al. 2000). The fact that these genes were induced in infected cells

independently of autocrine IFN signaling, is not unexpected, because IRF-3, which is

activated in the infected cells, can induce these genes directly. The surprising observation is that a cohort of genes was induced only in U4C cells, indicating that Jak1 or IFN-dependent signaling somehow negatively regulates the expression of these genes in Wt cells. The implications and the mechanism of this regulation need to be further explored in the future. To distinguish between the needs for IFN-dependent signaling and other putative functions of Jak1, we also analyzed the gene induction profiles in SeV- infected U2A cells and U3A cells (data not shown), which are also defective in IFN signaling because they lack functional IRF-9 and STAT1, respectively. The MHC class I genes, which were not induced by SeV in U4C cells, were not induced in U2A and U3A cells either, indicating that IFN-dependent signaling was needed for their induction. In contrast, the Jak1-repressed genes (Fig. 5-3), although induced in U4C cells, were not induced in U2A and U3A cells (data not shown), suggesting that it is the absence of Jak1 or another mutation specific to the cell line, and not the loss of IFN signaling that lead to

SeV-mediated induction of this class of genes in U4C cells. These results indicate that autocrine IFN signaling and Jak1 per se can influence the pattern of cellular gene- induction by SeV-infection.

For assessing the relative contributions of NFκB and IRF-3, two major transcription factors activated by SeV-infection, we took advantage of other mutant lines generated by us and our colleagues. The 2F-SR line was derived from 2fTGH cells by overexpressing a dominant negative mutant of IκB. NFκB cannot be activated by IL-1,

TNF-α, dsRNA, or SeV infection in these cells, as judged by electrophoretic mobility

148 shift assays (Peters, Smith et al. 2002). Consequently, these cells are functionally NFκB- null. As expected, we observed that many genes could not be induced by SeV in the absence of NFκB signaling (Fig. 5-4). Some of these genes, such as IKKα, NFκB p105, and A20, encode proteins that are involved in the NFκB signaling pathway itself.

Similarly, another set of genes was completely dependent on IRF-3 (Fig. 5). For identifying them, we used p2.1 cells, derived for U4C, which express little IRF-3 (Peters,

Smith et al. 2002). Many of these genes fall in the category of viral-stress inducible

genes, or VSIGs (Sarkar and Sen) because they can be independently induced by IFN,

dsRNA, or virus infection. Although the same cis-element, ISRE, present in their

promoter is responsible for their induction by the three agents, different trans-acting

factors mediate the process. Induction of these genes by IFNs is mediated by ISGF3, a

trimeric complex of IRF-9, STAT1, and STAT2. Induction by dsRNA and viruses, on

the other hand, is mediated by activated IRF-3, which also binds to ISRE. However, the

signaling pathways, triggered by dsRNA and viruses, that lead to IRF-3 activation

overlap only partially (Servant, Grandvaux et al. 2002). Thus it appears that VSIGs can

be induced by multiple inducers using multiple signaling pathways. There were also

genes, that were induced when either the NFκB pathway or the IRF-3 pathway were inoperative (Figs. 5-4 and 5-5). These genes probably get induced through other transcription factors activated by virus infection.

One such NFκB and IRF-3-independent gene is NOXA. NOXA, also known as

PMAIP1, originally discovered as a mediator of -induced apoptosis (Oda, Ohki et al.

2000), has been shown to be induced by a constitutively active IRF-3 mutant (IRF-3 5D)

(Grandvaux, Servant et al. 2002; Hiscott, Grandvaux et al. 2003). However in the

149 TABLE 5-1. Genes induced or repressed by SeV infection in various cell linesa Encoded Proteins Unigene ID Symbol 2fTGH U4C 2F-SR p2.1 U4C.2 2'-5'-oligoadenylate synthetase 2 (69-71 kD) Hs.414332 OAS2 I U U R U Activating transcription factor 3 Hs.460 ATF3 I I U I I Apoptosis Inhibitor 1 Hs.75263 cIAP1 I I U I U Apoptosis Inhibitor 2 Hs.127799 cIAP2 I I U U U Bcl-2 binding component 3 Hs.87246 BBC3 (PUMA) I I U U I Chemokine (C-X-C motif) ligand 10 Hs.2248 IP10 (CXCL10) I I U I I Colony stimulating factor 1 (macrophage) Hs.173894 CSF1 I I U I I Fibroblast growth factor 2 (basic) Hs.284244 FGF2 I U I U U Fucosyltransferase 2 Hs.292999 FUT2 IIIII General transcription factor IIF, polypeptide 1 Hs.68257 GTF2F1 I UUUU GTP cyclohydrolase 1 (dopa-responsive dystonia) Hs.86724 GCH1 I UUUU IFN-induced protein with tetratricopeptide repeats 1 Hs.20315 ISG56 I I I U I IFN-induced protein with tetratricopeptide repeats 2 Hs.293797 ISG54 I I I U I IFN-induced protein with tetratricopeptide repeats 4 Hs.181874 ISG60 I UUUU Inhibin, beta A (activin A, activin AB alpha polypeptide) Hs.727 INHBA I I U U U Inhibitor of kappa B, alpha Hs.81328 IκBα IIUII Interferon regulatory factor 1 Hs.80645 IRF-1 I I U U U Interferon regulatory factor 2 Hs.83795 IRF-2 I UUUU Interferon, alpha-inducible protein (clone IFI-15K) Hs.833 ISG15 I I I U I Interleukin 6 Hs.93913 IL-6 I I U U I Interleukin 8 Hs.624 IL-8 I I U I I Jun B proto-oncogene Hs.515157 JUNB I I U I U Macrophage Inflammatory Protein 1, alpha Hs.73817 MIP1α (CCL3) IIUII Mannose-binding lectin (protein C) 2 Hs.2314 MBL2 I I R I I MHC, class I, A Hs.181244 HLA-A I U I U I MHC, class I, C Hs.277477 HLA-C I U U I I MHC, class I, F Hs.110309 HLA-F I U U U I MHC, class I, G Hs.73885 HLA-G I U U U I Monocyte Chemotactic Protein 1 Hs.303649 MCP1 (CCL2) I UUUU Natural killer cell transcript 4 Hs.943 NK4 I U U U I Nexin, plasminogen activator inhibitor type 1 Hs.82085 PAI1 (SERPINE1) IIIII NFκB p105 Hs.83428 NFκB1 IIUII Nuclear factor (erythroid-derived 2)-like 3 Hs.22900 NFE2L3 (NRF3) I I U I U Nucleoside phosphorylase Hs.75514 NP I U I U I Phorbol-12-Myristate-13-Acetate-Induced Protein 1 Hs.96 NOXA (PMAIP1) IIIII Protein tyrosine phosphatase, non-receptor type 11 Hs.83572 PTPN11 (SHP2) I I U U I Protein tyrosine phosphatase, receptor type, Kappa Hs.79005 PTPRK I UUUU Short stature homeobox Hs.105932 SHOX I U U U I Similar to S. cerevisiae SSM4 Hs.380875 TEB4 IIIIU Stanniocalcin Hs.25590 STC1 I I I U U Syndecan 4 (amphiglycan, ryudocan) Hs.252189 SDC4 I I U I I Tight junction protein 2 (zona occludens 2) Hs.75608 TJP2 IIIII TNF receptor-associated factor 1 Hs.2134 TRAF1 I I R U U Transcription factor AP-2 gamma Hs.61796 ERF1 (TFAP2C) I U U I U Tumor necrosis factor, alpha-induced protein 3 Hs.211600 TNFAIP3 (A20) I I U I I UDP-N-acetyl-alpha-D-galactosamine:polypeptide N- Hs.130181 GALNT2 I U I U U acetylgalactosaminyltransferase 2 Vascular endothelial growth factor C Hs.79141 VEGFC I I I U U Viperin Hs.17518 CIG5 I U I U I CCCH type, antiviral 1 Hs.35254 ZC3HAV1 I I I U U a 'I' denotes induced, 'U' denotes unchanged, 'R' denotes repressed.

150 context of induction by SeV, IRF-3 does not appear to play an important role in NOXA

induction (Fig. 5-5 and 5-6 B, D). In the absence of STAT1, NOXA induction through

the p53 response element in its promoter is impaired (Townsend, Scarabelli et al. 2004);

likewise SeV fails to induce NOXA in STAT1-/- U3A cells (data not shown). This

finding suggests that SeV may induce a subset of genes through the IRF-3-, NFκB-, and

IFN-independent activation of STAT1 or p53. It also raises the possibility that the

constitutively active IRF-3 5D mutant and virally activated IRF-3 do not induce

equivalent sets of genes. Active SeV infection causes additional phosphorylation of IRF-

3 beyond that required for the IRF-3 mediated induction of ISG56 (Collins, Noyce et al.

2004). These additional modifications to IRF-3 might very well impair its ability to

mediate the induction of other genes, such as NOXA.

Further explorations of the role of IRF-3 in gene induction by SeV produced

interesting results. Using two related cell lines that expressed different levels of IRF-3,

we could demonstrate that, although some genes were induced equally well in both,

others were induced better in U4C.2 cells, which express a higher level of IRF-3 (Fig. 5-

6A). Thus, it seems that at least in the 2fTGH cell lineage, IRF-3 expression is the

limiting factor and the degree of gene induction can be modulated easily by changing the cellular concentration of IRF-3. The above conclusion was confirmed by a more quantitative assay using a series of cell lines expressing increasing amounts of IRF-3

(Fig. 5-7). Similar results were obtained with the same cell lines when the inducer was dsRNA, not SeV (Peters, Smith et al. 2002). Among the genes induced by SeV in U4C.2 cells were the MHC genes, which were not induced in U4C cells but were induced in

2fTGH cells (Figs. 5-3 and 5-6). Thus, a high level of IRF-3 could mediate SeV-

151 induction of genes which otherwise required IFN-signaling. Other genes, such as

Follistatin, were induced by SeV very strongly in U4C.2 cells, whereas the induction was minimal in U4C or 2fTGH cells (Fig. 5-6A and C).

The most surprising observation was that a higher level of IRF-3 expression negatively affected the degree of induction of some genes (Fig. 5-6B and 6E). Using the

A20 gene as the sentinel for this group of genes, the same trend was observed in Wt

HT1080 cells by overexpressing IRF-3 (Fig. 5-7). This proved that repression of SeV- induced gene expression by IRF-3 was not restricted to just U4C cells and their derivatives. However, the mechanism by which this occurs is not clear, and merits further investigation.

152 CHAPTER 6

INHIBITION OF INDUCED GENE EXPRESSION BY IRF-3

ABSTRACT

Interferon regulatory factor 3 (IRF-3) belongs to a family of transcription factors which regulate many immune and stress-related genes through an interferon stimulated response element (ISRE). IRF-3 has been shown to mediate gene by dsRNA induction through both TLR3 and RIG-I. In the previous chapter, the discovery, by cDNA microarray, of a novel role for IRF-3 in the repression of a subset of Sendai virus (SeV) induced genes was discussed. Here, that work is expanded upon to show that expression of A20 mRNA in SeV-infected cells is inversely proportional to the amount of IRF-3 those cells produced. IRF-3 repression of A20 induction by dsRNA signaling through TLR3, and

TNFα was also observed. Repression of A20 mRNA had the functional consequence of repressing even basal A20 protein expression. Further, transcriptional activation of IRF-

3 did not appear to be required for it to repress A20 expression. The anti-apoptotic genes c-IAP1, c-IAP2, NAIP and TRAF1 were verified to be repressed by IRF-3, similarly to

A20. Analysis of the promoters of all five genes revealed NFκB sites in all, but no

ISREs. However, IκB release, p65 phosphorylation and induction of other NFκB dependent genes did not appear to be inhibited by IRF-3.

153 INTRODUCTION

The transcription factor interferon regulatory factor 3 (IRF-3) is critical for the

induced transcription of many genes in response to stresses such as viral infection. IRF-3

-/- mice are more susceptible to viral infection than wild-type mice (Sato, Suemori et al.

2000). Recognition of viral dsRNA is one of the key mechanisms by which a cellular

response is initiated, and IRF-3 has been shown to be especially important for gene

induction through the dsRNA receptors TLR3 and RIG-I. In the absence of IRF-3, many

genes fail to be induced through these receptors. These include the cytokine IFNβ along with many other genes shown to have direct anti-viral effects, such as FUT2 and ISG56.

In chapters 4 and 5, we identified IRF-3-dependent genes induced by dsRNA through TLR3 or SeV through RIG-I by comparing mutant cell lines either under- or over-expressing IRF-3, by using cDNA microarrays. We were able to identify IRF-3 dependent genes by their augmented expression in IRF-3 overexpressing cells. Quite unexpectedly, overexpression of IRF-3 also inhibited the SeV-induced expression of mRNA a cohort of genes, including the attenuator of dsRNA signaling, A20. This is the first example of gene down-regulation by IRF-3, which until now has been thought to only induce gene transcription.

Although originally identified as a TNFα-induced gene, A20 is also induced through toll-like receptors and virus infection. In our microarray studies it consistently scored as one of the most strongly induced genes. Most import for the current study, it was the gene whose expression following SeV infection was most dramatically impaired.

We therefore used it as a model for studying the mechanism of IRF-3 mediated gene

154 repression. In addition to being a good model, A20 protein has potential functional implications for viral infection. A20 functions to inhibit both NFκB and IRF-3 mediated signaling responses (Saitoh, Yamamoto et al. 2005). In this way its induction serves as a means of negative feedback regulation for stress induced cellular responses. In addition,

through its negative regulatory role, A20 also has anti-apoptotic effects.

Here the initial findings on the mechanism by which IRF-3 is able to negatively

regulate gene expression are reported. Using a series of cell lines expressing increasing

levels of IRF-3, we demonstrated that the degree of induction of A20 mRNA, upon SeV

infection, was inversely proportional to the cellular level of IRF-3, whereas that of ISG56

mRNA was directly proportional. Our results indicate that, in addition to A20, IRF-3

inhibits the induced expression of a number of genes, including TRAF1, c-IAP1, c-IAP2

and NAIP. IRF-3 does this through a novel mechanism that does not involve its

activation, binding to a conventional ISRE site, or a general inhibition of NFκB

signaling. Further, we show that impairment of A20 mRNA expression results in an

inhibition of protein expression as well. Finally we identify different physiological

settings where regulation of A20 and other IRF-3 inhibited genes may play a functional

role in the cellular response to virus.

155 RESULTS

Expression of A20 is inhibited by IRF-3. To test the possibility that IRF-3 was acting as a negative regulator of SeV-induced gene expression in a dose-dependent manner, five cell lines, all derived from the U4C line, expressing increasing levels of IRF-3 protein were used (Fig. 6-1A). The cell lines were infected with SeV, and the levels of the ISG56 and

A20 mRNAs were measured by RPA before and after virus infection. Actin mRNA levels were used for normalization of the data. Neither mRNA was detectable in any uninfected cell line. ISG56 mRNA was barely induced in P2.1 cells, which expressed the lowest level of IRF-3, and its induction levels increased with increasing levels of IRF-3 expression. Conversely, A20 mRNA was most strongly induced in P2.1 cells and very poorly induced in cells expressing high levels of IRF-3 (Fig. 6-1B).

Since the results showed that induced A20 mRNA expression was inhibited by high cellular levels of IRF-3, we wanted to determine if this inhibition was also evident at the protein level. Expression of A20 protein was analyzed by Western blot in the wild- type and IRF-3 overexpressing cells, U4C and U4C.2, respectively, at different times following SeV infection. As can be seen in figure 6-2, A20 is induced by SeV in U4C cells (panel A) but not in U4C.2 cells (panel B), as expected based on A20 mRNA expression in these cells. Interestingly, basal expression of A20 can be detected in U4C cells, but not in U4C.2 cells, suggesting that even constitutive A20 expression may by inhibited by IRF-3. In contrast to A20, p56 induction by SeV in U4C.2 cells is much higher and more sustained than it is in U4C cells. Activated IRF-3 is degraded in SeV- infected cells. In U4C cells all endogenous IRF-3 is gone by seven hours, in contrast to

156

Figure 6-1. Modulation of A20 mRNA and ISG56 mRNA expression by cellular

levels of IRF-3. Cell lines, derived from U4C and p2.1 cells, expressing different levels

of IRF-3 protein were infected with SeV and analyzed for their expression of A20 mRNA and ISG56 mRNA. (A) Western blot showing the relative levels of IRF-3 expression in the different cell lines relative to actin. (B) Percent maximum fold-induction of A20 mRNA (white) and ISG56 mRNA (black) in cells six hours after infection, normalized to actin mRNA expression as determined by RPA.

157 (A) p2.1 U4C p2.1.6 U4C.2 p2.1.17 IRF-3

Actin (B)

100 A20 mRNA ISG56 mRNA 80

60

40 mRNA expression 20

0 p2.1 U4C p2.1.6 U4C.2 p2.1.17 Cell lines

158

Figure 6-2. Effect of IRF-3 on expression of A20 protein. Western blots for p56, A20 and IRF-3 expression in whole cell lysates from U4C cells (panel A), or U4C.2 cells

(panel B) following infection with SeV for 7, 15 or 28 hours.

159 (A) U4C U4C.2 +SeV (h) 071571528 0 p56

A20

IRF-3

(B) U4C.2 +SeV (h) 0 71528 p56

A20

IRF-3

160 U4C.2 cells where IRF-3 expression remains elevated due presumably due to continued

high levels of expression of exogenous IRF-3 mRNA.

IRF-3 inhibits SeV induced expression of anti-apoptotic genes. In addition to A20, many other SeV induced genes whose expression is repressed by IRF-3 have anti-apoptotic functions, including c-IAP1 and c-IAP2. As a result, we sought to verify that induction of these genes by SeV was impaired by IRF-3, as well as to determine if any other genes involved in apoptosis were similarly regulated. With the help of Dr. Kristi Peters we compared the expression of a panel of anti-apoptotic genes in SeV infected U4C and

U4C.2 cells by RNAse protection assay. In complete agreement with our microarray data, TRAF1, c-IAP1 and c-IAP2 were all expressed in SeV infected U4C cells, but not

U4C.2 cells (Fig. 6-3). In addition NAIP, which was not among the genes on our microarray, also showed the same behavior. Both cell lines expressed low basal levels of

TRPM2 and TRAF3 mRNA, which were up-regulated upon infection with SeV. Both messages appeared to be expressed slightly better in U4C compared to U4C.2 cells, but the difference was not nearly as great as seen with TRAF1 and c-IAP2. XIAP, TRAF2, and TRAF4 mRNAs were all expressed constitutively and were unaffected by SeV infection.

Inhibitory effects of IRF-3 not dependent on activation. SeV not only activates both IRF-

3 and NFκB, but also changes the cell physiology in many other ways over the course of infection. We were curious as to whether IRF-3 inhibited A20 expression induced by stimuli other than virus. To address this question we looked at A20 mRNA induction by

161

Figure 6-3. Effect of IRF-3 on expression of anti-apoptotic genes. The expression of mRNAs from multiple anti-apoptotic genes was assessed by RPA in U4C and U4C.2 cells. RNA from U4C and U4C.2 cells infected with SeV for six hours, along with RNA from uninfected U4C.2 cells was isolated and assessed by RPA for expression of multiple anti-apoptotic genes. In specific expression of XIAP, NAIP, TRPM2 , c-IAP1 and 2, as well as TRAFs 1-4 was determined. As a control L32 and GAPDH expression are shown. Asterixes (*) indicate mRNAs clearly repressed upon IRF-3 overexpression.

162 _ + + : SeV

XIAP TRAF1 * TRAF2 TRAF4 NAIP *

CIAP1 * CIAP2 *

TRPM2

TRAF3

L32

GAPDH

U4C.2 U4C

163 TNF-α and dsRNA signaling via TLR3 in U4C and U4C.2 cells. Like SeV, dsRNA can activate IRF-3, while TNFα cannot. The latter can be seen in figure 6-4A, where treatment with TNFα failed to induce p56 expression in either U4C or U4C.2 cells.

When A20 mRNA expression was analyzed by RPA, only a two-fold decrease in A20 levels was seen between U4C and U4C.2 cells following TNF-a treatment (Fig. 6-4B). In contrast, a 10-fold decrease is seen following Sendai infection. Similar to Sendai, a seven-fold difference in A20 levels is seen between cell lines following dsRNA treatment, which like Sendai, can activate IRF-3. Taken together these results suggest

that IRF-3 can inhibit A20 expression induced by multiple stimuli and is not a virus-

specific phenomenon. Further, activation of IRF-3 it is not required to inhibit A20 induction, although it may serve to augment the degree of inhibition.

IRF-3 represses induction of a subset of NFκB dependent genes. As shown in figure 6-

1B, the regulation of ISG56 and A20 induction following SeV infection is very different.

While ISG56 is induced through IRF-3, the promoter of the A20 gene contains multiple

κB sites, but no consensus ISRE (Laherty, Perkins et al. 1993). Accordingly, A20 induction by SeV was identified by cDNA microarray as NFκB-dependent (Fig. 5-4). In figure 6-5A, the contrasting dependence of A20 on NFκB, and ISG56 on IRF-3 for induction by SeV is highlighted by RPA. In 2F-SR cells, which express the IκB super repressor and cannot signal through NFκB, A20 is not induced while ISG56 expression is unaffected. By comparison, ISG56 expression is impaired in p2.1 cells, which express little IRF-3, while A20 expression is augmented slightly.

164

Figure 6-4. A20 induction by dsRNA and TNFα repressed by IRF-3. (A) A20

mRNA expression was measured in U4C cells (odd lanes) or U4C.2 cells (even lanes), 6

hours after infection with SeV (lanes 3 and 4), addition of dsRNA to the media (lanes 5

and 6) or treatment with TNFα (lanes 7 and 8). (B) p56 expression in U4C and U4C.2 cells following addition of dsRNA to the media or treatment with TNFα for six hours was assessed by western blot.

165 (A) 12 34 5678

A20 mRNA

actin mRNA

Untreated SeV dsRNA TNF-α

(B)

dsRNA TNF dsRNA TNF p56

U4C U4C.2

166 In addition to A20, TRAF1, NAIP, and c-IAPs 1 and 2 are all known to require

NFκB for induction. It was possible therefore, that IRF-3 overexpression was inhibiting

the activation of NFκB by competing for signaling components required by both for activation. Examination of NFκB activation in U4C.2 cells made it unlikely that this was the case. Analysis of NFκB release by gel shift assay revealed that NFκB is activated equally as well, if not better, in U4C.2 cells than in U4C cells (Fig. 6-5B).

Phosphorylation of p65 on serines 276 and 536 is another step that occurs during NFκB activation. Western blotting using phosphorylation site-specific antibodies against both residues revealed both Ser536 and Ser276 were phosphorylated equally as well in U4C and U4C.2 cells (Fig. 6-5C). Together these results suggested that IRF-3 was not inhibiting NFκB activation.

Since it was still possible that the full transcriptional activation of NFκB could be inhibited without affecting its release from IκB or serine phosphorylation, we wanted to determine if any NFκB dependent genes were induced by SeV in IRF-3 overexpressing cells. To answer this question we referred to our previous microarray data. Beginning with 50 genes originally determined to be induced by SeV in U4C cells and at least one other cell line, we divided them further based on whether or not they were also induced by SeV in U4C.2 cells. Previously we had separately identified genes induced in NFκB

signaling deficient cells by microarray. The overlap between genes determined to be

NFκB-dependent with those expressed or not expressed in SeV infected U4C.2 cells is shown in a Venn diagram as well as in groups in figure 6-6. Twenty-eight of the 49 genes induced in U4C cells were induced in U4C.2 cells, meaning that IRF-3 did not impair their expression. Of these 28 genes, 20 were NFκB dependent, thus providing

167

Figure 6-5. NFκB activation in IRF-3 overexpressing cells. (A) RPA showing the expression of A20 and ISG56 messages in uninfected or SeV-infected cells. The expression of both A20 and ISG56 was compared in wild-type (2fTGH), NFκB null (2F-

SR), IFN signaling null (U4C), IRF-3 depleted (p2.1) cells. (B) NFκB release following

SeV infection was measured by EMSA. Whole cell lysates from U4C and U4C.2 cells were collected following SeV infection for the times indicated. (C) Phosphorylation of p65 on serines 276 and 536 following SeV infection was compared in U4C and U4C.2 cells by western blot using phosphorylation site-specific antibodies. Total p65 expression is shown as a control.

168 (A) _ _ _ _ _ SeV + + + + + A20 mRNA

ISG56 mRNA

actin mRNA

2fTGH 2F-SRU4C P2.1 P2.1.17

(B) U4C U4C.2 0 1.5 3 6 0 1.5 3 6 hr +SeV

NFκB

(C) U4C U4C.2 +SeV (h) 00.75 1.5 3.0 0 0.75 1.5 3.0 Ser276

Ser536

p65

169 evidence that IRF-3 overexpression did not universally inhibit NFκB signaling.

However, 17 of the 21 genes not induced by SeV in U4C.2 cells were also NFκB- dependent. As expected, A20, TRAF1, c-IAP1 and c-IAP2 were among these. The remaining genes, MARCH6, AP-2 gamma, VEGFC and FGF2, were apparently repressed by IRF-3, yet induced independently of NFκB. The repression of these four genes suggests that IRF-3 may be suppressing the expression of non-NFκB-dependent genes as well.

Promoter analysis of IRF-3 repressed genes. To gain more insight into whether IRF-3 repressed gene induction by inhibiting their transactivation by NFκB, we compared the promoters, to 700 base pairs upstream of the transcription start site, of our five RPA- verified, NFκB-dependent genes and four unverified NFκB independent genes. The basic core ISRE binding motif for IRF-3 is GAAANNGAAA. None of the nine promoters contained such a motif. The closest ISRE sites in A20 were 5kb upstream and

4kb downstream of the transcription start site. Thus, any repressive effects of IRF-3 on gene induction are probably not exerted through its direct interaction with the A20 promoter.

As expected, all five NFκB-dependent genes had putative κB binding sites within

400 base pairs upstream of their transcription start sites. A20 had three putative κB sites within 700 base pairs of the transcription start site, at -236, -248 and -455. The two κb sites closest to the transcription start site have already been shown to be required for A20 transcriptional induction. Strikingly, the promoters of both TRAF1 and c-IAP2 contained

κB sites identical to the proximal A20 κB site, GGGGATTTCC. While neither the two

κB sites in promoter of c-IAP1 nor the two in the promoter of NAIP matched exactly,

170

Figure 6-6. Dependence on NFκB of IRF-3 inhibited genes. Venn diagram showing the dependence on IRF-3 induced and repressed genes on NFκB. Upper left circle represents SeV induced gene that are repressed by IRF-3 expression. Upper right circle is NFκB dependent genes. Bottom circle contains genes induced through IRF-3. Genes are grouped and labeled according to which category they fell into on the Venn diagram.

171 IRF-3 repressed, NF-kB IRF-3 repressed, NF-kB DEPENDENT genes INDEPENDENT genes A20 Nf e2-like3 MARCH6 TRAF1 Integrin A2 AP-2 gamma CIAP1 SCYA3 Vasular endothelial growth CIAP2 Kallmann Syndrome factor C junB Inhibin beta A Fibroblast growth factor 2 ATPase,Na /K IRF-1 transporting, gamma 1 FLJ23329 Pol gamma Ptp, non-receptor 11 Stat induced stat inhib 2 Noonan PI3K gamma

NOT INDUCED NFkappaB IN U4C.2 DEPENDENT

4 17 0

0 0 20 NF-kB DEPENDENT genes, not repressed by IRF-3 MBL2 8 Syndecan 4 Leucine -rich, glioma inactivated 1 POU domain, class 3, TF4 BCL -2 binding component 3 INDUCED IN U4C.2 IFI-6-16 Colony stimulating factor 1 NF-kB INDEPENDENT genes, IkB alpha not repressed by IRF-3 P105 IP10 ISG54 ISG_03B21 ISG15 Short stature homeobox ISG56 IL-15 receptor, alpha NOXA LIM domain only 2 FUT2 PTPnon -receptor type I Thrombospondin 1 IL-8 Serine proteinase inhib clade E Matrix metalloproteinase 13 (nexin ) GRO3 oncogene TJP2 Complement component 2 ATF3

World: induced > 2- fold in U4C and at least 1 other cell line

172 they were similar. Figure 6-7 shows an alignment of the κB sites from all five promoters.

A promoter search identified two putative κB sites in the promoter of MARCH6, which are shown in figure 6-7. A literature search revealed NFκB along with p38 is needed for

VEGFC expression (Tsai, Shiah et al. 2003). However no putative κB sites were found in the promoters of FGF2 or AP-2 gamma (Li, Wang et al. 2002). These results suggest that IRF-3 repressed, NFκB-dependent genes do share a more conserved κB binding site, but also indicate that IRF-3 repression may not be dependent on that site.

Gene repression by IRF-3 is cell type- and inducer-dependent. To see if IRF-3 repressed gene induction in other cell types, we compared A20 expression following SeV infection in 293 and wt11 cells with two derived lines overexpressing IRF-3, z3p and 333p, respectively (Fig. 6-8A). Surprisingly, A20 mRNA was expressed at higher levels in IRF-

3 overexpressing cells than it was in their parental cell lines (Fig. 6-8B). Over 2.9-fold more A20 message was expressed in SeV infected z3p cells than in 293 cells, although only 20% more A20 was expressed in 333p cells than wt11 cells. However, when ISG56 expression was also looked at in these same four cell lines, only a 4.5-fold and a 2-fold increase in expression was seen, respectively, between 293 or wt11 cells and their IRF-3 overexpressing counterparts (Fig. 6-8C). This increase was much lower than that observed by overexpressing IRF-3 in the HT1080 background, suggesting that IRF-3 expression in 293 cells may normally be close to saturating levels.

Although dsRNA is a much poorer inducer of A20 than Sendai virus, both exhibit similar behavior in U4C and U4C.2 cells. However, our microarray data suggested that unlike SeV, added dsRNA was able to induce A20 in IRF-3 overexpressing p2.1.17 cells

173

Figure 6-7. Alignment of NFκB binding sites in IRF-3 repressed genes. Shown here

are confirmed and putative NFκB binding sites from the promoters of IRF-3 repressed genes. To the right of each sequence is given the approximate position of the site relative to the transcription start site of the gene it regulates. A consensus site was generated based on the residues conserved throughout all five genes or present in at least five of nine sites. Putative κB sites are also listed for MARCH6 and VEGFc, two of the four genes which were indicated to be NFκB independent by microarray.

174 A20 GGGGATTTCC -230 GGGACTTTCC -245

TRAF1 GGGGATTTCC -3 AGGGCTTTCC -80 c-IAP2 GGGGATTTCC -170

c-IAP1 GGGCGTTTCC -372

NAIP TGGAGATTCC -298

CONSENSUS: GGGGNTTTCC

MARCH6 GGGAGCTCCC -70 GGGAGGTTCC -380

Basic consensus NFkappaB binding site: GG(A/G)NN(C/T)(C/T)CC

175

Figure 6-8. A20 repression by IRF-3 is cell type specific. (A) Western blot for IRF-3 expression. Wt11 cells were derived from HEK293 cells and express exogenous TLR3. z3p and 333p cells were derived from HEK293 and wt11 cells respectively transfected with an IRF-3 expression plasmid. (B) Comparison of A20 mRNA expression in untreated and SeV infected cells. (C) Expression of ISG56 mRNA in SeV infected cells was also measured by RPA.

176 (A) WT11 z3p 333p IRF-3

Actin

(B) 293 z3p WT11 333p -+-+- + -+Sendai A20 mRNA

Actin mRNA

(C) 293 z3p WT11 333p

561 mRNA

Actin mRNA

177 as well as it did in p2.1 cells. Testing this by RPA confirmed that A20 is still induced by

dsRNA in p2.1.17 cells at levels comparable to those seen in p2.1 cells, while in the same

experiment its expression was reduced from U4C to U4C.2 cells (Fig. 6-9A).

Unlike U4C cells, p2.1 cells normally express little IRF-3. This is not due to an

actual defect in IRF-3 itself, but an unknown mutation that causes its rapid turnover. If

SeV infection were modifying IRF-3 in some way that dsRNA signaling through TLR3

did not, it might account for the differences seen with the two inducers in the p2.1

background. IRF-3 modification by SeV and added dsRNA was therefore compared by

two-dimensional gel electrophoresis. As can be seen in figure 6-9B, IRF-3 from SeV-

infected cells is shifted higher and more to the left, indicating the presence of additional

modification compared to IRF-3 activated by dsRNA through TLR3.

Differential expression of IRF-3. IRF-3 is expressed constitutively and cannot be

induced (Lowther, Moore et al. 1999). Since the ability of IRF-3 to repress gene

induction is dependent on its expression, it is unclear when, under physiological

circumstances, such an effect would occur. One possibility is that IRF-3 is differentially

expressed in different cell types, as we saw with HT1080 and 293 cells. To address this

possibility, expression of IRF-3 mRNA in different tissues was referenced from a microarray gene expression database (Su, Cooke et al. 2002). Of the tissues looked at, more than twice as much IRF-3 mRNA was expressed in the blood and lymph nodes than in heart and brain (Fig. 6-10). Among individual cell types, immune cells expressed far more IRF-3 than other cells, with T-cells expressing well over ten-fold more IRF-3

mRNA than skeletal muscle. While mRNA expression is not necessarily indicative of

178

Figure 6-9. Differential phosphorylation of IRF-3 by SeV and added dsRNA. (A)

A20 mRNA expression in SeV infected cells was analyzed by RPA. RNA isolated from parental cell lines U4C and p2.1 were compared with their IRF-3 overexpressing

counterparts, U4C.2 and P2.1.17. (B) IRF-3 phosphorylation was analyzed by two-

dimensional gel electrophoresis. Whole cell lysate from untreated HT1080 cells, and

nuclear lysates from SeV infected cells and cells treated by adding dsRNA were

collected, subjected to two-dimensional electrophoresis, followed by western blotting for

IRF-3. pH increases from left to right in the first dimension.

179 (A) U4C U4C.2 P2.1 P2.1.17 A20 mRNA

Actin mRNA

(B)

Untreated

+ dsRNA

+ SeV

180

Figure 6-10. Expression of IRF-3 varies between tissues and cell types. (A)

Differential expression of IRF-3 mRNA between tissues and cell types as determined by affymetrix array. Data for expression of IRF-3 mRNA was taken from the GeneAtlas

Ver2 dataset and represent the average of multiple samples analyzed on the human

U133A affymetrix array and normalized by gcRNA.

181 Blood Bonemarrow Brain Heart Kidney Liver Lung Lymphnode Ovary Testis Thymus

Adipocytes CardiacMyocytes PancreaticIslets SkeletalMuscle SmoothMuscle

CD8+T cells CD4+T cells CD19+B cells CD56+NKCells BDCA4+Dentritic_Cells CD14+Monocytes

0 1000 2000 3000 4000 5000

Relative IRF-3 mRNA expression

182 protein expression, these results strongly suggest that IRF-3 is differentially expressed

between cell types.

Most of our work on IRF-3 mediated gene repression has been done using cells

expressing far more IRF-3 than normal. However during the course of SeV infection

cellular IRF-3 was completely degraded (Fig. 6-2A and B). In line with this, we

previously observed an increased A20 mRNA expression in p2.1 cells compared with

their parental line, U4C (Fig 6-1). The same effect was seen in HT1080 cells. When

IRF-3 expression was silenced by siRNA, SeV induced the expression A20 mRNA

almost twice as well.

DISCUSSION

Here it is reported that SeV-induced expression of A20, TRAF1, c-IAP1, c-IAP2 and NAIP mRNAs are repressed by IRF-3 in a manner inversely proportional to its cellular level (Fig 6-1 and 6-3). For A20, and probably the other messages as well, impaired mRNA expression results in decreased protein expression (Fig. 6-2). Further, repression of A20 induction was not unique to SeV, but was seen with TNFα and dsRNA signaling through TLR3 as well.

Using cells expressing the IκB super-repressor, we found that induction of all these genes was dependent upon NFκB. Since some common signaling components are required for both NFκB and IRF-3 activation, we thought it possible that IRF-3 and

NFκB competed for the same upstream activating factors. Surprisingly, the ability of

183 IRF-3 to impair A20 mRNA expression extended to situations where IRF-3 is not

normally activated as well. We tested the ability of TNF-α, which should not activate

IRF-3, to induce A20 expression in IRF-3 overexpressing cells. While A20 expression in response to TNFα was impaired in these cells, it was not inhibited as much as with SeV

and dsRNA (Fig. 6-4). This suggests that IRF-3 transcriptional activation augments, but

is not required for its negative regulation of A20 expression.

In the context of TLR4 signaling, Wietek et al. have shown that the p65 subunit of

NFκB co-localizes with IRF-3 on the ISRE and helps to promote gene induction (Wietek,

Miggin et al. 2003). While the authors did not observe a requirement for p65 in TLR3- mediated signaling, it is possible that in conditions of excess IRF-3, p65 is being sequestered by IRF-3 and prevented therefore from inducing transcription through binding to κB sites. However, closer analysis of our microarray data revealed that the

induced expression of many NFκB-dependent genes were not decreased following IRF-3

overexpression, nor was there a decrease in NFκB activation as determined by EMSA or

p65 phosphorylation (Fig. 6-5B and C). In fact, it appears that the NFκB response might

be more robust in the IRF-3 overexpressing cells.

These results provide some hints, but no real insight into the mechanism of IRF-3

mediated message repression. Because only a subset of NFκB-regulated genes have their

expression repressed by IRF-3, it is possible that specificity is determined by κB binding

sites within the promoters of IRF-3 repressed genes. Even single nucleotide differences

within κB binding sites can affect their regulation by NFκB (Natoli 2004). A20, TRAF1

and cIAP2 all contain identical κB binding sites, and are the three genes whose induced

expression is most inhibited by IRF-3 (Fig. 6-7 and 6-3). The consensus NFκB binding

184 site is GG(A/G)NN(C/T)(C/T)CC. We were able to construct a more specific κB site

from the IRF-3 repressed genes, gGGgNtTTCC (lower case nucleotides were not 100%

conserved). Another gene which is NFκB regulated and repressed by IRF-3 is c-jun;

analysis of its promoter indicates that its κB site meets the criteria for an IRF-3 repressed

gene, GGGGCTTTCC (Mathas, Hinz et al. 2002). Conversely the two κB sites of IκBα, which is not repressed by IRF-3, do not meet the criteria, GGAAATTCCC and

GGGGAAGTCC (Leung, Hoffmann et al. 2004).

It is still of concern that two IRF-3 repressed genes, FGF2 basic and AP-2 gamma appeared to be NFκB-independent in their regulation. However, that these genes are indeed repressed by IRF-3 must first be confirmed by another method besides microarray.

Transcriptional repression of gene induction by IRF-3 could occur by a number of different mechanisms. IIRF-3 could directly bind within repressed genes to inhibit their induction. If this is the case, transctiptionally-inactive IRF-3 should be detected in the nuclei of TNFα-treated cells bound to A20 and other genes.

Alternatively, IRF-3 could sequester a co-factor needed for transcription of IRF-

3-repressed gene. One such example is GRIP1 which has recently been shown by Reily, et al. to interact with IRF-3 (Reily, Pantoja et al. 2006). Although the authors report that

GRIP1 binding to IRF-3 positively affects gene induction through ISRE elements, GRIP1 also functions with (GR) as a selective co-repressor of AP1- and

NFκB-dependent gene induction. Increased IRF-3 expression leads to decreased GR- mediated gene repression, which is restored by GRIP1 overexpression. It is possible

185 IRF-3 binding to GRIP1 or a co-activator required for A20 induction could be involved in

the mechanism by which IRF-3 represses gene expression.

Another completely different possibility, which we have not ruled out, is that IRF-

3 may function to repress mRNA expression post-transcriptionally by affecting message

stability. The mRNA transcripts for many IRF-3 repressed genes have AU-rich 3’

untranslated regions, indicating they are normally short-lived. However, if IRF-3

expression selectively decreases message stability, it must be independent of induced

transcriptional activation, since in IRF-3 overexpressing cells A20 expression in response to TNFα is repressed despite the lack of IRF-3-induced transcription.

The ability of IRF-3 to repress gene induction appears to be somewhat cell-type specific as well. When IRF-3 was overexpressed in the HEK293 background, there was a gain instead of a loss in A20 expression after SeV infection (Fig. 6-8). However the usual increase in ISG56 induction that also accompanies IRF-3 overexpression did not occur either. This led us to believe that whatever differs between HEK293 and HT1080 cells affects IRF-3 signaling overall and not just gene repression.

The p2.1 cell line represents another case where IRF-3 repression of A20 was altered. In the p2.1 background, A20 mRNA expression by SeV, but not dsRNA was impaired by IRF-3 overexpression. This was unexpected, since in the parental line of p2.1, U4C, overexpression of IRF-3 inhibited A20 mRNA expression in response to both dsRNA and SeV. P2.1 cells have a defect which increases the turnover of IRF-3, causing them to express less IRF-3 than U4C cells. When we looked at IRF-3 modification by

SeV and dsRNA signaling via TLR3 by 2-dimensional electrophoresis we saw a differential phosphorylation. IRF-3 from SeV infected cells was more negatively charged

186 than in dsRNA-treated cells, indicating that it had additional phosphorylation or other modification (Fig. 6-8B). It is possible that the additional modification of IRF-3 by SeV accounted for the difference in A20 expression seen in SeV and dsRNA treated p2.1.17 cells.

We have shown here that the ability of A20 to be induced by virus infection,

TNFα and dsRNA is inversely dependent on the level of expression of IRF-3 (Fig. 6-4).

While IRF-3 is constitutively expressed and is not induced through an ISRE, its expression level can vary greatly between cell types. Immune effector cells, such as T cells and dendritic cells, naturally express far more IRF-3 than cells with low turnover, such as skeletal muscle (Fig. 6-10). The same is true at the tissue level. IRF-3 expression in the brain is one-third of that seen in the lymph nodes.

A20 is normally expressed within cells at low levels and functions to negatively regulate NFκB and IRF-3 signaling. Its further induction upon stimulation is critical in order to prevent a sustained, potentially pathological, signaling response. Therefore the potential implications of its regulation by IRF-3 are as follows: When stimulated by a pathogenic or immune signal, such as viral infection, immune cells expressing higher levels of IRF-3 will express less A20, and therefore have sustained NFκB and IRF-3 signaling responses. Conversely A20 would be expressed better in virus-infected skeletal muscle or brain cells, since they express relatively little IRF-3. As a result, these cells would mount an attenuated, less pathological signaling response. Such a model makes sense, in order to protect sensitive tissues from damage.

It is also possible that viruses might use this system of feedback to their advantage. Over the course of infection by viruses, such as SeV, cellular IRF-3 is

187 degraded, which would not only directly impair sustained gene induction by IRF-3, but also lead to increased A20 expression. This would be advantageous to viral replication, by serving to further impair the cellular anti-viral response. As we have shown, A20 mRNA expression is increased in cells with less IRF-3 (Fig 6-10B). Since IRF-3 represses the anti-apoptotic genes TRAF1, c-IAP1, c-IAP2 and NAIP in addition to A20, these genes could be expressed more in cells where IRF-3 has been degraded as well.

188 CHAPTER 7

DISCUSSION AND PERSPECTIVE

Although the results presented within chapters 3 through 6 of this dissertation all have their individual implications, together they provide some interesting perspectives on dsRNA signaling and viral-PAMP signaling, in general.

BASAL VERSUS INDUCED dsRNA RESPONSIVENESS. U3A cells are unable to respond to extracellular dsRNA due to a defect in basal TLR3 expression. However, this defect does not prevent IFN from inducing dsRNA responsiveness by up-regulating

TLR3 expression. This has a number of potential physiological implications:

Under resting conditions, most non-immune cell types have very low basal TLR3 expression, and respond poorly to extracellular dsRNA. As a result, these cells will not be overly sensitive to non-pathological dsRNA in their surroundings. Upon stimulation with IFN, generated through either the adaptive or innate immune responses, TLR3 will be up-regulated, thus rendering an increased responsiveness to extracellular dsRNA. If and when pathological dsRNA is sensed, cells with induced dsRNA-responsiveness would then be able to mount a dsRNA-induced innate immune response on top of that already induced by IFN. Pathological sources of dsRNA need not even be viral in origin, but could also arise from necrotic or even apoptotic tissue. In this way, the inducible expression of PRRs, like TLR3, may serve as an on/off switch for PAMP recognition in most cell types, perhaps even preventing the occurrence of an innate auto-immune response.

189 It is of consequence that type II IFN, or IFNγ, is able to induce TLR3 expression,

and thereby control responsiveness to extracellular dsRNA. Since IFNγ is highly induced

as part of the adaptive immune response, its ability to induce TLR3 expression may be

considered one of the few examples of the adaptive immune response regulating the

innate response. This could be especially important in the context of re-infection with a pathogen and immune memory. Re-infection by the same pathogen leads to an immediate adaptive immune response, because of immune memory, or the ability of the adaptive immune system to remember previous challenges. If IFNγ produced by the memory adaptive immune response leads to increased TLR3 expression, then the adaptive immune system can essentially confer its immune memory to the , which has none of its own. What role up-regulation of the innate immune response by a memory adaptive immune response plays in clearing viral infections has never been studied, but is a potentially important question.

DISTINCT COMPARTMENTS FOR dsRNA RECOGNITION BY TLR3 AND RIG-I. In

chapter 3 we showed that dsRNA added to the media is unable to induce ISG56 in cells

with an intact RIG-I signaling pathway. Conversely, the RIG-I dominant negative, RIG-

IC blocked ISG56 induction by SeV in cells where the TLR3 signaling pathway was

intact. This tells us that TLR3 and CARD-helicase proteins, like RIG-I and Mda5, are

unable to see dsRNA within the same cellular compartment. Since both RIG-I and TLR3

mediated signaling activate IRF-3, NFκB and ATF2, on the surface it appears that the

cell has simply evolved distinct ways of inducing identical stress responses to the same

stimuli in different cellular compartments. However, the microarray data in chapter five

190 suggests that, despite inducing some common genes, dsRNA-signaling through TLR3 and SeV signaling through RIG-I have distinct gene induction profiles from one another.

This observation is further bolstered by the fact that TLR3-mediated dsRNA signaling does not induce apoptosis, while RIG-I mediated signaling does.

Further, analysis of the phosphorylation of IRF-3 in response to SeV, and the addition of dsRNA to the culture media is different, as shown in chapter 6. SeV infection led to increased phosphorylation of IRF-3 compared to dsRNA signaling through TLR3.

It remains to be determined whether the difference observed is really between TLR3 and

RIG-I signaling, or simply the result of comparing dsRNA treatment to SeV infection.

Since dsRNA in the TLR3 compartment is distinct from dsRNA in the RIG-I compartment, it is possible that additional dsRNA signaling proteins in either compartment contribute to the differential induction of genes. Transfecting dsRNA leads to signaling through RIG-I instead of TLR3. PKR is activated much more strongly through the transfection of dsRNA than it is by adding dsRNA to the media. This makes sense if we accept RIG-I to be the cytosolic receptor for dsRNA, since PKR is also a cytosolic protein. However, the addition of dsRNA to the media leads to the phosphorylation of PKR in HT1080 cells, while RIG-I signaling is not induced by dsRNA in the media. This would suggest that PKR is somehow recognizing dsRNA in both compartments, or is being activated by dsRNA in the media by a means other than direct binding. Differential activation of PKR may even account for the differences seen in the gene induction profiles of dsRNA in the media and SeV.

191 DYNAMIC LANDSCAPE OF dsRNA-MEDIATED GENE INDUCTION. Much of the work detailed in chapters 4 and 5 deals with identifying which dsRNA-induced genes require IRF-3 or NFκB for their induction. A thorough knowledge of which transcription factors mediate the induction of individual genes by dsRNA will certainly be vital information for future researchers attempting to therapeutically manipulate specific aspects of the innate immune system to have. However, perhaps the most immediately useful information to come out of this dissertation is the realization of just how complex and variable the dsRNA signaling response can be. The variability arises, not only from differences in basal signaling component expression between cell types, but also from the secondary fluctuation in the expression and activation of signaling components resulting from initial gene induction. In the end, the overall signaling response to dsRNA seen in vivo, is the compilation of many dynamic interactions between different cell types signaling through a multitude of signaling pathways. Dramatic differences in the overall dsRNA signaling response will result depending upon how, and within which cell types, dsRNA is initially recognized and secondary cytokine signaling proceeds. Consequently, it is not possible to diagram a universally applicable model for the dsRNA signaling response.

This dissertation addresses two examples of cell type specific variability that affect the dsRNA signaling response. As discussed in chapter 3, and already touched

upon in this discussion, variability in responsiveness to extracellular dsRNA results from

differences between cell types in basal TLR3 expression. The second example is

discussed within chapters 5 and 6, where it is shown that cellular expression levels of

IRF-3 can modulate the profile of genes induced by dsRNA, both positively and

192 negatively. Expression of both TLR3 and IRF-3 has been shown to naturally vary

between cell types. While differences in the expression of these two proteins alone could account for large differences in the dsRNA signaling response, they are by no means the only signaling pathway components that vary between cell types.

Many of the genes initially induced by dsRNA can alter the subsequent signaling response. Type I interferons represent the most classical example of this. Chapter 5 shows that in the absence of the IFN signaling pathway, many genes normally induced only six hours after SeV infection failed to get induced. This indicates their induction by

dsRNA is indirect, through IFN signaling. Likewise, induction of the murine equivalents

of ISG54 and ISG56 by dsRNA was severely impaired in bone marrow-derived

macrophages from mice deficient in IFN signaling. However, IFNs are just a few of

many cytokines induced by dsRNA, which can potentially affect the dsRNA signaling

response.

When protein synthesis is inhibited by cycloheximide, the gene induction profile

in GRE cells is substantially altered. Since GRE cells lack the IFN locus, the differences

cannot be attributed to the absence of autocrine IFN signaling. Cytokines, other than

IFNs, induced by dsRNA and viruses include TNFα, IL-6 and IL-8 to name a few.

Transcription factors, including AP-2, p105, ATF3 and various IRF family members,

along with many proteins responsible for negative feedback regulation of signaling, such

as A20 and IκBα, were also induced by dsRNA. It is likely that inhibition of the

expression, and therefore the subsequent effects, of many of these proteins accounts for

the differences in gene induction profiles seen between cells with and without

193 cycloheximide treatment. This underscores the role indirect gene induction plays in shaping the dsRNA signaling response.

The relationship between basal IRF-3 expression and signaling attenuation by induced A20 provides a good example of how direct and indirect signaling cooperate to modulate the overall dsRNA signaling response. Differential induction of A20 between cell types plays a role in length of the overall signaling response. A20 is required for attenuation of NFκB activity induced through TNFα (Lee, Boone et al. 2000) and both

NFκB and IRF-3 activity induced by dsRNA (Boone, Turer et al. 2004). Consequently, induction of A20 is important in limiting the signaling response to dsRNA. Since the cellular expression of IRF-3 is inhibitory to A20 expression, as shown in chapter 6, cells expressing high levels of IRF-3 would have sustained NFκB and IRF-3 responses. In this way, IRF-3 expression indirectly influences the length of the signaling response, through its modulation of A20.

Cellular expression of IRF-3 may also have an indirect role in controlling apoptosis. In addition to A20, many of the genes repressed by increasing levels of IRF-3 expression, such as TRAF1 c-IAP1 and c-IAP2, have anti-apoptotic functions. The proteins encoded by these genes may protect infected cells from the effects of pro- apoptotic pathways, also activated by dsRNA signaling or virus infection. It has been reported that apoptosis of SeV-infected cells is mediated by IRF-3 (Heylbroeck,

Balachandran et al. 2000). Providing further evidence of a role for IRF-3 in programmed cell death, expression of a constitutively active IRF-3 mutant, with 5 phosphomimetic

Ser/Thr to Asp mutations at residues396/398/402/404/405, will cause cells to undergo apoptosis. Together with the results presented in this dissertation, this suggests that IRF-

194 3 may have dual functions in facilitating apoptosis, by not only directly activating a pro-

apoptotic signal, but also by indirectly suppressing the induction of anti-apoptotic genes.

Since A20 inhibits IRF-3 signaling, its expression serves to prevent the induction of a pro-apoptotic signal by IRF-3. Taken in conjunction with our findings, this suggests that IRF-3 and A20 essentially work in opposition to each other. The functional consequence of this is that the determinant of whether or not a virus-infected cell will

undergo apoptosis, as opposed to prolonged viral replication may be the effective ratio of

IRF-3 to A20 expression within the cell at the time of infection.

FINAL THOUGHTS. These last few pages contain vastly oversimplified models of how

the expression of proteins involved in dsRNA signal transduction function in relation to

one another. However, these models do begin to illustrate the complexity of the dsRNA

signaling response, and how subtle differences in the expression of a single protein, basal

or induced, can account for large variations in the signaling cascade and final gene

induction profile. By generating new therapeutics which inhibit or stimulate specific

factors in the signaling cascade, it should be possible to augment the expression of

proteins which actively inhibit viral replication, while minimizing the adverse effects of

the immune response. Therein lays the potential to greatly improve upon the main

limitation of many of today’s immuno-stimulatory therapeutics, which through less

discriminate activation of the immune response cause the debilitating, “flu-like”

symptoms associated most with viral infection.

195 Bibliography

Akira, S., S. Uematsu and O. Takeuchi (2006). "Pathogen recognition and innate immunity." Cell 124(4): 783-801.

Alcamo, E., J. P. Mizgerd, B. H. Horwitz, R. Bronson, A. A. Beg, M. Scott, C. M. Doerschuk, R. O. Hynes and D. Baltimore (2001). "Targeted mutation of TNF receptor I rescues the RelA-deficient mouse and reveals a critical role for NF- kappa B in leukocyte recruitment." J Immunol 167(3): 1592-600.

Alexopoulou, L., A. C. Holt, R. Medzhitov and R. A. Flavell (2001). "Recognition of double-stranded RNA and activation of NF-kappaB by Toll-like receptor 3." Nature 413(6857): 732-8.

Algarte, M., H. Nguyen, C. Heylbroeck, R. Lin and J. Hiscott (1999). "IkappaB-mediated inhibition of virus-induced beta interferon transcription." J Virol 73(4): 2694-702.

Andrejeva, J., K. S. Childs, D. F. Young, T. S. Carlos, N. Stock, S. Goodbourn and R. E. Randall (2004). "The V proteins of paramyxoviruses bind the IFN-inducible RNA helicase, mda-5, and inhibit its activation of the IFN-beta promoter." Proc Natl Acad Sci U S A 101(49): 17264-9.

Bandyopadhyay, S. K., G. T. Leonard, Jr., T. Bandyopadhyay, G. R. Stark and G. C. Sen (1995). "Transcriptional induction by double-stranded RNA is mediated by interferon-stimulated response elements without activation of interferon- stimulated gene factor 3." J Biol Chem 270(33): 19624-9.

Barnes, B. J., A. E. Field and P. M. Pitha-Rowe (2003). "Virus-induced heterodimer formation between IRF-5 and IRF-7 modulates assembly of the IFNA enhanceosome in vivo and transcriptional activity of IFNA genes." J Biol Chem 278(19): 16630-41.

Barnes, B. J., J. Richards, M. Mancl, S. Hanash, L. Beretta and P. M. Pitha (2004). "Global and distinct targets of IRF-5 and IRF-7 during innate response to viral infection." J Biol Chem 279(43): 45194-207.

Bell, J. K., J. Askins, P. R. Hall, D. R. Davies and D. M. Segal (2006). "The dsRNA binding site of human Toll-like receptor 3." Proc Natl Acad Sci U S A 103(23): 8792-7.

Boone, D. L., E. E. Turer, E. G. Lee, R. C. Ahmad, M. T. Wheeler, C. Tsui, P. Hurley, M. Chien, S. Chai, O. Hitotsumatsu, E. McNally, C. Pickart and A. Ma (2004). "The ubiquitin-modifying enzyme A20 is required for termination of Toll-like receptor responses." Nat Immunol 5(10): 1052-60.

196 Boyle, K. A., R. L. Pietropaolo and T. Compton (1999). "Engagement of the cellular receptor for glycoprotein B of human cytomegalovirus activates the interferon- responsive pathway." Mol Cell Biol 19(5): 3607-13.

Brzozka, K., S. Finke and K. K. Conzelmann (2005). "Identification of the rabies virus alpha/beta interferon antagonist: phosphoprotein P interferes with phosphorylation of interferon regulatory factor 3." J Virol 79(12): 7673-81.

Chang, Y. E. and L. A. Laimins (2000). "Microarray analysis identifies interferon- inducible genes and Stat-1 as major transcriptional targets of human papillomavirus type 31." J Virol 74(9): 4174-82.

Chen, L. F. and W. C. Greene (2004). "Shaping the nuclear action of NF-kappaB." Nat Rev Mol Cell Biol 5(5): 392-401.

Chin, K. C. and P. Cresswell (2001). "Viperin (cig5), an IFN-inducible antiviral protein directly induced by human cytomegalovirus." Proc Natl Acad Sci U S A 98(26): 15125-30.

Choe, J., M. S. Kelker and I. A. Wilson (2005). "Crystal structure of human toll-like receptor 3 (TLR3) ectodomain." Science 309(5734): 581-5.

Collins, S. E., R. S. Noyce and K. L. Mossman (2004). "Innate cellular response to virus particle entry requires IRF3 but not virus replication." J Virol 78(4): 1706-17.

Covert, M. W., T. H. Leung, J. E. Gaston and D. Baltimore (2005). "Achieving stability of lipopolysaccharide-induced NF-kappaB activation." Science 309(5742): 1854- 7.

Cusson-Hermance, N., S. Khurana, T. H. Lee, K. A. Fitzgerald and M. A. Kelliher (2005). "Rip1 mediates the Trif-dependent toll-like receptor 3- and 4-induced NF- {kappa}B activation but does not contribute to interferon regulatory factor 3 activation." J Biol Chem 280(44): 36560-6. de Bouteiller, O., E. Merck, U. A. Hasan, S. Hubac, B. Benguigui, G. Trinchieri, E. E. Bates and C. Caux (2005). "Recognition of double-stranded RNA by human toll- like receptor 3 and downstream receptor signaling requires multimerization and an acidic pH." J Biol Chem 280(46): 38133-45. de Veer, M. J., M. Holko, M. Frevel, E. Walker, S. Der, J. M. Paranjape, R. H. Silverman and B. R. Williams (2001). "Functional classification of interferon-stimulated genes identified using microarrays." J Leukoc Biol 69(6): 912-20.

Der, S. D., A. Zhou, B. R. Williams and R. H. Silverman (1998). "Identification of genes differentially regulated by interferon alpha, beta, or gamma using oligonucleotide arrays." Proc Natl Acad Sci U S A 95(26): 15623-8.

197

DiDonato, J. A., M. Hayakawa, D. M. Rothwarf, E. Zandi and M. Karin (1997). "A cytokine-responsive IkappaB kinase that activates the transcription factor NF- kappaB." Nature 388(6642): 548-54.

Diebold, S. S., M. Montoya, H. Unger, L. Alexopoulou, P. Roy, L. E. Haswell, A. Al- Shamkhani, R. Flavell, P. Borrow and C. Reis e Sousa (2003). "Viral infection switches non-plasmacytoid dendritic cells into high interferon producers." Nature 424(6946): 324-8.

Edelmann, K. H., S. Richardson-Burns, L. Alexopoulou, K. L. Tyler, R. A. Flavell and M. B. Oldstone (2004). "Does Toll-like receptor 3 play a biological role in virus infections?" Virology 322(2): 231-8.

Eisen, D. P. and R. M. Minchinton (2003). "Impact of mannose-binding lectin on susceptibility to infectious diseases." Clin Infect Dis 37(11): 1496-505.

Elco, C. P., J. M. Guenther, B. R. Williams and G. C. Sen (2005). "Analysis of genes induced by Sendai virus infection of mutant cell lines reveals essential roles of interferon regulatory factor 3, NF-kappaB, and interferon but not toll-like receptor 3." J Virol 79(7): 3920-9.

Elco, C. P. and G. C. Sen (2007). "Stat1 Required for Interferon-Inducible but Not Constitutive Responsiveness to Extracellular dsRNA." J Interferon Cytokine Res 27(5): 411-24.

Erlandsson, L., R. Blumenthal, M. L. Eloranta, H. Engel, G. Alm, S. Weiss and T. Leanderson (1998). "Interferon-beta is required for interferon-alpha production in mouse fibroblasts." Curr Biol 8(4): 223-6.

Escalante, C. R., J. Yie, D. Thanos and A. K. Aggarwal (1998). "Structure of IRF-1 with bound DNA reveals determinants of interferon regulation." Nature 391(6662): 103-6.

Fitzgerald, K. A., E. M. Palsson-McDermott, A. G. Bowie, C. A. Jefferies, A. S. Mansell, G. Brady, E. Brint, A. Dunne, P. Gray, M. T. Harte, D. McMurray, D. E. Smith, J. E. Sims, T. A. Bird and L. A. O'Neill (2001). "Mal (MyD88-adapter-like) is required for Toll-like receptor-4 signal transduction." Nature 413(6851): 78-83.

Fitzgerald, K. A., D. C. Rowe, B. J. Barnes, D. R. Caffrey, A. Visintin, E. Latz, B. Monks, P. M. Pitha and D. T. Golenbock (2003). "LPS-TLR4 signaling to IRF- 3/7 and NF-kappaB involves the toll adapters TRAM and TRIF." J Exp Med 198(7): 1043-55.

Foy, E., K. Li, R. Sumpter, Jr., Y. M. Loo, C. L. Johnson, C. Wang, P. M. Fish, M. Yoneyama, T. Fujita, S. M. Lemon and M. Gale, Jr. (2005). "Control of antiviral

198 defenses through hepatitis C virus disruption of retinoic acid-inducible gene-I signaling." Proc Natl Acad Sci U S A 102(8): 2986-91.

Frevel, M. A., T. Bakheet, A. M. Silva, J. G. Hissong, K. S. Khabar and B. R. Williams (2003). "p38 Mitogen-activated protein kinase-dependent and -independent signaling of mRNA stability of AU-rich element-containing transcripts." Mol Cell Biol 23(2): 425-36.

Funami, K., M. Matsumoto, H. Oshiumi, T. Akazawa, A. Yamamoto and T. Seya (2004). "The cytoplasmic 'linker region' in Toll-like receptor 3 controls receptor localization and signaling." Int Immunol 16(8): 1143-54.

Geiss, G., G. Jin, J. Guo, R. Bumgarner, M. G. Katze and G. C. Sen (2001). "A comprehensive view of regulation of gene expression by double-stranded RNA- mediated cell signaling." J Biol Chem 276(32): 30178-82.

Geiss, G. K., M. Salvatore, T. M. Tumpey, V. S. Carter, X. Wang, C. F. Basler, J. K. Taubenberger, R. E. Bumgarner, P. Palese, M. G. Katze and A. Garcia-Sastre (2002). "Cellular transcriptional profiling in -infected lung epithelial cells: the role of the nonstructural NS1 protein in the evasion of the host innate defense and its potential contribution to pandemic influenza." Proc Natl Acad Sci U S A 99(16): 10736-41.

Gohda, J., T. Matsumura and J. Inoue (2004). "Cutting edge: TNFR-associated factor (TRAF) 6 is essential for MyD88-dependent pathway but not toll/IL-1 receptor domain-containing adaptor-inducing IFN-beta (TRIF)-dependent pathway in TLR signaling." J Immunol 173(5): 2913-7.

Grandvaux, N., M. J. Servant, B. tenOever, G. C. Sen, S. Balachandran, G. N. Barber, R. Lin and J. Hiscott (2002). "Transcriptional profiling of interferon regulatory factor 3 target genes: direct involvement in the regulation of interferon-stimulated genes." J Virol 76(11): 5532-9.

Guillot, L., R. Le Goffic, S. Bloch, N. Escriou, S. Akira, M. Chignard and M. Si-Tahar (2005). "Involvement of toll-like receptor 3 in the immune response of lung epithelial cells to double-stranded RNA and influenza A virus." J Biol Chem 280(7): 5571-80.

Guo, J., D. J. Hui, W. C. Merrick and G. C. Sen (2000). "A new pathway of translational regulation mediated by eukaryotic initiation factor 3." Embo J 19(24): 6891-9.

Guo, J., K. L. Peters and G. C. Sen (2000). "Induction of the human protein P56 by interferon, double-stranded RNA, or virus infection." Virology 267(2): 209-19.

Guo, Y., D. W. Stacey and M. Hitomi (2002). "Post-transcriptional regulation of cyclin D1 expression during G2 phase." Oncogene 21(49): 7545-56.

199

Guo, Z., S. Garg, K. M. Hill, L. Jayashankar, M. R. Mooney, M. Hoelscher, J. M. Katz, J. M. Boss and S. Sambhara (2005). "A distal regulatory region is required for constitutive and IFN-beta-induced expression of murine TLR9 gene." J Immunol 175(11): 7407-18.

Gurova, K. V., J. E. Hill, O. V. Razorenova, P. M. Chumakov and A. V. Gudkov (2004). "p53 pathway in renal cell carcinoma is repressed by a dominant mechanism." Cancer Res 64(6): 1951-8.

Hacker, H., V. Redecke, B. Blagoev, I. Kratchmarova, L. C. Hsu, G. G. Wang, M. P. Kamps, E. Raz, H. Wagner, G. Hacker, M. Mann and M. Karin (2006). "Specificity in Toll-like receptor signalling through distinct effector functions of TRAF3 and TRAF6." Nature 439(7073): 204-7.

Haller, O. and G. Kochs (2002). "Interferon-induced mx proteins: dynamin-like GTPases with antiviral activity." Traffic 3(10): 710-7.

Haller, O., G. Kochs and F. Weber (2006). "The interferon response circuit: induction and suppression by pathogenic viruses." Virology 344(1): 119-30.

Hayden, M. S. and S. Ghosh (2004). "Signaling to NF-kappaB." Genes Dev 18(18): 2195-224.

Heinz, S., V. Haehnel, M. Karaghiosoff, L. Schwarzfischer, M. Muller, S. W. Krause and M. Rehli (2003). "Species-specific regulation of Toll-like receptor 3 genes in men and mice." J Biol Chem 278(24): 21502-9.

Hemmi, H., O. Takeuchi, S. Sato, M. Yamamoto, T. Kaisho, H. Sanjo, T. Kawai, K. Hoshino, K. Takeda and S. Akira (2004). "The roles of two IkappaB kinase- related kinases in lipopolysaccharide and double stranded RNA signaling and viral infection." J Exp Med 199(12): 1641-50.

Heylbroeck, C., S. Balachandran, M. J. Servant, C. DeLuca, G. N. Barber, R. Lin and J. Hiscott (2000). "The IRF-3 transcription factor mediates Sendai virus-induced apoptosis." J Virol 74(8): 3781-92.

Hiscott, J., N. Grandvaux, S. Sharma, B. R. Tenoever, M. J. Servant and R. Lin (2003). "Convergence of the NF-kappaB and interferon signaling pathways in the regulation of antiviral defense and apoptosis." Ann N Y Acad Sci 1010: 237-48.

Hoebe, K., X. Du, P. Georgel, E. Janssen, K. Tabeta, S. O. Kim, J. Goode, P. Lin, N. Mann, S. Mudd, K. Crozat, S. Sovath, J. Han and B. Beutler (2003). "Identification of Lps2 as a key transducer of MyD88-independent TIR signalling." Nature 424(6950): 743-8.

200 Hoffmann, A., A. Levchenko, M. L. Scott and D. Baltimore (2002). "The IkappaB-NF- kappaB signaling module: temporal control and selective gene activation." Science 298(5596): 1241-5.

Hofmann, K., P. Bucher and J. Tschopp (1997). "The CARD domain: a new apoptotic signalling motif." Trends Biochem Sci 22(5): 155-6.

Honda, K., H. Yanai, H. Negishi, M. Asagiri, M. Sato, T. Mizutani, N. Shimada, Y. Ohba, A. Takaoka, N. Yoshida and T. Taniguchi (2005). "IRF-7 is the master regulator of type-I interferon-dependent immune responses." Nature 434(7034): 772-7.

Horng, T., G. M. Barton and R. Medzhitov (2001). "TIRAP: an adapter molecule in the Toll signaling pathway." Nat Immunol 2(9): 835-41.

Hui, D. J., C. R. Bhasker, W. C. Merrick and G. C. Sen (2003). "Viral stress-inducible protein p56 inhibits translation by blocking the interaction of eIF3 with the ternary complex eIF2.GTP.Met-tRNAi." J Biol Chem 278(41): 39477-82.

Jiang, Z., T. W. Mak, G. Sen and X. Li (2004). "Toll-like receptor 3-mediated activation of NF-kappaB and IRF3 diverges at Toll-IL-1 receptor domain-containing adapter inducing IFN-beta." Proc Natl Acad Sci U S A 101(10): 3533-8.

Jiang, Z., M. Zamanian-Daryoush, H. Nie, A. M. Silva, B. R. Williams and X. Li (2003). "Poly(I-C)-induced Toll-like receptor 3 (TLR3)-mediated activation of NFkappa B and MAP kinase is through an interleukin-1 receptor-associated kinase (IRAK)- independent pathway employing the signaling components TLR3-TRAF6-TAK1- TAB2-PKR." J Biol Chem 278(19): 16713-9.

Karin, M. (1995). "The regulation of AP-1 activity by mitogen-activated protein kinases." J Biol Chem 270(28): 16483-6.

Kato, H., O. Takeuchi, S. Sato, M. Yoneyama, M. Yamamoto, K. Matsui, S. Uematsu, A. Jung, T. Kawai, K. J. Ishii, O. Yamaguchi, K. Otsu, T. Tsujimura, C. S. Koh, E. S. C. Reis, Y. Matsuura, T. Fujita and S. Akira (2006). "Differential roles of MDA5 and RIG-I helicases in the recognition of RNA viruses." Nature.

Kawai, T., K. Takahashi, S. Sato, C. Coban, H. Kumar, H. Kato, K. J. Ishii, O. Takeuchi and S. Akira (2005). "IPS-1, an adaptor triggering RIG-I- and Mda5-mediated type I interferon induction." Nat Immunol 6(10): 981-8.

Kobayashi, K., N. Inohara, L. D. Hernandez, J. E. Galan, G. Nunez, C. A. Janeway, R. Medzhitov and R. A. Flavell (2002). "RICK/Rip2/CARDIAK mediates signalling for receptors of the innate and adaptive immune systems." Nature 416(6877): 194-9.

201 Kurt-Jones, E. A., M. Chan, S. Zhou, J. Wang, G. Reed, R. Bronson, M. M. Arnold, D. M. Knipe and R. W. Finberg (2004). "Herpes simplex virus 1 interaction with Toll-like receptor 2 contributes to lethal encephalitis." Proc Natl Acad Sci U S A 101(5): 1315-20.

Laherty, C. D., N. D. Perkins and V. M. Dixit (1993). "Human T cell leukemia virus type I Tax and phorbol 12-myristate 13-acetate induce expression of the A20 zinc finger protein by distinct mechanisms involving nuclear factor kappa B." J Biol Chem 268(7): 5032-9.

Lamb, R. A. and D. Kolakofsky (2001). "Paramyxoviridae: The Viruses and Their Replication, p.1305-1339. In D. M. Knipe, P. M. Howley (ed.), Field's virology, 4th ed. Lippincott Williams & Wilkins, Philadelphia, PA."

Leaman, D. W., A. Salvekar, R. Patel, G. C. Sen and G. R. Stark (1998). "A mutant cell line defective in response to double-stranded RNA and in regulating basal expression of interferon-stimulated genes." Proc Natl Acad Sci U S A 95(16): 9442-7.

Lee, E. G., D. L. Boone, S. Chai, S. L. Libby, M. Chien, J. P. Lodolce and A. Ma (2000). "Failure to regulate TNF-induced NF-kappaB and cell death responses in A20- deficient mice." Science 289(5488): 2350-4.

Lee, H. K., S. Dunzendorfer, K. Soldau and P. S. Tobias (2006). "Double-stranded RNA- mediated TLR3 activation is enhanced by CD14." Immunity 24(2): 153-63.

Leung, T. H., A. Hoffmann and D. Baltimore (2004). "One nucleotide in a kappaB site can determine cofactor specificity for NF-kappaB dimers." Cell 118(4): 453-64.

Levy, D. E. and I. J. Marie (2004). "RIGging an antiviral defense--it's in the CARDs." Nat Immunol 5(7): 699-701.

Li, K., E. Foy, J. C. Ferreon, M. Nakamura, A. C. Ferreon, M. Ikeda, S. C. Ray, M. Gale, Jr. and S. M. Lemon (2005). "Immune evasion by hepatitis C virus NS3/4A protease-mediated cleavage of the Toll-like receptor 3 adaptor protein TRIF." Proc Natl Acad Sci U S A 102(8): 2992-7.

Li, M., Y. Wang, Y. Yu, M. Nishizawa, T. Nakajima, S. Ito and P. Kannan (2002). "The human transcription factor activation protein-2 gamma (AP-2gamma): gene structure, promoter, and expression in mammary carcinoma cell lines." Gene 301(1-2): 43-51.

Li, X. D., L. Sun, R. B. Seth, G. Pineda and Z. J. Chen (2005). "Hepatitis C virus protease NS3/4A cleaves mitochondrial antiviral signaling protein off the mitochondria to evade innate immunity." Proc Natl Acad Sci U S A 102(49): 17717-22.

202

Lin, R., P. Genin, Y. Mamane, M. Sgarbanti, A. Battistini, W. J. Harrington, Jr., G. N. Barber and J. Hiscott (2001). "HHV-8 encoded vIRF-1 represses the interferon antiviral response by blocking IRF-3 recruitment of the CBP/p300 coactivators." Oncogene 20(7): 800-11.

Linder, P., P. F. Lasko, M. Ashburner, P. Leroy, P. J. Nielsen, K. Nishi, J. Schnier and P. P. Slonimski (1989). "Birth of the D-E-A-D box." Nature 337(6203): 121-2.

Lowther, W. J., P. A. Moore, K. C. Carter and P. M. Pitha (1999). "Cloning and functional analysis of the human IRF-3 promoter." DNA Cell Biol 18(9): 685-92.

Lund, J. M., L. Alexopoulou, A. Sato, M. Karow, N. C. Adams, N. W. Gale, A. Iwasaki and R. A. Flavell (2004). "Recognition of single-stranded RNA viruses by Toll- like receptor 7." Proc Natl Acad Sci U S A 101(15): 5598-603.

Malmgaard, L. (2004). "Induction and Regulation of IFNs During Viral Infections." J Interferon Cytokine Res 24(8): 439-54.

Marie, I., J. E. Durbin and D. E. Levy (1998). "Differential viral induction of distinct interferon-alpha genes by positive feedback through interferon regulatory factor- 7." Embo J 17(22): 6660-9.

Mathas, S., M. Hinz, I. Anagnostopoulos, D. Krappmann, A. Lietz, F. Jundt, K. Bommert, F. Mechta-Grigoriou, H. Stein, B. Dorken and C. Scheidereit (2002). "Aberrantly expressed c-Jun and JunB are a hallmark of Hodgkin lymphoma cells, stimulate proliferation and synergize with NF-kappa B." Embo J 21(15): 4104-13.

Matikainen, S., J. Pirhonen, M. Miettinen, A. Lehtonen, C. Govenius-Vintola, T. Sareneva and I. Julkunen (2000). "Influenza A and sendai viruses induce differential chemokine gene expression and transcription factor activation in human macrophages." Virology 276(1): 138-47.

Matikainen, S., J. Siren, J. Tissari, V. Veckman, J. Pirhonen, M. Severa, Q. Sun, R. Lin, S. Meri, G. Uze, J. Hiscott and I. Julkunen (2006). "Tumor necrosis factor alpha enhances influenza A virus-induced expression of antiviral cytokines by activating RIG-I gene expression." J Virol 80(7): 3515-22.

Matsumoto, M., K. Funami, M. Tanabe, H. Oshiumi, M. Shingai, Y. Seto, A. Yamamoto and T. Seya (2003). "Subcellular localization of Toll-like receptor 3 in human dendritic cells." J Immunol 171(6): 3154-62.

McKendry, R., J. John, D. Flavell, M. Muller, I. M. Kerr and G. R. Stark (1991). "High- frequency mutagenesis of human cells and characterization of a mutant

203 unresponsive to both alpha and gamma interferons." Proc Natl Acad Sci U S A 88(24): 11455-9.

McWhirter, S. M., K. A. Fitzgerald, J. Rosains, D. C. Rowe, D. T. Golenbock and T. Maniatis (2004). "IFN-regulatory factor 3-dependent gene expression is defective in Tbk1-deficient mouse embryonic fibroblasts." Proc Natl Acad Sci U S A 101(1): 233-8.

Medzhitov, R., P. Preston-Hurlburt, E. Kopp, A. Stadlen, C. Chen, S. Ghosh and C. A. Janeway, Jr. (1998). "MyD88 is an adaptor protein in the hToll/IL-1 receptor family signaling pathways." Mol Cell 2(2): 253-8.

Meylan, E., K. Burns, K. Hofmann, V. Blancheteau, F. Martinon, M. Kelliher and J. Tschopp (2004). "RIP1 is an essential mediator of Toll-like receptor 3-induced NF-kappa B activation." Nat Immunol 5(5): 503-7.

Miyakoshi, J., K. D. Dobler, J. Allalunis-Turner, J. D. McKean, K. Petruk, P. B. Allen, K. N. Aronyk, B. Weir, D. Huyser-Wierenga, D. Fulton and et al. (1990). "Absence of IFNA and IFNB genes from human malignant glioma cell lines and lack of correlation with cellular sensitivity to interferons." Cancer Res 50(2): 278- 83.

Mossman, K. L., P. F. Macgregor, J. J. Rozmus, A. B. Goryachev, A. M. Edwards and J. R. Smiley (2001). "Herpes simplex virus triggers and then disarms a host antiviral response." J Virol 75(2): 750-8.

Nakaya, T., M. Sato, N. Hata, M. Asagiri, H. Suemori, S. Noguchi, N. Tanaka and T. Taniguchi (2001). "Gene induction pathways mediated by distinct IRFs during viral infection." Biochem Biophys Res Commun 283(5): 1150-6.

Napolitani, G., A. Rinaldi, F. Bertoni, F. Sallusto and A. Lanzavecchia (2005). "Selected Toll-like receptor agonist combinations synergistically trigger a T helper type 1- polarizing program in dendritic cells." Nat Immunol 6(8): 769-76.

Natoli, G. (2004). "Little things that count in transcriptional regulation." Cell 118(4): 406-8.

Navarro, L., K. Mowen, S. Rodems, B. Weaver, N. Reich, D. Spector and M. David (1998). "Cytomegalovirus activates interferon immediate-early response gene expression and an interferon regulatory factor 3-containing interferon-stimulated response element-binding complex." Mol Cell Biol 18(7): 3796-802.

Nehyba, J., R. Hrdlickova, J. Burnside and H. R. Bose, Jr. (2002). "A novel interferon regulatory factor (IRF), IRF-10, has a unique role in immune defense and is induced by the v-Rel oncoprotein." Mol Cell Biol 22(11): 3942-57.

204 Niikura, T., R. Hirata and S. C. Weil (1997). "A novel interferon-inducible gene expressed during myeloid differentiation." Blood Cells Mol Dis 23(3): 337-49.

Oda, E., R. Ohki, H. Murasawa, J. Nemoto, T. Shibue, T. Yamashita, T. Tokino, T. Taniguchi and N. Tanaka (2000). "Noxa, a BH3-only member of the Bcl-2 family and candidate mediator of p53-induced apoptosis." Science 288(5468): 1053-8.

Oganesyan, G., S. K. Saha, B. Guo, J. Q. He, A. Shahangian, B. Zarnegar, A. Perry and G. Cheng (2006). "Critical role of TRAF3 in the Toll-like receptor-dependent and -independent antiviral response." Nature 439(7073): 208-11.

Oshiumi, H., M. Matsumoto, K. Funami, T. Akazawa and T. Seya (2003). "TICAM-1, an adaptor molecule that participates in Toll-like receptor 3-mediated interferon-beta induction." Nat Immunol 4(2): 161-7.

Pellegrini, S., J. John, M. Shearer, I. M. Kerr and G. R. Stark (1989). "Use of a selectable marker regulated by alpha interferon to obtain mutations in the signaling pathway." Mol Cell Biol 9(11): 4605-12.

Peters, K. L., H. L. Smith, G. R. Stark and G. C. Sen (2002). "IRF-3-dependent, NFkappa B- and JNK-independent activation of the 561 and IFN-beta genes in response to double-stranded RNA." Proc Natl Acad Sci U S A 99(9): 6322-7.

Poltorak, A., X. He, I. Smirnova, M. Y. Liu, C. Van Huffel, X. Du, D. Birdwell, E. Alejos, M. Silva, C. Galanos, M. Freudenberg, P. Ricciardi-Castagnoli, B. Layton and B. Beutler (1998). "Defective LPS signaling in C3H/HeJ and C57BL/10ScCr mice: mutations in Tlr4 gene." Science 282(5396): 2085-8.

Qin, B. Y., C. Liu, S. S. Lam, H. Srinath, R. Delston, J. J. Correia, R. Derynck and K. Lin (2003). "Crystal structure of IRF-3 reveals mechanism of autoinhibition and virus-induced phosphoactivation." Nat Struct Biol 10(11): 913-21.

Reich, N. C. (2002). "Nuclear/cytoplasmic localization of IRFs in response to viral infection or interferon stimulation." J Interferon Cytokine Res 22(1): 103-9.

Reily, M. M., C. Pantoja, X. Hu, Y. Chinenov and I. Rogatsky (2006). "The GRIP1:IRF3 interaction as a target for glucocorticoid receptor-mediated immunosuppression." Embo J 25(1): 108-17.

Rock, F. L., G. Hardiman, J. C. Timans, R. A. Kastelein and J. F. Bazan (1998). "A family of human receptors structurally related to Drosophila Toll." Proc Natl Acad Sci U S A 95(2): 588-93.

Rudd, B. D., E. Burstein, C. S. Duckett, X. Li and N. W. Lukacs (2005). "Differential role for TLR3 in respiratory syncytial virus-induced chemokine expression." J Virol 79(6): 3350-7.

205

Rudd, B. D., J. J. Smit, R. A. Flavell, L. Alexopoulou, M. A. Schaller, A. Gruber, A. A. Berlin and N. W. Lukacs (2006). "Deletion of TLR3 alters the pulmonary immune environment and mucus production during respiratory syncytial virus infection." J Immunol 176(3): 1937-42.

Saitoh, T., M. Yamamoto, M. Miyagishi, K. Taira, M. Nakanishi, T. Fujita, S. Akira, N. Yamamoto and S. Yamaoka (2005). "A20 is a negative regulator of IFN regulatory factor 3 signaling." J Immunol 174(3): 1507-12.

Sarkar, S. N., K. L. Peters, C. P. Elco, S. Sakamoto, S. Pal and G. C. Sen (2004). "Novel roles of TLR3 tyrosine phosphorylation and PI3 kinase in double-stranded RNA signaling." Nat Struct Mol Biol 11(11): 1060-7.

Sarkar, S. N. and G. C. Sen "Novel functions of proteins encoded by Viral Stress- Inducible Genes." Pharmacol Ther., in press.

Sarkar, S. N., H. L. Smith, T. M. Rowe and G. C. Sen (2003). "Double-stranded RNA signaling by Toll-like receptor 3 requires specific tyrosine residues in its cytoplasmic domain." J Biol Chem 278(7): 4393-6.

Sato, M., H. Suemori, N. Hata, M. Asagiri, K. Ogasawara, K. Nakao, T. Nakaya, M. Katsuki, S. Noguchi, N. Tanaka and T. Taniguchi (2000). "Distinct and essential roles of transcription factors IRF-3 and IRF-7 in response to viruses for IFN- alpha/beta gene induction." Immunity 13(4): 539-48.

Sato, S., M. Sugiyama, M. Yamamoto, Y. Watanabe, T. Kawai, K. Takeda and S. Akira (2003). "Toll/IL-1 receptor domain-containing adaptor inducing IFN-beta (TRIF) associates with TNF receptor-associated factor 6 and TANK-binding kinase 1, and activates two distinct transcription factors, NF-kappa B and IFN-regulatory factor-3, in the Toll-like receptor signaling." J Immunol 171(8): 4304-10.

Schoenemeyer, A., B. J. Barnes, M. E. Mancl, E. Latz, N. Goutagny, P. M. Pitha, K. A. Fitzgerald and D. T. Golenbock (2005). "The interferon regulatory factor, IRF5, is a central mediator of toll-like receptor 7 signaling." J Biol Chem 280(17): 17005- 12.

Schulz, O., S. S. Diebold, M. Chen, T. I. Naslund, M. A. Nolte, L. Alexopoulou, Y. T. Azuma, R. A. Flavell, P. Liljestrom and C. Reis e Sousa (2005). "Toll-like receptor 3 promotes cross-priming to virus-infected cells." Nature 433(7028): 887-92.

Servant, M. J., N. Grandvaux and J. Hiscott (2002). "Multiple signaling pathways leading to the activation of interferon regulatory factor 3." Biochem Pharmacol 64(5-6): 985-92.

206 Seth, R. B., L. Sun, C. K. Ea and Z. J. Chen (2005). "Identification and characterization of MAVS, a mitochondrial antiviral signaling protein that activates NF-kappaB and IRF 3." Cell 122(5): 669-82.

Sha, W. C., H. C. Liou, E. I. Tuomanen and D. Baltimore (1995). "Targeted disruption of the p50 subunit of NF-kappa B leads to multifocal defects in immune responses." Cell 80(2): 321-30.

Sharma, S., B. R. tenOever, N. Grandvaux, G. P. Zhou, R. Lin and J. Hiscott (2003). "Triggering the interferon antiviral response through an IKK-related pathway." Science 300(5622): 1148-51.

Shim, J. H., C. Xiao, A. E. Paschal, S. T. Bailey, P. Rao, M. S. Hayden, K. Y. Lee, C. Bussey, M. Steckel, N. Tanaka, G. Yamada, S. Akira, K. Matsumoto and S. Ghosh (2005). "TAK1, but not TAB1 or TAB2, plays an essential role in multiple signaling pathways in vivo." Genes Dev 19(22): 2668-81.

Silverman, R. H. (1994). "Fascination with 2-5A-dependent RNase: a unique enzyme that functions in interferon action." J Interferon Res 14(3): 101-4.

Siren, J., J. Pirhonen, I. Julkunen and S. Matikainen (2005). "IFN-alpha regulates TLR- dependent gene expression of IFN-alpha, IFN-beta, IL-28, and IL-29." J Immunol 174(4): 1932-7.

Sizemore, N., S. Leung and G. R. Stark (1999). "Activation of phosphatidylinositol 3- kinase in response to interleukin-1 leads to phosphorylation and activation of the NF-kappaB p65/RelA subunit." Mol Cell Biol 19(7): 4798-805.

Strahle, L., D. Garcin, P. Le Mercier, J. F. Schlaak and D. Kolakofsky (2003). "Sendai virus targets inflammatory responses, as well as the interferon-induced antiviral state, in a multifaceted manner." J Virol 77(14): 7903-13.

Strassheim, D., K. Asehnoune, J. S. Park, J. Y. Kim, Q. He, D. Richter, K. Kuhn, S. Mitra and E. Abraham (2004). "Phosphoinositide 3-kinase and Akt occupy central roles in inflammatory responses of Toll-like receptor 2-stimulated neutrophils." J Immunol 172(9): 5727-33.

Su, A. I., M. P. Cooke, K. A. Ching, Y. Hakak, J. R. Walker, T. Wiltshire, A. P. Orth, R. G. Vega, L. M. Sapinoso, A. Moqrich, A. Patapoutian, G. M. Hampton, P. G. Schultz and J. B. Hogenesch (2002). "Large-scale analysis of the human and mouse transcriptomes." Proc Natl Acad Sci U S A 99(7): 4465-70.

Sun, J., K. E. Duffy, C. T. Ranjith-Kumar, J. Xiong, R. J. Lamb, J. Santos, H. Masarapu, M. Cunningham, A. Holzenburg, R. T. Sarisky, M. L. Mbow and C. Kao (2006). "Structural and Functional Analyses of the Human Toll-like Receptor 3: ROLE OF GLYCOSYLATION." J Biol Chem 281(16): 11144-51.

207

Sun, Y. and D. W. Leaman (2005). "Involvement of Noxa in cellular apoptotic responses to interferon, double-stranded RNA, and virus infection." J Biol Chem 280(16): 15561-8.

Takahasi, K., N. N. Suzuki, M. Horiuchi, M. Mori, W. Suhara, Y. Okabe, Y. Fukuhara, H. Terasawa, S. Akira, T. Fujita and F. Inagaki (2003). "X-ray crystal structure of IRF-3 and its functional implications." Nat Struct Biol 10(11): 922-7.

Tanner, N. K. and P. Linder (2001). "DExD/H box RNA helicases: from generic motors to specific dissociation functions." Mol Cell 8(2): 251-62.

tenOever, B. R., M. J. Servant, N. Grandvaux, R. Lin and J. Hiscott (2002). "Recognition of the measles virus nucleocapsid as a mechanism of IRF-3 activation." J Virol 76(8): 3659-69.

tenOever, B. R., S. Sharma, W. Zou, Q. Sun, N. Grandvaux, I. Julkunen, H. Hemmi, M. Yamamoto, S. Akira, W. C. Yeh, R. Lin and J. Hiscott (2004). "Activation of TBK1 and IKKvarepsilon kinases by vesicular stomatitis virus infection and the role of viral ribonucleoprotein in the development of interferon antiviral immunity." J Virol 78(19): 10636-49.

Tissari, J., J. Siren, S. Meri, I. Julkunen and S. Matikainen (2005). "IFN-alpha enhances TLR3-mediated antiviral cytokine expression in human endothelial and epithelial cells by up-regulating TLR3 expression." J Immunol 174(7): 4289-94.

Tiwari, R. K., J. Kusari and G. C. Sen (1987). "Functional equivalents of interferon- mediated signals needed for induction of an mRNA can be generated by double- stranded RNA and growth factors." Embo J 6(11): 3373-8.

Torrence, P. F., R. K. Maitra, K. Lesiak, S. Khamnei, A. Zhou and R. H. Silverman (1993). "Targeting RNA for degradation with a (2'-5')oligoadenylate-antisense chimera." Proc Natl Acad Sci U S A 90(4): 1300-4.

Town, T., D. Jeng, L. Alexopoulou, J. Tan and R. A. Flavell (2006). "Microglia recognize double-stranded RNA via TLR3." J Immunol 176(6): 3804-12.

Townsend, P. A., T. M. Scarabelli, S. M. Davidson, R. A. Knight, D. S. Latchman and A. Stephanou (2004). "STAT-1 interacts with p53 to enhance DNA damage-induced apoptosis." J Biol Chem 279(7): 5811-20.

Tsai, P. W., S. G. Shiah, M. T. Lin, C. W. Wu and M. L. Kuo (2003). "Up-regulation of vascular endothelial growth factor C in breast cancer cells by heregulin-beta 1. A critical role of p38/nuclear factor-kappa B signaling pathway." J Biol Chem 278(8): 5750-9.

208 Unterstab, G., S. Ludwig, A. Anton, O. Planz, B. Dauber, D. Krappmann, G. Heins, C. Ehrhardt and T. Wolff (2005). "Viral targeting of the interferon-{beta}-inducing Traf family member-associated NF-{kappa}B activator (TANK)-binding kinase- 1." Proc Natl Acad Sci U S A 102(38): 13640-5.

Van Antwerp, D. J., S. J. Martin, T. Kafri, D. R. Green and I. M. Verma (1996). "Suppression of TNF-alpha-induced apoptosis by NF-kappaB." Science 274(5288): 787-9.

Watling, D., D. Guschin, M. Muller, O. Silvennoinen, B. A. Witthuhn, F. W. Quelle, N. C. Rogers, C. Schindler, G. R. Stark, J. N. Ihle and et al. (1993). "Complementation by the protein tyrosine kinase JAK2 of a mutant cell line defective in the interferon-gamma signal transduction pathway." Nature 366(6451): 166-70.

Werner, S. L., D. Barken and A. Hoffmann (2005). "Stimulus specificity of gene expression programs determined by temporal control of IKK activity." Science 309(5742): 1857-61.

Whitmore, M. M., M. J. DeVeer, A. Edling, R. K. Oates, B. Simons, D. Lindner and B. R. Williams (2004). "Synergistic activation of innate immunity by double- stranded RNA and CpG DNA promotes enhanced antitumor activity." Cancer Res 64(16): 5850-60.

Wietek, C., S. M. Miggin, C. A. Jefferies and L. A. O'Neill (2003). "Interferon regulatory factor-3-mediated activation of the interferon-sensitive response element by Toll- like receptor (TLR) 4 but not TLR3 requires the p65 subunit of NF-kappa." J Biol Chem 278(51): 50923-31.

Williams, B. R. (2001). "Signal integration via PKR." Sci STKE 2001(89): RE2.

Yamamoto, M., S. Sato, H. Hemmi, S. Uematsu, K. Hoshino, T. Kaisho, O. Takeuchi, K. Takeda and S. Akira (2003). "TRAM is specifically involved in the Toll-like receptor 4-mediated MyD88-independent signaling pathway." Nat Immunol 4(11): 1144-50.

Yoneyama, M., M. Kikuchi, K. Matsumoto, T. Imaizumi, M. Miyagishi, K. Taira, E. Foy, Y. M. Loo, M. Gale, Jr., S. Akira, S. Yonehara, A. Kato and T. Fujita (2005). "Shared and unique functions of the DExD/H-box helicases RIG-I, MDA5, and LGP2 in antiviral innate immunity." J Immunol 175(5): 2851-8.

Yoneyama, M., M. Kikuchi, T. Natsukawa, N. Shinobu, T. Imaizumi, M. Miyagishi, K. Taira, S. Akira and T. Fujita (2004). "The RNA helicase RIG-I has an essential function in double-stranded RNA-induced innate antiviral responses." Nat Immunol 5(7): 730-7.

209 Zhu, H., J. P. Cong, G. Mamtora, T. Gingeras and T. Shenk (1998). "Cellular gene expression altered by human cytomegalovirus: global monitoring with oligonucleotide arrays." Proc Natl Acad Sci U S A 95(24): 14470-5.

210