Defining the Biochemical Factors Regulating IFITM3-Mediated Antiviral Activity

Home , Antiviral protein, Dispanin, Ifitm3, Translational regulation, UBE1L

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Nicholas Chesarino

Graduate Program in Biomedical Sciences

The Ohio State University

2016

Dissertation Committee:

Jacob Yount, Advisor

Amal Amer

Ian Davis

Joanne Turner

Mark Wewers

Nicholas Michael Chesarino

2016

Abstract

Influenza virus infects 5 to 20 percent of the global population each year, and is one of the top ten causes of human deaths. Furthermore, the ability of influenza viruses to evolve into novel strains, such as the 2009 pandemic H1N1 virus, leaves the human population exposed to more devastating pandemics, underscoring the importance of uncovering broadly-acting immune inhibitors of viral infection. For this reason, our lab focuses on a cellular protein known as interferon-induced transmembrane protein 3

(IFITM3), which inhibits the fundamental cellular entry process of all strains of influenza virus tested to date. Importantly, several studies have linked a polymorphism in IFITM3, which leads to a truncated variant of the protein, to increased influenza severity. Despite the essential role of IFITM3 in inhibiting influenza virus infections, a better understanding of its mechanism of action, cellular trafficking, and post-translational regulation are needed in order to develop IFITM3-based antiviral therapies.

Through investigation of Fyn kinase-mediated phosphorylation of IFITM3, we discovered that IFITM3 traffics to the plasma membrane and enters into the endocytic pathway, where it blocks influenza virus from entering the cell. We learned that this trafficking is disrupted upon IFITM3 phosphorylation, which we found to block an endocytic motif, sequestering IFITM3 at the plasma membrane and leaving cells more vulnerable to viral infection. In the course of these studies, we found that

ii phosphorylation of IFITM3 also blocks a NEDD4 E3 ubiquitin ligase recognition motif, preventing its ubiquitination in addition to its endocytosis. Research into the NEDD4-

IFITM3 interaction uncovered several key mechanisms of IFITM3 regulation and

activity. We determined that NEDD4 and IFITM3 directly interact, leading to poly-

ubiquitination of IFITM3 and ultimately degradation in the lysosome. We demonstrated

the ability to manipulate basal levels of IFITM3 by knocking down or knocking out

NEDD4, and showed that downregulating NEDD4 greatly increases resistance to

infection by several viruses tested.

Finally, prediction of IFITM3 secondary structure using a bioinformatics

approach allowed us to identify a highly-conserved, short amphipathic helix within a

hydrophobic region previously thought to be a transmembrane domain. Consistent with

the known ability of amphipathic helices to alter membrane properties, we show that this

helix and its amphipathicity are required for inhibition of influenza virus, Zika virus,

vesicular stomatitis virus, Ebola virus, and human immunodeficiency virus infections by

IFITM3. Further, we specifically show that the amphipathic helix is required for

blockade of influenza virus hemagglutinin-mediated membrane fusion by IFITM3.

Overall, this work has revealed several mechanistic insights into the IFITM proteins that

should prove useful in the development of broadly-acting, novel IFITM-based antiviral therapies.

iii

Acknowledgments

I would like to thank my advisor, Dr. Jacob Yount, for his guidance and

mentorship during my graduate career. The trust and freedom he provided me in his lab fostered an environment where I could learn and develop as a scientist. I truly feel fortunate for the chance to work with him and sharing our many successes together. I hope to continue to share mutual success as colleagues for many years to come.

Thank you to the members of my dissertation committee, Dr. Amal Amer, Dr. Ian

Davis, Dr. Joanne Turner, and Dr. Mark Wewers. Each of these members readily agreed to participate in my committee, which I understand is no small task. I greatly appreciate the time and commitment each one of them spent on my development as a scientist. I am humbled to have the support from these experts in their respective fields.

To the current and previous members of the Yount Lab, Temet McMichael,

Jocelyn Hach, Dr. Lizhi Zhang, and Adam Kenney. This experience would not have been nearly the same without you all. Each one of you has helped me become the scientist I am today through our philosophical musings, scientific discussions, and comic relief. I will cherish the friendships I have made in this lab and am thankful to know you all.

Finally, I sincerely thank my family for the love and support they have given me throughout my life. My late father, Daniel Chesarino, instilled in me an irresistible drive

iv to learn and problem solve through our countless hours of working with puzzles and playing board games. My mother, Kathryn Chesarino, has tirelessly supported me through all my endeavors, and has always encouraged me to persevere. My sister,

Carolyn Chesarino, will always be my friendly rival, the one person I have always strived to emulate, and the person I look up to most. Without your guidance, I would not be where I am today.

Vita

2008...... St. Peter Chanel High School

2012...... B.S. Biology, John Carroll University

2012 to present ...... Graduate Research Fellow, Department of

Microbial Infection and Immunity, The Ohio

State University College of Medicine

Publications

Chesarino NM, Compton AA, McMichael TM, Zhang L, Soewarna V, Doering R, Davis M, Schwartz O, and Yount JS (2016). IFITM3 restriction of virus infection requires an amphipathic helix. In preparation.

Chesarino NM, McMichael TM, and Yount JS (2015). E3 Ubiquitin Ligase NEDD4 Promotes Influenza Virus Infection by Decreasing Levels of the Antiviral Protein IFITM3. PLoS Pathogens 11(8):e1005095. PMID: 26263374.

Melvin WJ, McMichael TM, Chesarino NM, Hach JC, and Yount JS (2015). IFITMs from Mycobacteria Confer Resistance to Influenza Virus When Expressed in Human Cells. Viruses 7(6):3035-52. PMID: 26075508.

Woods PS, Tazi MF, Chesarino NM, Amer AO, and Davis IC (2015). TGF-β-induced IL-6 prevents development of acute lung injury in influenza A virus-infected F508del CFTR-heterozygous mice. American Journal of Physiology: Lung Cellular and Molecular Physiology. 308(11):L1136-44. PMID: 25840995.

Wu Y, Ma J, Woods PS, Chesarino NM, Liu C, Lee LJ, Nana-Sinkam SP, and Davis IC (2015). Selective targeting of alveolar type II respiratory epithelial cells by anti- surfactant protein-C antibody-conjugated lipoplexes. Journal of Controlled Release 203:140-149. PMID: 25687308. vi

Chesarino NM, Hach JC, Chen JL, Zaro BW, Rajaram MV, Turner J, Schlesinger LS, Pratt MR, Hang HC, and Yount JS (2014). Chemoproteomics reveals Toll-like receptor fatty acylation. BMC Biology 12:91. PMID: 25371237.

Chesarino NM, McMichael TM, and Yount JS (2014). Regulation of the trafficking and antiviral activity of IFITM3 by post-translational modifications. Future Microbiology 9(10):1151-1163. Review. PMID: 25405885.

Chesarino NM, McMichael TM, Hach JC, and Yount JS (2014). Phosphorylation of the Antiviral Protein IFITM3 Dually Regulates its Endocytosis and Ubiquitination. Journal of Biological Chemistry 289(17):11986-11992. PMID: 24627473.

Johansen JR, Bohunická M, Lukešová A, Hrčková K, Vaccarino MA, and Chesarino NM (2014). Morphological and molecular characterization within 26 strains of the genus Cylindrospermum (Nostocaceae, Cyanobacteria), with descriptions of three new species. Journal of Phycology 1(50): 187-202).

Hach JC, McMichael T, Chesarino NM, and Yount JS (2013). Palmitoylation on conserved and nonconserved cysteines of murine IFITM1 regulates its stability and anti- influenza A virus activity. Journal of Virology 87(17): 9923-9927. PMID: 23804635.

Fields of Study

Major Field: Biomedical Sciences

vii

Table of Contents

Abstract ...... ii

Acknowledgments...... iv

Vita ...... vi

Publications ...... vi

Fields of Study ...... vii

Table of Contents ...... viii

List of Tables ...... xi

List of Figures ...... xii

Chapter 1: Introduction ...... 1

1.1: Introduction to the influenza viruses ...... 1

1.1.1: Global impact ...... 1

1.1.2: Biology and Pathogenesis ...... 2

1.1.3: Antigenic drift and shift...... 6

1.1.4: Current interventions ...... 8

1.2: Host defense against influenza A virus ...... 11

1.2.1: Anatomical and physiobiological barriers ...... 11 viii

1.2.2: Viral detection and innate immunity ...... 12

1.2.3: The interferon-stimulated genes ...... 14

1.2.4: Adaptive immunity and viral clearance ...... 16

1.3: The IFITMs as broad inhibitors of viral infection ...... 18

1.3.1: Evolutionary origins ...... 18

1.3.2: Discovery of antiviral activity ...... 19

1.3.4: Proposed mechanisms of action ...... 23

1.3.5: The Complex Topology of IFITM3 ...... 28

1.3.6: Regulation of IFITMs by Post-translational modifications ...... 33

Chapter 2: Phosphorylation of IFITM3 dually regulates its endocytosis and ubiquitination ...... 39

2.1: Abstract ...... 39

2.2: Introduction ...... 40

2.3: Materials and Methods ...... 41

2.4: Results ...... 44

2.5: Discussion ...... 59

2.6: Contributions and Acknowledgements ...... 61

Chapter 3: Identification of E3 ubiquitin ligase NEDD4 as a negative regulator of

IFITM3 ...... 63

3.1: Abstract ...... 63

3.2: Introduction ...... 64

3.3: Materials and Methods ...... 66

3.4: Results ...... 73

3.5: Discussion ...... 87

3.6: Contributions and Acknowledgements ...... 90

Chapter 4: IFITM3 restriction of virus infection requires an amphipathic helix ...... 92

4.1: Abstract ...... 92

4.2: Introduction ...... 93

4.3: Materials and Methods ...... 95

4.4: Results ...... 101

4.5: Discussion ...... 124

4.6: Contributions and Acknowledgements ...... 128

Chapter 5: Conclusions ...... 130

5.1: IFITM3 phosphorylation: Insights and further questions ...... 130

5.2: IFITM3 ubiquitination: Overexpression without infection ...... 132

5.3: Understanding the broad antiviral activity of the IFITM3 amphipathic helix .... 133

Bibliography ...... 135

List of Tables

Table 1: Key differences in the human-relevant influenza viruses ...... 3

Table 2: Comprehensive list of viruses tested against the IFITM proteins ...... 22

Table 3: Mean hydrophobic moment calculations for WT IFITM3 amino acids 59-68 and the amphipathic helix mutants tested in this study ...... 113

List of Figures

Figure 1: Model of influenza virus entry and IFITM3 antiviral activity...... 27

Figure 2: Membrane topologies for IFITM3 and SYNDIG1 that are supported by experimental evidence ...... 32

Figure 3: IFITM3 phosphorylation by Fyn ...... 46

Figure 4: IFITM3 is active in the absence of Fyn ...... 47

Figure 5: Tyrosine 20 of IFITM3 is necessary for complete antiviral activity against influenza virus ...... 49

Figure 6: IFITM3 tyrosine 20 phosphorylation or mutation affects IFITM3 localization including plasma membrane accumulation ...... 51

Figure 7: Identification of a YxxΦ endocytosis motif in IFITM3 ...... 53

Figure 8: The YEML motif from hIFITM3 causes internalization of CD4 ...... 55

Figure 9: Unmodified tyrosine 20 is necessary for proper IFITM3 ubiquitination ...... 58

Figure 10: IFITM3 is ubiquitinated by NEDD4 ...... 73

Figure 11: The IFITM3 PPxY motif is required for ubiquitination by NEDD4 ...... 77

Figure 12: NEDD4 catalytic activity is required for IFITM3 ubiquitination ...... 78

Figure 13: NEDD4 ubiquitinates IFITM3 in vitro ...... 80

Figure 14: NEDD4 knockout decreases IFITM3 ubiquitination and protects cells from virus infection ...... 83

xii

Figure 15: NEDD4 regulates cellular susceptibility to influenza virus infection by controlling IFITM3 levels ...... 84

Figure 16: NEDD4 knockdown in human lung cells increases IFITM3 levels and resistance to influenza virus infection...... 86

Figure 17: Hydrophobic domain 2 of IFITM3 is not essential for antiviral activity ..... 103

Figure 18: Secondary structure prediction reveals an amphipathic helix within IFITM3

...... 106

Figure 19: Structural prediction programs suggest helical domains in IFITM3 ...... 108

Figure 20: IFITM3 amphipathicity is required for its antiviral function ...... 110

Figure 21: Cellular localization of IFITM3 with lysosomes is unaffected by mutations in its amphipathic helix ...... 115

Figure 22: Mutations of the amphipathic helix do not prevent IFITM3 colocalization with internalized influenza virus ...... 116

Figure 23: S-palmitoylation of IFITM3 occurs independently of its amphipathic helix 117

Figure 24: Antiviral activity of IFITM1 requires its amphipathic helix ...... 120

Figure 25: The amphipathic helix and its localization are important for IFITM3 inhibition of influenza virus hemagglutinin-mediated membrane fusion ...... 123

Figure 26: Control imaging for transfections presented in Figure 25 ...... 124

xiii

Chapter 1: Introduction

1.1: Introduction to the influenza viruses

1.1.1: Global impact

Influenza viruses are a major threat to human health, causing annual epidemics that lead to an estimated 3 to 5 million deaths worldwide [1]. While extensive measures

are taken to combat seasonal influenza outbreaks, vaccinations are ineffective at

combating the emergence of novel influenza A viruses, such as the influenza A H1N1

virus that quickly grew pandemic in 2009. A staggering 60 million influenza-related illnesses are thought to have been caused by the 2009 H1N1 virus, and 90% of hospitalizations and deaths from this virus occurred in people younger than 65 years of age [2, 3]. In addition, two novel avian influenza A viruses are being closely monitored due to the severe morbidity and mortality of the disease they cause: the avian influenza A

H7N9 virus, which presents with a 40% mortality rate and pneumonia in most cases and has emerged in four outbreaks in East Asia since 2013 [4, 5], and the highly-pathogenic avian influenza A H5N1 virus that has caused outbreaks around the world and presents with a 50% mortality rate [4, 6].

An often-underappreciated consequence of influenza viruses is the substantial economic burden that results from influenza-related disease. Molinari et. al. (2007) estimated that, based on the 2003 United States population statistics, seasonal influenza

can cost an average of $10.4 billion in direct medical costs, $16.3 billion in projected lost earnings due to illness and death, and a total economic burden at a striking $87.1 billion

[7]. A more recent estimation model by Mao et. al. (2012), based on county-level economic impacts of influenza, offered a more conservative burden estimate; $29.12 billion in the United States annually, with 65% coming from indirect costs (time off work, death, etc.) and 35% from direct medical costs [8]. Although the actual economic burden of the 2009 H1N1 influenza pandemic has yet to be determined, several statistical models exist to estimate the potential costs of future influenza pandemics. Verikios et. al. (2011) predicted economic losses ranging from ~0.3 – 4.0% of the global gross domestic product at its peak (~$200 billion to $2.6 trillion) [9]. Importantly, all of these studies directly address how crucial medical intervention and prophylaxis are in dramatically reducing influenza-related economic burden [6-9].

1.1.2: Biology and Pathogenesis

Influenza viruses are represented among three genera, Influenzavirus A,

Influenzavirus B, and Influenzavirus C in the Orthomyxoviridae family [10]. Though reassortment events (the mixing of genetic material between viruses) occurs within each of the three genera, genetic mixing does not occur between genera [11]. The key features distinguishing types A, B, and C are detailed in Table 1. In brief, influenza A and B viruses cause similar disease in humans [12], whereas influenza C viruses lead to only mild upper respiratory tract illness [12, 13]. Importantly, influenza A viruses are the only type that can generate pandemic strains, and account for the majority of influenza cases

[14]. For these reasons, the focus of our studies, and henceforth this document, are

primarily paid to the influenza A viruses.

Influenzavirus A Influenzavirus B Influenzavirus C Subtypes Yes No No Disease Severity Mild to Severe Mild to Severe Mild Hosts Humans, birds, mammals Humans, seals Humans, pigs Pandemics Yes No No Epidemics Yes Yes No RNA Segments 8 8 7 Membrane HA,NA,M2 HA,NA,NB,BM2 HEF,CM2 Proteins Host:Viral Sialic acid:HA Sialic acid:HA Sialic acid:HEF Binding Entry Pathway Endocytosis Endocytosis Endocytosis

Table 1: Key differences in the human-relevant influenza viruses HA; hemagglutinin. NA; neuraminidase. M2; matrix-2 protein. NB; Glycoprotein NB (ion channel). BM2; BM2 protein (proton transport). HEF; hemagglutinin-esterase- fusion glycoprotein. CM2; CM2 protein (ion channel). Table adapted from references [11-15].

The influenza A viruses are further divided into subtypes due to the variety of surface markers, hemagglutinin (HA) and neuraminidase (NA), that each strain can possess [16]. In all, there are 18 HA subtypes and 11 NA subtypes identified to date, potentially allowing for 198 unique strains of influenza A virus [17]. As a result, the

World Health Organization mandated a streamlined strain designation system for influenza A, B, and C viruses upon the following criteria: 1. Identification of influenza type (A, B, or C), 2. Isolation source (if non-human), 3. Geographical origin, 4. Strain number, and 5. Year of isolation. Specifically, influenza A viruses require the HA and

NA description in parentheses [18]. As an example, the most common strain for influenza research, PR8, is an influenza A virus with the antigens H1N1; it was isolated

from a human sputum in Puerto Rico in 1934, and passed through ferrets eight times [19].

These conditions give rise to the strain name influenza A/Puerto Rico/8/1934 (H1N1)

[18, 19].

Influenza viruses are negative-sense RNA viruses with genes expressed from 8 genomic segments (or 7 for influenza C), and encode at least 11 genes [11]. Influenza

virions are enveloped with the lipid membrane derived from the host cell in which it was

packaged, and is coated with hemagglutinin (HA), neuraminidase (NA), and matrix 2

(M2) proteins [11, 20, 21]. The matrix 1 (M1) protein forms a layer just under the

envelope, coating the viral ribonucleoprotein (RNP) complexes [20]. As aerosolized

virions are inhaled by its host, influenza A viruses utilize the HA protein to bind to sialic

acid, which is expressed on host cell glycoproteins [21]. Affinity of the HA receptor to

sialic acid is dependent on the subtype of the HA molecule, as well as the changes in

affinity resulting from antigenic drift [22]. Directly following HA:sialic acid binding is

the internalization of the virion through endocytosis [21], an active intracellular transport

process designed to internalize macromolecules and break them down into simple

compounds [23]. As such, the pH of endosomes actively declines while trafficking to the

perinuclear regions of the cell [24]. Late endosomal bodies are designed to fuse with

lysosomes, large acidic vesicles regarded as the hydrolytic center of the cell [25]. Fusion

with lysosomes leads to the final breakdown of cellular waste and foreign material to be

recycled in biosynthetic pathways [25].

Influenza viruses hijack the endosomal pathway in order to enter the cell.

Specifically, the decreasing pH in the influenza-containing endosomal compartment

triggers a conformational change in HA, releasing a large store of potential energy that is

used to mediate the fusion of the influenza envelope and host endosomal membrane [26-

30]. Concurrently, low pH triggers activation of the M2 channel, which imports protons

into the virion core and dissociates M1 from the RNP [31, 32]. Free RNPs are now

trafficked into the cytoplasm in the perinuclear space, and immediately imported into the

nucleus to begin replication [33].

Viral RNPs in influenza consist of the three polymerase proteins: polymerase

basic protein 1 (PB1), polymerase basic protein 2 (PB2), and polymerase acidic protein

(PA). Loaded into the RNPs are the RNA segments encased by nucleoprotein (NP).

RNPs perform transcription and replication in the nucleus, hijacking the host’s RNA

processing machinery [20]. The nonstructural 1 (NS1) protein assists in mRNA

translation and antagonizes the host antiviral response [11, 34], while the nonstructural 2

(NS2) protein performs nuclear export of RNPs [11, 20]. Viral mRNAs in the cytoplasm

are translated to proteins in the host ribosome, and subsequent products may either traffic

back into the nucleus to assist in replication, or be packaged and transported by the Golgi apparatus to the plasma membrane. Nascent virions assemble at the plasma membrane, and begin to bud off from the cell upon formation; they can collect aberrantly at the cell

membrane as HA molecules bind to sialic acid on both the cell surface and other virions,

and are freed from the host cell (and each other) after cleavage of sialic acid by

neuraminidase (NA) [35].

Influenza virus is capable of infecting and replicating in epithelial cells lining the

upper and lower respiratory tract, with tropism to the lower respiratory tract being

correlated with more severe and fatal disease [36]. Signs and symptoms include a sudden onset of high fever, cough, headache, and malaise, and can persist an average of 7 to 10 days, with fatigue and weakness lasting several weeks [21]. This group of symptoms is often referred to a “influenza-like illness” due to a number of respiratory pathogens that can result in similar disease; in fact, only about 1/3 of patients experiencing influenza- like illness test positive for influenza A virus infection [17]. Influenza infection is most prevalent in children, although disease severity is greatest in high-risk groups such as the neonates and infants, diabetics, the elderly, pregnant women, sufferers of cardiopulmonary disease, and the otherwise immunosuppressed. Severe influenza-related complications in high-risk groups can include pneumonia, myocarditis, encephalitis, multi-organ failure, sepsis, and death [17]. To add to the concern, pandemic influenza viruses are more likely to present with much higher morbidity and mortality, especially in healthy, low-risk groups, than seasonal influenza. These complications can arise in as little as 48 hours after the onset of symptoms, coinciding with the peak production of viral particles around that time [37, 38]. The unpredictability in disease severity of strains on a yearly basis underscores the importance of aggressive research and development into broadly acting anti-influenza prophylaxes and interventions.

1.1.3: Antigenic drift and shift

Influenza A viruses actively undergo two kinds of evolution: antigenic drift, and antigenic shift. Antigenic drift, a phenomenon also shared by the influenza B and C viruses, is the gradual accumulation of mutations in the HA and, less commonly, NA

proteins [39]. This occurs mechanistically due to the viral polymerase having no

proofreading ability [11]. Evolution by antigenic drift allows influenza to overcome the

host immune system, and several studies have linked this mechanism to a positive

selection of influenza strains capable of evading the host immune response at both the

individual and population levels [40-43]. Antigenic drift is responsible for the vast variability in vaccine efficacy to this day, and many groups posit that understanding this process can lead to the development of universal vaccines [44-49].

Antigenic shift, in contrast, is unique to the influenza A viruses due to both its segmented RNA genome and large human, swine, and avian reservoirs [50].

Components of segmented genomes from distinct strains may reassort if intersected in a common host cell [50-53]. Importantly, this mechanism gives rise to novel strains that the host population is immunologically naïve to. The strikingly-complex origin story of the 2009 H1N1 pandemic strain (officially referred to as influenza A(H1N1)pdm09) is a fitting example of antigenic shift. In the 1998 flu season, components of an unknown avian influenza A strain, a human H3N2 strain, and a North American classical swine

H1N1 strain all combined in swine to generate a swine triple-reassortant influenza A

H1N2 virus (36). In 2009, another reassortant event between this strain and a Eurasian swine H1N1 strain suddenly gave rise to the highly contagious influenza A(H1N1)pdm09 virus [54]. Extensive genomic analyses of historical pandemic influenza viruses provide strong evidence for antigenic shift as the likely driver of previous and future influenza pandemics [14, 54-56].

1.1.4: Current interventions

Current interventions to combat influenza viruses can be delineated into

prophylactic and therapeutic categories. By far, our best means of influenza prophylaxis

are seasonal vaccinations, which contain a minimum of antigens against three types of flu

viruses (trivalent vaccine): an influenza A (H1N1), an influenza A (H3N2), and an

influenza B strain [57]. Vaccines containing antigens from four viruses (quadrivalent

vaccines) contain the above plus an additional influenza B strain [58]. Vaccination gives

rise to protective antibodies mostly targeting the highly-variable epitopes of the HA, and

less commonly, the NA proteins [59]. Standard trivalent vaccines utilize influenza viruses grown in eggs and inactivated, although egg-free, recombinant trivalent vaccines exist for those allergic to eggs [60, 61]. Options exist for the immunocompromised and elderly to boost vaccine efficacy; use of adjuvants has been approved for the 2016-2017

flu season, and trivalent vaccines with higher concentration of viral antigens are also

available [62]. The strains utilized in each seasonal vaccine are chosen by the World

Health Organization after careful consideration by over 100 influenza surveillance

centers around the world, and decisions for vaccine composition are made in February for

the Northern Hemisphere, and September for the Southern Hemisphere [63].

Overall, vaccinations provide an extreme benefit to the communities in which

they are introduced. These benefits are especially realized in the elderly population, as

several studies have correlated use of the seasonal vaccine with decreased severity,

complications, hospitalizations, and illness events [64-66]. In addition, vaccination

significantly reduces influenza-related admissions to the intensive care unit for children

[67]. Finally, pregnant women who receive the vaccine reduce their rate of infection by about 50% [68], and infants born to vaccinated mothers receive significant protection as well [69, 70]. However, the efficacy of vaccines is dependent on an overwhelming amount of variables; the most important of which is the predicted strains included in the vaccine. As noted earlier, antigenic drift is a continuous process in influenza viruses, and small changes to the epitopes targeted by the immune system can greatly impact the efficacy of the vaccine. For this reason, vaccine efficacy is highly variable; a prime example is the vaccine efficacy of the 2015-2016 season (47%; 95%CI: 39%-53%) compared to the 2014-2015 season (19%; 95%CI: 10%-27%) [71, 72]. Despite the obvious benefits influenza vaccines provide, antigenic drift of viral strains ushers the need to expand our strategies in combating influenza.

Several direct anti-influenza medications have been developed and can be grouped by their mechanism of action: NA inhibitors, M2 ion channel blockers, nucleoside analogs, NP inhibitors, and PB2/PA inhibitors [73]. All of these compounds are small molecule inhibitors of crucial events in influenza infection and replication. NA inhibitors include oseltamivir, zanamivir, peramivir, and laninamivir, and all have similar functions as competitive inhibitors of influenza A and B NA, blocking the interaction between NA and sialic acid and preventing new viruses from budding away from the cell.

The efficacy of NA inhibitors is highly dependent on time of administration (within 48 hours of disease onset), and is useful in alleviating influenza-like symptoms [73].

However, the true anti-influenza efficacy of these drugs is under debate, as the effect of

NA inhibitors vary significantly with the time of treatment, dose, virus strain, and model

used [74-76]. The M2 ion channel blockers, amantadine and rimantadine, only inhibit

influenza A viruses and work by preventing electrostatic disruption of the viral core,

thereby disabling import of viral RNA into the cytoplasm [73]. Amantadine and

rimantadine have fallen out of favor due to side effects affecting the central nervous

system, immediate derivation of resistant virus that can be transmitted, and lack of

overall therapeutic benefit [77-80].

The less-common classes of small molecule influenza inhibitors target the

replication machinery of influenza viruses. The nucleoside analogs, favipiravir and

ribavirin, inhibit RNA-dependent RNA polymerase activity for several RNA viruses, and

Ribavirin is also capable of inhibiting some DNA viruses [73, 81]. The antiviral

mechanism of these analogs is thought to be prevention of RNA elongation, and potential mutagenesis of the virus [81-83]. However, these analogs are rarely used due to their mutagen potential [73, 83]. The NP inhibitors nucleozin, naproxen, and RK424 all disrupt NP-NP oligomerization, NP-RNA interaction, and NP cellular trafficking, which leads to impaired virus replication and assembly [73, 84-87]. Finally, the PB2 and PA inhibitors, VX-787 and L-742001, respectively, block influenza mRNA synthesis [88].

VX-787, which is currently undergoing a phase 2b clinical trial, blocks the cap-snatching

mechanism that influenza uses to add a host pre-mRNA 5’-cap needed for viral mRNA transcription [89, 90]. Though protection by VX-787 is promising, the production of resistance variants has already been detected [91]. L742001 works in a similar mechanism to VX-787 by blocking the PA endonuclease used to generate 5’-capped oligonucleotides used as primers for viral mRNA transcription [73]. So far, serial

passages have not derived resistant influenza strains, though an in vitro mutational

analysis showed potential resistance mutations in PA [92].

The current prophylactic and therapeutic strategies we have against influenza

virus, in all, have targeted nearly every critical part of viral entry, replication, assembly,

and budding. Unfortunately, nearly all of these options have generated, or have the likely

potential to generate, transmissible resistance mutants. Advancements of small molecule

inhibitors directly targeting influenza virus are exceedingly promising, though much

more work has to be done in ensuring resistance and creation of more virulent strains are

minimized. The rapid and adaptive evolution of influenza is an extreme hurdle in both

vaccine development and anti-influenza small molecule inhibitor development, as

evidenced by the sheer variability in efficacy of these treatments. Though continued

work in these fields is greatly encouraged, it is imperative that we equally focus on the

host immune response to influenza infection, as the secrets to developing broad anti-

influenza therapies may also be uncovered by those studies.

1.2: Host defense against influenza A virus

1.2.1: Anatomical and physiobiological barriers

Influenza virus is most commonly spread through exposure to aerosolized respiratory droplets from an infected to a non-infected individual, either from direct exchange through coughing or sneezing, or by indirect transfer to the respiratory system by fomites [93, 94]. The skin provides an effective barrier to influenza infection, as the virus is unable to penetrate the dead cells of the epidermis. Should influenza virus enter

the respiratory tract, it encounters a mucus layer, protecting the epithelial cells of the

respiratory tract against foreign particles and invaders through a complex assortment of polysaccharides, glycoproteins, lipoproteins, lipids, and hydrolytic enzymes [95]. For

example, the mucosal layer of the respiratory tract houses sialic acid-containing mucins that can bind up influenza A viruses through the interaction of sialic acid and NA; the greater the concentration of sialic acid in the respiratory mucosa, the less successful influenza A is at infecting host epithelial cells [96, 97]. However, the sialidase activity of

NA can reduce the sialic acid in the mucosa, and sufficient inoculating titer of influenza

A virus will subvert this natural barrier and lead to infection of epithelial cells [97].

1.2.2: Viral detection and innate immunity

Epithelial cells - Successful infection of airway epithelial cells by influenza virus triggers what is known as the interferon response, which is named after the resulting viral interference pathways initiated [98, 99]. The interferon response is one of the first lines of defense against all invading pathogens, and plays a critical role in the control and resolution of influenza infection. Influenza virus is able to avoid detection in the epithelial endosome, establishing infection before antiviral measures are triggered.

However, the processes of viral mRNA synthesis and RNA replication reveal infection to the host cell by two different means: 1) Detection of 5’-triphosphate viral RNA by retinoic acid-inducible gene I (RIG-I) [100], and 2) Activation of the NRLP3

inflammasome by M2 ion channel-mediated proton flux. RIG-I is a critical initiator of type I interferon (IFN) in epithelial cell; upon binding to viral RNA, RIG-I undergoes a

conformational change and binds to mitochondrial antiviral signaling protein (MAVS)

[101]. From there, MAVS signaling activates key immune transcription factors NF-κB

and interferon regulatory factor 3 (IRF3), leading to the transcription of the pro- inflammatory cytokines pro-IL-1β and pro-IL-18, as well as type I IFN and interferon stimulated genes (ISGs), respectively [102-105]. The importance of RIG-I and subsequent downstream signaling events is underscored by the influenza A NS1 protein, which has evolved to inhibit RIG-I function [106].

The NLRP3 inflammasome is also critical for viral detection and response in

epithelial cells [107]. As viral proteins are transported through the trans-Golgi network,

the M2 ion channel exports protons from the Golgi into the cytoplasm, which neutralizes

the pH of the trans-Golgi network and prevents low pH-induced fusion of HA to Golgi membranes; the resulting flux of protons into the cytoplasm is detected by, and activates, the NLRP3 inflammasome [108]. Activation of the NLRP3 leads to homodimerization, binding to the adaptor protein ASC, and the binding and cleaving of pro-caspase-1 [109].

Activated caspase-1 is then able to cleave pro-IL-1β and pro-IL-18, which were produced following RIG-I activation, and thereby the cleavage products IL-1β and IL-18 can be secreted [107].

Alveolar macrophages – Macrophages are relatively large, bone marrow-derived innate immune cells that engulf foreign material and debris through a process known as phagocytosis [110]. Alveolar macrophages, along with other phagocytic cells such as monocytes and neutrophils, are important for the clearance of virus-infected cells and debris [111]. Like epithelial cells, alveolar macrophages can be infected with influenza 13

virus, and detect viral infection through RIG-I and NLRP3. In addition, macrophages consuming influenza-infected cells detect viral dsRNA via the phagosomal Toll-Like

Receptor 3 (TLR3) [107]. TLR3 activation, much like RIG-I activation, leads to the downstream activation of NF-κB and IRF3 and subsequent production of pro- inflammatory cytokines, type I IFNs, and ISGs [103, 107].

Plasmacytoid dendritic cells (pDCs) – pDCs are the only cell type in the lung that is

able to detect influenza virus without active infection or replication of the virus taking

place first [112]. They express endosomal TLR7, which recognizes the ssRNA genomes

contained within the virion, and activation of TLR7 either initiates IRF7 activation

directly, or initiates NF-κB activation when bound to MyD88 [113, 114]. The activation of IFR7, like IRF3, results in the production of type I IFNs and ISGs [115].

Other innate immune cells – Chemokines that are secreted by the cells mentioned above recruit the help of additional immune cells, natural killer cells, monocytes, and neutrophils, to the site of infection [116]. Natural killer cells recognize the differential expression of surface receptors in infected epithelial cells, and induce safe and controlled cell death by apoptosis [117]. The apoptotic blebs are then cleared by phagocytic

monocytes, neutrophils, and macrophages [118].

1.2.3: The interferon-stimulated genes

ISGs are critical in the cellular protection of cells actively undergoing infection as

well as healthy cells neighboring infected cells. The result of type I IFNs, and the ISGs

that are produced, is an antiviral state of the cell that is designed to inhibit viral infection,

replication, assembly, and budding [119]. Over 5% of the total cellular genes are

transcribed in response to interferon, and the importance of many of these genes in protecting against influenza has been well-established [120]. Some ISGs directly act on

viral components. For example, MXA binds to NP and prevents nuclear import of the

virus [121]. ISG15, a ubiquitin mimetic, directly modifies viral proteins in a process called ISGylation [122]. TRIM22, an E3 ubiquitin ligase, targets NP for degradation.

Finally, RNase L is produced specifically to cleave viral RNA [123]. Other ISGs alter

host cell properties to prevent efficient virus infection and production. For example,

cholesterol 25-hydroxylase (CH25H) converts cholesterol to 25-hydroxycholesterol, a

molecule that directly blocks viral fusion [124]. Protein kinase R (PKR) phosphorylates

EIF2α, effectively blocking new translation [125]. Viperin interferes with lipid raft

formation, preventing viral budding [126]. Finally, the IFITM proteins, the major focus

of the following chapters and reviewed in Chapter 1.4, broadly inhibit viral fusion by

altering endosomal membranes.

1.2.4: Adaptive immunity and viral clearance

Often in the case of influenza infection, the innate immune system is overwhelmed and unable to clear an established infection. As such, a functional adaptive immunity needs to be recruited to achieve this aim.

Antigen-presenting cells (APCs) – The burden of alerting the adaptive immune response of an established infection rests on the APCs, which are cells that process antigens engulfed by phagocytosis, and present these antigens on their cell surface to

“show” to adaptive immune cells. The lung-relevant APCs include alveolar macrophages, dendritic cells (DCs) and interstitial macrophages [127]. DCs are highly motile, and respond to cytokines and chemokines secreted by infected epithelial cells to activate, take up viral antigen, and traffic to the draining lymph nodes. Activated DCs also increase expression of co-stimulatory and adhesion molecules to more effectively communicate with adaptive immune cells in the lymph node [112, 128-131].

Importantly, DCs utilize cross-presentation, which enables them to process and present extracellular antigens with MHC class I molecules to cytotoxic CD8+ T cells, in addition to presenting antigens via MHC class II to CD4+ T cells, which would otherwise only be accomplished through infection of the APC [132].

T cell activation and effector functions - Respiratory DC migration to the nearest draining lymph node serves to concentrate the influenza antigens away from the site of infection, thereby greatly increasing the odds of antigen recognition by influenza-specific naïve CD4+ and CD8+ T cells [133]. Cytokine signals, co-stimulatory molecules, and successful antigen matching between the DC and naïve T cell leads to activation,

proliferation, and differentiation of T effector cells [133, 134]. The activation of respiratory DCs in response to influenza infection is skewed towards inflammatory, and the influx of these inflammatory DCs lead to the differentiation of antigen-specific CD4+

T helper cells to the inflammatory TH1 effector cell type [135]. Activated TH1 effector cells are thought to mainly support the activation and differentiation of B cells for the purposes of antibody production, but are also proficient at forming a strong memory subset [136]. CD8+ effector functions are critical in the resolution of influenza infection, as they efficiently and specifically target influenza-infected cells for apoptosis [137]. To this aim, epithelial and immune cells infected and actively producing virus are effectively removed from the lung without further damaging uninfected neighbors.

Resolution of inflammation following viral clearance - The lung environment requires an anti-inflammatory, immunosuppressive environment in order to maintain homeostatic balance. However, the processes required for viral clearance are indeed inflammatory, causing unintended tissue damage to the lung epithelium. Therefore, the inflammatory response must be kept in perfect balance to facilitate complete viral clearance while minimizing host damage. T regulatory (Treg) cells are antigen-specific cells that keep the

TH1-induced inflammation in check; in fact, evidence shows a direct balance between the inflammatory severity of TH1 cells and the immunosuppressive activity of Treg cells, actively preventing a positive-feedback loop of damaging inflammation [138-140].

Furthermore, Treg cell-secreted IL-10, a potent anti-inflammatory cytokine, is essential for ramping down the immune response following viral clearance [141]. CD4+ and CD8+ effector T cells also have important regulatory functions once reaching the infected lung;

CD4+ effector T cell-secreted IL-2, and IL-27 from resident mononuclear cells, lead to

significant increase of both CD4+ and CD8+ effector T cell-secreted IL-10 [142]. This secretion does not occur in the lymph node – rather, the infected lung microenvironment seems to promote this unique effect [141, 142]. Finally, inflammation throughout and after influenza infection is partially limited by epithelial cell-secreted TGFβ, which has immunosuppressive effects and is constitutively expressed in the uninfected lung [143].

Following suppression of pro-inflammatory mediators, TGFβ and IL-10 trigger the apoptosis of infiltrating neutrophils, and promote immunosuppressive and tissue repair characteristics in macrophages [111, 144-146].

1.3: The IFITMs as broad inhibitors of viral infection

1.3.1: Evolutionary origins

The interferon-induced transmembrane proteins (IFITMs), first mentioned in

Chapter 1.3.2, are a group of proteins within the Dispanin gene family, which was formed around the 14 human proteins that contain a common ‘two transmembrane’ structure; that is, two separate α-helical domains [147]. The Dispanins originate in a common ancestor of choanoflaggellates and metazoa, most likely from a horizontal gene transfer from bacteria. Through gene duplication events, the Dispanins further expanded from chordates to form the five IFITM proteins expressed in modern humans (IFITMs 1,

2, 3, 5, and 10) [147, 148]. Despite their name, only IFITM 1-3 are interferon-inducible

[149, 150]. IFITM5, which first appeared in bony fish, is constitutively expressed in osteoblasts, and is required for proper bone mineralization [151, 152]. The clinical 18

significance of IFITM5 was established in a study finding a variant in the IFITM5 gene that leads to bone malformations associated with osteogenesis imperfecta type V [152].

IFITM10, originally arising in the tetrapods and the most conserved IFITM, is expressed primarily in the adrenal gland, and its function is still unknown [147, 148]. As a result of the ancient origins of IFITM proteins, IFITM paralogues exist in most vertebrates, including mammals, birds, reptiles, fish, and amphibians [147]. This studies presented hereafter focus on IFITMs 1-3, due to their established role as broad antiviral inhibitors.

1.3.2: Discovery of antiviral activity

Although the interferon-inducibility of IFITMs had been known since 1984 [149], the antiviral activity of IFITMs was not discovered and established until 2009 [153].

Brass et al performed an siRNA knockdown screen to identify novel host factors affecting influenza A virus infection, finding that IFITM3 knockdown substantially increased susceptibility of cells to infection by two different influenza strains (A/PR/8/34

(H1N1) and A/NWS/33 (H1N1)). Overexpressing the IFITMs in retrovirally-transduced cell lines, the authors established that IFITMs conferred significant protection to these cells against a wide array of influenza viruses, including A/Udorn/72 (H3N2),

A/Brisbane/59/07 H1N1, A/Uruguay/716/07 H3N2, A/Aichi/2/68 H3N2, Hong Kong

1968 pandemic virus, and pseudotyped particles coated with HA1, 3, 5, and 7. These results clearly demonstrate the broad anti-influenza activity that the IFITMs possess, and pinpointed this activity at the viral entry point using the pseudotyped particles.

Furthermore, none of the influenza variants tested against the IFITMs was resistant,

possibly suggesting an antiviral mechanism broad enough to protect against all current

and future influenza variants. Importantly, the IFITMs also conferred resistance to the

significant human pathogens dengue virus and West Nile virus, providing further

significance of the IFITM proteins in restricting viral infection [153].

Soon after, two subsequent studies verified the findings by Brass et al through

independent means. In the first of these, a large-scale palmitoylome-profiling of IFN- treated dendritic cells led to the discovery that IFITM3 antiviral activity was almost entirely dependent on its palmitoylation (reviewed further in Chapter 1.3.5), revealing an important consideration in determining the mechanism of IFITM antiviral activity [154].

The second study identified IFITM3 as a broadly-acting antiviral effector as part of a larger overexpression screen of over 380 interferon-stimulated genes, and further expanding list of known viruses inhibited by IFITM3 (yellow fever virus and vesicular stomatitis virus) [155]. While IFITMs, and IFITM3 especially, were proving to be potent antiviral factors in vitro, the in vivo relevance of these proteins was yet to be determined.

In 2012, Everitt et al provided significant evidence that IFITM3 was critical in

restricting influenza virus in both mouse and human. This study utilized Ifitm3-/- mice,

and found that mice infected with mild doses of influenza suffered greater morbidity and

mortality than their wild-type counterparts. Mice lacking this single factor present with

greater weight loss, higher viral titers, severe lung pathology, and decreased overall

survival. The same work also highlighted a polymorphism in human IFITM3 that is

predicted to alter a splice site acceptor region, the minority SNP rs12252-C. The authors

propose that this polymorphism may lead to the production of a splice variant lacking the

first 21 amino acids, and discovered that overexpressing this variant lost almost all antiviral activity conferred by wild-type IFITM3. Though the sample size was low, the authors showed an enrichment of the rs12252-C/C (homozygous for the IFITM3 variant) among patients hospitalized in the United Kingdom for influenza-positive illness during the 2009 H1N1 pandemic. Taken together, this study simultaneously provided in vivo evidence of IFITM3 as a potent anti-influenza effector in both mouse and man [156].

Furthermore, several additional studies by independent groups have shown evidence of association between the rs12252-C allele and increased severity in influenza [157-160] and HIV-1 [161] disease progression.

Since its identification of potent antiviral activity, the IFITM proteins, most notably IFITM3, have been studied extensively in multiple labs around the world. As a result, the implication of the IFITM proteins as broad antiviral inhibitors has expanded dramatically, showing promise that future IFITM-based therapies can be effective against human and animal pathogens beyond influenza. A summary of this research, and the viruses that IFITM proteins are able to inhibit, is presented in Table 2 below.

pH- Pseudotyped Restricts IFITMs Family Virus dependent (P) or Live (L) References Infectivity restricting Fusion Virus Enveloped

Orthomyxoviridae Influenza A virus Yes; pH <6 Yes P L 1-3 [153]

Influenza B virus Yes; pH <6 Yes L 1-3 [156]

Flaviviridae West Nile virus Yes; pH >6 Yes P 1-3 [153]

Dengue virus Yes; pH <6 Yes P 3 [153]

Hepatitis C virus Yes; pH >6 Yes P L 1 [162]

Zika virus Yes; pH >6 Yes L 1,3 [163]

Vesicular Rhabdoviridae Yes; pH >6 Yes P L 1-3 [164] Stomatitis virus

Rabies virus Yes; pH <6 Yes P 2-3 [165]

Lagos Bat virus Yes; pH <6 Yes P 2-3 [165]

Yes; Filoviridae Marburg virus requires Yes P L 1-3 [166] lysosome Yes; Ebola virus requires Yes P L 1-3 [166]

lysosome Yes; Coronaviridae SARS coronavirus requires Yes P L 1-3 [166] lysosome Fusion not Retroviridae HIV-1 Yes P L [167, 168] required Moloney leukaemia Fusion not No P L None [153] virus required Jaagsiekte sheep Yes; pH >6 Yes P 1 [169] retrovirus Arenaviridae Lassa virus Yes; pH >6 No P None [153] Machupo virus Yes; pH >6 No P None [153]

Lymphocytic choriomeningitis Yes; pH >6 No P None [153]

virus

Semliki Forest Alphaviridae Yes; pH >6 Yes L 2,3 [170] virus Sindbis Virus Yes; pH >6 Yes L [170]

Continued

Table 2: Comprehensive list of viruses tested against the IFITM proteins Adapted from Smith et al, 2014 [173]

Table 2 Continued

Bunyaviridae La Crosse virus Yes; pH <6 Yes L 1-3 [171]

Hantaan virus Yes; pH <6 Yes L 1-3 [171]

Andes virus Yes; pH <6 Yes L 1-3 [171]

Rift Valley fever Yes; pH <6 Yes L 2,3 [171] virus Crimean-Congo haemorrhagic fever Yes; pH <6 No L None [171]

virus Non-enveloped

Reoviridae Reovirus Yes; pH <6 Yes L 3 [172]

1.3.4: Proposed mechanisms of action

Although the importance of the IFITM proteins both in vitro and in vivo is well

appreciated, generation of IFITM-based therapies has been hindered by a lack of understanding of how IFITM3 confers its antiviral activity. An important consideration in determining this mechanism is the range of viruses that IFITM3 is able (or not able) to inhibit. IFITM3 is known to inhibit viral infections by preventing virus fusion with endolysosomal membranes [168, 169, 174, 175], in agreement with the Brass et al 2009 study using retroviruses pseudotyped with distinct viral surface proteins that showed

IFITM3 inhibition of infection is dependent upon the viral glycoprotein used for entry and fusion [153]. Comprehensive lists of viruses shown to be inhibited by IFITMs confirm that the majority of viruses inhibited by IFITM3, including influenza virus,

SARS coronavirus, dengue virus, and West Nile virus, among many others, share a common ability to enter cells via endocytosis and utilize pH-dependent fusion 23

mechanisms (Table 2) [173, 176]. Conversely, Sendai virus and Moloney leukemia virus, which fuse at the cell surface, are only modestly affected by IFITM3 [153, 177].

Additionally, treatment of SARS coronavirus with trypsin allows the virus to fuse at the cell surface by eliminating the need for its proteolytic activation by cathepsin L in lysosomes, resulting in virus evasion of IFITM3-mediated restriction [166].

IFITM3 dimerizes [178] and localizes with markers of endosomes and lysosomes, and these compartments are characterized by increased size and decreased pH in cells expressing IFITM3 [166, 174, 179, 180]. Yount et al utilized a hyperactive IFITM3 mutant lacking ubiquitination sites to find that Rab5- and Rab7-positive endosomes and acidic LAMP1-positive lysosomes are aberrantly merged upon IFITM3 expression, providing a compartment for viral degradation (and explaining why we refer to IFITM3- positive compartments as endolysosomes) [179]. This compartment also stains positive for the autophagosome marker LC3 [174, 179], and has characteristics of multivesicular bodies when visualized by electron microscopy [181], indicating that the effects of

IFITM3 on the endocytic pathway are multifaceted. Indeed, influenza virus particles have been visualized to enter cells expressing IFITM3 but are eliminated by degradation prior to their fusion/escape from IFITM3-positive endolysosomes [174, 182]. The expansion and alteration of the endolysosomal compartment in IFITM3-expressing cells and inhibition of influenza virus were found to be independent of the canonical autophagy protein 5-dependent autophagy induction pathway [179], but may involve an interaction between IFITM3 and the vacuolar ATPase [180].

The confluence of data discussed above supports a model whereby viruses are

degraded in acidic endolysosomes in IFITM3-expressing cells (Figure 1). However, this model contains a paradox in that influenza virus fusion is normally triggered by lowered pH in late endosomes and yet fusion is apparently prevented in low-pH endolysosomes when IFITM3 is present (Figure 1) [169, 174, 175, 182]. Two primary theories have emerged to explain inhibition of viral fusion by IFITM3: 1) Theory 1 stems from the observation that increased cholesterol levels are present within endolysosomes of cells expressing IFITM3 as compared to control cells [175, 181]. Changes in cholesterol were shown to be dependent upon an interaction of IFITM3 with vesicle-associated membrane protein-associated protein A (VAPA) that interrupts a homeostatic partnership between

VAPA and a cholesterol trafficking protein [181]. Proponents of this theory posit that dysregulation of cholesterol homeostasis is responsible for virus inhibition. However, more recent publications have challenged this notion showing minimal effects of either cholesterol depletion or modulation of VAPA levels on IFITM3 activity [175, 182].

Likewise, Neiman-Pick type C1 fibroblasts, which contain high levels of cholesterol in the late endosomal compartment, show little change in influenza virus infectivity when compared to control cells [181, 182]. Thus, though potentially complementary to other antiviral functions of IFITM3, it remains unclear whether cholesterol dysregulation is essential for the primary mechanism of action of IFITM3. 2) Theory 2 has arisen from two independent studies demonstrating decreases in membrane fluidity in cells expressing IFITMs [169, 182]. The necessity of membrane fluidity for efficient fusion of a multitude of viruses has been appreciated for decades [183], and though IFN-induced

enzymes can influence membrane lipid composition [184], IFITM3 would be the first cellular antiviral effector that limits infection through directly altering the “rigidity” of cell membranes [185]. Discovering the molecular details through which IFITM3 alters the structural features of endolysosomal membranes to inhibit virus fusion, and deciphering the potential role that cholesterol dysregulation may play in this process are exciting areas of future research.

Figure 1: Model of influenza virus entry and IFITM3 antiviral activity. Influenza virus enters cells via receptor-mediated endocytosis and fuses in the acidic late endosomal compartment, depositing its genomic contents into the cytosol, thereby avoiding lysosomal hydrolases. In cells expressing IFITM3, viruses are localized to an enlarged acidic and degradative compartment staining positive for endosomal and lysosomal markers. Virus fusion is inhibited by alterations to the endolysosomal membrane imposed by IFITM3, and virions are subsequently degraded. IFITM3 is depicted as a dimer. EE, early endosome; LE, late endosome; L, lysosome; EL, endolysosome.

1.3.5: The Complex Topology of IFITM3

The membrane topology of IFITM3 is a subject of controversy, and evidence

suggests that it adopts multiple conformations, with the antivirally-active topology being a topic of debate. Understanding this characteristic of active IFITM3 is essential for understanding its biophysical interactions with cellular membranes, its interactions with other cellular proteins, and ultimately its mechanism of antiviral action. However,

IFITM3’s presence at multiple cellular compartments including the plasma membrane,

ER, autophagosomes, and multivesicular endolysosomes, along with its potential for multiple topologies, has made traditional topology mapping with protease digestion and protection assays difficult to interpret. Thus, other methods, including mapping of PTMs, have been used to infer its membrane topology.

IFITM3 is predicted to be a dual-pass transmembrane protein with its N- and C- termini facing into the lumen of cellular organelles such as the ER and endolysosomes, or into the extracellular space (Figure 2) [186]. Indeed, epitope tags at both termini can be detected extracellularly by antibody staining of non-permeabilized cells [153, 164, 187].

Additionally, antibodies that bind the region linking the two hydrophobic domains require detergent permeabilization to stain IFITM3-expressing cells, indicating that this region is entirely intracellular [186]. Likewise, the positioning of the three conserved S- palmitoylated cysteines is consistent with this dual transmembrane topology [154].

Overall, these data would suggest that some portion of IFITM3 indeed exists as a dual- pass transmembrane protein (Figure 2). However, fractionation of cellular cytoplasmic and membrane compartments indicated that a significant portion of IFITM3 is present in

the cytoplasm, i.e., it is not strongly membrane-associated [179]. The lack of an ER

signal sequence at the N-terminus also makes it unclear precisely how efficiently IFITM3

membrane insertion might occur. These fractionation results were reminiscent of those

obtained for the palmitoylated protein caveolin-1, in which its hydrophobic segment does not fully traverse the membrane, but rather forms a proposed hairpin-like structure within the membrane known as an intramembrane domain [188].

As mentioned earlier, an evolutionary study of IFITMs identified 9 additional human genes that are likely related to IFITMs [147]. We examined whether topology mapping of any of the protein products for these genes had been previously performed.

Membrane topology data for one of the IFITM relatives known as synapse differentiation-inducing gene protein 1 (SYNDIG1) indicated that the second hydrophobic stretch in this protein is an intramembrane domain (Figure 2) [189].

Topology mapping for the additional eight relatives of the IFITMs remains to be performed. Nonetheless, this work provided a precedent for the presence of intramembrane domains within this family of proteins, and is also in accord with a previous prediction that the second hydrophobic domain of IFITMs may not be a true transmembrane domain due to the lack of sufficient anchoring residues at their C-termini, and because IFITMs from some species are truncated in this region [186].

The possibility that the transmembrane domains of IFITM3 might actually be intramembrane domains was first proposed based first on the observations that a portion of IFITM3 is not strongly membrane-associated, and that IFITM3 palmitoylation increases its membrane affinity [179]. Additionally, analysis of post-translational

modifications (PTMs), which are reviewed further in the next section, supported the notion that the N-terminus of active IFITM3 is oriented toward the cytoplasm rather than extracellularly. First, ubiquitination of Lys-24 likely occurs in the cytoplasm since this is dogmatically the location of ubiquitin ligases [179]. Second, the conserved tyrosine 20

residue is phosphorylated by the cytoplasmic kinase Fyn [167, 190]. Thus, the

cytoplasmic orientation of the N-terminus is essential for proper IFITM3 localization and

activity. It might also be concluded from these data that IFITM3 stained with N- terminus-directed antibodies at the cell surface of non-permeabilized cells does not represent antivirally-active IFITM3. It was also demonstrated that the N-terminus of an

IFITM3 construct appended with a myristoylation motif is modified by cytoplasmic myristoylation machinery, and that this modification supported IFITM3 membrane association and antiviral activity [179]. In addition, a naturally occurring N-linked glycosylation consensus sequence in the N-terminus, 2-NHT-4, is only minimally modified by ER-lumenal glycosylation machinery [179, 187]. Further glycosylation sites engineered within both the N- and C-terminal regions were mostly unmodified, agreeing with a cytoplasm-facing orientation of these domains for the majority of this protein

[179]. Thus, the overwhelming data in support of active IFITM3 having a cytoplasmically oriented N-terminus has resulted in a general consensus that the first hydrophobic stretch within active IFITM3 is likely acting as an intramembrane domain

(Figure 2) [176, 178, 179, 187, 191].

Conflicting evidence exists concerning whether the second hydrophobic stretch within active IFITM3 fully traverses the membrane bilayer. IFITM3 staining at the cell

surface is stronger when staining for C-terminal epitope tags as compared to N-terminal tags, suggesting that a more significant fraction of the second hydrophobic domain fully spans the plasma membrane [187, 191]. It is also reported that extensions of the C- terminus are partially cleaved in lysosomes [187]. This work suggested that a significant fraction of cellular IFITM3 exists as a type II transmembrane protein (Figure 2).

Supporting this topology model, untagged human IFITM5 was recently demonstrated to adopt a type II transmembrane topology [192]. As IFITM5 has a significantly extended

C-terminal domain compared to IFITM3 and localizes primarily to the cell surface, the relevance of this topology determination in relation to IFITM3 is not clear. Likewise, the importance of the type II transmembrane conformation in the antiviral activity of IFITM3 remains to be demonstrated, particularly since the percentage of cellular IFITM3 that adopts this topology has not been determined, and because this model was developed using IFITM3 constructs with C-terminal epitope tags that more than double the length of the C-terminal domain, potentially altering its ability to anchor a transmembrane region

[187, 191]. Conversely, IFITM3 appended with a C-terminal CaaX box is prenylated by cytoplasmic enzymes, and is active against influenza virus [179]. Likewise, the murine

IFITM1 protein has a non-conserved cysteine present within its C-terminus that is cytoplasmically S-palmitoylated and contributes to antiviral activity [177]. Overall, additional experiments involving endogenous IFITM3 or creatively engineered constructs that report on both IFITM3 topology and activity are required to conclusively determine the antivirally-active IFITM3 membrane topology.

Figure 2: Membrane topologies for IFITM3 and SYNDIG1 that are supported by experimental evidence Data have been described supporting each of the shown topology models. Hydrophobic segments are shown in red. The IFITM3 dual-transmembrane topology is predicted based on hydropathy plotting. However, IFITM3 in which its first hydrophobic segment adopts an intramembrane conformation is required for N-terminal access to cytoplasmic components necessary for its proper localization and activity. Whether the second hydrophobic segment in active IFITM3 traverses both membrane leaflets remains unclear. Yellow circles represent conserved cysteines that are known to be palmitoylated on IFITM3. Note that the cysteines in the first hydrophobic segment of SYNDIG1 appear earlier within the segment than IFITM3 cysteines making them more deeply embedded in the membrane and less likely to be palmitoylated. The third cysteine in SYNDIG1 is absent, agreeing with its experimentally determined topology. Potential changes in membrane properties induced by intramembrane domains are graphically depicted by altered membrane curvature.

1.3.6: Regulation of IFITMs by Post-translational modifications

Palmitoylation – Palmitoylation is the enzymatic addition of a 16-carbon fatty acid to

proteins, most often occurring on cysteine residues (known as S-palmitoylation in reference to the thioester linkage between the fatty acid and the sulfur-bearing side chain)

[193]. This hydrophobic modification results in the membrane attachment of otherwise cytoplasmic proteins [194, 195]. Palmitoylation can also have a variety of effects on proteins that are already membrane-associated, including targeting to specific cellular compartments, modulation of protein interactions or stability, and modulation of protein structure or positioning in membranes [196]. Yount et al identified IFITM3 in a chemical proteomics screen designed to detect immunity-related proteins that are post- translationally palmitoylated, and confirmed IFITM3 S-palmitoylation occurring on three cysteines within or adjacent to its two hydrophobic domains at positions 71, 72, and 105

[154, 197]. We were able to confirm IFITM3 S-palmitoylation occurring on three cysteines within or adjacent to its two hydrophobic domains at positions 71, 72, and 105

[154, 179, 198]. A palmitoylation-deficient, triple cysteine-to-alanine IFITM3 mutant showed a more diffuse pattern of localization than WT IFITM3, indicating that S- palmitoylation promotes IFITM3 clustering, and also suggested that, despite its two hydrophobic stretches predicted to be transmembrane domains, a fraction of IFITM3 may not be associated with membranes [154]. Cellular fractionation experiments indicated that a significant amount of endogenous IFITM3 is not strongly membrane-bound, and that membrane association is increased by S-palmitoylation [179]. Importantly, over-

expressed IFITM3 was able to inhibit influenza virus infection in a variety of cells types,

and this effect was largely lost upon mutation of its S-palmitoylated cysteines [154].

Alternative lipid modification sites added to the N- and C-termini of IFITM3

(myristoylation and prenylation, respectively) were able to rescue the membrane affinity

and antiviral activity of the palmitoylation-deficient IFITM3 mutant, suggesting that S- palmitoylation does not act by inducing a structural change in the IFITM3 hydrophobic regions, but rather, acts by increasing membrane affinity [179]. Thus, studies of IFITM3

S-palmitoylation would indicate that proper membrane anchoring, and potentially also

clustering of IFITM3, are the features mediated by S-palmitoylation that are essential for

its antiviral function. S-palmitoylation is conserved on all IFITMs that have been tested

to date (human IFITMs 1, 2, 3 [154] and 5 [199], and mouse IFITMs 1 [177] and 3

[154]), and indeed the palmitoylated cysteines are among the most conserved IFITM

residues present throughout evolution [147]. Interestingly, palmitoylation of murine

IFITM1 is essential for preventing its proteasomal degradation [177], and palmitoylation

of human IFITM5 is necessary for its interaction with a factor regulating its ability to

promote bone formation [199], suggesting that, though S-palmitoylation is a conserved

modification, it may have distinct effects on different IFITMs. Palmitoyltransferase(s)

responsible for modifying IFITM3 and other IFITMs have not yet been identified. These

enzymes may themselves represent novel virus susceptibility factors if polymorphisms in

the human population exist that render them unable to modify IFITM3, thus warranting

their identification and study.

Ubiquitination – Ubiquitination is the addition of the 9 kDa ubiquitin polypeptide to proteins on lysine residues. Upon overexposure of IFITM3 western blots, specific bands above the 15 kDa expected molecular weight can be observed at approximately 24 and 33 kDa [179, 190], corresponding to shifts that would be predicted for mono- and di- ubiquitination, respectively. By immunoprecipitating IFITM3 and examining these upper bands by tandem mass spectrometry, ubiquitin peptides and lysine-containing IFITM3 peptides that were covalently modified with a di-glycine motif, which is the characteristic signature of ubiquitination that remains following trypsin digestion, were identified

[179]. Several large-scale mass spectrometry studies aiming to identify ubiquitinated proteins have also identified various IFITMs as substrates for ubiquitination [200-203].

Mass spectrometry results were confirmed by anti-ubiquitin blotting of immunoprecipitated IFITM3, which revealed that IFITM3 is also poly-ubiquitinated with both lysine-48 and lysine-63 linkages [179]. These polyubiquitin linkages, which often control protein turnover or trafficking, respectively [204], suggest that ubiquitination of

IFITM3 has multiple effects.

IFITM3 contains four conserved lysines. IFITM3 constructs in which three of the four lysines were mutated to alanine allowed the demonstration that ubiquitination can occur at all four positions, though lysine 24 appeared to be the most strongly-modified residue particularly with lysine-64 polyubiquitin chains [179]. However, lysine-24 mutagenesis resulted in no overt changes in antiviral activity, localization, or abundance of IFITM3, indicating that other lysines may functionally compensate for the role of ubiquitination on this residue [167, 178, 179]. In fact, a novel phenotype was only

observed for IFITM3 when all four lysines were mutated to alanine. Lysine-less IFITM3 showed increased stability in pulse-chase experiments, and localized entirely to endolysosomes, losing the overlap with ER markers that is observed for a portion of WT

IFITM3 [179]. This may suggest that ubiquitinated IFITM3 is recruited to an ER- proximal site for degradation. Hyperactivity was also observed for the ubiquitination- deficient, lysine-less IFITM3 mutant in inhibiting influenza virus correlating with its complete localization to endolysosomes, thereby further supporting a model of antiviral activity in which endolysosomal localization is essential [179]. The specific roles of the different linkages of polyubiquitin chains on IFITM3 remain to be decrypted, and likewise, the ubiquitin ligase(s) responsible for modifying IFITM3 have yet to be identified. As the available data suggest that ubiquitination is a negative regulator of

IFITM3 stability and activity, these enzymes, once identified, may represent possible therapeutic targets for improving IFITM3-based antiviral immune responses.

Methylation – In addition to being ubiquitinated, lysine 88 of IFITM3 can also be monomethylated by Set7 [205], a lysine methyltransferase that was initially reported to methylate histone H3 [206]. Aside from the predominant methylation of chromatin and its primary localization in the nucleus, Set7 also methylates several cellular, non-histone proteins and the HIV protein Tat [207-211]. Thus, IFITM3 is one of a growing list of proteins reportedly regulated by Set7. Monomethylation of lysine 88 was discovered using mass spectrometry, and was confirmed upon development of a lysine 88- methylation-specific anti-IFITM3 monoclonal antibody [205]. In this study, methylation was increased by Set7 overexpression, correlating with a loss of antiviral activity.

Conversely, knockdown of Set7 decreased IFITM3 methylation and resulted in enhanced

IFITM3-mediated restriction of influenza virus and vesicular stomatitis virus.

Interestingly, infection with either of these viruses increased IFITM3 methylation measured as early as 4 hours post infection, while IFN treatment decreased methylation.

Since the net result of virus infection is increased methylation, despite the presumed induction of type I IFNs, this may suggest that these viruses actively promote IFITM3 methylation as an immune evasion strategy.

The clear importance of IFITM3 methylation, as established by experiments modulating levels of Set7 [205], make it surprising that mutation of lysine 88 does not result in major changes to IFITM3 activity against influenza virus, dengue virus, or vesicular stomatitis virus [178, 205]. As lysine 88 can also be ubiquitinated [179], the

prevention of these two negative regulatory modifications through lysine 88 mutation

might have been expected to increase antiviral activity. However, analogously to the

ability of multiple IFITM3 lysines to be ubiquitinated, compensatory methylation may

occur at another lysine in the absence of lysine 88, which will require further

investigation. Continued study to determine what effect methylation has on IFITM3

protein-protein interactions, stability, or modification with other PTMs will aid in

mechanistically characterizing the effects of this modification. Despite the withstanding

questions, Set7 is capable of affecting IFITM3 antiviral activity, and this knowledge

broadens the repertoire of IFITM3-modulating enzymes that represent potential targets

for antiviral therapy.

Phosphorylation – As part of a study to further understand the apparent antiviral defect

in the rs12252-C/C IFITM3 variant, Jia et al revealed the importance of the 21 N-

terminal amino acids that are removed from the minor variant. They found that by

removing the first 21 amino acids, substantial dysregulation of cellular trafficking

resulting in relocation from endolysosomes to the cell periphery [167]. This work found

that, while the rs12252-C/C variant was ineffective at limiting influenza, which fuses at

endosomes, it was just as effective as wild-type IFITM3 in restricting pH-independent

HIV-1. Thus, the antiviral activity of the IFITM3 variant seems to be preserved, but the

N-terminal region mediates the cellular trafficking of IFITM3 necessary to restrict pH- dependent viruses in the endosome. Finally, this work showed that phosphorylation of

IFITM3 on tyrosine 20 of human IFITM3, and is mediated specifically by the Src-family protein-tyrosine kinase Fyn [167]. Fyn is cotranslationally myristoylated and postranslationally palmitoylated, rapidly targeting and anchoring it to the cytoplasmic leaflet of the plasma membrane [212, 213], and its direct interaction with IFITM3 supports the notion that IFITM3 naturally traffics to the plasma membrane [167]. Jia et al developed tyrosine phosphorylation-deficient IFITM3 mutants (Y20A and ∆Y20), and by eliminating tyrosine 20 phosphorylation, noted a similar disruption of cellular localization reminiscent of the rs12252-C/C variant localization along with a significant defect in antiviral activity against vesicular stomatitis virus [167]. We sought to confirm these results in influenza virus, investigate the mechanism by which tyrosine phosphorylation regulates cellular trafficking and antiviral activity, and determine if post- translational crosstalk exists between phosphorylation and the other IFITM3 PTMs.

Chapter 2: Phosphorylation of IFITM3 dually regulates its endocytosis and

ubiquitination

2.1: Abstract

Interferon-inducible transmembrane protein 3 (IFITM3) is essential for innate defense against influenza virus in mice and humans. IFITM3 localizes to endolysosomes where it prevents virus fusion, though mechanisms controlling its trafficking to this cellular compartment are not fully understood. We determined that both mouse and human IFITM3 are phosphorylated by Fyn kinase on tyrosine 20, and that mouse IFITM3 is also phosphorylated on the non-conserved tyrosine 27. Phosphorylation led to a

cellular redistribution of IFITM3, including plasma membrane accumulation. Mutation

of tyrosine 20 caused a similar redistribution of IFITM3 and resulted in decreased antiviral activity against influenza virus, while tyrosine 20 mutation of mouse IFITM3

showed minimal effects on localization or activity. Using Fyn knockout cells, we also found that IFITM3 phosphorylation is not a requirement for its antiviral activity.

Together these results indicated that tyrosine 20 is part of an endocytosis signal that can be blocked by phosphorylation or by mutation of this residue. Further mutagenesis narrowed this endocytosis-controlling region to four residues conforming to a YxxΦ endocytic motif, which, when transferred to CD4, resulted in its internalization from the cell surface. Additionally, we found that phosphorylation of IFITM3 by Fyn and

mutagenesis of tyrosine 20 both resulted in decreased IFITM3 ubiquitination. Overall,

these results suggest that modification of tyrosine 20 may serve in a cellular checkpoint

controlling IFITM3 trafficking and degradation, and demonstrate the complexity of post-

translational regulation of IFITM3.

2.2: Introduction

The interferon-inducible transmembrane proteins (IFITMs) inhibit cellular

infection by a wide range of significant viral pathogens [153, 164, 166, 168, 171, 172,

178, 179, 214]. IFITM3 is particularly important for restriction of influenza virus, as

IFITM3 knockout mice succumb to sub-lethal doses of virus [156, 215], and a deleterious polymorphism in the human IFITM3 gene has been associated with increased severity of infection in at least three independent studies [156, 216, 217]. Despite the clear importance of IFITM3 in restricting influenza virus, many questions remain regarding its mechanism of action, cellular trafficking patterns, and regulation by cellular enzymes.

IFITM3 localizes to acidic compartments staining positive for endosomal and

lysosomal markers [166, 174, 179], where it prevents viral fusion through an unknown

mechanism [168, 169, 174]. Experiments with pseudotyped viruses demonstrated that

inhibition of viruses by IFITM3 is dependent upon the viral fusion glycoprotein used for

cellular entry [153]. Likewise, nearly all of the viruses shown to be inhibited by IFITM3 enter cells via endocytosis [176, 218]. Conversely, Sendai virus, which fuses at the cell

surface, is largely unaffected by IFITM3 [177]. Similarly, exogenous protease treatment

of SARS coronavirus allows it to fuse at the cell surface, thereby bypassing its pH-

dependent activation in lysosomes and restriction by IFITM3 [166]. Thus,

endolysosomes appear to be the site of antiviral action by IFITM3, and the targeting

signals that control IFITM3 trafficking to and from this compartment warrant further

study.

IFITM3 is a highly-regulated protein with at least four post-translational modifications occurring on multiple residues reported to date. We first reported palmitoylation of IFITM3 occurring on three cysteines that are essential for proper membrane anchoring and antiviral activity [154, 179, 197, 198]. We also reported ubiquitination of IFITM3 occurring on four lysines [179]. This modification negatively regulates IFITM3 by targeting the protein away from endolysosomes for degradation

[179]. Set7-dependent methylation on a single lysine has also been described to negatively regulate IFITM3 antiviral activity through an unknown mechanism [205].

Finally, IFITM3 phosphorylation by the tyrosine-protein kinase Fyn on tyrosine 20 has been reported [167]. Mutation of tyrosine 20 resulted in decreased antiviral activity against vesicular stomatitis virus and an accumulation of IFITM3 at the plasma membrane [167]. These intriguing findings prompted us to further investigate the role of phosphorylation in controlling IFITM3 trafficking and in restricting influenza virus.

2.3: Materials and Methods

Cell Culture, Transfections, and Western Blotting – Mouse Embryonic Fibroblasts

(MEFs) deficient in SRC, YES, and FYN kinases (SYF cells, ATCC), WT MEFs, and

HEK293T cells were cultured in DMEM supplemented with 4.5 g/L D-Glucose, L-

Glutamine, 110 mg/L sodium pyruvate, and 10% fetal bovine serum (Thermo Scientific) at 37ºC and 5% CO2 in a humidified incubator. For Western blotting, cells were plated for 90% confluency in 6-well plates 24 h prior to transfecting with 2 µg of indicated plasmid per well using Lipofectamine 2000 (Life Technologies). Cells for microscopy were plated for 50% confluency on glass coverslips in 12-well plates, and transfected with 1 µg of indicated plasmid per well. IFITM3 constructs were expressed from the pCMV-HA or pCMV-myc vectors as described previously [154]. Mutants were made using the QuikChange Multi Site-Directed Mutagenesis Kit (Stratagene). Plasmid expressing human FYN was provided by Dr. Marilyn Resh (Memorial Sloan Kettering

Cancer Center) [212], and plasmid expressing FLAGubiquitin was provided by Drs.

Rebecca Delker and F. Nina Papavasiliou (Rockefeller University). Control siRNA or

IFITM3 siRNA (Ambion, 4390816) was transfected into SYF cells using Lipofectamine

RNAiMax transfection reagent (Life Technologies).

For Western blotting, cells were lysed with 1% Brij buffer (0.1 mM triethanolamine,

150 mM NaCl, 1% BrijO10, pH 7.4) containing EDTA-free protease inhibitor cocktail.

For phospho-tyrosine westerns, cells were treated with sodium orthovanadate (Sigma) for

1 h prior to harvesting, and PhosSTOP phosphatase inhibitor cocktail was added to the lysis buffer (Roche).

Immunoprecipitations were performed using EZview Red Anti-c-myc or Anti-HA

Affinity Gel (Sigma), as indicated. Western blotting was performed with anti-p-Tyr-100

(Cell Signaling, 9411), anti-myc (Clontech, 51826), anti-FYN (Cell Signaling, 4023), anti-actin (Abcam, ab3280), anti-FLAG (Sigma, F7425), anti-HA (Clontech, 631207),

anti-IFITM3 (Abcam, ab65183), and anti-GAPDH (Invitrogen, 398600). Primary antibodies were all used at a 1:1000 dilution.

Fluorescence Microscopy – Cells were fixed for 10 min with 3.7% paraformaldehyde, permeabilized with 0.1% Triton X-100 in PBS for 10 min, and blocked for 10 min with

2% FBS in PBS. Primary antibodies, anti-myc (1:500), antiFYN (1:500), and anti-CD4

(directly conjugated to Alexa Fluor 488, 1:100, BD Pharmingen, 557695), and Alexa

Fluor-labeled secondary antibodies (Life Technologies) were diluted 1:1000 in 0.1%

Triton X-100 in PBS for staining of permeabilized cells. Cells were treated with antibodies sequentially for 20 min at room temperature, and washed five times with 0.1%

Triton X-100 in PBS after each antibody treatment. For nonpermeabilized cells, anti-

CD4 was added to live cells in ice-cold PBS for 15 minutes, followed by five washes with ice-cold PBS, and paraformaldehyde fixation. Glass slides were mounted in

ProLong Gold Antifade Reagent containing DAPI (Life Technologies). Images were captured using a Fluoview FV10i Confocal Microscope (Olympus).

Infections and Flow Cytometry – Influenza virus A/PR/8/34 (H1N1) was propagated in

10-day embryonated chicken eggs for 48 h at 37ºC and titrated using Madin-Darby canine kidney cells as described previously [219]. HEK293T were infected at a multiplicity of infection (MOI) of 2.5 for 6 h. SYF cells were infected at an MOI of 5 for

24 h. Infected cells were washed with PBS and harvested in 0.25% Trypsin EDTA.

Cells were fixed in 3.7% paraformaldehyde for 10 min and permeabilized with 0.1%

Triton X-100 for 10 min. Cells were stained with anti-myc and antiinfluenza nucleoprotein (NP) (Abcam ab20343, 1:333) directly conjugated to Alexa Fluor 488 and

Alexa Fluor 647, respectively, using a 100 µg antibody labeling kit (Life Technologies).

All antibodies were diluted in 0.1% Triton X-100 in PBS, and cells were stained for 20 min. Cells were washed twice with 0.1% Triton X-100 in PBS after each antibody treatment. PBS was used for final resuspension of cells for cytometric analysis using a

FACSCanto II flow cytometer (BD Biosciences).

2.4: Results

The previous finding that mutagenesis of the phosphorylated residue tyrosine 20 results in decreased antiviral activity of IFITM3 might suggest that phosphorylation is necessary for the function of IFITM3 [167]. In addressing this hypothesis, we first sought to confirm that human (h) IFITM3 is phosphorylated, and also to examine whether this modification is conserved on murine (m) IFITM3, which possesses both tyrosine 20 and also a non-conserved tyrosine nearby at amino acid position 27 (Figure 3A). For these experiments, we utilized a murine fibroblast cell line deficient in Src, Yes, and Fyn kinases (SYF cells), in which we observed nearly undetectable tyrosine phosphorylation of myc-tagged IFITM3 (Figure 3, B and C). Upon Fyn co-expression, tyrosine phosphorylation of both mIFITM3 and hIFITM3 could be detected, in agreement with the conclusion published previously showing that Fyn can indeed phosphorylate hIFITM3

[167] (Fig. 3, B and C). Tyrosine mutagenesis indicated that hIFITM3 is phosphorylated

solely on tyrosine 20 (Figure 3B), whereas mIFITM3 phosphorylation can occur on both tyrosine 20 and 27 (Figure 3C).

It is important to note that sodium orthovanadate pretreatment of live cells to inhibit phosphatases was required to visualize IFITM3 phosphorylation, as reported previously [167]. This observation suggests that IFITM3 phosphorylation is a dynamic and reversible process that involves one or multiple unknown phosphatases. Also of note are the reproducibly-observed upper bands in anti-myc blots that likely represent phosphorylated IFITM3 (Figure 3, B and C). This band does not correlate directly with

anti-phospho-tyrosine blots, therefore suggesting that additional phosphorylation sites on

serine or threonine residues may exist on IFITM3. Investigation into these alternative

phosphorylation sites and potential phosphatases modifying IFITM3 are of particular

interest in our future studies. Nonetheless, tyrosine-specific phosphorylation clearly

occurs primarily on tyrosine 20 and 27.

Figure 3: IFITM3 phosphorylation by Fyn A) Alignment of hIFITM3 and mIFITM3 amino acids 15-30. Tyrosines are highlighted with grey shading. B,C) SYF cells were co-transfected overnight with indicated myc- tagged hIFITM3 (B) or mIFITM3 (C) constructs and either vector control or plasmid expressing Fyn. Cells were then treated for 1 h with sodium orthovanadate. Cell lysates were subjected to anti-Fyn and anti-actin western blotting or were subjected to anti-myc immunoprecipitation followed by blotting for anti-phospho-tyrosine (p-Tyr) and anti- myc. Blots are representative of at least three experiments.

Fyn-mediated phosphorylation of IFITM3 is not required for anti-influenza virus activity. Having confirmed tyrosine phosphorylation on both mIFITM3 and hIFITM3 by

Fyn, we sought to test whether endogenous IFITM3 is active in SYF cells. As Fyn is knocked out in these cells, it is inferred that endogenous IFITM3 is lacking phosphorylation. We and others [153, 179] have shown that murine fibroblast lines express a basal amount of active IFITM3 that limits infection of these cells. Thus, we performed siRNA knockdown of IFITM3 in SYF cells (Figure 4A) and, subsequently, examined infection by influenza virus. The percentage of infected SYF cells increased significantly upon IFITM3 knockdown, indicating that endogenous IFITM3 is still active in these cells, and that IFITM3 tyrosine phosphorylation by Fyn is not a requirement for its antiviral function (Figure 4B). 46

Figure 4: IFITM3 is active in the absence of Fyn A,B) SYF cells were transfected for 18 h with control siRNA (siCont) or siRNA targeting IFITM3 (siIFITM3). A) Cells were collected just prior to infection for confirmation of IFITM3 knockdown by anti-IFITM3 western blotting. Anti-GAPDH blotting served as a loading control. B) Following siRNA treatment, cells were infected with influenza virus strain PR8 at an MOI of 5 for 24 h. Cells were then fixed and stained with anti-influenza NP to measure the percentage of cells that were infected using flow cytometry. Results shown are an average of six samples from two independent experiments. The average percent infection of siCont cells was set to 1 for calculation of relative fold infection. Error bars represent standard deviation. Student’s t-test was used to calculate the indicated p-value.

The tyrosine 20 residue of IFITM3 is required for complete antiviral activity. We

next examined several of our mIFITM3 and hIFITM3 tyrosine mutants for the ability to

inhibit influenza virus compared with WT IFITM3. HEK293T cells are a commonly used

cell line for the analysis of IFITM3 mutants because they express no detectable amount

of endogenous IFITM3, are readily transfectable, and retain the ability to be highly

infected with influenza virus even after transfection, allowing a large dynamic range for

observations of differences in antiviral protective effects between various IFITM3

mutants [154, 179, 220]. Tyrosine 20 mutants of both mIFITM3 and hIFITM3 are

expressed at levels similar to their respective WT proteins (Figure 5A), but lose anti- influenza virus activity, as determined by an established flow cytometry assay that measures the percentage of cells infected in each condition (Figure 5B) [154, 179, 220].

Interestingly, mutation of tyrosine 27, although phosphorylatable (Figure 3C), has no apparent effect on antiviral activity of mIFITM3, and does not significantly compensate for mutation of tyrosine 20 (Figure 5B). These results demonstrate that tyrosine 20 is a critical amino acid for the anti-influenza virus activity of both mIFITM3 and hIFITM3, while tyrosine 27 on mIFITM3 is not.

Figure 5: Tyrosine 20 of IFITM3 is necessary for complete antiviral activity against influenza virus A,B) HEK293T cells were transfected overnight with the indicated myc-tagged IFITM3 constructs or vector control, and were infected for 6 h with influenza virus strain PR8 at an MOI of 2.5. Cells were then fixed and stained with anti-myc and anti-influenza NP for confirming expression of IFITM3 and for measuring the percentage of cells that were infected, respectively, using flow cytometry. A) Myc-positive cells were gated based on lack of staining in the vector control and analyzed in histogram form to confirm comparable expression of IFITM3 and tyrosine mutants. B) Myc-positive cells were gated as in A and analyzed for percent infection based on anti-influenza nucleoprotein staining using non-infected samples as a baseline for gating. Results shown are an average of at least nine samples from a minimum of three independent experiments. The average percent infection of cells expressing WT mIFITM3 was set to 1 for calculation of relative fold infection. Error bars represent standard deviation. Results are presented on a log2 scale for ease of visualizing differences between WT IFITM3 and tyrosine mutants. Student’s t-test was used to calculate the indicated p-values.

Tyrosine Phosphorylation of IFITM3 leads to plasma membrane accumulation.

Previous imaging of hIFITM3 tyrosine 20 mutants demonstrated that mutation of this

residue results in retention of IFITM3 at the plasma membrane [167, 178]. However,

these experiments did not address whether phosphorylation of tyrosine 20, or the tyrosine

20 residue itself, is necessary for IFITM3 internalization. We reasoned that imaging

IFITM3 in SYF cells lacking Fyn, or co-expressing IFITM3 with Fyn, would allow

effects of phosphorylation on IFITM3 localization to be observed. Although IFITM3 in

the absence of Fyn was localized in punctate clusters intracellularly, as is normally

observed for IFITM3 staining in most cell types [153, 174, 178, 179], IFITM3 in the

presence of overexpressed Fyn was redistributed, including accumulation at the plasma

membrane (Figure 6, A and B). This effect was seen for both mIFITM3 and hIFITM3,

and mimics the effect we also observed for tyrosine 20 mutants (Figure 6, A and B).

Interestingly, tyrosine 27 mutation of mIFITM3 had no apparent effect on its cellular

distribution, agreeing with the lack of effect observed on antiviral activity for this mutant

compared with WT mIFITM3 (Figures 5 and 6B). Retention of IFITM3 at the plasma

membrane through a direct interaction with Fyn is a possible interpretation of these

results. However, it has been observed previously that the interaction between Fyn and

IFITM3 depends on the presence of tyrosine 20 [167], and tyrosine 20 mutants also show

plasma membrane accumulation, arguing against a direct role for interaction with Fyn in

sequestering IFITM3 at the plasma membrane. Therefore, we conclude that both tyrosine

20 phosphorylation and mutation result in a similar change in the cellular distribution of

IFITM3, including plasma membrane accumulation.

Figure 6: IFITM3 tyrosine 20 phosphorylation or mutation affects IFITM3 localization including plasma membrane accumulation A,B) SYF cells were co-transfected with the indicated myc-tagged hIFITM3 (A) or mIFITM3 (B) constructs and either vector control or plasmid expressing Fyn. Representative merged fluorescent confocal microscopy images for nuclear staining (DAPI, blue) and anti-myc staining (red) are shown. Insets depicting anti-Fyn staining (grayscale) are shown for some conditions in order to confirm Fyn expression or the lack thereof.

Identification of a putative IFITM3 endocytosis motif. Our imaging results suggest

that tyrosine 20 may be part of an endocytosis signal that can be blocked by

phosphorylation or perturbed by mutating tyrosine 20 (Figure 6). Previous IFITM3

mutagenesis has shown that the 17-PPN-19 residues immediately prior to tyrosine 20 are

dispensable for antiviral activity [167]. Likewise, we observed that mutation of lysine 24

does not affect IFITM3 localization or activity [179]. This narrows the hIFITM3

endocytic motif to the region 20-YEML-23 (Figure 7A). Interestingly, this sequence

conforms to the pattern of a YxxΦ motif (where X is any amino acid, and Φ is valine,

leucine, or isoleucine) that is involved in the adaptor protein complex-mediated

endocytosis of a multitude of membrane proteins [221]. Our experiments with mIFITM3

also support that this motif is an endocytosis signal because the 20-YERI-23 sequence of mIFITM3 also conforms to this pattern, whereas the 27-YEVA-30 motif involving tyrosine 27 does not (Figure 7A). This agrees with our data that tyrosine 27, although phosphorylatable, does not play a major role in IFITM3 cellular distribution or antiviral activity (Figures 5 and 6B). To test the hypothesis that the 20-YEML-23 sequence in

IFITM3 is an endocytosis signal, we made a conservative mutation of the Φ residue in hIFITM3 to valine (L23V), and a non-conservative mutation to the polar residue glutamine (L23Q). Imaging of these mutants revealed a similar localization of the L23V mutant, which preserves the canonical YxxΦ pattern, whereas the L23Q mutant was redistributed, including visible plasma membrane localization similar to what we saw previously with tyrosine 20 mutants (Figures 6A and B, and 7B).

Figure 7: Identification of a YxxΦ endocytosis motif in IFITM3 A) Alignment of amino acids 15-30 of hIFITM3 and mIFITM3. Tyrosines are highlighted with grey shading and the conserved YxxΦ motif is labeled and underlined. B) SYF cells were transfected overnight with the indicated myc-tagged hIFITM3 constructs. Representative merged fluorescent confocal microscopy images for nuclear staining (DAPI, blue) and anti-myc staining (red) are shown.

To confirm that the YEML sequence of hIFITM3 functions as an endocytosis

signal, we transferred this tetrapeptide to the cytoplasmic C-terminal region of CD4,

which normally localizes in part to the plasma membrane. Anti-CD4 staining of the cell

surface of non-permeabilized cells showed outlining of cells transfected with a plasmid

encoding WT myc-tagged CD4 (Figure 8A). Under the same conditions, minimal

staining of a CD4 construct containing the YEML peptide was observed (Figure 8A),

although this construct was expressed strongly and localized in internal compartments, as

indicated by staining of permeabilized cells with anti-CD4 (Figure 8A) or anti-myc

(Figure 8B). Overall, these results indicate that the YEML motif from hIFITM3 causes a 53

robust internalization of CD4 (Figure 8). Taken together, our results provide evidence that the hIFITM3 YEML motif functions as an endocytosis signal that can be regulated by Fyn-mediated phosphorylation.

Figure 8: The YEML motif from hIFITM3 causes internalization of CD4 A,B) MEFs were transfected overnight with myc-tagged CD4 or CD4-YEML constructs. A) Non-permeabilized or permeabilized cells were stained with anti-CD4 antibody for fluorescent confocal microscopy. Representative merged images for nuclear staining (DAPI, blue) and anti-CD4 staining (green) are shown. B) Permeabilized cells were stained with anti-myc antibody for confocal fluorescent microscopy. Representative merged images for nuclear staining (DAPI, blue) and anti- myc staining (red) are shown.

Tyrosine 20 phosphorylation regulates IFITM3 ubiquitination. Having previously

discovered IFITM3 ubiquitination [179] and noting the often-observed cross-talk

between phosphorylation and ubiquitination [222], we sought to investigate a potential

link between these two modifications on IFITM3. In our previous studies of IFITM3, we

found that, upon overexposure of IFITM3 western blot analyses, bands above the

expected IFITM3 molecular weight could be observed [179]. We identified this banding

pattern as ubiquitination through the use of mass spectrometry, anti-ubiquitin blotting,

and lysine mutagenesis [179]. Here we employed a straightforward assay whereby we

transfected mIFITM3 and hIFITM3 into HEK293T cells, with or without overexpression

of Fyn, and examined the banding pattern of IFITM3 by western blotting. Interestingly,

bands that we identified previously as mono- and di-ubiquitinated IFITM3 [179] were

diminished upon Fyn overexpression for both mIFITM3 and hIFITM3, whereas the bands

at the expected molecular weight were largely unaffected by Fyn (Figure 9A). We then

compared the banding patterns for WT IFITM3 and tyrosine mutants. Tyrosine 20

mutants of both mIFITM3 and hIFITM3 showed a decreased intensity of ubiquitinated

bands, whereas a tyrosine 27 mutant of mIFITM3 was ubiquitinated similarly to WT mIFITM3 (Figure 9B). A ubiquitination-deficient lysine-less mutant of mIFITM3

(termed Ub∆, note that the myc tag contains one lysine) we generated previously [179] was included as a control in this experiment to confirm that the higher molecular weight bands we observed indeed represent ubiquitination (Figure 9B).

To further visualize IFITM3 ubiquitination, including poly-ubiquitinated IFITM3, and to be sure that the lysine residue present within the myc epitope tag was not altering

our results, we co-transfected HA-tagged mIFITM3 and tyrosine mutants with FLAG- ubiquitin and performed anti-HA immunoprecipitation followed by blotting for HA and

FLAG. mIFITM3 constructs were chosen for this experiment because of the availability

of the mIFITM3-Ub∆ construct, which served as a negative control, and because we also

previously generated a palmitoylation-deficient mIFITM3 construct (Palm∆) that does

not have a defect in ubiquitination and served as an additional control [179].

Ubiquitination patterns observed in this experiment agreed with our previous results, in

that tyrosine 20 mutants showed a partial defect in ubiquitination, whereas tyrosine 27

mutation had no effect compared with WT mIFITM3 (Figure 9C). Overall, these data

suggest that the phosphorylatable residue tyrosine 20 is involved in promoting IFITM3

ubiquitination in addition to its role in promoting endocytosis.

Figure 9: Unmodified tyrosine 20 is necessary for proper IFITM3 ubiquitination A) 293T cells were co-transfected overnight with the indicated myc-tagged IFITM3 constructs and vector control or plasmid expressing Fyn. Anti-myc and anti-fyn blotting were performed. B) 293T cells were transfected overnight with the indicated myc-tagged IFITM3 constructs and anti-myc blotting was performed. A,B) Anti-myc blots were overexposed to allow visualization of ubiquitinated IFITM3. Single- and double-headed arrows indicate mono-ubiquitinated and di-ubiquitinated IFITM3, respectively. IFITM3 at the expected molecular weight of 15kDa serves as a loading control. C) 293T cells were co-transfected overnight with the indicated HA-tagged IFITM3 constructs and plasmid expressing FLAG-tagged ubiquitin. Anti-HA immunoprecipitation was performed followed by anti-HA and anti-FLAG blotting. Results in A-C are representative of at least three experiments.

2.5: Discussion

The post-translational regulation of IFITM3 is rich with complexity. At least eight distinct amino acids within this small, 15 kDa protein are modified with at least four different PTMs, including phosphorylation, palmitoylation, ubiquitination, and methylation [154, 167, 179, 205]. We demonstrate here for the first time that there is cross-regulation of PTMs on IFITM3, namely phosphorylation of IFITM3 by Fyn at

tyrosine 20 decreases IFITM3 ubiquitination (Figure 9). This is in contrast to the

previously-observed independent nature of IFITM3 ubiquitination with respect to its

palmitoylation [179].

Crosstalk between phosphorylation and ubiquitination has been extensively

documented for multiple other proteins [222]. Particularly, phosphorylation often serves

as a signal for modification by E3 ubiquitin ligases. For example, the CBL family of E3

ubiquitin ligases recognize phosphorylated tyrosines on their substrates [223], which coincidentally include Fyn [224]. Thus, we expected that IFITM3 phosphorylation would promote ubiquitination. However, we observed the opposite in that IFITM3

phosphorylation by Fyn, led to a defect in ubiquitination, and mutating the Tyrosine 20

phosphorylation site resulted in the same effect (Figure 9).

The observed decrease in ubiquitination of IFITM3 upon phosphorylation may

have at least two possible explanations. First, IFITM3 endocytosis may be required for

ubiquitination to occur. We have found that phosphorylation blocks an endocytosis motif

that is necessary for proper localization and full antiviral activity of IFITM3 (Figures 3-

8). Our imaging data indicate that IFITM3 traffics to the plasma membrane and is either

retained there upon phosphorylation or is internalized (Figures 6,7). Since the

submission of this study as a manuscript, a second group independently reported

identification of the 20-YEML-23 motif of hIFITM3 as a critical determinant for its

endocytosis [191], thus validating and complementing our observations regarding

phosphorylation of this motif. If ubiquitination takes place at the endolysosome or another cellular compartment, it may be inhibited by retention of IFITM3 at the plasma membrane upon tyrosine 20 mutations or phosphorylation. Second, tyrosine 20 may be part of an additional motif recognized by ubiquitin ligases, again explaining the similar results we obtained when either blocking tyrosine 20 by phosphorylation or by mutation

(Figure 9). The sequence 17-PPNY-20 in both mIFITM3 and hIFITM3 presents a potential E3 ligase interaction motif, as it conforms to the PPxY pattern recognized by

HECT E3 ubiquitin ligases (Figure 7A) [225]. Both of these findings should aid significantly in the future identification of the IFITM3 ubiquitin ligase(s) responsible for modifying IFITM3.

It remains to be determined what effect relocalization of IFITM3 upon tyrosine 20 mutation or upon increasing Fyn activity would have on its range of viral restriction. For instance, IFITM1, which lacks a YxxΦ motif and localizes in part to the plasma membrane, has been described to have an overlapping but somewhat distinct specificity for different viruses as compared to IFITM3 [166, 171, 218]. Likewise, it will be interesting to examine what effect drugs that inhibit Src-family tyrosine kinases such as

Fyn would have on IFITM3 activity, as phosphorylation of IFITM3 by Fyn has both potentially negative and positive effects through modulating endocytosis and

ubiquitination, respectively. Overall, the continued study of IFITM3 post-translational

modifications, their crosstalk, and their mechanisms of regulating antiviral activity will

be important for understanding and controlling IFITM3 biology for combating influenza

and other viruses.

2.6: Contributions and Acknowledgements

Experiments were conceived by Nicholas Chesarino and Jacob Yount. Jacob

Yount performed Fyn knockdown experiments (Figure 4) and the CD4 endocytosis assay

(Figure 8), and Nicholas Chesarino performed the remaining experiments. Technical assistance was provided by Jocelyn Hach and Temet McMichael. The manuscript was written by Nicholas Chesarino and Jacob Yount, with editorial assistance from Jocelyn

Hach and Temet McMichael.

This study was supported by funding from the NIH/NIAID (grant R00AI095348 to Dr. Jacob Yount) and from The Ohio State University Public Health Preparedness for

Infectious Disease program. I was supported by The Ohio State University Systems and

Integrative Biology Training Program (NIH/NIGMS grant T32GM068412). Thank you to Dr. Marilyn Resh of the Memorial Sloan Kettering Cancer Center for providing Fyn constructs and Drs. Rebecca Delker and F. Nina Papavasiliou of the Rockefeller

University for providing the FLAG-Ubiquitin-expressing plasmid. We also thank Dr.

Howard Hang of the Rockefeller University for helpful discussions.

This work was published in the Journal of Biological Chemistry on April 25, 2014, issue

289, pages 11986-11992. Citation of this manuscript is as follows:

Chapter 3: Identification of E3 ubiquitin ligase NEDD4 as a negative regulator of

IFITM3

3.1: Abstract

Interferon (IFN)-induced transmembrane protein 3 (IFITM3) is a cell-intrinsic factor that limits influenza virus infections. We previously showed that IFITM3 degradation is increased by its ubiquitination, though the ubiquitin ligase responsible for this modification remained elusive. Here, we demonstrate that the E3 ubiquitin ligase

NEDD4 ubiquitinates IFITM3 in cells and in vitro. This IFITM3 ubiquitination is dependent upon the presence of a PPxY motif within IFITM3 and the WW domain- containing region of NEDD4. In NEDD4 knockout mouse embryonic fibroblasts, we observed defective IFITM3 ubiquitination and accumulation of high levels of basal

IFITM3 as compared to wild type cells. Heightened IFITM3 levels significantly protected NEDD4 knockout cells from infection by influenza A and B viruses. Similarly, knockdown of NEDD4 in human lung cells resulted in an increase in steady state IFITM3 and a decrease in influenza virus infection, demonstrating a conservation of this NEDD4- dependent IFITM3 regulatory mechanism in mouse and human cells. Consistent with the known association of NEDD4 with lysosomes, we demonstrate for the first time that steady state turnover of IFITM3 occurs through the lysosomal degradation pathway.

Overall, this work identifies the enzyme NEDD4 as a new therapeutic target for the

prevention of influenza virus infections, and introduces a new paradigm for up-regulating

cellular levels of IFITM3 independently of IFN or infection.

3.2: Introduction

Interferon (IFN)-induced transmembrane protein 3 (IFITM3) is a small 15 kDa

protein that restricts cellular infection by influenza virus [153, 154, 166]. IFITM3 is

active against all strains of influenza virus that have been tested to date, regardless of

serotype or species of origin [153, 166, 176, 177, 218], and it similarly inhibits many

other medically important viruses such as HIV, SARS coronavirus, and Ebola virus [166,

168, 176, 226, 227]. Confirming its importance in vivo, IFITM3 knockout mice succumb to sublethal doses of influenza virus [156, 215]. Likewise, IFITM3 is the only known protein for which a genetic polymorphism present in a significant percentage of certain human populations is associated with severe influenza virus infections [156, 159, 216,

217]. In the cell, IFITM3 localizes to endosomes and lysosomes [174, 179, 228], and

traps endocytosed virus particles within these degradative compartments by impeding the

formation of the virus fusion pore [169, 174, 175]. Yet, even with this potent mechanism

by which IFITM3 limits infections, influenza virus remains a significant health concern

[7, 20]. This may be explained by the fact that IFITM3 is present at low levels within

most cells at steady state and is induced by IFNs only after infection has already been

established [149, 153, 215]. The inability to up-regulate IFITM3 levels independently of

infection or IFNs is a challenge preventing the field from harnessing the activity of

IFITM3 for infection prevention.

We previously showed that ubiquitination increases the rate of IFITM3 turnover

within the cell [179]. A non-ubiquitinated lysine-to-alanine mutant of IFITM3 possessed

enhanced antiviral activity and a longer half-life as compared to WT IFITM3 [179].

These findings indicated that inhibition of IFITM3 ubiquitination could augment the activity and/or levels of endogenous IFITM3, thus offering a strategy for exploiting

IFITM3 therapeutically or prophylactically against viral infections. The identification of the E3 ubiquitin ligase(s) capable of modifying IFITM3 among the more than 600 annotated E3 ligases in the human genome will be an important step toward validating this antiviral strategy.

Through our work studying tyrosine phosphorylation of IFITM3, we discovered that phosphorylation at tyrosine 20 (Y20) inhibited IFITM3 ubiquitination [190]. This led us to posit that phosphorylation of Y20 may block an E3 ubiquitin ligase recognition signal. Indeed, Y20 is part of a highly conserved PPxY motif (where P = proline, x = any amino acid, and Y = tyrosine, Figure 10A) [229]. PPxY motifs are commonly recognized by WW (characterized by two tryptophan residues spaced approximately 20 amino acids apart) domains of NEDD4-family E3 ubiquitin ligases, of which there are nine family members [225]. We chose to focus first on NEDD4, the prototypical member of this family, for several reasons: 1) NEDD4 and IFITM3 both have ubiquitous expression patterns while several other NEDD4-family members are tissue specific (BioGPS.org

[230]), 2) Like IFITM3, many of the known NEDD4 substrates are membrane proteins and are associated with endosomal and lysosomal pathways [225, 231], 3) IFITM3 and

NEDD4 are both S-palmitoylated, suggesting that they may localize to similar membrane

subdomains [154, 232], and 4) NEDD4 is reported to be inhibited by ISG15 [233-235],

an IFN-inducible protein, thus providing an intriguing model whereby IFN might induce

IFITM3 expression while also inhibiting its ubiquitination. Herein, we provide results

demonstrating the ability of NEDD4 to ubiquitinate IFITM3 and identify a unique role

for NEDD4 in decreasing steady state IFITM3 abundance, leading to increased cellular

susceptibility to influenza virus infection.

3.3: Materials and Methods

Cell Culture, Transfections, and siRNA Knockdowns — All cell lines used in these

studies (HEK293T, A549, NCI-H358, NCI-H2009, and MEFs) were cultured in DMEM supplemented with 4.5 g/liter D-glucose, L-glutamine, 110 mg/liter sodium pyruvate, and

10% fetal bovine serum (Thermo Scientific) at 37 °C and 5% CO2 in a humidified

incubator. HEK293T and A549 cells were purchased from ATCC. NCI-H358 and NCI-

H2009 cells were obtained from the ATCC and provided to us by Dr. Gustavo Leone

(The Ohio State University). NEDD4 WT and KO MEFs used in this study were

generated by Dr. Hiroshi Kawabe (Max Planck Institute) [236] and were kindly provided

to us by Dr. Matthew Pratt (University of Southern California) who also generated the

retrovirally reconstituted control cell lines. For western blotting, cells were plated for

90% confluency in 6-well plates for 24 h prior to transfection with 2 µg/well of plasmids

using Lipofectamine 2000 (Invitrogen). For microscopy, MEFs were plated for 50%

confluency on glass coverslips in 12-well plates for 24 h prior to overnight treatment with

IFN-α (BEI Resources). IFITM3 constructs were expressed from the pCMV-HA or

pCMV-myc vectors (Clontech) as described previously [154, 177, 190]. IFITM3 mutants

were made using the QuikChange Multi site-directed mutagenesis kit (Stratagene).

Plasmids expressing HA-hNEDD4 and HA-hNEDD4-C867A were obtained from

Addgene (plasmids 27002 and 26999, deposited by Dr. Joan Massagué, Memorial Sloan

Kettering Cancer Center) [237], and plasmids expressing FLAG-NEDD4, FLAG-

NEDD4-∆WW, and HA-CBL-B were kindly provided by Dr. Jian Zhang (The Ohio State

University). FLAG-SMURF1 and FLAG-SMURF2 were obtained from Addgene

(plasmids 11752 and 11746, desposited by Jeff Wrana, University of Toronto) [238].

The plasmid expressing FLAG-ISG15 was obtained from Addgene (plasmid 12443,

deposited by Dr. Dong-Er Zhang, University of California San Diego) [239].

IFITM3 knockdown in MEFs was performed using Silencer Select Ifitm3 siRNA

(Ambion, catalog no. 4390816) and negative control (Ambion, catalog no. 4390844).

Human NEDD4 knockdown in A549, NCI H358, and NCI H2009 cells was performed

using Dharmacon ON-TARGETplus SMARTpool Human NEDD4 (GE Healthcare,

catalog no. L-007178-00) and Dharmacon ON-TARGETplus Control Pool Non-targeting control (GE Healthcare, catalog no. D-001810-10-20). siRNAs were transfected into cells using Lipofectamine RNAiMax transfection reagent (Invitrogen). Transfection of siRNA was performed for 24 h for mIFITM3 knockdown, and 48 h for NEDD4 knockdown. For

Western blotting, cells were lysed with 1% Brij buffer (0.1 mM triethanolamine, 150 mM

NaCl, 1% BrijO10 (Sigma), pH 7.4) containing EDTA-free protease inhibitor mixture

(Roche) and 25 µM MG132 (Sigma). Immunoprecipitations were performed using

EZview Red anti-c-myc or anti-HA affinity gel (Sigma), or with Protein G Plus Agarose

Suspension (Calbiochem) in conjunction with anti-mIFITM3.

Co-Immunoprecipitation (Co-IP) Assays – Co-immunoprecipitation assays were adapted from a previously described protocol [240]. HEK293T cells were co-transfected overnight with plasmids expressing myc-tagged IFITM3 and FLAG-NEDD4. Cells were washed twice with PBS, lysed on ice in Triton X-100 lysis buffer (50 mM Hepes, pH 7.5,

150 mN NaCl, 1% Triton X-100, 10% glycerol, 1.5 mM MgCl2, 1.0 mM EGTA, 10

µg/mL leupeptin, 10 µg/mL aprotinin, 10 µg/mL pepstatin, and 1 mM PMSF) for 5 min, and centrifuged at 1,000 rcf (g) (5 min, 4ºC). 50 µg of cell lysate was set aside for each sample in order to evaluate, via Western blotting, expression of myc-hIFITM3, FLAG-

NEDD4, and GAPDH as a loading control. Equal concentrations of cell lysate were

immunoprecipitated using 15 µL EZview Red anti-c-myc or anti-FLAG affinity gel

(Sigma) per sample for 1 h at 4ºC with gentle nutation. Immunoprecipitations were washed three times with lysis buffer and examined by Western blotting with both anti-

myc and anti-FLAG for each immunoprecipitate.

Western Blotting and Antibodies — Western blotting was performed with anti-myc

(Developmental Studies Hybridoma Bank at the University of Iowa, deposited by Dr. J.

Michael Bishop, catalog no. 9E 10), anti-HA (Clontech, catalog no. 631207), anti-

hIFITM3 (Proteintech Group, catalog no. 11714-1-AP), anti-mIFITM3 (Abcam, catalog

no. ab65183), anti-NEDD4 (Millipore, catalog no. 07-049), anti-FLAG (Sigma, catalog 68

no. F7425), anti-actin (Abcam, catalog no. ab3280), or anti-GAPDH (Invitrogen, catalog

no. 398600) antibodies. All primary antibodies were used at a 1:1000 dilution.

Secondary antibodies, Goat Anti-Mouse IgG, HRP conjugate (Millipore catalog no. 12-

349), Goat Anti-Rabbit IgG, HRP-linked (Cell Signaling, catalog no. 70745), and Goat

Anti-Mouse, IgG1 Gamma 1 Heavy Chain Specific (SouthernBiotech, catalog no. 1070-

05, specifically used for detecting immunoprecipitated protein ubiquitination) were all diluted at 1:20,000.

In Vitro Ubiquitination — HEK293T cells were transfected overnight with plasmid expressing HA-hIFITM3. Protein collected from all wells of one 6-well plate was

immunoprecipitated using anti-HA affinity gel and was washed extensively.

Immunoprecipitated protein on affinity gel was resuspended in PBS. 10% of the

retrieved protein was used in each reaction containing 500 µM ubiquitin or ubiquitin

mutants (Boston Biochem, catalogue nos. U-100H, UM-K630, or UM-K480), 0.5 µM

UbcH5b E2 ligase (Boston Biochem, catalogue no. E2-622), 100 nM UBE1 E1 ligase

(Boston Biochem, catalogue no. E305), and 1x Ubiquitin Conjugation Reaction Buffer

containing ATP (Boston Biochem, catalogue no SK-10) in the presence or absence of 100

ng recombinant human NEDD4 (Sigma, catalogue no. SRP0226). Reactions were

allowed to proceed at 37 °C for 1 h and were stopped by boiling for 5 min. The reactions

were then diluted 1:100 in ice cold 1% Brij buffer, and IFITM3 was re-

immunoprecipitated at 4 °C using newly added anti-HA affinity gel prior to Western blot

analysis.

Fluorescence Microscopy — Cells were fixed for 10 min with 3.7% paraformaldehyde,

permeabilized with 0.1% Triton X-100 in PBS for 10 min, and blocked for 10 min with

2% FBS in PBS. Primary antibodies, anti-mIFITM3 (Fragilis, Abcam, catalogue number

ab15592) (1:500), anti-NEDD4 (1:500), and anti-LAMP1 (Santa Cruz Biotechnology, catalogue number sc-19992), and Alexa Fluor-labeled anti-mouse and anti-rabbit secondary antibodies (Life Technologies, 1:1000) or anti-rat DyLight 550-labeled secondary antibody (Abcam, catalogue number ab96888, 1:1000) were diluted in 0.1%

Triton X-100 in PBS. Cells were treated with antibodies sequentially for 20 min at room temperature and washed five times with 0.1% Triton X-100 in PBS after each antibody treatment. Glass slides were mounted in ProLong Gold antifade reagent containing DAPI

(Life Technologies). Images were captured using a Fluoview FV10i confocal microscope

(Olympus).

Infections and Flow Cytometry — Influenza viruses A/Puerto Rico/8/1934 (H1N1, PR8), a PR8 reassortant virus possessing the hemagglutinin and neuraminidase genes from

A/Aichi/2/1968 (H3N2, X-31), A/Victoria/361/2011 (H3N2), and B/Texas/06/2011 were

propagated in 10-day embryonated chicken eggs (purchased as day 0 eggs from Charles

River Laboratories) for 48 h at 37 °C as described previously [219]. PR8 and X-31 were

provided to us by Drs. Bruno Moltedo and Thomas Moran (Mount Sinai School of

Medicine) and the 2011 virus isolates were obtained from BEI Resources sponsored by

the NIH/NIAID. SeV expressing green fluorescent protein (SeV-GFP)[102] was

generated by Dr. Dominique Garcin (University de Geneve) and provided to us by Dr.

Mark Peeples (Nationwide Children’s Hospital Research Institute). SeV-GFP was

propagated in 10-day embryonated chicken eggs for 40 h at 37 °C as described previously [241]. VSV G-pseudotyped retrovirus expressing green fluorescent protein

was generated by transfection of the viral vector pLenti-CMV-GFP-puro (Addgene plasmid 17448, deposited by Dr. Eric Campeau)[242] and packaging plasmids (provided by Dr. Li Wu, The Ohio State University) along with plasmid expressing VSV G into

HEK293T cells. Direct inhibition of GFP production by IFITM3 was not expected since this is driven by the CMV immediate early promoter and bypasses retrovirus-specific expression machinery[243]. Media was changed 18 h post-transfection, and media containing virus was then harvested 48 h post-transfection. Virus-containing media was centrifuged at 1200 x g for 5 min, filtered with 0.45 µm filters, frozen, stored at -80 °C, and used for infection at a dose that provided approximately 70% infection of WT MEFs.

MEFs were infected with IAV PR8, X-31, and H3N2 2011 strains, SeV and IBV at a multiplicity of infection of 5.0 or 10.0. MEFs were infected for 24 h, except in the case of VSV G-pseudotyped retrovirus infections, which were analyzed after 48 h of infection.

A549 cells were infected with IAV strain PR8 at a multiplicity of infection of 2.5 for 6 h.

Infected cells were washed with PBS and harvested in 0.25% trypsin EDTA. Cells were fixed in 3.7% paraformaldehyde for 10 min and permeabilized with 0.1% Triton X-100

for 10 min. IAV infected cells were stained with anti-influenza nucleoprotein (Abcam,

catalog no. ab20343, 1:333) directly conjugated to Alexa Fluor 647 using a 100 µg

antibody labeling kit (Life Technologies). IBV infected cells were stained with anti-IBV

nucleoprotein (Thermo Scientific catalogue no. MA1-80712, 1:1000) followed by anti- mouse secondary antibodies conjugated directly to Alexa Fluor 488 (Life Technologies).

Measurement of SeV and VSV G-pseudotyped retrovirus infection rates was done by detecting virus-encoded green fluorescent protein. All antibodies were diluted in 0.1%

Triton X-100 in PBS, and cells were stained for 20 min. Cells were washed three times with 0.1% Triton X-100 in PBS after each antibody treatment. PBS was used for final resuspension of cells for flow cytometric analysis using a FACSCanto II flow cytometer

(BD Biosciences). Results were analyzed using FlowJo software.

3.4: Results

Figure 10: IFITM3 is ubiquitinated by NEDD4 A) Alignment of IFITM3 N-terminal amino acids from various species. Bold and underlined text highlights the conserved PPxY motif. B) Mouse embryonic fibroblasts (MEFs) were stimulated overnight with IFNα (160 units/mL) overnight to ensure production of IFITM3, and imaged by fluorescent confocal microscopy with staining for endogenous IFITM3, NEDD4, LAMP1, and nuclei (DAPI). Images were taken with a 60x objective and 2.5x zoom. Pseudocolored merged images in different staining combinations are shown. C-E), HEK293T cells were co-transfected with plasmids expressing IFITM3 and epitope tagged ubiquitin ligases, NEDD4, CLB-B, SMURF1 and SMURF2, as indicated. Cell lysates were immunoprecipited with anti-myc resin, and examined by Western blotting with anti-myc and anti-ubiquitin (Ub) antibodies. Western blots of cell lysates with anti-HA (C) or anti-FLAG (D,E) antibodies were performed to confirm expression of the ubiquitin ligases. Anti-GAPDH Western blotting was performed to confirm comparable loading.

NEDD4 co-localizes with IFITM3 at lysosomes. To explore the possibility that

NEDD4 ubiquitinates IFITM3, we first examined whether IFITM3 and NEDD4 are in

proximity to one another within cells. We stimulated mouse embryonic fibroblasts

(MEFs) with IFN-α to induce abundant expression of IFITM3. By performing

immunofluorescence microscopy imaging of endogenous IFITM3 and NEDD4, we

detected co-localization of these two proteins (Figure 10B). Further co-localization of

these proteins with endogenous LAMP1, a lysosomal marker, indicates that NEDD4 and

IFITM3 may interact at lysosomes (Figure 10B).

NEDD4 overexpression increases IFITM3 ubiquitination. We next examined the

effect of overexpressing HA-tagged human NEDD4 (HA-NEDD4) on IFITM3 ubiquitination. We observed a significant increase in myc-hIFITM3 ubiquitination when

HA-NEDD4 was expressed as compared to the transfection control (Figure 10C). On the

contrary, no increase in IFITM3 ubiquitination was seen upon overexpression of HA-

tagged human CBL-B, another E3 ubiquitin ligase that has been reported to interact with

NEDD4 [244], and that is associated with regulation of immune responses [245] (Figure

10C). Additionally, we examined the effect of overexpressing FLAG-tagged NEDD4 on

IFITM3 ubiquitination in comparison to FLAG- tagged SMURF1 and SMURF2, both of

which are members of the NEDD4-family of ubiquitin ligases [225]. FLAG-NEDD4

caused an increase in IFITM3 ubiquitination while FLAG- SMURF1 and FLAG-

SMURF2 were unable to robustly modify IFITM3 (Figure 10D). These results

demonstrate that NEDD4 possesses a degree of specificity for IFITM3 that is lacking for

CBL-B and the NEDD4-family members, SMURF1 and SMURF2.

The IFITM3 PPxY motif is required for ubiquitination by NEDD4. As previously mentioned, NEDD4-family ubiquitin ligases possess two to four characteristic WW domains that interact with proline-rich motifs, including PPxY motifs, on substrate proteins [246]. NEDD4 has four WW domains, and IFITM3 contains a highly conserved

PPxY motif within its N-terminus (Figure 10A). To test whether these domains are required for IFITM3 ubiquitination, we generated an IFITM3 mutant in which each residue of the PPxY motif (17-PPNY-20 in IFITM3) was mutated to alanine (designated

17-20A), and utilized a FLAG-NEDD4 mutant in which its four WW domains were deleted (designated ∆WW). Upon co-overexpression of WT FLAG-NEDD4 with myc- hIFITM3, ubiquitination of IFITM3 was increased as expected, while the ∆WW mutant was unable to increase IFITM3 ubiquitination (Figure 10E). In fact, FLAG-NEDD4-

∆WW partially decreased steady state IFITM3 ubiquitination, perhaps indicating a dominant negative effect (Figure 10E). Moreover, the 17-20A mutant of IFITM3 showed less ubiquitination than WT IFITM3, and was unaffected by overexpression of NEDD4

(Figure 10E).

The IFITM3 PPxY motif shares its tyrosine with an overlapping YxxΦ motif known to be involved in the trafficking of IFITM3 from the plasma membrane to endosomes [167, 191, 247]. Thus, in order to be certain that the results we observed for the 17-20A mutant of IFITM3 was not because of interference with the YxxΦ motif, we 75

tested additional PPxY mutants in which the two prolines were mutated to alanine (myc- hIFITM3-P17,18A) or in which the tyrosine was mutated to alanine (myc-hIFITM3-

Y20A). Upon co-overexpression of FLAG-NEDD4, the ubiquitination of both of these mutants was only minimally increased as compared to the robust increase in ubiquitination of WT IFITM3 (Figure 11A). Interestingly, a truncated variant of IFITM3 missing its first 21 amino acids, including the PPxY motif, is prevalent in certain human populations. As previously discussed, this variant is associated with severe influenza virus infections [156, 159, 217] and more rapid progression of HIV-related disease [161].

A myc-hIFITM3 construct lacking these first 21 amino acids (∆1-21) was, as expected, largely unaffected in terms of ubiquitination by overexpression of FLAG-NEDD4 (Figure

11B), identifying a potentially important difference between the truncated and full-length

IFITM3 variants.

Next, since NEDD4 has been shown to physically interact with the PPxY motifs of its substrate proteins [225], we examined whether or not NEDD4 and IFITM3 co- immunoprecipitate with one another. We found that myc-hIFITM3 and FLAG-NEDD4 indeed co-immunoprecipitate (Figure 11C), suggesting a physical interaction.

Importantly, this interaction was greatly diminished between FLAG-NEDD4 and the

P17,18A mutant of IFITM3 (Figure 11C). In sum, these results indicate that the IFITM3

PPxY motif is required for a strong interaction with, and ubiquitination by, NEDD4.

Figure 11: The IFITM3 PPxY motif is required for ubiquitination by NEDD4 A-C) HEK293T cells were co-transfected with plasmids expressing myc-hIFITM3 or FLAG-NEDD4 as indicated. A-B) Cell lysates were immunoprecipited with anti-myc resin, and examined by Western blotting with anti-myc and anti-ubiquitin (Ub). Western blotting of cell lysate with anti-FLAG antibodies was performed to confirm expression of NEDD4. Western blotting with anti-GAPDH antibodies was performed to confirm comparable loading. C) Cell lysates were immunoprecipitated with anti-myc or anti- FLAG resin, and co-immunoprecipitation was examined by Western blotting with both anti-myc and anti-FLAG antibodies for each immunoprecipitate. Western blots of cell lysates with anti-myc and anti-FLAG antibodies were performed to confirm expression of IFITM3 and NEDD4, respectively. Anti-GAPDH Western blotting was performed to confirm comparable loading.

NEDD4 catalytic activity is required for IFITM3 ubiquitination. To determine whether a non-enzymatic activity of NEDD4 might be mediating its effect on IFITM3 ubiquitination, we tested a catalytically-inactive NEDD4 point mutant. We found that this mutant was unable to increase IFITM3 ubiquitination, establishing that catalytic activity of NEDD4 is indeed required for its ability to increase IFITM3 ubiquitination (Figure

12). Since murine (m)IFITM3 also possesses a PPxY motif (Figure 10A), we tested the

ability of NEDD4 to affect mIFITM3 modification. Like hIFITM3, we observed an

increase in mIFITM3 ubiquitination when HA-NEDD4 was co-overexpressed, and

observed no effect of the catalytic mutant (Figure 12), suggesting a possible evolutionary

conservation of NEDD4 modification of IFITM3 in mice and humans. These data further

implicate NEDD4 as an E3 ubiquitin ligase capable of enzymatically modifying mouse

and human IFITM3.

Figure 12: NEDD4 catalytic activity is required for IFITM3 ubiquitination HEK293T cells were transfected with the indicated mouse or human IFITM3 constructs and were co-transfected with plasmids expressing HA-NEDD4 or a catalytically inactive HA-NEDD4-C867A mutant. IFITM3 was immunoprecipitated with anti-myc resin and subjected to anti-myc and anti-ubiquitin (Ub) Western blotting. Cell lysates were probed with anti-NEDD4 antibodies to confirm expression of NEDD4 constructs. Anti-GAPDH staining served as a loading control.

NEDD4 can modify IFITM3 in vitro. While NEDD4 overexpression experiments

suggest that NEDD4 directly ubiquitinates IFITM3 (Fig 10C-E, 11A-B, 12), this effect could be indirect. We therefore tested the ability of purified NEDD4 to ubiquitinate

immunoprecipitated IFITM3 in vitro in order to confirm that NEDD4 can directly modify

IFITM3. HA-hIFITM3 was incubated with purified NEDD4, enzymatic cofactors, and

ubiquitin. We then re-immunoprecipitated IFITM3 and subjected it to anti-ubiquitin

western blotting. Our results show that NEDD4 is capable of robustly ubiquitinating

IFITM3 in vitro (Figure 13). Additionally, we employed ubiquitin mutants that could

only be added via lysine 48 (K48) or lysine 63 (K63) linkages in order to examine

whether NEDD4 preferentially utilizes one of these polyubiquitination linkages for

modifying IFITM3. While both K48 and K63 linkages could be added to IFITM3 by

NEDD4, we observed a preference for the K48 linkage in long polyubiquitin chains,

which is traditionally associated with protein degradation (Figure 13). These results are

consistent with our previous findings using linkage-specific anti-ubiquitin antibodies, which demonstrated that while both K48 and K63 ubiquitin linkages could be detected on

IFITM3, K48 linkages are more prevalent [179]. These data are also consistent with our previous results indicating that ubiquitination of IFITM3 promotes its turnover [179].

Figure 13: NEDD4 ubiquitinates IFITM3 in vitro HA-hIFITM3 was added to reactions containing NEDD4-compatible E1 and E2 ubiquitin ligases, ubiquitin (WT or mutants in which only K48 or K63 were not mutated), and reaction buffer containing ATP in the presence or absence of NEDD4. The reaction was allowed to proceed for 1 h at 37ºC, and IFITM3 was re-immunoprecipitated and subjected to Western blotting with anti-ubiquitin (Ub) and anti-HA antibodies. IgG H.C. indicates detection of the heavy chain of the immunoglobulin used for immunoprecipitation.

NEDD4 knockout decreases IFITM3 ubiquitination and increases resistance to viral infection. In order to examine the effects of NEDD4 on endogenous IFITM3, we examined NEDD4 WT and knockout (KO) mouse embryonic fibroblasts (MEFs) [236].

We also utilized KO MEFs reconstituted with NEDD4 via retroviral transduction.

Remarkably, Western blotting of lysates from NEDD4 KO cells showed an increase in steady state IFITM3 levels as compared to WT cells, while NEDD4 reconstitution decreased IFITM3 to WT levels (Figure 14A). To examine the requirement for NEDD4 in ubiquitinating IFITM3, we immunoprecipitated IFITM3 from large quantities of lysate from both WT and KO cells, expecting that the immunoprecipitation reagents would be 80

saturated, thus providing us with comparable amounts of IFITM3 for examination of ubiquitination. Indeed, IFITM3 from NEDD4 KO cells was ubiquitinated much less than

IFITM3 from WT cells (Figure 14B). These results demonstrate that NEDD4 is required for proper steady state ubiquitination of IFITM3, and that the absence of NEDD4 results in cellular accumulation of unmodified IFITM3.

Given the increase in baseline IFITM3 levels, we predicted that NEDD4 KO cells would be more resistant to influenza virus infection. We observed that NEDD4 KO

MEFs were in fact significantly less susceptible to infections with influenza A virus

(IAV) subtypes H1N1 and H3N2 (PR8 and X-31 strains, respectively) compared to WT control cells (Figure 14C). The decreased susceptibility of KO cells was returned to WT levels of infection upon NEDD4 reconstitution (Figure 14C). We also verified that the enhanced resistance of NEDD4 KO cells to influenza virus infection included resistance to recently circulating strains. NEDD4 KO cells were significantly less susceptible than

WT cells to infection by both influenza B virus (IBV) and IAV H3N2 strains isolated in

2011 (Figure 14D). We also examined retrovirus pseudotyped with the vesicular stomatitis virus (VSV) G protein, which is reported to be inhibited by IFITM3 [153, 164,

167, 181, 191]. As expected, the percent of NEDD4 KO cells infected with VSV G- pseudotyped virus was significantly less than WT cells (Figure 14D). Sendai virus

(SeV), a parainfluenza virus that primarily fuses at the cell surface [248] and is thus only minimally affected by IFITM3 [220], was also tested. Unlike IAV, IBV, and VSV G- pseudotyped retrovirus, SeV was not appreciably affected by NEDD4 KO (Figure 14D).

Thus, the pattern of virus restriction we observed is consistent with protection of NEDD4

KO cells by IFITM3.

To confirm that the increased resistance of NEDD4 KO cells to influenza virus infection was due to increased levels of basal IFITM3, we knocked down IFITM3 in

NEDD4 WT and KO cells for 24 hours prior to infection. Knockdown was verified through western blotting of cell lysates prepared at the time of infection (Figure 15A).

Importantly, knockdown of IFITM3 in both NEDD4 WT and KO MEFs resulted in an increase in influenza virus susceptibility, and largely eliminated the resistance of NEDD4

KO cells to infection (Figure 15B). Overall, these experiments demonstrate that NEDD4 promotes cellular susceptibility to influenza virus infection by decreasing levels of

IFITM3.

Figure 14: NEDD4 knockout decreases IFITM3 ubiquitination and protects cells from virus infection A) Cell lysates from NEDD4 WT and KO MEFs were subjected to anti-IFITM3, anti- NEDD4, and anti-actin immunoblotting to evaluate NEDD4 levels in each cell line and the effect of NEDD4 on endogenous IFITM3 levels. Retroviral reconstitution of the indicated cells with an empty retrovirus control or retrovirus expressing NEDD4 is denoted by + Vector and + NEDD4, respectively. B) 2 mg of protein from NEDD4 WT and KO cell lysates were immunoprecipitated for endogenous IFITM3 and examined by Western blotting with anti-ubiquitin and anti-IFITM3 antibodies. C) The indicated cell lines were infected for 24 h with influenza A virus (IAV) PR8 and X-31 strains at an MOI of 5. Cells were then fixed and stained with anti-influenza NP to measure the percentage of cells infected using flow cytometry. D) NEDD4 WT and KO MEFs were infected with influenza virus 2011 isolates (IAV or influenza B virus (IBV)) or Sendai virus (SeV) at an MOI of 5 for 24 h, or were infected with VSV G-pseudotyped retrovirus (VSV G-pseudo) for 48 h. Cells were then fixed and stained with anti- influenza NP in the case of influenza virus infections or were examined for GFP positivity in the case of VSV G-pseudo or SeV to measure the percentage of cells infected using flow cytometry. C,D) Non-infected samples were used as a baseline for gating of infected cells. Results shown are representative of at least three independent experiments, each performed with triplicate samples. The average percent infection of WT NEDD4 cells was set to 1 for the calculation of relative percent infection. Error bars represent standard deviation of triplicate samples. * Indicates a p-value less than 0.001 calculated by Student’s t-test in comparison to values for NEDD4 WT cells.

Figure 15: NEDD4 regulates cellular susceptibility to influenza virus infection by controlling IFITM3 levels A,B) NEDD4 WT and KO MEFs were transfected for 24 h with control siRNA (siCont) or siRNA targeting IFITM3 (siIFITM3). A) Cells were collected just prior to infection for confirmation of IFITM3 knockdown by anti-IFITM3 Western blotting, with anti-actin blotting serving as a loading control. B) Following siRNA treatment, cells were infected with influenza A virus strain PR8 at an MOI of 10 for 24 h. Cells were then fixed and stained with anti-influenza NP to measure the percentage of cells infected using flow cytometry. Results shown are representative of three independent experiments, each with samples run in triplicate. Error bars represent standard deviation. * Indicates p-value less than 0.001 calculated with Student’s t-test for comparison of samples denoted by horizontal lines.

NEDD4 knockdown in human lung cells increases IFITM3 levels and resistance to influenza virus infection. To extend our results to more relevant human lung cells, we utilized the A549 human alveolar epithelial cell line to study the role of NEDD4 in the regulation of steady state IFITM3 levels. Knockdown of NEDD4 with siRNA in A549 cells led to a significant increase in endogenous IFITM3 compared to non-targeting control siRNA (Figure 16A). As expected, NEDD4 knockdown led to a significantly greater resistance to IAV infection (Figure 16B and C). Importantly, we found that the relationship between NEDD4 knockdown and increased IFITM3 levels was preserved in two additional human lung cell lines (Figure 16D). Taken together with experiments

presented in Figures 12, 14, and 15, these data confirm an evolutionary conservation between mouse and man in the regulation of cellular IFITM3 levels by NEDD4. This work also identifies NEDD4 as a novel target in human cells for improving resistance to influenza virus infection independently of IFNs or adaptive immunity.

IFITM3 is turned over by the lysosomal degradation pathway. The degradative pathway involved in the turnover of steady state IFITM3 has not been previously investigated. Since NEDD4 is known to associate with the endosomal and lysosomal system and to target several of its substrates for lysosomal degradation [225], our results identifying NEDD4 as the primary ubiquitin ligase for IFITM3 would suggest that perhaps IFITM3 is degraded in lysosomes. To test this hypothesis, we utilized chloroquine and bafilomycin, which inhibit endosomal and lysosomal acidification and thus the activation of pH-dependent lysosomal proteases. We observed that treatment of

A549 lung cells with these two inhibitors caused an accumulation of IFITM3 (Figure

16E). Similarly, treatment with leupeptin, an inhibitor of specific lysosomal proteases resulted in a similar increase in IFITM3 levels (Figure 16E). This is in contrast to the treatment of cells with the proteasomal inhibitor MG132, which consistently caused a modest decrease in IFITM3 levels, perhaps due to up-regulation of lysosomal degradation pathways when proteasome activity is inhibited. Overall, these experiments demonstrate that, consistent with the co-localization of IFITM3 and NEDD4 at lysosomes (Figure

10B), and the ubiquitination of IFITM3 by NEDD4 (Figures 10C-E, 11-13, 14B),

IFITM3 is turned over by the lysosomal degradation pathway.

Figure 16: NEDD4 knockdown in human lung cells increases IFITM3 levels and resistance to influenza virus infection A-C) A549 cells were transfected for 48 h with control siRNA (siControl) or siRNA targeting human NEDD4 (siNEDD4). A) Cells were collected just prior to infection for confirmation of NEDD4 knockdown by anti-NEDD4 Western blotting, with anti-actin blotting serving as a loading control. Anti-IFITM3 blotting demonstrates an increase in IFITM3 upon NEDD4 knockdown. B,C) Following siRNA treatment, cells were infected with influenza virus strain PR8 at an MOI of 2.5 for 6 h. Cells were then fixed and stained with anti-influenza NP to measure the percentage of cells infected using flow cytometry. Results shown are representative of three independent experiments, with samples run in triplicate. Error bars represent standard deviation of triplicate samples. * Indicates a p-value less than 0.0001 calculated with Student’s t-test. D) NCI-H358 and NCI-H2009 cells were transfected for 48 h with siControl or siNEDD4. Cell lysates were subjected to immunoblotting with anti-NEDD4 to confirm NEDD4 knockdown, anti- IFITM3 to demonstrate increase in endogenous IFITM3 upon NEDD4 knockdown, and anti-GAPDH as a loading control. E) A549 cells were treated with equal volumes Dimethyl Sulfoxide (DMSO) as a control, MG132 (10 µM), Chloroquine (10 µM), Bafilomycin A1 (1 µM), or Leupeptin (100 µM) for 24 h. Cells were also treated with IFNα (100 units/mL) for comparison. Cell lysates were subjected to anti-IFITM3 immunoblotting to evaluate endogenous IFITM3 levels with each treatment. Western blotting with anti-GAPDH served as a loading control.

3.5: Discussion

Our previous work established that ubiquitination promotes the turnover of

IFITM3 [179]. Thus, identification of the IFITM3 ubiquitin ligase would provide a

potential target for improving IFITM3-mediated resistance to virus infections. In the

course of studying regulation of IFITM3 endocytosis by phosphorylation, we made the

serendipitous discovery that the amino acid Y20 within IFITM3 is involved in regulating

IFITM3 ubiquitination [190], which led us to identify the involvement of the IFITM3

PPxY motif in its ubiquitination by NEDD4 (Figures 10E, 11). NEDD4 knockdown or

knockout in human or mouse cells, respectively, resulted in substantially greater levels of

steady-state IFITM3 (Figures 14A, 15A, 16A,D). This accumulation of unmodified

IFITM3 is consistent with the observed decrease in IFITM3 ubiquitination in NEDD4

KO cells (Figure 14B).

An additional intriguing aspect of our finding that IFITM3 steady state levels are

regulated by NEDD4 is the previously described role of the IFN effector ISG15 in

inhibiting NEDD4 [233, 234]. ISG15 is a ubiquitin-like protein that specifically binds to

NEDD4, blocking its productive interaction with Ubiquitin-E2 ligase complexes [233,

234]. The importance of this pathway was highlighted by two independent studies

demonstrating that ISG15 blocks NEDD4-mediated monoubiquitination of the VP40

matrix protein of Ebola virus, thereby inhibiting the budding of Ebola virus-like particles

[233, 234]. Importantly, several studies have implicated ISG15 as a critical antiviral effector against influenza A and B viruses [249-251]. Two studies have demonstrated

conjugation of ISG15 onto the influenza A virus NS1 protein by the E3 ligase HERC5,

and found that ISGylation of influenza A virus NS1 antagonizes virus replication [251,

252]. Interestingly, influenza B virus NS1 specifically blocks human ISG15 conjugation by preventing ISG15 interaction with the ISG15 activating enzyme UbE1L, effectively counteracting its antiviral effect [250, 253-255]. We posit that high levels of IFITM3 attained after IFN stimulation result from both IFITM3 gene induction, as well as increased IFITM3 protein stability as a result of ISG15 inhibition of NEDD4. We are currently investigating this exciting potential synergistic link between ISG15 and

IFITM3.

Our work demonstrates that NEDD4 is required for proper basal ubiquitination of

IFITM3 (Figure 14B). However, our results would also suggest that additional ubiquitin

ligases are also able to modify IFITM3, particularly when IFITM3 is present at high

levels. This is supported by detection of partial ubiquitination of our various IFITM3-

PPxY mutants (Figures 10E, 11A,B) and by detection of modest IFITM3 ubiquitination in NEDD4 KO cells (Figure 14B). The identities of secondary ubiquitin ligases for

IFITM3 are still unknown. Of particular interest are the ubiquitin ligases capable of modifying the truncated ∆1-21 splice variant of human IFITM3, which was not significantly ubiquitinated by NEDD4 (Figure 11B). Identifying the ubiquitin ligases that modify this disease-associated variant may implicate crucial differences in the stability and degradative pathways potentially underlying the defect possessed by this protein.

Nonetheless, our data clearly implicate NEDD4 as the primary E3 ubiquitin ligase for

IFITM3, and demonstrate that NEDD4 is essential for maintaining low steady state

IFITM3 levels.

This current work is in contrast to a prior study that concluded the IFITM3 PPxY motif was not involved in regulating the levels or antiviral activity of overexpressed

IFITM3 [167]. However, this previous work did not directly assess ubiquitination of

IFITM3 upon mutation of the PPxY motif. Additionally, our experiences studying

IFITM3 ubiquitination here and in our prior work suggest that when examining overexpressed IFITM3 constructs, ubiquitination has only subtle effects on total protein

levels detected by Western blotting despite significant effects on the IFITM3 half-life

[179, 247]. Thus, overexpression likely masked any effects of mutating the PPxY motif

on the parameters previously tested [167].

Our study has uncovered a novel mechanism by which NEDD4 indirectly

promotes cellular entry of influenza virus by decreasing IFITM3 levels (Figures 14-16).

Although this work is the first of its kind to identify NEDD4 as a negative regulator of

IFITM3 levels, NEDD4 is well described to be necessary for the replication of several

important RNA viruses. For example, NEDD4 interacts with proline-rich motifs in the

viral late budding domains of Ebola virus [256], rabies virus [257], and HIV [258].

Mono-ubiquitination of these domains promotes efficient budding and viral egress

necessary for productive viral spread. However, it remains to be determined whether

inhibition of NEDD4 will serve as an effective in vivo antiviral strategy, particularly

since NEDD4 is a developmentally essential molecule as demonstrated by the embryonic

lethality of NEDD4 KO mice [259]. Likewise, NEDD4 has been implicated in regulation

of insulin-like growth factor signaling [260], T-cell-mediated immunity [261, 262], and

tumor suppression [263]. On the other hand, neuron- and skeletal muscle-specific

NEDD4 KO mice are viable [264, 265], and NEDD4 is naturally inhibited by ISG15

during virus infections [233, 234], perhaps suggesting that short-term inhibition of

NEDD4 can occur without adverse effects. Additional experimentation will be needed to

answer these vital questions, and this will be aided by the development of selective

NEDD4 inhibitors, which is an area of active investigation [266, 267]. Overall, our study

identifies inhibition of NEDD4 as a novel strategy for preventing infection by influenza

virus and other IFITM3-sensitive viruses through the increased accumulation of the

antiviral restriction factor IFITM3.

3.6: Contributions and Acknowledgements

Experiments were conceived by Nicholas Chesarino and Jacob Yount. Jacob

Yount performed the in vitro ubiquitination experiment (Figure 13), and Nicholas

Chesarino performed the remaining experiments. Temet McMichael provided technical

assistance, propagated viruses used in this study, and provided editorial assistance. The

manuscript was written by Nicholas Chesarino and Jacob Yount.

This study was supported by funding from the NIH/NIAID (grants R00AI095348

and R56AI114826 to Dr. Jacob Yount) and from The Ohio State University Public Health

Preparedness for Infectious Disease program. Nicholas Chesarino was supported by The

Ohio State University Systems and Integrative Biology Training Program (NIH/NIGMS

grant T32GM068412). Temet McMichael was supported by the American Society for

Microbiology Robert D. Watkins Graduate Research Fellowship. Thank you to Dr.

Matthew Pratt of the University of Southern California for providing NEDD4 WT and

KO MEFs, and reconstituted control cell lines.

This work was published in PLOS Pathogens on August 11, 2015. Citation of this manuscript is as follows:

Chapter 4: IFITM3 restriction of virus infection requires an amphipathic helix

4.1: Abstract

Interferon-induced transmembrane protein 3 (IFITM3) is a cellular protein that blocks virus fusion with cellular membranes. IFITM3 has been suggested to alter membrane curvature, though its exact mechanism of action is unclear. Prediction of

IFITM3 secondary structure using several bioinformatics tools allowed us to identify a highly conserved, short amphipathic helix within an enigmatic hydrophobic region that was previously thought to be a transmembrane domain, and more recently hypothesized to be an intramembrane domain. Consistent with the known ability of amphipathic helices to alter membrane properties, we show that this helix and its amphipathicity are

required for inhibition of influenza virus, Zika virus, vesicular stomatitis virus, Ebola

virus, and human immunodeficiency virus infections by IFITM3. The homologous

amphipathic helix within IFITM1 is also required for its inhibition of infection,

suggesting that IFITMs possess a conserved mechanism of antiviral action. Further, we

specifically show that the amphipathic helix is required for blockade of influenza virus

hemagglutinin-mediated membrane fusion by IFITM3. Overall, our results provide the

first evidence that IFITMs utilize an amphipathicity-based mechanism for inhibiting virus

fusion. This work should aid future biophysical studies of membrane changes mediated

by IFITMs, and may inspire novel amphipathic antiviral interventions.

4.2: Introduction

Interferon-induced transmembrane protein 3 (IFITM3) restricts cellular infection

by numerous pathogenic viruses, including Ebola virus and Zika virus [153, 155, 162,

164-166, 168, 170-172, 214, 243, 268-273], and is particularly well characterized as a

restriction factor active against influenza virus infection in vivo [153-155, 166, 176].

Knockout of IFITM3 in mice results in increased severity of influenza virus infections

[156, 215]. Likewise, a polymorphism in the human IFITM3 gene that is thought to result in IFITM3 mislocalization has been associated with severe cases of influenza [156,

159, 216, 217]. This human IFITM3 polymorphism has also been linked to more rapid

HIV-associated disease progression in comparison to infected patients with the more common IFITM3 alleles [161]. Though the importance of IFITM3 in antiviral defense is well documented in vitro and in vivo, its mechanism of action is not clearly defined.

The examination of IFITM3 susceptibility of numerous viruses has revealed that

IFITM3 is generally able to inhibit infections by viruses that enter cells through endocytosis [176, 218]. Indeed, IFITM3 primarily localizes to endosomes and lysosomes

[174, 179], and IFITM3 mutants that alternatively localize to the plasma membrane possess a diminished ability to inhibit influenza virus infection [167, 178, 190, 191].

Taken together, these findings indicate that IFITM3 acts within the endolysosomal pathway. Studies of influenza virus entry and fusion demonstrated that while viruses are endocytosed with standard kinetics in cells expressing IFITM3, viruses are unable to fuse with the endosome membrane, and are ultimately eliminated, likely by degradation in

endolysosomes [174, 175]. Membrane labeling using fluorescent lipids has demonstrated

that IFITM3 and other IFITMs decrease membrane fluidity, suggesting that IFITM3

counteracts viral manipulations of cellular membranes that occur during the fusion

process [169, 182]. The precise amino acid sequence elements of IFITM3 required for

this function are not known.

A previous report observed that stable cell lines overexpressing IFITM3 show

increased accumulation of intracellular cholesterol [181]. Given the known ability of

cholesterol to decrease membrane fluidity, this observation provided a compelling

hypothesis for how IFITM3 might affect membranes and block virus fusion. The

accumulation of cholesterol was explained by the interaction of IFITM3’s hydrophobic

domain (HD) 2 with vesicle-associated membrane protein-associate protein A (VAPA),

thereby blocking the basal interaction between VAPA and cholesterol trafficking

machinery [181]. While the interaction between VAPA and IFITM3 has been reproducibly observed [274], a role for VAPA or cholesterol in the antiviral mechanism of action of IFITM3 has been disputed by at least four laboratories [175, 182, 268, 270].

Additionally, we previously reported a requirement for S-palmitoylation of cysteine

residues within HD1 for antiviral activity of IFITM3 and other IFITMs, suggesting that

HD1 has a prominent, but largely undefined, role in IFITM3 activity [154, 177, 197,

198]. Thus, inconsistencies with the current mechanistic model for IFITM3 function

prompted us to evaluate specific amino acid determinants of its antiviral activity. Here,

we identify an amphipathic helix within HD1, adjacent to S-palmitoylation sites, and

confirm that this region and its amphipathicity are required for maximal inhibition of

viruses by IFITM3.

4.3: Materials and Methods

Structural Prediction Programs — Amino acid sequences and alignments were gathered

using Uniprot (www.uniprot.org). Structural prediction for human IFITM3 was performed using the Iterative Threading Assembly Refinement (I-TASSER) bioinformatics method without additional restraints or templates [275, 276]

(www.zhanglab.ccmb.med.umich.edu/I-TASSER). Additional analyses were performed using NetSurfP version 1.1 [277](www.cbs.dtu.dk/services/NetSurfP), PredictProtein

[278, 279](www.predictprotein.org), and POLYVIEW-2D [280](polyview.cchmc.org), each with default settings. De novo peptide structure prediction was performed using

PEP-FOLD 3.1 software [281, 282] (bioserv.rpbs.univ-paris-diderot.fr/services/PEP-

FOLD).

Cell Culture, Transfections, Transductions, and Plasmids — HEK293T and HeLa cells

(purchased from ATCC) were cultured in DMEM supplemented with 4.5 g/liter D- glucose, L-glutamine, 110 mg/liter sodium pyruvate (Thermo Fisher Scientific), and 10% fetal bovine serum (Thermo Fisher Scientific) at 37 °C and 5% CO2 in a humidified

incubator. CHME ‘4x4’ cells were cultured similarly and were generated by transducing

CHME cells with lentiviral constructs encoding CD4 and CXCR4 as done previously

[283]. 24 h prior to transfection, cells were plated to obtain 90% confluency for western

blotting and 50% for microscopy. Transfections were performed using Lipojet

(Signagen), using 400 ng of plasmid and 2 µL Lipojet per well of 12-well plates, and

1000 ng and 4 µL per well of 6-well plates. IFITM1 and IFITM3 constructs were expressed from the pCMV-HA or pCMV-myc vectors (Clontech) and were described

previously [154, 190, 229]. IFITM3 and IFITM1 mutants were made using the

QuikChange Multi site-directed mutagenesis kit (Stratagene). The Influenza A virus

hemagglutinin protein from the A/Puerto Rico/8/34 strain coding DNA sequence was

generated by RT-PCR from infected cells and inserted into the pCAGGS vector using

NheI and BglII enzyme sites. IFITM3 and mutants were subcloned into the pQCXIP

retroviral vector using BamHI and EcoRI sites for stable transduction of cell lines as

previously described [167, 168, 284].

Western Blotting and Antibodies — For Western blotting, cells were lysed with 1% Brij

buffer (0.1 mM triethanolamine, 150 mM NaCl, 1% BrijO10 (Sigma, catalog no. P6136),

pH 7.4) containing cOmplete EDTA-free protease inhibitor cocktail (Sigma, catalog no.

11873580001). 10 µg per lysate were resolved on 4-20% Mini-PROTEAN TGX Precast

Protein Gels (BIO-RAD, catalog no. 4561096). Western blotting was performed with

anti-myc (Developmental Studies Hybridoma Bank at the University of Iowa, deposited

by Dr. J. Michael Bishop, catalog no. 9E 10), anti-HA (Clontech, catalog no. 631207), or

anti-GAPDH (Invitrogen, catalog no. 398600) antibodies. All primary antibodies were

used at a 1:1000 dilution in Tris-buffered saline solution with 1% Tween 20 (TBST).

Secondary antibodies, Goat Anti-Mouse IgG, HRP conjugate (Millipore catalog no. 12- 96

349) and Goat Anti-Rabbit IgG, HRP-linked (Cell Signaling, catalog no. 70745) were

both diluted at 1:10,000 in TBST.

Fluorescence Microscopy — Cells grown on slides were transfected for 24 h prior to

being washed twice with PBS, and fixed for 10 min with 3.7% paraformaldehyde. After

fixation, cells were permeabilized with 0.1% Triton X-100 in PBS (PBST) for 10 min,

and blocked for another 10 min with 2% FBS in PBS. Primary antibodies, anti-IFITM3

(Proteintech Group, catalog no. 11714-1-AP), anti-CD63 (Developmental Studies

Hybridoma Bank at the University of Iowa, deposited by Drs. J. T. August and J. E. K.

Hildreth, catalog no. H5C6), and anti-IAV HA (Sino Biological Inc., catalog no. 11684-

RP01) were used at a 1:500 dilution in PBST, and incubated for 30 min at 4 °C. Slides were washed five times with PBST prior to adding Alexa Fluor-conjugated anti-mouse and anti-rabbit secondary antibodies (Thermo Fisher Scientific) at 1:1000 dilutions in

PBST for 30 min at 4 °C. Slides were again washed five times with PBST, rinsed with

PBS, and mounted in ProLong Gold antifade mountant with DAPI (Thermo Fisher

Scientific, catalog no. P36931) for 24 h at room temperature. Images were captured using a Fluoview FV10i confocal microscope (Olympus).

Viruses and Infections — Influenza viruses A/Puerto Rico/8/1934 (H1N1, PR8), a PR8 reassortant virus possessing the hemagglutinin and neuraminidase genes from

A/Aichi/2/1968 (H3N2, X-31), and B/Texas/06/2011 (IBV) were propagated in 10-day embryonated chicken eggs (purchased as day 0 eggs from Charles River Laboratories) for

48 h at 37 °C as described previously. Ebola virus Zaire strain glycoprotein (GP)-

pseudotyped retrovirus expressing green fluorescent protein (EBOV) was generated using

the pLenti-CMV-GFP-puro retroviral vector along with transfection of packaging

plasmids and GP-expressing plasmid (BEI Resources) into HEK293T cells. Stocks of

HIV-1 were similarly produced via CaPO4 transfection of HEK293T cells with pNL4-3.

Zika virus strain HD7878 was propagated in Vero cells. Transiently-transfected

HEK293T cells were infected with PR8 (MOI 2.5, 6 h), IBV, (MOI 5.0, 24 h), X-31

(MOI 5.0, 6 h), or VSV (MOI 1.0, 8 h). In the case of EBOV infection, cells were for 48

h to allow for integration and the expression of GFP. IFITM3 stably expressing

HEK293T cell lines were infected with PR8 at an MOI of 0.2 for 18 h, or with Zika virus

at an MOI of 1.0 for 72 hours. IFITM3 stably expressing CHME ‘4x4’ cell lines were

infected with 10 ng p24 equivalents of HIV-1 for 48 h.

Flow Cytometry— Infected cells were washed with PBS and harvested in 0.25% trypsin

EDTA (Thermo Fisher Scientific). Cells were fixed in 3.7% paraformaldehyde for 10 min and permeabilized with 0.1% Triton X-100 for 10 min. For all infections, cells were stained with anti-IFITM3, anti-myc, or anti-HA, and the appropriate Alexa Fluor conjugated anti-rabbit secondary (Thermo Fisher Scientific, 1:1000), to allow for segregation of transfected cells. IAV-infected cells were stained with anti-influenza nucleoprotein (Abcam, catalog no. ab20343, 1:333) directly conjugated to Alexa Fluor

647 using a 100 µg antibody labeling kit (Life Technologies). IBV-infected cells were stained with anti-IBV nucleoprotein (Thermo Scientific catalogue no. MA1-80712,

1:1000) followed by anti-mouse secondary antibodies conjugated directly to Alexa Fluor

488 (Life Technologies). Measurement of VSV and EBOV-pseudotyped retrovirus

infection rates was done by detecting virus-encoded green fluorescent protein. Zika virus and HIV-1 infections were detected using anti-Zika virus envelope protein (Kerafast catalog no. EVU302) or anti-Gag (Beckman Coulter item no. 6604667) antibodies, respectively. All antibodies were diluted in 0.1% Triton X-100 in PBS, and cells were stained for 30 min at 4 °C. Cells were washed three times with 0.1% Triton X-100 in PBS

after each antibody treatment. PBS was used for final resuspension of cells for flow

cytometric analysis using a FACSCanto II flow cytometer (BD Biosciences). Results were analyzed using FlowJo software. For cholesterol staining, transfected cells were treated similarly followed by blockig with 5% BSA in PBS and staining with 1 µg/mL

Nile Red (Sigma, catalog no. N3013) for 2 h at 4 °C. After staining with Nile Red, cells

were resuspended in 5% BSA in PBS and analyzed immediately by flow cytometry.

Circular Dichroism — Circular Dichroism was performed as previously reported [285].

Peptide consisting of the IFITM3 59-69 sequence (VWSLFNTLFM) was synthesized

with >97% purity by Thermo Fisher Scientific, and solubilized in buffer containing 25

mM Sodium Phosphate, pH 7.4, with 25 mM SDS added. 300 µL of a 50 µM peptide

solution was added to a 0.1 cm path length quartz cuvette. Spectra were collected

between wavelengths 190-260 nm on a Jasco J-815 CD Spectropolarimeter using a 0.5

nm step resolution, 100 nm/min scan time, response time of 0.5 s, and a bandwidth of 1

nm. Readings were measured in triplicate. 99

Cell-Cell Fusion Assay — HeLa cells grown on slides were transfected overnight with the indicated combinations of IFITM3 mutants and IAV HA. Following transfection, cells were treated with a dilute trypsin treatment (1 µg/mL in PBS) for 5 min at room temperature to proteolytically-activate HA. Cells were then incubated with freshly prepared DMEM containing 25 mM MES at pH 5.0 for 2 min to trigger pH-dependent

HA-mediated cell-cell fusion. Cells were then rinsed with PBS and incubated with standard tissue culture overnight 37 °C. Cells were fixed for 10 min with 3.7% paraformaldehyde, mounted in ProLong Gold antifade mountant with DAPI, and imaged by bright field and epifluorescence microscopy using an EVOS FL Cell Imaging System

(Thermo Fisher Scientific).

Palmitoylation Assay — Palmitoylation assays on IFITM3 were performed as previously described [154, 198, 228, 232, 286, 287]. HEK293T cells transfected for myc-tagged

IFITM3 constructs, metabolically-labeled with 50 µM 17-Octadecynoic Acid (alk-16,

Cayman Chemical, catalog no. 90270) for 2 h. Cell were lysed in 1% Brij buffer, and lysates were subjected to anti-myc immunoprecipitation using EZview Red Anti-C-Myc

Affinity Gel (Sigma, catalog no. E6654). IFITM3 was reacted with azido-rhodamine using copper(I)-catalyzed alkyne-azide cycloaddition to fluorescently tag alk-16-labeld protein. The final product was resolved on a 4-20% Mini-PROTEAN TGX Precast

Protein Gel (BIO-RAD) and imaged by fluorescence gel scanning on a Typhoon 9410

(GE Healthcare). Comparable loading was verified using anti-myc western blotting.

100

4.4: Results

The second hydrophobic domain of IFITM3 is not essential for antiviral activity.

Since IFITM3 is able to block virus membrane fusion [169, 175], we posited that its

membrane-interacting regions likely play a role in its antiviral activity. IFITMs from

diverse species possess two characteristic Hydrophobic Domains (HDs) that are thought to mediate interactions with membranes [147]. Alignment of these domains from mouse

and human IFITMs 1-3, which each possess antiviral activity when overexpressed in

human cells [153], demonstrates a strong conservation of HD1, with 17 of 21 identical

residues between these six IFITMs, while only 4 of 21 residues of HD2 are conserved

(Figure 17A). Interestingly, while HD2 is characterized as a transmembrane domain

[187, 288], HD1 does not traverse the membrane, and its membrane association pattern

remains unknown [179, 187, 229, 288]. We observed that a truncation of IFITM3

eliminating residues 88-133 (IFITM3-∆88-133), which removed HD2, maintained a significant ability to inhibit influenza virus infection when transiently expressed in

HEK293T cells (Figure 17B), despite decreased expression levels as compared to WT

IFITM3 (Figure 17C). This is in contrast to a triple cysteine-to-alanine IFITM3 mutant

(C71A, C72A, and C105A) lacking S-palmitoylation (IFITM3-∆Palm), which lost the

majority of its antiviral activity in this assay as previously reported [154, 179, 228, 289]

(Figure 17B and C). Our results indicate that HD2, previously reported to be required for

interaction of IFITM3 with VAPA, and disruption of cholesterol homeostasis [181], is

not essential for inhibition of influenza virus infection by IFITM3, though amino acids

101

88-133 likely contribute to antiviral activity by enhancing protein stability (Figure 17C).

As a control, we stained cells with the cholesterol and lipid stain Nile Red, and found that

cells transiently transfected and expressing IFITM3 showed a slight increase in Nile Red

fluorescence as compared to control cells (Figure 17D and E). As expected, this increase

was not seen in cells expressing IFITM3-∆88-133 (Figure 17D and E). Overall, these results suggest that IFITM3 antiviral activity can be decoupled from the robust cholesterol accumulation previously reported for IFITM3-overexpressing stable cell lines

[181], indicating the existence of an additional mode of antiviral action.

102

Figure 17: Hydrophobic domain 2 of IFITM3 is not essential for antiviral activity A) Amino acid alignments of the first and second hydrophobic domains of mouse and human IFITMs 1-3. Pamitoylated cysteines are highlighted in red. Conserved amino acids are marked with asterisks. B,C) HEK293T cells were transfected overnight with empty vector, full-length myc-tagged human IFITM3 (WT), IFITM3 truncated atfter amino acid 87 (∆88-133), or palmitoylation-deficient IFITM3 with cysteines 71, 72, and 105 mutated to alanine (∆Palm). For B, Cells were infected with H1N1 influenza A virus strain PR8 at an MOI of 2.5 for 6 h, and then fixed, permeabilized, and stained with anti- IFITM3 to identify transfected cells and with anti-influenza NP to quantify the

Continued

103

Figure 17 Continued percentage of infected cells by flow cytometry. Results shown are representative of more than three independent experiments each performed in triplicate. Error bars represent standard deviation. * Indicates a p-value less than 0.0001 calculated by Student’s t-test. For C, Cell lysates were subjected to Western blotting to determine expression of IFITM3 (anti-myc) with anti-GAPDH serving as a control for comparable loading. D,E) HEK293T cells were transfected overnight with the indicated IFITM3 constructs or vector control. Cells were fixed, permeabilized, and stained for IFITM3 (anti-myc) and cholesterol (Nile Red). D) Example histograms showing fluorescence intensity of transfected cells, as measured by flow cytometry. E) The averaged mean fluorescence intensity of transfected cells from triplicate samples. Error bars represent standard deviation. Results shown in D and E are representative of 3 similar experiments. * Indicates a p-value less than 0.01 calculated by Student’s t-test.

Secondary structure prediction for IFITM3 reveals a putative amphipathic helix.

We performed a structural prediction for human IFITM3 using the Iterative Threading

Assembly Refinement (I-TASSER) bioinformatics method [275, 276]. The prediction of secondary structure revealed regions of alpha helical structure with high confidence, roughly corresponding to HD1 and HD2 (Figure 18A). Indeed, as expected for a transmembrane domain, HD2 was predicted to exist as an uninterrupted alpha helix

(Figure 18A). On the contrary, two distinct alpha helices separated by unstructured residues were predicted for HD1 (Figure 18A). Similar results were obtained using additional secondary structure prediction programs such as NetSurfP [277] (Figure 19A),

PredictProtein [278, 279] (Figure 19B), and POLYVIEW-2D [280] (Figure 19C).

Adapting this helical secondary structure to the currently preferred intramembrane model of HD1 in which these alpha helices would form a hairpin loop structure (Figure 18B)

[187, 290] was not favorable. In this model, cysteines 71 and 72 would be located within the lipid bilayer, potentially inaccessible to the cytoplasmic activity of the enzymes mediating their S-palmitoylation, which is critical for antiviral activity [154, 289]. As 104

such, our structural prediction prompted us to consider HD1 as potentially being made up

of distinct structured regions rather than as a single domain.

Peptide structural prediction for the first predicted helix within HD1 (amino acids

59-68, sequence VWSLFTNLFM) using the PEP-FOLD prediction program [281, 282]

again suggested an alpha helical structure for this peptide, and indicated that this helix is

amphipathic, possessing both a hydrophobic and a primarily hydrophilic face (Figure 18C

and D). Examination of a synthetic peptide corresponding to this region using circular dichroism spectroscopy revealed that this peptide is capable of forming alpha helical

secondary structure (Figure 18E). Importantly, an empirical determination of IFITM3

secondary structure by NMR spectroscopy was recently published that largely confirms

these structural predictions, though the exact boundary of this short alpha helix was not

defined because examination of IFITM3 in this study began at amino acid position 60

[290]. Further, the data from this study was adapted to the prevailing intramembrane

model for IFITM3, and the amphipathic nature of this helix was not noted [290].

Nonetheless, this published work, along with our circular dichroism results (Figure 18E),

allow us to confidently conclude that the IFITM3 region spanning residues 59-68 is capable of forming alpha helical secondary structure.

Since amphipathic alpha helices wedge into only one leaflet of the membrane

bilayer and alter membrane properties [291-294], and because of the close proximity of

this helix to critical S-palmitoylated cysteines that may assist in the targeting or

anchoring of the helix to membranes [154, 289] (Figure 18B), we hypothesized that the

helix and its amphipathicity contribute to the ability of IFITM3 to inhibit viral infection.

105

Figure 18: Secondary structure prediction reveals an amphipathic helix within IFITM3 A) Results of secondary structure prediction for the human IFITM3 amino acid sequence by the Iterative Threading Assembly Refinement (I-TASSER). Predicted helical regions are shaded in different colors. The amphipathic helix of interest in this study is shaded orange. B) Membrane topology models for IFITM3. Colored regions correspond to predicted helices from A. S-palmitoylation of cysteines 71 and 72 is represented by red lines. C,D) Different views of the helix predicted for amino acids 59-68 (VWSLFTNLFM) produced using the PEP-FOLD prediction program displaying both a hydrophobic (C) and primarily hydrophilic (D) face. E) Circular dichroism spectra collected for a synthetic peptide corresponding to IFITM3 amino acids 59-68. F,G) HEK293T cells were transfected overnight with empty vector, full-length human IFITM3 (WT), IFITM3 lacking the helix at residues 59-68 (∆59-68), or palmitoylation-deficient IFITM3 with cysteines 71, 72, and 105 mutated to alanine (∆Palm). For F, cells were Continued

106

Figure 18 Continued infected with H1N1 influenza A virus strain PR8 at an MOI of 2.5 for 6 h, and then were fixed, permeabilized, and stained with anti-IFITM3 to identify transfected cells and with anti-influenza NP to quantify the percentage of infected cells by flow cytometry. Results shown are representative of more than three independent experiments each performed in triplicate. Error bars represent standard deviation. * Indicates a p-value less than 0.0001 calculated by Student’s t-test. For G, cell lysates were subjected to Western blotting to confirm similar expression of IFITM3 (anti-myc) with anti-GAPDH serving as a control showing comparable loading.

107

Figure 19: Structural prediction programs suggest helical domains in IFITM3 Secondary structure predictions for the human IFITM3 amino acid sequence using the A) NetSurfP, B) PredictProtein and C) POLYVIEW-2D online programs. *Denotes the predicted amphipathic helix region of interest examined in this study.

The IFITM3 amphipathic helix is required for anti-influenza A virus activity. As an initial test of the hypothesis that amphipathicity of IFITM3 drives its antiviral activity, we generated an IFITM3 construct lacking the putative amphipathic helix (IFITM3-∆59-68).

In infection assays, IFITM3-∆59-68 was impaired in its ability to inhibit H1N1 influenza

A virus infection similarly to IFITM3-∆Palm (Figure 18F and G). Depiction of residues

59-68 on a helical wheel diagram further confirmed, and allowed visualization of, the amphipathic nature of this sequence (Figure 20A). Since our results suggested that this 108

amphipathic helix within IFITM3 is required for antiviral activity, we next generated a

series of mutant constructs to alter or eliminate its amphipathicity. We chose to utilize

alanine substitutions for the residues of the hydrophilic face of the helix (S61, N64, and

T65) (Figure 20A) because alanine would provide weak hydrophobicity in these positions

and should minimally impact secondary structure. Single alanine substitution mutants

maintained most of their antiviral activity, confirming that IFITM3 structure was not

generally altered (Figure 20B). Double alanine mutants at these positions were decreased

in their ability to inhibit H1N1 influenza A virus infection to different degrees depending

on the locations of the respective residues within the helix and in relation to one another

(Figure 20A and B). Indeed, an IFITM3-S61A,N64A mutant lost a significant portion of

its activity, suggesting that these two positions together are particularly important for antiviral activity (Figure 20B). A triple-alanine mutant, which fully eliminates all hydrophilic residues, and thus completely eliminates amphipathicity, also lost the majority of its antiviral activity against influenza A virus (Figure 20B). These initial

results are consistent with the existence of a functionally critical amphipathic helix within

IFITM3, with residues S61, N64, and T65 comprising the hydrophilic face of the helix.

109

Figure 20: IFITM3 amphipathicity is required for its antiviral function A) Visualization of IFITM3 residues 59-68 (VWSLFTNLFM) on a helical wheel projection plot created using HELIQUEST software. The series of mutants generated to test the amphipathicity of this helix is also shown. Hydrophobic residues are displayed as gray or yellow, while hydrophilic residues are displayed as pink or purple. Arrows represent the magnitude and orientation of the mean hydrophobic moment value calculated by HELIQUEST software. Numerical values of the mean hydrophobic moments are shown in Supplemental Table 1. B-D) HEK293T cells were transfected overnight with empty vector or indicated myc-IFITM3 mutants. For B, cells were infected with IAV PR8 at an MOI of 2.5 for 6 h, and then fixed, permeabilized, and Continued

110

Figure 20 Continued stained with anti-IFITM3 to identify transfected cells, and with anti-influenza NP to quantify the percentage of infected cells by flow cytometry. Average results from at least three experiments, each performed in triplicate, are shown. Error bars represent standard deviation of the mean. * Indicates a p-value less than 0.0001 and # indicates a p-value less than 0.01 in comparison to WT IFITM3 calculated by Student’s t-test. For C, cell lysates were subjected to Western blotting to determine expression of IFITM3 (anti-myc) with anti-GAPDH serving as a control showing comparable loading. For D, cells were infected with influenza B virus strain B/Texas/06/2011 (IBV) at an MOI of 5 for 24 h, Influenza A virus strain X-31 at an MOI of 5 for 6 h, vesicular stomatitis virus expressing GFP (VSV) at an MOI of 1.0 for 8 h, or Ebola virus Zaire strain glycoprotein- pseudotyped lentivirus expressing GFP (EBOV) for 48 h. Cells were then fixed, permeabilized and stained with anti-myc to identify transfected cells. For IBV and H3N2 infections, cells were also stained with specific anti-influenza virus nucleoprotein antibodies to quantify the percent infections by flow cytometry. Alternatively, GFP positive cells were quantified by flow cytometry for VSV and EBOV infections. Infection percentages were normalized relative to a value of 100% set for vector control infections. Results shown are averages of triplicate samples from a representative of at least three independent experiments and error bars represent standard deviation. E) Flow cytometry histograms of anti-IFITM3 staining for stable HEK293T and CHME ‘4x4’ cell lines transduced with empty lentivirus (vector control), or lentivirus expressing WT IFITM3, IFITM3-∆59-68, or IFITM3-S61A,N64A,T65A. F) HEK293T cell lines as in E were infected with H1N1 influenza A virus strain PR8 (H1N1) at an MOI of 0.2 for 18 h or with Zika virus strain HD7878 (ZIKAV) at an MOI of 1.0 for 72 h. CHME ‘4x4’ cell lines as in E were infected with 10 ng p24 equivalents of HIV-1 prepared from molecular clone NL4.3 for 48 h. Infected cells were then stained respectively with anti- influenza virus nucleoprotein, anti-Zika virus envelope protein, or anti-HIV-1 p24 antibodies for quantification of percent infection by flow cytometry. Infection percentages were normalized relative to a value of 100% for vector control infections. Results shown are averages of four independent experiments for each virus and error bars represent standard deviation of the mean.

Next, we generated single mutants of the hydrophobic residues L62, F63, and F67 to the hydrophilic amino acid glutamine. Mutation of L62, which is located at the border of the hydrophilic and hydrophobic face of the helix (Figure 20A), to glutamine did not decrease antiviral activity (Figure 20B). Calculation of the mean hydrophobic moment

[295] for this construct using HELIQUEST software [296], while not particularly

111

informative for our previous alanine substitutions due to the weak hydrophobicity of alanine, indicated that amphipathicity was actually increased for the L62Q mutation in comparison to WT IFITM3, potentially explaining its strong antiviral activity (Figure

20A, arrows, and Table 3). In contrast, individual mutations of the most hydrophobic residues of the helix, F63 and F67, to glutamine significantly decreased antiviral activity

(Figure 20B), while also decreasing the mean hydrophobic moment of the helix (Figure

20A, arrows, and Table 3), further suggesting that disruption of amphipathicity decreases antiviral activity. Finally, we generated an IFITM3 variant in which a phenylalanine was inserted at position 65 (IFITM3-insert65F), which would significantly disrupt both the hydrophilic and hydrophobic faces of the helix (Figure 20A). This single amino acid insertion mutant of IFITM3 corresponded to the lowest mean hydrophobic moment calculation among all of our mutant constructs (Figure 20A, arrows, and Table 3), and lost the majority of its ability to inhibit influenza virus infection (Figure 20B).

Importantly, protein levels of the mutants examined in our experiments were comparable to WT IFITM3, and minor differences in expression did not correlate with loss of activity

(Figure 20C).

112

Table 3: Mean hydrophobic moment calculations for WT IFITM3 amino acids 59- 68 and the amphipathic helix mutants tested in this study Calculated using HELIQUEST software. Values possess arbitrary units and can range from 0 to 3.26 with higher values indicating stronger amphipathicity. Note that alanine substitutions did not significantly affect the mean hydrophobic moments since alanine is still significantly less hydrophobic than the amino acids making up the hydrophobic face of the helix.

As controls, we also examined the cellular distribution of several IFITM3 amphipathicity mutants that lost a significant portion of their antiviral activity, and found that they maintained colocalization with the lysosomal markers CD63 and LAMP1-GFP similarly to WT IFITM3, indicating that their loss of function could not be explained by mislocalization (Figure 21). We also confirmed that internalized influenza virus colocalizes with IFITM3 mutants despite their defect in inhibiting infection, further indicating that localization does not account for their decreased activity (Figure 22).

Additionally, we measured S-palmitoylation of a similar set of IFITM3 constructs, and found that they were S-palmitoylated comparably to WT IFITM3 (Figure 23), demonstrating that they maintain interaction with palmitoyltransferases and that their loss

113

of antiviral activity cannot be explained by a defect in S-palmitoylation. Overall, our results support a model whereby IFITM3 amphipathicity at amino acids 59-68 is required for robust inhibition of H1N1 influenza A virus infection.

114

Figure 21: Cellular localization of IFITM3 with lysosomes is unaffected by mutations in its amphipathic helix HeLa cells were transfected for 24 h with the indicated myc-tagged IFITM3 constructs and A,B) were fixed, permeabilized, and stained with anti-myc and anti-CD63, or C) were co-transfected with LAMP1-GFP and were fixed, permeabilized, and stained with anti-myc alone. Example fluorescent confocal microscopy images used for colocalization analysis are shown in A. Manders Overlap Coefficient for IFITM3 overlap with CD63 (B) or with LAMP1-GFP (C) was calculated using the JACoP plugin for ImageJ. For each bar in B and C, overlap from at least 10 random cell fields from two experiments, each showing at least 10 cells were quantified and averaged. Error bars represent standard deviation of the mean.

115

Figure 22: Mutations of the amphipathic helix do not prevent IFITM3 colocalization with internalized influenza virus HeLa cells were transfected with the indicated IFITM3 and mutant constructs overnight. Cells were infected with influenza A virus (IAV) strain PR8 in ice cold PBS at an MOI of 50 for 30 minutes at 4 oC to allow virus to attach to cells. Unattached virus was then removed with 3 ice-cold PBS washes. Warm media at 37 oC was then added to cells to allow virus internalization for 45 minutes. Cells were then fixed and stained with DAPI to visualize nuclei, anti-IFITM3 to identify transfected cells, and with anti-influenza virus NP to visualize virus. Representative bright field and confocal fluorescent images are shown. Zoomed regions show internalized virus staining overlapping with WT and mutant IFITM3 constructs.

116

Figure 23: S-palmitoylation of IFITM3 occurs independently of its amphipathic helix HEK293T cells were transfected overnight with the indicated myc-tagged IFITM3 WT or mutant constructs. Cells were metabolically labeled with 50 µM alk-16 (palmitoylation chemical reporter) for 2 h. IFITM3 was purified from cell lysates by anti-myc immunoprecipitation, reacted with azido-rhodamine using copper(I)-catalyzed alkyne- azide cycloaddition to fluorescently tag alk-16 labeled protein, and imaged by fluorescence gel scanning. Comparable loading was verified using anti-myc western blotting.

The IFITM3 amphipathic helix is required for inhibition of multiple virus infections. To test the generality of the requirement for IFITM3 amphipathicity in its inhibition of virus infection, we examined IFITM3-S61A,N64A,T65A, which lacks amphipathicity, for ability to inhibit additional strains of influenza virus and other viruses. Indeed, in comparison to WT IFITM3, the amphipathicity mutant exhibited significantly decreased inhibition of infection by recent H3N2 influenza A virus and influenza B virus isolates (Figure 20D). Likewise, this mutant lost much of its ability to inhibit vesicular stomatitis virus and Ebola virus pseudotyped retroviral particles, both of which are well established to be inhibited by IFITM3 [153, 164, 166, 167, 181, 191, 271].

117

We next sought to confirm that the amphipathic helix is required for virus

inhibition when IFITM3 is stably expressed in cells. We transduced HEK293T cells as well as CHME cells that were previously engineered to express the CD4 and CXCR4 receptors necessary for HIV-1 infection (termed CHME ‘4x4’ cells). We obtained cell

lines expressing similar levels of WT IFITM3 and two amphipathicity mutants for

comparison with vector control cells (Figure 20E). We first confirmed that the stable

HEK293T lines showed a similar restriction pattern of inhibition of H1N1 influenza A virus to that obtained using transient transfection methods (Figure 20F). Specifically, cells stably expressing WT IFITM3 were highly restrictive to influenza virus infection, and this restriction activity was lost upon deletion or mutation of the amphipathic helix

(Figure 20F). We further utilized these cell lines to examine infection by Zika virus strain HD7878. At three days post infection, we observed strong inhibition of Zika virus infection in cells expressing WT IFITM3, confirming a recent report of inhibition of Zika virus by IFITMs [163], while cells expressing IFITM3 amphipathicity mutants were infected at rates similar to vector control cells (Figure 20F). Using CHME ‘4x4’ cells, we examined HIV-1 infection rates, and again observed that restriction of HIV-1 by

IFITM3 is dependent upon an intact amphipathic helix (Figure 20F). Interestingly, the cell line expressing IFITM3-S61A,N64A,T65A was infected by HIV-1 at a significantly higher rate in all four of our experiments, suggesting a possible dominant-negative effect of this mutant in CHME cells, though we have not followed up on this observation.

Overall, these results demonstrate that amphipathicity of IFITM3 is broadly required for restriction of virus infections.

118

Virus restriction by IFITM1 requires amphipathicity. IFITMs 1 and 3 possess a similar ability to inhibit virus infections when overexpressed in vitro [153, 166, 177,

270], though IFITM1 does not appear to compensate for a loss of IFITM3 in vivo [156,

215], potentially owing to distinct cellular sorting signals within IFITM3 versus IFITM1

[167, 190, 191, 297]. Nonetheless, we noted that human IFITM1 possesses an amphipathic sequence of amino acids (residues 38-47) identical to the amphipathic helix in IFITM3. In infection experiments, we observed that the ability of human IFITM1 to restrict H1N1 influenza A virus was significantly decreased when its corresponding amphipathic helix was deleted, or when the hydrophilic residues of the helix (S40, N43,

T44) were mutated to alanine (Figure 24A and B). These results indicate a conserved mechanism of antiviral action requiring amphipathicity for IFITMs 1 and 3.

119

Figure 24: Antiviral activity of IFITM1 requires its amphipathic helix A,B) HEK293T cells were transfected overnight with empty vector or the indicated HA- tagged IFITM1 constructs. For A, cells were infected with H1N1 influenza A virus strain PR8 at an MOI of 2.5 for 6 h, and then fixed, permeabilized, and stained with anti-HA to identify transfected cells, and with anti-influenza NP to quantify the percentage of infected cells by flow cytometry. Results shown are the average of triplicate samples from an experiment representative of three similar experiments. Error bars represent standard deviation. * Indicates a p-value less than 0.0001 calculated by Student’s t-test. For B, cell lysates were subjected to Western blotting to determine expression of IFITM3 (anti-HA) with anti-GAPDH serving as a control showing comparable loading.

The IFITM3 amphipathic helix is required for inhibition of influenza HA-mediated membrane fusion. Since inhibition of influenza virus infection by IFITM3 is conferred by its ability to prevent virus hemagglutinin (HA)-mediated membrane fusion [169, 174,

175], we sought to test whether the IFITM3 amphipathic helix is required specifically for inhibition of fusion. When influenza virus HA is produced in cells, membrane fusion can be artificially triggered between cells by treatment with low pH media, mimicking conditions in the late endosome. Cell-cell fusion assays were previously employed to study inhibition of virus-mediated fusion by IFITM3 and other IFITMs [169]. Indeed,

IFITM3 expression was previously shown to partially block HA-mediated cell-cell 120

fusion, while IFITM1, which has a stronger localization at the plasma membrane, more

significantly reduced cell-cell fusion [169]. We utilized a similar system along with an

IFITM3 point mutant at tyrosine 20 (Y20A), which lacks a functional YxxΦ endocytosis

motif and accumulates at the plasma membrane [167, 178, 190, 191]. We incubated

HeLa cells transfected with an influenza HA-expressing construct with a short trypsin

treatment to proteolytically-activate HA at the cell surface. This was immediately

followed by an incubation of the cells with media at pH 5.0 to trigger pH-dependent fusion. After overnight incubation in standard media, we then fixed and analyzed the cells by microscopy. We observed that cells transfected with vector controls remained fully confluent and did not show multinucleated fused cells (Figure 25A and B).

However, we observed two expected effects of HA expression that occurred only after

the fusion-inducing treatments. First, we qualitatively observed increased cell death as

indicated by loss of cell confluence and the presence of rounded, loosely adherent cells

(Figure 25A). Second, we quantified multiple multinucleated fused cells present on

average per cell field (Figure 25B). As previously reported [169], these phenotypes were

partly prevented when IFITM3 was co-expressed with HA, though this effect was not

statistically significant for the number of cell fields that were quantified, and this trend

was not seen for IFITM3-∆59-68 (Figure 25A and B). Remarkably, the plasma

membrane-localized mutant, IFITM3-Y20A, was able to almost fully reverse the cell

death and cell fusion phenotypes, while IFITM3-Y20A-∆59-68, lacking the amphipathic

helix, lost this ability (Figure 25A and B). As controls, we confirmed that HA was

expressed in the presence of IFITM3 and mutants (Figure 26). HA was most prevalent in 121

intracellular compartments in permeabilized cells likely representing the ER and Golgi, and the Y20A mutants showed a more dispersed cellular distribution than WT IFITM3 as expected (Figure 26). These experiments demonstrate that IFITM3 is able to inhibit HA- mediated cell-cell fusion, particularly when localized at the plasma membrane. Consistent with inhibition of overall virus infection (Figure 18F, 20B, D, and F), the amphipathic helix is required for this fusion inhibition (Figure 25).

122

Figure 25: The amphipathic helix and its localization are important for IFITM3 inhibition of influenza virus hemagglutinin-mediated membrane fusion A,B) HeLa cells transfected with the indicated combinations of WT IFITM3 or IFITM3 mutants and Influenza A Virus (IAV) hemagglutinin (HA) were treated with a short, dilute trypsin wash to proteolytically activate HA at the cell surface, immediately followed by an incubation with media at pH 5.0 to trigger pH-dependent HA-mediated cell-cell fusion. After an overnight incubation in standard media, cells were fixed, stained with DAPI, and imaged by bright field and epifluorescence microscopy. A) Representative portions of cell fields are shown. DAPI staining is shown in magenta in the merged images. Yellow arrows point to example multinucleated cells. B) Average quantification of multinucleated cells in cell field images. For each condition more than 14 cell fields from two distinct experiments were quantified. *Indicates a p-value less than 0.0005 calculated by Student’s t-test. C) A theoretical model of the IFITM3 mechanism of antiviral action. In cells lacking IFITM3, HA (red rectangles) induces membrane fusion between the viral membrane and the host endosome membrane. In the presence of IFITM3 (shown as blue triangles to represent the wedging mechanism of the amphipathic helix), HA begins the process of virus fusion but local accumulation of IFITM3 counteracts virus-induced changes in membrane curvature, preventing formation of the fusion pore.

123

Figure 26: Control imaging for transfections presented in Figure 25 HeLa cells were transfected with the indicated IFITM3 constructs along with influenza A virus hemagglutinin (IAV HA), or with vector controls. Cells were fixed, permeabilized, and stained with anti-IFITM3 (red) and anti-HA (green), and analyzed by confocal microscopy. Merged images of IFITM3 and HA are shown.

4.5: Discussion

Through bioinformatics structural prediction for IFITM3, and functional assessment of IFITM3 mutants, we have identified a previously unappreciated amphipathic helix “hidden” within the larger HD1, which was previously hypothesized to exist as a transmembrane domain [153, 186, 298], and more recently as an intramembrane domain [179, 187, 290]. This helix is necessary for complete inhibition of influenza virus infection and for inhibition of influenza HA-mediated membrane fusion (Figure 18F, 20B,D,F). We have also reported that S-palmitoylation at C71, C72, and C105 promotes antiviral activity [154, 179, 228, 229], with C72 being the most essential site of modification for inhibition of influenza virus infection [178, 289]. The exact function of this post-translational modification has been challenging to define [197,

229], though given the close proximity of the S-palmitoylated residues to the amphipathic helix, we speculate that their function may be to target or anchor this helix to membranes

(Figure 18B, amphipathic model). Indeed, lipidated amino acids are commonly found 124

near amphipathic helices [299-302]. Further, this membrane-anchoring model is consistent with our demonstration that both amphipathicity and S-palmitoylation are required for full antiviral activity of IFITM3 (Figure 18,20).

Examination of previously performed evolutionary analysis and alignments of

IFITMs from multiple species confirms strong conservation of hydrophilic residues in positions corresponding to S61, N64, and T65 of human IFITM3, while, as expected, remaining positions within the amphipathic helix are predominantly occupied by hydrophobic amino acids [148, 303]. Such evolutionary conservation of the amphipathic helix further supports its importance in IFITM3 functionality. Indeed, deletion of the human IFITM1 amphipathic helix or loss of its amphipathicity both decreased the ability of IFITM1 to inhibit influenza virus infection, confirming a conserved antiviral function for this helix (Figure 24).

During the completion of our study, the first empirically determined secondary structure information was published for IFITM3 [290]. This report utilized site directed spin labeling involving systematic mutagenesis of each IFITM3 amino acid to cysteine, which they began at residue 60. Importantly, this work identified a short helix corresponding to the amphipathic helix that we identify here, confirming that this region indeed forms alpha helical structure [290]. This group further observed that this helix does not fully insert into detergent micelles [290]. While they concluded that their data are consistent with an intramembrane domain structure for HD1, these results are alternatively explained by our discovery that this region forms an amphipathic helix involved in IFITM3 antiviral activity (Figure 18-20). Since our analysis of the IFITM3

125

amphipathic helix was based on structural prediction, and the aforementioned empirical structure determination was based on limited mutagenesis, we acknowledge that we cannot be certain that the helix begins or ends at precisely positions 59-68, though the addition of one residue on either end of the helix would maintain its amphipathic nature.

Amino acid N69 is on the border of the amphipathic helix and was previously shown to be essential for antiviral activity, as an IFITM3-N69D mutant lost its ability to inhibit influenza virus infection [178]. We have also found that IFITM3-N69A and

N69Q mutants lose a significant portion of their antiviral activity (not shown). Since asparagines are commonly observed at the beginning or end of alpha helices due to their ability to hydrogen bond with the peptide backbone, we propose that an asparagine is specifically required at this position for properly terminating the amphipathic helix. This is further supported by the near-universal conservation of an asparagine in this position throughout evolution [148]. As noted in our structural predictions in Figures 17A, 18, and 19, and in the recent empirical secondary structure determination of IFITM3 [290], an additional alpha helix is formed only a few residues away from the amphipathic helix

(Figure 18A,B). Interestingly, this additional helix contains phenylalanine residues previously determined to be involved in IFITM3 dimerization [178, 228], suggesting that this helix represents the IFITM3-IFITM3 interaction interface. This nearby dimerization motif may serve to amplify membrane alterations mediated by the amphipathic helix.

IFITM3 has been shown to possess multiple antiviral mechanisms in the context of HIV infection. For instance, in addition to restriction of HIV fusion and entry [168,

243, 284], IFITM3 incorporates into viral particles and limits their infectious potential

126

[226, 227]. Further, IFITM3 has been suggested to directly interact with the HIV envelope glycoprotein to impair its processing and incorporation into virions [304].

Determining the role of S-palmitoylation and the amphipathic helix in these processes in the future should provide important information regarding the mechanisms underlying these additional restriction activities of IFITM3.

Amphipathic helices within proteins often detect, induce, or stabilize regions of membrane curvature by increasing the occupied area of only one layer of the membrane bilayer through a partial insertion, i.e., wedging, mechanism [291-293]. Interestingly, viruses, including influenza virus, also utilize amphipathic segments for alteration of membrane properties during budding and fusion [305-308]. It is possible that IFITM3 may be recruited to regions of membrane curvature generated by the virus fusion process, perhaps to counteract or stabilize virus-induced membrane alterations (Figure 25C).

Indeed, it has been reported that influenza virus HA-mediated membrane hemifusion is observed in the presence of IFITM3, but that complete fusion pores are not formed [175].

A model involving IFITM3 accumulation at sites of fusion would be consistent with the ability of IFITM3 to partially inhibit infection even when present at low levels, and with its increasing inhibition efficiency as its levels rise [153, 179, 309]. Such a model is also consistent with the saturability of IFITM3 activity at high virus MOIs [178, 182, 191].

Development of new systems for high resolution or live cell imaging of IFITM3 will be required to explore this local concentration hypothesis. Likewise, studies determining whether the amphipathic helix of IFITMs can also affect virus budding or scission in addition to virus entry will also be exciting future directions of this work. Overall, our

127

study provides the first evidence for an amphipathicity-based mechanism of antiviral

action for IFITM3, and provides a new model for influenza virus inhibition that may be

further scrutinized in the context of additional infections, and exploited in the future for

antiviral therapeutic or prevention approaches.

4.6: Contributions and Acknowledgements

Experiments were conceived by Nicholas Chesarino and Jacob Yount. Jacob

Yount performed the bioinformatics analyses and cell-cell fusion assays, Lizhi Zhang performed palmitoylation and IFITM1 experiments, Alex Compton and Olivier Schwartz created stable cell lines and performed Zika and HIV-1 infections, and Nicholas

Chesarino performed the remaining experiments. Nicholas Chesarino, Jacob Yount,

Temet McMichael, Lizhi Zhang, Victoria Soewarna, Rachel Doering, and Matthew Davis helped generate IFITM3 mutants. The manuscript was written by Nicholas Chesarino and Jacob Yount, with editorial assistance by Alex Compton and Olivier Schwartz.

This study was supported by funding from the NIH/NIAID (grants R00AI095348 and R56AI114826 to Dr. Jacob Yount) and from The Ohio State University Public Health

Preparedness for Infectious Disease program. Nicholas Chesarino was supported by The

Ohio State University Systems and Integrative Biology Training Program (NIH/NIGMS grant T32GM068412) and The Ohio State University Presidential Fellowship. Temet

McMichael was supported by the American Society for Microbiology Robert D. Watkins

Graduate Research Fellowship.

128

This work is being prepared as a manuscript for publication.

Chesarino NM, Compton AA, McMichael TM, Zhang L, Soewarna V, Doering R, Davis M, Schwartz O, and Yount JS (2016). IFITM3 restriction of virus infection requires an amphipathic helix. In preparation.

129

Chapter 5: Conclusions

5.1: IFITM3 phosphorylation: Insights and further questions

The work presented here focuses on understanding the cellular mechanisms regulating IFITM3 trafficking and antiviral activity. The depth and complexity of

IFITM3 regulation is truly astounding, and future endeavors to uncover these pathways will be fruitful. At the genesis of this dissertation, IFITM3 was simply known to be induced by interferon, localize to endolysosomes, and prevent viruses from infecting the cell. By studying IFITM3 phosphorylation (Chapter 2), we added key details to what we now know about IFITM3 trafficking. This study demonstrated, for the first time, that

IFITM3 naturally traffics to the plasma membrane, where it enters the endocytic pathway. Knowing that IFITM3 coats endocytic membranes at the very beginning of the pathway explains why we see IFITM3-coated endolysosomes with markers for both early and late endosomes. We showed that Fyn kinase regulates the trafficking of IFITM3 by sequestering it at the cell surface. This is a dynamic process, strongly suggesting active dephosphorylating occurs by an undiscovered phosphatase. As we know dephosphorylation must occur at the cell surface to initiate endocytosis of IFITM3, we can narrow down our future search of IFITM3 phosphatases by those that localize to the plasma membrane. This study also suggests that Fyn kinase, being a negative regulator of IFITM3 trafficking, might be a promising target for antiviral therapy. Indeed, several

130

FDA-approved therapies exist for the inhibition of Src-family kinases, including Fyn, for the treatment of chronic myelogenous leukemia. It would be interesting to see how these inhibitors affect IFITM3 trafficking and antiviral activity against influenza virus.

The Y20A mutant of IFITM3 has been used by ourselves and others to study the function of IFITM3 more clearly at the cell surface. Use of this mutant as a tool has gathered significant insight into how the wild-type variant behaves in relation to the rs12252 C/C disease variant. Most notably, Compton et al (2016) found that, in the case of HIV-1 infection, the rs12252 C/C variant is actually more protective against viral entry at the cell surface, at a functional trade-off of increasing susceptibility to flu [284]. This study and ours would then suggest that wild-type IFITM3 trafficking and primary localization may be capable of being controlled by the cell in order to place IFITM3 at the point of the cell where its antiviral activity will be most effective. For example, if some cellular process exists that upregulates Fyn-mediated phosphorylation while downregulating phosphatase activity, a higher percentage of IFITM3 will result at the cell surface, allowing greater protection against viruses like HIV-1. Therefore, it is imperative that we identify the phosphatase(s) modifying IFITM3, as controlling its activity may allow us to “toggle” IFITM3 localization based on the identity of the virus we want to protect against.

131

5.2: IFITM3 ubiquitination: Overexpression without infection

Previous studies of IFITM3 ubiquitination identified that ubiquitination is a mostly-degradative process for IFITM3. Therefore, there was strong motivation to

identify what the E3 ubiquitin ligase modifying IFITM3 was in order to control its

expression levels for therapeutic benefit. Studying IFITM3 phosphorylation led to the

serendipitous re-discovery of the PPxY endocytosis motif, which a previous study ruled

out as unimportant due to the type of overexpression experiment used. The implication

of the PPxY motif on IFITM3 ubiquitination allowed us to identify NEDD4 as the primary E3 ubiquitin ligase almost immediately (Chapter 3). One of the most important discoveries to come from this study was that the steady-state levels of IFITM3 were malleable without induction by IFN or infection. This is extremely critical for the future proposals of IFITM3-based therapies, because it demonstrates that levels can be increased without inducing the full-on interferon response.

The basal levels of IFITM3 seem to be kept in check by the ubiquitous-expressed

NEDD4. It would stand to reason, then, that the interferon response would not only increase transcription of the IFITM3 gene, but would also prevent its post-translational degradation by somehow inhibiting IFITM3 ubiquitination. Indeed, NEDD4 has been shown by three independent groups to be directly inhibited by the interferon-induced gene 15 (ISG15). Although the data is not presented here, we have performed pilot experiments that clearly demonstrate ISG15-mediated inhibition of NEDD4 prevents the ubiquitination of IFITM3, and increases steady-state levels of IFITM3 in cells. This is an exciting avenue of research that our lab is currently pursuing.

132

Finally, implication of NEDD4 in IFITM3 ubiquitination has inspired us with a

new strategy to upregulate IFITM3 in a novel mouse model of IFITM3 overexpression.

Previous attempts to overexpress IFITM3 in a genetic mouse model by the addition of ifitm3 gene copies have failed, likely due to the more ubiquitous expression of NEDD4 at basal levels (not published). On the other hand, NEDD4 knockout mice are embryonic lethal, preventing us from upregulating IFITM3 in his manner. Therefore, our strategy for creating an IFITM3 overexpression mouse model is to mutate the prolines of the mouse IFITM3 PPxY motif to alanines, thereby blocking the interaction between IFITM3 and NEDD4, preventing IFITM3 ubiquitination, and upregulating basal levels of

IFITM3. We envision this mouse model to be resistant to doses of influenza that would be lethal to their wild-type counterparts. Furthermore, we will be able to study the physiological consequences of long-term IFITM3 overexpression. In the long term, we hope to use this model to characterize the role IFITM3 may play in autoimmune, inflammatory, and other infectious diseases. Should we find that IFITM3 overexpression is safe for the organism, our mouse model can set the precedent for overexpressing

IFITM3 in poultry and swine livestock, both of which express a PPxY-containing

IFITM3, as a way to significantly dampen the circulation of influenza strains that would reach humans.

5.3: Understanding the broad antiviral activity of the IFITM3 amphipathic helix

One of the greatest mysteries in IFITM3 biology is the mechanism by which the

IFITM proteins confer their antiviral activity. We are the first to identify the 10 amino

133

acid amphipathic helix in the IFITM proteins as the main domain regulating antiviral activity (Chapter 4), significantly narrowing down the exact mechanism that the IFITMs act through. Although we know that IFITM3 is capable of altering membrane properties, and that amphipathic helices are well-characterized to alter membrane properties, we have yet to pinpoint what exactly is being targeted by IFITM3 at the virus:host interface.

Active investigation into the mechanism is ongoing in our lab.

134

Bibliography

1. Influenza (Seasonal) Fact Sheet 211. 2014 [cited 2016.

2. Shrestha, S.S., et al., Estimating the burden of 2009 pandemic influenza A (H1N1) in the United States (April 2009-April 2010). Clin Infect Dis, 2011. 52 Suppl 1: p. S75-82.

3. Swerdlow, D.L., L. Finelli, and C.B. Bridges, 2009 H1N1 influenza pandemic: field and epidemiologic investigations in the United States at the start of the first pandemic of the 21st century. Clin Infect Dis, 2011. 52 Suppl 1: p. S1-3.

4. Influenza at the human-animal interface – Summary and assessment, 20 July to 3 October 2016. 2016 [cited 2016.

5. Gao, H.N., et al., Clinical findings in 111 cases of influenza A (H7N9) virus infection. N Engl J Med, 2013. 368(24): p. 2277-85.

6. Lai, S., et al., Global epidemiology of avian influenza A H5N1 virus infection in humans, 1997-2015: a systematic review of individual case data. Lancet Infect Dis, 2016. 16(7): p. e108-18.

7. Molinari, N.A., et al., The annual impact of seasonal influenza in the US: measuring disease burden and costs. Vaccine, 2007. 25(27): p. 5086-96.

8. Mao, L., et al., Annual economic impacts of seasonal influenza on US counties: spatial heterogeneity and patterns. Int J Health Geogr, 2012. 11: p. 16.

9. Verikios, G., et al., Assessing Regional Risks From Pandemic Influenza: A Scenario Analysis. The World Economy, 2016. 39(8): p. 1225-1255.

10. Virus Taxonomy: 2015 Release. 2015.

11. Palese, P. and M.L. Shaw, Orthomyxoviridae: the viruses and their replication. Fields virology, 2007. 2: p. 1647-1689.

12. Bouvier, N.M. and P. Palese, The biology of influenza viruses. Vaccine, 2008. 26 Suppl 4: p. D49-53. 135

13. Matsuzaki, Y., et al., Clinical features of influenza C virus infection in children. Journal of Infectious Diseases, 2006. 193(9): p. 1229-1235.

14. Taubenberger, J.K. and D.M. Morens, Influenza: the once and future pandemic. Public health reports, 2010: p. 16-26.

15. Yuanji, G., et al., Isolation of influenza C virus from pigs and experimental infection of pigs with influenza C virus. Journal of General Virology, 1983. 64(1): p. 177-182.

16. Laver, W. and R. Valentine, Morphology of the isolated hemagglutinin and neuraminidase subunits of influenza virus. Virology, 1969. 38(1): p. 105-119.

17. Update: influenza activity--United States and worldwide, 1999-2000 season, and composition of the 2000-01 influenza vaccine, in MMWR. Morbidity and mortality weekly report. 2000, Centers for Disease Control and Prevention. p. 375.

18. Memorandums, M.-L., A revision of the system of nomenclature for influenza viruses: a WHO memorandum. Bulletin ofthe World Health Organization, 1980. 58(4): p. 585-591.

19. Gerber, P., C.G. Loosli, and D. Hamre, ANTIGENIC VARIANTS OF INFLUENZA A VIRUS (PR8 STRAIN) I. THEIR DEVELOPMENT DURING SERIAL PASSAGE IN THE LUNGS OF PARTIALLY IMMUNE MICE. The Journal of experimental medicine, 1955. 101(6): p. 627-638.

20. Taubenberger, J.K. and J.C. Kash, Influenza virus evolution, host adaptation, and pandemic formation. Cell Host Microbe, 2010. 7(6): p. 440-51.

21. Taubenberger, J.K. and D.M. Morens, The pathology of influenza virus infections. Annual review of pathology, 2008. 3: p. 499.

22. Obenauer, J.C., et al., Large-scale sequence analysis of avian influenza isolates. Science, 2006. 311(5767): p. 1576-1580.

23. Lakadamyali, M., M.J. Rust, and X. Zhuang, Endocytosis of influenza viruses. Microbes and infection, 2004. 6(10): p. 929-936.

24. Chakraborty, A. and N.R. Jana, Clathrin to Lipid Raft-Endocytosis via Controlled Surface Chemistry and Efficient Perinuclear Targeting of Nanoparticle. The journal of physical chemistry letters, 2015. 6(18): p. 3688-3697.

25. Gruenberg, J., The endocytic pathway: a mosaic of domains. Nature reviews Molecular cell biology, 2001. 2(10): p. 721-730.

136

26. Joshi, S.B., R.E. Dutch, and R.A. Lamb, A core trimer of the paramyxovirus fusion protein: parallels to influenza virus hemagglutinin and HIV-1 gp41. Virology, 1998. 248(1): p. 20-34.

27. Sieczkarski, S.B. and G.R. Whittaker, Dissecting virus entry via endocytosis. Journal of General Virology, 2002. 83(7): p. 1535-1545.

28. Pelkmans, L. and A. Helenius, Insider information: what viruses tell us about endocytosis. Current opinion in cell biology, 2003. 15(4): p. 414-422.

29. Conner, S.D. and S.L. Schmid, Regulated portals of entry into the cell. Nature, 2003. 422(6927): p. 37-44.

30. Chen, J., et al., Structure of the hemagglutinin precursor cleavage site, a determinant of influenza pathogenicity and the origin of the labile conformation. Cell, 1998. 95(3): p. 409-417.

31. Hayden, F.G., Update on influenza and rhinovirus infections, in Antiviral Chemotherapy 5. 1999, Springer. p. 55-67.

32. Monto, A.S., et al., Zanamivir in the prevention of influenza among healthy adults: a randomized controlled trial. Jama, 1999. 282(1): p. 31-35.

33. Wasilenko, J.L., et al., NP, PB1, and PB2 viral genes contribute to altered replication of H5N1 avian influenza viruses in chickens. Journal of virology, 2008. 82(9): p. 4544-4553.

34. Hale, B.G., et al., The multifunctional NS1 protein of influenza A viruses. Journal of General Virology, 2008. 89(10): p. 2359-2376.

35. Baigent, S.J. and J.W. McCauley, Glycosylation of haemagglutinin and stalk- length of neuraminidase combine to regulate the growth of avian influenza viruses in tissue culture. Virus research, 2001. 79(1): p. 177-185.

36. Wright, P.F. and R.G. Webster, Orthomyxoviruses. Fields virology, 2001. 1: p. 1533-1579.

37. Beveridge, W.I.B., Influenza: the last great plague. An unfinished story of discovery. 1977: Heinemann Educational Books Ltd., 48 Charles Street, London W1X 8AH.

38. Simonsen, L., The global impact of influenza on morbidity and mortality. Vaccine, 1999. 17: p. S3-S10.

137

39. Webster, R., et al., Molecular mechanisms of variation in influenza viruses. Nature, 1982. 296(5853): p. 115-121.

40. Bush, R.M., et al., Positive selection on the H3 hemagglutinin gene of human influenza virus A. Molecular biology and evolution, 1999. 16(11): p. 1457-1465.

41. Suzuki, Y., Natural selection on the influenza virus genome. Molecular biology and evolution, 2006. 23(10): p. 1902-1911.

42. Fitch, W.M., et al., Positive Darwinian evolution in human influenza A viruses. Proceedings of the National Academy of Sciences, 1991. 88(10): p. 4270-4274.

43. Bean, W., et al., Evolution of the H3 influenza virus hemagglutinin from human and nonhuman hosts. Journal of virology, 1992. 66(2): p. 1129-1138.

44. Wiersma, L.C., G.F. Rimmelzwaan, and R.D. de Vries, Developing Universal Influenza Vaccines: Hitting the Nail, Not Just on the Head. Vaccines (Basel), 2015. 3(2): p. 239-62.

45. Sheikh, Q.M., et al., Towards the knowledge-based design of universal influenza epitope ensemble vaccines. Bioinformatics, 2016.

46. Ren, H. and P. Zhou, Epitope-focused vaccine design against influenza A and B viruses. Curr Opin Immunol, 2016. 42: p. 83-90.

47. Krause, J.C. and J.E. Crowe, Jr., Committing the Oldest Sins in the Newest Kind of Ways-Antibodies Targeting the Influenza Virus Type A Hemagglutinin Globular Head. Microbiol Spectr, 2014. 2(5).

48. Krammer, F., Novel universal influenza virus vaccine approaches. Curr Opin Virol, 2016. 17: p. 95-103.

49. de Vries, R.D., A.F. Altenburg, and G.F. Rimmelzwaan, Universal influenza vaccines: a realistic option? Clin Microbiol Infect, 2016.

50. Nelson, M.I. and E.C. Holmes, The evolution of epidemic influenza. Nature reviews genetics, 2007. 8(3): p. 196-205.

51. Holmes, E.C., et al., Whole-genome analysis of human influenza A virus reveals multiple persistent lineages and reassortment among recent H3N2 viruses. PLoS Biol, 2005. 3(9): p. e300.

52. Lindstrom, S.E., N.J. Cox, and A. Klimov, Genetic analysis of human H2N2 and early H3N2 influenza viruses, 1957–1972: evidence for genetic divergence and multiple reassortment events. Virology, 2004. 328(1): p. 101-119. 138

53. Barr, I., et al., An influenza A (H3) reassortant was epidemic in Australia and New Zealand in 2003. Journal of Medical Virology, 2005. 76(3): p. 391-397.

54. Garten, R.J., et al., Antigenic and genetic characteristics of swine-origin 2009 A (H1N1) influenza viruses circulating in humans. science, 2009. 325(5937): p. 197-201.

55. Scholtissek, C., et al., On the origin of the human influenza virus subtypes H2N2 and H3N2. Virology, 1978. 87(1): p. 13-20.

56. Kawaoka, Y., S. Krauss, and R.G. Webster, Avian-to-human transmission of the PB1 gene of influenza A viruses in the 1957 and 1968 pandemics. Journal of virology, 1989. 63(11): p. 4603-4608.

57. Bridges, C.B., et al., Prevention and control of influenza. Recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR. Recommendations and reports: Morbidity and mortality weekly report. Recommendations and reports/Centers for Disease Control, 2003. 52(RR-8): p. 1- 34; quiz CE1-4.

58. Reed, C., et al., Public health impact of including two lineages of influenza B in a quadrivalent seasonal influenza vaccine. Vaccine, 2012. 30(11): p. 1993-1998.

59. Lambert, N.D., et al., Understanding the immune response to seasonal influenza vaccination in older adults: a systems biology approach. Expert review of vaccines, 2012. 11(8): p. 985-994.

60. Chen, Z., et al., Generation of live attenuated novel influenza virus A/California/7/09 (H1N1) vaccines with high yield in embryonated chicken eggs. Journal of virology, 2010. 84(1): p. 44-51.

61. Bright, R.A., et al., Influenza virus-like particles elicit broader immune responses than whole virion inactivated influenza virus or recombinant hemagglutinin. Vaccine, 2007. 25(19): p. 3871-3878.

62. Key Facts About Seasonal Flu Vaccine. 2016 [cited 2016; Available from: http://www.cdc.gov/flu/protect/keyfacts.htm.

63. Selecting Viruses for the Seasonal Influenza Vaccine. 2016 [cited 2016; Available from: http://www.cdc.gov/flu/about/season/vaccine-selection.htm.

64. Havers, F.P., et al., Case-control study of vaccine effectiveness in preventing laboratory-confirmed influenza hospitalizations in older adults, United States, 2010-11. Clinical Infectious Diseases, 2016: p. ciw512.

139

65. Talbot, H.K., et al., Effectiveness of influenza vaccine for preventing laboratory- confirmed influenza hospitalizations in adults, 2011-2012 influenza season. Clinical infectious diseases, 2013: p. cit124.

66. Talbot, H.K., et al., Effectiveness of seasonal vaccine in preventing confirmed influenza-associated hospitalizations in community dwelling older adults. Journal of Infectious Diseases, 2011. 203(4): p. 500-508.

67. Ferdinands, J.M., et al., Effectiveness of influenza vaccine against life-threatening RT-PCR-confirmed influenza illness in US children, 2010–2012. Journal of Infectious Diseases, 2014. 210(5): p. 674-683.

68. Thompson, M.G., et al., Effectiveness of seasonal trivalent influenza vaccine for preventing influenza virus illness among pregnant women: a population-based case-control study during the 2010-2011 and 2011-2012 influenza seasons. Clinical infectious diseases, 2013: p. cit750.

69. Benowitz, I., et al., Influenza vaccine given to pregnant women reduces hospitalization due to influenza in their infants. Clinical Infectious Diseases, 2010. 51(12): p. 1355-1361.

70. Madhi, S.A., et al., Influenza vaccination of pregnant women and protection of their infants. New England Journal of Medicine, 2014. 371(10): p. 918-931.

71. Chung, J.R., et al., Seasonal effectiveness of live attenuated and inactivated influenza vaccine. Pediatrics, 2016: p. peds. 2015-3279.

72. Chung, B.F.a.J. Influenza Vaccine Effectiveness, Including LAIV vs IIV in Children and Adolescents, US Flu VE Network, 2015-16. 2016.

73. Yen, H.-L., Current and novel antiviral strategies for influenza infection. Current Opinion in Virology, 2016. 18: p. 126-134.

74. Zarogiannis, S.G., et al., Comparison of ribavirin and oseltamivir in reducing mortality and lung injury in mice infected with mouse adapted A/California/04/2009 (H1N1). Life sciences, 2012. 90(11): p. 440-445.

75. Govorkova, E.A., et al., Assessment of the efficacy of the neuraminidase inhibitor oseltamivir against 2009 pandemic H1N1 influenza virus in ferrets. Antiviral research, 2011. 91(2): p. 81-88.

76. Gupta, Y.K., M. Meenu, and P. Mohan, The Tamiflu fiasco and lessons learnt. Indian journal of pharmacology, 2015. 47(1): p. 11.

140

77. Hayden, F.G., et al., Emergence and apparent transmission of rimantadine- resistant influenza A virus in families. New England Journal of Medicine, 1989. 321(25): p. 1696-1702.

78. Deyde, V.M., et al., Surveillance of resistance to adamantanes among influenza A (H3N2) and A (H1N1) viruses isolated worldwide. Journal of Infectious Diseases, 2007. 196(2): p. 249-257.

79. Bright, R.A., et al., Incidence of adamantane resistance among influenza A (H3N2) viruses isolated worldwide from 1994 to 2005: a cause for concern. The Lancet, 2005. 366(9492): p. 1175-1181.

80. Alves Galvão, M.G., M.A. Rocha Crispino Santos, and A.J. Alves da Cunha, Amantadine and rimantadine for influenza A in children and the elderly. The Cochrane Library, 2012.

81. Furuta, Y., et al., Favipiravir (T-705), a novel viral RNA polymerase inhibitor. Antiviral research, 2013. 100(2): p. 446-454.

82. Sangawa, H., et al., Mechanism of action of T-705 ribosyl triphosphate against influenza virus RNA polymerase. Antimicrobial agents and chemotherapy, 2013. 57(11): p. 5202-5208.

83. Arias, A., L. Thorne, and I. Goodfellow, Favipiravir elicits antiviral mutagenesis during virus replication in vivo. Elife, 2014. 3: p. e03679.

84. Lejal, N., et al., Structure-based discovery of the novel antiviral properties of naproxen against the nucleoprotein of influenza A virus. Antimicrobial agents and chemotherapy, 2013. 57(5): p. 2231-2242.

85. Kao, R.Y., et al., Identification of influenza A nucleoprotein as an antiviral target. Nature biotechnology, 2010. 28(6): p. 600-605.

86. Gerritz, S.W., et al., Inhibition of influenza virus replication via small molecules that induce the formation of higher-order nucleoprotein oligomers. Proceedings of the National Academy of Sciences, 2011. 108(37): p. 15366-15371.

87. Su, C.-Y., et al., High-throughput identification of compounds targeting influenza RNA-dependent RNA polymerase activity. Proceedings of the National Academy of Sciences, 2010. 107(45): p. 19151-19156.

88. Reguera, J., P. Gerlach, and S. Cusack, Towards a structural understanding of RNA synthesis by negative strand RNA viral polymerases. Current opinion in structural biology, 2016. 36: p. 75-84.

141

89. Clark, M.P., et al., Discovery of a novel, first-in-class, orally bioavailable azaindole inhibitor (VX-787) of influenza PB2. Journal of medicinal chemistry, 2014. 57(15): p. 6668-6678.

90. Byrn, R.A., et al., Preclinical activity of VX-787, a first-in-class, orally bioavailable inhibitor of the influenza virus polymerase PB2 subunit. Antimicrobial agents and chemotherapy, 2015. 59(3): p. 1569-1582.

91. McKimm-Breschkin, J.L. and A.M. Fry, Meeting report: 4th ISIRV antiviral group conference: Novel antiviral therapies for influenza and other respiratory viruses. Antiviral research, 2016. 129: p. 21-38.

92. Stevaert, A., et al., Mutational analysis of the binding pockets of the diketo acid inhibitor L-742,001 in the influenza virus PA endonuclease. Journal of virology, 2013. 87(19): p. 10524-10538.

93. Tellier, R., Review of aerosol transmission of influenza A virus. Emerg Infect Dis, 2006. 12(11): p. 1657-1662.

94. Tellier, R., Aerosol transmission of influenza A virus: a review of new studies. Journal of the Royal Society Interface, 2009: p. rsif20090302.

95. McCormack, F.X. and J.A. Whitsett, The pulmonary collectins, SP-A and SP-D, orchestrate innate immunity in the lung. The Journal of clinical investigation, 2002. 109(6): p. 707-712.

96. Baum, L. and J. Paulson, Sialyloligosaccharides of the respiratory epithelium in the selection of human influenza virus receptor specificity. Acta histochemica. Supplementband, 1989. 40: p. 35-38.

97. Cohen, M., et al., Influenza A penetrates host mucus by cleaving sialic acids with neuraminidase. Virology journal, 2013. 10(1): p. 1.

98. Isaacs, A., J. Lindenmann, and R.C. Valentine, Virus interference. II. Some properties of interferon. Proceedings of the Royal Society of London B: Biological Sciences, 1957. 147(927): p. 268-273.

99. Isaacs, A. and J. Lindenmann, Virus interference. I. The interferon. Proceedings of the Royal Society of London B: Biological Sciences, 1957. 147(927): p. 258- 267.

100. Hornung, V., et al., 5'-Triphosphate RNA is the ligand for RIG-I. Science, 2006. 314(5801): p. 994-997.

142

101. Kato, H., et al., Cell type-specific involvement of RIG-I in antiviral response. Immunity, 2005. 23(1): p. 19-28.

102. Strahle, L., et al., Activation of the beta interferon promoter by unnatural Sendai virus infection requires RIG-I and is inhibited by viral C proteins. J Virol, 2007. 81(22): p. 12227-37.

103. Le Goffic, R., et al., Cutting Edge: Influenza A virus activates TLR3-dependent inflammatory and RIG-I-dependent antiviral responses in human lung epithelial cells. The Journal of Immunology, 2007. 178(6): p. 3368-3372.

104. Kowalinski, E., et al., Structural basis for the activation of innate immune pattern-recognition receptor RIG-I by viral RNA. Cell, 2011. 147(2): p. 423-435.

105. Jiang, F., et al., Structural basis of RNA recognition and activation by innate immune receptor RIG-I. Nature, 2011. 479(7373): p. 423-427.

106. Gack, M.U., et al., Influenza A virus NS1 targets the ubiquitin ligase TRIM25 to evade recognition by the host viral RNA sensor RIG-I. Cell host & microbe, 2009. 5(5): p. 439-449.

107. Pothlichet, J., et al., Type I IFN triggers RIG-I/TLR3/NLRP3-dependent inflammasome activation in influenza A virus infected cells. PLoS pathog, 2013. 9(4): p. e1003256.

108. Ichinohe, T., I.K. Pang, and A. Iwasaki, Influenza virus activates inflammasomes via its intracellular M2 ion channel. Nature immunology, 2010. 11(5): p. 404- 410.

109. Allen, I.C., et al., The NLRP3 inflammasome mediates in vivo innate immunity to influenza A virus through recognition of viral RNA. Immunity, 2009. 30(4): p. 556-565.

110. Sibille, Y. and H.Y. Reynolds, Macrophages and polymorphonuclear neutrophils in lung defense and injury. American Review of Respiratory Disease, 1990. 141(2): p. 471-501.

111. Hashimoto, Y., et al., Evidence for phagocytosis of influenza virus-infected, apoptotic cells by neutrophils and macrophages in mice. The Journal of Immunology, 2007. 178(4): p. 2448-2457.

112. Kaminski, M.M., et al., Plasmacytoid dendritic cells and Toll-like receptor 7- dependent signalling promote efficient protection of mice against highly virulent influenza A virus. Journal of General Virology, 2012. 93(3): p. 555-559.

143

113. Diebold, S.S., et al., Innate antiviral responses by means of TLR7-mediated recognition of single-stranded RNA. Science, 2004. 303(5663): p. 1529-1531.

114. Pang, I.K., P.S. Pillai, and A. Iwasaki, Efficient influenza A virus replication in the respiratory tract requires signals from TLR7 and RIG-I. Proceedings of the National Academy of Sciences, 2013. 110(34): p. 13910-13915.

115. Kawai, T., et al., Interferon-α induction through Toll-like receptors involves a direct interaction of IRF7 with MyD88 and TRAF6. Nature immunology, 2004. 5(10): p. 1061-1068.

116. Luster, A.D., R. Alon, and U.H. von Andrian, Immune cell migration in inflammation: present and future therapeutic targets. Nature immunology, 2005. 6(12): p. 1182-1190.

117. Vivier, E., et al., Functions of natural killer cells. Nature immunology, 2008. 9(5): p. 503-510.

118. Savill, J., et al., A blast from the past: clearance of apoptotic cells regulates immune responses. Nature Reviews Immunology, 2002. 2(12): p. 965-975.

119. Staeheli, P., Interferon-induced proteins and the antiviral state. Advances in virus research, 1990. 38: p. 147-200.

120. Schneider, W.M., M.D. Chevillotte, and C.M. Rice, Interferon-stimulated genes: a complex web of host defenses. Annual review of immunology, 2014. 32: p. 513.

121. Turan, K., et al., Nuclear MxA proteins form a complex with influenza virus NP and inhibit the transcription of the engineered influenza virus genome. Nucleic acids research, 2004. 32(2): p. 643-652.

122. Lenschow, D.J., et al., IFN-stimulated gene 15 functions as a critical antiviral molecule against influenza, herpes, and Sindbis viruses. Proc Natl Acad Sci U S A, 2007. 104(4): p. 1371-6.

123. Silverman, R.H., Viral encounters with 2′, 5′-oligoadenylate synthetase and RNase L during the interferon antiviral response. Journal of virology, 2007. 81(23): p. 12720-12729.

124. Liu, S.-Y., et al., Interferon-inducible cholesterol-25-hydroxylase broadly inhibits viral entry by production of 25-hydroxycholesterol. Immunity, 2013. 38(1): p. 92- 105.

144

125. Schulz, O., et al., Protein kinase R contributes to immunity against specific viruses by regulating interferon mRNA integrity. Cell host & microbe, 2010. 7(5): p. 354-361.

126. Hinson, E.R. and P. Cresswell, The N-terminal amphipathic α-helix of viperin mediates localization to the cytosolic face of the endoplasmic reticulum and inhibits protein secretion. Journal of Biological Chemistry, 2009. 284(7): p. 4705- 4712.

127. Holt, P., Regulation of antigen-presenting cell function (s) in lung and airway tissues. European respiratory journal, 1993. 6(1): p. 120-129.

128. Banchereau, J. and R.M. Steinman, Dendritic cells and the control of immunity. Nature, 1998. 392(6673): p. 245-252.

129. GeurtsvanKessel, C.H., et al., Clearance of influenza virus from the lung depends on migratory langerin+ CD11b− but not plasmacytoid dendritic cells. The Journal of experimental medicine, 2008. 205(7): p. 1621-1634.

130. Guermonprez, P., et al., Antigen presentation and T cell stimulation by dendritic cells. Annual review of immunology, 2002. 20(1): p. 621-667.

131. Langenkamp, A., et al., Kinetics of dendritic cell activation: impact on priming of TH1, TH2 and nonpolarized T cells. Nature immunology, 2000. 1(4): p. 311-316.

132. Helft, J., et al., Cross-presenting CD103+ dendritic cells are protected from influenza virus infection. The Journal of clinical investigation, 2012. 122(11): p. 4037-4047.

133. Jakubzick, C., et al., Modulation of dendritic cell trafficking to and from the airways. The Journal of Immunology, 2006. 176(6): p. 3578-3584.

134. Kim, T.S., et al., Antigen persistence and the control of local T cell memory by migrant respiratory dendritic cells after acute virus infection. The Journal of experimental medicine, 2010. 207(6): p. 1161-1172.

135. Nakano, H., et al., Blood-derived inflammatory dendritic cells in lymph nodes stimulate acute T helper type 1 immune responses. Nature immunology, 2009. 10(4): p. 394-402.

136. Topham, D.J. and P.C. Doherty, Clearance of an influenza A virus by CD4+ T cells is inefficient in the absence of B cells. Journal of virology, 1998. 72(1): p. 882-885.

145

137. Lawrence, C.W. and T.J. Braciale, Activation, differentiation, and migration of naive virus-specific CD8+ T cells during pulmonary influenza virus infection. The Journal of Immunology, 2004. 173(2): p. 1209-1218.

138. Campbell, D.J. and M.A. Koch, Phenotypical and functional specialization of FOXP3+ regulatory T cells. Nature Reviews Immunology, 2011. 11(2): p. 119- 130.

139. Koch, M.A., et al., The transcription factor T-bet controls regulatory T cell homeostasis and function during type 1 inflammation. Nature immunology, 2009. 10(6): p. 595-602.

140. Ohkura, N. and S. Sakaguchi, Maturation of effector regulatory T cells. Nature immunology, 2011. 12(4): p. 283-284.

141. Sun, J., et al., Effector T cells control lung inflammation during acute influenza virus infection by producing IL-10. Nature medicine, 2009. 15(3): p. 277-284.

142. Sun, J., et al., CD4+ T cell help and innate-derived IL-27 induce Blimp-1- dependent IL-10 production by antiviral CTLs. Nature immunology, 2011. 12(4): p. 327-334.

143. Williams, A.E., et al., TGF-β prevents eosinophilic lung disease but impairs pathogen clearance. Microbes and infection, 2005. 7(3): p. 365-374.

144. Mantovani, A., et al., Macrophage plasticity and polarization in tissue repair and remodelling. The Journal of pathology, 2013. 229(2): p. 176-185.

145. Novak, M.L. and T.J. Koh, Phenotypic transitions of macrophages orchestrate tissue repair. The American journal of pathology, 2013. 183(5): p. 1352-1363.

146. Tsunawaki, S., et al., Deactivation of macrophages by transforming growth factor-β. 1988.

147. Sallman Almen, M., et al., The dispanins: a novel gene family of ancient origin that contains 14 human members. PloS one, 2012. 7(2): p. e31961.

148. Zhang, Z., et al., Evolutionary Dynamics of the Interferon-Induced Transmembrane Gene Family in Vertebrates. PloS one, 2012. 7(11).

149. Friedman, R.L., et al., Transcriptional and posttranscriptional regulation of interferon-induced gene expression in human cells. Cell, 1984. 38(3): p. 745-55.

146

150. Reid, L.E., et al., A single DNA response element can confer inducibility by both alpha-and gamma-interferons. Proceedings of the National Academy of Sciences, 1989. 86(3): p. 840-844.

151. Hanagata, N., et al., Characterization of the osteoblast-specific transmembrane protein IFITM5 and analysis of IFITM5-deficient mice. Journal of bone and mineral metabolism, 2011. 29(3): p. 279-290.

152. Shapiro, J.R., et al., Phenotypic variability of osteogenesis imperfecta type V caused by an IFITM5 mutation. Journal of Bone and Mineral Research, 2013. 28(7): p. 1523-1530.

153. Brass, A.L., et al., The IFITM proteins mediate cellular resistance to influenza A H1N1 virus, West Nile virus, and dengue virus. Cell, 2009. 139(7): p. 1243-54.

154. Yount, J.S., et al., Palmitoylome profiling reveals S-palmitoylation-dependent antiviral activity of IFITM3. Nat Chem Biol, 2010. 6(8): p. 610-4.

155. Schoggins, J.W., et al., A diverse range of gene products are effectors of the type I interferon antiviral response. Nature, 2011. 472(7344): p. 481-485.

156. Everitt, A.R., et al., IFITM3 restricts the morbidity and mortality associated with influenza. Nature, 2012. 484(7395): p. 519-23.

157. Yang, X., et al., Interferon-inducible transmembrane protein 3 genetic variant rs12252 and influenza susceptibility and severity: a meta-analysis. PloS one, 2015. 10(5): p. e0124985.

158. Zhang, Y.-H., et al., Interferon-induced transmembrane protein-3 genetic variant rs12252-C is associated with severe influenza in Chinese individuals. Nature communications, 2013. 4: p. 1418.

159. Xuan, Y., et al., IFITM3 rs12252 T>C polymorphism is associated with the risk of severe influenza: a meta-analysis. Epidemiol Infect, 2015. 143(14): p. 2975-84.

160. Wang, Z., et al., Early hypercytokinemia is associated with interferon-induced transmembrane protein-3 dysfunction and predictive of fatal H7N9 infection. Proc Natl Acad Sci U S A, 2014. 111(2): p. 769-74.

161. Zhang, Y., et al., Interferon-induced transmembrane protein-3 rs12252-C is associated with rapid progression of acute HIV-1 infection in Chinese MSM cohort. Aids, 2015. 29(8): p. 889-94.

162. Wilkins, C., et al., IFITM1 Is a Tight Junction Protein That Inhibits Hepatitis C Virus Entry. Hepatology, 2013. 57(2): p. 461-469. 147

163. Savidis, G., et al., The IFITMs Inhibit Zika Virus Replication. Cell reports, 2016. 15(11): p. 2323-30.

164. Weidner, J.M., et al., Interferon-induced cell membrane proteins, IFITM3 and tetherin, inhibit vesicular stomatitis virus infection via distinct mechanisms. J Virol, 2010. 84(24): p. 12646-57.

165. Smith, S.E., et al., Chicken Interferon-Inducible Transmembrane Protein 3 Restricts Influenza Viruses and Lyssaviruses In Vitro. Journal of virology, 2013. 87(23): p. 12957-12966.

166. Huang, I.C., et al., Distinct patterns of IFITM-mediated restriction of filoviruses, SARS coronavirus, and influenza A virus. PLoS Pathog, 2011. 7(1): p. e1001258.

167. Jia, R., et al., The N-terminal region of IFITM3 modulates its antiviral activity by regulating IFITM3 cellular localization. J Virol, 2012. 86(24): p. 13697-707.

168. Lu, J., et al., The IFITM proteins inhibit HIV-1 infection. J Virol, 2011. 85(5): p. 2126-37.

169. Li, K., et al., IFITM proteins restrict viral membrane hemifusion. PLoS Pathog, 2013. 9(1): p. e1003124.

170. Weston, S., et al., Alphavirus Restriction by IFITM Proteins. Traffic, 2016. 17(9): p. 997-1013.

171. Mudhasani, R., et al., IFITM-2 and IFITM-3 but not IFITM-1 restrict Rift Valley fever virus. J Virol, 2013. 87(15): p. 8451-64.

172. Anafu, A.A., et al., Interferon-inducible transmembrane protein 3 (IFITM3) restricts reovirus cell entry. J Biol Chem, 2013. 288(24): p. 17261-71.

173. Smith, S., et al., IFITM proteins-cellular inhibitors of viral entry. Current opinion in virology, 2014. 4C: p. 71-77.

174. Feeley, E.M., et al., IFITM3 inhibits influenza A virus infection by preventing cytosolic entry. PLoS Pathog, 2011. 7(10): p. e1002337.

175. Desai, T.M., et al., IFITM3 restricts influenza A virus entry by blocking the formation of fusion pores following virus-endosome hemifusion. PLoS Pathog, 2014. 10(4): p. e1004048.

176. Perreira, J.M., et al., IFITMs restrict the replication of multiple pathogenic viruses. J Mol Biol, 2013. 425(24): p. 4937-55.

148

177. Hach, J.C., et al., Palmitoylation on conserved and non-conserved cysteines of murine IFITM1 regulates its stability and anti-influenza A virus activity. Journal of virology, 2013.

178. John, S.P., et al., The CD225 domain of IFITM3 is required for both IFITM protein association and inhibition of influenza A virus and dengue virus replication. J Virol, 2013. 87(14): p. 7837-52.

179. Yount, J.S., R.A. Karssemeijer, and H.C. Hang, S-palmitoylation and ubiquitination differentially regulate interferon-induced transmembrane protein 3 (IFITM3)-mediated resistance to influenza virus. J Biol Chem, 2012. 287(23): p. 19631-41.

180. Wee, Y.S., et al., Interferon-inducible transmembrane proteins of the innate immune response act as membrane organizers by influencing clathrin and v- ATPase localization and function. Innate immunity, 2012. 18(6): p. 834-45.

181. Amini-Bavil-Olyaee, S., et al., The antiviral effector IFITM3 disrupts intracellular cholesterol homeostasis to block viral entry. Cell Host Microbe, 2013. 13(4): p. 452-64.

182. Lin, T.Y., et al., Amphotericin B Increases Influenza A Virus Infection by Preventing IFITM3-Mediated Restriction. Cell reports, 2013.

183. Kohn, A., Early interactions of viruses with cellular membranes. Advances in virus research, 1979. 24: p. 223-76.

184. Schoggins, J.W. and G. Randall, Lipids in innate antiviral defense. Cell host & microbe, 2013. 14(4): p. 379-85.

185. Liu, S.Y., D.J. Sanchez, and G. Cheng, New developments in the induction and antiviral effectors of type I interferon. Current opinion in immunology, 2011. 23(1): p. 57-64.

186. Smith, R.A., et al., Expression of the mouse fragilis gene products in immune cells and association with receptor signaling complexes. Genes Immun, 2006. 7(2): p. 113-21.

187. Bailey, C.C., et al., Interferon-induced transmembrane protein 3 is a type II transmembrane protein. J Biol Chem, 2013. 288(45): p. 32184-93.

188. Uittenbogaard, A., Y. Ying, and E.J. Smart, Characterization of a cytosolic heat- shock protein-caveolin chaperone complex. Involvement in cholesterol trafficking. The Journal of biological chemistry, 1998. 273(11): p. 6525-32.

149

189. Kalashnikova, E., et al., SynDIG1: an activity-regulated, AMPA- receptor- interacting transmembrane protein that regulates excitatory synapse development. Neuron, 2010. 65(1): p. 80-93.

190. Chesarino, N.M., et al., Phosphorylation of the Antiviral Protein IFITM3 Dually Regulates its Endocytosis and Ubiquitination. The Journal of biological chemistry, 2014.

191. Jia, R., et al., Identification of an endocytic signal essential for the antiviral action of IFITM3. Cell Microbiol, 2014. 16(7): p. 1080-93.

192. Patoine, A., et al., Topological Mapping of BRIL Reveals a Type II Orientation and Effects of Osteogenesis Imperfecta Mutations on Its Cellular Destination. Journal of bone and mineral research : the official journal of the American Society for Bone and Mineral Research, 2014.

193. Tom, C.T. and B.R. Martin, Fat chance! Getting a grip on a slippery modification. ACS chemical biology, 2013. 8(1): p. 46-57.

194. Linder, M.E. and R.J. Deschenes, Palmitoylation: policing protein stability and traffic. Nat Rev Mol Cell Biol, 2007. 8(1): p. 74-84.

195. Resh, M.D., Palmitoylation of ligands, receptors, and intracellular signaling molecules. Sci STKE, 2006. 2006(359): p. re14.

196. Blaskovic, S., M. Blanc, and F.G. van der Goot, What does S-palmitoylation do to membrane proteins? The FEBS journal, 2013.

197. Yount, J.S., M.M. Zhang, and H.C. Hang, Emerging roles for protein S- palmitoylation in immunity from chemical proteomics. Curr Opin Chem Biol, 2013. 17(1): p. 27-33.

198. Yount, J.S., G. Charron, and H.C. Hang, Bioorthogonal proteomics of 15- hexadecynyloxyacetic acid chemical reporter reveals preferential targeting of fatty acid modified proteins and biosynthetic enzymes. Bioorg Med Chem, 2012. 20(2): p. 650-4.

199. Tsukamoto, T., et al., Role of S-palmitoylation on IFITM5 for the interaction with FKBP11 in osteoblast cells. PloS one, 2013. 8(9): p. e75831.

200. Wagner, S.A., et al., A proteome-wide, quantitative survey of in vivo ubiquitylation sites reveals widespread regulatory roles. Molecular & cellular proteomics : MCP, 2011. 10(10): p. M111 013284.

150

201. Wagner, S.A., et al., Proteomic analyses reveal divergent ubiquitylation site patterns in murine tissues. Molecular & cellular proteomics : MCP, 2012. 11(12): p. 1578-85.

202. Kim, W., et al., Systematic and quantitative assessment of the ubiquitin-modified proteome. Molecular cell, 2011. 44(2): p. 325-40.

203. Chen, T., et al., mUbiSiDa: a comprehensive database for protein ubiquitination sites in mammals. PloS one, 2014. 9(1): p. e85744.

204. Kulathu, Y. and D. Komander, Atypical ubiquitylation - the unexplored world of polyubiquitin beyond Lys48 and Lys63 linkages. Nature reviews. Molecular cell biology, 2012. 13(8): p. 508-23.

205. Shan, Z., et al., Negative regulation of interferon-induced transmembrane protein 3 by SET7-mediated lysine monomethylation. J Biol Chem, 2013. 288(49): p. 35093-103.

206. Wang, H., et al., Purification and functional characterization of a histone H3- lysine 4-specific methyltransferase. Molecular cell, 2001. 8(6): p. 1207-17.

207. Kurash, J.K., et al., Methylation of p53 by Set7/9 mediates p53 acetylation and activity in vivo. Molecular cell, 2008. 29(3): p. 392-400.

208. Munro, S., et al., Lysine methylation regulates the pRb tumour suppressor protein. Oncogene, 2010. 29(16): p. 2357-67.

209. Pagans, S., et al., Characterization of HIV Tat modifications using novel methyl- lysine-specific antibodies. Methods, 2011. 53(1): p. 91-6.

210. Esteve, P.O., et al., A methylation and phosphorylation switch between an adjacent lysine and serine determines human DNMT1 stability. Nature structural & molecular biology, 2011. 18(1): p. 42-8.

211. Liu, X., et al., Methyltransferase Set7/9 regulates p53 activity by interacting with Sirtuin 1 (SIRT1). Proceedings of the National Academy of Sciences of the United States of America, 2011. 108(5): p. 1925-30.

212. Alland, L., et al., Dual myristylation and palmitylation of Src family member p59fyn affects subcellular localization. J Biol Chem, 1994. 269(24): p. 16701-5.

213. van't Hof, W. and M.D. Resh, Rapid plasma membrane anchoring of newly synthesized p59fyn: selective requirement for NH2-terminal myristoylation and palmitoylation at cysteine-3. The Journal of cell biology, 1997. 136(5): p. 1023- 35. 151

214. Jiang, D., et al., Identification of five interferon-induced cellular proteins that inhibit west nile virus and dengue virus infections. J Virol, 2010. 84(16): p. 8332- 41.

215. Bailey, C.C., et al., Ifitm3 limits the severity of acute influenza in mice. PLoS Pathog, 2012. 8(9): p. e1002909.

216. Wang, Z., et al., Early hypercytokinemia is associated with interferon-induced transmembrane protein-3 dysfunction and predictive of fatal H7N9 infection. Proceedings of the National Academy of Sciences of the United States of America, 2013. 111(2).

217. Zhang, Y.H., et al., Interferon-induced transmembrane protein-3 genetic variant rs12252-C is associated with severe influenza in Chinese individuals. Nat Commun, 2013. 4: p. 1418.

218. Diamond, M.S. and M. Farzan, The broad-spectrum antiviral functions of IFIT and IFITM proteins. Nat Rev Immunol, 2013. 13(1): p. 46-57.

219. Moltedo, B., et al., Unique type I interferon responses determine the functional fate of migratory lung dendritic cells during influenza virus infection. PLoS Pathog, 2011. 7(11): p. e1002345.

220. Hach, J.C., et al., Palmitoylation on conserved and nonconserved cysteines of murine IFITM1 regulates its stability and anti-influenza A virus activity. J Virol, 2013. 87(17): p. 9923-7.

221. Pandey, K.N., Functional roles of short sequence motifs in the endocytosis of membrane receptors. Front Biosci (Landmark Ed), 2009. 14: p. 5339-60.

222. Hunter, T., The age of crosstalk: phosphorylation, ubiquitination, and beyond. Mol Cell, 2007. 28(5): p. 730-8.

223. Huang, F. and H. Gu, Negative regulation of lymphocyte development and function by the Cbl family of proteins. Immunol Rev, 2008. 224: p. 229-38.

224. Ghosh, A.K., et al., Biochemical basis for the requirement of kinase activity for Cbl-dependent ubiquitinylation and degradation of a target tyrosine kinase. J Biol Chem, 2004. 279(34): p. 36132-41.

225. Rotin, D. and S. Kumar, Physiological functions of the HECT family of ubiquitin ligases. Nat Rev Mol Cell Biol, 2009. 10(6): p. 398-409.

226. Compton, A.A., et al., IFITM proteins incorporated into HIV-1 virions impair viral fusion and spread. Cell Host Microbe, 2014. 16(6): p. 736-47. 152

227. Tartour, K., et al., IFITM proteins are incorporated onto HIV-1 virion particles and negatively imprint their infectivity. Retrovirology, 2014. 11: p. 103.

228. Melvin, W.J., et al., IFITMs from Mycobacteria Confer Resistance to Influenza Virus When Expressed in Human Cells. Viruses, 2015. 7(6): p. 3035-52.

229. Chesarino, N.M., T.M. McMichael, and J.S. Yount, Regulation of the trafficking and antiviral activity of IFITM3 by post-translational modifications. Future Microbiol, 2014. 9(10): p. 1151-63.

230. Wu, C., et al., BioGPS: an extensible and customizable portal for querying and organizing gene annotation resources. Genome Biol, 2009. 10(11): p. R130.

231. Martin-Serrano, J., et al., HECT ubiquitin ligases link viral and cellular PPXY motifs to the vacuolar protein-sorting pathway. J Cell Biol, 2005. 168(1): p. 89- 101.

232. Chesarino, N.M., et al., Chemoproteomics reveals Toll-like receptor fatty acylation. BMC Biol, 2014. 12: p. 91.

233. Malakhova, O.A. and D.E. Zhang, ISG15 inhibits Nedd4 ubiquitin E3 activity and enhances the innate antiviral response. J Biol Chem, 2008. 283(14): p. 8783-7.

234. Okumura, A., P.M. Pitha, and R.N. Harty, ISG15 inhibits Ebola VP40 VLP budding in an L-domain-dependent manner by blocking Nedd4 ligase activity. Proc Natl Acad Sci U S A, 2008. 105(10): p. 3974-9.

235. Pincetic, A., et al., The interferon-induced gene ISG15 blocks retrovirus release from cells late in the budding process. J Virol, 2010. 84(9): p. 4725-36.

236. Fouladkou, F., et al., The ubiquitin ligase Nedd4-1 is dispensable for the regulation of PTEN stability and localization. Proc Natl Acad Sci U S A, 2008. 105(25): p. 8585-90.

237. Gao, S., et al., Ubiquitin ligase Nedd4L targets activated Smad2/3 to limit TGF- beta signaling. Mol Cell, 2009. 36(3): p. 457-68.

238. Kavsak, P., et al., Smad7 binds to Smurf2 to form an E3 ubiquitin ligase that targets the TGF beta receptor for degradation. Molecular cell, 2000. 6(6): p. 1365-1375.

239. Kim, K.I., et al., Interferon-inducible ubiquitin E2, Ubc8, is a conjugating enzyme for protein ISGylation. Molecular and cellular biology, 2004. 24(21): p. 9592- 9600.

153

240. Milkereit, R. and D. Rotin, A role for the ubiquitin ligase Nedd4 in membrane sorting of LAPTM4 proteins. PLoS One, 2011. 6(11): p. e27478.

241. Lopez, C.B., et al., Sendai virus infection induces efficient adaptive immunity independently of type I interferons. J Virol, 2006. 80(9): p. 4538-45.

242. Campeau, E., et al., A versatile viral system for expression and depletion of proteins in mammalian cells. PLoS One, 2009. 4(8): p. e6529.

243. Chutiwitoonchai, N., et al., Characteristics of IFITM, the newly identified IFN- inducible anti-HIV-1 family proteins. Microbes Infect, 2013. 15(4): p. 280-90.

244. Magnifico, A., et al., WW domain HECT E3s target Cbl RING finger E3s for proteasomal degradation. J Biol Chem, 2003. 278(44): p. 43169-77.

245. Liu, Q., et al., E3 ubiquitin ligase Cbl-b in innate and adaptive immunity. Cell Cycle, 2014. 13(12): p. 1875-84.

246. Scheffner, M. and S. Kumar, Mammalian HECT ubiquitin-protein ligases: biological and pathophysiological aspects. Biochim Biophys Acta, 2014. 1843(1): p. 61-74.

247. Chesarino, N.M., et al., Phosphorylation of the antiviral protein interferon- inducible transmembrane protein 3 (IFITM3) dually regulates its endocytosis and ubiquitination. J Biol Chem, 2014. 289(17): p. 11986-92.

248. Hoekstra, D., et al., Characterization of the fusogenic properties of Sendai virus: kinetics of fusion with erythrocyte membranes. Biochemistry, 1985. 24(18): p. 4739-45.

249. Lenschow, D.J., et al., IFN-stimulated gene 15 functions as a critical antiviral molecule against influenza, herpes, and Sindbis viruses. Proceedings of the National Academy of Sciences, 2007. 104(4): p. 1371-1376.

250. Lai, C., et al., Mice lacking the ISG15 E1 enzyme UbE1L demonstrate increased susceptibility to both mouse-adapted and non-mouse-adapted influenza B virus infection. J Virol, 2009. 83(2): p. 1147-51.

251. Zhao, C., et al., ISG15 conjugation system targets the viral NS1 protein in influenza A virus-infected cells. Proc Natl Acad Sci U S A, 2010. 107(5): p. 2253- 8.

252. Tang, Y.J., et al., Herc5 Attenuates Influenza A Virus by Catalyzing ISGylation of Viral NS1 Protein. Journal of immunology, 2010. 184(10): p. 5777-5790.

154

253. Yuan, W. and R.M. Krug, Influenza B virus NS1 protein inhibits conjugation of the interferon (IFN)-induced ubiquitin-like ISG15 protein. Embo j, 2001. 20(3): p. 362-71.

254. Zhao, C., et al., Interferon-induced ISG15 pathway: an ongoing virus-host battle. Trends Microbiol, 2013. 21(4): p. 181-6.

255. Versteeg, G.A., et al., Species-specific antagonism of host ISGylation by the influenza B virus NS1 protein. J Virol, 2010. 84(10): p. 5423-30.

256. Harty, R.N., et al., A PPxY motif within the VP40 protein of Ebola virus interacts physically and functionally with a ubiquitin ligase: Implications for filovirus budding. Proceedings of the National Academy of Sciences of the United States of America, 2000. 97(25): p. 13871-13876.

257. Harty, R.N., et al., Rhabdoviruses and the cellular ubiquitin-proteasome system: a budding interaction. J Virol, 2001. 75(22): p. 10623-9.

258. Strack, B., et al., A role for ubiquitin ligase recruitment in retrovirus release. Proc Natl Acad Sci U S A, 2000. 97(24): p. 13063-8.

259. Fouladkou, F., et al., The ubiquitin ligase Nedd4-1 is required for heart development and is a suppressor of thrombospondin-1. J Biol Chem, 2010. 285(9): p. 6770-80.

260. Cao, X.R., et al., Nedd4 controls animal growth by regulating IGF-1 signaling. Sci Signal, 2008. 1(38): p. ra5.

261. Guo, H., et al., E3 ubiquitin ligase Cbl-b regulates Pten via Nedd4 in T cells independently of its ubiquitin ligase activity. Cell Rep, 2012. 1(5): p. 472-82.

262. Yang, B., et al., Nedd4 augments the adaptive immune response by promoting ubiquitin-mediated degradation of Cbl-b in activated T cells. Nat Immunol, 2008. 9(12): p. 1356-63.

263. Xu, C., C.D. Fan, and X. Wang, Regulation of Mdm2 protein stability and the p53 response by NEDD4-1 E3 ligase. Oncogene, 2015. 34(3): p. 281-9.

264. Kawabe, H., et al., Regulation of Rap2A by the ubiquitin ligase Nedd4-1 controls neurite development. Neuron, 2010. 65(3): p. 358-72.

265. Nagpal, P., et al., The ubiquitin ligase Nedd4-1 participates in denervation- induced skeletal muscle atrophy in mice. PLoS One, 2012. 7(10): p. e46427.

155

266. Han, Z., et al., Small-molecule probes targeting the viral PPxY-host Nedd4 interface block egress of a broad range of RNA viruses. J Virol, 2014. 88(13): p. 7294-306.

267. Mund, T., et al., Peptide and small molecule inhibitors of HECT-type ubiquitin ligases. Proc Natl Acad Sci U S A, 2014. 111(47): p. 16736-41.

268. Wrensch, F., M. Winkler, and S. Pohlmann, IFITM proteins inhibit entry driven by the MERS-coronavirus spike protein: evidence for cholesterol-independent mechanisms. Viruses, 2014. 6(9): p. 3683-98.

269. Benfield, C., et al., Bat and pig Interferon-Induced Transmembrane Protein 3 restrict cell entry by influenza virus and lyssaviruses. The Journal of general virology, 2015.

270. Narayana, S.K., et al., The Interferon-induced Transmembrane Proteins, IFITM1, IFITM2, and IFITM3 Inhibit Hepatitis C Virus Entry. The Journal of biological chemistry, 2015. 290(43): p. 25946-59.

271. Wrensch, F., et al., Interferon-Induced Transmembrane Protein-Mediated Inhibition of Host Cell Entry of Ebolaviruses. The Journal of infectious diseases, 2015. 212 Suppl 2: p. S210-8.

272. Poddar, S., et al., The interferon-stimulated gene IFITM3 restricts infection and pathogenesis of arthritogenic and encephalitic alphaviruses. Journal of virology, 2016.

273. Munoz-Moreno, R., et al., Antiviral Role of IFITM Proteins in African Swine Fever Virus Infection. PloS one, 2016. 11(4): p. e0154366.

274. Peng, T. and H.C. Hang, Bifunctional fatty acid chemical reporter for analyzing S-palmitoylated membrane protein-protein interactions in mammalian cells. Journal of the American Chemical Society, 2015. 137(2): p. 556-9.

275. Yang, J., et al., The I-TASSER Suite: protein structure and function prediction. Nature methods, 2015. 12(1): p. 7-8.

276. Yang, J. and Y. Zhang, I-TASSER server: new development for protein structure and function predictions. Nucleic acids research, 2015. 43(W1): p. W174-81.

277. Petersen, B., et al., A generic method for assignment of reliability scores applied to solvent accessibility predictions. BMC structural biology, 2009. 9: p. 51.

278. Rost, B., G. Yachdav, and J. Liu, The PredictProtein server. Nucleic acids research, 2004. 32(Web Server issue): p. W321-6. 156

279. Yachdav, G., et al., PredictProtein--an open resource for online prediction of protein structural and functional features. Nucleic acids research, 2014. 42(Web Server issue): p. W337-43.

280. Porollo, A.A., R. Adamczak, and J. Meller, POLYVIEW: a flexible visualization tool for structural and functional annotations of proteins. Bioinformatics, 2004. 20(15): p. 2460-2.

281. Thevenet, P., et al., PEP-FOLD: an updated de novo structure prediction server for both linear and disulfide bonded cyclic peptides. Nucleic acids research, 2012. 40(W1): p. W288-W293.

282. Shen, Y., et al., Improved PEP-FOLD Approach for Peptide and Miniprotein Structure Prediction. Journal of chemical theory and computation, 2014. 10(10): p. 4745-58.

283. Lepelley, A., et al., Innate sensing of HIV-infected cells. PLoS pathogens, 2011. 7(2): p. e1001284.

284. Compton, A.A., et al., Natural mutations in IFITM3 modulate post-translational regulation and toggle antiviral specificity. EMBO reports, 2016.

285. Haney, E.F., et al., Design of a novel tryptophan-rich membrane-active antimicrobial peptide from the membrane-proximal region of the HIV glycoprotein, gp41. Beilstein journal of organic chemistry, 2012. 8: p. 1172-84.

286. Charron, G., et al., Alkynyl-farnesol reporters for detection of protein S- prenylation in cells. Molecular bioSystems, 2011. 7(1): p. 67-73.

287. Yount, J.S., M.M. Zhang, and H.C. Hang, Visualization and Identification of Fatty Acylated Proteins Using Chemical Reporters. Current Protocols in Chemical Biology, 2011. 3: p. 65-79.

288. Weston, S., et al., A membrane topology model for human interferon inducible transmembrane protein 1. PloS one, 2014. 9(8): p. e104341.

289. Percher, A., et al., Mass-tag labeling reveals site-specific and endogenous levels of protein S-fatty acylation. Proceedings of the National Academy of Sciences of the United States of America, 2016.

290. Ling, S., et al., Combined approaches of EPR and NMR illustrate only one transmembrane helix in the human IFITM3. Scientific reports, 2016. 6: p. 24029.

157

291. Cornell, R.B. and S.G. Taneva, Amphipathic helices as mediators of the membrane interaction of amphitropic proteins, and as modulators of bilayer physical properties. Current Protein & Peptide Science, 2006. 7(6): p. 539-552.

292. Prinz, W.A. and J.E. Hinshaw, Membrane-bending proteins. Critical reviews in biochemistry and molecular biology, 2009. 44(5): p. 278-91.

293. Drin, G. and B. Antonny, Amphipathic helices and membrane curvature. FEBS letters, 2010. 584(9): p. 1840-7.

294. Suetsugu, S., S. Kurisu, and T. Takenawa, Dynamic shaping of cellular membranes by phospholipids and membrane-deforming proteins. Physiological reviews, 2014. 94(4): p. 1219-48.

295. Eisenberg, D., R.M. Weiss, and T.C. Terwilliger, The helical hydrophobic moment: a measure of the amphiphilicity of a helix. Nature, 1982. 299(5881): p. 371-4.

296. Gautier, R., et al., HELIQUEST: a web server to screen sequences with specific alpha-helical properties. Bioinformatics, 2008. 24(18): p. 2101-2.

297. Li, K., et al., A sorting signal suppresses ifitm1 restriction of viral entry. The Journal of biological chemistry, 2014.

298. Siegrist, F., M. Ebeling, and U. Certa, The small interferon-induced transmembrane genes and proteins. Journal of interferon & cytokine research : the official journal of the International Society for Interferon and Cytokine Research, 2011. 31(1): p. 183-97.

299. D'Souza-Schorey, C. and P.D. Stahl, Myristoylation is required for the intracellular localization and endocytic function of ARF6. Experimental cell research, 1995. 221(1): p. 153-9.

300. Donaldson, J.G., Arfs and membrane lipids: sensing, generating and responding to membrane curvature. The Biochemical journal, 2008. 414(2): p. e1-2.

301. Nakanishi, H., P. de los Santos, and A.M. Neiman, Positive and negative regulation of a SNARE protein by control of intracellular localization. Molecular biology of the cell, 2004. 15(4): p. 1802-15.

302. Brewer, K.D., et al., Reluctance to membrane binding enables accessibility of the synaptobrevin SNARE motif for SNARE complex formation. Proceedings of the National Academy of Sciences of the United States of America, 2011. 108(31): p. 12723-8.

158

303. Bailey, C.C., et al., IFITM-Family Proteins: The Cell's First Line of Antiviral Defense. Annual review of virology, 2014. 1: p. 261-283.

304. Yu, J., et al., IFITM Proteins Restrict HIV-1 Infection by Antagonizing the Envelope Glycoprotein. Cell reports, 2015. 13(1): p. 145-56.

305. Podbilewicz, B., Virus and cell fusion mechanisms. Annual review of cell and developmental biology, 2014. 30: p. 111-39.

306. Smrt, S.T., A.W. Draney, and J.L. Lorieau, The Influenza Hemagglutinin Fusion Domain Is an Amphipathic Helical Hairpin That Functions by Inducing Membrane Curvature. Journal of Biological Chemistry, 2015. 290(1): p. 228-238.

307. Roberts, K.L., et al., The Amphipathic Helix of Influenza A Virus M2 Protein Is Required for Filamentous Bud Formation and Scission of Filamentous and Spherical Particles. Journal of virology, 2013. 87(18): p. 9973-9982.

308. Schmidt, N.W., et al., Influenza Virus A M2 Protein Generates Negative Gaussian Membrane Curvature Necessary for Budding and Scission. Journal of the American Chemical Society, 2013. 135(37): p. 13710-13719.

309. Chesarino, N.M., T.M. McMichael, and J.S. Yount, E3 Ubiquitin Ligase NEDD4 promotes influenza virus infection by decreaseing levels of the antiviral protein IFITM3. PLoS pathogens, 2015.

159