Natural killer cell activation, trafficking, and contribution to immune responses to viral pathogens

University of Iowa Iowa Research Online Theses and Dissertations 2013 Natural killer cell activation, trafficking, and contribution to immune responses to viral pathogens Lindsey Elizabeth Carlin University of Iowa Copyright 2013 Lindsey Carlin This dissertation is available at Iowa Research Online: http://ir.uiowa.edu/etd/1302 Recommended Citation Carlin, Lindsey Elizabeth. "Natural killer cell activation, trafficking, and contribution to immune responses to viral pathogens." PhD (Doctor of Philosophy) thesis, University of Iowa, 2013. http://ir.uiowa.edu/etd/1302. Follow this and additional works at: http://ir.uiowa.edu/etd Part of the Immunology of Infectious Disease Commons

NATURAL KILLER CELL ACTIVATION, TRAFFICKING, AND CONTRIBUTION TO IMMUNE RESPONSES TO VIRAL PATHOGENS by Lindsey Elizabeth Carlin A thesis submitted in partial fulfillment of the requirements for the Doctor of Philosophy degree in Immunology in the Graduate College of The University of Iowa August 2013 Thesis Supervisor: Associate Professor Kevin L. Legge

Copyright by LINDSEY ELIZABETH CARLIN 2013 All Rights Reserved

Graduate College The University of Iowa Iowa City, Iowa CERTIFICATE OF APPROVAL PH.D. THESIS This is to certify that the Ph.D. thesis of Lindsey Elizabeth Carlin has been approved by the Examining Committee for the thesis requirement for the Doctor of Philosophy degree in Immunology at the August 2013 graduation. Thesis Committee: Kevin L. Legge, Thesis Supervisor Vladimir Badovinac Jonathan Heusel Aloysius Klingelhutz Stanley Perlman

To Merlot Joe ii

ACKNOWLEDGMENTS I need to thank my mentors, Dr. Jon Heusel and Dr. Kevin Legge, for their support, time, and guidance. Getting through my first day as a graduate student to my defense date was a long, laborious process. I would not have made it without their help. Thank you to the rest of my committee members: Dr. Stanley Perlman, Dr. Al Klinglehuts, and Dr. Vladimir Badovinac, for making meetings less scary than I anticipated, providing helpful feedback, and for support and encouragement throughout this process. I would also like to thank Dr. Steve Varga and Dr. Gail Bishop for allowing me to rotate through their labs and for developing my skills as a scientist. I would like to thank the members of Dr. Heusel s laboratory Colleen Fullenkamp, Dr. Natasha Guseva, and Paul Naumann for their technical support. I would also like to thank members of Dr. Legge s laboratory: Dr. Jodi McGill, Emily Hemann, Emma Hornick. Not only for their technical assistance and advice, but also for their support and friendship. Several other graduate students made completion of my dissertation possible: Nicole Chapman, Sabrina Scroggins, Dr. Anna Peters, Dr. Erik Brincks, Dr. Alexander Boyden, Dr. Elaine Hamarstrom, Dr. Laura Fraczek, Dr. Prajwal Gurung, Britnie Thomas, Breanna Scorza, Farrah Steinke, Sean Duong, Shaniya Kahn, Ann Janowski, and the other immunology graduate students. Thank you for your encouragement, advice, and friendship. Finally, I would like to thank my family, whose love and support have made this possible. My father, for being a wonderful role model and my sounding board throughout graduate school. My mother, for her warmth and encouragement. My siblings, for helping me laugh through the tough times. And John, for keeping me sane and supporting me in my goals. iii

ABSTRACT Natural killer (NK) cells are a critical component of the immune response against viral infections. NK cell depletion prior to murine cytomegalovirus (MCMV) infections results in increased susceptibility to infection in several mouse strains. The mechanism of protection in C57Bl/6 mice is dependent on the activation of NK cells by Ly49H recognition of m157. Our previous studies have examined important residues of m157 for Ly49H recognition, as well as the contribution of m157 glycosylation to NK cell activation. However, what role the glycophosphatidyl inositol (GPI) anchor of m157 plays in Ly49H activation was unknown. Here we demonstrate that the GPI anchor of m157 regulates the surface expression of the protein. While the GPI anchor was not required for recognition of m157 by the activating or inhibitory Ly49 receptors, expression of GPI-anchored m157 resulted in greater receptor downregulation on NK cells, as well as increased NK cell cytotoxicity compared to transmembrane m157. In addition to MCMV infections, NK cells have been shown to participate in the immune response to influenza A virus (IAV). However the exact role of NK cells in IAV infection is less clear, as some studies have found NK cells to be protective, while others have shown that NK cells cause lethal immunopathology. It is likely that the severity of IAV infection may dictate the NK cell response to IAV infection (i.e. protective vs. immunopathogenic). Herein we show that NK cell accumulation in IAV-infected lungs and lung-draining lymph nodes (DLN) is regulated by the severity of IAV infection, where there is increased NK cell accumulation in the lungs during high dose IAV infection, and greater NK cell accumulation in the DLN in low dose IAV infections. Despite significant NK cell recruitment to the lung during IAV infection, as well as iv

previously published studies demonstrating the importance of NK cells to IAV immunity, NK cell depletion prior to IAV infection did not result in a significant change in morbidity or mortality. Interestingly, NK cell depletion resulted in a significantly greater number of CD4 T cells in the IAV infected lung. Further, both CD4 and CD8 T cells in NK-depleted mice showed increased IFN-γ production. Finally, while not statistically significant, NK cell depletion resulted in a trend toward greater protection from heterosubtypic IAV challenge infections. Taken together these results suggest that NK cells may either regulate the adaptive immune response to IAV infection through suppression of CD4 and CD8 T cells, or that the T cell response to IAV infection is able to compensate for the loss of NK cells. Moreover, while NK cell suppression of T cell function during a primary IAV infection does not result in increased susceptibility to primary IAV infections, NK cell regulation of adaptive immune responses may suppress the memory T cell response, and therefore leave the host more susceptible to secondary infections. Overall the studies presented herein demonstrate a complex role for NK cells in the immune response against viral infections. Ly49H + NK cells directly kill MCMVinfected cells and m157-bearing targets, but NK cell activation is regulated by ligand density, as well as the ligand membrane anchor. Additionally, NK cells suppress adaptive immune responses during a primary IAV infection, resulting in changes to the T cell response during both primary and memory responses. v

TABLE OF CONTENTS LIST OF FIGURES... viii LIST OF ABBREVIATIONS...x CHAPTER I. GENERAL INTRODUCTION...1 Natural Killer Cells...1 NK Cells in Murine Cytomegalovirus Infections...4 Glycophosphatidylinositol Anchor...5 Influenza Virus...7 NK Cells in IAV Infections...9 NK Cell Trafficking...13 Purpose of Study...15 II. THE GLYCOPHOSPHATIDYLINOSITOL ANCHOR OF THE MCMV EVASIN, M157, FACILITATES OPTIMAL CELL SURFACE EXPRESSION AND LY49 RECEPTOR RECOGNITION...21 Introduction...21 Materials and Methods...24 Mice...24 Construction of Transmembrane m157...24 Cell Lines, Retroviral Transduction, and Flow Cytometry...25 CPRG Assay for β-galactosidase Activity...26 Intracellular Cytokine Staining...26 Chromium Release Assay...27 Results...27 Expression of m157 as a Transmembrane Protein...27 Transmembrane m157 306 CD4 Activates Ly49H- and Ly49HI 129 - Expressing Reporter Cells...28 Transmembrane m157 is Less Potent in Stimulating NK Cell Cytotoxicity Compared to GPI-Anchored wt-m157 and is Associated with Impaired Ly49H Downregulation...30 Discussion...32 III. NK CELLS ARE RECRUITED TO THE IAV INFECTED LUNG VIA CXCR3...42 Introduction...42 Materials and Methods...44 Mice...44 Influenza Virus Infections...45 NK Cell Purification...45 Flow Cytometry...45 NK Cell Proliferation Assay...46 Drug Treatments...47 ELISA...47 Results...48 vi

NK Cells Accumulate in the Lung and Lung Draining Lymph Nodes During IAV Infections...48 NK Cell Accumulation in the Lung During IAV Infection is Not Due to Proliferation...50 Increased NK Cell Apoptosis in the Spleen of High Dose IAV- Infected Mice...52 NK cells are Recruited to the Lung and DLN via Chemokine Receptors...53 CXCR3 Expression on NK Cells Increases NK Cell Recruitment to the Lungs and DLN of IAV-Infected Mice...55 CCR5 is Not Required for NK Cell Recruitment to the IAV- Infected Lung...56 Discussion...58 IV. NK CELLS ARE NOT REQUIRED FOR HOST SURVIVAL DURING PRIMARY IAV INFECTIONS, BUT REGULATE THE PRIMARY T CELL RESPONSE AND INFLUENCE MEMORY CD8 T CELL DEVELOPMENT AND RESPONSES TO A HETEROSUBTYPIC IAV CHALLENGE...85 Introduction...85 Materials and Methods...88 Mice...88 Influenza Virus Infections...89 NK Cell Depletion...89 Flow Cytometry and Intracellular Cytokine Staining...89 CD8 T Cell Adoptive Transfers...90 Results...91 NK cells Suppress T Cell Cytokine Production During the Primary Immune Response to IAV...91 NK Cell Depletion Results in Significantly More CD4 T Cells in the Lung During IAV Infection...93 NK Cells do not Contribute to Protection Against Primary IAV Infections...94 NK Cell Depletion Decreases Mortality Associated with a Heterosubtypic IAV Challenge Infection...94 NK Cells do not Significantly Change the Phenotype of Memory CD8 T Cells During IAV Infection...96 Discussion...97 V. CONCLUSIONS...116 Synopsis...116 Future Directions...118 Conclusions...129 REFERENCES...132 vii

LIST OF FIGURES Figure 1. NK cell maturation, subsets, and chemokine receptor expression....18 Figure 2. Steps in GPI anchor synthesis....20 Figure 3. Expression of m157 as a transmembrane protein results in lower surface expression than GPI-anchored m157....37 Figure 4. Transmembrane m157 stimulates both activating and inhibitory Ly49 reporter cells....39 Figure 5. Transmembrane m157 expression impairs NK cell Ly49H receptor downregulation and cytotoxicity, but not IFN-γ production....41 Figure 6. Kinetics of NK cell accumulation during high and low dose IAV infections....64 Figure 7. Kinetics of NK cell distribution are similar between BALB/c and C57Bl/6 mice....66 Figure 8. Small frequency of NK cells proliferate in the lung during low and high dose IAV infections....68 Figure 9. Significantly more NK cells undergo apoptosis in the spleen during high compared to low dose IAV infections....70 Figure 10. Blockade of chemokine receptor-mediated recruitment results in decreased NK cell accumulation in the IAV-infected lung....72 Figure 11. Inhibition of chemokine receptor signaling in NK cells results in decreased accumulation in the lungs and draining lymph nodes of IAVinfected mice....74 Figure 12. CXCR3 is expressed on NK cells during IAV infection....76 Figure 13. CXCR3 expression by NK cells increases NK cell trafficking to the lung and DLN during IAV infection....78 Figure 14. CXCR3 ligands are upregulated in the lungs during IAV infection and expressed at higher levels in the lungs of high dose-infected mice....80 Figure 15. CCL5 is upregulated in the lungs during high and low dose IAV infections....82 Figure 16. While pulmonary NK cells express CCR5, blocking CCR5-mediated signaling does not affect pulmonary NK cell numbers....84 Figure 17. BALB.B6-CT6 mice express the NK1.1 epitope and their NK cells can be depleted with the anti-nk1.1 antibody clone PK136....103 viii

Figure 18. NK cell depletion does not affect the number of IAV-specific CD8 T cells....105 Figure 19. CD8 T cell effector functions are slightly increased in NK cell depleted mice....107 Figure 20. NK cell depletion results in significantly more CD4 T cells and IFN-γ + CD4 T cells in the IAV infected lung....109 Figure 21. NK cell depletion does not significantly affect morbidity or mortality....111 Figure 22. NK cell depletion during a primary IAV infection enhances survival during heterosubtypic challenge IAV infection....113 Figure 23. NK cell depletion does not affect the total number of memory CD8 T cells, but results in a trend toward increased T EM cell numbers in the lung....115 ix

LIST OF ABBREVIATIONS Ag CMV CTLD DC DLN ECTV GPI HA IAV IFN i.n. i.p. i.v. LCMV LPS mab MFI MHC MIP MPEC NK pdc PIG- PI-PLC RSV antigen cytomegalovirus C-type lectin-like domain dendritic cell draining lymph node ectromelia virus glycophosphatidylinositol hemagglutinin influenza A virus interferon intranasal intraperitoneal intravenous lymphocytic choriomeningitis virus lipopolysaccharide monoclonal antibody mean fluorescence intensity major histocompatibility complex macrophage inflammatory protein memory precursor effector cell natural killer plasmacytoid dendritic cell phosphatidylinositol glycan biosynthesis phosphatidylinositol phospholipase C respiratory syncytial virus x

SLEC T CM T EM TNF WT short-lived effector cell central memory T cell effector memory T cell tumor necrosis factor wild type xi

1 CHAPTER I: GENERAL INTRODUCTION Natural Killer Cells Natural killer (NK) cells are innate lymphocytes of the immune system. They are so named as they were found to naturally have cytotoxic properties against tumor cells without any prior priming, as required by CD8 T cells (1-3). In addition to cytotoxic properties, NK cells are also potent secretors of cytokines, notably IFN-γ, TNF, and MIP- 1α. Subsequent to initial studies, NK cells have been found to elicit responses against many types of target cells, including virus infected cells, allogeneic hematopoietic stem cells, naïve dendritic cells (DC), and activated CD8 T cells (4-8). The activation of NK cells is tightly regulated by an array of activating and inhibitory receptors, including activating receptors NKG2D, NKp46, and CD16, and inhibitory receptors KLRG1, CD94/NKG2-A, and 2B4 (9). In addition to these receptors, expressed by both human and murine NK cells, the mouse expresses the Ly49 family of lectin-like receptors, which is composed of both activating and inhibitory receptors (10). The majority of Ly49 receptors are inhibitory and recognize major histocompatibility complex (MHC) class I and class I-like molecules. Recognition of self (most often MHC class I) on a target cell by an NK cell results in an inhibitory signal. In contrast to inhibitory receptors, recognition of ligands by activating receptors results in an activation signal. The balance of inhibitory and activating signals that an NK cell simultaneously receives leads to either NK cell activation (i.e. greater activating receptor stimulation), or no activation (i.e. greater inhibitory stimuli) (9, 11). MHC class I can be downregulated by cells under stress, in transformed cells, and virally infected cells. The downregulation of self on these cells results in increased susceptibility to NK cell-mediated killing, as the balance

2 of activating and inhibitory signals received by the NK cell then favors activation (12). Downregulation of self molecules alone is not enough for NK cell activation, as NK cell activation requires signaling through an activating receptor to trigger cytokine production and cytolytic activity (13). Among the many activating receptors expressed by murine NK cells are members of the Ly49 family, including Ly49D B6, whose only known ligand is MHC class I on Chinese hamster ovary cells, as well as Ly49P MA/My and Ly49H B6, which will be discussed in more detail later in this chapter (14, 15). NK cells are generated from common lymphoid progenitors, which also give rise to T and B cells (16), and develop primarily in the bone marrow (17). During development, NK progenitors upregulate two components of the IL-15 receptor: IL-2Rβ and IL2Rγ (18, 19). The third chain of the IL-15 receptor, IL-15Rα, is expressed only on a small subset of NK cells (20). IL-15Rα expression on NK cells is not required for IL-15 receptor-mediated signaling, but is instead expressed by other cells (e.g. dendritic cells, epithelial cells) for trans-presentation of IL-15 (21). IL-15-mediated signaling through this receptor is required for both NK cell development, as well as survival in the periphery (22-24). In addition to the IL-15 receptor, as NK cell precursors develop into immature NK cells, they gain expression of several activating and inhibitory receptors. These include NKG2D, an important activating receptor for several tumor models and infections (25-27), NKp46, an activating receptor capable of binding influenza virus hemagglutinin (28), and both activating and inhibitory members of the Ly49 family (18, 29). During development NK cells also begin to express proteins that can be utilized for the identification of NK cells, including the activating receptor Nkrp1c (epitope NK1.1 in C57Bl/6 mice), and the integrin CD49b (mab clone DX5). Nkrp1c and CD49b are

3 expressed on NK cells as well as another population of innate lymphocytes- NKT cells. However, NK and NKT cells can be differentiated by their expression of CD3ε; NKT cells express CD3ε and NK cells do not. Overall, while mature NK cells are a heterogeneous population with different chemokine receptor expression, tissue distribution, and effector functions, they can generally be classified into subsets by their expression of CD56 and CD16 in humans, and CD27 and CD11b in mice and humans (Fig 1) (30). CD27 - CD11b - have been described in humans as immature NK cells with the ability to differentiate into the three following NK cell subsets (31). As the cells mature they gain CD27 expression. CD27 + CD11b - NK cells produce cytokines upon activation, but possess very little cytolytic capacity (31, 32). This subset of NK cells is most concentrated within lymph nodes and liver, but can be found in small numbers in many tissues (32, 33). Further maturation results in the upregulation of CD11b. CD27 + CD11b + NK cells are distributed throughout the body and have the most pronounced production of cytokines and release of cytolytic granules upon activation (32). In the last stage of maturation NK cells lose CD27 expression. When irradiated mice are reconstituted with bone marrow, the CD27 - CD11b + NK cells are the last to appear, and adoptively transferred CD27 - CD11b + NK cells do not differentiate into other NK cell subsets, suggesting that they are the most mature subset (34, 35). These mature NK cells are found in high concentrations within the lung (75-80% of the pulmonary NK cell population), followed by blood (~65%), and spleen (~50%) (32). CD27 - CD11b + NK cells have reduced cytokine production and cytotoxicity compared to CD27 + CD11b + NK cells. The reduced effector functions of the

4 CD27 - CD11b + may be explained by the upregulation of the inhibitory receptor KLRG1 on this NK cell subset (32, 36). NK Cells in Murine Cytomegalovirus Infections Cytomegalovirus (CMV) belongs to the betaherpes subset of the Herpesviridae family. The virus is composed of double-stranded DNA, has wide tissue tropism, strict species specificity, and can establish lytic and latent life cycles (37, 38). Although the human and murine CMV (MCMV) have evolved in separate hosts, they share 68 predicted proteins (39), and infection results in similar high viral titers within the liver and spleen (40). NK cells have been demonstrated to contribute significantly to the control of CMV infection, as both mice and humans are particularly susceptible to the infection in the absence of functional NK cells (38, 41, 42). To date, murine NK cells contribute to resistance to MCMV through four mechanisms, Cmv1-4. Cmv1 and Cmv3 are the most well characterized and both require activating Ly49 receptors: Ly49H for Cmv1, and Ly49P along with the MHC haplotype H2(K) for Cmv3 (37, 43). Cmv2 in New Zealand White mice and Cmv4 in wild-derived PWK/Pas mice are both multigenic mechanisms of viral control and are dependent on NK cells (44, 45). More recent studies have identified multiple mouse strains expressing activating Ly49 receptors that can recognize the MCMV product m04/gp36 in the context of H-2D k, including Ly49D2 in the PWK/Pas mouse strain, suggesting that Cmv4 may not be significantly different from Cmv3, which was originally described in MA/My mice (45, 46). Among the known MCMV resistance mechanisms, Cmv1 is the only one dependent on a single gene. Cmv1-mediated resistance is dependent on expression of the Ly49H activating receptor, which directly recognizes the MCMV-encoded MHC-like

5 protein m157, as depletion of Ly49H + NK cells or deletion of m157 from MCMV renders the resistant C57Bl/6 strain susceptible to MCMV infection (37). Further, when the laboratory strain of MCMV is passaged through resistant C57Bl/6 mice, escape mutations arise in m157 (47). Moreover, strains of MCMV isolated from wild mice often contain mutations in m157 that allow the virus to escape detection by Ly49H, suggesting that m157 did not evolve to interact with activating Ly49 receptors (47). Indeed, although m157 can be recognized by the activating Ly49H receptor in C57Bl/6 mice, the MHC class I-like ligand is also recognized by inhibitory Ly49s (48, 49). The inhibitory Ly49I receptor in the 129 strain recognizes m157 from laboratory strains of MCMV, while m157 in viruses isolated from wild mice can interact with inhibitory Ly49 receptors in several mouse strains, including Ly49C in C57Bl/6 mice (48, 49). These data, along with the MHC class I-like structure of m157, suggest that m157 likely evolved as a MHC decoy ligand. More specifically, while MCMV infection downregulates the host MHC to escape detection by antigen-specific CD8 T cells, the expression of a decoy MHC class I- like ligand (i.e. a negative signal) would result in NK cell recognition of MCMV-infected cells as self, allowing the virus to escape NK cell-mediated detection (50). Notably, m157 is expressed as a glycophosphatidylinositol (GPI) anchored protein (51), the importance of which will be discussed in the next section and explored in Chapter II. Glycophosphatidylinositol Anchor The GPI anchor is a glycolipid structure that is added at the C-terminus of proteins post-translationally (52). This modification allows for the protein to be inserted in the outer leaflet of the cell membrane, as opposed to transmembrane proteins, which span both leaflets of the cell membrane (53). The GPI anchor is found in several

6 eukaryotic cell types, including plant, protozoan, yeast, and animal (52). In addition to being highly conserved, GPI anchors are critical for survival, as defects in GPI biosynthesis are embryonic lethal in mammals (54). While the main purpose of the GPI anchor is to attach proteins to the cell membrane, the GPI anchor may also be important in a number of other functions, including protein localization within membrane microdomains, regulation of surface expression, and signal transduction (55-59). The GPI anchor is a post-translational modification, and GPI precursors are first synthesized in the endoplasmic reticulum before being bound to protein (53). First, N- acetylglucosamine (GlcNAc) is transferred from uridine diphosphate onto phosphatidylinositol (PI) by the glycosyltransferase GPI-GlcNAc transferase (Fig 2, step 1) (60, 61). The resulting compound, GlcNac-PI, is then de-n-acetylated by the ERmembrane phosphatidylinositol glycan biosynthesis, class L (PIG-L) protein to generate GlcN-PI (Fig 2, step 2) (53, 61). GlcN-PI is then flipped from the cytoplasmic side of the ER to the lumen where an acyl chain is added by the acyltransferase PIG-W to the inositol ring to form GlcN-(acyl)PI (Fig 2, steps 3 and 4) (53, 61). Next, two mannose (Man) groups are transferred sequentially from dolichol-phosphate-mannose (Dol-P- Man) to GlcN-(acyl)PI by GPI mannosyltransferases PIG-M and PIG-V, respectively (53, 61, 62). Transfer of the two mannose sugars generates Man-Man-GlcN-(acyl)PI (Fig 2 steps 5 and 6). Ethanolamine phosphate (EtNP) is then added to the 2-position of Man- 1 by the EtNP transferase PIG-N to form Man-(EtNP)Man-GlcN-(acyl)PI (Fig 2 step 7) (53, 61). A third mannose is transferred from Dol-P-Man by PIG-B, and an EtNP side chain is added to this mannose by PIG-O to form the mature GPI anchor precursor (Fig 2 steps 8 and 9) (53, 61). This final EtNP will make the amide bond with the C-terminus of

7 the protein (53). Proteins that are to be expressed as GPI-anchored proteins express an ω- site, the amino acid residue to which the GPI precursor will bond (53). While there is no consensus sequence for GPI anchor attachment, there are common features shared among GPI anchored proteins (53). The known ω-site amino acids are G, A, S, N, D, and C, all of which have small side-chains (63). The ω + 2 amino acid also has side chains, and there is a small hydrophilic sequence with at least 6 amino acids beginning at ω + 3 (64). Finally, there is a hydrophobic sequence at the C-terminus long enough to span the ER membrane (65). The GPI anchored protein is cleaved just before the ω-site and the GPI precursor is attached by subunits of the GPI transamidase (61). In addition to the importance of the GPI anchor in expressing host proteins within cell membranes, to date both dengue virus and MCMV have been found to encode proteins that are processed with a GPI anchor. While the importance of the GPI anchor for dengue virus non-structural protein 1 (NS1) has not been established, crosslinking of NS1 results in tyrosine phosphorylation within the host cell (66). Whether signal transduction via NS1 occurs during infection is not yet known. The role and importance of the GPI anchor in the MCMV-encoded m157 protein will be explored in Chapter II. Influenza Virus Influenza viruses are members of the orthomyxoviridae family, and are divided into three groups, A, B, and C (67). While strains from all three groups can infect humans, influenza A virus (IAV) strains infect the widest range of hosts, and will be the focus of this section. The genome of group A influenza viruses is composed of 8 segments of negative sense RNA, which encode 12 proteins (68, 69). Importantly, the segmented genome allows for new strains of IAV to emerge when segments from two or

8 more influenza viruses infect the same cell and reassort into a new virion. This phenomenon, termed antigenic shift, is what led to the emergence of the first pandemic strain of IAV of the 21 st century, 2009 H1N1, and the most recent avian IAV strain to infect humans, H7N9 (70, 71). Influenza viruses are classified by their surface protein types: hemagglutinin (HA) and neuraminadase (NA) (72). For instance, an influenza virus expressing HA 5 and NA 3 would be classified as H5N3. The HA protein of IAV is cleaved to HA 1 and HA 2 by host proteases (73). While most influenza virus strains encode HA that can be cleaved by proteases expressed only within the lung (giving the virus its tropism), some strains of influenza express HA that can be cleaved by a wider range of host proteases, including the high pathogenicity avian strain H5N1 (74). These viruses can cause systemic infections, as they are not limited by host protease expression patterns. Once cleaved by a protease, the HA subunits bind to sialic acid receptors on epithelial cells within the lung to allow the virion to enter the cell through endocytosis (75). Binding of sialic acids by HA can be species specific, as some influenza viruses preferentially bind alpha 2,6 sialic acid receptors on respiratory tract cells (human), and some bind alpha 2,3 sialic acid receptors (avian) (76). Notably, swine and some species of terrestrial birds express both alpha 2,3 and 2,6 sialic acid receptors within the respiratory tract, and can be infected with both human and avian strains of influenza (77, 78). The dual expression of alpha 2,3 and 2,6 sialic acids makes these animals an important host for virus reassortment, as cells can be infected by both alpha 2,3 and alpha 2,6 sialic acid-tropic viruses. Once HA 1 binds to a target cell, the virus is endocytosed, and the low ph within the endosome allows for a conformation change in HA 2 (75, 79). This exposes the fusion domain of HA 2, allowing the virus to fuse to the

9 endosomal membrane. The IAV M2 ion channel then acidifies the virus core, permitting the release of the viral genome and core proteins into the cytoplasm of the cell (80). The viral RNA dependent RNA polymerase proteins (PB1, PB2, and PA), nucleocapsid protein (NP), and viral RNA are bound together in a complex (the RNP), which is translocated from the cytoplasm to the cell nucleus (81). Once inside the nucleus, the negative-sense vrna must be transcribed into positive-sense RNA for both virus replication and for translation into viral proteins. In order to transcribe complementary mrna strands from viral RNA, the viral polymerase requires 5 methylguanosine caps on the negative sense viral genomic RNA. Capping of viral mrna is mediated by PB2, which snatches caps from host mrna (82, 83). Viral mrna transcripts are exported to the cytoplasm where they are translated into protein, and the newly synthesized HA, NA, and M2 proteins are localized to the plasma membrane where they are expressed on the surface (84). The new viral RNA is packaged with NP and the viral polymerase proteins (PB1, PB2, and PA). The newly associated RNP and M1 proteins are transported to the assembly site, and the newly packaged virion buds from the cell, becoming enveloped in the infected cell membrane with HA, NA, and M2. The mature virion is released from the cells when NA cleaves the sialic acid residues binding the HA of the newly formed virion to the cell surface. Once the new virion is released, host proteases can cleave HA into HA 1 and HA 2, allowing for the process to begin again (73). NK Cells in IAV Infections NK cells are activated during many different viral infections, and activation of NK effector functions is often crucial for pathogen control and protection (41, 85, 86). Several studies have examined the importance of NK cells to protection from IAV

10 infection, but the role for NK cells to immunity in IAV is not well understood. Early studies showed that NK cells within the lung have higher cytolytic activity following IAV infection (87), and are important for survival, as their depletion from hamsters or mice with anti-asialo GM 1 antibody resulted in increased morbidity and mortality compared to undepleted controls (88). This increased protection mediated by NK cells may be due, in part, to expression of the NK cell-specific activating receptor NKp46, which can recognize HA from several strains of H1 influenza (28, 89-91). In contrast, NKp46 has poor recognition of type 5 HA (90). Instead, H5 is recognized by the activating receptor NKp44 (92). Recognition of H1 by NKp46 and H5 by NKp44 results in activation of both human and mouse NK cells. (28, 89, 91-93). In addition to direct receptor activation of NK cells by HA-bearing target cells, type I interferon produced by IAV-infected epithelial cells plays an important role in NK cell activation during IAV infection, as mice deficient for the IFN-α receptor (IFNAR) showed decreased NK cell activation (94). The role for type I IFNs in NK cell activation was further shown to be direct, as IFNAR deficient NK cells were not activated after transfer into infected WT mice (94). The contribution of other inflammatory cytokines to NK cell activation and subsequent NK cell contribution to immunity against IAV is controversial. One study found that there were no deficits in NK cell activation in the absence of IL-12 or IL-18 receptors (94), while another study found that IL-18 could augment the NK cytolytic response (95). The discrepancies between these two studies could be accounted for by differences in inoculation route (i.v. vs. i.n.), source of NK cells (splenic vs. pulmonary), time after infection that NK cell activation was observed (9 hr. vs. 24-48 hr.), and measurement of activation (flow cytometry for activation markers vs. cytotoxicity assay),

11 respectively. Despite discrepancies in cytokine activation of NK cells during H1N1 IAV infection, the importance of NK cell activation to protective immunity has been illustrated in NKp46-deficient mice. NKp46 deficient mice had decreased survival compared to WT mice, similar to what was seen in NK depleted mice (88, 91). Conversely, more recent studies have observed that NK cells contribute to immunopathology (96, 97). Depletion of NK cells from mice by either anti-asialo GM 1 or anti-nk1.1 mabs resulted in decreased pro-inflammatory cytokines within the bronchoalveolar lavage fluid (BALF), decreased cell infiltration, and increased survival (96, 97). These studies used a much higher inoculum of IAV than the aforementioned studies (88, 91), which may explain the differences observed in survival. In addition to recognition of IAV infected cells, NK cells can also contribute to IAV immunity by modulating the responses of other immune cells. NK cells have been shown to be important for the development of an IAV-specific cytolytic CD8 T cell response in both humans and mice, as CD8 T cells primed in the absence of NK cells were less cytotoxic against IAV infected cells (98). Other studies have shown that NK cells are crucial for the migration of CD8 T cells to the lung-draining lymph node (DLN) during IAV infection, and that this T cell recruitment is dependent on NK cell IFNγ production (99). In addition to enhancing the CD8 T cell responses, NK cells also contribute to the DC response to IAV (99). In the absence of NK cells, antigen uptake by DC is reduced, and fewer DC traffic to the DLN. DC uptake of antigen (Ag) and trafficking to DLN is dependent on NK cell production of perforin and IFN-γ (99). A few recent studies have begun to examine how to enhance the NK cell response during IAV infection in order to boost immunity (100, 101). IFN-γ seems to be a critical

12 facilitator for NK cell-mediated protection from IAV. Mice treated with soluble T. gondii antigens have increased survival during H5N1 infections. The protection was found to be dependent on NK cells, which produced more IFN-γ than NK cells in untreated control mice (100). IFN-γ treatment has also been shown to enhance survival in NK-sufficient mice, but not NK-depleted mice during IAV infection (101). IFN-γ treatment enhanced NK cell numbers within the blood, spleen, bone marrow, and lung, resulted in increased NK cell proliferation, and enhanced both IFN-γ production and cytotoxicity by NK cells (101). Despite evidence that NK cells are protective in IAV infections, IAV utilizes several mechanisms to escape detection by NK cells (102-106). While, as previously mentioned, the NK cell receptor NKp46 can recognize HA, mutations in HA have been observed, allowing IAV to escape NKp46-mediated detection (105). Further, overexposure of NK cells to HA can also lead to impaired NK cell functions, as HA can cause downregulation of the CD3 ζ chain, a protein required for NKp46-mediated signaling (103). In addition to affecting the activating NKp46 receptor, IAV can also modulate NK cell inhibitory receptors. While several other viruses downregulate MHC class I as a means to escape CD8 T cell detection, which leaves the cells vulnerable to NK cell detection, IAV-infected cells do not show decreased MHC class I expression in vitro (106). Instead, MHC class I proteins are reorganized into lipid raft microdomains, resulting in enhanced interactions with inhibitory ligands on NK cells and less effective killing of IAV infected cells (102). Finally, IAV can escape NK cell-mediated killing through direct infection of NK cells (104). While infection of NK cells does not result in

13 productive IAV replication, it does induce NK cell apoptosis, effectively reducing the anti-iav immune response (104). Taken together, these studies suggest that the role for NK cells in IAV immunity is complex, and may depend on multiple factors including the influenza HA subtype, dose of infection, and production of pro-inflammatory cytokines. The contribution of NK cells to IAV immunity is explored in Chapters III and IV. NK Cell Trafficking NK cells express a wide variety of chemokine receptors, and the expression of some of these receptors is dependent on the maturation state of NK cells, contributing to the differential tissue distribution discussed previously (33). More immature NK cells (CD27 + CD11b - ) express CXCR3 and CXCR6, which can aid in lymph node and liver localization (Fig 1) (33). As NK cells mature (CD27 - CD11b + ), CXCR3 expression decreases, and cells lose CXCR6 expression but gain CX 3 CR1 and S1P5 expression, the latter of which aids in egress from the bone marrow and lymph nodes (Fig 1) (33, 107, 108). In addition to these specialized receptors, other chemokine receptors, including CCR2 and CCR5, are expressed on all NK subsets found in the periphery (33) and are important for NK cell trafficking to sites of infection, as discussed below. Likewise, chemokine receptors have an important role in homeostatic NK cell trafficking (33). CXCR4 is expressed on immature NK cells within the bone marrow, and must be desensitized for NK cell egress from the bone marrow (108). S1P5 is also important for NK cell egress from bone marrow as well as lymph nodes, as S1P5 deficient mice have an increased number of NK cells within these organs, and decreased NK cell numbers in the blood, spleen, and lungs (107, 109). Once NK cells have entered the periphery,

14 individual chemokine receptors direct them to different organs. CXCR6 is important for NK cell accumulation within the liver, while CXCR3 is important for NK cell recruitment to the naïve lung (110, 111). While studies performed in naïve mice determine how NK cells migrate during homeostasis and give us insight into what chemokine receptors are important for NK cell migration to certain tissues, evidence from different infection models suggests that NK cell recruitment to specific organs may be more complex during infection, and may require different chemokine receptors depending on the infection (107-118). The importance of NK cell trafficking during infection has been illustrated in several infection models (112-118). NK cell trafficking, when halted, can result in higher pathogen burden and/or increased susceptibility (112-114, 116). CCR5 is important for recruiting NK cells to the liver in both Toxoplasma gondii and MCMV infections (113, 116). In T. gondii infections, NK cells traffic to the liver where they produce IFN-γ and protect from infection. Reduced NK cell migration to the liver in CCR5 -/- mice results in lower levels of IFN-γ, higher parasite burden, and increased morbidity (113). This increase in susceptibility was reduced upon transfer of CCR5 sufficient NK cells, illustrating the importance of NK cell trafficking during infection (113). Similarly, NK cells recruited to the liver via CCR5 during MCMV infection produce MIP-1α and IFN-γ to create a pro-inflammatory, anti-viral environment, resulting in reduced viral titers (116). In addition to recruitment to the site of infection, NK cell recruitment to lymphoid tissues can be critical for host resistance to infection. CXCR3 has been shown to be important for NK cell recruitment to the DLN after stimulation with mature DC (119),

15 and cowpox infection (115). While the latter study did not examine the effects of NK cell trafficking on survival, studies in aged mice infected with ectromelia virus (ECTV) have shown the importance of NK cell recruitment to the DLN for increasing survival during infectious disease (112). Reduced trafficking of NK cells due to age resulted in increased susceptibility to ECTV due to a defective CD8 T cell response. Transfer of NK cells from younger mice restored NK cell trafficking to the lymph nodes during ECTV infection, and restored the CD8 T cell response, resulting in increased resistance to ECTV in aged mice (120). Interestingly, NK cells are most concentrated in mucosal tissue, including the lung, which is the primary organ infected by IAV (33). Upon pulmonary infections, NK cells further accumulate within the lung (91). CXCR3 is important in recruiting NK cells to the naïve lung, however it has also been shown to play a role in NK cell recruitment to the lung during Bordetella bronchiseptica infection (118). During Aspergillus fumigatus infections, CCR2 has instead been identified as important for NK cell recruitment to the lung (114). Interestingly, CCR2 deficiency results in decreased NK cell numbers within the BALF of IAV-infected mice, but not within the lung parenchyma, illustrating that the mechanism of NK cell recruitment to infected organs may depend on the infectious agent (117). NK cell trafficking and recruitment during IAV infection will be further discussed and examined in Chapter III. Purpose of Study NK cells play a critical role in the immune response against several different viruses, most notably those of the herpesviridae family. NK cell-mediated protection against MCMV is dependent on the interaction between Ly49H on NK cells and m157 on

16 MCMV infected cells. Previous studies have examined key amino acids within m157 that allow for recognition by Ly49H, and the contribution of m157 glycosylation to the Ly49H-m157 interaction to better understand the parameters of NK cell ligand-receptor interactions to NK cell activation. Interestingly, m157 is one of the two known viral proteins attached to the surface by a GPI anchor (51, 66), but the importance of this attachment to the interaction with NK cell receptors is not known. While the contribution of NK cells to immunity against herpesvirses has been well described, the contribution of NK cells to immunity against other viruses is not well understood. NK cells are more concentrated within the lung than any other organ, suggesting that NK cells could play an important role against pulmonary pathogens. Indeed, several studies have highlighted the importance of NK cells in the immune response against IAV, although it is unclear if NK cells contribute to protection or immunopathology. The purpose if this dissertation is to answer the following questions: 1.) Does the GPI anchor of m157 contribute to inhibitory and activating Ly49 receptor binding? 2.) What is the mechanism of NK cell trafficking during IAV infection and does it differ between high and low dose infections? 3.) How do NK cells contribute to the development of the IAV-specific CD8 T cell responses during IAV infections?

Figure 1. NK cell maturation, subsets, and chemokine receptor expression. In the murine system, NK cell subsets can be differentiated by CD11b and CD27 expression. (A) Immature NK cells are CD27 + CD11b - and uniquely express CXCR6. As NK cells mature they gain CD11b and become (B) CD27 + CD11b +. This subset loses CXCR6 expression, but begins to express S1P5. Finally NK cells downregulate CD27 to become (C) CD27 - CD11b +. Notably, this subset has low CXCR3 expression, but has the highest expression for CX 3 CR1 and S1P5. 17

Adapted from Gregoire et. al. 2007 18

Figure 2. Steps in GPI anchor synthesis. The GPI anchor is synthesized prior to addition to protein. (1) N-acetylglucosamine is added onto phosphatidylinositol (PI) on the cytoplasmic side of the endoplasmic reticulum (ER) (GlcNac-PI), and is then (2) de-nacetylated (GlcN-PI). (3) GlcN-PI is flipped to the lumen of the ER and (4) an acyl chain is added (GlcN-(acyl)PI). One mannose group is added in steps (5) and (6) (Man-Man- GlcN-(acyl)PI). (7) Ethanolamine phosphate (EtNP) is added to the first mannose group (Man-(EtNP)Man-GlcN-(acyl)PI). (8) A third mannose group is added, and (9) an EtNP sidechain is sequentially added ((EtNP)Man-Man-(EtNP)Man-GlcN-(acyl)PI). The GPI anchor is then fully mature, and can be further modified, or bound to a protein (10). 19

Adapted from Kinoshita et. al. 2008 20

21 CHAPTER II: THE GLYCOPHOSPHATIDYLINOSITOL ANCHOR OF THE MCMV EVASIN, M157, FACILITATES OPTIMAL CELL SURFACE EXPRESSION AND LY49 RECEPTOR RECOGNITION Introduction Natural killer (NK) cells are innate cytotoxic lymphocytes that participate in the immune responses against a wide variety of microbial pathogens, and also display potent anti-tumor responses (121-126). The activation of NK cells is tightly regulated by an array of activating and inhibitory receptors expressed on their surface, including those of the Ly49 family in rodents (encoded by Klra genes) (127). Inhibitory Ly49 receptors recognize MHC class I and class I-like ligands in both trans and cis, and control the activation threshold for NK cells at the level of the NK cell immune synapse (128-131). Further, there is promiscuity among the Ly49 receptors for multiple ligands, with a hierarchy of affinities and corresponding inhibitory potency (127, 132, 133). The activating Ly49 receptors are both fewer in number and, to date, show a reduced ligand complexity compared to the inhibitory Ly49 receptors. Prominent among the activating Ly49 receptor ligands is the m157 product of murine cytomegalovirus (MCMV), a cell surface glycoprotein attached to the surface of MCMV-infected cells through a glycophosphatidylinositol (GPI) anchor (48, 51). Like MHC class I, m157 is polymorphic. While m157 expressed by laboratory MCMV strains engages the activating Ly49H receptor of C57BL/6 mice (Ly49H B6 ), forming the molecular basis of the potent Cmv-1 resistance trait in this mouse strain, it also is a cognate ligand for inhibitory Ly49 receptors, including Ly49I from 129 mice (Ly49I 129 ) (48, 134, 135). In wild strains of MCMV, additional m157 variants have been identified

22 that do not engage Ly49H B6 or Ly49I 129, but are ligands for other inhibitory receptors from various mouse strains, evidence that m157 arose as a decoy ligand, or viral evasin, by exploiting the inhibitory Ly49 receptor-self MHC class I interaction (48, 49, 136). Notably, as the GPI addition site is highly conserved, all of these m157 variants are predicted to be GPI-associated proteins, raising the possibility that the GPI anchor supports a critical function for m157. GPI-anchored proteins penetrate only a single layer of the cell membrane, and are usually found in lipid raft microdomains rich in cholesterol and sphingomyelin (52, 137). In contrast, transmembrane proteins span both layers of the cell membrane and are typically excluded from lipid rafts (138, 139). While the presence of a GPI anchor for a virally encoded protein is rare, and thus a unique feature for m157, endogenous GPIassociated proteins are well established ligands for other NK cell receptors most notably NKG2D, an activating lectin-like receptor expressed on all NK cells and a subset of T cells in humans and mice (140-143). NKG2D ligands are a diverse group of cell surface transmembrane and GPI-associated proteins whose expression is normally restricted developmentally, among discrete tissues, or increased in response to viral infection or other cell stresses, increasing cell susceptibility to NK cell cytotoxicity (144-146). Notably, among the retinoic acid early transcript (Raet) gene families of mice (including Rae1α ε, H60a-c, and MULT-1), and humans (RAET1E, G, and L, and UL16-binding proteins, ULBP1-3), investigators have examined the role of GPI anchors for NKG2D recognition and activation. For example, H60c is GPI-linked, restricted to the skin, and shows the lowest affinity for mouse NKG2D among the H60 proteins; yet, H60c has comparable potency to transmembrane H60a and H60b in activating NK cell

23 cytotoxicity when expressed in BaF3-transduced targets (147). Independent, and more direct examinations of the role of the GPI anchor for the human NKG2D ligands ULBP-1 and ULBP-2 revealed curiously discordant findings. While ULBP-1 may be expressed stably at the cell surface as a transmembrane fusion protein, it is significantly less potent in activating NK cell cytotoxicity, and the authors concluded that NKG2D ligand distribution within the membrane influences the NK cell-target cell interaction (148). Although ULBP1-3 are normally GPI-anchored proteins, ULBP2 is unique in that a minor fraction exists on the surface as a transmembrane protein. Although the transmembrane ULBP2 isoform shows delayed protein maturation and ultimately is expressed at a lower cell surface density compared to GPI-anchored ULBP2, the potency in activating NKG2D-mediated cytotoxicity is essentially equivalent (149). The activating Ly49H receptor is a member of the lectin-like NK cell receptor family, but unlike NKG2D, Ly49H has only a single known ligand MCMV m157. That inhibitory Ly49 receptors recognize endogenous MHC class I ligands, and the observation that m157 and its wild MCMV strain variants engage multiple inhibitory Ly49 receptors, has led to the idea that MCMV uses m157 as a decoy ligand to evade NK cell mediated immunity, and that Ly49H evolved as a means to defeat this viral evasion mechanism (48, 49). Our previous work has revealed that the Ly49H-m157 interaction is unique in comparison to other related lectin-like receptors and their ligands. Ly49H recognition of m157 is absolutely dependent on m157 residues Ile153 and Lys161; residues that lie outside of the prototypical sites 1 and 2 for inhibitory Ly49 receptor recognition of MHC class I ligands (150). In addition, N-glycosylation on m157 stabilizes the receptor-ligand interaction, which may be important for prolonged signaling

24 at low ligand densities for m157 on MCMV-infected cells (151). We wished to extend this analysis by examining the contribution of the GPI anchor to m157 expression and its capacity to functionally engage cognate Ly49 receptors. Herein, we report that MCMV m157 may be expressed as a transmembrane fusion protein, albeit at a reduced cell surface density, and that transmembrane m157 is recognized by both Ly49H B6 and Ly49I 129. However, the potency of transmembrane m157 in activating NK cell cytotoxicity, in particular, is reduced, suggesting that the GPI-anchor of m157 is important for robust surface expression and likely contributes to a stronger and/or prolonged signaling interaction with Ly49H + NK cells. Materials and Methods Mice C57Bl/6 and B6.129S7-Rag1 tmmom /J (RAG1-deficient) mice, used as a source for CD4 DNA or fresh splenic NK cells, respectively, were maintained in the University of Iowa Animal Care Unit barrier facility. All mouse experiments were conducted according to protocols approved by the University of Iowa Institutional Animal Care and Use Committee. Care was taken to minimize animal suffering. Construction of Transmembrane m157 RNA was harvested from spleens of C57Bl/6 mice using TRIzol, per manufacturer s instructions. CD4 cdna was prepared by RT-PCR using the following primers: 5 - CTCAGAATTCCCAACCAACAAG-3, 5 - GAGGGAAACCGTGGCCGGTTGTG-3, 5 -GTCAAGATGGACTCCAGGATC-3,