Rational Design of Live Attenuated Respiratory Syncytial Vaccines by Inhibiting

mRNA Cap Methylation and RNA N6-methyladenosine

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the

Graduate School of The Ohio State University

By

Miaoge Xue, M.S.

Graduate Program in Comparative and Veterinary Medicine

The Ohio State University

2020

Dissertation Committee:

Jianrong Li, DVM, PhD, Advisor

Prosper N. Boyaka, PhD

Mark E. Peeples, PhD

Shan-Lu Liu, MD, PhD

Copyrighted by

Miaoge Xue

2020

2

Abstract

Human respiratory syncytial virus (RSV), a member of family , is the leading causative agent of respiratory tract in , with high morbidity and mortality seen in infants, children, the elderly, and immunocompromised individuals. The clinical signs and symptoms associated with RSV range from mild respiratory problems to severe coughs, , and . Despite the enormous economic losses and emotional burdens these cause, vaccines and anti-viral drugs are currently not available. Since the first discovery of

RSV in 1956, approaches to generate vaccines employing viral proteins or inactivated vaccines have failed either due to a lack of immunogenicity or the potential for causing enhanced pulmonary disease upon natural infection with the same virus.

A live attenuated vaccine is one of the most promising vaccine candidates for RSV. However, it has been a challenge to identify an RSV vaccine strain that has an optimal balance between attenuation and immunogenicity. Many approaches to develop live RSV vaccine candidates either affect immunogenicity, genetic stability, or lead to insufficient attenuation or poor virus growth in vitro. In addition, RSV encodes two nonstructural NS1 and NS2 proteins which suppress innate immune response, and the subsequent adaptive immune response. Therefore, identification of new targets for attenuating RSV and exploration of new approaches to improve the innate and adaptive immune responses of live attenuated vaccine candidates are urgently needed. I hypothesize that viral mRNA cap methylation and RNA N6-methyladenosine (m6A) are excellent targets to ii rationally attenuate RSV for development of live attenuated vaccines. The rationale for this hypothesis is that mRNA cap methylation and RNA m6A methylation is essential for RNA

stability, , and innate immune evasion.

In Chapter 2, I generated three live attenuated RSV vaccine candidates by alteration of S-

adenosylmethionine (SAM) binding site in the methyltransferase (MTase) region of the large (L)

polymerase protein of RSV. Recombinant rgRSVs carrying a single (G1853A or G1857A) or a

double mutation (G1857A-G1853A) in the SAM binding site were genetically stable and grew to

high titers in cultured cells. All three recombinant viruses were highly attenuated for replication in

primary, well differentiated, human bronchial epithelial (HBE) cultures, and upper and lower

respiratory tracts of cotton rats. All three attenuated strains elicited comparable levels of

neutralizing antibody to the parental rgRSV in cotton rats. Despite high attenuation in vitro and in

vivo, all immunized cotton rats were completely protected against RSV infection in both lungs and

nasal turbinates. Taken together, mutations in the SAM binding site of L protein represent a

promising approach to generate live attenuated vaccine candidates for RSV.

In Chapter 3, I aimed to enhance the genetic stability of live attenuated vaccines. I inserted or

deleted amino acid in the flexible hinge region between mRNA capping region and a cap

methyltransferase region in the L protein and found that the flexible hinge region of RSV L protein

is tolerant to amino acid deletion or insertion. Recombinant RSVs carrying a single and double

deletion or a single alanine insertion were highly defective in replication and spread in vitro, in the

ex vivo lung model (human bronchial epithelial cultures) and in cotton rats. Importantly, these

recombinant viruses elicited high levels of neutralizing antibody and provided complete protection

against RSV replication. Therefore, amino acid deletions or insertions in the hinge region of the L

iii

protein can serve as a novel approach to rationally design genetically stable, highly attenuated and

immunogenic live vaccine candidates for RSV.

In Chapter 4, I discovered viral RNA internal m6A methylation upregulates RSV replication and

pathogenesis and identify viral m6A methylation as a target for rational design of live attenuated

vaccine candidates for RSV. m6A is the most prevalent internal modification of mRNAs in most

eukaryotes and many viral . I found that RSV RNAs are modified by m6A within discreet

regions and that these modifications enhance and pathogenesis. The G gene

transcript contains the most m6A modifications. Recombinant RSV variants expressing G

transcripts that lack particular clusters of m6A display reduced replication in A549 cells, primary

well differentiated human airway epithelial cultures, and respiratory tracts of cotton rats. One of

the m6A-deficient variants is highly attenuated yet retains high immunogenicity in cotton rats.

In Chapter 5, I aimed to enhance the innate and adaptive immunity of live attenuated RSV

vaccines. I found that m6A-deficient RSV virion RNA induced significantly higher type I

interferon (IFN) compared to m6A-suficient virion RNA. Mechanistically, m6A-deficient RSV

virion RNA induces higher expression of RIG-I, binds more efficiently to RIG-I, enhances RIG-I

ubiquitination and IRF3 phosphorylation, leading to enhanced RIG-I mediated IFN signaling.

Subsequently, we inactivated the m6A sites in the G gene of and/or antigenome using

synonymous mutations, and the resultant recombinant RSVs were defective in m6A methylation,

attenuated in vitro and in vivo, and induced a higher neutralizing antibody and T cell immune

responses in mice. In addition, animals immunized with these m6A-deficient RSVs were

completely protected from RSV challenge. The results suggest that inhibition of viral m6A

iv methylation may be a novel approach to enhance innate and adaptive immunity against RSV infection.

In summary, I have developed a panel of live attenuated RSV vaccine candidates by inhibiting viral mRNA cap methylation or RNA internal m6A methylation. These live attenuated vaccine candidates are genetically stable, sufficiently attenuated in vitro and in vivo, and highly immunogenic in animal models. These studies will not only contribute to our understanding of

RSV replication and , but will also lead to a new approach to attenuate RSV and the development of a panel of new, improved live vaccine candidates for RSV which will, in the future, lead directly to trials in nonhuman primates.

v

Dedication

Dedicated to my beloved parents Xiaowen Xue, and Debi Fu

and all my supportive friends

for their unconditional love during this journey

vi

Acknowledgments

First, I owe my deepest gratitude to my advisor, Dr. Jianrong Li, for offering me such a great opportunity to pursue this journey in his lab. His guidance, patience and encouragement were of immense help for me. More importantly, his hardworking and passion to science made him such a good model for me and will be cherished for the rest of my life.

I am also grateful to my committee members Drs. Shan-Lu Liu, Mark Peeples and Prosper Boyaka, to their time and constructive comments on my committee meetings, candidacy exam and projects overall. Their knowledge in virology and immunology has expanded my perspectives in research and cultivated my critical thinking.

I would also like to extend a sincere thanks to all of the members of the Li lab, past and present, whom have supported me and always available for my laugh and cry. Drs. Yuanmei Ma and Yu

Zhang had helped me out at the first few days when I came here, I will never forget the night that we ran away together from the gas station in downtown Columbus. Many thanks to Dr. Mijia Lu,

Xueya liang, Yunjian Lu, Dr. Deng Pan and Dr. Erin DiCaprio for their thoughtful suggestions and discussions on projects as well as my graduate life. I am also grateful for all of the faculty, staff, and students of the Department of Veterinary Biosciences that have contributed to my success. More specifically people in Dr. Stefan Niewiesk’s lab: Olivia Hard and Devra Huey for their direct support with my project.

vii

Finally, to my parents and family, thank you for your love and support. In particular, I desire to thank my husband, Anzhong Li, who is both the best colleague in lab and at home.

viii

Vita

2008………………………………… Bashu High School, Chongqing, China

2012…………………………...... B.S., Biology, Xiamen University, Fujian, China

2015…………………………………. M.S., Biochemistry and Molecular Biology, Xiamen

University, Fujian, China

2016-2020…………………………… Ph.D. Student, Veterinary Biosciences, The Ohio State

University, Columbus, Ohio

Publications

1. Lu M, Zhang Z, Xue M, Zhao BS, Harder O, Li A, Liang X, Gao TZ, Xu Y, Zhou J, Feng Z, Niewiesk S, Peeples ME, He C, Li J. N6-methyladenosine modification enables viral RNA to escape recognition by RNA sensor RIG-I. Nat Microbiol. 2020 Feb3;.doi: 10.1038/s41564-019-0653-9. PubMed PMID: 32015498. 2. Xue M, Zhao BS, Zhang Z, Lu M, Harder O, Chen P, Lu Z, Li A, Ma Y, Xu Y, Liang X, Zhou J, Niewiesk S, Peeples ME, He C, Li J. Viral N6-methyladenosine upregulates replication and pathogenesis of human respiratory syncytial virus.Nature Communications. 2019 Oct 9;10(1):4595. doi: 10.1038/s41467-019-12504-y. PubMed PMID: 31597913. 3. Craig K, Dai X, Li A, Lu M, Xue M, Rosas L, Gao TZ, Niehaus A, Jennings R, Li J. A Lactic Acid Bacteria (LAB)-Based Vaccine Candidate for Human . Viruses. 2019 Mar 2;11(3). doi: 10.3390/v11030213. PubMed PMID: 30832363; PubMed Central PMCID: PMC6466309. 4. Li A, Yu J, Lu M, Ma Y, Attia Z, Shan C, Xue M, Liang X, Craig K, Makadiya N, He JJ, Jennings R, Shi PY, Peeples ME, Liu SL, Boyaka PN, Li J. A Zika virus vaccine expressing premembrane-envelope-NS1 polyprotein. Nature Communications. 2018 Aug

ix

3;9(1):3067. doi: 10.1038/s41467-018-05276-4. PubMed PMID: 30076287; PubMed Central PMCID: PMC6076265. 5. Li Y, Xue M, Yu L, Luo G, Yang H, Jia L, Zeng Y, Li T, Ge S, Xia N. Expression and characterization of a novel truncated VP4 for the development of a recombinant rotavirus vaccine. Vaccine. 2018 Apr 12;36(16):2086-2092. doi: 10.1016/j.vaccine.2018.03.011. Epub 2018 Mar 16. PubMed PMID: 29555220. 6. Xue M, Yu L, Jia L, Li Y, Zeng Y, Li T, Ge S, Xia N. Immunogenicity and protective efficacy of rotavirus VP8* fused to cholera toxin B subunit in a mouse model. Human Vaccine Immunotherapy. 2016 Nov;12(11):2959-2968. doi: 10.1080/21645515.2016.1204501. Epub 2016 Jul 19. PubMed PMID: 27435429; PubMed Central PMCID: PMC5137547. 7. Xue M, Yu L, Che Y, Lin H, Zeng Y, Fang M, Li T, Ge S, Xia N. Characterization and protective efficacy in an animal model of a novel truncated rotavirus VP8 subunit parenteral vaccine candidate. Vaccine. 2015 May 21;33(22):2606-13. doi: 10.1016/j.vaccine.2015.03.068. Epub 2015 Apr 14. PubMed PMID: 25882173. 8. Li T, Lin H, Yu L, Xue M, Ge S, Zhao Q, Zhang J, Xia N. Development of an enzyme- linked immunospot assay for determination of rotavirus infectivity. Journal of Virological Methods. 2014 Dec; 209:7-14. doi: 10.1016/j.jviromet.2014.08.012. Epub 2014 Aug 27. PubMed PMID: 25172048. 9. Zhang S, Lin Y, Wang J, Wang P, Chen J, Xue M, He S, Zhou W, Xu F, Liu P, Chen P, Ge S, Xia N. A convenient nucleic acid test on the basis of the capillary convective PCR for the on-site detection of 71. The Journal of molecular diagnostics. 2014 Jul;16(4):452-8. doi: 10.1016/j.jmoldx.2014.04.002. Epub 2014 May 22. PubMed PMID: 24858492. 10. Zhang S, Xue M, Zhang J, Chen Q, Chen J, Wang Z, Zhou W, Chen P, Xia N, Ge S. A one-step dipstick assay for the on-site detection of nucleic acid. Clin Biochem. 2013 Dec;46(18):1852-6. doi: 10.1016/j. Clinical biochemistry. 2013.10.013. Epub 2013 Oct 22. PubMed PMID: 24161476. 11. Miaoge Xue, Rongzhang Wang, Olivia Harder, Phylip Chen, Mijia Lu, Hui Cai, Anzhong Li, Xueya Liang, Ryan Jennings, Krista La Perle, Stefan Niewiesk, Mark Peeples, Jianrong Li. Sable attenuation of human respiratory syncytial virus for live vaccines by deletion and insertion of amino acids in the hinge region between the mRNA capping and methyltransferase domains of the large polymerase protein. Journal of Virology. 2019. In Revision. 12. Anzhong Li, Miaoge Xue, Mijia Lu, Zayed Attia, Jingyou Yu, Chao Shan, Xueya Liang, Ryan Jennings, Pei-Yong Shi, Shan-Lu Liu, Prosper N Boyaka, Stefan Niewiesk, Jianrong Li*. Vesicular stomatitis virus expressing NS1 protein of Zika virus provides partial protection against Zika virus infection in mice. Journal of Virology, 2019. Accepted. 13. Miaoge Xue, Rongzhang Wang, Yu Zhang, Mijia Lu, Xueya Liang, Mark Peeples, Stefan Niewiesk, Jianrong Li*. Rational design of human respiratory syncytial virus live attenuated vaccines by inhibiting viral mRNA cap methyltransferases. Journal of Virology, 2019. In Revision.

x

14. Miaoge Xue, Mijia Lu, Xueya Liang, Mark Peeples, Stefan Niewiesk, Jianrong Li*. Rational design of human respiratory syncytial virus live attenuated vaccines by alteration of the putative RNA binding site in large polymerase protein. Manuscript in Preparation. 15. Miaoge Xue, Mijia Lu, Anzhong Li, Zijie Zhang, Chuan He, Jianrong Li*. Non-segmented negative-strand RNA viruses utilize m6A as a means to escape innate immune surveillance. Manuscript in Preparation.

Fields of Study

Comparative and Veterinary Medicine

xi

Table of Contents

Abstract ...... ii Dedication ...... vi Acknowledgments...... vii Vita ...... ix List of Tables ...... xiv List of Figures ...... xv Chapter 1. Literature review ...... 1 1.1. Origin, pathogenesis, and clinical and epidemiologic features of human respiratory syncytial virus ...... 1 1.2. Virology of RSV ...... 5 1.3. Interferon signaling ...... 29 1.4. Host innate immunity against RSV infection ...... 31 1.5. Antiviral drug against RSV ...... 32 1.6. RSV vaccine development ...... 34 1.7. Introduction to internal RNA methylation ...... 40 Chapter 2. Rational design of human respiratory syncytial virus live attenuated vaccine candidates by alteration of the S-adenosylmethionine (SAM) binding site in the large polymerase protein ...... 55 2.1 Abstract ...... 55 2.2 Introduction ...... 56 2.3 Materials and Methods ...... 57 2.4 Results ...... 64 2.5 Discussion ...... 88 Chapter 3: Stable attenuation of human respiratory syncytial virus for live vaccines by deletion and insertion of amino acids in the hinge region between the mRNA capping and methyltransferase domains of the large polymerase protein...... 94 3.1 Abstract ...... 94 3.2 Introduction ...... 94 3.3 Materials and Methods ...... 95 3.4 Results ...... 100

xii

3.5 Discussion ...... 124 Chapter 4. Viral N6-methyladenosine upregulates replication and pathogenesis of human respiratory syncytial virus ...... 130 4.1 Abstract ...... 130 4.2 Introduction ...... 130 4.3 Materials and Methods ...... 133 4.4 Results ...... 141 4.5 Discussion ...... 181 Chapter 5. Enhanced innate and adaptive immune responses of human respiratory syncytial virus by inhibiting viral N6-methyladenosine modification ...... 187 5.1 Abstract ...... 187 5.2 Introduction ...... 188 5.3 Materials and Methods ...... 190 5.4 Results ...... 196 5.5 Discussion ...... 233 Chapter 6. Summary and future directions ...... 236 6.1. Summary ...... 236 6.2. Future directions ...... 238 References ...... 242

xiii

List of Tables

Table 1.1. Agents in development for treatment of RSV ...... 34 Table 1.2. Recombinant live-attenuated RSV vaccines...... 37 Table 1.3. The regulators of m6A in different viruses...... 48 Table 1.4. Comparison of m6A functions in HIV-1 life cycle...... 50 Table 2.1. Replication of rgRSV carrying mutations in the SAM binding site in cotton rats ...... 82 Table 2.2. Immunogenicity of rgRSV mutations in cotton rats in short term study ...... 86 Table 2.3. Immunogenicity of rgRSV mutations in cotton rats in long term study ...... 88

xiv

List of Figures

Figure 1.1. Phylogeny of the family Pneumoviridae...... 2 Figure 1.2. Human respiratory syncytial virus virion...... 6 Figure 1.3. Overview of RSV and RNA replication...... 8 Figure 1.4. Structure of VSV-L...... 13 Figure 1.5. Structure of the RSV L Core Bound to Tetrameric P...... 15 Figure 1.6. A schematic view of RSV life cycle...... 17 Figure 1.7. Binding and entry of RSV into the host cell...... 18 Figure 1.8. Fusion process between the RSV envelope and cellular membrane...... 20 Figure 1.9. Nucleotide positions in the leader region of genomic RNA that are important for transcription (top) and RNA replication (bottom)...... 23 Figure 1.10. Model for the RSV assembly and budding...... 24 Figure 1.11. Comparison of mRNA cap formation in eukaryotic cells and VSV...... 27 Figure 1.12. Proposed model for mRNA cap methylation in VSV and eukaryotic cells...... 27 Figure 1.13. The signaling pathways of type I interferon...... 30 Figure 1.14. A schematic view of the structure and localization of methylated nucleosides in eukaryotic mRNA...... 41 Figure 1.15. Diverse molecular functions of m6A...... 42 Figure 2.1 Examination of the function of L mutants using a minigenome assay...... 66 Figure 2.2. Characterization of rgRSVs carrying mutations in the SAM binding site of the L protein...... 70 Figure 2.3. Viral protein expression by rgRSV mutants...... 72 Figure 2.4. Interferon β (IFN-β) production of rgRSV mutants in A549 cells...... 73 Figure 2.5. rgRSV mutants spread more slowly in HBE cultures...... 75 Figure 2.6. rgRSV production in HBE cultures. HBE cultures were inoculated with 400 TCID50 of each rgRSV...... 76 Figure 2.7. Cytokine production in RSV mutant infected HBE cultures...... 79 Figure 2.8. Neutralizing antibody responses of rgRSV mutants in cotton rats in a short-term immunization experiment...... 85 Figure 2.9. Neutralizing antibody responses of rgRSV mutants in cotton rats in a long-term immunization experiment...... 87 Figure 3.1. Design of RSV L deletion and insertion mutants...... 102 Figure 3.2. Examination of the function of L deletion and insertion mutants using a minigenome assay...... 103 Figure 3.3. Characterization of rgRSVs carrying deletion and insertion in the flexible hinge region of the L protein...... 106 xv

Figure 3.4. F and G protein expression...... 108 Figure 3.5. rgRSV production in HBE cultures...... 111 Figure 3.6. rgRSV mutants spread more slowly in HBE culture...... 113 Figure 3.7. Interferons, IP-10 and IL-6 production in RSV mutant-infected HBE cultures...... 116 Figure 3.8. Replication of RSV mutant in cotton rats...... 120 Figure 3.9. Immunogenicity of rgRSV mutants in cotton rats...... 123 Figure 4.1. The RSV genome and antigenome/mRNAs are m6A methylated...... 144 Figure 4.2. RSV infection alters the methylome of host transcripts in HeLa cells...... 147 Figure 4.3. RSV infection alters the methylome of host transcripts in A549 cells...... 149 Figure 4.4. YTHDF1, 2, 3 (reader) proteins promote RSV replication, gene expression, and progeny virus production...... 153 Figure 4.5. Knockdown of endogenous YTHDF1, 2, 3 (reader) proteins diminishes RSV gene expression...... 156 Figure 4.6. Effects of m6A writer proteins on RSV gene expression...... 159 Figure 4.7. Effects of m6A eraser proteins on RSV gene expression...... 162 Figure 4.8. RSV infection does not alter the m6A reader, writer, or eraser protein distribution in cells...... 165 Figure 4.9. m6A reader protein binds to RSV genomic RNA and mRNA...... 167 Figure 4.10. m6A-abrogating RSV mutants have defects in replication in immortalized cells. . 170 Figure 4.11. m6A-abrogating RSV mutants have defects in replication in HBE culture...... 173 Figure 4.12. Pathogenicity and immunogenicity of m6A-mutated rgRSVs in cotton rats...... 176 Figure 4.13. The attenuated phenotype of m6A mutated rgRSVs is m6A-related...... 179 Figure 5.1. Fold change of m6A level in each virion RNA of virus grown from METTL3- knockout U2OS cells or wild type U2OS cells...... 198 Figure 5.2. Fold change of m6A level in virion RNA of RSV mutant virus...... 200 Figure 5.3. m6A-deficient virion RNAs induce higher type I IFN responses...... 202 Figure 5.4. m6A-deficient viral RNA enhances expression of molecules that involved in type I IFN signaling pathway...... 205 Figure 5.5. IFN response in A549 cells transfected with m6A-deficient RSV virion RNA...... 208 Figure 5.6. IFN production in A549 deficient cells stimulated with m6A-deficient SeV virion RNA...... 209 Figure 5.7. IFN production in A549 deficient cells stimulated with m6A-deficient MeV virion RNA...... 210 Figure 5.8. Affinity binding assay of RIG-I with RSV RNA...... 212 Figure 5.9. Affinity binding assay of RIG-I with SeV and MeV RNA...... 213 Figure 5.10. In vitro ubiquitination analysis of RIG-I...... 215 Figure 5.11. IFN response in A549 after viral infection...... 217 Figure 5.12. m6A-dificient RSVs are attenuated in replication in A549 cells...... 219 Figure 5.13.Viral expression and RNA synthesis of RSV mutant...... 221 Figure 5.14. Virus replication in HBE cultures...... 223 Figure 5.15. Body weight (a) and neutralizing antibodies (b) of mice that infected with m6A- deficient and wild type RSV...... 226 Figure 5.16. RSV-specific Th1 cell response triggered by pre-fusion F protein of RSV...... 228 Figure 5.17. RSV-specific Th2 cell response triggered by pre-fusion F protein of RSV...... 228 xvi

Figure 5.18. RSV-specific Tfh (a) and Th17 (b) cell response triggered by pre-fusion F protein of RSV...... 229 Figure 5.19. Replication of m6A-deficient RSV in cotton rats...... 231 Figure 5.20. Immunogenicity of m6A-deficient and wild type RSV...... 232

xvii

Chapter 1. Literature review

1.1. Origin, pathogenesis, and clinical and epidemiologic features of human respiratory

syncytial virus

1.1.1. Discovery and classification of Respiratory syncytial virus

Human respiratory syncytial virus (RSV) was first isolated in 1955 from a laboratory chimpanzee.

In 1955, there was an outbreak of colds and coryza among a group of chimpanzees and the agent

isolated from those chimpanzees was known as chimpanzee coryza agent (CAA), and now

recognized as RSV [1, 2]. Soon after the discovery, the same virus was identified from infants and

children with respiratory illnesses. Subsequent epidemiology studies showed that RSV is a

common respiratory disease in humans, particular infants, children, the elderly, and the

immunocompromised individuals. Since its discovery, the biochemical and molecular

characterization of RSV remained poorly understood until 1981, at when the molecular tools for

studying RSV became available [3].

RSV was belonging to the genus Pneumovirus, subfamily Pneumovirinae, family

Paramyxoviridae in the order of , but was reclassified into the genus

Orthopneumovirus in the family of Pneumoviridae in 2016 [4]. The Pneumoviridae is a family of large enveloped non-segmented negative-strand RNA viruses. The Pneumoviridae has only two genera, and , which includes three and two species,

respectively (Figure 1.1.). The genus orthopneumovirus include human orthopneumovirus (e.g.

1 RSV), murine orthopneumovirus (e.g. Pneumonia virus of mice, PVM), and bovine

orthopneumovirus (e.g. bovine RSV) species, causing respiratory diseases in human, and

bovine, respectively. Based on the genetic diversity, RSV is further divided into subgroups A1,

A2, B1 and B2. The genus Metapneumovirus includes (e.g. human

metapneumovirus, hMPV) and (e.g. avian metapneumovirus, aMPV) species, causing respiratory diseases in humans and birds respectively [4]. Together, the family

Pneumoviridae includes many important viruses that cause respiratory diseases in human or animals.

Figure 1.1. Phylogeny of the family Pneumoviridae.

2

Source derived from website: https://talk.ictvonline.org/ictv-reports/ictv_online_report/negative-

sense--viruses/mononegavirales/w/pneumoviridae/737/resources-pneumoviridae.

1.1.2. Clinical feature, transmission and epidemiology of RSV

RSV infects individuals of all ages, with high morbidity and mortality seen in infants, children, the elderly, and immunocompromised individuals. RSV is the no.1 cause of pediatric respiratory tract infection, and the no.2 cause of respiratory infection in the elderly, only second to virus [5]. Epidemiological studies suggest that 69% of infants are infected by RSV during the first year of life and 83% during the second year. By age of 5, nearly 100% children have been infected

[6]. The clinical signs and symptoms associated with RSV infection range from mild respiratory

problems to severe coughs, bronchiolitis, and pneumonia. Severe RSV infection can result in

wheezing and later in life and death. Age is the biggest risk co-factor for RSV infection,

but underlying conditions such as smoke exposure and history of asthma can also increase the risk

for infection [7]. Most RSV infections are mild and don’t require medical attention. However, some infections are more severe and result in outpatient visits to emergency departments or hospitalization.

Even though RSV is commonly seen as a childhood disease, RSV infections are frequent and occur multiple times throughout life of older children and adults [8]. Most recurrent infections are

symptomatic upper respiratory tract illnesses, as well as wheezing and low-grade fever, that tend

to be more severe and prolonged than the average cold.

RSV can spread quickly within daycare, families and many other groups of children via close

contact with infected individuals or their secretions. This observation provides two possible

mechanisms for RSV spread: 1) through sneezing, coughing, quiet breathing, and the large droplet 3 aerosols in the surroundings; and 2) through the infectious secretions from infected individuals

[9]. The survival time of RSV in the nasal secretions of infants varies between 30 min when on cloth and paper tissue to as long as 6 h when on countertops, and it extended if the humidity is low, as the winter respiratory season.

1.1.3. Burden of RSV diseases

RSV infection is largely restricted to the human despites the fact that it can also infect animals including (such as mice and cotton rats), sheep, bovine, and monkeys. However, those animals are semi-permissive to RSV infections and there are no spread and transmission between them. Chimpanzees are the only animal host in which RSV can replicate efficiently to allow the transmission from animal to animal, and to produce respiratory tract diseases [10]. In human, RSV pathogenesis a complex process which is dependent on the interaction of viral and host determinants. Disease can range from mild to lethal and can encompass a wide range of acute upper and lower respiratory tract disease manifestations, from mild at one extreme to bronchiolitis and pneumonia at the other.

In high-income countries, young infants and children are at the highest risk for severe RSV infection and hospitalization. It has been reported that bronchiolitis was the leading cause of all hospitalization for infants. Besides the huge burden of hospitalization, the outpatient disease with

RSV infection increase clinical and economic burden [11, 12]. Mortality associated with RSV infection is relatively rare in young infants in developed countries. In older children and adults, the mortality is the lowest. In the US, there are less than one thousand cases of death reported per year. Economically, an estimated total cost of 2.6 billion for hospitalizations for infants and more than 1 billion for elderly in the US per year.

4

In developing countries, RSV is the most frequent cause of mortality in post neonatal infants. It is

estimated that lower respiratory tract illness (LRTI) causes more than 200,000 death in children

under the age of five yearly and 90% of these fatalities is due to the limited viral diagnostic

capacity. The factors that contribute to the characteristic of RSV infection in developing countries

are unknown. However, it is likely that the poor access to health care, defectiveness in sanitation

system, and gaps in supportive care are the factors.

1.2. Virology of RSV

1.2.1. Virion

RSV is an enveloped virus. Two forms of RSV virion have been described. Virions produced in

vitro predominately contains 100–350 nm sized spherical particles whereas long filamentous form

of virions are 60–200 nm in diameter and up to 10 μm in length (Figure 1.2.). Both spherical and filamentous forms of virion are infectious [4, 13]. There are three viral transmembrane surface glycoproteins: the large glycoprotein G, the fusion protein F, and the small hydrophobic SH protein which are anchored on the surface of RSV envelop. These surface glycoproteins form transmembrane homo-oligomers that appear as short (11–16 nm) surface spikes. The matrix M

protein is present on the inner face of the envelope. A ribonucleoprotein (RNP) is located inside

of the particle which consists of a nucleocapsid (N) encapsidated non-segmented negative-strand

RNA genome complex (N-RNA) tightly bound by viral RNA-dependent RNA polymerase

proteins, the catalytic large (L) protein and the accessory phosphoprotein P and M2-1 proteins

[13].

5

Figure 1.2. Human respiratory syncytial virus virion.

Electron micrograph of filamentous RSV virion (A) and spherical RSV virion (B). (C) The structure of RSV [4, 13].

6

1.2.2. Viral RNA and proteins

The RSV genome (Figure 1.2.) is a single-stranded non-segmented negative-sense RNA of 13.2-

15.3 nt. RNA genome replication produces the antigenome which is a full-length complementary copy of the genome. Both genome and antigenome lack 5’ cap structure and 3’ poly A tails, and are completely encapsidated by the N protein to form a nuclease-resistant helical nucleocapsids.

These RNA templates are protected in nucleocapsids from degradation and recognition by the innate immune system, and only released temporarily during replication and transcription. The

RSV genome starts with a 44-nt extragenic leader region at its 3’ end and ends up with a 155-nt extragenic trailer region at its 5’ end (Figure 1.3.). The genome encodes 10 viral genes that are arranged sequentially, beginning with the non-structural protein (NS1): 3’ NS1-NS2-N-P-M-SH-

G-F-M2-L 5’. Each gene starts with a 9-nt highly conserved gene-start (GS) signal and terminates with 12–14-nt moderately conserved gene-end (GE) signal followed by 4–7 U residues that encode the poly A tail. Each gene is separated by intergenic regions and encodes a corresponding mRNA which is capped and guanine N-7 and ribose 2’-O methylated at 5’ end and polyadenylated at 3’ end. The intergenic regions are vary in sequence and length between strains and appear to be unimportant spacers. Each mRNA contains a single ORF, except for M2 which has two overlapping ORFs encoding the M2-1 and M2-2 [14].

7

Figure 1.3. Overview of RSV transcription and RNA replication.

During transcription, the polymerase enters the negative-sense genome at its 3’ end executes transcription to synthesize leader RNA, which is neither capped nor polyadenylated, and 10 mRNAs that are capped and methylated at the 5’end and polyadenylated at 3’end in a polar gradient manner. During replication, the polymerase initiates at the extreme 3’ end of the genome and synthesize a full-length complementary antigenome, which in turn serves as a template for synthesis of full-length progeny [14].

RSV encodes 11 proteins including two nonstructural proteins (NS1 and NS2), three surface

glycoproteins (F, G, and SH), N, P, L, M2-1, and M2-2. Among these 11 proteins, the N, P, M, F,

and L proteins are conserved in relative gene order throughout the family Pneumoviridae. The

NS1 and NS2 proteins are unique for the genus orthopneumovirus. All viruses in the family

Pneumoviridae encode M2-1 and M2-2 proteins which are produced from two overlapping ORFs in M2 gene.

8

The nonstructural protein NS1 (139 amino acids) and NS2 (124 amino acids) may have synergistic

functions by forming a complex, and they are found to interfere with innate immune response.

Studies have shown that NS1 and NS2 could block IFN signaling by causing the proteasomal

degradation of STAT2 [15]. RSV infection induces a relatively low level of IFN in vitro and in

vivo. It is likely that the strong IFN inhibitory effects of NS1 and NS2 may be the reason. In fact,

recombinant RSVs lacking NS1 and/or NS2 induce more IFN production than wild-type (wt) RSV

in cell cultures. In addition, these proteins play roles in inhibiting thus increasing cell

life and the viral yield [16].

The N protein (391 amino acids) is essential for virus replication and assembly. It binds tightly to

both RSV genome and antigenome to form helical nucleocapsids, which serving as the templates

for transcription and replication, and protecting its genome and antigenome RNA being degraded

by nuclease [17].

The P protein (241 amino acids) plays a critical role in bridging N, M2-1, and L proteins to

polymerase complex which is thought as an essential polymerase co-factor [18]. As a

phosphorylated protein, several serine and threonine residues of P protein are phosphorylated at

different levels of turnover, which are related to life cycle of virus including extracellular viral

particle formation and viral transcription [19].

The M protein (256 amino acids) is associated with viral inclusion body and plasma membrane and is essential for virus budding. It is not necessarily needed for the formation of viral filaments, but the filaments remain stunned and immature in the absence of M protein. The structural studies on M protein revealed that there is a large positively charged area in its surface which mediates the association of nucleocapsids and the negatively charged plasma membrane. In the early stage

9

of infection, the M protein is seen in the nucleus which indicating its functions in inhibiting host

transcription [20].

The SH protein (64 amino acids) is relatively small and is a transmembrane protein. Its ability in

modifying membrane permeability generates pore-like structures on membrane to provide channel like functions. Besides, its roles in budding and reducing apoptosis are also reported [21].

There are two major glycoproteins on the surface of RSV, the attachment glycoprotein (G) and the fusion protein (F). They play key roles in the initiation of virus infection. The G protein (298 amino acids) is responsible for the attachment with the cells by targeting the receptors on the cell surface.

There are two forms of G protein: the secreted form which accounts for 80% of released G protein and the membrane anchored form. The N terminus of G protein is responsible for the membrane anchor, and the secreted G initiated its translation at the second AUG in the ORF thus missing the membrane anchor by following proteolytic trimming [22]. In addition to its role in attachment, the

G protein also helps RSV to evade immune response [23]. The CXC3 motif in G protein is thought to mimic the CXC3 chemokine fractalkine thus reducing the immune cells flood into lungs of RSV infected animals [24]. The secreted form of G proteins interferes with the binding of neutralizing antibody due to its role in acting as an decoy. The F protein (574 amino acids) is responsible for viral penetration, entry, and formation. Upon receptor binding and other unknown triggering factors, the F protein undergoes a significant conformational change, transiting from its pre-fusion form to post-fusion form. The former form is thought to play more critical roles in inducing the neutralizing antibody, which is essential for protection. There are two cleavage sites on F protein which cleaves inactive F0 into F1 and F2 [22]. In addition, the F protein is able to initiate signaling transduction that involved in immune response by binding with TLR4

10

[25]. The F and G proteins are the only viral neutralization and are the major protective

antigens. Therefore, many antiviral drug development (such as fusion inhibitors) and monoclonal

antibody therapy mostly focused on these two proteins.

Due to the overlapped translation, there are two proteins (M2-1 and M2-2) produced from M2 mRNA. The M2-1 protein (194 amino acids) is unique to Pneumoviridae and it is an essential

component in RSV RdRp by interacting with P and L proteins to support RNA synthesis [18].

Besides binding with P protein, it also contacts with M protein to mediate its transportation to

inclusion body. In addition, M2-1 possesses a zinc binding activity, which is essential for M2-1

function. The M2-2 protein (88 or 90 amino acids) plays a role in regulating RNA synthesis by reducing its transcription. The studies on virus-like particle with M2-2 suggests that expression of

M2-2 promotes the packing efficiency. Deletion of M2-2 delays the replication of virus in vitro and in vivo [26].

As the largest protein among RSV viral proteins, the L protein (2165 amino acids) is the catalytic

subunit for RNA-dependent RNA polymerase (RdRp), harboring many critical functions such as

nucleotide polymerization, mRNA cap addition, mRNA cap methylation, and [25,

27]. The structure and function relationship of RSV L protein remains poorly understood due to

the difficulties in protein production and purification and the lack of robust in vitro biochemical

techniques. Most of our understanding of RSV L came from the studies of L protein of vesicular

stomatitis virus (VSV) because of their high similarity. Sequence alignment between the L proteins

of representative members of each family within NNS RNA viruses has identified six conserved

regions numbered I to VI (CRs I–VI). Many of enzymatic activities of L protein have been assigned to these conserved regions. For example, the nucleotide polymerization activity has been

11 assigned to the CR III and this region is also required for polyadenylation [28]. The mRNA capping enzyme has been mapped to the CR V whereas mRNA cap methyltransferases are located in CR

VI [28]. Although the exact functions of other three conserved regions (CRs I, II, and IV) are unknown, studies in Sendai virus (SeV) have shown that CR I is critical in binding P protein and

CR II may bind to the RNA template [29, 30]. In 2015, the electron cryo-microscopy (cryo-EM) structure of VSV L has been determined. The VSV L protein organizes as an RdRp domain, an mRNA capping domain, and a methyltransferase domain, a connector between the capping and methyltransferase domains, and a C-terminal domain (Figure 1.4.) [31]. Most recently, the cryo-

EM structure of RSV L protein was solved by two separate researcher groups. Gilman et al. presented a 3.2-A° cryo-EM structure of the RSV L polymerase in complex with the P phosphoprotein components of the core viral replication machinery in 2019. In their work, they also investigated the inhibitor escape mutants and mutations observed in live attenuated vaccine candidates which would facilitate the development of therapeutic agents for RSV [32]. Similarly,

Cao et al. presented a 3.67 Å cryo-EM structure of the RSV polymerase (L:P) complex in 2020.

By further analysis and comparison with VSV RNA polymerases at different stages, it provided enriched insights into the interrelationship and the evolutionary implications of the RSV polymerase [33]. Unfortunately, both groups failed to get the density information of the CD,

MTase, and CTD domain of RSV L protein.

12

Figure 1.4. Structure of VSV-L.

13

(A) Domain organization of VSV-L. The polymerase domain (RdRp) is in cyan; capping domain

(Cap), green; connector domain (CD), yellow; methyl transferase (MT), orange; C-terminal

domain (CTD), red. Amino-acid residue numbers indicate functional domain boundaries. Flexible

linkers 1 and 2 connect Cap to CD and CD to MT domain, respectively. Conserved regions within

L proteins of non-segmented negative-strand (NNS) RNA viruses are labeled CR I – VI. Asterisks

indicate the position of active site residues. (B) Ribbon diagram of VSV-L polypeptide chain;

domains colored as in (A). (C) Substrate channels and internal cavities of VSV-L, depicted as white surface enclosed by the structure in ribbon representation. In this orientation, the entrance to the template channel leading to the active site faces down; the channel runs between the RdRp and capping domains. Nucleotides can access the RdRp active site through the channel in the foreground [31].

14

Figure 1.5. Structure of the RSV L Core Bound to Tetrameric P.

RSV L is shown in ribbons colored in cool colors and each monomer of tetrameric P is shown in ribbons colored in a unique warm color. The RNA-dependent RNA polymerase domain (RdRp) is colored dark blue, and the polyribonucleotidyltransferase (PRNTase) domain is shown in light blue. The linear maps show the portions of L (top) and P (bottom) that are resolved in the structure as colored bars. The connector domain (CD), methyltransferase domain (MTase) and C-terminal domain (CTD) of L are not observed in the structure. Arrows denote conserved residues required for function. NTD, N-terminal domain; OD, oligomerization domain. The linear schematics for L and P are not drawn to scale [32].

15

1.2.3. Life Cycle of RSV

RSV life cycle (Figure 1.6.) starts with the attachment and entry which are mediated by the G and

F proteins [34]. Entry occurs by fusion of viral envelop with the cell plasma membrane. In addition, clathrin-mediated endocytosis is also reported as a mechanism for RSV enter the host cells. Upon entry, the RNP complex is immediately released into the cytoplasm and replication/transcription occurs in a structure called inclusion body in the cytoplasm. Replication yields antigenome and progeny genomes whereas transcription yields mRNAs. RSV assembly and budding is a process that occurs at the plasma membrane. Three surface proteins and M protein are involved in the trafficking and lipid rafts. In polarized cells, this process occurs at the apical surface. Infected cells develop syncytia that are a major viral cytopathic effect and lead to the destruction of the monolayer, but those are much less evident in well-differentiated, polarized epithelium in vitro and in vivo. In general, RSV life cycle is divided into three major stages: entry (attachment and fusion), replication and gene expression, and release (assembly and budding), which will be discussed in detail below.

16

Figure 1.6. A schematic view of RSV life cycle.

1.2.3.1. Attachment and entry

There have been many candidate cellular receptors for RSV entry, including annexin II [35], CX3 chemokine receptor 1 (CX3CR1) [36], calcium-dependent lectins [37], and heparan sulfate proteoglycans (HSPGs) [38] which bind with RSV G protein and epidermal growth factor (EGF) receptor [39], Toll-like receptor 4 (TLR4) [40], intercellular adhesion molecule 1 (ICAM-1) [41], and nucleolin [42] which bind with RSV F protein (Figure 1.7.) [34]. Some of them are only specific to certain strains of RSV. G protein plays the primary role in the attachment to the host . In immortalized cells, the G protein binds with glycosaminoglycans (GAGs) to initiate the entry [43]. However, GAGs are not present in surface of human well-differentiated,

17

primary bronchial epithelial (HBE) culture, suggesting that the receptors for RSV entry in vitro

and in vivo may be different. In HBE culture, it has been demonstrated that CX3CR1 is the

attachment receptor [44]. Interestingly, the G gene can be deleted from RSV genome and the resultant recombinant virus is viable [45]. This observation, combined with the fact that RSV F

alone can mediate fusion, suggests that F can bind to cellular receptors. In fact, F protein was

shown to bind to TLR4 and trigger the signaling for RSV entry. However, TLR4 alone is not

sufficient for RSV entry, it has been proposed that CX3CR1 and HSPG also bind to G protein

which tethers the virus particle to the cell surface and triggers the fusion. In addition, host

micropinocytosis of RSV provides an alternative pathway for the virus entry, however, this process

is less understood and receptor involved is unclear [46].

Figure 1.7. Binding and entry of RSV into the host cell. 18

Candidate receptors of RSV (A) such as TLR4, CX3CR1, and HSPG (B) bind to the RSV-G glycoprotein and act to tether the virus particle to the cell surface. Cell surface nucleolin may also be involved in the entry process (C) by triggering fusion of the virus and host cell membranes by binding to the RSV-F fusion glycoprotein (D). The virion fuses with the cell membrane and enters the cell, one of the last events of virus entry that must take place for successful replication of RSV in the host cell (E). Host cell macropinocytosis of RSV is also a route of entry for RSV (F). It is unclear which receptors are involved in this process (G). Internalization of the virion (H) is dependent on actin rearrangement, phosphatidylinositol 3-kinase activity, and host cell (I) early endosomal Rab5+ vesicles where proteolytic cleavage of the RSV-F protein triggers delivery of the contents into the host cells by fusion of the virus and endosomal membranes (J) [47].

RSV F is a class I membrane fusion protein, which mediate virus envelop-cell and cell-cell fusion

(Figure 1.8.) [47]. In the virion membrane, F protein is a trimer and is in a metastable, prefusion form. Prior to fusion, the F protein stays as a “spring-loaded” trimer with the major antigenic site exposed which is called as pre-F state. Upon triggering (such as binds to a receptor nucleolin or other triggering factors), the pre-F undergoes a dramatic conformational change during which the antigenic sites are lost and the fusion peptide inserts into the host cell membrane. Thus, the viral and host membranes get closely contact and the pore-like structures forms in the membrane which allows the viral nucleocapsid pass through the channel and enter the host cells. Similar to paramyxoviruses and many pneumoviruses, the fusion of RSV is pH independent and RSV- infected cells can further fuse with neighboring cells to generate multinucleated cells called syncytia. Although our knowledge about the mechanisms of fusion has increased recent years,

19

there are many questions remain unknown. For example, how does fusion occur at the right time

at right place, what are the triggers of fusion, and how does the process initiate?

Figure 1.8. Fusion process between the RSV envelope and cellular membrane.

20

The RSV envelope has multiple protruding RSV-F fusion glycoproteins, anchored via transmembrane domains (A). In the prefusion state, RSV-F exists as a spring-loaded trimer with the major neutralization epitopes shown at the N-terminal region. The major antigenic site Ø exists only on the prefusion trimer and is lost after fusion. Interaction between the RSV-F trimer and a receptor may cause RSV-F to undergo a dramatic conformational shift (B), which leads to insertion of the fusion peptide into the host cell membrane (C) and forcing of the viral and host membranes into close contact (D). Although only two RSV-F monomers are depicted for simplicity, the combined force of multiple RSV-F conformational shifts is required to overcome the thermodynamic barrier of mixing membranes and establish a stable fusion pore for viral nucleocapsid delivery (E) [47].

1.2.3.2. RNA replication and gene expression

Upon enter into the host cells, the RNA complex is released into the cytoplasm where the viral replication complexes consist of viral and certain host proteins are formed. Unlike positive strand

RNA viruses, the naked RNA genome of negative-sense RNA viruses is not infectious. The

genome of RSV is completely encapsidated by N protein formed N-RNA complex which serves

as active templates for RNA replication and mRNA transcription. During replication, the RdRp

complex (L, P, and M2-1) recognizes N-RNA template and carries out two distinct RNA synthesis:

(i) replication to produce a full-length complementary antigenome, encapsidated by N protein,

which in turn serves as templates to synthesize progeny genome; and (ii) transcription, which

yields capped, methylated, and polyadenylated mRNAs. The transcription of RSV initiated by the

polymerase encounters the leader region (Le) at 3’ end of genome [48]. It happens in a polarized

manner that under the control of cis-acting gene start (GS) and gene end (GE) sequences which 21

flank each gene of RSV. The RNA synthesis begins when the RdRp binds with GS region and

stops at the recognition of GE signal, where the polymerase polyadenylates and release the mRNA.

The sequence of GS of each RSV mRNA and many other related viruses is highly conserved, with

only one GS is slightly different in sequence, other nine GS of RSV are exactly the same [49]. The

GS contains multifunctional regions that direct polymerase to cap and methylate the nascent

mRNA. Due to the polarized gradient of transcription, each mRNA is different in abundance, with

decreasing amount from most enriched NS1 mRNA to the least L mRNA. For replication, it

initiates at the first nucleotide at the 3’ end of genome (Figure 1.9.) [50]. In contrast to the mRNA

synthesis, the polymerase ignores the GS and GE signals in between of each gene and produces a

full-length positive-sense antigenome, which serves as the template for the generation of more

genome. Neither genome nor antigenome is capped and polyadenylated but are encapsidated by

the N protein.

Large cytoplasmic protein inclusions, which are defined as the inclusion bodies, are the distinct

structures formed during RSV infection. About 6 h post infection, small inclusions begin to form

as the same time as viral protein synthesis starts. The size of inclusions increases as RSV

replication proceeds. Viral N, P, L, and M2-1 protein together with viral genome are co-localized in these inclusion bodies [51]. Studies have found that a number of host proteins including some antiviral and signaling proteins such as MAVS and MDA5 are also observed colocalizing in the inclusion bodies [52]. The presence of those proteins in the inclusion bodies supports the hypothesis that the inclusion bodies play a critical role in favoring RSV replication by preventing host cell responses.

22

Figure 1.9. Nucleotide positions in the leader region of genomic RNA that are important for transcription (top) and RNA replication (bottom).

Important residues present in positions 1–11 are indicated with open boxes, note that those that are important for transcription are a subset of those important for RNA replication. A region that increases the efficiency of transcription is indicated with a dashed box. The GS signal of the first gene, necessary for transcriptional initiation but not involved in RNA replication, is underlined in the diagram for transcription. A region that contains an apparent encapsidation signal necessary to produce full-length replication products is indicated with a shaded box. Sequences are in negative- sense [50].

1.2.3.3. Assembly and budding

RSV generated nascent genomic RNAs in the inclusion bodies remain in the cytoplasm, while the translated G and F proteins traffic through the secretory pathway to the apical surface of the host

23

cells. Then the RNP complex is guided by M proteins to the surface where they “meet” with the

glycoproteins. Viral proteins are assembled into filaments that contain both RSV genomic RNA

and structural proteins at the cell surface then separated from the host cell (Figure 1.10.) [53]. In this process, cytoskeleton is thought to play the key roles in trafficking and interactions with M protein. Literatures have found that nucleocapsids of RSV utilize microfilaments to transport to the assembly sites at the cell surface, and this movement is strictly mediated by M protein to guarantee that only mature nucleocapsids move to the plasma membrane [53]. F proteins may also be involved in this process but needs more exploration. At the assembly sites, the M-RNP (+/- F) complex interact with glycoproteins to form the budding particles [54].

Figure 1.10. Model for the RSV assembly and budding.

24

a, The M-containing RNPs of RSV utilise microfilaments (stress fibres) as a means of transport to reach assembly sites; this movement is facilitated by the M protein interaction with microfilaments. At the assembly site, the RNPs associate with bundles of actin filaments and the envelope glycoproteins to complete viral assembly and initiate budding. b, The growth of the viral

filament is supported by microfilaments and actin-modulatory proteins; mature virions are released

through as yet unknown mechanism of membrane scission [53].

1.2.4. mRNA capping and methylation in RSV

RSV produces mRNAs which are capped, methylated and polyadenylated. In eukaryote, there are

three different types of mRNA cap. Cap 0 is defined as the first guanosine of cap structure is

methylated by a guanine N-7 (G-N-7) methyltransferase to yield m7GpppN-RNA. The G-N-7

methylated cap can be further methylated by the ribose-2’-O methyltransferase at the first

7 transcribed nucleotide to form m GpppN1m-RNA which is called cap 1. In some cases, the second

transcribed nucleotide is methylated at the 2’-OH group of ribose moiety, thus forming

7 m GpppN1mpN2m-RNA (cap 2). Although most caps in nature including RSV mRNA cap are of

the cap 1, some plant viruses are found to have cap 0 structure [55]. NNS RNA viruses utilizes

unique mechanisms in capping and methylation. In eukaryotic cells, mRNA capping is an early

posttranscriptional event which is critical for stability, transportation as well as translation of

mRNA [55]. A series of enzyme reactions are involved in the cap formation (Figure 1.11.) [56].

First, the 5’ triphosphate end (pppN) of the newly synthesized mRNA is converted into the 5’ diphosphate end (ppN) by an RNA triphosphatase (RTPase). Then, an RNA guanylyltransferase

(GTase) transfers GMP that derived from GTP to 5’ ppN-RNA via 5’-5’ triphosphate linkage to

25

yield the cap structure. Different from the conventional eukaryotic capping pathway, the cap

structure of NNS viral mRNAs was achieved by a novel polyribonucelotidyltransferase (PRNTase)

which transfers a monophosphate RNA onto a Guanylyl Diphosphate (GDP) acceptor through a

covalent L-RNA intermediate (Figure 1.11.) [56]. Specifically, in the first step, the GTPase hydrolyzes GTP to GDP. Secondly, the PRNTase facilitates the covalent linkage of the 5’ end of viral mRNA with viral polymerase (L protein) after removal of pyrophosphate and the cap structure is formed after GDP attacks the phosphodiester bond between L-protein and mRNA.

Following cap formation, mRNA is sequentially methylated during which the S-adenosyl-L- methionine (SAM) serves as the donor. The capped eukaryotic mRNA is typically methylated by two conventional steps. First, the capping guanylate is methylated by a G-N-7 methyltransferase

(MTase) to yield cap 0, then the cap 0 structure can be further methylated at ribose-2’-O position by a ribose-2’-O MTase to form cap 1. In this conserved process, the G-N-7 methylation occurs prior to 2’-O methylation, and they are carried out by two separate enzymes, each containing its own binding site for SAM. Different from in eukaryotic cells, viruses encode their own methylation machinery (Figure 1.12.) [57]. For NNS viruses such as VSV, G-N-7 and 2’-O methyltransferase (MTase) activities are encoded by the single conserved region VI located in the viral large protein (L). This unusual dual activity MTase methylates viral mRNA at an unconventional order, in which 2’-O methylation precedes and facilitates the G-N-7 methylation.

This is different with all known mRNA methylation including many other viruses.

26

Figure 1.11. Comparison of mRNA cap formation in eukaryotic cells and VSV.

(A) VSV mRNA cap formation, (B) Cellular mRNA cap formation [56].

Figure 1.12. Proposed model for mRNA cap methylation in VSV and eukaryotic cells.

(A) VSV MTase model. (B) Cellular MTase model [57]. 27

Studies in early 70s have clearly shown that G-N-7 methylation of mRNA cap structure is essential for mRNA stability and translation. Specifically, G-N-7 methylation of mRNA cap structure is essential for the recognition of the rate limiting factor for translation initiation, eIF-4E [58].

However, the biological function of 2’-O methylation remains unclear until recently. Studies on

West Nile Virus (WNV) showed that the 2’-O methylation of 5’ cap structure functions as a molecular marker to escape the detection by host innate immunity [59]. Recombinant WNV defective in 2’-O MTase was attenuated in C57BL/6 mice, but restored the pathogenicity in knockout mice that are lacking type I interferon (IFN) signaling pathway, demonstrating that WNV utilizes the 2’-O methylation on cap structure to evade immune response through escape of suppression of IFIT [59]. It has been also reported that RNA cap 2’-O methylation in mouse hepatitis virus (MHV) [60] and human 229E had a similar function but via a different mechanism [61]. Specifically, the 2’-O methylation facilitates viral evasion from host innate immune detection by RNA sensor MDA5 [61]. Thus, the 2’-O methylation serves as a molecular signature for host cells to discriminate self and non-self RNA. However, whether 2’-O methylation plays a similar function in NNS RNA viruses including RSV is unknown.

RNA cap methylation has been widely used as a target for live attenuated vaccines. It has been reported that MTase-defective VSV mutants were able to induce high level of VSV-specific antibodies in mice and thus provided full protection against challenge of VSV Indiana serotype

[62]. In addition to VSV, studies on other NNS viruses such as Sendai virus [63] and hMPV [64], and positive strand RNA virus, such as have also found that MTase-defective viruses were attenuated in cell culture as well as in a1nimal models [65]. For live attenuated vaccine, it has been technically challenging to isolate a virus with low virulence but maintaining high

28

immunogenicity. Introducing mutations into the MTase site could be an ideal approach because

the MTase site locates at the L protein, which is not a neutralizing antibody target. Therefore, virus

lacking MTase would likely be attenuated without defective in immunogenicity.

1.3. Interferon signaling

Interferons (IFNs) are a group of cytokines released by host cells in response to the virus infection

and regulate both innate and adaptive immune systems. There are more than twenty known IFN

genes and proteins in animals and humans, and they are divided into three classes including types

I, II, and III IFN [66]. The type I IFN presents in humans are IFN-α, IFN-β, IFN-ε, IFN-κ, and

IFN-ω, and their production is prohibited by interleukin-10 [67]. The type II IFN only consists of

IFN-γ which is known as an immune interferon and is activated by interleukin-12 [67]. The type

III IFN encodes IFN-λ and it signals through a receptor complex consisting of IL10R2 [67].

Among them, the type I IFN is thought to be responsible for regulating and activating immune

responses and carry the most potent ability to resist to viral infection.

IFN-α and IFN-β are the most studied members of the type I IFNs. Both of them are induced

following the recognition between pathogen-associated molecular pattern (PAMP) and pattern

recognition receptors (PRR) after viral infection and are transcriptionally regulated by the host

cells [68]. There are three major signaling pathways are involved in IFNα/β production in humans

(Figure 1.10.). In general, the RIG-I/MDA5 pathway is activated by the RNA virus infection. The

released viral RNA is recognized by RIG-I and/or MDA5 then promoting the phosphorylation of

downstream IRF3/7, which is translocated to nucleus to induce IFN production [69]. The second

pathway is the TLR3/4-mediated pathway which is involved in the recruitment of the adaptor

protein TRIF [70]. The third pathway described here is triggered by TLR7/8 and TLR9 which are

29

located in the endosome and the activation of IRF5/7 is responsible for the production of IFN [71].

Following the stimulation, type I IFN is secreted and binds to type I interferon receptor (IFNAR) which stimulates the JAK1-STAT pathway leading to the formation of the ISGF3 complex consists of STAT1-STAT2 dimers and IRF9. The ISFG3 complex then regulates the expression of IFN- stimulated genes (ISG) by targeting the IFN-stimulated response elements (ISRE) in its promoters.

This autocrine/paracrine feed-back process creates an antiviral state in the surrounding cells.

Figure 1.13. The signaling pathways of type I interferon.

30

Source derived from website: https://www.invivogen.com/review-type1-ifn-production.

1.4. Host innate immunity against RSV infection

The innate immune response is the first protection line against RSV infection. The initiation of innate immune response depends on the recognition between PAMP on the virus and PRRs from host cells. Since RSV is a negative-sense RNA virus, it generates both ssRNA and dsRNA species during replication which are detected by PRRs. PRRs such as toll-like receptors (TLR), RIG-I-like

receptors (RLR), and NOD-like receptors (NLR) play key roles in establishing the antiviral state

by inducing type I IFN and chemokines that promote inflammation and direct the recruitment of

immune cells.

1.4.1. TLR

Multiple TLRs, such as TLR2, TLR3, TLR4, and TLR7, are associated with RSV infection. TLR2

is found to act as a functional receptor for RSV and activate innate immune response by promoting

chemokine production and dendritic cells thus controls the viral replication in lungs [72]. TLR3 is

an intercellular receptor to recognize dsRNA. TLR3 is capable of detecting dsRNA that formed

during RSV replication and mediates the downstream signaling through TRIF and drives the

production of IFN [73]. TLR3 also plays important roles in balancing Th1 and Th2 type responses.

Activation of TLR3 elevated the Th1 type response whereas deletion of TLR3 leads to the

increased Th2-biased response [74]. TLR4 functions depend on the interaction with RSV F protein.

Activation by F protein leads to NFκB-mediated innate immune response and inflammations [75].

TLR7 recognizes most ssRNA viruses including RSV and it mediates T cell response that enhances

immunopathogenesis and mucus production after RSV infection [76].

31

1.4.2. RLR

MDA5 and RIG-I are two critical RNA sensor molecules which recognize viral RNA in most cell

types. RIG-I detects ssRNA (genome and antigenome) bearing 5’ triphosphate whereas MDA5

recognizes the long dsRNA formed during viral replication [77]. They are upregulated during RSV

infections and activate the downstream NF-κB and IRF3 pathways signaling through MAVS

adaptor protein which localizes to the mitochondrial membrane [77]. Production of cytokines such

as IFN is inhibited by NS1 and NS2 during RSV infection because those proteins decrease the

interaction between RIG-I and MAVS [78].

1.4.3. NLR

Similar to RLR receptors, NOD2 is reported to recognize ssRNA virus and it functions through

activation of IRF3 by translocated to the mitochondria where it interacts with MAVS. RSV

infection can also increase the expression of NOD2 which may due to the influence of IFN

feedback [72].

1.4.4. F and G protein

Besides RSV RNA, viral proteins such as G and F are also involved in the regulation of innate

immune response to RSV infection. The F protein is able to signal through TLR4 and CD14 to

induce IFN which helps to build antiviral state [25]. The G protein modulates SOCS expression to inhibit type I IFN and interferon-stimulated gene (ISG)-15 expression in RSV infection, and that

SOCS induction is linked to toll-like receptor (TLR) signaling by RSV F protein [79].

1.5. Antiviral drug against RSV

Development of a safe and effective antiviral drug for RSV is urgently needed for high-risk

populations particularly infants who are born prematurely and those with underlying disorders

32 such as chronic lung disease (CLD), congenital heart disease (CHD), and other birth defects. The only immunoprophylaxis agent used for RSV therapy is the palivizumab, which is a monoclonal antibody. In 1998, FDA approved palivizumab to prevent RSV infection for high-risk populations.

It was shown that palivizumab was effective in reducing hospitalizations and preventing serious lower respiratory tract infections in high-risk infants. However, palivizumab is very expensive and are not affordable for low-income families and developing countries. It is estimated that a single course of palivizumab can cost $1500 to $4300 per month per child [80].

Ribavirin is the only FDA-approved antiviral drug for aerosol treatment of serious RSV infections in hospitalized children. Ribavirin is a guanosine analogue that has broad-spectrum activity against several RNA and DNA viruses including RSV [81], influenza [82], parainfluenza [83], adenovirus

[84], [85] , Lassa fever [86], and Hantaan viruses [87]. Ribavirin has been used to treat severe RSV infections but also controversial due to its potential toxicity to exposed person and health care workers, the high cost, and unpredictability, it is not recommended to use ribavirin as a good treatment for RSV infections [81].

There are currently multiple antivirals being investigated in clinical trials (Table 1.1.) [88]. They are generally divided into four groups: 1) Monoclonal or polyclonal antibodies, most of which are against the more conserved F protein, and few are designed to target G protein; 2) antisense anti-

RSV drugs, usually by using interfering RNAs to target the viral proteins; 3) fusion inhibitors, which prevent viral fusion and entry; and 4) polymerase inhibitors which inhibit RSV replication.

33

Table 1.1. Agents in development for treatment of RSV infections.

1.6. RSV vaccine development

Vaccination is the most effective strategy to combat infectious diseases. Currently, there is no

FDA-approved vaccine for RSV despites the fact that great efforts have been contributed to the

development of a safe and efficacious vaccine. In general, inactivated and live attenuated vaccines are the two most common strategies used in vaccine development. For safety reasons, an

inactivated vaccine is often preferred. Soon after the discovery of RSV, a formalin-inactivated

RSV vaccine (FI-RSV) was developed and tested in humans. Unfortunately, this vaccine not only

failed to protect the vaccinated children but induced enhanced respiratory disease (ERD) upon

natural infection with RSV. Eighty percent of the vaccinated children were hospitalized following 34

natural RSV infection, and two children died [89]. The ERD was also observed in FI vaccines of

other pneumoviruses (such as human metapneumovirus, hMPV) [90] and paramyxoviruses (such

as human parainfluenza virus 3, PIV3) [91]. Clearly, FI vaccine is not safe for RSV, hMPV, and

PIV3 all of which cause upper and lower respiratory tract infection in humans. In early 1960’s,

research efforts were quickly switched to the development of alternative RSV vaccines.

1.6.1. Live attenuated vaccines

Live attenuated RSV vaccine candidates become one of the most promising vaccines for RSV

because they do not appear to prime for ERD as the data suggested in animal models and human

clinical trials. From 1970-1990, a panel of cold passaged (cp) and temperature sensitive (ts) RSVs

have been identified by repeatedly passing RSV at low temperature or inducing mutations using

chemical mutagens [92] [93]. Several of these live attenuated vaccines have been tested clinically.

None of them were found to have satisfactory attenuation characteristics including the safety,

genetic stability, and efficacy.

Until 1995 when the reverse genetic system for RSV was developed, it provided the capability to

manipulate the virus genome by introducing mutations into full-length cDNA clone. The recovery

of virus involves co-transfection of cells with a plasmid encoding a copy of the RSV antigenome

and supporting plasmids expressing N, P, M2-1, and L proteins. The expressed RSV antigenome and supporting proteins assemble into nucleocapsids that launch a productive infection [94].

Therefore, a number of candidates with attenuating point mutations were identified and evaluated to be promising (Table 1.2.) [95]. However, since the aggregate attenuating effect is not always precisely the sum of the individual mutations, and mutations are sometimes incompatible, the resulting level of attenuation cannot be predicted precisely. In addition, gene stability is always a

35

big concern for RSV mutant with point mutations. To further enhance the gene stability, studies have found that deletions of RSV genes could be a better strategy and with some combinations, they can restrict viral replication in vivo. Due to the association with neutralization and protection, deletion of G gene was not considered as an ideal attenuating mutation. The ΔNS1 and ΔM2-2 mutations could be appropriate mutation since they conferred an approximates level of attenuation thought to be necessary for a live RSV vaccine [96]. The ΔSH and ΔNS2 mutations were

associated with a 10-fold and 100-fold restriction of virus replication in chimpanzees respectively, which need to be combined with other mutations in order to make a suitably attenuated vaccine candidate [97]. Because the antagonistic roles of the NS1 and NS2 proteins in type I interferon

(IFN) production and signaling, the increased IFN responses and antiviral effect would restrict the viral replication for the NS1/NS2 deletion viruses [98].

One potential limitation of live-attenuated RSV vaccines is that the magnitude of the antibody level is generally associate with the replication of the virus. Therefore, it would be difficult to balance between the level of antibodies that required for protection and the level of replication that required for safety. However, with improved understanding of RSV gene function, we might able to find novel strategies to de-link replication and immune response, either by enhancing antigen production while limiting viral replication, or by deletion the genes involved in immune response to enhance the immunogenicity.

36

Table 1.2. Recombinant live-attenuated RSV vaccines.

37

1.6.2. Protein-based vaccines

Protein-based RSV vaccines are approaches that had been thought to be safer to be used in infants,

pregnant women, and immunocompromised persons compared to the live attenuated vaccines.

Currently, the protein-based RSV vaccines fall into three major categories: purified protein

vaccine, peptide vaccine, and virus-like particles vaccine. The design of those vaccines is mostly

based on the intact F protein or both F and G proteins, peptide fragments of F and G protein, and

particles of various types containing F and/or G proteins. It is known that F protein is the major

target for inducing neutralizing antibody. Both post-fusion and pre-fusion F protein has been tested

in animal models. Since the F protein on the surface of RSV virions is pre-fusion form, it is

proposed that the immunogenicity of pre-fusion F is better than the post-fusion F protein. The earliest developed protein vaccine candidates were purified F protein (PFP) isolated from RSV infected VERO cells. However, limited preclinical studies in animal models are described for these

candidates [99]. Subsequently, candidates containing both F and G proteins as well as M protein

were co-purified and found not to induce ERD upon RSV challenge in cotton rats [100]. Studies

on those candidates focused on safety and immunogenicity but failed to test efficacy. There was

little indication that those candidates provided sufficient protection. For peptide vaccine, BBG2Na

is a candidate with sequences from amino acid 130-230 of the G protein fused to the albumin-

binding domain of the Streptococcus G protein. Immunization of mice triggered antibody

response, and of cotton rats, this candidate only protected infection for lower respiratory tract but

not upper respiratory tract [101]. Two good examples of RSV VLP candidates are Newcastle

disease virus (NDV) based VLPs [102] and influenza M1 based VLPs [103]. Both of them are

containing the G and F protein of RSV, and were tested to stimulate robust neutralizing antibodies

38

as well as T cell response. Even though protein-based vaccines are capable of inducing neutralizing

antibody and protection in animal models, it is not clear whether the effects of a Th2-associated T

cell response would be overcome in humans, thus the risk of enhanced disease still exists for the

application of subunit protein vaccines.

1.6.3. Nucleic acid-based vaccines

Nucleic acid vaccines, such as naked DNA or mRNA, are approved to be worthy of continued

development and testing in humans. DNA and mRNA are easy to manufacture and purify and it

can be produced rapidly and efficiently on a commercial scale. In addition, the DNA or mRNA

itself is not immunogenic and the endogenously synthesized protein are thought more “authentic’’

since they are modified by the host. Thus, they can stimulate both humoral and cellular responses

that similar as triggered by natural infection. Early in mice work showed that plasmids carrying

the RSV F gene were able to induce strong neutralizing antibodies and lung cytokine expression

that was comparable to live RSV infection [104]. The RSV G DNA vaccine was proved to confer

protection in the lower respiratory tract against RSV challenge, and to induce a balanced Th1/Th2

anti-G response [105]. Besides DNA vaccine, study showed that i.m. administration with very low

dose of self-amplifying mRNA encoding the RSV F protein, mice and cotton rats induced very high titers of IgG1 and interferon (IFN)-producing CD4 and CD8 T cells [106].

1.6.4. Vectored vaccines

The RSV vectored vaccines including bacteria vectors and some replication deficient viral vectors, such as adenovirus or VSV vectors containing one or more RSV gene inserts.

Bacterial vectors display series of attractive features including the capacity to induce systemic immune responses. A S. carnosus vector was used to express three different peptides derived from

39

the RSV G protein (residues 130–230) on the bacterial surface. It was found to elicit significant

RSV G specific IgG responses with balanced IgG1/IgG2a responses in mice and half of those vaccinated mice were protected from intranasal RSV challenge [107].

Viral vectored RSV vaccine is currently utilizing adenovirus and non-pathogenic viral genomes as delivery systems. Adeno-vectored RSV vaccine candidates GlaxoSmithKline’s ChAd155-RSV

(GSK3389245A) and GSK3003891A were shown both safe and immunogenic in clinical trial of healthy, non-pregnant women aged 18–45 years clinical trial. However, further clinical evaluation was canceled due to the instability of the Pre-F antigen during manufacturing [108]. Adenoviruses of serotype 26 (Ad26) are engineered to comprise a nucleotide sequence encoding RSV F protein, which showed efficacy against RSV in mice and cotton rats [109]. Other , such as

modified Vaccinia Virus Ankara (MVA), is also used for RSV vectored vaccine. MVA-BN

(modified Vaccinia Ankara—Bavarian Nordic) is undergoing investigation and it was able to elicit

broad antibody and T cell responses to both RSV subtypes that lasted 6 months. Even though viral

vectors have many attractive features, they are not a panacea. Pre-existing immunity to the vector

may block the transduction, concerns over vector pathogenicity are always present, and in some

cases large-scale manufacturing is challenging [110].

1.7. Introduction to internal RNA methylation

RNA modification occurs to all living organisms. Currently, more than 150 different types of

chemical modifications have been discovered in eukaryotic RNA. With the recent development of

RNA sequencing methodology and techniques, several RNA chemical modifications including

RNA methylation have been relatively well studied, which significantly enhances our knowledge

on the biological functions of these modifications. RNA methylation is a reversible post-

40

transcriptional modification. It has been found in different RNAs including tRNA, rRNA, mRNA,

snRNA, snoRNA, miRNA, and viral RNA [111]. To date, six different methylations on mRNA,

including methylation occurs at the N1 (m1A) and N6 (m6A) atoms in adenosine, N3 (m3C) and

C5 (m5C) in cytidine, N7 in guanosine, and at the 2’-OH of ribose (Fig. 1.14.) [112].

Figure 1.14. A schematic view of the structure and localization of methylated nucleosides in

eukaryotic mRNA.

The bold line represents the coding sequence, and the thin lines are 5’ and 3’ untranslated regions

(UTRs) [112].

Currently, the most abundant and well-studied internal RNA modification is the N6-

methyladenosine (m6A) [113]. It widely exists in different eukaryotic species including yeast, plants, flies, and mammals. Despites the fact that it was discovered in the 1970s, its functions

remained unknown until recently. In 2012, an m6A-seq technique involved in enrichment of m6A- modified RNA using antibody-based immunoprecipitation followed by high-throughput sequencing made it possible to map m6A in RNA [114]. Since then, more than 10,000 m6A peaks

have been validated in human transcripts by using this method. The consensus RNA motif of 41

RRACH (R = A or G; H = A, U, or C) is found to carry m6A methylation and they are enriched in

regions such as long exons, near 3’ untranslated regions and stop codons [114]. The m6A

modification is a dynamic and reversible process and the balance is controlled by m6A

methyltransferases (“writers”) and demethylases (“erasers”) (Figure 1.15.) [115]. The functions of m6A are mediated by a panel of m6A-binding proteins called m6A “readers”.

Figure 1.15. Diverse molecular functions of m6A.

42

In eukaryotic cells, RNA m6A level is dynamically regulated by “writers” and “erasers”, and recognized by “readers” in direct or indirect ways. The diversity of cellular processes and functions involving m6A is mainly contributed by various “readers”. Potential functions of m6A in nuclear

RNA include: a, mRNA alternative splicing; b, secondary structure switching; c, mRNA export; d, pri-miRNA processing; e, mRNA stability; and f, XIST-dependent X chromosome inactivation.

The functions of m6A in cytoplasmic RNA include: g and h, enhances mRNA translation efficiency; and i, accelerates mRNA decay [115].

1.7.1. Methyltransferases/Writers

The internal m6A methylation is installed by a multicomponent S-adenosyl-L-methionine (SAM)- dependent methyltransferase complex composed of Methyltransferase Like 3 (METTL3),

METTL14, Wilms Tumor 1 Associated Protein (WTAP), KIAA1429, RNA Binding Motif Protein

15 (RBM15), and zinc finger CCCH domain-containing protein 13 (ZC3H13) [116]. Among these writer proteins, METTL3 and METTL14 are the major enzymes which function as the catalytic subunit and activator, respectively. METLL3 is a highly conserved protein which is located mainly in nucleus and regulates RNA metabolism in nucleus. However, other studies have shown that

METTL3 is also distributed in the cytoplasmic although the abundance is lower than that in nucleus. In vitro study reveals that METTL3 preferably methylates the sequences of GAC and

AAC in RNA. Besides its methyltransferase activity, METTL3 has been demonstrated to have functions in promoting translation [117], cell programming [118], modulating embryonic development [119], and many other immune associated processes [120]. METTL14 forms a stable heterodimer with METTL3 and provides an RNA-binding scaffold for METTL3-METTL14

43 methyltransferase complex. METTL14 functions as a methyltransferase activator and itself alone is not able to catalyze the transfer of methyl groups. In addition, METTL14 harbors many other biological functions such as embryonic development and maturation. Deletion of METTL14 in embryonic stem cells of mice led to dysregulation of plenty of gene expression enriched in embryo development pathways which are differentiation related or essential for mouse early development

[121]. Studies also showed that METTL3 and METTL14 are potential tumor suppressors.

Depletion of METTL3 and METTL14 promotes the tumor growth and progression.

1.7.2. Demethylases/Erasers

The discovery of m6A demethylases was much later than m6A writer. The first m6A demethylase, fat mass and obesity-associated protein (FTO), was identified in 2011 by Jia et al [122]. Two years later, the second demethylase, α-ketoglutarate-dependent dioxygenase alkB homolog 5

(ALKBH5), was discovered [123]. The finding of demethylases revealed that post-transcriptional modification is reversible. FTO was originally thought to play a role in obesity in human and have the ability to demethylate N3-methylthymidine in RNA or DNA. In 2011, it was found that depletion of FTO dramatically increased m6A levels of mRNA, leading to the discovery that FTO is an m6A demethylase. Additionally, FTO is involved in the regulation of cell growth, suggesting that FTO may play roles in tumor progression [124]. Unlike METTL3 and METTL14 which act synergistically as m6A methylase, ALKBH5 functions independently with FTO. Similar to

METTL3 and METTL14, ALKBH5 and FTO are mainly distributed in the nucleus and regulate nuclear export of mRNA. In addition, ALKBH5 plays a critical role in immune response to viral infections by sequestering the m6A demethylated antiviral transcripts in the nucleus thus inhibiting the production of interferons [125].

44

1.7.3. m6A binding proteins/Readers

The functions of m6A are medicated by m6A binding proteins, termed m6A readers. These m6A binding proteins were discovered using approaches of affinity chromatography and mass spectrometry. Most of the reader proteins are members of YT521-B homology (YTH) family proteins and they can be generally divided into two categories: nuclear m6A readers and cytoplasmic m6A readers. YTHDC1 is an example of nuclear m6A readers which preferentially recognizes the GG (m6A) C sequences. It plays important roles in splicing and nuclear exportation

[126, 127]. Another potential nuclear reader protein is HNRNPA2B1 which belongs to the

HNRNP family, but its functionaries are being investigated. It was reported that HNRNPA2B1 can affect splicing by directly binding to the m6A [128]. However, structural study suggests that it functions through an “m6A switch” mechanism instead of direct binding [129]. There are four known cytoplasmic readers including YTHDF1, YTHDF2, YTHDF3, and YTHDC2. YTHDF1 can promote the translation efficiency by involved in the translation initiation machinery [130].

YTHDF2 has a unique function in RNA degradation because it is associated with deadenylation and decapping enzymes which are essential in RNA decay. The YTHDF2 mediated RNA decay has been found to impact the zygotic development in vivo based on the studies on zebrafish and mice [131]. The function of YTHDF3 is likely mediated by interacting with other reader proteins.

The interaction of YTHDF3 and YTHDF1 regulates mRNA translation whereas interaction with

YTHDF2 plays a role in RNA decay [132] [133]. YTHDC2 is a 160 kDa protein which contains multiple helicase domains. It impacts meiosis prophase I thereby affecting the spermatogenesis and oogenesis [134]. Additionally, it shows a variety of capabilities in regulating RNA binding.

45

The fact that YTHDC2 lacked binding ability to m6A in HEK cells indicates that it may binds m6A

under unique circumstances or in certain cell types [135].

1.7.4. m6A and viral infections

Viruses are intracellular parasites and they must rely on host cells to synthesize their own genetic

materials and proteins. Viruses that replicate in nucleus likely contain m6A modification in their

RNA because host m6A writer and eraser proteins are mainly localized in the nucleus. In early

1970s, it was found that mRNAs produced by several DNA viruses (such as simian virus 40,

adenovirus, and type 1) contain m6A methylation, which is not surprising as

DNA viruses replicate in the nucleus [136] [137, 138]. In early 1980s, it was found that RNAs of

Rous virus (a ) and influenza virus contained internal m6A modifications, which

is also not surprisingly as replication cycle of these viruses goes through nuclear phase [139, 140].

Although it is known that RNA of these viruses containing m6A methylation, the roles of m6A

methylation in viral RNA is not understood until 2016. The first evidence that m6A methylation

plays a role in virus life cycle came from the study of the HIV in 2016. Using high-throughput

m6A-seq technique, it was found that HIV RNA was m6A modified. Subsequently, the authors

determined the effects of knockdown and overexpression of m6A writer, eraser, and/or reader

proteins on HIV replication and gene expression. It was found that YTHDF1–3 proteins recognize

m6A-modified HIV-1 RNA and inhibit HIV-1 infection in cell lines and primary CD4+ T-cells.

Silencing the m6A writers decreased HIV-1 Gag protein expression in virus-producing cells, while silencing the m6A erasers showed the opposite functions [141]. Since then, there are more than 80

papers on viral m6A have been published. Interestingly, RNAs of many RNA viruses also contain

m6A methylation, even though they replicate entirely in the cytoplasm and don’t have access to

46 the nucleus. It is concluded that viral RNA m6A methylation can play either proviral or antiviral roles in virus life cycle, which is dependent on specific virus types, and in some cases, cell types and/or replication stages (Table 1.3.).

Virus type Molecule Change Biological function Reference

HIV METTL3/METTL14 Down Suppress virus replication [142]

ALKBH5 Down Promote virus replication

YTHDF1–3 Up Promote virus replication [142] [143]

Suppress virus replication [141]

EV71 METTL3 Down Decrease in virus titer [144]

FTO Down Promote in virus titer

YTHDF2/3 Down Suppress virus replication

IAV METTL3 Down Suppress IAV replication [145]

YTHDF2 Up Promote IAV replication

KSHV METTL3 Down Suppress viral lytic replication [146]

FTO Down Promote viral lytic replication

YTHDC1 Up Promote ORF50 pre-mRNA splicing

YTHDF2 Down Suppress transcription in ORF50 [147]

HBV METTL3/14 Down Promote HBc/s protein expression and [148] the half-life of pgRNA

FTO/ALKBH5/YTHDF2–3 Down Suppress HBc/s protein expression and the half-life of pgRNA

HCV METTL3/14 Down Enhance titer of HCV [149]

FTO Down Decrease viral titer

YTHDF1–3 Up Suppress HCV replication

ZIKV METTL3/14 Down Increase viral titer [150]

ALKBH5/FTO Down Decrease viral titer

47

Virus type Molecule Change Biological function Reference

YTHDF2 Up Suppress ZIKV replication

SV40 METTL3 Down Reduce SV40 replication [151]

YTHDF2 Up Enhance replication of SV40

EBV METTL14 Up Promote growth and proliferation of [152] EBV-infected cells Table 1.3. The regulators of m6A in different viruses.

1.7.4.1. m6A on RNA viruses

(1) HIV. During HIV-1 replication, viral mRNAs are subject to a non-canonical form of nuclear export, splicing and cap-dependent and independent forms of translation. Given the fact that m6A play roles in those process, it is conceivable that the m6A is also involved in the epitranscriptomic regulation of HIV-1 gene expression. Prior to this dissertation, there are four reports have studied the potential functions of m6A in HIV infection. In these studies, the groups are unconformity in the locations, effects, and mechanisms of m6A in HIV-1 RNA, despite the general agreed-upon conclusions (Table 1.4.).

In 2016, there are three studies investigated the roles of m6A on HIV-1 infection. By utilizing different bioinformatic methods, all three studies reported shared enriched m6A modifications at

3’ 1.4 kb of the 9.2 kb of the HIV-1 RNA genome. Lichinchi and colleagues identified 14 m6A methylation peaks in both the 5’ and 3’ UTRs, coding sequences and splicing regulatory sequences in HIV-1 genomic RNA (gRNA) [142]. Kennedy et al. found 0-2 additional m6A in 3’ UTR, while

Tirumuru and colleagues only reported one further m6A peak at 5’ UTR. To address the functions of m6A in HIV-1 infection, Lichinchi and colleagues quantified the RNA levels of HIV-1 envelope glycoprotein GP120 and protein level of viral capsid protein p24 at 72 h post-infection under the

48

depletion of METTL3, METTL14 and ALKBH5. It was found that knockdown of writer protein

lead to the decreased GP120 and p24 levels. Conversely, a dramatic increase in GP120 and p24

was seen in ALKBH5-depleted cells. Similar as reported in study of Lichinchi et al., Kennedy and colleagues observed enhanced production of the HIV-1 mRNAs , Tat and Rev, as well as increased protein levels of p55 Gag, p24 and Nef when overexpressed the reader proteins

YTHDF1–3. And their further work showed the reader proteins bind m6A residues within the 3′

UTR of HIV-1 transcripts and enhance expression of viral mRNAs in cis [143]. In agreement with

the findings of Lichinchi et al. and Kennedy et al. on writer and eraser proteins, Tirumuru et al.

depleted writer protein METTL3/METTL14 finding a reduction in protein expression of p55 Gag

and p24 and the knockdown of eraser protein FTO/ALKBH5 showed the opposite effect.

Interestingly, overexpression of m6A reader proteins YTHDF1-3 in HeLa cells was found to inhibit

HIV-1 infection and downregulation of Gag protein, whereas around 10-fold elevation in HIV-1

infectivity was observed when YTHDF1-3 were knockdown [141]. These contradictory results

associated with modulation of reader protein was argued as that modified HIV-1 strain with firefly

luciferase was used by the Tirumuru and colleagues. However, in the later work of Lu et al. using

wild type HIV-1 excluded the influence from methylation of firefly luciferase RNA [153].

Together, these studies suggest m6A play profound regulation roles in HIV-1 replication despite

some differences.

49

Key conclusions Mechanisms of action Main m6A sites Cell types used m6A Reference detection technologies

m6A modification, The m6A abundant HIV-1 3’UTR Human CD4+ PA-m6A-seq; [143] and the resultant sites recruit the cellular CEMSS T-cells PAR-CLIP recruitment of YTHDF m6A “reader” infected with HIV-1 YTHDF proteins, are proteins to enhance NL4.3 genome, HIV- major positive HIV-1 protein and 1-expressig 293T regulators of HIV-1 RNA expression, and cells mRNA expression. virus replication

m6A modification of YTHDF1–3 proteins 5’UTR, 3’UTR HIV-1-NL4.3 m6A-seq; [141] HIV-1 RNA increase inhibit HIV-1 infection and several infected Jurkat cells, CLIP-seq; HIV-1 Gag protein by blocking viral internal positions primary CD4+ T- LC-MS expression; YTHDF reverse transcription of HIV-1 cells, HEK293T cells proteins inhibited and promoting and HeLa cells HIV-1 post-entry degradation of viral infection RNA

The addition of m6A The presence of m6A In coding and MT4 T-cells infected MeRIP-seq [142] group in HIV-1 favor the binding of non-coding with HIV-1 LAI gRNAs enhance HIV- Rev to the RRE in regions, splicing strain, HEK293T 1 infection and viral HIV-1 infected cells junctions, and cells replication splicing regulatory sequences of HIV-1

YTHDF1-3 proteins m6A reader proteins Undetected HeLa or CD4+ cells None [153] inhibit HIV-1 YTHDF1–3 inhibit overexpressing each infection and viral HIV-1 infection by YTHDF protein production decreasing viral gRNA infected with HIV-1 and early reverse NL4.3, HEK293T transcription products cells Table 1.4. Comparison of m6A functions in HIV-1 life cycle.

(2) Influenza virus. Although it is known that influenza virus RNA contains internal m6A methylation in 1985, the role of m6A methylation in influenza virus replication remains unknown until 2017. Courtney and colleagues mapped epitranscriptomic m6A modifications on both the

IAV mRNA and vRNA strands, and demonstrated that with m6A on its RNA, IAV is able to replicate and transcript efficiently. Inhibition m6A modifications by either using non-toxic dose of

DAA treatment (an inhibitor of m6A addition), or knockout of METTL3 showed a reduction in

IAV protein NS1 and M2 expression. Moreover, IAV with depletion of m6A on both strands of 50 the hemagglutinin (HA) segment were attenuated in pathogenicity in mice. Those findings confirmed the positive roles of m6A in IAV gene expression [145].

(3) . The are a family of positive, single-stranded, enveloped RNA viruses, which include Yellow Fever virus (YFV), Dengue Fever virus (DFV), Japanese encephalitis virus (JEV), West Nile virus (WNV), Zika virus (ZIKV), and virus (HCV).

There are two publications in late 2016 providing unexpected evidence of m6A modulate HCV and ZIKA virus. Gokhale et al. carried out m6A mapping and analyses in cells infected with HCV and identified 19 peaks in the total 9.6 kb HCV RNA genome. siRNA mediated depletion of

METTL3 and METTL14 in Huh7 cells significantly enhanced the abundance of HCV NS5A protein, whereas depletion of FTO lead to the decrease in HCV NS5A levels. In addition, the authors revealed that the relocalization of reader proteins at viral assembly sites retained m6A modified HCV RNA, thus reducing HCV particle production. To date, only one research article was published reporting the relevance of ZIKV to m6A modification [149]. Lichinchi et al. mapped

ZIKA virus and identified twelve prominent m6A peaks in ZIKA RNA and half were located in the region encoding the NS5 protein and the 3′ UTR region. Similar as found in HCV, the depletion of writer proteins enhanced ZIKV production while silencing of erasers proteins inhibited ZIKA infection. Different as seen in HCV that m6A machinery decreased HCV production by affecting the assembly of virus, m6A influent ZIKV production by impacting replication of virus [150].

(4) EV71. Enterovirus 71 (EV71) is a single-stranded RNA virus that belongs to Picornaviridae family and is one of the top pathogens that causes hand-foot-and-mouth disease (HFMD). Recent study form Hao et al. demonstrated that EV71 is m6A modified and viral infection drives the alteration of expression and localization of m6A related proteins. Silencing METTL3 significantly

51 reduced the genomic copy numbers of EV71 RNA and increased by FTO gene depletion. In addition, the authors revealed that overexpression of METTL3 could enhance sumoylation and ubiquitination of viral RdRp 3D protein thus increasing self-stability of 3D and facilitate EV71 replication. Taken together, m6A modifications positively regulate EV71 replication [144].

1.7.4.2 m6A on DNA viruses

(1) KSHV. Kaposi’s sarcoma-associated herpesvirus (KSHV) is a double stranded DNA virus that undergoes distinct latent and lytic life cycles. To date, KSHV was found to contain m6A on its transcripts and it showed different effects on KSHV life cycle owing to different cell types by three different groups. Ye and colleagues found the enhanced expression of lytic gene ORF50 and

ORF57 when inhibiting the FTO either by knockdown or using meclofenamic acid (MA). While blocking m6A or silencing METTL3 abolished two lytic genes as well as virion production. Their further study on the pre-mRNA of a KSHV lytic switch gene ORF50 demonstrated the positive function of m6A in regulating ORF50 (RTA) RNA splicing through binding of the YTHDC1 [146].

In another report, Hesser et al. revealed a proviral manner of m6A in KSHV production in iSLK.219 and iSLK.BAC16 cells by modulate the writer and reader protein. Silencing METTL3 and YTHDF2 dramatically impaired generation of progeny virus. Interestingly, in KSHV infected

B cells, depletion of METTL3 enhanced expression of ORF50 protein which suggests that m6A regulating KSHV replication in a cell-type dependent manner [147]. In the third study conducted by Tan and colleagues, it was found that m6A modifications on KSHV are highly conserved among different cell types and infection systems. Consistent with findings in work of Hesser et al., the authors found the inhibition of YTHDF2 in KSHV lytic replication by facilitating the degradation of viral lytic transcripts. The mechanism insights provided by this work greatly helped enriching

52 the knowledge in the changes of viral and cellular m6A modifications during KSHV latent and lytic infection [154]. In sum, m6A modifications in KSHV is nonequivalent owing to different cell types during lytic replication.

(2) SV40. As another well studies DNA virus, simian virus 40 (SV40) was reported to be positively regulated by m6A but it is phase dependent. The distribution of m6A on SV40 transcripts in the

“late” region was reported in 1979, but its location and functions were identified until recently. In

2018, Tsai et al. found that overexpression of YTHDF2 speeded up the viral replication as well as enlarged the plaque size in BSC40 cells infected with SV40, while blocking endogenous YTHDF2 or METTL3 brought the opposite effect. Those results indicate a provial role of m6A in SV40 life cycle. In addition, by introducing mutations to sites of m6A residues on the viral early and late transcripts, the authors determined that the m6A modifications on the late SV40 transcripts played the key role in viral gene expression and replication [151].

(3) HBV. virus (HBV) infection leads to the chronic hepatitis. Prominently, m6A was found in both HBV transcripts in patients liver tissues and HBV-expression cells. Imam et al. observed that m6A located in HBV 3’ UTRs negatively affecting the expression of their corresponding proteins by destabilizing these RNAs. Under the condition that lacking of METTL3 and METTL14 or YTHDF2, Imam and colleagues measured the stability of HBV transcripts and they observed a two-fold increase in the half-life of pgRNA components. In addition, they also measured the core-associated DNA levels which indicates the reverse transcription of HBV pgRNA in cells missing METTL3 and METTL14 or FTO and significantly reduced reverse transcription was observed in cells lacking writer protein but enhanced in cells missing FTO. All those results suggest that the modulating function of m6A is mediated by YTHDF proteins either

53 through impacting the stability of HBV transcripts or pregenomic RNA reverse transcription. By conducting m6A-seq on uninfected and HBV-expressing hepatocytes and Imam and colleagues were able to identify a single m6A peak at position A1907 in the HBV genome and other m6A sites within a 3′ epsilon stem loop present in all HBV transcripts. 3’ m6A-defecient mutants pgRNA was assessed to be more stable than wild type counterpart. And the A1907C mutation was found to cause a mismatch within the epsilon stem loop structure. Therefore, deficient in m6A might alter the structure of RNA which impact on RNA stability and reverse transcription. As a result, Imam et al. proved that methylation of A1907 in HBV RNA affects the virus life cycle, and m6A regulates HBV RNAs in more than one way, hinging on its position in the RNA [148].

54

Chapter 2. Rational design of human respiratory syncytial virus live attenuated vaccine

candidates by alteration of the S-adenosylmethionine (SAM) binding site in the large

polymerase protein

2.1 Abstract

S-adenosylmethionine (SAM) binding site in the large polymerase protein of RSV has been shown critical for RSV transcription and replication. Here, we generated three live attenuated RSV vaccine candidates by alteration of S-adenosylmethionine (SAM) binding site in the methyltransferase (MTase) region of the large (L) polymerase protein of RSV. Recombinant rgRSVs carrying a single (G1853A or G1857A) or a double mutation (G1857A-G1853A) in the

SAM binding site were genetically stable and grew to high titers in cultured cells. All three recombinant viruses were highly attenuated for replication in primary, well differentiated HBE cultures, and upper and lower respiratory tracts of cotton rats. All three attenuated strains elicited comparable levels of neutralizing antibody to the parental rgRSV in cotton rats. Despite high attenuation in vitro and in vivo, all immunized cotton rats were completely protected against RSV infection in both lungs and nasal turbinates. Taken together, mutations in the SAM binding site of

L protein represent a promising approach to generate live attenuated vaccine candidates for pneumoviruses.

55

2.2 Introduction

Human respiratory syncytial virus (RSV) is the most important cause of upper and lower

respiratory tract infection of infants, young children, the elderly, and immunocompromised

individuals. Despite tremendous efforts, no vaccine or antiviral drug is yet available for RSV.

Because an inactivated RSV vaccine was not safe for humans, research efforts quickly switched

to the development of a live attenuated RSV vaccine. Several approaches have yielded promising

live attenuated RSV vaccine candidates which are currently in preclinical or clinical trials [155-

157]. Although some of these candidates are promising, exploration of new approaches to

attenuate RSV for vaccine purposes is still needed.

Inhibition of viral mRNA cap methylation is a novel approach to attenuate RNA viruses. Since the

early 1970s, it was known that mRNA cap G-N-7 methylation is required for recognition of the

cap by the rate-limiting factor for translation initiation, eIF-4E [58, 158]. However, it was not until

2012 that it was learned that ribose 2’-O methylation of the mRNA cap serves as a molecular

signature for discrimination of self and non-self RNA by the innate immune system, and that it protects RNAs from decapping and degradation [60, 61, 158]. Thus, inhibition of mRNA cap methylation will likely reduce RNA stability, decrease protein translation, and initiate innate immune detection, all of which would result in virus attenuation. For non-segmented negative- sense (NNS) RNA viruses (e.g. VSV, RSV, and hMPV), the mechanism of mRNA cap methylation is different from the host cell process in that a single MTase active site in CR VI of L protein is essential for both G-N-7 and ribose 2’-O MTases [57, 159, 160], and these two MTases share a

single SAM binding site [161]. Previously, we showed that mutations in the MTase active site

(KDKE motif) resulted in recombinant VSVs with much more dramatic attenuation than mutations

56 in the SAM binding site [62]. This is due to the fact that mutations in the KDKE motif abolished both G-N-7 and 2’-O methylation whereas mutations in the SAM binding site retained partial

MTase activity, either G-N-7 and/or 2’-O methylation [159, 161]. As a consequence, recombinant viruses carrying mutations in the KDKE motif were highly attenuated and grew poorly in cell culture, which limited their potential use as live vaccine candidates [62]. In contrast, recombinant viruses carrying mutations in the SAM binding site typically replicated to high titers despite high attenuation both in vitro and in vivo [62]. This has been demonstrated in several NNS RNA viruses including VSV [62], hMPV [64], avian metapneumovirus (aMPV) [162], and measles virus [163].

We hypothesize that recombinant RSVs (rRSVs) with mutations in the SAM binding site will be good live attenuated vaccine candidates. For this purpose, amino acids in the SAM binding site of RSV L protein were mutated to alanine and rRSVs were recovered. Two recombinant viruses (rgRSV-G1853A and G1857A) grew to high titers in Vero cells whereas two other viruses (rgRSV-G1855A and D1972A) had significant defects in growth. Subsequently,

G1853A and G1857A were combined and an rRSV with double mutations (rgRSV-G1853A-

G1857A) were generated. The single or double RSV mutants were genetically stable, replicated to reasonably high titer in cell culture, were significantly attenuated in primary well differentiated human bronchial epithelial (HBE) culture and cotton rats yet retained high immunogenicity.

2.3 Materials and Methods

2.3.1 Cell cultures

HeLa (ATCC CCL-2), A549 (ATCC CCL-185), Vero (ATCC CRL-CCL81), Vero E6 cells

(ATCC CRL-1586), and HEp-2 (ATCC CCL-23) cell lines were purchased from the American

57

Type Culture Collection (Manassas, VA) and were grown in Dulbecco's modified Eagle's medium

(DMEM; Life Technologies) supplemented with 10% FBS. All cell lines used in this study were

free of mycoplasma, as confirmed by the LookOut Mycoplasma PCR Detection Kit (Sigma).

Primary, well-differentiated human bronchial epithelial (HBE) cultures were grown on collagen coated Transwell inserts (Corning Incorporated, Corning, NY), as previously described (38). Upon reaching confluency and forming tight junctions, the apical medium was removed and the basal medium was replaced with Pneumacult (Stem Cell Technologies, Inc.) to stimulate differentiation at the air-liquid interface over the following 3 weeks.

2.3.2 Virus preparation

Recombinant RSV (rRSV) A2 strain and recombinant RSV A2 strain containing a green fluorescence protein (GFP) gene between the leader sequence and NS1 gene (rgRSV) was propagated and titrated in HEp-2 cells. The sequences of rRSV and rgRSV are identical except that rgRSV expresses GFP gene. For infection experiments in HBE culture and cotton rats, all

RSV stocks were purified. Briefly, 10 T150 flasks of HEp-2 cells were infected by RSV at an MOI of 0.1, and culture supernatant and cells harvested at 72 h post-infection were clarified by centrifugation at 10,000 ×g for 30 min. Virus was concentrated through a 35% (wt/vol) sucrose cushion by centrifugation at 30,000×g for 2 h at 4°C in a Ty 50.2 rotor (Beckman). The pellet was resuspended in DMEM with 10% trehalose. Viral titer was determined by TCID50 assay.

2.3.3 Plasmids, site-directed mutagenesis and recovery of RSV

Plasmid (RW30) encoding the full-length antigenomic cDNA of RSV strain A2 with GFP inserted between the leader and the NS1 gene, and support plasmids expressing RSV A2 strain N protein

(pTM1-N), P protein (pTM1-P), L protein (pTM1-L), and M2-1 protein (pTM1-M2-1) were

58

generously provided by Dr. Peter Collins, NIAID, Bethesda, MD. The L mutants in MTase active

site (K1831A, D1936A, K1973A, and D2004A) and SAM binding site (G1853A, G1855A,

G1857A, and D1912A) were introduced into the pTM1-L and RW30 plasmids using QuikChange

site-directed mutagenesis kit (Stratagene, La Jolla, CA). All plasmids and mutations were

confirmed by DNA sequencing. To recover RSV, HEp-2 cells were infected with MVA-T7 at an

MOI of 10, then transfected with 1.2 µg of plasmid RW30 or RW30 mutant, 0.4 µg of pTM1-N,

0.2 µg of pTM1-P, 0.1 µg of pTM1-M2-1, and 0.1 µg of pTM1-L using the Lipofectamine 3000 reagent (Life Technologies). At day 4 post-transfection, the cells were harvested using scrapers and were co-cultured with new flask of HEp-2 cells at 50 to 60% confluence. When an extensive

cytopathic effect (CPE) was observed, the cells were subjected to three freeze-thaw cycles,

followed by centrifugation at 4,000×g for 10 min. The supernatant was subsequently used to infect

new HEp-2 cells. The successful recovery of the rgRSV was confirmed by the presence of green

fluorescent cells, followed by RT-PCR and sequencing. Recombinants rgRSVs carrying mutations

in the MTase active site were named rgRSV-K1831A, D1936A, K1973A, and D2004A.

Recombinants rgRSVs carrying mutations in the SAM binding site were named rgRSV-G1853A,

G1855A, G1857A, and D1912A. Recombinant rgRSV carrying double mutations was named rgRSV-G1853A-G1857A.

2.3.4 RSV minigenome assay

The minigenome assay was performed in HEp-2 cells as described previously. Briefly, confluent

HEp-2 cells were infected with vaccinia virus MVA-T7 (kindly provided by Dr. Bernard Moss,

NIAID, Bethesda, MD) at an MOI of 10, followed by transfection with 0.2 µg of the minigenome plasmid together with 0.4 µg of pTM1-N or pTM1-N mutant, 0.2 µg of pTM1-P, 0.1 µg of pTM1-

59

M2-1, and 0.1 µg of pTM1-L using the Lipofectamine 3000 reagent (Life Technologies). At 6 h

post-transfection, the medium was replaced with DMEM containing 2% FBS. At day 2 post-

transfection, expression of GFP in the transfected cells was visualized by fluorescence microscopy

and quantified by flow cytometry.

2.3.5 RT-PCR and sequencing

All plasmids, viral mutants and stocks, and virus isolates from the nasal turbinates and lungs of

cotton rats were sequenced to confirm virus identity. Viral RNA was extracted from 100 µl of each

recombinant virus or nasal and lung homogenates using an RNeasy minikit (Qiagen, Valencia,

CA). A 1.5-kb DNA fragment spanning the CR-VI of the L gene was amplified by RT-PCR. Also, the entire RSV genome was amplified using six overlapping fragments by RT-PCR. The PCR products were purified using Qiagen PCR Purification kit and sequenced using a sequencing primer at The Ohio State University Plant Microbe Genetics Facility to confirm the presence of the desired mutations.

2.3.6 Characterization of recombinant RSV in cell culture

GFP is a good indicator for rgRSV replication, thus we monitored the fluorescence of each recombinant virus at indicated time points. Vero or HEp-2 cells were infected with rgRSV or mutants at an MOI of 0.1 or 1, and GFP expression was monitored at the indicated times by fluorescence microscopy. At the indicated time points, cells were trypsinized and fixed in 4 % of paraformaldehyde solution and the number of GFP-positive cells quantified by flow cytometry.

For Viral replication kinetics, confluent HEp-2 cells in 12-well-plate were infected with wild-type rgRSV or mutant rgRSV at an MOI of 1.0. After 1 h of adsorption, the inoculum was removed and the cells were washed three times with DMEM. Fresh DMEM (supplemented with 2% FBS) was

60

added, and the infected cells were incubated at 37°C. At different time points post-inoculation, the supernatant and cells were harvested by three freeze-thaw cycles, followed by centrifugation at

1,500 × g at 4 oC for 15 min. The virus titer was determined by TCID50 assay in HEp-2 cells. For analyses RSV protein expression, confluent A549 cells were infected with rgRSV and rgRSV mutants at an MOI of 0.1. At 18, 24, and 48h postinfection, the cell culture supernatant was removed and the cells were lysed in 150 μl of RIPA buffer (Abcam) supplemented with protease inhibitor cocktail (Sigma-Aldrich). One hundred microliters of the cell lysate were denatured at

99°C for 10 min and analyzed on a 10% polyacrylamide bis-Tris gel. The separated protein was transferred to a Hybond-P polyvinylidene difluoride membrane (Amersham Biosciences,

Pittsburgh, PA) using a Trans-Blot SD semidry transfer cell (Bio-Rad, Hercules, CA). Membranes were blocked with 5% skim milk in PBST and subsequently probed with anti-RSV serum

(Virostat), F (Abcam) or anti-β-actin (Proteintech) diluted in PBST-milk, followed by incubation with horseradish peroxidase-conjugated anti-mouse IgG monoclonal antibody (Thermo Scientific) diluted to 1:5,000 in PBST-milk. Membranes were developed with a chemiluminescence substrate

(Thermo Scientific) and exposed to Biomax MR film (Kodak) for visualization of the proteins.

Actin was used as a loading control.

2.3.7 Replication, spread, and cytokine production in HBE culture

Purified virions were titrated on HEp-2 cells and were diluted in DMEM. The apical surface of well-differentiated HBE cells in Transwells was washed 5 times with 100 µl DMEM and the basal medium was changed before adding the GFP-expressing rgRSV virus and its mutants (400

TCID50) to the apical surface. Fluorescent cells were visualized and photographed with an EVOS2 fl inverted fluorescence microscope (Life Technologies). The amount of the HBE culture

61

expressing GFP-positive cells was determined from a digital image using Image J Software. At

indicated times, 100 µl of DMEM was added to the apical surface of HBE cultures, gently rocked

for 30 min, and collected to quantify virus release and apically released cytokines and chemokines.

The 0.5 ml of medium on the basal side of the filter was collected to assess basally released

cytokines and chemokines.

2.3.8 Genetic stability of rgRSV mutants in cell culture

Ninety percent confluent Vero cells in T25 flasks were infected with each rgRSV mutant at an

MOI of 0.1. When extensive CPE was observed (day 3-5 post-inoculation), the cell culture supernatant was harvested and used for the next passage in Vero cells. Each rgRSV mutant was repeatedly passaged 12 times in Vero cells. At each passage, a 1.5 kb DNA fragment covering CR-

VI of the L gene was amplified by RT-PCR and sequenced. At passage 12, the entire genome of each recombinant virus was amplified by RT-PCR using six overlapping fragments and sequenced.

2.3.9 Replication and immunogenicity of rgRSV mutants in cotton rats

Twenty-five 6-week-old specific-pathogen-free (SPF) male cotton rats (Envigo, Indianapolis, IN) were randomly divided into 5 groups (5 cotton rats per group). Prior to virus inoculation, the cotton rats were anesthetized with isoflurane. The cotton rats in groups 1 to 2 were inoculated with

2.0×105 TCID50 of parental rRSV and rgRSV respectively. The cotton rats in groups 3 to 5 were inoculated with 2.0×105 TCID50 of three rgRSV mutants, rgRSV-G1857A, G1853A, and

G1853A-G1857A. Each cotton rat was inoculated intranasally with a volume of 100 μl. At day 4 post-infection, the cotton rats were sacrificed via carbon dioxide inhalation. The left lung and nasal turbinates were collected for virus titration and the right lung was collected for histological analysis.

62

For short-term immunogenicity study, cotton rats were challenged with rgRSV at week 4 post-

immunization. Briefly, twenty-five six-week-old female cotton rats (Envigo) were randomly divided into five groups (5 cotton rats per group). Cotton rats in groups 1-4 were intranasally inoculated with 2.0×105 TCID50 of three rgRSV mutants (rgRSV-G1857A, G1853A, and

G1853A-G1857A) and rgRSV, respectively. Cotton rats in groups 5 were mock-infected with PBS

and served as unvaccinated challenged control. After immunization, the cotton rats were evaluated

daily for any possible abnormal reaction and blood samples were collected from each cotton rat

weekly by orbital sinus blood sampling, and serum was used for detection of neutralizing

antibodies. At 4 weeks post-immunization, the cotton rats in all groups were challenged with

2.0×105 TCID50 of parental rgRSV via intranasal route, and evaluated twice daily for the presence

of any abnormal reactions. At 4 days post-challenge, all cotton rats were euthanized by CO2 asphyxiation, and their lungs and nasal turbinates were collected for virus titration. The immunogenicity of rgRSV mutants was assessed based on their ability to trigger neutralizing antibody and the ability to prevent rgRSV replication in lungs and nose. For long-term

immunogenicity study, cotton rats were challenged with rgRSV at day 112 post-immunization.

Three viruses (rgRSV, rgRSV-G1857A, and rgRSV-G1853A) at two different doses (105 and 103

TCID50) were used in this experiment. After immunization, blood samples were collected from

each cotton rat biweekly for detection of neutralizing antibodies.

2.3.10 Determination of RSV-neutralizing antibody

RSV-specific neutralizing antibody titers were determined using a plaque reduction neutralization

assay (35). Briefly, cotton rat sera were collected by orbital sinus blood sampling weekly until

challenge. The serum samples were heat inactivated at 56°C for 30 min. Twofold dilutions of the

63

serum samples were mixed with an equal volume of DMEM containing approximately 50

TCID50/well rgRSV in a 96-well plate, and the plate was incubated at room temperature for 1 h

with constant rotation. The mixtures were then transferred to confluent HEp-2 cells in a 96-well

plate in triplicate. After 1 h of incubation at 37°C, the virus-serum mixtures were removed and the

cells were overlaid with 0.75% methylcellulose in overlay media (1× MEM, 2% FBS, Sodium

bicarbonate, 25mM HEPES, 1% L-Glutamine, 1% Pen Strep) and incubated for another 3 days

before counting the fluorescent foci. The numbers of foci at each serum dilution were plotted and

the 50% plaque reduction titer was used as the RSV-specific neutralizing antibody titer.

2.3.11 Statistical analysis

Quantitative analysis was performed by either densitometric scanning of autoradiographs or by

using a phosphorimager (Typhoon; GE Healthcare, Piscataway, NJ) and ImageQuant TL software

(GE Healthcare, Piscataway, NJ). Statistical analysis was performed by one-way multiple

comparisons using SPSS (version 8.0) statistical analysis software (SPSS Inc., Chicago, IL). A P

value of <0.05 was considered statistically significant.

2.4 Results

2.4.1 Mutations in the MTase active site and SAM binding site in the CR-VI of RSV L protein

are functional in a minigenome assay.

The CR-VI of the L proteins of NNS RNA viruses encodes both mRNA cap MTase activities.

Sequence alignment clearly showed that a K-D-K-E tetrad, which functions as the MTase catalytic

site, is conserved in the CR-VI of the L proteins of cytoplasmic replicating NNS RNA viruse. The

corresponding amino acid residues in RSV L are K1831, D1936, K1973, and D2004. The SAM- dependent MTase superfamily typically contains a conserved G-rich motif for binding the SAM

64

molecule, the methyl donor for RNA methylation. Sequence analysis found that aa residues

corresponding to the GxGx…D motif of the RSV L protein includes G1853, G1855, G1857, and

D1912. Thus, these amino acid residues in the MTase active site and SAM binding site were

individually mutated to alanine in a plasmid encoding RSV L gene (pTM1-L). A total of eight

RSV L mutants were generated. A minigenome assay was performed to examine whether these L-

protein mutants were functional. The minigenome (pMINI-RSV-GFP) RNA contains the RSV leader at its 3’ end and trailer sequence at its 5’ end. It also contains the RSV-specific gene start (gs) and gene end (ge) transcription control elements flanking the GFP gene for GFP

mRNA expression. HEp-2 cells were infected with a weakened vaccinia virus expressing T7 RNA

polymerase (MVA-T7) at an MOI of 10, followed by transfection with the RSV minigenome

(pMINI-RSV-GFP) together with support plasmids pTM1-N, pTM1-P, pTM1-M2-1, pTM1-L, or

pTM1-L carrying a mutated L protein. At 48 h postinfection, the cells expressing GFP were imaged

by fluorescence microscopy. Wild type pTM1-L had strong GFP expression. No GFP signal was

detected when pTM1-L was not included in the transfection (Figure 2.1). All pTM1-L carrying

mutations in the MTase active site and SAM binding site decreased GFP expression (Figure 2.1).

These results showed that all RSV L mutants were functional but had defects in minigenome

replication.

65

Figure 2.1 Examination of the function of L mutants using a minigenome assay.

66

Confluent HEp-2 cells were transfected with the minigenome plasmid together with pTM1-N, pTM1-P, pTM1-M2.1, pTM1-L, or an pTM1-L mutant using Lipofectamine 3000. GFP expression was visualized by fluorescence microscopy and quantified by flow cytometry at 48 h post- transfection. Error bars represent SD from n = 3 independent experiments. *P<0.5, **P < 0.01,

***P < 0.001, ****P < 0.0001.

2.4.2 Recovery of rgRSV carrying mutations in the SAM binding sites

Since all L mutants were functional, we next introduced all these mutations individually into plasmid RW30 which encodes the full-length antigenomic cDNA of RSV strain A2 with GFP inserted between the leader and the NS1 gene. The inserted GFP reporter gene is a useful tool for monitoring RSV infection and spread as a measure of attenuation. Using the reverse genetics system, we recovered all eight recombinant viruses. Green cells were readily detected when these recombinant viruses were passaged in HeLa cells. However, we realized that four rgRSV mutants in the MTase active site and two rgRSV mutants (G1855A and D1912A) in the SAM binding site had approximately 2 log defects in growth compared to the parental rgRSV, even after 6-7 passages in cell culture. Thus, we decided not to pursue these mutants further due to the low viral titer.

However, two rgRSV mutants (G1853A and G1857A) in the SAM binding site had comparable titers to rgRSV.

Genetic instability has been one of major issues in the development of an RSV live attenuated vaccine. Thus, we decided to combine these two mutations (G1853A and G1857A) in order to reduce the chances of reversion to the unattenuated phenotype. An L gene carrying both mutations

(pTM1-L-G1853A and G1857A) was generated and tested in the minigenome assay and found to be functional (Figure 2.1). Recombinant rgRSV carrying the double mutation (rgRSV-G1853A- 67

G1857A) was recovered and found to replicate to a high titer during passage in cell culture.

Subsequently, each rgRSV mutant was plaque purified, and a large stock of virus was prepared.

To test whether these rgRSV mutants were genetically stable, we successively passed these recombinant viruses in Vero cells. At passage 12, the entire genome was amplified by RT-PCR and sequenced. Sequencing results showed that all rgRSV mutants contained the desired mutation in CR-VI of the L-protein gene and no additional mutations were found in the genome indicating that these rgRSV mutants were genetically stable in cell culture.

2.4.3 Recombinant rgRSV carrying mutations in the SAM binding site are attenuated in cell culture

We next compared the replication kinetics of these rgRSV mutants with the parental rgRSV in

Vero and HEp-2 cells. Briefly, Vero and HEp-2 cells were infected with each recombinant virus at an MOI of 0.1 or 1. At the indicated time points, the spread of the GPF signal was recorded, the supernatant and cells were harvested and the virus titer was determined by TCID50 assay. The spread of GFP expression and CPE were both significantly delayed in rgRSV mutants in both Vero

(Figure 2.2A) and HEp-2 (Figure 2.2B) cells. For rgRSV, most cells were green at 48 h post- inoculation and cells were killed by 72 post-inoculation. However, for the mutants, maximal green signal was observed at 60 h post-inoculation but cells survived until 84 post-inoculation.

Quantification of GFP-expressing cells by flow cytometry showed significantly fewer GFP- positive cells in cells infected by the rgRSV mutants than cells infected by the rgRSV parent

(Figure 2.2C). All three rgRSV mutants had significant delays in viral replication compared to the parental rgRSV (P < 0.05) (Figure 2.2D). Parental rgRSV reached peak titer (6.86±0.14

68 log10PFU/ml) at 60 h post-inoculation (hpi) whereas all rgRSV mutants reached a similar peak titer but at 72 h post-inoculation.

We next determined viral protein synthesis in virus-infected cells. Briefly, A549 cells were infected by each rgRSV mutant, cell lysates were harvested at 12, 24, and 48 h post-infection and viral proteins were analyzed by Western blot. All mutants had a significant delay in viral G and F protein expression compared to the parental rgRSV (Figure 2.3). At 48h post-infection, rgRSV-

G1857A and G1853A expressed similar amounts of viral protein whereas rgRSV-G1857A-

G1853A expressed less, compared to the parental rgRSV. These results showed that rgRSV carrying mutations in the SAM binding site were significantly attenuated in cell culture. The degree of attenuation is ranked as following: rgRSV-G1853A-G1857A > rgRSV-G1853A > rgRSV-G1857A.

69

Figure 2.2. Characterization of rgRSVs carrying mutations in the SAM binding site of the L protein.

70

(A) Delayed GFP expression by rgRSV mutants in Vero cells. Confluent Vero cells were infected by each rgRSV at an MOI of 0.1, and GFP expression was monitored at the indicated times by fluorescence microscopy. (B) Delayed GFP expression by rgRSV mutants in HEp-2 cells.

Confluent HEp-2 cells were infected by each rgRSV at an MOI of 0.1, and GFP expression was monitored. Only rgRSV-G1857A was shown. (C) Quantification of GFP positive cells by flow cytometry. Confluent HEp-2 cells were infected by each rgRSV (MOI of 1.0), at indicated time point, cells were trypsinized and GFP positive cells were quantified by flow cytometry. Data are the average of three independent experiments ± standard deviation. (D) Single-step growth curve of rgRSV mutants. HEp-2 cells in 12-well-plates were infected with each recombinant rgRSV at an MOI of 1.0. After adsorption for 1 h, the inocula were removed and the infected cells were washed 3 times with Opti-MEM medium. Fresh DMEM medium containing 2% FBS was added and the cells were incubated at 37°C for various times. The supernatant and cells were harvested by three freeze-thaw cycles, followed by centrifugation at 1,500 × g at 4oC for 15 min at the indicated intervals. The viral titer was determined by TCID50 assay in HEp-2 cells. Viral titers are the geometric mean titer (GMT) of three independent experiments ± standard deviation. Error bars represent SD from n = 3 independent experiments. *P<0.5, **P < 0.01, ***P < 0.001,

****P < 0.0001.

71

Figure 2.3. Viral protein expression by rgRSV mutants.

A549 cells were infected with the parental rgRSV or rgRSV mutants at an MOI of 0.1. At 18, 24, and 48 h post-inoculation, total cell lysates were harvested and subjected to Western blotting using a monoclonal antibody against RSV F or G protein.

2.4.4 Interferon β (IFN-β) induction by rgRSV mutants in A549 cells

It was previously shown that deficient in mRNA cap methylation stimulated a higher type I interferon [61]. Here we compared IFN-β production of rgRSV mutants with rgRSV.

Briefly, A549 cells were infected with each recombinant virus at an MOI of 1.0, IFN-β level in supernatants were measured at days 1-4 post-infection. The peak level of IFN-β responses occurred at day 2 post-infection for all recombinant viruses (Figure 2.4). There was no significant difference among rgRSV-G1853A-G1857A, rgRSV-G1857A, and rgRSV (P>0.05) (Figure 2.4). However, rgRSV-G1853A had significantly higher IFN-β at days 2 and 3 post-infection compared to rgRSV

(P<0.05) (Figure 2.4).

72

Figure 2.4. Interferon β (IFN-β) production of rgRSV mutants in A549 cells.

Confluent A549 cells in 12-well-plates were infected with each recombinant virus at an MOI of

1.0, IFN-β level in supernatants were measured at days 1-4 post-infection. Error bars represent SD from n = 3 independent experiments. *P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

2.4.5 Recombinant rgRSV carrying mutations in the SAM binding site are attenuated in

replication and spread in primary well differentiated human bronchial epithelial (HBE)

cultures

HBE cultures are an ideal model to test the replication and spread of rgRSV mutants. These

cultures are pseudostratified and polarized, closely resembling the airway epithelium in vivo with 73

regard to morphology and function, including mucus production and ciliary motion [164].

Infection spreads from an infected ciliated cell to neighboring ciliated cells, usually in a counter-

clockwise fashion, due to the concerted ciliary beat, likely mimicking RSV infection and spread

in human airways [164]. HBE cultures were inoculated with 400 TCID50 of each rgRSV mutant,

and the dynamics of viral gene expression (GFP production) and spread was imaged by

microscopy. By the time individual rgRSV-G1857A mutant infected cells could be detected (36 hpi), the parental rgRSV had already spread to downstream cells, directed by the ciliary motion.

Parental rgRSV reached maximal spread by day 3 post-inoculation but rgRSV-G1857A did not until day 8 post-inoculation. Recombinant rgRSV-G1853A and rgRSV-G1853A-G1857A were the most delayed viruses, having few detectable green cells until day 3, and slow viral spreading even at day 10 post-inoculation (Figure 2.5A). The level of GFP expression for each virus is

quantified by Image J. The defect of spread is ranked as following: rgRSV-G1853A-G1857A >

rgRSV-G1853A > rgRSV-G1857A (Figure 2.5B).

The amount of infectious virus released from HBE culture was also monitored. Briefly, apical

washes were collected from HBE culture for virus titration by TCID50. As shown in Figure 2.6,

7.91 rgRSV released a high amount of virus in HBE culture with a peak titer of 10 TCID50 at day 6

post-inoculation. All three rgRSV mutants had significant delays and defects in virus release in

6.63 HBE culture. Recombinant rgRSV-G1857A had a peak titer of 10 TCID50 at day 8 post- inoculation whereas rgRSV-G1853A and rgRSV-G1853A-G1857A had a peak titer of 105.02

5.41 TCID50 and 10 TCID50 days 8 and 14 post-inoculation respectively. Taken together, these results

showed that rgRSV-G1853A and rgRSV-G1853A-G1857A were the most attenuated viruses and

rgRSV-G1857A had moderate attenuation in HBE culture.

74

Figure 2.5. rgRSV mutants spread more slowly in HBE cultures.

75

(A) Infection and spreading of rgRSVs in HBE cultures. HBE cultures were inoculated with 400

TCID50 of each rgRSV. At the indicated times, the entire transwell was imaged by fluorescence microscopy. A representative image from three transwells at each time point for each rgRSV are shown. (B) Quantification of GFP expression in infected HBE cultures. The GFP signal was quantified from a digital image by Image J software for each rgRSV-infected culture on the day they reached maximum infection. Fold increases in GFP intensity of each rgRSV mutant-infected

HBE culture, relative to mock-infected HBE were calculated, and data were expressed as the mean of three transwells ± standard deviation. *P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 2.6. rgRSV production in HBE cultures. HBE cultures were inoculated with 400 TCID50 of each rgRSV.

76

Apical washes were collected every 2 days until day 14 post-inoculation. Infectious virus in washes

was quantified by the TCID50 assay. Viral titers are the geometric mean titer (GMT) of three independent experiments ± standard deviation. *P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

2.4.6 Recombinant rgRSV carrying mutations in the SAM binding site induce the release of

different mediators from primary HBE cultures

During natural RSV infection or vaccination with a live attenuated RSV, cytokines or chemokines

may be induced and released from the airway epithelium. Some of these mediators may affect the immune response to the vaccine. To assess some of the more common mediators, we sampled the apical fluids. Multiple mediators were measured simultaneously with the Legendplex Human Anti-

Virus Response Panel (13-plex) bead assay. Because the mutants grew more slowly than the

parental rgRSV and because the levels of mediators released are very likely to correlate with the

number of infected cells in a well, we decided to assay the apical and basal surfaces at the point of

maximal RSV infection, determined from the images (Figure 2.5A), and their quantification

(Figure 2.5B).

Interestingly, a number of cytokines including IFN-ɣ, IL-10, IFN-α2, IL-1β, TNF-α, and IL-12 were significantly higher in all three rgRSV mutants than rgRSV (P<0.05) (Figure 2.7). Among the mediators assessed, IL-6 and GM-CSF stand out for being released more highly from HBE cultures infected with rgRSV-G1853A and rgRSV-G1853A-G1857A (P<0.05), both of which share the G1853A mutation and replicate and spread much less efficiently than the parental rgRSV or the single mutant, rgRSV-G1857A. Likewise, rgRSV-G1853A and rgRSV-G1853A-G1857A infected HBE released IL-6 basolaterally almost exclusively whereas the small amount of IL-6

77

released by the parental rgRSV infected HBE is released apically. Because the basal medium is

five times the volume of the apical medium, the amount of basal IL-6 and GM-CSF released into the basal medium would be multiplied by five for an accurate comparison. Conversely, much less interferon (IFN) λ2/3 is released from HBE infected with all three mutants than from HBE infected with rgRSV (P<0.05). Interestingly, IFN λ1 is released basally by HBE infected with all of these

viruses, and apically by rgRSV or rgRSV-G1857A infected HBE. But IFN λ1 is not released apically by either rgRSV-G1853A or rgRSV-G1853A-G1857A infected HBE. Finally, IP-10, an

IFN-induced protein, is released both basally and apically from HBE infected with all of these viruses. Clearly, some SAM-binding mutants induce different levels of mediators or release them in different directions.

78

Continued

Figure 2.7. Cytokine production in RSV mutant infected HBE cultures.

79

Figure 2.7 Continued

The apical and basolateral fluid from the infected HBE cultures were sampled as described in

Figure 2.5, at the time that each mutant-infected culture reached its maximum GFP expression

(indicated below each pair of bars). IL-1β (A), IL-6 (B), TNF-α (C), IL-12 (D), IFN-α2 (E), GM-

80

CSF (F), IFN-β (G), IL-10 (H), IFN-ɣ (I), IP-10 (J), IFN-λ1 (K), and IFN-λ2/3 (L) levels were assayed using the Legendplex Human Anti-Virus Response Panel (Biolegend) bead assay.

Basolateral values were multiplied by 5 to account for the dilution in 5 times more liquid. Data are the mean of three transwells ± standard deviation. *P<0.5, **P < 0.01, ***P < 0.001,

****P < 0.0001.

2.4.7 Recombinant rgRSVs with mutations in the SAM binding site are highly attenuated in cotton rats

We next determined whether rgRSVs with mutations in the SAM binding site were attenuated in cotton rats, the best small-animal model for RSV. As shown in Table 2.1, both rRSV and rgRSV replicated efficiently in the nasal turbinates and lungs of all five cotton rats. The wild type rRSV

3.97 4.30 had mean viral titers of 10 and 10 TCID50/g tissue in nasal turbinate and lung, respectively.

3.85 4.63 The rgRSV had mean viral titers of 10 and 10 TCID50/g tissue in nasal turbinate and lung, respectively. These results indicate that the insertion of GFP in rRSV did not significantly affect viral replication in cotton rats. In contrast, all the rgRSV mutants were significantly attenuated in replication in cotton rats. Infectious RSV was not detectable in the nasal turbinates of all cotton rats infected with rgRSV mutants whereas low levels of virus were detected in the lungs. 4 out 5 cotton rats infected with rgRSV-G1857A had detectable RSV with a mean titer of 102.32. Similarly,

4 out 5 cotton rats infected with rgRSV-G1853A had detectable RSV with a mean titer of 102.10.

The double mutant rgRSV-G1853A-G1857A was the most attenuated virus and only 2 out of 5 cotton rats had detectable virus with a mean titer of 102.30. These results demonstrate that rgRSVs

81

with mutations in the SAM binding site were highly attenuated in replication in both the upper and

lower respiratory tracts in cotton rats.

Titer in nasal turbinate B Titer in lung C Group A % of Mean titer % of Mean titer infected log infected log (TCID50/g) animals (TCID50/g) animals rRSV 5/5 3.97 ± 0.88a 5/5 4.30 ± 0.22a rgRSV 5/5 3.85 ± 0.77a 5/5 4.63 ± 0.36a rgRSV-G1857A 0/5 ND 4/5 2.32 ± 0.35b rgRSV-G1853A 0/5 ND 4/5 2.10 ± 0.42b rgRSV-G1853A- 0/5 ND 2/5 2.30 ± 0.32b G1857A DMEM 0 ND 0 ND

Table 2.1. Replication of rgRSV carrying mutations in the SAM binding site in cotton rats

5 A: Cotton rats were inoculated intranasally with DMEM, 2.0×10 TCID50 of wild-type rRSV,

rgRSV, or rgRSV mutants. At day 4 post immunization animals were euthanized. Nasal turbinate

and lung tissues were collected for virus titration by TCID50 assay.

B: “ND” indicates that infectious virus was not detectable.

C: Viral titers are the geometric mean titer (GMT) of animals with detectable RSV ± standard deviation. The detection limit is 2.0 log TCID50/g tissue. Value within a column followed by the

different lowercase letters (a or b) are significantly different (P<0.05).

82

2.4.8 Recombinant rgRSVs with mutations in the SAM binding site provide complete protection against rgRSV challenge in both short-term and long-term immunization study

Since all three rgRSV mutants were highly attenuated, we next determined whether they were immunogenic in cotton rats. As shown in Figure 2.8, all three rgRSV mutants triggered high levels of neutralizing antibodies that were comparable to those generated after rgRSV immunization at weeks 3 and 4 post-immunization (P > 0.05, t-test) although antibody titers of rgRSV mutants at weeks 1 and 2 were significantly lower than rgRSV (P<0.05, t-test). No RSV-specific antibody was detected in the unvaccinated control. At week 4 post-immunization, the immunized animals

5 were challenged with 2.0 × 10 TCID50 of rgRSV, and cotton rats were terminated at day 4 post- challenge (Table 2.2). DMEM-immunized but RSV challenged groups had a titer of 4.03 ± 0.20 and 5.50 ± 0.25 log10 TCID50 in nasal turbinate and lung, respectively. In contrast, no infectious

RSV was detected in nasal turbinate or lungs of cotton rats vaccinated with rgRSV, rgRSV-

G1853A, rgRSV-G1857A, or rgRSV-G1853A-G1857A after challenge. Thus, cotton rats were completely protected against RSV challenge in both the upper and lower respiratory tracts.

To further determine the protective efficacy of these rgRSV mutants, we performed a long-term

5 3 study in which cotton rats were immunized with lower doses (10 or 10 TCID50) of rgRSV,

5 rgRSV-G1857A, or rgRSV-G1853A, and were challenged with 2.0×10 TCID50 of rgRSV at day

112 after immunization. This allowed us to compare the dose effects and the dynamics of antibody

5 response over a longer time period (Figure 2.9). At a dose of 10 TCID50, there was no significant difference in antibody responses (P>0.05, ANOVA) between rgRSV and rgRSV-G1857A although rgRSV had slightly higher antibody responses than rgRSV-G1857A (Figure 2.9A).

Antibody titers of rgRSV-G1853A at weeks 4 and 8 were significantly lower than rgRSV (P<0.05,

83 t-test). However, there was no difference in the three other time points (weeks 2, 6, and 10, P>0.05, t-test). ANOVA test showed that there was no significance difference among these three groups

3 (P>0.05). At a dose of 10 TCID50, no significant difference was observed between rgRSV and rgRSV-G1857A from weeks 2-6 (Figure 2.9B) (P>0.05, t-test). However, rgRSV-G1857A immunized animals had significantly lower antibody titers than those immunized with rgRSV, at weeks 8 and 10 (P<0.05, t-test). In addition, rgRSV-G1853A had significantly lower antibody levels than rgRSV from weeks 4-10 (P<0.05). However, ANOVA test showed that there was no significance difference among these three groups (P>0.05). Taken together, these data demonstrate that (i) the ability of each rgRSV mutant to trigger and maintain neutralizing antibody is dose- dependent; and (ii) both rgRSV-G1857A and rgRSV-G1853A triggered similar levels of neutralizing antibody compared to rgRSV although antibody titers of these two mutants at some time points were relatively lower than those of rgRSV. Overall, antibody responses can be ranked rgRSV > rgRSV-G1857A > rgRSV-G1853A. At day 112 post-immunization, all cotton rats were challenged with rgRSV and terminated at day 4 post-challenge (Table 2.3). The unimmunized control group had a viral titer of 4.90 ± 0.40 and 5.20 ± 0.30 log10 TCID50 in their nasal turbinate and lungs, respectively. All vaccinated groups were completely protected from RSV challenge. No infectious RSV was detected in nasal turbinate or lungs. Thus, all immunized animals, even those immunized with 103 pfu of the mutant viruses, were completely protected from RSV challenge at

112 days post-immunization.

84

Figure 2.8. Neutralizing antibody responses of rgRSV mutants in cotton rats in a short-term

immunization experiment.

5 6-week-old SPF cotton rats were inoculated intranasally with 2.0×10 TCID50 of each rgRSV.

Blood samples were collected from each rat weekly by retro-orbital bleeding. The RSV-

neutralizing antibody titer was determined using a plaque reduction neutralization assay, as

described in Materials and Methods. Antibody titers are the geometric mean titer (GMT) of five cotton rats ± standard deviation. *P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

85

Titer in nasal turbinate B Titer in lung Group A Protection % of Mean titer % of Mean titer rate C infected log infecte log animals (TCID50/g) d (TCID50/g) animal s DMEM 5/5 4.03 ± 0.20 5/5 5.50 ± 0.25 0% rgRSV 0/5 ND 0/5 ND 100% rgRSV-G1857A 0/5 ND 0/5 ND 100% rgRSV-G1853A 0/5 ND 0/5 ND 100% rgRSV-G1853A- 0/5 ND 0/5 ND 100% G1857A

Table 2.2. Immunogenicity of rgRSV mutations in cotton rats in short term study

5 A: Animals were immunized intranasally with DMEM or 2.0×10 TCID50 of rgRSV or rgRSV

5 mutants. At day 28 post immunization, animals were challenged with 2.0×10 TCID50 of rgRSV.

Group DMEM indicates that cotton rats were inoculated with DMEM and challenged with rgRSV (unvaccinated and challenged control). Group rgRSV indicates that cotton rats were immunized with rgRSV and challenged with rgRSV.

B: “ND” indicates that infectious virus was not detectable. Viral titers are the geometric mean titer (GMT) of 5 animals ± standard deviation. The detection limit is 2.0 log TCID50/g tissue.

C: Protection indicates lack of viral replication in lung and nasal turbinate.

86

Figure 2.9. Neutralizing antibody responses of rgRSV mutants in cotton rats in a long-term immunization experiment.

5 3 6-week-old SPF cotton rats were inoculated intranasally with 10 or 10 TCID50 of each rgRSV.

Blood samples were collected from each rat biweekly by retro-orbital bleeding until week 10. The

RSV-neutralizing antibody titer was determined using a plaque reduction neutralization assay.

Antibody titers are the geometric mean titer (GMT) of four cotton rats ± standard deviation.

*P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

87

Titer in nasal turbinate Titer in lung A Group Dose B Protection rate C % of Mean titer % of Mean titer infected log infecte log animals (TCID50/g) d (TCID50/g) animal s DMEM 4/4 4.90 ± 0.40 4/4 5.20 ± 0.30 0% 5 rgRSV 10 TCID50 0/4 ND 0/4 ND 100% 3 rgRSV 10 TCID50 0/4 ND 0/4 ND 100% 5 rgRSV-G1857A 10 TCID50 0/4 ND 0/4 ND 100% 3 rgRSV-G1857A 10 TCID50 0/4 ND 0/4 ND 100% 5 rgRSV-G1853A 10 TCID50 0/4 ND 0/4 ND 100% 3 rgRSV-G1853A 10 TCID50 0/4 ND 0/4 ND 100% Table 2.3. Immunogenicity of rgRSV mutations in cotton rats in long term study

5 3 A: Animals were immunized intranasally with DMEM, 1.0×10 TCID50, or 1.0×10 TCID50 of rgRSV or rgRSV mutants. At day 112 post-immunization, animals were challenged with 2.0×105

TCID50 of rgRSV. Group DMEM indicates that cotton rats were inoculated with DMEM and challenged with rgRSV (unvaccinated and challenged control).

B: “ND” indicates that infectious virus was not detectable.

C: Protection indicates lack of viral replication in lung and nasal turbinate.

2.5 Discussion

In this study, we generated three rgRSVs carrying mutations in the SAM binding site in the L protein, and examined the effects of these mutations on viral replication, attenuation, and immunogenicity. We found that one of rgRSV mutants (rgRSV-G1857A) was moderately attenuated in HBE culture and cotton rats whereas rgRSV-G1853A and the combined double mutant (rgRSV-G1853A-G1857A) were highly attenuated. All three rgRSV mutants were

88 genetically stable in cell culture, grew to a high titer in cultured cells, triggered high levels of neutralizing antibody, and provided complete protection against RSV replication in both the upper

5 5 and lower respiratory tract of cotton rats. At high doses (2×10 and 10 TCID50), both rgRSV-

G1857A and rgRSV-G1853A induced similar levels of neutralizing antibody compared to rgRSV.

3 At a lower dose (10 TCID50) of immunization, rgRSV-G1857A and rgRSV-G1853A had relatively lower antibody responses at some time points. However, there was no significant difference among rgRSV-G1857A, rgRSV-G1853A, and rgRSV (P>0.05, ANOVA). These results suggest that rgRSVs with mutations in the SAM binding site had comparable or slightly less immunogenicity than the parental rgRSV. Interestingly, in a parallel experiment, one of rhMPVs carrying point mutations in the SAM binding site (rhMPV-G1700A) induced higher neutralizing antibodies compared to the parental rhMPV at a low immunization dose (103 PFU) although the difference was not significant (P>0.05, ANOVA). This suggests that rhMPV-G1700A is more immunogenic than rhMPV. Collectively, amino acid substitutions in the SAM binding site of the

L protein represent a promising approach to attenuate RSV and hMPV for the purpose of vaccine development.

A live attenuated vaccine is one of the most promising vaccine strategies for RSV. However, it has been a challenge to identify a genetically stable attenuated RSV strain that has an optimal balance between attenuation and immunogenicity [95, 165, 166]. Many RSV attenuated strains were either genetically unstable, insufficiently attenuated, overly attenuated, lacked sufficient immunogenicity, or have safety issues [95]. In addition, it has proven difficult to identify an RSV strain which has a greater immunogenicity than wild type RSV. One of the most promising candidates which has been tested in both nonhuman primates and human clinical trials is the M2-

89

2 protein-deleted RSV (RSV-ΔM2-2) [156, 167]. Interestingly, deletion of M2-2 from RSV

genome alters the balance of RNA synthesis such that genome replication is decreased and gene

transcription is increased, which results in increased viral protein expression [167]. Recombinant

vaccine candidate with ΔM2-2 (RSV MEDI ΔM2-2) had a greater neutralizing antibody response

than the parental RSV MEDI, the previous leading live attenuated RSV candidate vaccine in

human trials [157]. However, the antibody responses to RSV-ΔM2-2 was lower than those of wild-

type RSV in nonhuman primates [167]. This suggests that RSV-ΔM2-2 still had a lower

immunogenicity compared to wild-type RSV.

MTase is an excellent target for attenuating viruses for vaccine purpose [62, 65]. The SAM-

dependent MTase utilizes methyl donor SAM to transfer a methyl group to an RNA acceptor. A

unique feature of mRNA cap methylation in NNS RNA viruses is that the CR-VI domain of the L

protein possesses both G-N-7 and 2’-O MTases and these two MTases share a single SAM binding

site [159-161]. Thus, mutations in the SAM binding site of L protein will affect either G-N-7

and/or 2’-O methylation. Recombinant viruses carrying these mutations exhibited different

degrees of attenuation depending on the specific methyl group involved and the degree of the defect in mRNA cap methylation [62]. This approach likely enhances the genetic stability of the attenuated strain, as multiple amino acid substitutions in MTase region of L protein can be

combined, reducing the possibility of reversion. Another advantage is that viral mRNA deficient

in 2’-O methylation is recognized as non-self RNA and induces a higher type I interferon response

[60, 61], which may promote a better adaptive immune response.

In this study, we compared the immunogenicity of rgRSVs carrying mutations in the SAM binding

5 site with the parental RSV. At a dose of 2×10 TCID50, there was no significant difference in

90 antibody titers at weeks 3 and 4 post-immunization between the live attenuated vaccine candidates and the parental rgRSV (P>0.05) although the antibody titers of all three vaccine candidates at weeks 1 and 2 post-immunization were lower than the parental rgRSV (P<0.05). After challenge with rgRSV, all vaccinated groups were completely protected from RSV replication in nose and lungs. This led to a long-term vaccination study with a goal of comparing the antibody response

5 3 and protection at the lower vaccination doses (10 and 10 TCID50). Our results at the lowest

3 immunization dose (10 TCID50) showed that rgRSV-G1857A had higher antibody responses than rgRSV-G1853A but both mutants induced a somewhat lower antibody response than the parental rgRSV at some time points, perhaps reflecting the degree of virus attenuation. For example, rgRSV-G1853A is more attenuated in cell culture, HBE culture, and cotton rats than rgRSV-

G1857A. However, there was no significant difference among three groups (P>0.05). Also, cotton rats immunized with both doses of both attenuated viruses were completely protected from RSV challenge at 112 days post-immunization. This suggests that, despite the high attenuation, the immune responses triggered by both rgRSV-G1857A and rgRSV-G1853A were sufficient to protect cotton rats from virus replication in both upper and lower respiratory tract.

All of the SAM binding mutations in rgRSV, inoculated at an MOI of 1.0 grew slower than rgRSV in Vero cells, but reached similar peak titers by 2.5 dpi, and therefore should be economical to produce as vaccine candidates. These mutants spread more slowly in HBE cultures, particularly those containing the G1853A mutation, indicating that this mutation is more attenuating than the

G1857A mutation. Nevertheless, these mutants were protective, even at the low virus dose of 103

TCID50 in cotton rats. The double mutant might be a preferable vaccine candidate since both mutations are attenuating and reversion at one of these positions would not result in pathogenic

91

virus. In addition, because G1857A is a partially attenuating mutation that does not reduce virus

yield in Vero cells, it could be combined with other attenuating mutations to reduce the likelihood

of reversion.

The mediators induced in HBE cultures by these weakened viruses may suggest how they stimulate

the immune system so robustly without high level virus replication. In fact, a number of cytokines including IFN-ɣ, IL-10, IFN-α2, IL-1β, TNF-α, and IL-12 were significantly higher in all three

rgRSV mutants than rgRSV in HBE culture (P<0.05). Both viruses with the G1853A mutation

induced more GM-CSF than rgRSV or rgRSV-G1857A. GM-CSF recruits and activates antigen

presenting cells, promoting antigen capture and T cell activation [168-171]. These mutant viruses

also induced more IL-6, an inflammatory mediator. Although too much inflammation is likely part

of the problem in infants with severe RSV disease, contributing to the narrowing of the airways,

some inflammation is required to deliver immune cells to the site of infection to initiate the immune

response. Compared to rgRSV, all three mutants induced much less IFN-λ 1 and 2/3. The reduced

IFN might be useful for a vaccine candidate if IFN prevents B cell memory, as recent reports with

chronic lymphocytic choriomeningitis virus (LCMV) infection of mice have found [172, 173].

The dramatic attenuation of G1853A carrying virus infections in HBE cultures is likely related to

their reduced mRNA cap methylation, both the reduced ability of its mRNAs to translate viral

proteins and the induction of IFN by these mRNAs. Because these infections were initiated with a

low MOI, the mutants replicated and spread slowly and the level of IFN detected in the apical and

basal fluids of the HBE cultures remained low. In A549 cells inoculated at a higher MOI, G1853A

containing viruses did induce more IFN-β than WT or the other two mutants. The double mutant

92 also produced much less viral protein under these conditions, probably because of its more severe defect in cap methylation. Less viral replication and mRNA production would induce less IFN.

There are a few limitations in our study. In the long-term immunization study, we did not measure the neutralizing antibody responses after 10 weeks due to the unexpected injury of cotton rats caused by frequent bleeding. It will be interesting to investigate the dynamics of antibody responses using a booster vaccination strategy in future studies. In addition, we only compared the

5 3 antibody responses of two virus doses (10 and 10 TCID50). We found complete protection against

3 RSV infection 112 days post-immunization even with a low immunizing dose (10 TCID50) even though less neutralizing antibody had been induced. It will be of interest to determine the minimal vaccination dose and the minimal neutralizing antibody required for complete protection against

RSV in cotton rats. Finally, it is difficult to compare the levels of mediators released by HBE cultures infected with viruses that grow and spread at such different rates. We chose to examine their levels at the times when each virus reached its maximum infection levels, determined by viral

GFP expression. However, the differences in mediator expression in HBE cultures suggest possible reasons for the strong antiviral response in the cotton rats despite poor replication by the mutant viruses.

In summary, we showed that alteration of amino acid residues in the SAM binding site represents a promising approach to attenuate RSV and hMPV for the purpose of vaccine development.

Despite a high attenuation phenotype, rRSV mutants in SAM binding site retained high immunogenicity. Future studies will focus on the strategies which can enhance the immunogenicity of rgRSVs with mutations in the SAM binding site.

93

Chapter 3: Stable attenuation of human respiratory syncytial virus for live vaccines by

deletion and insertion of amino acids in the hinge region between the mRNA capping and

methyltransferase domains of the large polymerase protein

3.1 Abstract

The large (L) polymerase protein of RSV organizes as a core ring-like domain containing the RNA dependent RNA polymerase and an appendage of globular domains containing an mRNA capping region and a cap methyltransferase region which are linked by a flexible hinge region. Here, we found that the flexible hinge region of RSV L protein is tolerant to amino acid deletion or insertion.

Recombinant RSVs carrying a single and double deletion or a single alanine insertion were

genetically stable, highly attenuated in immortalized cells, and defective in replication and spread

in primary, well differentiated, human bronchial epithelial (HBE) cultures. The replication of these

recombinant viruses was highly attenuated in the upper and lower respiratory tracts of cotton rats.

Importantly, these recombinant viruses elicited high levels of neutralizing antibody and provided

complete protection against RSV replication. Taken together, amino acid deletions or insertions in

the hinge region of the L protein can serve as a novel approach to rationally design genetically

stable, highly attenuated and immunogenic live vaccine candidates for RSV.

3.2 Introduction

RSV belongs the family Pneumoviridae in the order of Mononegavirales. Similar to all other non- segmented negative-sense (NNS) RNA viruses, RSV encodes a large (L) polymerase protein with 94 molecular weight of 220 kDa which possesses all the enzymatic activities for genome replication, mRNA transcription, and mRNA modifications [174, 175]. A recent structural study of the L protein of vesicular stomatitis virus (VSV) within the related revealed that the L protein organizes as a core ring-like domain containing the RNA dependent RNA polymerase and an appendage of globular domains containing a capping region (CR V) and a cap methyltransferase region (CR VI) which are linked by a flexible hinge region [31]. Interestingly,

(measles virus, MeV [176]; rinderpest virus, RPV [177]; and canine distemper virus, CDV [178]) within the family , as well as VSV within the Rhabdoviridae [179], were shown to tolerate in-frame insertion of the entire enhanced GFP (eGFP) at the region between CR V and

CR VI in the L protein.

In this study, we found that the flexible hinge region between CR V and CR VI of RSV L protein was not only tolerant to amino acid insertion but also tolerates amino acid deletion. Recombinant

RSVs (rRSVs) carrying a single or double deletion, or an alanine insertion in this flexible region, grew to a high titer, were genetically stable, and sufficiently attenuated yet retained high immunogenicity in cotton rats. Therefore, these rRSVs deletion and insertion mutants are highly promising vaccine candidates for RSV.

3.3 Materials and Methods

3.3.1 Cell lines

HeLa (ATCC CCL-2), A549 (ATCC CCL-185), Vero (ATCC CRL-CCL81), and HEp-2 (ATCC

CCL-23) cell lines were purchased from the American Type Culture Collection (Manassas, VA) and were grown in Dulbecco's modified Eagle's medium (DMEM; Life Technologies) supplemented with 10% FBS. Primary, well-differentiated human bronchial epithelial (HBE)

95 cultures were grown on collagen coated Transwell inserts (Corning Incorporated, Corning, NY) at an air-liquid interface, as previously described [164]. Upon reaching confluency and forming tight junctions, the apical medium was removed and cultures were maintained at the air-liquid interface for 4 to 6 weeks to generate well-differentiated, polarized cultures. All cell lines used in this study were free of mycoplasma, as confirmed by the LookOut Mycoplasma PCR Detection Kit (Sigma).

2.3.2 Preparation and characterization of recombinant RSV

Full-length infectious cDNA clones of RSV (R2 strain) were used to generate mutant and wild type viruses. The L deletions (ΔD1557, ΔM1558, and ΔD1557-ΔM1558) and insertion (1557-A-

1558) were introduced into the pTM1-L and RW30 plasmids using QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). A standard recovery protocol was used to rescue the recombinant rgRSVs as described previously. The protocols for virus purification, minigenome assay, TCID50 assay, viral protein expression analysis, growth kinetics and genetic stability were reported previously.

3.3.3 RT-PCR and sequencing

All plasmids, viral mutants and stocks, and virus isolates from the nasal turbinates and lungs of cotton rats were sequenced to confirm virus identity. Viral RNA was extracted from 100 µl of each recombinant virus using an RNeasy minikit (Qiagen, Valencia, CA). A 1.0-kb DNA fragment spanning the flexible region between CR V and CR VI of the L gene was amplified by RT-PCR.

Also, the entire RSV genome was amplified using six overlapping fragments by RT-PCR. The

PCR products were purified and sequenced using a sequencing primer at The Ohio State University

Plant Microbe Genetics Facility to confirm the presence of the designed mutations.

96

3.3.4 Replication, spreading, and cytokine production in HBE culture

Purified virions were titrated on HEp-2 cells and were diluted in HBE cell medium. The apical surface of well-differentiated HBE cells in Transwells was washed with PBS for 2 h and the basal medium was changed before adding the virus (400 or 2,000 TCID50) was added to the apical chamber of the Transwell. Fluorescent cells were visualized with an EVOS2 fl inverted fluorescence microscope (Life Technologies). At indicated time, 100 µl of PBS was added to the apical surface of HBE culture, gently rocked for 30 min, and collected for virus titration. The amount of the HBE culture expressing GFP-positive cells was determined from a digital image using Image J Software. In addition, the medium at the apical wash and the basolateral medium from HBE cultures was collected at days 2 and 5 post-inoculation for detection of cytokines (IFN-

β, IFN-λ1, IFN-λ2/3, IP-10, and IL-6) by a flow cytometer bead assay (LEGENDplexTM,

Biolegend, San Diego, CA).

3.3.5 Replication, pathogenesis and immunogenicity of rgRSV mutants in cotton rats

Thirty 4-week-old specific-pathogen-free (SPF) male cotton rats (Envigo, Indianapolis, IN) were randomly divided into 5 groups (5 cotton rats per group). Prior to virus inoculation, the cotton rats

5 were anesthetized with isoflurane. The cotton rats in group 1 were inoculated with 2.0×10 TCID50 of parental rgRSV and served as positive controls. The cotton rats in groups 2 to 5 were inoculated

5 with 2.0×10 TCID50 of four rgRSV mutants, rgRSV-ΔD1557, ΔM1558, ΔD1557-ΔM1558, and

1557-A-1558. Each cotton rat was inoculated intranasally with a volume of 100 μl. At day 4 post- infection, the cotton rats were sacrificed via carbon dioxide inhalation. The left lung and nasal turbinates were collected for virus titration and the right lung was collected for histological analysis.

For the immunogenicity study, twenty five 4-week-old female cotton rats (Envigo) were randomly

97

divided into five groups (5 cotton rats per group). Cotton rats in groups 1-4 were intranasally

5 inoculated with 2.0×10 TCID50 of three rgRSV mutants (ΔM1558, ΔD1557-ΔM1558, and 1557-

A-1558) and rgRSV, respectively. Cotton rats in groups 5 were mock-infected with DMEM and

served as unvaccinated challenged control. After immunization, the cotton rats were evaluated

daily for any possible abnormal reaction and blood samples were collected from each cotton rat

weekly by orbital sinus blood sampling, and serum was used for detection of neutralizing

antibodies. At 4 weeks post-immunization, the cotton rats in all groups were challenged with

5 2.0×10 TCID50 of parental rgRSV via intranasal route, and evaluated twice daily for the presence

of any clinical symptoms. At 4 days post-challenge, all cotton rats were euthanized by CO2

asphyxiation, and their lungs and nasal turbinates were collected for virus titration. The

immunogenicity of rgRSV mutants was assessed based on their ability to trigger neutralizing

antibody, the ability to prevent rgRSV replication in lungs and nose, and the ability to protect lung

from pathological changes.

3.3.6 Pulmonary histology

After sacrifice, the right lung of each animal was removed, inflated, and fixed with 4%

paraformaldehyde. Fixed tissues were routinely processed, embedded in paraffin and sectioned at

4 μm. Slides were then stained with hematoxylin and eosin (HE) and evaluated light

microscopically by a veterinary anatomic pathologist board certified by the American College of

Veterinary Pathologists. Sections were evaluated for peribronchiolitis and perivasculitis (0: none;

1: > 60% of bronchioles/vessels characterized by no or few infiltrating leukocytes [granulocytes,

lymphocytes and/or macrophages]; 2: > 40% of bronchioles/vessels characterized by focal

aggregate or cuff of 1 discrete layer of leukocytes and < 30% characterized by aggregates or cuff

98

of 2 or more discrete layers of leukocytes; 3: > 30% of bronchioles/vessels characterized by

aggregates or cuff of 2 or more discrete layers of leukocytes); bronchiolitis (absence [0] or

presence [1] of bronchiolar goblet cell hyperplasia and granulocytic exocytosis); alveolitis (0: none

to scattered individual macrophages and/or granulocytes; 2: < 5 nodular aggregates of macrophages and/or granulocytes; 2: 5-20 nodular aggregates of macrophages and/or granulocytes; 3: extensive to confluent nodular aggregates of macrophages and/or granulocytes in

< 60% of the tissue; 4: extensive to confluent nodular aggregates of macrophages or granulocytes in ≥ 60% of the tissue); and, interstitial pneumonia (0: no thickening of alveolar septae; 1: thickening of alveolar septae) using a modification of adapted criteria. [180, 181].

3.3.7 Sequence alignment and analysis

L protein sequences were aligned by ClustalW2. The RSV-L(YP009518860.1), VSV-L

(Q98776.1), RPV-L(NC006296.2), CDV-L(NP047207.1), and MV-L(Z66517.1) were included.

Structure modeling of RSV L protein. The Cryo-EM Structure of the L protein of vesicular

stomatitis virus (PDB: 5A22) was chosen as the template for RSV L protein structure prediction

using MODELLER program (Ver 9.20).

3.3.8 Statistical analysis

Quantitative analysis was performed by either densitometric scanning of autoradiographs or by

using a phosphorimager (Typhoon; GE Healthcare, Piscataway, NJ) and ImageQuant TL software

(GE Healthcare, Piscataway, NJ). Statistical analysis was performed by one-way multiple

comparisons using SPSS (version 8.0) statistical analysis software (SPSS Inc., Chicago, IL). A P

value of <0.05 was considered statistically significant.

99

3.4 Results

3.4.1 Deletion or insertion in the L protein inhibit its activity

Recent structural studies of VSV L protein showed that CR V and CR VI are connected by two

linkers (linker 1 and 2) separated by a connector domain (CD), suggesting that the region between

CR V and CR VI is flexible [31]. Amino acid sequence alignment and homology analysis of the

VSV and RSV L proteins showed that CRI-VI, linkers 1 and 2, and CD are conserved between

RSV and VSV L proteins (Figure 3.1A and B). Previously, in-frame insertion of EGFP into linker

2 of VSV L resulted in a functional L protein and a viable recombinant VSV [179]. In addition,

mutagenesis analysis in a region between amino acids 1450-1481 within the CD of VSV L affected

mRNA cap methylation [182], although mRNA cap methylase catalytic sites are located in CR VI

of the L protein [160, 183, 184]. We hypothesized that the CD of RSV L will tolerate amino acid

insertions and deletions. Based on the amino acid sequence alignment and homology analysis of

the VSV and RSV L proteins, we predicted that amino acid residues at positions 1557 and 1558

of RSV L may be most tolerant to deletion and insertion (Figure 3.1C). To generate mutants, we

first deleted single amino acids D1557 or M1558 from the RSV L gene resulting in L-ΔD1557 and

L-ΔM1558, respectively. Both D1557 and M1558 were also deleted in one virus to generate the double deletion mutant, L-ΔD1557-ΔM1558. To generate an insertion mutant, a single alanine residue was inserted between D1557 and M1558 resulting in L-1557-A-1558.

These RSV L mutants were tested in an RSV minigenome that expresses GFP and found to be

functional (Figure 3.2A) although these L mutants expressed less GFP relative to the wt L,

suggesting that the L deletion and L insertion mutants are capable of replicating genome and

transcribing mRNA, and thus may not be lethal to the virus. Quantitative analysis showed that L-

100

ΔD1557 expressed 90% as many GFP-positive cells compared to wt L (P <0.01) but had no significant difference in GFP intensity with wt L (P>0.05) (Figure 3.2B and C). L-ΔM1558, L-

ΔD1557-ΔM1558, and L-1557-A-1558 produced approximately 80%, 70%, and 50% as much

GFP as the wt minigenome expression, respectively (Figure 3.2B and C).

To further test the abilities of those L mutants in replication and transcription, we quantified the synthesized genome and mRNA of minigenome. It was found that L mutants generated significantly less genomic RNA and mRNA (Figure 3.2D and E) which suggested that the reduction on GFP expression of mutants carrying deletion or insertion is due to the defective of L protein in replicating genome and transcribing mRNA.

101

Figure 3.1. Design of RSV L deletion and insertion mutants.

(A) Conserved regions (CRs) in the VSV L protein. The nucleotide polymerization motif (GDN) in CR III, mRNA capping motif (GxxT[n]HR) in CR V, and mRNA cap methyltrasferase motif

(SAM binding GxGxG…D) in CR VI is indicated. (B) CRs in the RSV L protein. Based on the sequence alignment of RSV L (YP 009518860.1) and VSV L (Q98776.1), the CRs and their predicted amino acid positions are assigned. (C)(D)(E) Structure modeling of RSV L protein. The

Cryo-EM structure of the VSV L protein (PDB: 5A22) was used as the template for RSV L protein structure prediction using MODELLER program (Ver 9.20). Amino acids D1557 and M1558 are indicated.

102

Figure 3.2. Examination of the function of L deletion and insertion mutants using a minigenome assay. 103

(A) L deletion and insertion mutants diminished GFP expression. Confluent HEp-2 cells were

transfected with the minigenome plasmid together with pTM1-N, pTM1-P, pTM1-M2.1, pTM1-

L, or an pTM1-L mutant, using Lipofectamine 2000. GFP expression was visualized by a

fluorescence microscope at 48 h posttransfection. (B) Quantification of GFP-positive cells by flow

cytometry. HEp-2 cells were transfected with the minigenome and support plasmids. At 48h

posttransfection, cells were trypsinized and GFP-positive cells were counted by flow cytometry.

(C) Quantification of GFP density by flow cytometry. Relative level of genome of minigenome

(D) and mRNA (E). Data are the average of three independent experiments ± standard deviation.

*P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

3.4.2 rgRSV deletion and insertion mutants are attenuated in HEp-2 cells

We built these deletion and insertion mutations into an infectious cDNA clone of RSV A2 strain in which the GFP gene had been inserted between the leader and the NS1 gene. GFP expression allowed us to monitor replication and gene expression of the recombinant virus. All of the mutant viruses were successfully recovered and were named rgRSV-ΔD1557, ΔM1558, ΔD1557-

ΔM1558, and 1557-A-1558. All recombinant viruses were plaque purified and further passed 4-5

times in Vero cells. The entire genome was amplified by RT-PCR and sequenced. All retained the desired deletion or insertion in the L gene. No additional mutations were found elsewhere in their

genomes.

We next determined whether these rgRSV mutants were attenuated in HEp-2 cells. Briefly,

confluent HEp-2 cells were infected with each recombinant virus at an MOI of 1.0. The expression

of GFP and virus production kinetics were monitored. The parental rgRSV had a maximal level of

104

GFP expression at day 3 post-infection, developing extensive CPE by day 4. However, all rgRSV

mutants had delayed GFP expression and CPE in Vero cells, reaching maximal GFP expression at

day 4-5 post-inoculation. Extensive CPE was not observed for these rgRSV mutants until days 5-

6 post-inoculation.

Quantification by flow cytometry showed that significantly fewer GFP-positive cells were detected

in HEp-2 cells that were infected by each of the mutants at 24 and 48 h post-infection, compared

to rgRSV (P<0.05) (Figure 3.3A and B). We also compared the growth kinetics of these rgRSV

mutants with the parental rgRSV (Figure 3.3C). The parental rgRSV reached a peak titer of 107.02

TCID50 at 48 h post-inoculation. The rgRSV-ΔD1557 grew to a higher titer compared to rgRSV,

7.13 reaching a peak titer of 10 TCID50 at day 3 post-inoculation. The growth kinetics of rgRSV-

ΔM1558, ΔD1557-ΔM1558, and 1557-A-1558 had a significant delay. Those rgRSV mutants reached a peak titer at days 4-5 post-inoculation and had a 0.4-0.6 log reduced peak titer compared to rgRSV. These results demonstrated that all these rgRSV mutants had a significant delay in viral replication in HEp-2 cells.

105

Figure 3.3. Characterization of rgRSVs carrying deletion and insertion in the flexible hinge region of the L protein.

106

(A) Delayed GFP expression by rgRSV mutants in Vero cells. Confluent Vero cells were infected

by each rgRSV at an MOI of 0.1, and GFP expression was monitored at the indicated time by

fluorescence microscope. (B) Quantification of GFP positive cells by flow cytometry. Confluent

HEp-2 cells were infected by each rgRSV (MOI of 0.1), at indicated time point, cells were

trypsinized and GFP positive cells were quantified by flow cytometry. Data are the average of

three independent experiments ± standard deviation. (C) Single-step growth curve of rgRSV

mutants. HEp-2 cells in 12-well-plates were infected with each recombinant rgRSV at an MOI of

1.0. After adsorption for 1 h, the inocula were removed and the infected cells were washed 3 times

with Opti-MEM medium. Fresh DMEM medium containing 2% FBS was added and the cells were

incubated at 37°C for various times. The supernatant and cells were harvested by three freeze-thaw

cycles, followed by centrifugation at 1,500 × g at 4oC for 15 min at the indicated intervals. The

viral titer was determined by TCID50 assay in HEp-2 cells. Viral titers are the geometric mean

titer (GMT) of three independent experiments ± standard deviation. *P<0.5, **P < 0.01,

***P < 0.001, ****P < 0.0001.

3.4.3 Antigen expression of rgRSV deletion and insertion mutants in A549 cells

An ideal attenuated strain for vaccine purposes should generate sufficient amounts of viral antigens in order to trigger a strong immune response. Since F and G proteins are the two major surface glycoproteins responsible for inducing neutralizing antibody, we monitored the dynamics of F and

G protein expression in virus-infected cells (Figure 3.4). Briefly, confluent A549 cells were infected by each recombinant virus at an MOI of 0.1, cell lysates were collected at 18, 24, and 48 h post-inoculation, and were subjected to Western blot using monoclonal antibodies against F or

107

G. At 18 h post-infection, all rgRSV mutants had significant defects in both F and G protein

expression compared to rgRSV. However, rgRSV-ΔD1557 and rgRSV-ΔD1557-ΔM1558 reached a similar level of F and G expression with rgRSV at 24 and 48 h post-inoculation. Recombinant rgRSV-1557-A-1558 and rgRSV-ΔM1558 had less F and G expression at all three time points

compared to rgRSV. Consistent with the attenuation phenotype in cell culture, all rgRSV mutants had a delay in antigen expression but two mutants eventually reached similar levels of antigen expression, relative to rgRSV.

Figure 3.4. F and G protein expression.

A549 cells were infected with the parental rgRSV or rgRSV mutants at an MOI of 0.1. At 18, 24,

and 48 h post-inoculation, total cell lysates were harvested and subjected to Western blotting using

a monoclonal antibody against RSV F or G protein.

108

3.4.4 Genetic stability of rgRSV deletion and insertion mutants in Vero cells

We hypothesize that amino acid deletion and insertion in the L gene will enhance genetic stability

of rgRSV mutants. To test this hypothesis, we successively passed these recombinant viruses in

Vero cells 15 times. Briefly, confluent Vero cells were infected by each recombinant virus at an

MOI of 0.1, and cell culture supernatants were harvested when extensive CPE was observed. Cell

culture supernatants were used as inoculum for the next passage. At each passage, a 1.0 kb

fragment spanning the deletion or insertion mutations was amplified by RT-PCR and sequenced.

The result showed that all recombinant viruses retained the desired deletions or insertion at each passage. At passage 15, the entire genome was amplified by RT-PCR and sequenced. No additional mutations were found in the genome. This result suggests that rgRSV deletion and insertion mutants were genetically stable in cell culture.

3.4.5 Infection of rgRSVs and cytokine production in HBE culture

RSV infection has been studied mainly in immortalized cell lines. However, virus replication in vitro in immortalized cell lines may differ from replication in vivo in the human airway epithelium in many aspects such as receptors, entry and spread mechanisms. Primary, well differentiated HBE cultures have been shown to accurately represent the human airway epithelium, both in appearance and function. Similar to RSV infection in the airways of human lungs, RSV infects HBE cultures via the apical surface of the ciliated cells. Therefore, we tested replication and spread of rgRSV deletion and insertion mutants in HBE cultures. Briefly, HBE cultures were inoculated with 400

TCID50 of each recombinant virus and viral spread was detected by monitoring GFP expression

(Figure 3.5A). The parental rgRSV produced visible GFP positive cells at day 1, rapidly spreading

at days 2-4, and reaching a peak that involved most of the culture by day 6. All rgRSV mutants

109

had delays in spreading in HBE cultures. At the indicated times, mucus was washed off,

supernatants were collected, and the dynamics of virus release from HBE cultures was determined

7.86 by TCID50 (Figure 3.5B). The parental rgRSV reached a peak titer of 10 TCID50/ml at day 6

post-inoculation and decreased by day 8. All rgRSV mutants had a significant delay in virus released from HBE cultures. Recombinant rgRSV-ΔD1557 and rgRSV-ΔD1557-ΔM1558 reached

peak titers of 106.97 and 106.58, respectively, at day 6 whereas rgRSV-ΔM1558 and rgRSV-1557-

A-1558 peaked at day 8 with titers of 105.40/ml and 106.36/ml, respectively, before receding.

We also tested the replication of rgRSV mutants in HBE culture derived from a different lung

donor at a higher inoculation dose (2,000 TCID50). In this experiment, we did not disturb and wash

the mucus of HBE culture until 65 h post-inoculation. Significantly less GFP signal was observed

in these cultures for the rgRSV mutants (Figure 3.6A). Quantification of GFP by image J software shows significant delays in viral spreading for all rgRSV mutants, with some more than others

(Figure 3.6B). Based on the GFP and virus titer, the robustness of spread and virus replication in

HBE culture can be ranked as following: rgRSV >rgRSV-ΔD1557 > rgRSV-ΔD1557-ΔM1557 > rgRSV-1557-A-1558 > rgRSV-ΔM1558.

110

Figure 3.5. rgRSV production in HBE cultures.

111

(A) rgRSVs in HBE culture. HBE cultures were inoculated with 400 TCID50 of each rgRSV. At the indicated times, virus spread was monitored by fluorescence microscopy. Representative images at each time point were shown. (B) Virus release from rgRSV-infected HBE culture. Apical washes were collected every 2 days until day 14 post-inoculation. Infectious virus in washes was quantified by the TCID50 assay. Viral titers are the geometric mean titer (GMT) of three independent experiments ± standard deviation. *P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

112

Figure 3.6. rgRSV mutants spread more slowly in HBE culture.

113

(A) Spreading of rgRSVs in HBE culture. HBE cultures were inoculated with 2,000 TCID50 of

each rgRSV. At the indicated times, the entire Transwell was imaged by fluorescence microscopy.

At 65h post-inoculation, 100 µl of PBS was added to the apical surface of the HBE cultures.

Representative images of three Transwells at each time point for each rgRSV are shown. (B) GFP

expression in infected HBE cultures. The GFP signal was quantified from a digital image by Image

J software for each rgRSV-infected culture on the day they reached maximum infection. Fold

differences in GFP intensity of each rgRSV mutant-infected HBE culture, relative to mock-

infected HBE were calculated, and data were expressed as the mean of three Transwells ± standard deviation. *P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

We next determined cytokine production in HBE culture in response to rgRSV infection. We are

particularly interested in those cytokines involved in innate immunity which include IFN-β (type

I IFN response), IFN-λ 2, and 3 (type III IFN response), IP-10 (interferon-gamma inducible protein

10 kDa, also known as CXCL10 chemokine), and IL-6 (a signature cytokine for inflammatory

response). Briefly, HBE cultures were inoculated with 2,000 TCID50 of rgRSV or each mutant and

apical and basolateral fluids were sampled at days 2 and 5 post-inoculation for cytokines. 100 µl

of PBS was added to apical surface of HBE culture, and the liquid on the apical and baso-lateral

surface was collected. The amount of each cytokine was normalized by the degree of viral

spreading (GFP intensity). As shown in Figure 3.7A, there was no significant difference in IL-6

production at day 2 for medium harvested from the apical or basolateral surface. However, rgRSV-

ΔD1557-ΔM1558 and rgRSV-ΔM1558 had a significantly higher IL-6 compared to rgRSV

(P<0.05). All rgRSV mutants had less IP-10 at day 2 compared to rgRSV (Figure 3.7B). However,

114 rgRSV-ΔD1557 and rgRSV-1557A1558 had a higher IP-10 in basolateral samples (P<0.05), and rgRSV-ΔD1557-ΔM1558 and rgRSV-ΔM1558 had a similar level of IP-10 compared to rgRSV at day 5 (P>0.05). Recombinant rgRSV-ΔD1557-ΔM1558 had a similar level of IFN λ2 and 3

(P>0.05) whereas other mutants had less IFN λ2 and 3 at day 5 (Figure 3.7C) (P<0.05). All rgRSV mutant had less IFN-β in basolateral samples at day 2 relative to rgRSV (Figure 3.7D).

Interestingly, rgRSV-ΔD1557-ΔM1558 and rgRSV-ΔM1558 had a significantly higher IFN-β on apical surface compared to rgRSV (Figure 3.7D) (P<0.05). In addition, rgRSV-ΔD1557-ΔM1558 had similar level of IFN-β whereas all other mutants had less IFN-β at day 5 (P<0.05). These results demonstrated that rgRSV mutants had a delay in cytokine production at early time point

(day 2) but were capable of producing a high level of the cytokines involved in innate immunity.

115

Continued Figure 3.7. Interferons, IP-10 and IL-6 production in RSV mutant-infected HBE cultures.

116

Figure 3.7 Continued

117

The apical and basolateral fluid from the infected HBE cultures described in Figure 3.6 were sampled as described in Figure 3.5, at the time that each mutant-infected culture reached its maximum GFP expression (indicated below each pair of bars). IL-6 (A), IP-10 (B), IFN-λ2,3 (C),

IFN-β (D) levels were assayed using the Legendplex Human Anti-Virus Response Panel

(Biolegend) bead assay. Basolateral values were multiplied by 5 to account for the dilution in 5 times more liquid. Data were expressed as the mean of three Transwells ± standard deviation.

3.4.6 rgRSVs are attenuated in replication and pathogenicity in cotton rats

We next determined whether rgRSV deletion and insertion mutants are attenuated in vivo. The parental rgRSV replicated efficiently in the nasal turbinates and lungs of all five cotton rats.

Average viral titers of 4.77 ± 0.31 log10 TCID50/g and 4.82 ± 0.11 log10 TCID50/g were found in the nasal turbinate and lung, respectively (Figure 3.8A and B). The rgRSV deletion and insertion mutants had significantly reduced viral replication in the nasal turbinates and lungs. For the rgRSV

ΔD1557, average viral titers of 3.35 ± 0.43 log10 TCID50/g and 4.06 ± 0.25 log10 TCID50/g were found in the nasal turbinate and lung, respectively. For rgRSV-ΔD1557-ΔM1558, an average titer of 2.44 ± 0.16 log10 PFU/g was detected in the nasal turbinate, and 4 out of 5 rats had detectable virus in lung tissue, with a titer of 2.39 ± 0.24 log10 PFU/g. For rgRSV-1557-A-1558, 4 out of 5 cotton rats had detectable virus in the nasal turbinate, with an average titer of log 2.40 ± 0.32 log10

PFU/g, and 4 out of 5 rats had detectable virus in lung tissue, with a titer of log 2.45 ± 0.26 log10

PFU/g. Recombinant rgRSV-ΔM1558 was completely attenuated in cotton rats. There was no detectable infectious virus in either nasal turbinate or lung tissue. In order to confirm the replication of rgRSV-ΔM1558 in cotton rats, we have measured the genome RNA of each virus. Briefly, 150

118 ul of homogenized tissue was used for extraction of RNA followed by real-time RT-PCR quantification. Compared with wild type rgRSV, rgRSV-ΔM1558 and rgRSV-ΔD1557-ΔM1558 produced significantly less genome RNA in the lungs, while in nasal turbinates, all mutant virus showed significantly lower level of genome RNA. These results suggested that all mutant RSV including rgRSV-ΔM1558 were capable in replicating in cotton rats both in upper and lower respiratory tract. Histopathologic changes were comparably mild in lungs from all RSV-infected groups, and were generally characterized by few leukocytes surrounding bronchioles and blood vessels, evidence of occasional bronchiolar goblet cell hyperplasia with granulocyte exocytosis, and small numbers of discrete nodules of alveolar macrophages. This result demonstrated that the rgRSV-ΔM1558, ΔD1557-ΔM1558, and 1557-A-1558 mutants were highly attenuated in viral replication in both the upper and lower respiratory tracts in cotton rats, whereas rgRSV-ΔM1558 was moderately defective in cotton rats.

119

Figure 3.8. Replication of RSV mutant in cotton rats.

120

Viral load in lungs (A), nasal turbinates (B) and genome RNA level in lungs (C), nasal turbinates

(D) of cotton rats in each group 4 days after inoculation. Four-week-old SPF cotton rats were

5 inoculated intranasally with 2.0×10 TCID50 of each rgRSV. At day 4 post-infection, the cotton rats were sacrificed, and lungs and nasal turbinates were collected for virus titration by TCID50

assay and genome RNA detection by RT-realtime PCR using primer annealing to the leader and

GFP sequence of RSV. Viral titers and genome RNA copies are the geometric mean titer (GMT)

of 5 animals ± standard deviation. The detection limit is 2.0 log TCID50/g tissue.

3.4.7 Deletion and insertion mutants of rgRSV provide complete protection against RSV

infection

Since rgRSV-ΔM1558, ΔD1557-ΔM1558, and 1557-A-1558 were highly attenuated in vivo, we

next determined whether they are immunogenic. Briefly, 6-week-old female cotton rats were

5 intranasally immunized with 2×10 TCID50 of recombinant parental rgRSV, or rgRSV deletion or

insertion viruses. The parental rgRSV was used as a control. An ideal attenuated vaccine candidate

should retain a similar level of immunogenicity compared to the parental rgRSV. After

immunization, serum antibody levels were determined weekly by a plaque reduction neutralization

assay. Figure 3.9A showed the dynamics of neutralizing antibody responses following

immunization. Recombinant rgRSV-ΔM1558 and ΔD1557-ΔM1558 triggered levels of

neutralizing antibody comparable to those induced by rgRSV immunization (P > 0.05). However,

rgRSV-1557-A-1558 induced relatively lower antibody responses at weeks 3 and 4 compared to

rgRSV (P<0.05). No RSV-specific neutralizing antibody was detected in the unvaccinated control.

5 At week 4 post-immunization, cotton rats were challenged intranasally with 2×10 TCID50 of

rgRSV. At day 4 post-challenge, cotton rats were sacrificed, and viral replication in nasal 121 turbinates and lungs, and pulmonary histology were examined. No infectious RSV was detectable in either nasal turbinates or lungs from the animals immunized with rgRSV, rgRSV-ΔM1558,

ΔD1557-ΔM1558, 1557-A-1558 followed by challenging with rgRSV (Figure 3.9B).

Microscopically, all five groups had variable but mild pulmonary changes. No enhanced lung damage was found for the mutants. Therefore, these results showed that cotton rats immunized with rgRSV deletion and insertion mutants were protected from RSV challenge.

122

Figure 3.9. Immunogenicity of rgRSV mutants in cotton rats. 123

5 Four-week-old SPF cotton rats were inoculated intranasally with 2.0×10 TCID50 of each rgRSV.

Blood samples were collected from each rat weekly by retro-orbital bleeding. (A) Neutralizing

antibody production. The RSV-neutralizing antibody titer was determined using a plaque reduction

neutralization assay, as described in Materials and Methods. (B) Protection from RSV challenge.

5 At week 4 post-immunization, cotton rats were challenged with 2.0×10 TCID50 rgRSV. At day 4

post-challenge, the cotton rats were sacrificed, and lungs and nasal turbinates were collected for

virus titration by TCID50 assay. Viral titers are the geometric mean titer (GMT) of 5 animals ± standard deviation. The detection limit is 2.0 log TCID50/g tissue. *P<0.5, **P < 0.01,

***P < 0.001, ****P < 0.0001.

3.5 Discussion

Despite major efforts, there is no FDA-approved vaccine for RSV. In this study, we generated

three RSV live attenuated vaccine candidates by deleting and inserting single or double amino acid

residues in the flexible hinge region between CR V and CR VI of the L protein. We showed that

these vaccine candidates were genetically stable, highly attenuated in immortalized cells, HBE

cultures, and an animal model, and were capable of inducing a high levels of innate immune

cytokines in HBE culture and triggering a high level of neutralizing antibody and providing

complete protection against RSV infection in cotton rats. To the best of our knowledge, it is first

time that the flexible hinge region of L protein has been shown to tolerate amino acid deletion,

although it has been shown that this region is able to tolerate an in-frame GFP insertion for several

NNS RNA viruses.

124

The L protein is an important target for designing live attenuated vaccine candidates for RSV,

since the L protein is a multifunctional protein containing domains that perform at least 10

functions, including nucleotide polymerization, replication, transcription, mRNA capping, cap

methylation, and polyadenylation [31, 174, 175]. Suppression or partial suppression of these

enzymatic activities will lead to diminished viral replication and gene expression, which

potentially results in virus attenuation. Between 1960 and 1990, several RSV attenuated strains were developed by classic methods such as cold or warm passage (cp or ts), and chemical

mutagenesis [93, 185-190]. Interestingly, amino acid residue(s) responsible for the temperature-

sensitive (ts) phenotype of cp ts RSV mutants often mapped to the CRI-IV region of L protein

[191, 192], which is essential for polymerase activity. Consistent with these findings, random

substitutions of a cluster of charged amino acids to alanine in the L protein resulted in recombinant

RSVs that also exhibited a ts phenotype [193]. Interestingly, mutations responsible for host range

and ts phenotype of other NNS RNA viruses (such as VSV and Sendai virus) map to the flexible

region or CR VI in the L protein which affected the mRNA cap MTase activity [160, 182, 194].

Genetic stability is one of the major challenges for developing a live attenuated RSV vaccine. The primary advantage of using amino acid deletion and insertion is that this strategy can permanently retain the genetic stability, as it would not be possible for these deletion or insertion mutants to

revert to wildtype virus. In our study, no reversion or additional mutations were observed when

these RSV mutants were repeatedly passaged in Vero cells. In addition, RT-PCR fragments

amplified from nasal turbinate and lung tissues retained the desired deletion and insertion

mutations. These results suggest that these RSV mutants are genetically stable. Historically, many live attenuated RSV vaccine candidates exhibited good attenuation and immunogenicity but were

125 genetically unstable, leading to the reversion and de-attenuation when they were passed in cell culture, in animal models, or in preclinical or clinical trials. A cp ts RSV live vaccine candidate

(called MEDI-559) was one of the most promising candidates that had been advanced for human clinical trials [195]. However, it was genetically unstable when tested in children and infants in a clinical trial [196]. To solve this problem, a stabilized version of MEDI-559 (called RSV cps2) was generated by using alternative codons for mutations 248 (L codon 831) and 1030 (L codon

1321) [197]. Unexpectedly, this led to a spontaneous second-site compensatory mutation at position S1313 in the L protein, which resulted in the de-attenuation of the virus. The authors finally deleted the amino acid S1313 which resulted in a recombinant virus that had robust genetic stability. A virus containing the codon 1313 deletion combined with deletion of the NS2 gene is currently being developed as a potential RSV vaccine candidate [192]. Clearly, amino acid deletion enhanced genetic stability.

We found for the first time that the flexible hinge region tolerates amino acid deletion. Notably, a single or double deletion was sufficient to achieve attenuation, highlighting this strategy as a novel approach for rational design of live attenuated vaccines for RSV and perhaps for other NNS RNA viruses. Deletion of the amino acid at position 1557 (rgRSV-ΔD1557) moderately attenuated growth in immortalized cells and HBE cultures. In cotton rats, rgRSV-ΔD1557 exhibited 0.7 and

1.4 log reductions in viral replication in the lung and nasal turbinate, respectively. However, deletion of the M residue at the neighboring 1558 position (rgRSV-ΔM1558) achieved a highly attenuated phenotype. Interestingly, combining these two deletions into one virus did not have a further, synergetic effect on attenuation. Rather, this recombinant virus (rgRSV-ΔD1557-

ΔM1558) exhibited a level of attenuation between rgRSV-ΔD1557 and rgRSV-ΔM1558. If such

126

a double mutant were found to be attenuated in infants, it would have the theoretical safety feature

that if one of the deleted amino acids, L-ΔD1557 could be reinserted by the viral polymerase, the

resulting virus (L-ΔM1558) would be more attenuated and less likely to be selected in vivo. These

two single deletion mutants had delayed CPE, protein synthesis, and growth kinetics in

immortalized cells. Similarly, they were highly attenuated in virus spread and virus release in HBE

culture, a near in vivo airway model for RSV infection. Importantly, these two recombinant viruses

were highly attenuated in the cotton rat, the best available small animal model for RSV infection.

A near detection limit level of viral replication was observed in both nasal turbinates and lungs.

Importantly, cotton rats immunized with a single dose of each recombinant virus triggered a high

level of neutralizing antibody and were completely protected from RSV replication. In addition,

no enhanced damage was observed upon challenge. Another important characteristic of these two

RSV mutants is that they grow to a high titer in cell culture despite a high attenuation. Recombinant

rgRSV-ΔM1558 and rgRSV-ΔD1557-ΔM1558 only had a 0.5 and 0.3 log reduction in peak titer

compared to the parental rgRSV. Thus, it should be economically feasible to produce these vaccine

candidates. Therefore, these two RSV L deletion mutants have excellent attenuation characteristics

and immunogenicity and thus are promising vaccine candidates for RSV.

We also found that a single alanine insertion at the flexible hinge region was sufficient to attenuate

RSV. The resultant recombinant virus rgRSV-1557-A-1558 was highly attenuated in cell culture,

HBE culture, and cotton rats. Although it triggered a relatively lower neutralizing antibody response compared to the two RSV deletion mutants, it provided complete protection against RSV replication. Previously, in-frame insertion of GFP into this region has been reported for several L protein of NNS RNA viruses. In the case of MeV, c-Myc tag or EGFP tag was inserted in the

127

flexible region [176]. Recombinant MeV containing an EGFP insertion (240 amino acids) was more attenuated in cell culture than rMeV with c-Myc tag (6 amino acids) at the same position, suggesting that the size of insert correlates with the degree of attenuation [176]. For CDV, it was shown that insertion of EGFP at this position resulted in over-attenuation of the virus [178]. A single intranasal immunization with this recombinant virus provides partial protection against challenge with the virulent parental CDV virus [178]. Insertion of GFP into this flexible region of

VSV L resulted in a virus displaying unusual properties, including a temperature-sensitive growth phenotype, a defect in packaging of L-protein into virions, and a lack of virion-associated polymerase activity in vitro [179].

Although the exact mechanism(s) behind the attenuation of these RSV deletion and insertion

mutants is not clear, one possibility is that these insertion and deletion RSV L proteins reduced the

polymerase activity which leads to the synthesis of less genomic and antigenomic RNA as well as

mRNAs. In a minigenome assay, we found significantly less GFP expression for these L protein

deletion and insertion mutants, suggesting that they either affect replication, transcription, or both.

In RSV-infected cells, we found that synthesis of viral F and G proteins were significantly delayed

at early time points compared to the parental RSV. Similarly, the EGFP inserted into the PRV and

CDV L proteins significantly reduced polymerase activity, retaining 1-85% and 30-60% of wt

polymerase activity, respectively, depending on the concentration of L plasmid used for the

minigenome assay [178]. For VSV, insertion of EGFP into the flexible region of L (LEGFP) resulted

in a ts phenotype for polymerase activity [179]. The LEGFP had significantly reduced polymerase

activity at 37°C and exhibited no activity at 37.5°C in cultured cells. In addition, LEGFP had no

detectable polymerase activity in an in vitro RNA reconstitution assay using purified virions.

128

Another possible mechanism of viral attenuation by these L deletion and insertion mutants is that

they affect mRNA capping and/or methylase activities since they are located in the region between

the mRNA capping and methylase domains. In fact, it was previously shown that single amino

acid substitutions in the flexible region of VSV L significantly reduced both G-N-7 and ribose 2’-

O MTase activity [182]. This is interesting because MTase catalytic sites are located in CR VI, but not the flexible hinge region, of the L protein. Mutations in this hinge region would likely affect

the spacing between the polymerase and the methylation domains of the L protein, putting the

methyltransferase out of position to efficiently methylate the nascent mRNA cap. But it is also

possible that these mutations affect the conformation of the CR V domain enough to reduce its

capping activity. Future experiments will address whether these L deletion and insertion mutants

alter the mRNA capping, MTase or polymerization activities.

129

Chapter 4. Viral N6-methyladenosine upregulates replication and pathogenesis of human

respiratory syncytial virus

4.1 Abstract

N6-methyladenosine (m6A) is the most prevalent internal modification of mRNAs in most

eukaryotes. Here we show that RNAs of human respiratory syncytial virus (RSV) are modified

by m6A within discreet regions and that these modifications enhance viral replication and

pathogenesis. Knockdown of m6A methyltransferases decreases RSV replication and gene

expression whereas knockdown of m6A demethylases has the opposite effect. The G gene

transcript contains the most m6A modifications. Recombinant RSV variants expressing G

transcripts that lack particular clusters of m6A display reduced replication in A549 cells,

primary well differentiated human airway epithelial cultures, and respiratory tracts of cotton

rats. One of the m6A-deficient variants is highly attenuated yet retains high immunogenicity

in cotton rats. Collectively, our results demonstrate that viral m6A methylation upregulates

RSV replication and pathogenesis and identify viral m6A methylation as a target for rational

design of live attenuated vaccine candidates for RSV and perhaps other pneumoviruses.

4.2 Introduction

Internal N6-methyladenosine (m6A) is the most prevalent modification in the mRNA of most

eukaryotes [198, 199]. Although m6A was discovered in the 1970s [113, 136, 200, 201], its

biological functions remained a mystery for decades. In 2011, m6A demethylases were discovered 130

[122, 123]. It is generally believed that m6A methylation is reversible although studies also suggest

that substantial degree of demethylation did not occur in some mRNA exons [202]. Transcriptome-

wide mapping of m6A by high-throughput sequencing technology was reported subsequently [114,

203, 204]. A series of m6A-related enzymes and proteins were also discovered and characterized

in mammalian cells. Recent work has found that m6A regulates RNA metabolism, protein

translation, gene expression, and embryonic development in organisms ranging from plants to mice [130, 205-208].

As obligate intracellular parasites, viruses must synthesize their own genetic material and carry out their reproduction while avoiding innate immune surveillance by mimicking their host. Studies

from the early 1970s showed that viral RNAs of several DNA viruses, and influenza

virus contained internal m6A modifications [136-139, 209-211]. Although it is still controversial

whether viral m6A positively or negatively regulates HIV replication [141, 142, 212], it was shown

that m6A promotes gene expression of influenza virus [145] and simian virus 40 [151]. In the case

of Kaposi's sarcoma-associated herpesvirus (KSHV), the impact of m6A machinery on viral gene

expression is dependent on the cell type [146, 147, 154]. In contrast, m6A negatively regulates the

production of (HCV) and Zika virus[149]. However, the mechanism(s) by which

m6A methylation regulates the virus life cycle remain poorly understood. In addition, the role of

m6A in viral virulence, pathogenesis, and immunity has not been studied.

Non-segmented negative-sense (NNS) RNA viruses encompass a wide range of significant human,

animal, and plant pathogens. For many of these agents, there are no effective vaccines or antiviral drugs. The replication and gene expression strategy of NNS RNA viruses is unique. During

transcription, the polymerase transcribes the linear array of genes in the viral genome into 5-10

131

mRNAs that are capped and methylated at the 5’ end and polyadenylated at the 3’ end. During

replication, the RdRP initiates at the extreme 3’ end of the genome and ignores the gene junctions

to synthesize a full-length complementary antigenome which is encapsidated by N protein and

subsequently serves as template for the synthesis of full-length progeny genomes that are likewise

encapsidated by N protein [213, 214]. A good example of NNS RNA viruses is human respiratory

syncytial virus (RSV), a member of the Pneumoviridae [215]. RSV is the most important cause of

upper and lower respiratory tract infection of infants, young children, and immunocompromised

individuals and second only to influenza virus for the elderly [166]. Worldwide it is estimated that

RSV causes 3.4 million hospitalizations and between 66,000 and 199,000 deaths in children less

than 5 years of age [6]. Despite major efforts, no vaccine or antiviral drug is yet available for RSV

[166].

Here, we find that the RSV genome, antigenome, and mRNAs are m6A methylated internally at

specific sites which positively regulate RSV replication, gene expression, and virus production in

HeLa and A549 cells. Subsequently, the m6A sites in the viral G gene, the most abundant m6A

enrichment gene, are mutated in an infectious cDNA clone of RSV. The resultant m6A deficient

rgRSVs have significant defects in replication, gene expression, spread, and virus release in A549

cells and primary well differentiated human airway epithelial (HBE) cultures. These m6A mutated

rgRSVs are defective in viral replication in the upper and lower respiratory tract of cotton rats and

produce less pathology in the lungs. Importantly, we find that one m6A-mutated rgRSV is completely attenuated in cotton rats yet retains a wild-type level of immunogenicity. Collectively, these results reveal that m6A upregulates each step in the RSV replication cycle and viral

132

pathogenesis, and identify m6A as a new target for the rational design of live attenuated vaccine

candidates and antiviral drugs for RSV.

4.3 Materials and Methods

4.3.1 Cell lines

HeLa (ATCC CCL-2), A549 (ATCC CCL-185), Vero (ATCC CRL-CCL81), and HEp-2 (ATCC

CCL-23) cell lines were purchased from the American Type Culture Collection (Manassas, VA)

and were grown in Dulbecco's modified Eagle's medium (DMEM; Life Technologies)

supplemented with 10% FBS. HeLa cells overexpressing the empty vector (pPB-CAG), YTHDF1,

YTHDF2, or YTHDF3 were maintained in DMEM, 10% FBS and 1 µg/ml of puromycin every

passage to select for YTHDF1-3 overexpressing cells. Primary, well-differentiated human airway

epithelial (HBE) cultures were grown on collagen coated Transwell inserts (Corning Incorporated,

Corning, NY) at an air-liquid interface, as previously described [164]. Upon reaching confluency

and forming tight junctions, the apical medium was removed and cultures were maintained at the

air-liquid interface for 4 to 6 weeks to generate well-differentiated, polarized cultures. All cell

lines used in this study were free of mycoplasma, as confirmed by the LookOut Mycoplasma PCR

Detection Kit (Sigma).

4.3.2 Plasmids, site-directed mutagenesis and virus preparation

The pPB-CAG plasmid vector was used to overexpress the readers (YTHDF1-3), writers

(METTL3, METTL14), and erasers (FTO, ALKBH5) as described previously [141]. Plasmid

(RW30) encoding the full-length antigenomic cDNA of RSV strain A2 with GFP inserted between

the leader and the NS1 gene, and support plasmids expressing RSV A2 strain N protein (pTM1-

N), P protein (pTM1-P), L protein (pTM1-L), and M2-1 protein (pTM1-M2-1) were generously

133 provided by Dr. P.L. Collins, NIAID, Bethesda, MD. Mutations to the potential m6A sites in G gene were introduced into the RW30 plasmids using QuikChange site-directed mutagenesis kit

(Stratagene, La Jolla, CA). The m6A peaks in the G gene are clustered in three regions, 392-467nt,

567-660nt, and 716-795nt. Since it is known that m6A modified sites in RNA contain the conserved

Pu [G>A]m6AC[A/C/U] motif (Pu represents purine) [198], we searched for this motif in these three regions in G mRNA and identified 6, 7, and 4 potential m6A sites in regions G1, G2, and G3 respectively. The potential m6A sites mutants in G1 peak include 394-AGm6ACC-400; 401-

AAm6ACA-407; 418-AAm6ACA-424; 444-AAm6ACA-450; 455-AAm6ACA-461; 459-

AAm6ACC-465; mutants in G2 peak include 569-AAm6ACA-575; 576-AAm6ACC-582; 589-

AAm6ACC-595; 612-AAm6ACC-618; 625-AGm6ACA-631; 645-AAm6ACC-651; 652-

AAm6ACC-658; and mutants in G3 peak include 718-AAm6ACA-724; 722-AAm6ACA-728; 768-

GAm6ACT-774; 787-AAm6ACC-793. The A or C within the consensus m6A sites was mutated to a T or G in these sites without changing the encoded amino acid. Mutant G12 combined the mutations from G1 and G2. Mutant G123 was a combined the mutations from G1, G2, and G3. In addition, M-fold and Genscript software were used to predict that these mutations did not alter

RNA secondary structure or codon usage. All plasmids and mutations were confirmed by DNA sequencing.

Recombinant RSV were recovered using standard protocol. All viruses were propagated and titered in HeLa cells or A549 cells. Highly purified RSV were obtained using purification protocol as described previously.

134

4.3.3 m6A-seq

High-throughput sequencing of the RSV and host methylome was carried out using m6A-seq

(MeRIP-seq) as described previously [216]. For m6A-seq of the rgRSV genome and antigenome,

RNAs were extracted from purified rgRSV virions and purified with the RiboMinus Eukaryote

System v2 kit (Thermo Fisher). For m6A-seq of host transcripts, total RNAs were extracted from mock or rgRSV-infected HeLa or A549 cells and polyadenylated RNAs were isolated using

Dynabeads mRNA DIRECT Purification kit (Thermo Fisher). Purified RNAs were sonicated with

Bioruptor Pico (Diagenode) with 30 s ON 30 s OFF for 30 cycles, mixed with 1 µl of affinity

purified anti-m6A monoclonal antibody (NEB) in IPP buffer (150 mM NaCl, 0.1% NP-40, 10 mM

Tris-HCl, pH 7.4) and incubated for 2 h at 4°C. Enriched mRNA fragments were purified with

RNA Clean & Concentrator kit (Zymo) and used for library generation with TruSeq Stranded mRNA Library Prep kit (Illumina). Sequencing was carried out on Illumina HiSeq 4000 according

to the manufacturer’s instructions. Two replicates of RNA samples from virions, virus-infected cells, and mock-infected cells were subjected to m6A-seq. For data analysis, after removing the

adapter sequences, the reads were mapped to the human genome (hg38) and rgRSV genome and

antigenome by using Hisat2 [217] with peak calling as described [218]. Metagene analysis was

performed by R package Guitar [219]. Differential methylation analysis was performed with count

based negative binomial model implemented in QNB test [220].

4.3.4 Quantification of RSV RNA m6A level using LC-MS/MS

RSV RNA (250 mg) was extracted from highly purified rgRSV virions using an RNeasy Mini kit

(Qiagen) and purified twice with RiboMinus Eukaryote System v2 kit (Thermo Fisher). To

examine the purity of virion RNA, oligo d(T) was used for reverse transcription, followed by qPCR

135

for quantification for β-actin and viral N and G mRNAs. Virion RNA which was free of

contamination of host RNA and viral mRNAs was used for liquid chromatography-mass

spectrometry (LC-MS/MS), m6A antibody pulldown assay, and m6A-seq. Purified RNA was

digested and subjected to a quantitative analysis of the m6A level using LC-MS/MS as previously described [122].

4.3.5 Quantification of RSV RNA m6A level using anti-m6A antibody

A549 cells in T150 flasks were transfected with 10 µg of plasmids encoding Mettl3 and Mettl14,

AlkBH5, or vector pCAGGS. For siRNA transfection, A549 cells in T75 flasks were transfected

with 150 pmole of siRNA targeting Mettl3 and Mettl14, AlkBH5, or control siRNA. At 24 h post- transfection, the transfected cells were infected with rgRSV at an MOI of 0.1. At 42 h post-

infection, cell culture supernatants (containing RSV particles) were harvested. RSV particles were

pelleted and purified through ultracentrifugation. Virion RNA was extracted from highly purified

RSV virions. Antigenome was quantified by real-time RT-PCR. Each amount of antigenome was

bound to strip wells using a RNA high binding solution, and m6A was detected using a specific

capture anti-m6A antibody (Abcam, ab185912) and then quantified colorimetrically by reading the

absorbance in a microplate spectrophotometer at a wavelength of 450 nm. A standard curve was

generated using known m6A methylated RNA (range from 0.02 to 1 ng of m6A) as a positive

control. The m6A content was calculated from each RNA samples based on their OD450 values.

The percent of changes was calculated by dividing m6A contents in viral RNA from the treated

group by those from the control group.

136

4.3.6 Host Cell Gene Differential Expression analysis

Host cell differential gene expression was analyzed by R package DESeq2 [219] using wald-test.

The significantly differentially expressed genes were reported at adjusted P value cutoff of 0.05.

4.3.7 Gene Ontogeny (GO) analysis

GO analysis was performed using the R package cluster Profiler [219]. Specifically, enrichKEGG function was called to analyze for enriched pathway and enrichMap function was called to generate network plot of enriched pathway.

4.3.8 siRNA and siRNA transfection

siRNAs against METTL3, METTL14, FTO, ALKBH5, YTHDF1, YTHDF2, YTHDF3 or non-

targeting AllStars negative control siRNA were purchased from Qiagen (Valencia, CA). All

siRNA transfections were performed using the Lipofectamine 3000 transfection reagent (Thermo-

Fisher) according to the manufacturer’s instructions.

4.3.9 Antibodies and Western blotting

The antibodies used in this study were anti-YTHDF1 (1:1000, Proteintech, Rosemont, IL), anti-

YTHDF2 (1:1000, Abcam, Cambridge, MA), anti-YTHDF3 (1:1000, Abcam, Cambridge, MA), anti-METTL 3 (1:1000, Proteintech, Rosemont, IL), anti-METTL 14 (1:1000, Proteintech,

Rosemont, IL), anti-ALKBH5 (1:1000, Sigma-Aldrich, St. Louis, MO), anti-FTO (1:1000,

Abcam, Cambridge, MA), and anti-RSV serum (1:400, Virostat, Westbrook, ME), F (1:3000,

Abcam, Cambridge, MA), anti-FLAG (1:3000, Sigma-Aldrich, St. Louis, MO), anti-Actin

(1:5000, Proteintech, Rosemont, IL) and anti-Tubulin (1:5000, Abcam, Cambridge, MA). Cells were harvested and lysed in RIPA buffer (Abcam, Cambridge, MA) supplemented with protease

137 inhibitor cocktail (Sigma-Aldrich). Western blotting was performed as described. Tubulin or actin was used as a loading control.

4.3.10 Immunofluorescence analysis (IFA) and confocal microscopy

For flow cytometry, A549 cells were infected with rgRSV or mutants at an MOI of 0.1 or 1, and

GFP expression was monitored by fluorescence microscopy. At the indicated time points, cells were trypsinized and fixed in 4 % of paraformaldehyde solution and the number of GFP-positive cells quantified by flow cytometry.

For IFA assay, mock or rgRSV-infected cells were fixed in acetone and methanol at the ratio of

1:1 for 30 min, and blocked with 5% milk in PBST. Slides were stained with all primary antibodies

(1:100), washed 3 times with PBST, and stained with conjugated Alexa Fluor secondary antibodies

Alexa Fluor 488/594 (Thermo-Fisher; 1:300), and mounted with SlowFade™ Diamond Antifade

Mountant with DAPI (Thermo-Fisher). Imaging was performed on an Olympus FV 1000 confocal microscopy system at The Ohio State University Campus Microscopy & Imaging Facility.

4.3.11 Real-time RT-PCR

RSV genome, antigenome, and mRNA were quantified by real-time RT-PCR. HeLa or A549 cells were infected with rgRSV or an rgRSV mutant at an MOI of 0.1. At 12, 18 and 24 post-infection, total RNA was isolated from cells using TRIzol (Life Technologies). Viral genome or antigenome copies were quantified by real-time RT-PCR using two primers specifically targeting the RSV leader sequence and GFP gene (Supplementary Table 8). Poly (A)-containing viral mRNAs were isolated from total RNA using a Dynabead mRNA isolation kit (Life Technologies) according to the manufacturer’s recommendations. Using the viral mRNAs as the template, the NS1 and G

138

mRNA copies were quantified by real-time RT-PCR using two primers targeting the viral NS1 and

G genes, respectively.

4.3.12 RNA-immunoprecipitation (RIP)

The RIP assay was performed as described previously [149]. Briefly, HeLa cells were infected

with rgRSV at MOI of 1.0 and cell extracts were harvested in polysome lysis buffer after 36 h

post-infection. RNP complexes were immunoprecipitated with anti-HA antibody conjugated to

magnetic beads (Sigma) or anti-YTHDF2 antibody overnight at 4°C, and washed five times with

ice-cold NT2 buffer. For the RIP with anti-YTHDF2 antibody, additional secondary antibody was

added. After the final wash, 10% of the beads were used for immunoblotting and the remaining

90% were used for RNA extraction using TRIzol (ThermoFisher).

4.3.13 Characterization of recombinant RSV in vitro and in vivo

Viral replication kinetics, viral protein expression level and genetic stability were tested as

described previously in vitro. Cotton rats were used to assess the replication, pathogenesis and

immunogenicity of each recombinant RSV. Thirty 6-week-old specific-pathogen-free (SPF) male

cotton rats (Envigo, Indianapolis, IN) were randomly divided into 6 groups (5 cotton rats per

group). Prior to virus inoculation, the cotton rats were anesthetized with isoflurane. The cotton rats

5 in group 1 were inoculated with 2.0×10 TCID50 of parental rgRSV and served as positive controls.

5 6 The cotton rats in groups 2 to 5 were inoculated with 2.0×10 TCID50 of four m A deficient rgRSV

mutants, rgRSV-G1, G2, G3, and G12. Each cotton rat was inoculated intranasally with a volume of 100 μl. At day 4 post-infection, the cotton rats were sacrificed via carbon dioxide inhalation.

The left lung and nasal turbinates were collected for virus titration as reported and the right lung

was collected for histological analysis.

139

For the immunogenicity study, twenty 6-week-old female cotton rats (Envigo) were randomly divided into five groups (5 cotton rats per group). Cotton rats in groups 1, 2, and 3 were intranasally

5 6 inoculated with 2.0×10 TCID50 of two m A deficient rgRSV mutants (rgRSV-G1 and G12) and rgRSV, respectively. Cotton rats in groups 4 were mock-infected with DMEM and served as unvaccinated challenged control. After immunization, the cotton rats were evaluated daily for any possible abnormal reaction and blood samples were collected from each cotton rat weekly by facial vein retro-orbital plexus sampling, and serum was used for detection of neutralizing antibodies. At

5 4 weeks post-immunization, the cotton rats in groups 2 to 5 were challenged with 2.0×10 TCID50 of parental rgRSV via intranasal route, and evaluated twice daily for the presence of any clinical symptoms. At 4 days post-challenge, all cotton rats were euthanized and evaluated as reported.

The immunogenicity of rgRSV mutants was assessed based on their ability to trigger neutralizing antibody (determined as previously described), the ability to prevent rgRSV replication in lungs and nose, and the ability to protect lung from pathological changes.

After sacrifice, the right lung of each animal was removed, inflated, and fixed with 4% neutral buffered formaldehyde. Fixed tissues were embedded in paraffin and a microtome used to generate

5 μm sections. Slides were then stained with hematoxylin-eosin (H&E) for the examination of histological changes by light microscopy. Histopathological changes were evaluated based on the extent of interstitial inflammation, edema, and peribronchiolar inflammation.

4.3.14 Statistical analysis

Quantitative analysis was performed by either densitometric scanning of autoradiographs or by using a phosphorimager (Typhoon; GE Healthcare, Piscataway, NJ) and ImageQuant TL software

(GE Healthcare, Piscataway, NJ). Statistical analysis was performed by one-way multiple

140

comparisons using SPSS (version 8.0) statistical analysis software (SPSS Inc., Chicago, IL). A P

value of <0.05 was considered statistically significant.

4.4 Results

4.4.1 The RSV genome and antigenome/mRNAs are m6A methylated

RSV has a NNS RNA genome of 15,222 nucleotides (RSV A2 strain). As is typical for NNS RNA viruses, replication of the viral genomic RNA (vgRNA) produces an exact, positive-sense full- length complementary RNA (cRNA) antigenome [174]. Both the genome and antigenome are encapsidated by the nucleocapsid (N) protein and both nucleocapsids can be packaged into virions,

as for many NNS RNA viruses [221]. To investigate whether RSV RNA contains m6A, RNA was extracted from highly purified virions grown in HeLa cells, and the purity of RNA was examined by real-time RT-PCR to ensure to be free of any contamination of host RNAs and viral mRNAs.

The presence of m6A in viral RNA was quantified by liquid chromatography-tandem mass

spectrometry (LC-MS/MS). We found that approximately 0.7% of the A bases were m6A

methylated in RSV viral RNAs, a somewhat higher level than the host mRNAs (0.1-0.4%).

To locate the m6A sites on RSV RNA, we sonicated virion RNA and subjected it to m6A-specific

antibody immunoprecipitation followed by high throughput sequencing (m6A-seq), then mapped

all the reads onto either the genome or antigenome sequence. Several m6A peaks were identified

on both strands of the viral RNA (Figure. 4.1a). The RSV antigenomic RNA contained major m6A

peaks in the regions complementary to the N, P, G, and F genes and in the regions complementary

to the two regulatory elements, the gene end (ge) sequence of N and the intergenic (ig) sequence

between the P and M genes in the genome (Figure. 4.1a). In the genomic RNA, eleven m6A peaks

were detected in the NS2, N, P, M, G, and L genes and four regulatory elements including the gene

141

start (gs) of NS2, ig between P and M, ge of M, ig between M and SH, ge of G, and ig between G and F (Figure. 4.1a). Since we used a recombinant RSV harboring GFP between the leader and the NS1 gene (rgRSV), we also searched whether GFP region contains m6A. An m6A peak with a

size of 60 nt was detected in GFP gene in genome. No m6A peak was found in GFP region in

antigenome. The G gene regions from both genome and antigenome have the strongest m6A

enrichment with peak size of 822 nt and 672 nt, respectively, indicating that there may be multiple

adjacent m6A sites in these regions. Together, these results confirm that both RSV genome and

antigenome RNAs contain m6A.

We also mapped m6A peaks in mRNAs purified from RSV-infected cells. To do this, total RNA was isolated from rgRSV-infected HeLa cells, enriched for mRNA by binding to oligo dT, and

subjected to m6A-seq. As RSV mRNAs contain poly(A) and are subsequently detected from

poly(dT)-enriched m6A-seq, we identified 16 m6A peaks from RSV mRNAs (Figure. 4.1b) which

were largely overlapped with those of antigenome (Figure. 4.1a). Interestingly, the G gene

transcript has the strongest m6A enrichment with 846 bp peak size. In addition, no m6A peak was detected in GFP mRNA in virus-infected HeLa cells.

We next performed m6A-seq of rgRSV grown in A549 cells, a relevant cell line for RSV infection.

Similar to HeLa cells, we found that RSV genome, antigenome, and mRNAs were m6A methylated

in A549 cells (Figure. 4.1c and d). For virion RNAs, a total of 9 and 15 m6A peaks were identified

in the genome and antigenome respectively (Figure. 4.1c). Similar to virions grown in HeLa cells,

the location of m6A peaks identified from genome and antigenome largely overlap. G gene regions from both genome and antigenome have the strongest m6A enrichment with 696 and 846 bp peak

size, respectively. An m6A peak was detected in GFP gene in antigenome but no m6A peak was

142

detected in GFP gene in genome. For the RNAs extracted from virus-infected cells, a total of 18

m6A peaks were identified in RSV mRNAs (Figure. 4.1d). Again, the G gene transcript has the strongest m6A enrichment with 1046 bp peak size. In addition, one m6A peak was found in GFP

mRNA in infected A549 cells.

We next analyzed the overlapping regions based on m6A-seq data from HeLa and A549 cells. For

virion RNA, six and four overlapping regions were identified in the genome (gs of NS2, NS2, N,

P, ig between P and M, and G) and antigenome (N, M, G, and F), respectively. For RNAs purified

from RSV-infected cells, 11 overlapping m6A peaks were also found in mRNAs, respectively.

Although there are some differences, the majority of m6A peaks are highly conserved between the

two cell lines suggesting that RSV utilizes the host m6A machinery to methylate these specific

sites.

143

Figure 4.1. The RSV genome and antigenome/mRNAs are m6A methylated. 144 a. Distribution of m6A peaks in the RSV antigenome and genome of virions grown in HeLa cells.

Confluent HeLa cells were infected by rgRSV at an MOI of 1.0, supernatant was harvested at 36 h post-infection. RSV virions were purified by sucrose gradient ultracentrifugation. Total RNAs were extracted from purified virions and were subjected to m6A-specific antibody immunoprecipitation followed by high-throughput sequencing (m6A-seq). A schematic diagram of partial RSV antigenome is shown, as most m6A peaks are clustered in these regions. The normalized coverage from m6A-seq of RSV RNA showing the distribution of m6A- immunoprecipitated (IP) reads mapped to the RSV antigenome (blue block) and genome (pink block). The baseline distributions for antigenome and genome from input sample are shown as a blue and pink line respectively. Data presented are the averages from two independent virion samples (n =2). b. Distribution of m6A peaks in the RSV mRNAs from RSV infected HeLa cells.

Confluent HeLa cells were infected by rgRSV at an MOI of 1.0, cell lysates were harvested at 36 h post-infection. Total RNAs were extracted from cell lysates, and were enriched for mRNA by binding to oligo dT, and subjected to m6A-seq. The distribution of m6A-immunoprecipitated (IP) reads were mapped to the RSV mRNAs (pink block). The baseline distributions for mRNAs from input sample are shown as a pink line. Data presented are the averages from two independent virus- infected HeLa cell samples (n= 2). c. Distribution of m6A peaks in the RSV antigenome and genome of virions grown in A549 cells. Data presented are the averages from two independent virion samples (n = 2). d. Distribution of m6A peaks in the RSV mRNAs from RSV-infected A549 cells. Data presented are the averages from two independent virus-infected A549 cell samples (n

=2)

145

4.2 RSV infection alters the m6A distribution of host RNAs

We next determined the effects of RSV infection on the abundance and distribution of m6A on

cellular transcripts. Metagene analysis showed that RSV-infected and mock-infected HeLa cells have m6A peaks enriched near the start and stop codons of open reading frames (Figure. 4.2A),

which is consistent with the known distribution of m6A sites on transcripts [114, 203, 204]. Unlike

the distribution of m6A peaks on mRNA, the peaks are mostly uniformly distributed on lncRNA

with slightly more enrichment at its 5’ end (Figure. 4.2B). The distribution of m6A peaks in each

annotation also recapitulate m6A site distribution [114, 203, 204] with the majority of peaks

residing in the CDS and 3’ UTR regions (Figure. 4.2B and D). Differential peak analysis using the

count based QNB test [220] identified 2256 differentially methylated peaks. Analysis of RNA-seq

data from the host cell (HeLa) revealed over 9,000 differentially expressed genes at an adjusted P

value cutoff of 0.05. These data suggest RSV infection significantly altered both the

epitranscriptome and the transcriptome of the host cells. Pathway enrichment analysis shows

differentially expressed genes are enriched in pathways including , metabolism,

autophagy, RNA synthesis and transport, and response to viral infection (Figure. 4.2E).

As expected, the distribution of m6A sites was highly conserved between cell lines (Figure. 4.3A-

D). RSV infection altered the expression of over 7,000 host cell genes in A549 cells (Figure. 4.2)

involved in a series of signal pathways (Figure. 4.3E) despite very few m6A peaks were found to

be differentially methylated. We also analyzed the differentially expressed genes which are

overlapped in pathways between HeLa and A549 cells. Many of these overlapped differentially

expressed genes are involved cell cycle, metabolism, TNFα signaling, and RNA transport (Figure.

146

4.3F). Therefore, RSV infection may have widespread effects on host gene expression partially

attributed to alteration of the deposition of m6A.

Continued

Figure 4.2. RSV infection alters the methylome of host transcripts in HeLa cells.

147

Figure 4.2 Continued

Total RNAs were isolated from mock-infected and rgRSV-infected HeLa cells. Poly(A) enriched mRNAs were purified and subjected to m6A-seq. (A) Metagene analysis of m6A peaks distribution along the human mRNA in control and infected HeLa cells. (B) Metagene analysis of m6A peak distribution on lncRNA. (C and D) Distribution of m6A peaks in the 5′ UTR, CDS, and 3′ UTR of host cell mRNA transcripts. Charts show the proportion of m6A peaks in the indicated regions in uninfected (C) and rgRSV-infected HeLa cells (D). (E) GO graphs showing pathway clusters from

148 differential expressed genes in rgRSV-infected HeLa cells. Data presented are the averages from duplicate samples (n = 2).

Continued

Figure 4.3. RSV infection alters the methylome of host transcripts in A549 cells.

149

Figure 4.3 Continued

150

Total RNAs were isolated from mock-infected and rgRSV-infected A549 cells. Poly(A) enriched

mRNAs were purified and subjected to m6A-seq. (A) Metagene analysis of m6A peak distribution

along the human mRNA in control and infected A549 cells. (B) Metagene analysis of m6A peak

distribution in lncRNA. (C and D) Distribution of m6A peaks in the 5′ UTR, CDS, and 3′ UTR of

host cell RNA transcripts. Charts show the proportion of m6A peaks in the indicated regions in

uninfected (C) and rgRSV-infected A549 cells (D). (E) GO graphs showing pathway clusters from

differential expressed genes in rgRSV-infected A549 cells. Data presented are the average results

from duplicate samples (n = 2). (F) GO graphs showing pathway clusters from differential

expressed genes which are overlapped between rgRSV-infected HeLa and A549 cells.

4.4.3 m6A reader proteins positively regulate RSV replication

To begin to explore the role of m6A modification in RSV replication and gene expression, we first

took advantage of HeLa cells that stably overexpress m6A “reader” proteins, YTHDF1, YTHDF2, and YTHDF3 (Figure. 4.4A). As shown in Figure. 4.4B, stronger GFP expression (more green cells, brighter cells) was observed in HeLa cells overexpressing YTHDF1-3 compared to the vector control. Quantification by flow cytometry showed that significantly more GFP-positive cells and higher GFP density were detected in HeLa cells overexpressing m6A reader proteins than

in the vector control (P<0.05) (Figure. 4.4C).

We next measured the expression of RSV F, G, and N proteins. As shown in Figure. 4.4D, more

F, G, and N proteins were detected in all three YTHDF-overexpressing HeLa cell lines.

Quantitative analysis showed a dramatic increase in viral protein expression during the first 12 h

although later time points were not as large (Figure. 4.4E, F, and G). Next, we measured the release of infectious virus particles in a single step growth curve. The RSV titer was significantly increased 151 in all three YTHDF-overexpressing cell lines (Figure. 4.4H) (P<0.05 or 0.01). Overexpression of

YTHDF2 had the most dramatic impact on virus production, increasing RSV titer by 1-2 logs compared to the vector control HeLa cells (P<0.05 or 0.01) (Figure. 4.4H).

The upregulating role of m6A reader proteins on RSV replication was also confirmed in HeLa cells transfected with plasmids expressing YTHDF1-3. Compared to HeLa cells stably overexpressing

YTHDF1-3 (Figure. 4.4D), transient overexpression of YTHDF1-3 led to a more robust enhancement of F and G protein synthesis and GFP expression.

We further analyzed viral replication and gene expression in A549 cells. Similar to the observations in HeLa cells, enhanced F, G, and N protein synthesis and GFP expression was detected when YTHDF1-3 proteins were overexpressed. We also tested RSV replication in Vero cells, the WHO-approved cell line for production of RSV live attenuated vaccine candidates.

Similarly, m6A reader protein (YTHDF1) enhanced RSV protein synthesis in Vero cells. Thus, we observed a pro-viral function for m6A in all three cell lines. It should be noted that overexpression of YTHDF1 protein (a representative of reader proteins) in A549 cells did not significantly affect the growth or survival of the cells.

152

Figure 4.4. YTHDF1, 2, 3 (reader) proteins promote RSV replication, gene expression, and progeny virus production.

153

(A) Detection of YTHDF1, 2 3 in HeLa cells stably overexpressing YTHDF1-3. Western blot

confirmed the overexpression of YTHDF1-3 proteins in HeLa cells using anti-Flag antibody. (B)

YTHDF1, 2, 3 enhance GFP expression in rgRSV-infected cells. HeLa cells stably overexpressing these YTHDF proteins were infected with rgRSV at an MOI of 0.1, and GFP expression was

monitored at the indicated times by fluorescence microscopy. Micrographs with 10× magnification

(scale bar of 100 μm) are shown. (C) YTHDF1, 2, 3 increase the number of GFP-positive cells

quantified by flow cytometry. (D) YTHDF1, 2, 3 enhance RSV protein expression. Total cell

extracts were harvested from rgRSV-infected HeLa cells at the indicated times and subjected to

Western blot using antibody against RSV N, F, or G protein. Western blots shown are the

representatives of three independent experiments. RSV F (F0 + F1) (E), G (F), and N (G) proteins

were quantified by Image J Software. Data are expressed as mean of three independent

experiments ± standard deviation. (H) YTHDF1, 2, 3 increases RSV progeny virus production.

The release of infectious RSV particles was monitored by a single-step growth curve. Virus titer

was measured by TCID50. (I) YTHDF1, 2, 3 enhances RSV genomic RNA replication. Total RNA

was purified from rgRSV-infected cells using TRizol, and genomic RNA was quantified by real-

time RT-PCR using specific primers annealing to the RSV leader sequence and GFP gene. (J)

YTHDF1, 2, 3 enhance mRNA transcription. Viral mRNA was separated from total RNA using

the Dynabeads mRNA isolation kit and quantified by real-time PCR using primers annealing to

the NS1 gene. (K) Ratio between mRNA and genomic RNA. The ratio between NS1 mRNA and

genomic RNA was calculated for each cell line. All results are from three independent

experiments. Flow cytometry data are expressed as mean ± standard deviation. RNA copy and

viral titer are the geometric mean titer (GMT) of three independent experiments ± standard

154

deviation. Data were analyzed using Student’s t-test and *P < 0.05; **P < 0.01; ***P < 0.001;

****P < 0.0001.

We also measured the RSV genomic RNA (the replication product) and mRNAs (the transcription

product) in HeLa cells by real-time RT-PCR. Overexpression of YTHDF1-3 significantly increased both RSV genomic RNA (Figure. 4.4I) and mRNA synthesis (Figure. 4.4J).

Overexpression of YTHDF1 and 3 did not alter the balance between the synthesis of genomic

RNA and mRNA whereas overexpression of YTHDF2 led to a more dramatic increase in replication than transcription (Figure. 4.4K). It appears that overexpression of YTHDF1-3

enhanced the ability of the RSV polymerase to both replicate and transcribe.

155

Figure 4.5. Knockdown of endogenous YTHDF1, 2, 3 (reader) proteins diminishes RSV gene expression.

156

HeLa cells were transfected with 150 pmole of siRNA targeting YTHDF1, 2 3 or control siRNA.

At 36 h post-transfection, cells were infected with rgRSV at an MOI of 0.5. (A) Immunoblot

analysis of YTHDF1, 2, 3 in HeLa cells transfected with siRNA. (B) Immunoblot analysis of RSV

G and F proteins. (C) Dynamics of GFP expression in YTHDF1, 2, 3 protein-depleted HeLa cells.

Micrographs with 10× magnification (scale bar of 100 μm) are shown. (D) Quantification of GFP-

positive cells by flow cytometry at 18 h post-inoculation. Fold of GFP signal compared to the

control is shown. Western blots and GFP images shown are the representatives of three

independent experiments. Flow cytometry data are expressed as mean ± standard deviation. Data were analyzed using Student’s t-test and *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

As a complementary approach, we also tested RSV replication and gene expression in HeLa cells

when m6A reader proteins were knocked down by siRNA. Counting live cells by flow cytometry

showed that siRNA targeting YTHDF1 did not significantly alter cell. Knockdown of individual,

endogenous YTHDF1-3 proteins (Figure. 4.5A) did significantly reduced viral F and G protein

synthesis (Figure. 4.5B) and GFP expression (Figure. 4.5C and D) relative to the control siRNA

transfected cells. Collectively, these results demonstrate that m6A binding proteins promote RSV

genome replication, mRNA transcription, and as a result, viral protein expression, and progeny

virus production.

4.4.4 m6A writer proteins positively regulate RSV replication

The internal m6A addition is catalyzed by host methyltransferases termed m6A writer proteins

(METTL3 and METTL14)[222]. We next examined the role of the m6A writer proteins in RSV

replication and protein expression. More F and G protein synthesis (Figure. 4.6A) and GFP

157

expression (Figure. 4.6B and C) were observed when METTL3 and METTL14 were overexpressed in HeLa cells. In contrast, less F and G proteins were synthesized (Figure. 4.6D)

and less GFP was expressed (Figure. 4.6E and F) when endogenous METTL3, METTL14, or both,

were knocked down. siRNA targeting METTL3 did not significantly alter cell survival. These

results suggest that modification of RSV RNA by m6A writers facilitates RSV replication and gene

expression.

Next, we determined whether alteration of m6A writer proteins can alter the level of m6A level in

viral RNA. As shown in Figure.4.6G and H, the m6A contents in viral RNA of virus from Mettl3

and Mettl14-overexpressed cells were significantly higher (approximately 26% increase) than viral

RNA of particles from vector-transfected cells. Knockdown of Mettl3 and Mettl14 led to 70% of

reduction in m6A content in viral RNA (Figure.4.6 I and J). These experiments suggest that

manipulation of m6A writer proteins lead to enhanced or reduced levels of m6A in viral RNAs,

which may in turn, affect viral replication and gene expression.

158

Figure 4.6. Effects of m6A writer proteins on RSV gene expression.

159

(A) Overexpression of m6A writer proteins enhances RSV gene expression. HeLa cells were

transfected with plasmids encoding METTLE3 and/or METTL14, followed by rgRSV infection at

an MOI of 0.5. At 18 h post-infection, cell lysates were harvested for Western blot analysis. (B)

Overexpression of m6A writer proteins enhances RSV expression of GFP. The GFP expression

was monitored by fluorescence microscopy. Representative micrographs with 10× magnification

(scale bar of 100 μm) at 18 h post-infection were shown. (C) Quantification of GFP-positive cells

by flow cytometry at 18 h post-infection. Fold of GFP signal compared to the control is shown.

(D) Knockdown of m6A writer proteins diminishes RSV gene expression. HeLa cells were transfected with siRNA targeting METTL3 and/or METTL14, followed by rgRSV infection at an

MOI of 0.5. At 18 h post-infection, cell lysates were harvested for Western blot analysis. (E)

Knockdown of m6A writer proteins diminishes GFP expression. Micrographs with 10×

magnification (scale bar of 100 μm) are shown. (F) Quantification of GFP-positive cells by flow cytometry. Fold of GFP signal compared to the control is shown. Flow cytometry data are

expressed as mean ± standard deviation. (G) Overexpression of m6A writer proteins enhances m6A

level in viral RNA. Confluent A549 cells were transfected with plasmids encoding Mettl3 and

Mettl14 or vector pCAGGS, followed by rgRSV infection. RSV particles were harvested and

purified. The m6A content in viral RNA was determined as described in Materials and Methods.

(H) Fold increase in m6A content in viral RNA compared to controls. (I) Knockdown of m6A writer

proteins reduces m6A level in viral RNA. A549 cells were transfected with siRNA targeting Mettl3

and Mettl14 or control siRNA, followed by rgRSV infection. RSV particles were harvested and

purified. The m6A content in viral RNA was determined. (J) Fold reduction in m6A content in viral

160

RNA compared to controls. Data were analyzed using Student’s t-test and *P < 0.05; **P < 0.01;

***P < 0.001; ****P < 0.0001.

4.4.5 m6A eraser proteins downregulated RSV replication

Internal m6A modifications are reversible and can be removed by m6A eraser proteins [122, 123].

We thus examined the effects of overexpression of eraser proteins AlkBH5 or FTO, or both

(Figure. 4.7). Overexpression of eraser proteins dramatically reduced RSV F and G protein

expression by 80- and 20-fold, respectively (Figure. 4.7A), and number of GFP cells by 1.3-2.7

times (Figure. 4.7B and C). Next, we knocked down AlkBH5 or FTO, or both, in HeLa cells, followed by rgRSV infection. siRNA targeting AlkBH5 did not significantly alter cell survival.

Knockdown of AlkBH5 and FTO enhanced the expression of F protein by 3-fold, G protein by 5-

fold (Figure. 4.7D), and number of GFP cells by 1.2-fold (Figure. 4.7E and F) compared to the

cells transfected with control siRNA. Therefore, overexpression of m6A eraser proteins negatively

regulated RSV replication and gene expression. We also determined whether overexpression and

knockdown of eraser proteins can affect m6A level of viral RNA. As shown in Figure. 4.7G and

H, the m6A contents in viral RNA of virus from AlkBH5-overexpressed cells led to 46% reduction.

Knockdown of AlkBH5 led to 2.12-fold increases in m6A abundance in viral RNA (Figure. 4.7I

and J). Therefore, manipulation of m6A eraser proteins can directly affect m6A levels in viral

RNAs, which may affect viral replication and gene expression.

161

Figure 4.7. Effects of m6A eraser proteins on RSV gene expression.

162

(A) Overexpression of m6A eraser proteins diminishes RSV gene expression. HeLa cells were

transfected with plasmids encoding ALKHB5 and/or FTO, followed by rgRSV infection at an MOI

of 0.5. At 18 h post-infection, cell lysates were harvested for Western blot analysis. (B)

Overexpression of m6A eraser proteins reduces GFP expression. The GFP expression was

monitored by fluorescence microscopy. Representative micrographs with 10× magnification (scale

bar of 100 μm) at 18 h post-infection were shown. (C) Quantification of GFP-positive cells by flow cytometry at 18 h post-infection. (D) Knockdown of m6A eraser proteins enhances RSV gene

expression. HeLa cells were transfected with siRNA targeting ALKHB5 and/or FTO, followed by

rgRSV infection at an MOI of 0.5. At 18 h post-infection, cell lysates were harvested for Western blot analysis. (E) Knockdown of m6A eraser proteins enhances GFP expression. Micrographs with

10× magnification (scale bar of 100 μm) are shown. (F) Quantification of GFP-positive cells by

flow cytometry. Fold of GFP signal compared to the control is shown. Western blots and GFP

images shown are representatives of three independent experiments. Flow cytometry data are

expressed as mean ± standard deviation. (G) Overexpression of m6A eraser protein reduces m6A

level in viral RNA. A549 cells were transfected with plasmid encoding ALKBH5 or pCAGGS,

followed by rgRSV infection. RSV particles were purified from supernatants. Virion RNA was

subjected to m6A-IP. (H) Fold reduction in m6A content in viral RNA compared to controls. (I)

Knockdown of m6A eraser protein increases m6A level in viral RNA. A549 cells were transfected

with siRNA targeting ALKBH5 or control siRNA, followed by rgRSV infection. RSV particles

were purified. Virion RNA was subjected to m6A-IP. Results are from three or four independent experiments. Data were analyzed using Student’s t-test and *P < 0.05; **P < 0.01; ***P < 0.001;

****P < 0.0001.

163

4.4.6 RSV infection does not alter the translocation of m6A proteins

To directly visualize the locations of the m6A reader, writer, and eraser proteins, mock and rgRSV infected HeLa cells were stained with antibodies specific to each m6A-related protein and analyzed by confocal microscopy. As shown in Figure. 4.8A, m6A reader proteins (YTHDF1-3) were

distributed in the cytoplasm in both mock and RSV-infected cells. In addition, viral N protein was

partially co-localized with reader protein (YTHDF2) (Figure. 4.8A). In contrast, the majority of m6A writer proteins (METTL3 and METTL14) and eraser protein (AlkBH5) were distributed in

the nucleus although a small fraction of these proteins was also found in the cytoplasm (Figure.

4.8B). Also, viral N protein was partially co-localized with writer protein (METTL3) (Figure.

4.8B). Another eraser protein, FTO, was exclusively located in the nucleus (Figure. 4.8C). Equal

amounts of m6A related proteins were detected in the cytoplasmic and nuclear fractions by Western

blot (Figure. 4.8D and E). Therefore, RSV infection does not significantly alter the distribution

pattern of m6A-related proteins in HeLa cells.

164

Figure 4.8. RSV infection does not alter the m6A reader, writer, or eraser protein distribution in cells.

165

HeLa cells were infected by rgRSV at an MOI of 10.0. At 24 h post-infection, mock- or rgRSV- infected cells were stained with anti-reader, writer, or eraser protein antibody (green) and anti-

RSV N protein antibody (red), and were analyzed by confocal microscope. Nuclei were labeled with DAPI (blue). Micrographs with 60× magnification (scale bar of 20 μm) are shown. (A) m6A reader protein YTHDF2; (B) m6A writer protein METTL3; and (C) m6A eraser protein FTO. (D)

Detection of m6A reader, writer, and eraser proteins by Western blot. Nuclear and cytoplasmic fractions were separated from mock- or rgRSV-infected HeLa cells, and were subjected to Western blot. Nuclear and cytoplasmic markers were indicated by Lamin A and α-Tubulin, respectively.

Representative results from three independent experiments are shown.

4.4.7 m6A reader proteins bind to both RSV genomic RNA and mRNA

Since the biological function of m6A is mediated by m6A binding proteins, we next determined whether YTHDF2 can directly bind to RSV RNAs in virus-infected cells. As expected, YTHDF2 was pulled down by YTHDF2-specfic antibody (Figure. 4.9A). Both RSV genomic RNA and N mRNA were efficiently precipitated as complexes, with YTHDF2 (Figure. 4.9B). We further confirmed this result by pulling down HA-tagged YTHDF2 from total cell lysates of HeLa cells overexpressing YTHDF2 with HA antibody (Figure. 4.9C). Similarly, significant amounts of RSV genomic RNA and N mRNA bound to YTHDF2 (Figure. 4.9D). Since G mRNA has the strongest m6A enrichment, we also determined the binding of m6A reader proteins to G mRNA. All reader proteins (YTHDF1, 2 and 3) were capable of binding to G mRNA (Figure. 4.9E).

166

Figure 4.9. m6A reader protein binds to RSV genomic RNA and mRNA.

167

HeLa cells stably expressing YTHDF2 and vector control HeLa cells were infected with rgRSV at an MOI of 1.0. At 24 h post-infection, cells were lysed and cytoplasmic extracts were

immunoprecipitated with an antibody against YTHDF2 (A) or an equivalent amount of HA-tag

(non-specific IgG control) (C). The amount of vgRNA and mRNA captured by the YTHDF2

antibody (B) or the HA-tag antibody (D) was quantified by real-time RT-PCR, as was the input

RNA, and graphed as the percentage of input. Data are representative of two experiments. (E) m6A

reader proteins bind to G mRNA. HeLa cells stably expressing YTHDF1, 2, and 3 and vector

control HeLa cells were infected with rgRSV at an MOI of 1.0. At 24 h post-infection, cells were lysed and cytoplasmic extracts were immunoprecipitated with an antibody against HA-tag

antibody. The amount of G mRNA bound to YTHDF1, 2, and 3 was quantified by real-time RT-

PCR. Data are representative of two or three experiments. The P value (Student’s t-test) for

YTHDF2 is **P = 0.001223238.

4.4.8 Abrogation of m6A sites results in attenuation of RSV

Based on m6A-seq, G gene and G mRNA have the most abundant m6A enrichment in both HeLa

and A549 cells (Figure. 4.1). Thus, we decided to mutate the m6A sites in the G gene region which

we found to be conserved in the m6A-seq from both HeLa and A549 cells. These mutations were

predicted to remove the m6A sites in the G mRNA and in the G gene region of the antigenome

without changing the amino acid they encoded. Recombinant RSVs carrying these mutations were

recovered by reverse genetics system. First, the potential m6A sites in regions 1, 2, and 3 were

mutated individually to produce rgRSV-G1, G2, and G3, respectively. Second, mutations of m6A

sites in regions 1 and 2 were combined to produce rgRSV-G12. Third, m6A sites in regions 1, 2,

168

and 3 were combined to produce rgRSV-G123. All m6A-mutated rgRSVs had various degrees of reduction in viral F and G protein synthesis in A549 cells compared to the parental rgRSV (Figure.

4.10A). Significantly less GFP expression was observed in m6A-mutated rgRSVs compared to

rgRSV at 48 h post-infection (Figure. 4.10B and C). Single-step growth curves showed that m6A-

mutated rgRSVs had delayed replication kinetics and had 0.5-1.5 log reductions in peak titer compared to rgRSV (Figure. 4.10D). Overall, mutants rgRSV-G1, G3, G12, and G123 had a moderate defect whereas rgRSV-G2 had a mild defect in replication. Real-time RT-PCR results showed that both rgRSV-G1 and G12 had defects in genome (Figure. 4.10E), NS1 (Figure. 4.10F), and G (Figure. 4.10 G) mRNA synthesis compared to rgRSV, and rgRSV-G1 had more defects than rgRSV-G12. Next, we calculated the percentage of reduction for NS1 and G mRNA. It was found that NS1 mRNA had significantly less reduction than G mRNA, suggesting that removal of the m6A from the G mRNA may accelerate its decay.

169

Figure 4.10. m6A-abrogating RSV mutants have defects in replication in immortalized cells.

170

(A) Immunoblot analysis of RSV proteins. Confluent A549 cells were infected with each rgRSV

at an MOI of 0.1, cell lysates were harvested at 18, 24, and 48 h post-infection, and RSV proteins were detected by specific antibodies against F and G protein. (B) GFP expression of m6A-deficient rgRSV mutants. Micrographs with 10× magnification (scale bar of 100 μm) are shown. (C)

Quantification of GFP-positive cells by flow cytometry. (D) Single step growth curve of m6A- deficient rgRSV mutants in A549 cells. Confluent A549 cells were infected with each rgRSV at an MOI of 1.0, supernatants were harvested, and viral titer was determined by TCID50 assay. (E)

RSV genomic RNA replication. At 18, 24, and 48 h post-infection, total RNA was purified from

rgRSV-infected cells using TRizol, and genomic RNA was quantified by real-time RT-PCR using

specific primers annealing to the RSV leader sequence and GFP gene. (F) RSV NS1 mRNA

transcription. Viral mRNA was separated from total RNA and quantified by real-time PCR using

primers annealing to the NS1. (G) RSV G mRNA transcription. Results are from three independent

experiments. Flow cytometry data are expressed as mean ± standard deviation. RNA copy and

viral titer are the geometric mean titer (GMT) of three independent experiments ± standard deviation. Western blots shown are the representatives of three independent experiments. Data were analyzed using Student’s t-test and *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

4.4.9 m6A-mutated rgRSVs are defective in replication in HBE

We next tested the replication and spread of m6A-deficient rgRSVs in HBE cultures, a near in vivo

model for lower airway infection. These cultures are pseudostratified and polarized, closely

resembling the in vivo airway epithelium morphology and function, including mucus production

and ciliary motion[164]. Briefly, HBE cultures were infected with 800 TCID50 of each recombinant

171

virus, and viral release and spread was monitored. As in A549 cells, m6A-mutated rgRSVs had a

delay in viral gene expression (GFP production) and spread (Figure. 4.11A and B). At day 4 post-

inoculation, m6A-mutated rgRSVs had fewer green cells compared to rgRSV. Although several

rgRSV mutants gradually increased at days 6 and 8, the density of green cells remained less than

for rgRSV. rgRSV-G1 was delayed in spreading but eventually spread to most susceptible cells at

day 8. rgRSV-G2 had a delay at the early time point (day 4) but had wild type level of spreading

at later times. Recombinant rgRSV-G12 was the most defective virus in HBE cultures, displaying

a weak GFP signal during the entire experimental period. In addition, m6A-mutated rgRSVs had delays in virus release in HBE culture with 1-2 log defects in virus yield (Figure. 4.11C). These results demonstrate that m6A-mutated rgRSVs were defective in replication and spread in this near

in vivo lung infection model.

172

Figure 4.11. m6A-abrogating RSV mutants have defects in replication in HBE culture.

173

6 (A) Spreading of m A mutated rgRSVs in HBE culture. HBE cultures were infected by 800 TCID50

of each rgRSV. At the indicated time, virus spreading was monitored by fluorescence microscopy.

Representative micrographs with 4× magnification (scale bar of 200 μm) at each time point are

shown. (B) Quantification of GFP signal in HBE culture. GFP signal was quantified by Image J

software, and data are expressed as mean ± standard deviation. (C) Virus release from m6A

mutated rgRSV-infected HBE culture. HBE cultures were infected by 800 TCID50 of each rgRSV.

After virus inoculation, supernatants were collected every 2 days until day 14 post-inoculation.

Infectious virus in supernatants was determined by TCID50 assay. Viral titers are the geometric

mean titer (GMT) of three independent experiments ± standard deviation. Data were analyzed

using Student’s t-test and *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

4.4.10 m6A-mutated rgRSVs are defective in replication in vivo

We tested replication and pathogenesis of four m6A-mutated rgRSV mutants, rgRSV-G1, G2, G3,

and G12, in cotton rats. The parental rgRSV replicated efficiently in the lungs (Figure. 4.12A) and

nasal turbinates (Figure. 4.12B) with average viral titers of 4.70 ± 0.10 log10 TCID50/g and 4.10 ±

0.10 log10 TCID50/g, respectively. Mutant rgRSV-G1 had a 7-fold reduction in replication in nasal

turbinate and lung titers, respectively (P<0.05). Mutant rgRSV-G2 had no significant reduction in

replication in lung (P>0.05) but 3-fold reductions in nasal turbinate (P<0.05). Mutant rgRSV-G12

had the most dramatic defect in replication, with reductions of 100- and 200-fold in viral titer in

nasal turbinate and lung, respectively. It should be noted that 4 out of 5 cotton rats had below

detection limit level of RSV replication in the nasal turbinate and 3 out of 5 cotton rats had below

detection limit level of RSV replication in lungs, suggesting that rgRSV-G12 is highly attenuated

174

in vivo. The rgRSV-G3 had 5-fold reductions in replication in nasal turbinate and lung (P<0.05).

Histologic examination showed that rgRSV caused moderate pulmonary histopathological

changes, including interstitial pneumonia and peribronchial lymphoplasmocytic infiltrates (Figure.

4.12 C). In contrast, m6A-deficient rgRSV mutants only had mild and less pulmonary

histopathological changes compared to rgRSV (Figure. 4.12C). These results showed that m6A-

deficient rgRSV mutants had significant reductions in viral replication in both the upper and lower

respiratory tracts in cotton rats and were less pathogenic compared to rgRSV. These results indicate that viral m6A upregulates viral replication and pathogenesis in vivo.

4.4.11 m6A-mutated rgRSVs are highly immunogenic in cotton rats

To determine whether defects in viral m6A methylation impair the immunogenicity of the virus, we evaluated the protection efficacy of a partially attenuated (rgRSV-G1) and highly attenuated

(rgRSV-G12) virus in cotton rats. Cotton rats immunized with parental rgRSV or m6A-mutated

rgRSVs did not have any detectable infectious virus in either the nasal turbinate or lung tissue after

challenge with rgRSV (Figure. 4.12D). In contrast, unvaccinated challenged controls had average

titers of 5.12 ± 0.28 and 4.27 ± 0.07 log10 PFU/g in the lung and nasal turbinate, respectively

(Figure. 4.12D). These results demonstrate that immunization with the rgRSV-G1 and G12

provided complete protection from challenge with rgRSV. Lung histology showed that

unvaccinated challenged controls had moderate histologic lesions. However, the vaccinated

challenged groups had only mild lesions in lungs. In addition, no enhanced lung damage was

observed. The most attenuated m6A-deficient virus (rgRSV-G12) triggered similar levels of

neutralizing antibody compared to rgRSV (P > 0.05) (Figure. 4.12E). The rgRSV-G1 had a lower

antibody titer at week 3 and 4 compared to rgRSV (Figure. 4.12E). These results demonstrate that

175 m6A-mutated rgRSV retained high immunogenicity and provided complete protection against

RSV infection in cotton rats.

Figure 4.12. Pathogenicity and immunogenicity of m6A-mutated rgRSVs in cotton rats.

176

(A) RSV titer in lungs. Four-week-old SPF cotton rats were inoculated intranasally with 2.0×105

TCID50 of each rgRSV. At day 4 post-infection, the cotton rats were sacrificed, and lungs and nasal

turbinates were collected for virus titration by TCID50 assay. Viral titers are the geometric mean

titer (GMT) of 5 animals ± standard deviation. Detection limit is 2.0 log TCID50/g tissue. (B) RSV

titer in nasal turbinates. (C) m6A mutated rgRSVs had less lung histopathological changes

compared to rgRSV. Representative pathological changes from each group are shown. Right lung

lobe of each cotton rat was fixed in 4% neutral buffered formaldehyde, embedded in paraffin,

sectioned at 5 µm, and stained with hematoxylin-eosin (HE) for the examination of histological

changes by light microscopy. Micrographs with 20× magnification (scale bar of 500 μm) are

shown. (D) m6A mutated rgRSV provides complete protection against RSV challenge. Four-week-

5 old SPF cotton rats were inoculated intranasally with 2.0×10 TCID50 of each rgRSV. At week 4

5 post-immunization, cotton rats were challenged with 2.0×10 TCID50 rgRSV. At day 4 post- challenge, the cotton rats were sacrificed, and lungs and nasal turbinates were collected for virus

titration by TCID50 assay. Viral titers are the geometric mean titer (GMT) of 5 animals ± standard deviation. The detection limit is 2.0 log TCID50/g tissue. (E) rgRSV induced a high level of

neutralizing antibody. Blood samples were collected from each rat weekly by retro-orbital

bleeding. The RSV-neutralizing antibody titer was determined using a plaque reduction

neutralization assay, as described in Materials and Methods. Data were analyzed using Student’s

t-test and *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

177

4.4.12 m6A-mutated rgRSVs are less dependent on host m6A machinery

If the attenuated phenotype of m6A-mutated rgRSVs is indeed m6A-dependent, alteration of host m6A machinery would have no or less of an impact on replication and gene expression. To address

this question, we tested the replication of rgRSV-G1 and G12 in A549 cells overexpressing

AlkBH5 which is an m6A eraser protein. Overexpression of AlkBH5 led to 70% and 42% reduction in RSV G and F protein synthesis in rgRSV-infected cells compared to vector control cells (Figure.

4.13A). We also observed a reduction in replication and protein expression of the m6A-mutated rgRSVs in AlkBH5 overexpressing cells, but the level of reduction was much less compared to the parental rgRSV. For example, only 17% and 10% reduction in RSV G and F protein synthesis was observed for rgRSV-G12, and 50% and 20% reduction in G and F protein was observed for rgRSV-

G1, respectively. We also tested the replication of rgRSV-G123 in m6A writer protein-depleted

A549 cells. Knockdown of MELL3 and METTL14 led to 32 and 22% reduction in RSV G and F

protein in rgRSV-infected A549 cells whereas only 25% and 8% reduction in G and F in rgRSV-

G123-infected A549 cells (Figure. 4.13B). Thus, these results showed that replication and gene

expression of m6A-mutated rgRSVs were less dependent on host m6A machinery, suggesting that

the attenuated phenotype of these mutants is likely due to the deficiency in m6A methylation of

the viral RNA.

4.4.13 m6A-mutated rgRSVs are deficient in m6A enrichment

To determine whether m6A sites are indeed missing from the G gene, A549 cells were infected by each m6A-mutated rgRSV, and polyadenylated mRNAs were isolated and subjected to m6A-seq.

As shown in Figure. 4.13C, the enrichment of m6A in the G mRNA of each m6A-mutated rgRSV significantly decreased compared to the G mRNA from the parental rgRSV, confirming that m6A

178

methylation in the G mRNA has indeed been significantly reduced. Our mutagenesis targeted the

positive-sense RNA which includes both antigenome and G mRNA. To determine whether antigenome is also defective in m6A methylation, virion RNA was extracted from highly purified virions of rgRSV, rgRSV-G12, rgRSV-G123, and rgRSV-ΔG (rgRSV with deletion of entire G gene), and m6A content in total RNA was measured. As shown in Figure.4.13D, m6A content in

virion RNA of rgRSV-G12, rgRSV-G123, and rgRSV-ΔG was significantly less compared to

those in rgRSV (P<0.05). In addition, there was no significant difference in m6A content between

rgRSV-G123, and rgRSV-ΔG (P>0.05). Collectively, these results demonstrate that m6A-mutated rgRSVs indeed have defects in m6A methylation.

Figure 4.13. The attenuated phenotype of m6A mutated rgRSVs is m6A-related. 179

(A) rgRSV-G1 and -G12 were less dependent on the m6A eraser protein. A549 cells were

transfected with a plasmid encoding ALKHB5. At 36 h post-transfection, cells were infected with

each rgRSV at an MOI of 0.5. At 18 h post-infection, cell lysates were harvested for Western blot analysis. (B) rgRSV-G123 expression was less dependent on m6A writer protein. A549 cells were

transfected with control siRNA or siRNA targeting METTL4 and METTL13. At 36 h post- transfection, cells were infected with each rgRSV at an MOI of 0.5. At 18 h post-infection, cell lysates were harvested for Western blot analysis. The density of Western blot was quantified by

Image J software, and the ratio of the protein bands was calculated. (C) Distribution of m6A peaks

on the RSV mRNAs from A549 cells infected by rgRSV and rgRSV-G123. Confluent A549 cells

were infected by each m6A-mutated rgRSV at an MOI of 1.0, cell lysates were harvested at 36 h post-infection. Total RNAs were extracted from cell lysates, and were enriched for mRNA by

binding to oligo dT, and subjected to m6A-seq. The distribution of m6A-immunoprecipitated (IP) reads were mapped to the RSV mRNAs (pink block). The baseline distributions for mRNAs from input sample are shown as a pink line. Data presented are the mean coverage from two independent virus-infected A549 cell samples (n = 2). Red arrow indicates the m6A enrichment in G mRNA.

(D) Virion RNA of m6A-mutated rgRSVs is defective in binding to anti-m6A antibody. Virion

RNA was extracted from highly purified RSV virions. Antigenome was quantified by real-time

RT-PCR. Each amount of antigenome was bound to strip wells using a RNA high binding solution,

and m6A was detected using a specific capture anti-m6A antibody and then quantified

colorimetrically by reading the absorbance in a microplate spectrophotometer at a wavelength of

450 nm. A standard curve was generated using known m6A methylated RNA (range from 0.02 to

180

1 ng of m6A) as a positive control. The m6A content was calculated from each RNA samples. Data were analyzed using Student’s t-test and *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

4.5 Discussion

The biological function of m6A methylation in viral RNAs has remained uncertain since its discovery 40 years ago. Recent studies have shown that viral m6A methylation can have pro-viral or anti-viral functions dependent on the specific virus species and cell type. Here, we showed that m6A modification positively regulates each step in the RSV replication cycle ranging from genome replication, mRNA transcription and viral protein synthesis, to progeny infectious particle production. We demonstrated that m6A regulates RSV replication and pathogenesis in an animal model. Furthermore, we provide evidence that m6A could be a target for the development of live attenuated vaccine candidates.

The m6A methylation of RNAs is modulated by writers, erasers, and readers in host cells. It should be noted that m6A methylation and its reader proteins may play distinct roles in a virus life cycle.

In this study, we showed that overexpression of both m6A reader and writer proteins positively regulated RSV replication while knockdown inhibited RSV gene expression and replication. The opposite was true for eraser proteins: overexpression decreased RSV gene expression and replication whereas knockdown increased them. In case of influenza virus, m6A methylation enhanced viral replication as did the reader, YTHDF2, but not YTHDF1 or 3 [145]. We found that overexpression of m6A writer proteins led to an increased level of m6A in viral RNA. The impact of m6A-related proteins on RSV replication may result from three aspects: a change in m6A levels in viral RNA which directly affects viral replication, a change host gene expression which

181 indirectly affect viral replication, or both. However, m6A writer and m6A reader proteins have been found to negatively regulate HCV production [149], opposite to RSV and influenza virus. In addition, the role of m6A reader proteins in the HIV life cycle is controversial [141, 142, 212].

Thus, m6A readers have distinct effects on the life cycles of different viruses, as they are multifunctional and play many important biological roles ranging from RNA stability, decay, and transport, to protein translation.

Our m6A-seq also found that the viral G mRNA has the most abundant m6A peaks among the 10

RSV mRNAs in both HeLa and A549 cells. In addition, the strongest m6A peaks in both the genome and the antigenome are located in the G gene region. Another interesting finding was that the positions of the m6A modifications in the genome and antigenome largely overlapped despite the fact that the sequence of the antigenome is complementary to the genome. However, it is unclear why antigenome and genome are m6A methylated because both of them are completely encapsidated by viral N protein. It was shown that RSV RNA synthesis occurs in cytoplasmic inclusion bodies [51]. In fact, m6A-related proteins are partially co-localized viral N protein

(Figure. 4.6A and B), suggesting that host m6A machinery may interact with RSV ribonucleoprotein (RNP) complex. It is possible that m6A methylation occurs prior to N encapsidation, or m6A methylation and encapsidation can occur concurrently.

Since G mRNA has the strongest m6A enrichment, we searched the three peaks in the G sequence for m6A motifs, identifying a total of 18 putative m6A sites. It is known that the G gene is the most genetically diverse RSV gene. However, bioinformatics analysis of 100 RSV strains found that those 18 m6A sites are highly conserved in the G gene, suggesting that m6A sites in the G gene may provide an evolutionary advantage for virus infection, replication, and spreading. Mutations

182 in these three m6A peaks in the G mRNA showed that peaks 1 and 3 play a major role in regulating

RSV replication whereas peak 2 plays a minor role, as recombinant rgRSV mutants in peak 1 and

3 (rgRSV-G1 and G3) had greater deficits in replication compared to mutants in peak 2 (rgRSV-

G2).

The G protein is primarily responsible for the attachment of RSV to host cells and plays a role in modulating innate immune responses [164, 223]. Although it is not essential for the production of infectious RSV, RSV G is necessary for full infectivity [224, 225]. The G protein also plays an important role in the assembly of filamentous virions[226]. It is likely that the abundant m6A modifications of the G mRNA enhances its stability, enabling more translation, which may leads to insertion of more G protein into virions and enhanced viral assembly and production of infectious virions. However, a portion of the G protein produced in a cell is released in a soluble form that affects leukocyte migration [227]. Enhanced G protein expression could enhance the production of soluble G protein, thereby affecting the immune response to RSV. It is also possible that m6A modification of viral RNAs facilitate the virus to escape the surveillance of host innate immunity to allow for efficient gene expression and virus replication. In this study, we also found that rgRSVs carrying mutations in the m6A sites in G not only decreased G protein expression but also reduced expression of other viral proteins (such as N and F proteins). Since RSV replication and transcription require ongoing protein synthesis, a reduction in viral G protein expression will affect the second round of viral infection, which results in a reduction of overall replication and transcription, and expression of all viral proteins.

Importantly, we found that viral m6A also modulates viral replication and pathogenesis in vivo.

However, the degree of attenuation in cell culture did not always match that in vivo. For example,

183

rgRSV-G1 and G12 had similar levels of attenuation in immortalized cells (A549 and HeLa cells).

In cotton rats, rgRSV-G1 replication was 7-fold reduced in the lung and nasal turbinate, respectively, whereas rgRSV-G12 had more than 100-fold reductions. Recombinant rgRSV-G2

was only mildly attenuated in cell culture. This recombinant had similar level of replication in

lungs (P>0.05), and only had 3-fold reduction in nasal turbinates (P<0.05). Therefore, it appears

that m6A sites in peaks 1 and 2 contributed synergistically to the highly attenuated phenotype of

rgRSV-G12 in vivo. The phenotype of these mutants in HBE culture seems to correlate better with the phenotype in cotton rats than in HeLa and A549 cells. Thus, HBE culture may be better system

to predict virus replication in vivo. Also, m6A-mutated rgRSVs had significantly less

histopathology compared to the parental rgRSV. These results demonstrate that m6A not only

modulates the virus life cycle in vitro but also regulates viral replication and pathogenesis in vivo.

Recent study in vesicular stomatitis virus (VSV) found that codon usage also affected genome

replication and mRNA transcription in a minigenome assay[228]. Specifically, purine/pyrimidine

content in RNA template modulates the stability of the polymerase complex, which results in

alteration of the activity of viral RNA synthesis[228]. In this study, we designed mutations in

predicted m6A sites to avoid as much as possible alterations to the predicted mRNA secondary structure and to avoid changes in the efficiency of gene expression of the new codon relative to the original codon. We also confirmed the loss of m6A in the predicted region of the G mRNA by

m6A sequencing and tested the functional consequences of reducing the m6A modifications.

Functional loss of m6A modifications was examined by comparing replication of the mutant

rgRSV in A549 cells overexpressing or depleted of m6A-related proteins. The m6A-mutated rgRSVs (G1, G12, and G123) were much less dependent on host m6A enzyme compared to the

184

parental rgRSV, confirming that the attenuated phenotype of m6A-mutated rgRSVs is due to the reduction of m6A sites in G mRNA. Removal of m6A sites in the mRNA also removes them from

the antigenome, but not from other viral mRNAs, other locations in the antigenome, or sites in the

genome. Therefore, rgRSVs lacking particular m6A peaks in the G gene would be partially but not

fully independent of host m6A enzymes.

A potentially important application of this study is in the rational design of live attenuated RSV

vaccine candidates by inhibiting m6A addition to the mRNA and antigenome, or perhaps the viral

genome. Currently, there is no FDA-approved vaccine for RSV despite the fact that it was first

isolated in 1953. A live attenuated vaccine, similar to the effective vaccines for the related measles

and viruses, would seem to be one of the most promising methods for protection from RSV

disease. However, it has been a challenge to strike the right balance between attenuation and

immunogenicity [166].

Although mutations in individual m6A peaks in the G mRNA were not sufficient to achieve

complete attenuation of RSV replication in vivo, the combination of m6A mutations in peaks 1 and

2 resulted in a recombinant virus that was sufficiently attenuated both in vitro and in vivo.

Importantly, cotton rats vaccinated with rgRSV-G12 had similar neutralizing antibody response

levels compared to parental rgRSV and were completely protected from rgRSV challenge. In

addition, no enhanced lung damage was observed. Thus, rgRSV-G12 may be a good live attenuated vaccine candidate for RSV. This study demonstrates that inhibition of m6A methylation

may be a novel method for rationally designing live attenuated vaccines. These m6A-mutated rgRSVs would also provide invaluable tools to understand the roles of m6A modification in the

innate immune response. In fact, it has been shown that internal m6A modification of in vitro

185 synthesized RNAs prevents recognition of the RNA by the host pattern recognition receptors TLR3 and RIG-I [229].

This study also provides a novel approach for enhancing viral titers in cell culture, an important consideration in the production of live attenuated vaccines. Attenuated viruses typically grow to lower titers than wild-type virus. In the case of RSV, a relatively large dose of vaccine candidate is required to induce a protective immune response in humans, making vaccine production expensive. One strategy might be to produce live attenuated vaccines in cells overexpressing one or more m6A reader or writer proteins, since overexpression of these host m6A machinery components enhance virus yield at least 10-fold. Such a boost in the production of a vaccine should greatly enhance its economic feasibility. Unfortunately, this approach would not be useful for m6A mutant attenuated viruses, such as the ones described in this report. It should be useful for viruses attenuated by any other method.

In summary, we mapped the internal m6A modifications in RSV RNAs and showed that m6A enhances RSV replication, gene expression, and virus production. In addition, we provide evidence that m6A upregulates RSV pathogenesis and virulence in vivo. These findings highlight viral m6A machinery as a possible novel target for rational design of live attenuated vaccines, for enhanced production of live attenuated vaccines, and for broad-spectrum antiviral drug discovery.

186

Chapter 5. Enhanced innate and adaptive immune responses of human respiratory

syncytial virus by inhibiting viral N6-methyladenosine modification

5.1 Abstract

Human respiratory syncytial virus (RSV) is the leading cause of respiratory tract infections in

humans. Currently, there is no FDA-approved vaccine for RSV. RSV encodes two nonstructural

proteins which suppress the interferon (IFN)-mediated antiviral response and dampen the subsequent adaptive immune responses. As a consequence, most live attenuated RSV vaccine candidates are less immunogenic than the wild type RSV. Here, we found that inhibition of viral

RNA N6-methyladenosine (m6A) enhances innate immunity and adaptive immunity of RSV. We

found that m6A-deficient RSV virion RNA induced significantly higher type I IFN compared to

m6A-suficient virion RNA. Mechanistically, m6A-deficient RSV virion RNA induces higher

expression of RIG-I, binds more efficiently to RIG-I, enhances RIG-I ubiquitination and IRF3

phosphorylation, leading to an enhanced RIG-I mediated IFN signaling. Notably, this enhanced

IFN signaling is universally conserved in other non-segmented negative-sense (NNS) RNA viruses

such as measles virus and Sendai virus. Subsequently, we inactivated the m6A sites in the G gene

of genome and/or antigenome using synonymous mutations, and the resultant recombinant RSVs

were defective in m6A methylation, attenuated in vitro and in vivo, and induced a higher

neutralizing antibody and T cell immune responses in mice. In addition, animals immunized with

these m6A-deficient RSVs were completely protected from RSV challenge. Our results suggest 187 that inhibition of viral m6A methylation may be a novel approach to enhance innate and adaptive immunity against RSV infection.

5.2 Introduction

One of the major challenges in the development of a live attenuated human respiratory syncytial virus (RSV) vaccine is how to enhance innate immune responses of attenuated RSV strains. A robust innate immune response is likely essential for inducing a strong adaptive immune response.

It is known that humans can be repeatedly infected by RSV in their life, which is probably due to the fact that RSV infection strongly inhibits innate immunity, leading to a blunted adaptive immune responses including neutralizing antibody and memory B- and T-cell immune responses

[230]. In animal models and human clinical trials, several live vaccine candidates lacked sufficient immunogenicity and/or failed to induce long-lasting neutralizing antibody and protection [231]

[195]. Therefore, strategies to enhance the innate immunity of live attenuated RSV vaccine are urgently needed.

Recent studies have shown that the NS1 and NS2 proteins of RSV impaired type I interferon (IFN) production and signaling pathway by interfering the recognition by RIG-I or promoting proteasomal degradation of STAT2 [15, 78, 232]. Thus, many research efforts have been focused on generating recombinant RSVs with deletions or mutations of NS1 and NS2. In fact, RSVs lacking NS1 (∆NS1) and/or NS2 (∆NS2) have been generated and tested in animal models. The

∆NS1 virus found to be highly attenuated and immunogenic in chimpanzees [96]. However, this virus replicates poorly in cell culture, particularly in interferon-deficient Vero cells, the WHO- approved cell lines for vaccine production, preventing the large-scale production of the vaccine

[232]. Compared to the ∆NS1 virus, ∆NS2 virus is less attenuated in chimpanzees thus may need

188 combine with additional attenuating mutations [97]. To date, it is unknown whether alteration of

RSV RNA modification can modulate innate immunity.

We hypothesize that inhibition of m6A methylation in RSV RNA will be a novel and new approach to enhance innate immunity. This hypothesis is based on the rationale that short synthetic RNA lacking posttranscriptional modifications is recognized by host innate immunity as non-self RNAs which may induce a higher type I IFN responses. It has been shown that short RNAs with m5C, m6A, m5U, s2U or Ψ suppressed Toll-like receptor 3 (TLR3), TLR7 and TLR8 signaling [233]. In addition, short RNAs (106 nucleotide poly U/UC sequence) derived from the 3′ untranslated region of hepatitis C virus containing m6A methylation inhibited in the RIG-I signaling pathway [229].

In 2020, our laboratory showed for the first time that human metapneumovirus (hMPV), a member of Pneumoviridae, lacking m6A methylation induced a significantly higher type I IFN and provided evidence that hMPV acquired m6A in its RNA as a means of mimicking cellular RNA to avoid detection by innate immunity. In that study, we generated m6A-deficient hMPVs by two approaches, either growing hMPV in A549 cells overexpressing m6A eraser protein ALKBH5 or mutating the major m6A peaks in viral G gene [234].

The objective of this study is to determine whether viral RNA m6A methylation can modulate both innate immunity and adaptive immunity in RSV. We first generated m6A-deficient RSV by growing RSV in U2OS cells lacking METTL3, the major catalytic subunit of m6A methyltransferase. We confirmed that viral RNA from RSV virions grown in METTL3-knockout

U2OS cells was indeed defective in m6A methylation compared to the viral RNA from RSV virions grown in wild type U2OS cells. We found that m6A-deficient virion RNA induced significantly higher type I IFN than m6A-suficient virion RNA. Mechanistically, m6A-deficient virion RNA

189 induces higher expression of RIG-I, binds more efficiently to RIG-I, and enhances ubiquitination of RIG-I and IRF3 phosphorylation, leading to an enhanced RIG-I mediated IFN signaling.

Notably, this enhanced IFN signaling is universally conserved in the representative members of

Paramyxoviridae (measles virus, MeV; and Sendai virus, SeV), and Pneumoviridae (RSV and hMPV). Subsequently, we inactivated the m6A sites in the G gene of genome and/or antigenome using synonymous mutations, and the resultant recombinant RSVs were defective in m6A methylation, and induced a higher type I IFN and a higher neutralizing antibody and T cell immune responses. In addition, animals immunized with these m6A-deficient RSVs were completely protected from RSV challenge. Our results suggest that viral m6A methylation modulate both innate and adaptive immunity.

5.3 Materials and Methods

5.3.1. Cell culture

HEp-2 (ATCC CCL-23) and A549 cells (ATCC CCL-185) was purchased from the American

Type Culture Collection. A549-Dual, A549-Dual KO-RIG-I, A549-Dual KO-MDA5, and A549-

Dual KO-MAVS-knockout cells were purchased from InvivoGen (San Diego, CA), and were supplemented with Normocin (100 μg ml−1), blasticidin (10 μg ml−1) and zeocin (100 μg ml−1).

The wild type U2OS cells and METTL3-knockout U2OS cell lines were generous gifts from Dr.

Yang Shi (Harvard University, Cambridge, MA). All cell lines were kept in Dulbecco’s modified

Eagle’s medium (DMEM; Life Technologies, Carlsbad, CA) supplemented with 10% FBS and were confirmed free from mycoplasma by the LookOut Mycoplasma PCR Detection Kit (Sigma,

St. Louis, MO).

190

5.3.2. Plasmids, site-directed mutagenesis, and recovery of rgRSV

Plasmid (RW30) encoding the full-length antigenomic cDNA of RSV strain A2 with GFP inserted

between the leader and the NS1 gene, and support plasmids expressing RSV A2 strain N protein

(pTM1-N), P protein (pTM1-P), L protein (pTM1-L), and M2-1 protein (pTM1-M2-1) were

generously provided by Dr. P.L. Collins, (NIAID, Bethesda, MD). Mutations to the potential m6A

sites in the G region of genome (GALL(-)) were designed and introduced into the RW30 plasmids

using the same strategy as described previously. Mutant GALL (+/-) was the combined m6A

mutations from the G gene of genome and the G gene region of antigenome. Recombinant RSVs

carrying GALL(-) and GALL (+/-) were rescued using a reverse genetics system as described

previously. All plasmids and viral mutants were confirmed by DNA sequencing.

5.3.3. Purification of viruses that are naturally defective in m6A

Ten T150 flasks of wild type U2OS cells and METTL3-knockout U2OS cells were infected with

rgRSV, measles virus (MeV), Sendai virus (SeV), vesicular stomatitis virus (VSV), or human

metapneumovirus (hMPV) at an MOI of 0.01. When extensive cytopathic effects (CPE) were

observed, cell culture supernatants were harvested and clarified by centrifugation at 10,000 × g for

30 min. Virus was concentrated through a 35% (wt/vol) sucrose cushion by centrifugation at

30,000 × g for 2 h at 4 °C in a Ty 50.2 rotor (Beckman, Brea, CA), except that MeV and SeV were

purified using 10% (wt/vol) sucrose cushion. The pellet was resuspended in DMEM with 10%

trehalose.

5.3.4. Purification of rgRSV and mutants

Recombinant RSV containing m6A mutations on antigenome (rgRSV-GALL(+)), genome

(rgRSV-GALL (-)), both genome and antigenome (rgRSV-GALL (+/-)), and rgRSV with G gene

191

deletion mutant (rgRSV-ΔG) (kindly provided by Dr. Mark Peeples at Nationwide Children’s

Hospital, Columbus, OH) were grown on 10 T150 flasks of HEp-2 cells. RSV virions were purified

using the procedure described above.

5.3.5. Viral RNA extraction and real-time RT–PCR

Virion RNA was isolated from the purified virions using TRIzol reagent (Life Technologies,

Carlsbad, CA) following the manufacturer’s instruction. Where indicated, total RNA was extracted

from virus-infected cells using TRIzol reagent. Viral genome and antigenome RNA copies were quantified using reverse transcript (RT) real-time PCR using TB-Green premix Ex Taq (TaKaRa,

Kusatsu, Shiga Prefecture, Japan).

5.3.6. Colorimetric quantification of viral m6A methylation

Total m6A content of virion RNA was quantified by m6A RNA Methylation Assay Kit (Abcam,

ab185912) based on the manufacturer’s instruction. The percent change was calculated by dividing

m6A levels in viral RNA from METTL3-KO U2OS cells by those from the wild type U2OS cells.

The m6A level of rgRSV mutants grown in A549 cells was calculated using the parental rgRSV

grown in A549 cells.

5.3.7. Characterization of the recombinant m6A RSV mutants

Analysis of replication and gene expression of each recombinant rgRSV was determined in A549

cells and human bronchial epithelial (HBE) cultures. The protocols used for TCID50 assay for virus

titration, Western blot for viral protein expression analysis, growth kinetics, and genetic stability

were reported previously.

192

5.3.8. Measurement of interferon

For viral RNA transfection, same copies of viral RNA were transfected into regular A549 or A549 dual KO cells using Lipofectamine 3000 (ThermoFisher Scientific, Waltham, MA). At different time points after transfection, culture medium was harvested for quantification of IFN-β by commercial ELISA following the manufacturer’s instruction. Known concentrations of human

IFN-β were used to generate the standard curve. For viral infection, A549 cells were infected by same MOI of each virus treated or untreated with cycloheximide (CHX), cell lysates were collected at indicated time points and total RNA was extracted by Trizol reagent following the manufacturer’s instruction. IFN mRNA was quantified by realtime RT-PCR and normalized with mRNA of β-actin.

5.3.9. Western blot

After RNA transfection, cell pellet was lysed in 1× RIPA buffer (Abcam, Cambridge, United

Kingdom) for Western blot. Protein was detected by using antibody against RIG-I, MDA5, IRF3 and IRF3p (Phospho S386). β-actin was used as the loading control.

5.3.10. Ubiquitination reaction

Ubiquitination assay was performed as previously described [235]. Specifically, purified RIG-I

(1.0 mM) was first incubated with purified RSV virion RNA (0.2, 20, or 148.2 ng/µl) in buffer A

(20m MHEPES, pH 7.5, 150mM NaCl, 1.5mM MgCl2, 2mM ATP and 5mM DTT) at RT for 15 min. The RIG-I: RNA complex (to the final RIG-I concentration of 0.5 mM) was then further incubated with 20 mM ubiquitin, 1 mM mE1, 5 mM Ubiquitin-conjugating enzyme E2 13(Ubc13),

2.5 mM Ubiquitin-conjugating enzyme E2 variant 1A (Uev1A) and 0.25 mM E3 RIPLET in buffer

B (50 mM Tris pH 7.5, 10 mM MgCl2, 5 mM ATP and 0.6 mM DTT) at 37 ℃ for 30 min.

193

Reactions were quenched with SDS loading buffer, boiled at 96 ℃ for 5 min, and analyzed on

SDS-PAGE by anti-RIG-I or anti-Ub Western blot.

5.3.11. Replication and immunogenicity of rgRSV in cotton rats

For assessing the replication of each recombinant RSV in cotton rats, twenty 6-week-old specific-

pathogen-free (SPF) male cotton rats (Envigo, Indianapolis, IN) were randomly divided into 4

5 groups (5 cotton rats per group). Groups 1 to 4 were intranasally inoculated with 2.0×10 TCID50

of rgRSV-GALL(+), rgRSV-GALL(-), rgRSV-GALL(+/-), and parental rgRSV respectively. At

day 4 post-infection, the cotton rats were sacrificed, and lungs and nasal turbinates were collected

for both virus isolation and histological analysis.

For the immunogenicity evaluation, five groups of cotton rats (5 per group) were included. Groups

5 1 to 5 were intranasally inoculated with 2.0×10 TCID50 of rgRSV-GALL(+), rgRSV-GALL(-),

rgRSV-GALL(+/-), parental rgRSV, and PBS respectively. All groups were evaluated daily for

any possible abnormal reaction and blood were collected by orbital sinus bleeding at weeks 2, 3,

and 4 for measurement of neutralizing antibody. At week 4 post-immunization, cotton rats in all

5 groups were challenged with 2.0×10 TCID50 of parental rgRSV via intranasal route. At day 4

post-challenge, all cotton rats were sacrificed and their lungs and nasal turbinates were collected

for virus titration as described previously.

5.3.12. Mice study

Twenty five of 4 to 6-week-old specific-pathogen-free (SPF) female BALB/c mice (Charles River

Laboratories, Wilmington, MA) were randomly divided into 5 groups (5 mice per group). Mice in

6 group 1 to 5 were intranasally inoculated with 1.0×10 TCID50 of rgRSV-GALL(+), rgRSV-

GALL(-), rgRSV-GALL(+/-), parental rgRSV, and PBS respectively. After inoculation, the

194

animals were evaluated daily for mortality, body weight change, and the presence of any symptoms

of RSV infection. Blood samples were collected from each mouse weekly by orbital sinus blood

sampling, for neutralizing antibody detection.

5.3.13. Analysis of RSV-specific T cell immune responses

To measure RSV F-specific T cell immune response, spleens were collected at week 4 post- immunization, spleen cells were isolated and prepared as previously described. The cell concentrations were adjusted to 3 × 106 cells/mL and 100 µl were added into each well (3 wells per spleen sample) of a 96-well microtiter plate and cultured either alone or in the presence of

50 µg/ml of RSV pre-fusion F protein (kindly provided by Dr. Mark Peeples at Nationwide

Children’s Hospital, Columbus, OH) for 5 days at 37 °C in a 5% CO2 atmosphere. Culture

supernatants were collected from each well and frozen at −80 °C until analysis of secreted

cytokines using the Bio-Plex Pro Mouse Cytokine Standard 23-Plex, Group I (Bio-Rad

Laboratories Inc, Hercules, CA) per manufacturer’s instructions. The frequencies of RSV-specific

Th1 (IFN-λ+CD4+CD3+ and TNF-α -+CD4+CD3+), Th2 cells (IL-4+CD4+CD3+, IL-

5+CD4+CD3+, IL-10+CD4+CD3+), Th17 (IL-17A+ CD4+ CD3+), and Tfh (IL-21+ CD4+

CD3+) cells were determined by intracellular staining with the corresponding anti-cytokine Abs

(at a dilution of 1:5000) after additional incubation in the presence of PMA and ionomycin. The

cells were then analyzed with the aid of an Attune flow cytometer and data were expressed as

mean % positive cells ± standard deviation.

195

5.3.14. Statistic analysis

All data and statistical analysis were performed by using GraphPad Prism 8.0 (GraphPad Software,

San Diego, CA). Data were analyzed using unpaired Student’s t-test and a P value of <0.05 was

considered statistically significant. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

5.4 Results

5.4.1. Virion RNA purified from virions grown in METTL3-knockout U2OS cells is defective

in m6A methylation.

We first generated RSV RNA that was naturally defective in m6A methylation. Although several

host proteins were capable of adding m6A methylation to RNA, METTL3-METTL14 complex is

the major host m6A methyltransferase. METTL3 is the methyltransferase catalytic subunit whereas

METTL14 is the activator. Viral RNA will likely be defective in m6A methylation when it grows

in a cell line lacking METTL3. To test this hypothesis, rgRSV was grown in wild type U2OS and

METTL3-knockout U2OS cells and RSV virions were purified. Virion RNA was extracted and

quantified by real-time RT-PCR. The m6A level in RSV RNA was measured by m6A RNA

Methylation Assay Kit. Western blot analysis of cell lysates confirmed that METTL3 was indeed

knocked out in METTL3-knockout U2OS cells (Figure 5.1a). As shown in Figure 5.1b, RSV RNA

purified from virions grown in METTL3-knockout U2OS cells had a significantly lower level of m6A content than virion RNA purified from virions grown in wild type U2OS cells

(****P<0.0001). However, virion RNA from METTL3-knockout U2OS cells was not completely defective in m6A methylation, suggesting that METTL3 is not the only m6A writer protein in

U2OS cells.

196

We realized that U2OS cells were susceptible for infection by many NNS RNA viruses. Thus, we decided to test whether m6A-deficient virion RNA of other NNS RNA viruses can be produced using a similar strategy. To do this, measles virus (MeV, Paramyxoviridae) and Sendai virus (SeV,

Paramyxoviridae) were grown in wild type or METTL3-knockout U2OS cells. Virion RNA of each virus from each cell line was purified and m6A content was measured. Interestingly, for all viruses tested, m6A content of virion RNA from METTL3-knockout U2OS cells had a variable degree of reduction compared to that of virion RNA grown in wild type cells. Quantitatively, SeV and MeV had approximately 77% (Figure 5.2c) and 78% reduction (Figure 5.2d) in m6A respectively. Collectively, we developed a novel method to generate virion RNA that was naturally defective in m6A methylation, as their nucleotide sequences were not altered.

197

Figure 5.1. Fold change of m6A level in each virion RNA of virus grown from METTL3-knockout

U2OS cells or wild type U2OS cells.

(a)Western blot showing METTL3 expression in METTL3-knockout U2OS cells and wild type

U2OS cells; Relative m6A level in virion RNA from (b) RSV (c) SeV (d) MeV that grown on

METTL3 KO/WT cells. Data shown are mean ± s.d. of n= 3 (c) or n= 6 (b,d) independent experiments. Statistical significance was determined by two-sided Student’s t-test. *P < 0.05;

**P < 0.01; ***P < 0.001; ****P < 0.0001.

198

5.4.2. Virion RNA of rgRSV mutants is defective in m6A methylation

In Chapter 4, we have performed m6A-seq for RSV RNA and found that most of m6A peaks

enriched at the G region of both RSV genome and antigenome/mRNA. In addition, we have

recovered several m6A-deficient RSV mutants with their m6A sites in the G gene region of

antigenome/mRNA mutated. One of these rgRSV mutants is rgRSV-G123 in which all putative

m6A sites in the G gene region in the positive-sense RNA (antigenome and mRNA) were mutated.

The rgRSV-G123 was renamed as rgRSV-GALL(+). In this chapter, we recovered two more

rgRSV mutants. The first rgRSV mutant is rgRSV-GALL(-), in which all the putative m6A sites

on the G gene of RSV genome were mutated. The second rgRSV mutant is rgRSV-GALL(+/-),

which is the combination of all the m6A mutations from both the G region of RSV genome and the G gene region of antigenome/mRNA. All recombinant viruses were purified, and their virion

RNA were extracted and measured by m6A quantification kit.

As shown in Figure 5.2., compared to the virion RNA of the parental rgRSV, m6A level in virion

RNA of rgRSV-GALL(+), rgRSV-GALL(-) and rgRSV-GALL(+/-) were reduced by 30% to 50%

(P< 0.05). In addition, the combined mutant rgRSV-GALL(+/-) was more defective in m6A than

rgRSV-GALL(+) and rgRSV-GALL(-). In this experiment, we also used rgRSV-ΔG which lacks

the G gene as a control. We found the m6A level of each rgRSV mutant was similar to that of

rgRSV-ΔG, which was consistent with the fact that most of m6A peaks are enriched at the G region

of RSV genome and antigenome. Thus, these experiments demonstrated that rgRSV-GALL(+),

rgRSV-GALL(-) and rgRSV-GALL(+/-) were indeed defective in m6A methylation.

199

**** **** **** 1.2 **** rgRSV-GALL(+) rgRSV-GALL(-) 1.0 rgRSV-GALL(+/-) rgRSV-ΔG A levelA

6 rgRSV 0.8

Fold of m 0.6

0.4 G Δ rgRSV rgRSV-

rgRSV-GALL(+)rgRSV-GALL(-) rgRSV-GALL(+/-)

Figure 5.2. Fold change of m6A level in virion RNA of RSV mutant virus.

Relative m6A level in virion RNA from rgRSV mutants that with the G gene deleted (rgRSV-ΔG)

or with putative m6A sites mutated on either antigenome/genome or both. Data shown are mean ± s.d. of n= 3 independent experiments. Statistical significance was determined by two-sided

Student’s t-test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

5.4.3. m6A-deficient virion RNA induces a higher type I IFN response

Having generated a panel of m6A-deficient virion RNAs either purified from virions grown in

METTL3-KO cells or by site-directed mutagenesis, we next tested their abilities in triggering type

I IFN by transfection of virion RNA into A549 cells. Under this condition, there is no RNA

replication or viral protein synthesis, thus eliminating the effects of viral proteins in IFN signaling.

200

Briefly, same amount of virion RNA from each virus was transfected into A549 cells, and IFN-β was measured at different time points. As shown in Fig.5.3b-d, RNA purified from RSV virions grown in METTL3-KO U2OS cells induced a significantly higher IFNβ than RNA purified from virions grown in wt U2OS cells at 16 h, 24 h and 40 h post-transfection at all three different doses of RNA. Importantly, this phenomenon occurred to other NNS RNA viruses including SeV and

MeV (Figure 5.3e and f). Next, we compared the IFN response of virion RNA of rgRSV mutants and the parental rgRSV. When same amount of virion RNA was transfected into A549 cells, virion

RNA of all rgRSV mutants induced a significantly higher IFN-β than that of rgRSV at 18, 24, and

48 h after transfection. Notably, rgRSV-GALL(+/-) triggered a significantly higher IFN than rgRSV-GALL(-) (Figure 5.3a.), suggesting that the degree of m6A defect is correlated with the level of IFN production. Taken together, these results demonstrated that (i) m6A-deficient RSV virion RNA induced significantly more IFN than m6A-sufficient RNA; and (ii) the higher IFN production of m6A-deficient RNA was applicable to other NNS RNA viruses.

201

Figure 5.3. m6A-deficient virion RNAs induce higher type I IFN responses.

202

(a), Comparison of IFN response virion RNA of rgRSV-GALL(+), rgRSV-GALL(-), rgRSV-

GALL(+/-) and WT rgRSV. A549 cells were transfected with of 2×108 RNA copies of each mutant virus and IFN response was measured at indicated time points. (b-d), A549 cells were transfected with 109 RNA copies (b), 2×108 RNA copies (c) and 4×107 RNA copies (d) of virion RNA of RSV grown on METTL3 KO/WT cells, IFN response was measured at indicated time points. e, IFN-β

production in A549 cells at 16 and 24 h after transfected with 105 RNA copies of SeV virion RNA of SeV grown on METTL3 KO/WT cells. f, IFN-β production in A549 cells at 16 and 24 h after transfected with 5×106 RNA copies of MeV virion RNA of MeV grown on METTL3 KO/WT cells. Error bars represent SD from n = 3 independent experiments. *P<0.5, **P < 0.01,

***P < 0.001, ****P < 0.0001.

5.4.4. m6A-deficient virion RNA induces a stronger type I IFN signaling pathway

For cytoplasmic replicating RNA viruses, it is known that RIG-I-like receptors (RLRs) including

RIG-I and MDA5 are involved in detection of viral RNA. Upon recognition, RIG-I and MDA5

undergo a significant conformational change that transits a signal to the downstream adaptor

protein MAVS, activating IRF3 phosphorylation and translocation to nucleus, leading to IFN

production. To further investigate the mechanism underlying the enhanced IFN response

associated with m6A-deficient virion RNA, we examined the expression of molecules involved in

type I IFN signaling pathway.

Briefly, A549 cells were transfected with increasing amounts of m6A-deficient or sufficient RSV

virion RNA, and the expression of RIG-I, MDA5, IRF3, and phosphorylated IRF3 was detected

by Western blot. Interestingly, RNA purified from RSV virions grown in METTL3-KO U2OS

203 cells induced significantly higher expression of RIG-I, MDA5, and IRF3 phosphorylation than

RNA purified from virions grown in WT U2OS cells (Figure 5.4a.). The difference was more obvious when a lower dose of RNA was used for transfection.

Using a similar approach, we tested whether this phenomenon is universal for SeV and MeV.

Similar analyses were conducted and it was found that both m6A-deficient SeV and MeV RNA could enhance phosphorylation of IRF3. Similar to RSV, m6A-deficient MeV virion RNA triggered stronger expression of RIG-I and MDA5, and IRF3 phosphorylation than m6A-sufficient

MeV virion RNA (Figure 5.4c). Interestingly, m6A-deficient SeV virion RNA had a similar level of RIG-I and MDA expression but had a higher IRF3 phosphorylation compared to m6A-suficient

SeV virion RNA (Figure 5.4b). Although the expression level of RIG-I and MDA5 may be different in different viruses, the higher IRF3 phosphorylation occurred to m6A-deficient virion

RNA of all tested viruses, which was consistent with the fact that they induced higher IFN production.

We also assessed the ability of virion RNA of rgRSV mutants in activating IFN signaling pathway.

Compared to virion RNA of the parental rgRSV, virion RNA of rgRSV-GALL(+/-) induced a higher expression of RIG-I and MDA5, and IRF3 phosphorylation (Figure 5.4d). Collectively, these results demonstrated that m6A-deficient virion RNA activated a higher type I IFN signaling than m6A-sufficient virion RNA.

204

Figure 5.4. m6A-deficient viral RNA enhances expression of molecules that involved in type I IFN signaling pathway.

205

a, m6A-deficient RSV RNA increases expression of RIG-I and MDA5 and induces higher IRF3

phosphorylation. A549 cells were transfected with virion RNA of RSV grown on METTL3

KO/WT cells at dose of 1.0×109, 2×108 and 4×107 RNA copies At indicated times, cell lysates

were analyzed by Western blotting using antibodies specific to RIG-I, MDA5, IRF3, IRF3

(phosphorylated at S386) or β-actin. Comparison of RIG-I, MDA5, and phosphorylated IRF3

triggered by m6A-deficient SeV virion RNA (b) and m6A-deficient MeV virion RNA (c). A549 cells were transfected with virion RNA of SeV grown on METTL3 KO or WT cells at dose of

2×105, 1×105, and 5×104 RNA copies; virion RNA of MeV grown on METTL3 KO or WT cells

at dose of 1×106, 5×105, and 2×105 RNA copies. At indicated times, cell lysates were analyzed by

Western blotting using antibodies specific to RIG-I, MDA5, IRF3, IRF3 (phosphorylated at S386)

or β-actin. d, Virion RNA of RSV mutant rgRSV-GALL(+/-) induced higher expression of RIG-

I, MDA5, and phosphorylated IRF3. A549 cells were transfected with virion RNA of rgRSV-

GALL(+/-) and parental rgRSV at dose of 2×108, 5×107, and 1×106 RNA copies. At indicated

times, cell lysates were analyzed by Western blotting using antibodies specific to RIG-I, MDA5,

IRF3, IRF3 (phosphorylated at S386) or β-actin. Western blots are representative of n= 3

biologically independent experiments.

5.4.5. RIG-I is a major RNA sensor that recognizes m6A-deficient virion RNA

To directly determine which RNA sensors are involved in the detection of m6A-deficient RNA,

we examined the IFN response of m6A-deficient virion RNA in A549 cells lacking RIG-I, MDA5,

or their downstream adaptor protein, MAVS. Briefly, wt A549-Dual or KO A549-Dual cells were

transfected with virion RNA grown in METTL3-KO U2OS cells or wt U2OS cells, and IFN-β was

206 detected by ELISA. As shown in Figure 5.5a., RSV virion RNA from METTL3 KO U2OS cells triggered significantly more IFN-β production than virion RNAs from WT U2OS cells in wild type

A549-Dual cells (P < 0.05; Student's t-test). Notably, when RIG-I was knocked out, no IFN-β was detected at 24 h post-transfection and a very low level of IFN-β was detected at 40 h post- transfection. In contrast, a considerable level of IFN-β was still detectable in MDA5-knockout

A549-Dual cells (Figure 5.5b) although there was a significant reduction compared to the control

A549-Dual cells. In addition, RSV virion RNA from METTL3 KO U2OS cells triggered significantly more IFN-β production than virion RNAs from WT U2OS cells in MDA5-KO A549-

Dual cells. IFN-β production was completely abrogated in A549-Dual cells lacking MAVS, which is consistent with the fact that MAVS is the downstream adaptor protein in the type I IFN signaling.

Therefore, these results demonstrate that RIG-I plays a major role and MDA5 plays a minor role in recognizing m6A-deficient RSV RNA.

Using a similar approach, we compared the IFN responses of m6A-deficient and sufficient virion

RNA of SeV and MeV RNA. Exact same pattern was observed for SeV and MeV RNA (Figure

5.6. and Figure 5.7.). IFN-β was completely abrogated in RIG-I and MAVS-knockout A549-Dual cells when transfected with m6A-deficient or sufficient virion RNA. In contrast, a high level of

IFN-β was still detectable in MDA5-knock out A549-Dual cells although there was a significant reduction. Similar to RSV, RIG-I plays a major role in recognizing m6A-deficient MeV and SeV

RNA.

207

Figure 5.5. IFN response in A549 cells transfected with m6A-deficient RSV virion RNA.

a–d, Confluent wild-type (a), MDA5-knockout (b), RIG-I-knockout (c) or MAVS-knockout (d)

A549 Dual cells were transfected with same amount (2×107 RNA copies) of RSV virion RNAs

from either METTL3 KO cells or METTL3 WT cells. Cell culture supernatants were harvested at

24 and 40 h after inoculation, IFN-β in the supernatant at indicated time points was measured by

commercial ELISA kit. Error bars represent SD from n = 3 independent experiments. *P<0.5,

**P < 0.01, ***P < 0.001, ****P < 0.0001.

208

Figure 5.6. IFN production in A549 deficient cells stimulated with m6A-deficient SeV virion RNA.

a–d, Confluent wild-type (a), MDA5-knockout (b), RIG-I-knockout (c) or MAVS-knockout (d)

A549 Dual cells were transfected with same amount (1×105 RNA copies) of SeV virion RNAs from either METTL3 KO cells or METTL3 WT cells. Cell culture supernatants were harvested at

24 and 40 h after inoculation, IFN-β in the supernatant at indicated time points was measured by

commercial ELISA kit. Error bars represent SD from n = 3 independent experiments. *P<0.5,

**P < 0.01, ***P < 0.001, ****P < 0.0001.

209

Figure 5.7. IFN production in A549 deficient cells stimulated with m6A-deficient MeV virion

RNA.

a–d, Confluent wild-type (a), MDA5-knockout (b), RIG-I-knockout (c) or MAVS-knockout (d)

A549 Dual cells were transfected with same amount (2×105 RNA copies) of MeV virion RNAs

from either METTL3 KO cells or METTL3 WT cells. Cell culture supernatants were harvested at

24 and 40 h after inoculation, IFN-β in the supernatant at indicated time points was measured by

commercial ELISA kit. Error bars represent SD from n = 3 independent experiments. *P<0.5,

**P < 0.01, ***P < 0.001, ****P < 0.0001.

210

5.4.6. Enhanced recognition of m6A-deficient viral RNA by RIG-I

In order to explore the possible mechanisms by which m6A inhibits the innate immune response to viral RNA, we directly compared the binding affinity of m6A-containing and m6A-deficient viral RNA to RIG-I protein as previously described [234]. Briefly, same amount of each viral RNA was incubated with cell lysate that containing Flag-tagged RIG-I protein. The amount of viral RNA pulled down by RIG-I was quantified by real-time PCR by using primers that annealing to genome or antigenome of each virus. As the quantity control, equal amount of RIG-I protein was detected in each sample, confirming that the pull-down efficiency of each group was the same (Figure 5.8.c and d). As shown in Figure 5.8. a and b, RIG-I pulled down significantly more RSV virion RNA

(genome and antigenome) from METTL3 KO U2OS cells than RSV virion RNA from wt U2OS cells. A 50-fold and 17-fold increase in RIG-I binding was observed for m6A-deficient antigenome and genome, respectively. Similarly, RIG-I pulled down significantly more virion RNA of rgRSV-

GALL(+/-) than that of rgRSV. In this case, a 4-fold and 3-fold increase in RIG-I binding was observed for antigenome and genome, respectively (Figure 5.8.e and f). These results suggest that m6A-deficient RSV RNA enhance the binding affinity to RIG-I.

211

Figure 5.8. Affinity binding assay of RIG-I with RSV RNA.

Pull down antigenome (a), genome (b), and RIG-I (c) of RSV RNA from METTL3 KO/WT cells.

Pull down antigenome (e) and genome (f) of m6A mutant and wild type RSV. d, western blot of pull-down RIG-I showing the equal pull down efficiency for mutant and wild type RNA. Error bars represent SD from n = 3 independent experiments. *P<0.5, **P < 0.01, ***P < 0.001,

****P < 0.0001.

Using a similar assay, we also tested the binding affinity of SeV and MeV RNA to RIG-I. Cell lysates with Flag-RIG-I was equally divided into four reactions, each of them was used for pulldown with different viral RNA. For SeV virion RNA, a 10-fold increase in RIG-I binding was observed for genome of m6A-deficient virion RNA compared to m6A-sufficient virion RNA

212

(Figure 5.9b). For MeV virion RNA, a 30-fold and 10-fold in RIG-I binding was observed for antigenome (Figure 5.9c) and genome (Figure 5.9c) of m6A-deficient RNA, respectively, compared to those of m6A-sufficent RNA. Thus, m6A-deficient virion RNA of SeV and MeV enhances the binding to RIG-I.

Figure 5.9. Affinity binding assay of RIG-I with SeV and MeV RNA.

213

a, One aliquot of beads was analyzed by Western blot. b, pulldown of SeV genome. c, pulldown

of MeV antigenome (c) and genome (d). Error bars represent SD from n = 3 independent

experiments. *P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

5.4.7. m6A-deficient RSV RNA enhances ubiquitination of RIG-I

Ubiquitination plays essential roles in RIG-I activation as well as type I IFN signaling pathway.

Upon binding to the RNA ligand, RIG-I must undergo significant conformational change for K63-

linked ubiquitination at multiple sites to reach full activity. Recent studies found that RIG-I

ubiquitination and activation is dependent on RIPLET [235]. Thus, we tested the abilities of m6A- deficient RSV RNA in promoting ubiquitination of RIG-I. In this experiment, a 42bp dsRNA was used as a positive control. As expected, incubation of purified RIG-I with 42bp dsRNA triggered a strong ubiquitination of RIG-I in the presence of RIPLET. No RIG-I ubiquitination was observed in the absence of RIPLET (Fig.5.10a), demonstrating that RIPLET is essential for RIG-I ubiquitination. As shown in Fig.5.10a, RSV RNA promoted RIG-I ubiquitination in a dose-

dependent manner only in the presence of RIPLET. RIG-I ubiquitination increased as dose of RSV

RNA increased. Although there was no significant difference in RIG-I ubiquitination at low doses

of RSV RNA (0.1 and 10 ng/µl), virion RNA from METTL3-KO U2OS cells induced a

significantly stronger RIG-I ubiquitination than virion RNA from wt U2OS cells at higher

concentration (74.1 ng/µl) (Figure 5.10a). Similarly, virion RNA of rgRSV-GALL(+/-) induced much more RIG-I ubiquitination of than virion RNA of rgRSV at doses of 10 and 74.1 ng/µl RSV

RNA (Figure 5.10b). These data demonstrate that m6A-deficient RSV RNA has higher ability in

214 triggering the ubiquitination of RIG-I, providing a mechanistic insight for its higher ability in triggering IFN production.

Figure 5.10. In vitro ubiquitination analysis of RIG-I.

215

1.0 uM of purified RIG-I were incubated with 1ng/ µg of 42 bp dsRNA, different doses of RSV virion RNA from METTL3 KO U2OS cells or wt U2OS cells (a), or different doses of virion RNA from rgRSV mutant or rgRSV (b). Ubiquitination of RIG-I was analyzed by anti-RIG-I blot.

5.4.8. m6A-deficient rgRSV infection triggers higher IFN

We have revealed that m6A-deficient virion RNA of RSV enhanced the IFN response. In order to connect the observation of virion RNA to viral infection, we have tested the IFN production in m6A-deficient rgRSV infection. It should be noted that IFN response is more complicated in during virus infection as several viral proteins (such as NS1 and NS2) and RNA replication regulate IFN production. In addition, RSV replication requires ongoing viral protein synthesis.

We tested the IFN responses of m6A-deficient rgRSV grown in METTL3-KO U2OS cells. First, we measured the IFN production in infected cells treated with cycloheximide (CHX). Application of CHX allows particle attachment, entry of nucleocapsids, and primary transcription, but not genome replication and ongoing viral protein synthesis [236]. Briefly, A549 cells were treated for

1 h with 50 μg/ml of CHX, and then infected by rgRSV grown from METTL3-KO U2OS or WT

U2OS cells at an MOI of 0.5. As shown in Figure 5.11a, when treated with CHX, m6A-deficient rgRSV grown from METTL3-KO U2OS cells produced significantly higher IFN-β mRNA than m6A-sufficient rgRSV grown in WT U2OS cells (P<0.05). It is likely that IFN response was not strongly activated because of the low quantities of incoming viral genome and antigenome when

CHX was applied. Interestingly, under normal virus infection (no CHX treatment), a more significant increase was observed in viral infection of rgRSV grown from METTL3-KO U2OS cells.

216

We also examined the IFN-β mRNA in A549 cells infected by rgRSV mutants. Briefly, confluent

A549 cells were infected with rgRSV-GALL(+), rgRSV-GALL(-),rgRSV-GALL(+/-) or parental rgRSV at an MOI of 0.5. Cell lysates were harvested at 18, 24, and 48h post-infection, total RNA was extracted, and IFN-β mRNA was quantified by real-time RT-PCR. As shown in Figure 5.11b, all m6A-deficient rgRSV mutants triggered significantly higher IFN- β mRNA than the parental rgRSV at the indicated time points. Taken together, those results indicate that m6A-deficient rgRSVs activate higher IFN responses than m6A-sufficient rgRSVs during viral infection.

Figure 5.11. IFN response in A549 after viral infection. 217 a, IFN-β mRNA expression in rgRSV-infected A549 cells with CHX or without CHX (UT). A549 cells in 24-well-plate were treated with 0 or 50 μg/ml CHX for 1 h, and then infected either with rgRSV grown from METTL3-KO U2OS or WT U2OS cells at MOI 0.5, and 500 ul of DMEM with 50 μg/ml CHX was added. b, IFN-β mRNA expression in rgRSV mutant-infected A549 cells.

A549 cells in 24-well-plate were infected with same amount of each m6A-deficient rgRSV mutant, total RNA was extracted at 18, 24 and 48h after infection, IFN-β mRNA was quantified by real- time RT-PCR and normalized with β-actin. Error bars represent SD from n = 3 independent experiments. *P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

5.4.9. m6A-deficient rgRSV mutants are attenuated in vitro

We next determined whether rgRSV mutants were attenuated in cell culture. Briefly, A549 cells were infected rgRSV, rgRSV-GALL(+), rgRSV-GALL(-) or rgRSV-GALL(+/-) at an MOI of 0.1, viral replication, gene expression, and RNA synthesis was analyzed. As shown in Fig.5.11a, all rgRSV mutants had significantly reduced GFP expression compared to the parental rgRSV. The combined mutant rgRSV-GALL(+/-) was the most defective virus for GFP expression Figure

5.11a). The number of GFP positive cells and the intensity of GFP were monitored by flow cytometry. As shown in Figure 5.12b and c, all rgRSV mutants produced much less GFP positive cells and had significantly weaker GFP density compared to the parental rgRSV. One-step growth curve showed that all rgRSV mutants had delayed replication kinetics and had 0.5 log reductions in peak titer compared to the rgRSV (Figure 5.11d).

218

Figure 5.12. m6A-dificient RSVs are attenuated in replication in A549 cells.

a, Delayed GFP expression by rgRSV mutants in A594 cells. Confluent A549 cells were infected by each rgRSV at an MOI of 0.1, and GFP expression was monitored at the indicated times by fluorescence microscopy. b-c, Quantification of GFP positive cells by flow cytometry. Confluent

A549 cells were infected by each rgRSV (MOI of 0.1), at indicated time point, cells were trypsinized and GFP positive cells (b) and GFP intensity (c) were quantified by flow cytometry.

d, Single-step growth curve of rgRSV mutants. A549 cells in 24-well-plates were infected with

each recombinant rgRSV at an MOI of 0.1. The viral titer was determined by TCID50 assay in

A549 cells. Viral titers are the geometric mean titer (GMT) of three independent experiments ±

standard deviation. Error bars represent SD from n = 3 independent experiments. *P<0.5,

**P < 0.01, ***P < 0.001, ****P < 0.0001. 219

We also determined viral protein expression in virus-infected cells. A549 cells were infected with

each virus at an MOI of 0.1, cells were lysed at 18, 24, and 48 h and viral protein expression was

analyzed by Western blot using antibody against G and F proteins. As shown in Figure 5.13a., all

rgRSV mutants had defects in G and F protein synthesis, particularly at the early time points. We

also examined viral RNA synthesis in virus-infected cells by real-time RT-PCR. All rgRSV mutants had significant reductions (1.0-1.5 log RNA copies) in both genome and G mRNA

synthesis (Figure 5.13b and c). Together, all rgRSV mutants were attenuated in A549 cells and

rgRSV-GALL(+/-) was the most attenuated virus.

We also tested replication of rgRSVs in human airway epithelial (HBE) cultures, the ex vivo model

for RSV infection. Briefly, HBE cultures were infected with 400 TCID50 of each virus, GFP

expression was monitored daily and the released virus was measured by TCID50 assay until day

14 after infection. As shown in Figure 5.14., all rgRSV mutants were defective in spreading and

GFP expression at days 1 and 2 but gradually reached a similar level at day 4 post-inoculation.

Again, rgRSV-GALL(+/-) was the most defective virus in spreading in HBE culture. In addition,

all rgRSV mutants had a delay in virus release in HBE cultures compared to rgRSV (Fig.5.14b).

220

Figure 5.13.Viral expression and RNA synthesis of RSV mutant.

221

a, A549 cells were infected with the parental rgRSV or rgRSV mutants at an MOI of 0.1. At 18,

24, and 48 h post-inoculation, total cell lysates were harvested and subjected to Western blotting

using a monoclonal antibody against RSV F or G protein. b, RSV genomic RNA replication. At

18, 24, and 48 h post-infection, total RNA was purified from rgRSV-infected cells using TRizol,

and genomic RNA was quantified by real-time RT-PCR using specific primers annealing to the

RSV leader sequence and GFP gene. c, RSV G mRNA transcription. Viral mRNA was quantified

by real-time PCR using primers annealing to the G. Results are from three independent

experiments. RNA copy and viral titer are the geometric mean titer (GMT) of three independent experiments ± standard deviation. Western blots shown are the representatives of three independent experiments. Data were analyzed using Student’s t-test and *P < 0.05; **P < 0.01;

***P < 0.001; ****P < 0.0001.

222

Figure 5.14. Virus replication in HBE cultures.

223

6 a, Spreading of m A mutated rgRSVs in HBE culture. HBE cultures were infected by 400 TCID50

of each rgRSV. At the indicated time, virus spreading was monitored by fluorescence microscopy.

Representative micrographs at each time point are shown. b, Virus release from m6A-deficient rgRSV-infected HBE culture. HBE cultures were infected by 400 TCID50 of each rgRSV. After

virus inoculation, supernatants were collected on day 1 and every 2 days until day 12 post- inoculation. Infectious virus in supernatants was determined by TCID50 assay. Viral titers are the

geometric mean titer (GMT) of three independent experiments ± standard deviation. Data were

analyzed using Student’s t-test and *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

5.4.10. m6A-deficient rgRSV mutants trigger stronger neutralizing antibody and T cell

responses.

We decided to test the adaptive immune responses of rgRSV mutants in mice because of the

availability of immunological reagents in mice, allowing us to analyze antibody and T cell immune

responses. It should be noted that high doses of RSV can cause clinical symptoms in BALB/C

mice including body weight loss, ruffled fur, and a hunched posture [237]. Briefly, mice in each

6 group were intranasally inoculated with 1.0×10 TCID50 of rgRSV-GALL(+), rgRSV-GALL(-),

rgRSV-GALL(+/-), parental rgRSV, and PBS respectively. Daily body weight change was

monitored and blood was collected every week after inoculation.

As shown in Figure 5.15a., all rgRSV mutants caused significant body weight losses in mice. Mice

in rgRSV-GALL(+) and rgRSV had similar body losses but were recovered at day 8. Mice in

rgRSV-GALL(-) and rgRSV-GALL(+/-) had a significantly less body losses compared to rgRSV

and were recovered at day 4. All mice in rgRSV group showed significant clinical signs including

224 ruffled fur. Interestingly, mice in rgRSV-GALL(+)groups showed no obvious clinical signs even though they had a similar level of body weight losses as rgRSV group. Mice in rgRSV-GALL(-) and rgRSV-GALL(+/-) group did not exhibit any clinical signs, suggesting that these two m6A- deficient rgRSV were attenuated in mice.

Weekly serum samples were collected from each mouse and neutralizing antibodies were measured (Figure 5.15b). Notably, all three m6A-deficient rgRSV groups showed significantly higher neutralizing antibodies at weeks 2 and 3 after virus inoculation. This result suggests that m6A-deficient rgRSV mutants are capable of inducing higher neutralizing antibodies in mice.

225

Figure 5.15. Body weight (a) and neutralizing antibodies (b) of mice that infected with m6A- deficient and wild type RSV.

Error bars represent SD from n = 5 cotton rats. *P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

226

At week 4 post-inoculation, mice were terminated and spleen cells were isolated for T cell assay.

Th1 cells play crucial roles in preventing virus infections by producing cytokines that involved in

the generation of complement fixing Abs and cytotoxic T cells. Flow cytometry analysis of

CD3+CD4+ cells producing Th1 cytokines (Figure 5.16.) indicated that mice immunized with all

three m6A-deficient rgRSVs producing significantly more IFN-γ T cells (CD4+IFN-γ+) than mice immunized with the parental rgRSV. Interestingly, TNF-α producing T cells (CD4+TNF-α+) were detected in the spleens of mice immunized rgRSV-GALL(+) and rgRSV-GALL(-), but not rgRSV-

GALL(+/-) or rgRSV. These results indicated that the production of T helper cells is relevant with virus replication efficiency.

Th2 cells trigger cytokines that involved in antibodies against extracellular pathogens. Follicular

T helper cells (Tfh) and Th17 cells are important in promoting affinity maturation of antibodies,

and the signature products are interleukin 21 and IL-17A, respectively. As shown in Figure 5.17.

and 5.18., rgRSV-GALL(+) triggered slightly higher CD4+IL-4+ than other groups. Both rgRSV-

GALL(+) and rgRSV-GALL(-) generated significantly more CD4+IL-5+ and CD4+IL-10+

compared to the parental rgRSV. All rgRSV mutants produced much stronger CD4+IL-10+ than

rgRSV. Additionally, slightly more Tfh cells was observed in rgRSV-GALL(-) group whereas

significantly higher Th17 responses (CD4+IL-17A+) were observed after immunized with rgRSV-

GALL(+) and rgRSV-GALL(-). These results demonstrate that both m6A-deficient rgRSV

mutants and rgRSV are capable of triggering RSV-specific T cell responses and m6A-deficient

rgRSVs enhanced most types of T cell responses compared to the parental rgRSV.

227

Figure 5.16. RSV-specific Th1 cell response triggered by pre-fusion F protein of RSV.

Proliferation of (a), IFN-γ+CD4+ and (b), TNF-α+CD4+. Error bars represent SD from n = 5 mice.

*P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 5.17. RSV-specific Th2 cell response triggered by pre-fusion F protein of RSV.

Proliferation of IL-4+CD4+(a), IL-5+CD4+ (b) and IL-10+CD4+ (c). Error bars represent SD from n = 5 mice. *P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

228

Figure 5.18. RSV-specific Tfh (a) and Th17 (b) cell response triggered by pre-fusion F protein of

RSV.

Error bars represent SD from n = 5 mice. *P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

5.4.11. m6A-deficient rgRSV mutants are attenuated in cotton rats and provide full protection against RSV challenge

We next determined the replication of rgRSV mutants in cotton rats, as RSV replicates more robustly in lungs and nasal turbinate of cotton rats than those of mice. Briefly, cotton rats were

5 intranasally inoculated 2×10 TCID50 of each rgRSV, cotton rats were terminated at day 4, and viral titer in lung and nasal turbinate were determined. Recombinant rgRSV-GALL(+), rgRSV-

GALL(-) and rgRSV-GALL(+/-) had 0.1-1 log and 1.1-1.4 log virus reduction in lungs and nasal turbinate respectively. Therefore, these rgRSV mutants had various degrees of attenuation in cotton rats and rgRSV-GALL(+/-) was the most attenuated mutant (Figure 5.19).

229

We also determined the immunogenicity of each rgRSV mutant in cotton rats. Briefly, cotton rats

5 in each group were intranasally immunized with 2×10 TCID50 of recombinant parental rgRSV-

GALL(+), rgRSV-GALL(-), rgRSV-GALL(+/-),or parental rgRSV. PBS was used as a control.

Blood was collected from each cotton rats every week and cotton rats were sacrificed at week 4 and analyzed as previously. The unimmunized but challenged groups had 4-5 log TCID50 viral titer in lungs and nasal turbinate. All immunized animals were completely protected from rgRSV challenge. No infectious RSV was detected in lung or nasal turbinate (Figure 5.20a). We also measured the neutralizing antibodies in the vaccinated cotton rats. As shown in Figure 5.20b, all three m6A-deficient rgRSV generated similar levels of neutralizing antibodies compared to the parental rgRSV. These data demonstrate that all three m6A-deficient rgRSV are highly immunogenic in in cotton rats.

230

Figure 5.19. Replication of m6A-deficient RSV in cotton rats.

231

Virus titer in lungs (a), and nasal turbinates (b). Four-week-old SPF cotton rats were inoculated intranasally with 2.0×105 TCID50 of each rgRSV. At day 4 post-infection, the cotton rats were sacrificed, and lungs and nasal turbinate were collected for virus titration by TCID50 assay. Viral

titers are the geometric mean titer (GMT) of 5 animals ± standard deviation. The detection limit is

2.0 log TCID50/g tissue. *P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 5.20. Immunogenicity of m6A-deficient and wild type RSV.

232

5 Four-week-old SPF cotton rats were inoculated intranasally with 2.0×10 TCID50 of each rgRSV.

Blood samples were collected from each rat weekly by retro-orbital bleeding. a, Protection from

5 RSV challenge. At week 4 post-immunization, cotton rats were challenged with 2.0×10 TCID50 rgRSV. At day 4 post-challenge, the cotton rats were sacrificed, and lungs and nasal turbinate were collected for virus titration by TCID50 assay. Viral titers are the geometric mean titer (GMT) of 5 animals ± standard deviation. The detection limit is 2.0 log TCID50/g tissue. b, Neutralizing antibody production. The RSV-neutralizing antibody titer was determined using a plaque reduction neutralization assay. *P<0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001.

5.5 Discussion

A live attenuated vaccine is one of the most promising vaccines for RSV. However, most live attenuated RSV vaccine candidates are less immunogenic than the wild type virus. One of the most important approaches to enhance the immunogenicity is to enhance the innate immunity of live attenuated RSV vaccines. In this study, we found that viral RNA m6A methylation modulated both innate and adaptive immune responses of RSV. Specifically, we found that m6A-deficient virion

RNA induced significantly higher type I IFN responses and m6A-deficient rgRSVs induced higher neutralizing antibody and T cell immune responses in mice and provided complete protection against RSV challenge. These results support the hypothesis that a higher innate immune response induced by RSV induces a stronger adaptive immune response. Thus, inhibition of viral m6A methylation is a novel approach to enhance both innate and adaptive immune responses of RSV vaccine candidates.

233

Another important finding from our study is that m6A methylation in RSV antigenome and genome

may serve as a means of mimicking host RNA to avoid detection by the innate immune system via

RIG-I mediated IFN signaling pathway. First, RSV virion RNA that is naturally defective in m6A

methylation triggered a higher type I IFN response. Second, RSV virion RNA with mutated m6A-

sites in G gene of antigenome and/or genome induced a significantly higher type I IFN response.

Third, RIG-I but not MDA5 plays a major role in recognizing m6A-deficient virion RNA. Fourth,

m6A methylation prevents the antigenome and genome from the recognition by RIG-I, thereby

inhibiting RIG-I-dependent production of type I IFN in virion RNA-transfected cells. Fifth, m6A-

deficient antigenome and genome RNA enhances the activation of the RIG-I signaling pathway,

including RIG-I expression, RIG-I binding affinity, RIG-I ubiquitination, and IRF3

phosphorylation, leading to an enhanced production of IFN. Our results suggest that m6A

methylation of the antigenome and genome is a molecular marker for host innate immunity to

discriminating self from non-self RNA through the RNA sensor RIG-I.

We also found that the roles of m6A methylation in innate immunity is universally conserved in

other non-segmented negative-strand (NNS) RNA viruses in the order Mononegavirales. Using

METTL3-knockout U2OS cells, we generated m6A-deficient virion RNA of SeV, MeV and

hMPV, and found that m6A-deficient virion RNA induced a significantly higher type I IFN through

RIG-I mediated signaling pathway. The NNS RNA viruses include many significant human and

animal pathogens, such as RSV, Ebola virus, Marburg virus, virus, Hendra virus, and Nipah

virus. For many of these viruses, there are no effective vaccines or anti-viral drugs. These viruses

share many common strategies in replication and gene expression. The finding that m6A-deficient

virion RNA induced a higher type I IFN suggests that inhibition of m6A methylation may serve as

234

a universal approach to enhance innate immune response of other NNS RNA viruses. The data

presented here expands our knowledge of interaction between virus and host by highlighting the

function of epitranscriptomic modifications on viral RNA, which also will be crucial to build the

fundamental knowledge to better understand immune evasion of NNS RNA viruses.

Like all other NNS RNA viruses, the genome of RSV is not naked but encapsidated with the nucleoprotein (N) to form a helical nucleocapsid (N-RNA complex), serving as a template to

synthesize antigenome and also encapsidated by N, which in turn synthesizes progeny genome

RNA. Two important questions have not been addressed in this study are how RIG-I gains access

to the N-encapsidated genome and/or antigenome and how RIG-I interacts with m6A-deficient or

sufficient RSV RNA. A possibility is that all of the genomes and antigenomes that are synthesized

are not encapsidated. Particularly early in the infectious cycle, when the concentration of the N

protein is low, some of these full-length RNA genomes and antigenomes may not be encapsidated,

enabling RIG-I access to both genome and antigenome. Another possibility is that m6A

methylation may enhance encapsidation of genome and antigenome. In addition, m6A reader

binding proteins will bind to the m6A methylated genome and antigenome which may escape the

recognition of RIG-I. Once genome and antigenome are not m6A methylated, encapsidation rate

may be reduced. At the same time, m6A reader binding proteins are not able to bind to these m6A-

deficient RNAs, leading to the exposure of these RNAs to RIG-I. In conclusion, we have found

that inhibition of m6A methylation in RSV genome and antigenome induced a significantly higher

innate immunity, which in turn induced a higher adaptive immune response against RSV infection.

This strategy can be used for development of a live attenuated RSV vaccine with higher

immunogenicity.

235

Chapter 6. Summary and future directions

6.1. Summary

Human respiratory syncytial virus (RSV) is the leading causative agent of acute viral respiratory

tract infections in infants, young children, the elderly, and immunocompromised individuals [6].

Despite major efforts, no vaccines or anti-viral drugs are currently available. Development of such

agents requires a better understanding of all aspects of virus life cycle. RNA modification is an essential issue in RSV replication and gene expression, which may be used as a target for the development of vaccines and antiviral drugs. In this dissertation work, we have revealed the critical roles of viral mRNA cap methylation in RSV replication and gene expression and found that inhibition of RSV mRNA cap methylation results in attenuation of RSV but retains high

immunogenicity. In addition to mRNA cap methylation, we have discovered that RSV genome,

antigenome, and mRNAs are post-transcriptionally methylated at internal adenosine residues to

form N6-methyladenosine (m6A). We found that viral m6A plays pro-viral roles in RSV replication and gene expression which can be used as a new target for attenuation of RSV for the development of live attenuated vaccines. Importantly, we also found that inhibition of viral m6A methylation

enhances innate immunity, which in turn enhances immunogenicity of RSV. Therefore, inhibition

of viral RNA modification may lead to improved live vaccine candidates for RSV.

236

(1) Identify mRNA cap methylation as an approach to attenuate RSV. We characterized the roles

of S-adenosylmethionine (SAM) binding site in the methyltransferase (MTase) region of the L

protein in RSV replication and gene expression. We found that mutations to the SAM binding site

in the L protein reduced the replication and transcription in a minigenome assay. We found that

recombinant RSVs carrying mutations in the SAM binding site grew to a high titer in cell culture,

were genetically stable and sufficiently attenuated in vitro and in vivo, highly immunogenic, and

induced compete protection against RSV challenge. We also proved the concept that combination

of multiple mutations in the SAM binding site led to a recombinant virus that is genetically stable,

sufficiently attenuated but highly immunogenic. Thus, targeting RSV mRNA cap MTase is a

novel strategy to balance the attenuation and immunogenicity of RSV live attenuated vaccine candidates.

(2) Identify a novel approach to enhance the genetic stability of RSV live attenuated vaccine

candidates. Genetic stability is one of the major challenges for developing a live attenuated RSV

vaccine. Historically, several live attenuated RSV vaccine candidates reverted into wild type virus

in clinical trials, raising a serious safety issue. We found for the first time that the flexible hinge

region in the conserved region V (CR V) and CR VI of RSV L protein tolerates amino acid deletion

or insertion. We found that a single or double amino acid deletion or a single alanine insertion at

the flexible hinge region was sufficient to attenuate RSV. Recombinant RSVs with amino acid

deletion or insertion are highly immunogenic and provide complete protection in cotton rats. This

strategy can enhance genetic stability of RSV mutants, as it is unlikely that these amino acid

deletions or insertions will be reverted back.

237

(3) Discover RSV RNA contains internal m6A methylation and identify viral m6A methylation as

a new target to attenuate RSV. Using high throughput m6A-seq technique, we found that for the first time that RSV genome, antigenome, and mRNAs contain m6A methylation and that viral m6A

methylation positively regulates viral replication and gene expression. We characterized the roles

of host m6A machinery in regulating RSV replication and gene expression. We generated a panel

of m6A-deficient rgRSVs that were attenuated and immunogenic in vivo. Therefore, our work

demonstrates that inhibition of m6A methylation may be a novel approach for rationally designing

RSV live attenuated vaccines.

(4) Discover the role of viral m6A methylation in innate and adaptive immune responses. One

major challenge in the development of a live attenuated RSV is how to enhance innate immune

responses of attenuated RSV strains. We found that m6A-deficient RSV virion RNA induced

significantly higher type I interferon (IFN) compared to m6A-suficient virion RNA. This enhanced

IFN signaling is universally conserved in other non-segmented negative-sense RNA viruses.

Recombinant m6A-defective RSVs were attenuated both in vitro and in vivo, and induced a higher

neutralizing antibody and T cell immune responses in mice. Thus, we found that inhibition of viral

m6A methylation is a novel approach to enhance both innate and adaptive immune responses of

RSV vaccine candidates.

6.2. Future directions

6.2.1. To determine the mechanism of mRNA cap methylation in RSV

In Chapters 2 and 3, we have identified mutations, deletions, and insertions in the MTase region

of RSV L protein, which led to attenuation of RSV. Presumably, these L mutants affect viral

mRNA cap MTases. For L deletion or insertion mutaznts in the hinge region between the mRNA

238

capping and methyltransferase regions, it may also affect mRNA capping. However, it has been a challenge to purify these L mutants to conduct biochemical capping and MTase assays. One future direction is to express L fragments containing capping (CR V) and MTase (CR VI) domains and to screen a functional L fragment for capping and MTase assays.

6.2.2. To precisely locate m6A on RSV RNA

In this work, we have mapped m6A on RSV RNA by high-throughput m6A sequencing. However, this method allows us to locate putative m6A sites in specific regions in RSV RNA thus has a

limited resolution. Mapping the exact locations of m6A methylation in viral RNA is of great value

for understanding the dynamics and functions of m6A methylation in viral replication and gene

expression. Recently, an advanced m6A sequencing method, allowing us to map the m6A to a single nucleotide level, has been published by Zhang et al. [238]. In the future, we will adopt this new method to precisely locate m6A in RSV RNA.

6.2.3. To define the functions of m6A in innate immunity of viral infection in vivo

In Chapter 5, we have revealed the function of m6A in immune response of RSV in cell culture.

However, it is unknown whether m6A-deficient RSV RNA is capable of inducing a higher innate immunity in vivo. One future experiment is to directly deliver the m6A-deficient RSV RNA into the lungs of mice or cotton rats and determine the IFN responses. Another experiment is to investigate whether m6A-deficient RSVs can induce higher innate immune responses in animals.

6.2.4. To determine roles of m6A in innate and adaptive immunity of other NNS RNA viruses

Although we have demonstrated that m6A-deficient virion RNA of SeV and MeV can trigger higher IFN responses in RNA-transfected cells, whether it holds true for viral infection in cells

and animals is unknown. Future experiments are to use advanced m6A-seq techniques to map the

239

m6A methylation to a single nucleotide resolution in viral RNA. We can use a similar strategy as

we used for RSV to obtain m6A-deficient mutants of SeV and MeV. Then, we will be able to define

the functions of viral m6A on other NNS RNA viruses and to further test our hypothesis that

utilizing m6A as a target for development of antivirals or vaccines for other NNS RNA viruses.

6.2.5. To determine the mechanisms of m6A regulating RSV infections

How exactly m6A affects the RSV infection remains elusive. From the virus aspect, Chapter 5 has

provided one possible explanation that m6A inhibits the binding and activation of RIG-I following

recognition of viral RNA thus abrogating the IFN production. However, there may be other

possible mechanisms. For example, m6A methylation on RSV mRNAs may enhance their stability and translation, which is similar to the functions of mRNA cap methylation. Alternatively, from the aspect of host, it was reported that RSV strongly up-regulated expression of a membrane receptor protein TRAIL and strongly sensitized cells to apoptosis induced by exogenous TRAIL

[239]. Thus, apoptotic death of infected cells may be a mechanism for reducing RSV replication.

Recent works on m6A writer protein METTL3 revealed that METTL3 is involved in apoptosis by

interfering with many apoptosis-related proteins [240]. Therefore, a future direction is to determine

the mechanisms how host m6A-related pathways affect RSV replication, gene expression, and

pathogenesis.

6.2.6. To determine whether combination of different mutations can offer synergetic effects

and lead to the development of an idea live attenuated vaccine

An ideal live attenuated RSV vaccine should be genetically stable, safe (sufficiently attenuated),

and efficacious (highly immunogenic and protective), grows to high titer, and induces high innate

immunity. In my thesis work, I developed a variety of mutations that may play different roles

240

toward the development of an ideal live attenuated vaccine. Mutations in the SAM binding site

can serve as a means to attenuate RSV whereas deletion and insertion can enhance the genetic

stability of the virus. The m6A mutations can serve as a novel approach to enhance the innate

immune response of RSV. One way is to combine the m6A mutations with the SAM binding site

mutation or the deletion and insertion mutations, which may further optimize the RSV vaccine

candidates.

In Chapters 4 and 5, we have generated several m6A-deficient RSVs that were capable of triggering similar or even higher neutralizing antibodies and RSV-specific T cell response in mice compared to the parental RSV. These results indicated that it could be achievable to generate a recombinant

RSV with a higher immunogenicity than wild type RSV. However, those m6A-deficient RSVs may not be sufficiently attenuated as live vaccine candidates. Therefore, the combination of m6A

mutations with other mutations (such as SAM binding site, deletion, or insertion) could be a

promising approach to achieve an ideal balance between attenuation and immunogenicity for live

attenuated RSV vaccines. In addition, it is well known that NS1 and NS2 proteins inhibit the IFN

production after RSV infections. Therefore, we can combine m6A mutations with NS1 and NS2

mutations to further enhance innate immune responses which may improve the immunogenicity.

241

References

1. Blount, R.E., Jr., J.A. Morris, and R.E. Savage, Recovery of cytopathogenic agent from chimpanzees with coryza. Proc Soc Exp Biol Med, 1956. 92(3): p. 544-9. 2. Chanock, R., B. Roizman, and R. Myers, Recovery from infants with respiratory illness of a virus related to chimpanzee coryza agent (CCA). I. Isolation, properties and characterization. Am J Hyg, 1957. 66(3): p. 281-90. 3. Hall, C.B. and R.G. Douglas, Jr., Modes of transmission of respiratory syncytial virus. J Pediatr, 1981. 99(1): p. 100-3. 4. Rima, B., et al., ICTV Virus Taxonomy Profile: Pneumoviridae. J Gen Virol, 2017. 98(12): p. 2912-2913. 5. Glezen, W.P., et al., Risk of primary infection and reinfection with respiratory syncytial virus. Am J Dis Child, 1986. 140(6): p. 543-6. 6. Nair, H., et al., Global burden of acute lower respiratory infections due to respiratory syncytial virus in young children: a systematic review and meta-analysis. Lancet, 2010. 375(9725): p. 1545-55. 7. Dubus, J.C., A.C. Bodiou, and V. Millet, [Respiratory allergy in children and passive smoking]. Arch Pediatr, 1999. 6 Suppl 1: p. 35S-38S. 8. Piedimonte, G. and M.K. Perez, Respiratory syncytial virus infection and bronchiolitis. Pediatr Rev, 2014. 35(12): p. 519-30. 9. Kulkarni, H., et al., Evidence of Respiratory Syncytial Virus Spread by Aerosol. Time to Revisit Infection Control Strategies? Am J Respir Crit Care Med, 2016. 194(3): p. 308-16. 10. Altamirano-Lagos, M.J., et al., Current Animal Models for Understanding the Pathology Caused by the Respiratory Syncytial Virus. Front Microbiol, 2019. 10: p. 873. 11. Hasegawa, K., et al., Trends in bronchiolitis hospitalizations in the United States, 2000- 2009. Pediatrics, 2013. 132(1): p. 28-36. 12. Parikh, K., M. Hall, and S.J. Teach, Bronchiolitis management before and after the AAP guidelines. Pediatrics, 2014. 133(1): p. -7. 13. Ke, Z., et al., The Morphology and Assembly of Respiratory Syncytial Virus Revealed by Cryo-Electron Tomography. Viruses, 2018. 10(8). 14. Collins, P.L., et al., Nucleotide sequences for the gene junctions of human respiratory syncytial virus reveal distinctive features of intergenic structure and gene order. Proc Natl Acad Sci U S A, 1986. 83(13): p. 4594-8. 15. Ling, Z., K.C. Tran, and M.N. Teng, Human respiratory syncytial virus nonstructural protein NS2 antagonizes the activation of beta interferon transcription by interacting with RIG-I. J Virol, 2009. 83(8): p. 3734-42.

242

16. Bitko, V., et al., Nonstructural proteins of respiratory syncytial virus suppress premature apoptosis by an NF-kappaB-dependent, interferon-independent mechanism and facilitate virus growth. J Virol, 2007. 81(4): p. 1786-95. 17. Bakker, S.E., et al., The respiratory syncytial virus nucleoprotein-RNA complex forms a left-handed helical nucleocapsid. J Gen Virol, 2013. 94(Pt 8): p. 1734-1738. 18. Oliveira, A.P., et al., Human respiratory syncytial virus N, P and M protein interactions in HEK-293T cells. Virus Res, 2013. 177(1): p. 108-12. 19. Villanueva, N., et al., The bulk of the phosphorylation of human respiratory syncytial virus phosphoprotein is not essential but modulates viral RNA transcription and replication. J Gen Virol, 2000. 81(Pt 1): p. 129-33. 20. Mitra, R., et al., The Human Respiratory Syncytial Virus Matrix Protein Is Required for Maturation of Viral Filaments. Journal of Virology, 2012. 86(8): p. 4432-4443. 21. Carter, S.D., et al., Direct visualization of the small hydrophobic protein of human respiratory syncytial virus reveals the structural basis for membrane permeability. FEBS Lett, 2010. 584(13): p. 2786-90. 22. McLellan, J.S., W.C. Ray, and M.E. Peeples, Structure and function of respiratory syncytial virus surface glycoproteins. Curr Top Microbiol Immunol, 2013. 372: p. 83-104. 23. Oshansky, C.M., et al., The host response and molecular pathogenesis associated with respiratory syncytial virus infection. Future Microbiol, 2009. 4(3): p. 279-97. 24. Harcourt, J., et al., Respiratory syncytial virus G protein and G protein CX3C motif adversely affect CX3CR1+ T cell responses. J Immunol, 2006. 176(3): p. 1600-8. 25. Funchal, G.A., et al., Respiratory syncytial virus fusion protein promotes TLR-4-dependent neutrophil extracellular trap formation by human neutrophils. PLoS One, 2015. 10(4): p. e0124082. 26. Jin, H., et al., Recombinant respiratory syncytial viruses with deletions in the NS1, NS2, SH, and M2-2 genes are attenuated in vitro and in vivo. Virology, 2000. 273(1): p. 210-8. 27. Fix, J., et al., The insertion of fluorescent proteins in a variable region of respiratory syncytial virus L polymerase results in fluorescent and functional enzymes but with reduced activities. Open Virol J, 2011. 5: p. 103-8. 28. Paesen, G.C., et al., X-ray structure and activities of an essential Mononegavirales L- protein domain. Nat Commun, 2015. 6: p. 8749. 29. Chandrika, R., et al., Mutations in conserved domain I of the Sendai virus L polymerase protein uncouple transcription and replication. Virology, 1995. 213(2): p. 352-63. 30. Smallwood, S., et al., Mutations in conserved domain II of the large (L) subunit of the Sendai virus RNA polymerase abolish RNA synthesis. Virology, 1999. 262(2): p. 375-83. 31. Liang, B., et al., Structure of the L Protein of Vesicular Stomatitis Virus from Electron Cryomicroscopy. Cell, 2015. 162(2): p. 314-327. 32. Gilman, M.S.A., et al., Structure of the Respiratory Syncytial Virus Polymerase Complex. Cell, 2019. 179(1): p. 193-204 e14. 33. Cao, D., et al., Cryo-EM structure of the respiratory syncytial virus RNA polymerase. Nat Commun, 2020. 11(1): p. 368. 34. Battles, M.B. and J.S. McLellan, Respiratory syncytial virus entry and how to block it. Nature Reviews Microbiology, 2019. 17(4): p. 233-245.

243

35. Malhotra, R., et al., Isolation and characterisation of potential respiratory syncytial virus receptor(s) on epithelial cells. Microbes Infect, 2003. 5(2): p. 123-33. 36. Johnson, S.M., et al., Respiratory Syncytial Virus Uses CX3CR1 as a Receptor on Primary Human Airway Epithelial Cultures. Plos Pathogens, 2015. 11(12). 37. Johnson, T.R., J.S. McLellan, and B.S. Graham, Respiratory syncytial virus glycoprotein G interacts with DC-SIGN and L-SIGN to activate ERK1 and ERK2. J Virol, 2012. 86(3): p. 1339-47. 38. Cagno, V., et al., Heparan Sulfate Proteoglycans and Viral Attachment: True Receptors or Adaptation Bias? Viruses, 2019. 11(7). 39. Currier, M.G., et al., EGFR Interacts with the Fusion Protein of Respiratory Syncytial Virus Strain 2-20 and Mediates Infection and Mucin Expression. PLoS Pathog, 2016. 12(5): p. e1005622. 40. Rallabhandi, P., et al., Respiratory syncytial virus fusion protein-induced toll-like receptor 4 (TLR4) signaling is inhibited by the TLR4 antagonists Rhodobacter sphaeroides lipopolysaccharide and eritoran (E5564) and requires direct interaction with MD-2. mBio, 2012. 3(4). 41. Behera, A.K., et al., Blocking intercellular adhesion molecule-1 on human epithelial cells decreases respiratory syncytial virus infection. Biochem Biophys Res Commun, 2001. 280(1): p. 188-95. 42. Tayyari, F., et al., Identification of nucleolin as a cellular receptor for human respiratory syncytial virus. Nature Medicine, 2011. 17(9): p. 1132-U140. 43. Hallak, L.K., et al., Glycosaminoglycan sulfation requirements for respiratory syncytial virus infection. J Virol, 2000. 74(22): p. 10508-13. 44. Aksoy, M.O., et al., CXCR3 surface expression in human airway epithelial cells: cell cycle dependence and effect on cell proliferation. Am J Physiol Lung Cell Mol Physiol, 2006. 290(5): p. L909-18. 45. Techaarpornkul, S., P.L. Collins, and M.E. Peeples, Respiratory syncytial virus with the fusion protein as its only viral glycoprotein is less dependent on cellular glycosaminoglycans for attachment than complete virus. Virology, 2002. 294(2): p. 296- 304. 46. Krzyzaniak, M.A., et al., Host cell entry of respiratory syncytial virus involves macropinocytosis followed by proteolytic activation of the F protein. PLoS Pathog, 2013. 9(4): p. e1003309. 47. Griffiths, C., S.J. Drews, and D.J. Marchant, Respiratory Syncytial Virus: Infection, Detection, and New Options for Prevention and Treatment. Clin Microbiol Rev, 2017. 30(1): p. 277-319. 48. Tremaglio, C.Z., et al., Respiratory Syncytial Virus Polymerase Can Initiate Transcription from Position 3 of the Leader Promoter. Journal of Virology, 2013. 87(6): p. 3196-3207. 49. Kuo, L., R. Fearns, and P.L. Collins, Analysis of the gene start and gene end signals of human respiratory syncytial virus: Quasi-templated initiation at position 1 of the encoded mRNA. Journal of Virology, 1997. 71(7): p. 4944-4953. 50. McGivern, D.R., P.L. Collins, and R. Fearns, Identification of internal sequences in the 3' leader region of human respiratory syncytial virus that enhance transcription and confer replication processivity. J Virol, 2005. 79(4): p. 2449-60. 244

51. Rincheval, V., et al., Functional organization of cytoplasmic inclusion bodies in cells infected by respiratory syncytial virus. Nat Commun, 2017. 8(1): p. 563. 52. Lifland, A.W., et al., Human respiratory syncytial virus nucleoprotein and inclusion bodies antagonize the innate immune response mediated by MDA5 and MAVS. J Virol, 2012. 86(15): p. 8245-58. 53. Shahriari, S., J. Gordon, and R. Ghildyal, Host cytoskeleton in respiratory syncytial virus assembly and budding. Virol J, 2016. 13(1): p. 161. 54. El Najjar, F., A.P. Schmitt, and R.E. Dutch, Paramyxovirus glycoprotein incorporation, assembly and budding: a three way dance for infectious particle production. Viruses, 2014. 6(8): p. 3019-54. 55. Ramanathan, A., G.B. Robb, and S.H. Chan, mRNA capping: biological functions and applications. Nucleic Acids Res, 2016. 44(16): p. 7511-26. 56. Ogino, T. and A.K. Banerjee, Unconventional mechanism of mRNA capping by the RNA- dependent RNA polymerase of vesicular stomatitis virus. Mol Cell, 2007. 25(1): p. 85-97. 57. Rahmeh, A.A., et al., Ribose 2'-O methylation of the vesicular stomatitis virus mRNA cap precedes and facilitates subsequent guanine-N-7 methylation by the large polymerase protein. J Virol, 2009. 83(21): p. 11043-50. 58. Both, G.W., A.K. Banerjee, and A.J. Shatkin, Methylation-dependent translation of viral messenger RNAs in vitro. Proc Natl Acad Sci U S A, 1975. 72(3): p. 1189-93. 59. Ray, D., et al., West Nile virus 5'-cap structure is formed by sequential guanine N-7 and ribose 2'-O methylations by nonstructural protein 5. J Virol, 2006. 80(17): p. 8362-70. 60. Daffis, S., et al., 2'-O methylation of the viral mRNA cap evades host restriction by IFIT family members. Nature, 2010. 468(7322): p. 452-6. 61. Zust, R., et al., Ribose 2'-O-methylation provides a molecular signature for the distinction of self and non-self mRNA dependent on the RNA sensor Mda5. Nat Immunol, 2011. 12(2): p. 137-43. 62. Ma, Y., et al., mRNA cap methylation influences pathogenesis of vesicular stomatitis virus in vivo. J Virol, 2014. 88(5): p. 2913-26. 63. Ogino, T., et al., Sendai virus RNA-dependent RNA polymerase L protein catalyzes cap methylation of virus-specific mRNA. J Biol Chem, 2005. 280(6): p. 4429-35. 64. Zhang, Y., et al., Rational design of human metapneumovirus live attenuated vaccine candidates by inhibiting viral mRNA cap methyltransferase. J Virol, 2014. 88(19): p. 11411-29. 65. Li, S.H., et al., Rational design of a flavivirus vaccine by abolishing viral RNA 2'-O methylation. J Virol, 2013. 87(10): p. 5812-9. 66. Sen, G.C., Viruses and interferons. Annu Rev Microbiol, 2001. 55: p. 255-81. 67. Samuel, C.E., Antiviral actions of interferons. Clin Microbiol Rev, 2001. 14(4): p. 778- 809, table of contents. 68. Mogensen, T.H., Pathogen recognition and inflammatory signaling in innate immune defenses. Clin Microbiol Rev, 2009. 22(2): p. 240-73, Table of Contents. 69. Kato, H., et al., Cell type-specific involvement of RIG-I in antiviral response. Immunity, 2005. 23(1): p. 19-28. 70. Schroder, M. and A.G. Bowie, TLR3 in antiviral immunity: key player or bystander? Trends Immunol, 2005. 26(9): p. 462-8. 245

71. Rifkin, I.R., et al., Toll-like receptors, endogenous ligands, and systemic autoimmune disease. Immunol Rev, 2005. 204: p. 27-42. 72. Kim, T.H. and H.K. Lee, Innate immune recognition of respiratory syncytial virus infection. BMB Rep, 2014. 47(4): p. 184-91. 73. Lambert, L., et al., Immunity to RSV in Early-Life. Front Immunol, 2014. 5: p. 466. 74. Monick, M.M., et al., Respiratory syncytial virus synergizes with Th2 cytokines to induce optimal levels of TARC/CCL17. J Immunol, 2007. 179(3): p. 1648-58. 75. Haynes, L.M., et al., Involvement of toll-like receptor 4 in innate immunity to respiratory syncytial virus. J Virol, 2001. 75(22): p. 10730-7. 76. Lund, J.M., et al., Recognition of single-stranded RNA viruses by Toll-like receptor 7. Proc Natl Acad Sci U S A, 2004. 101(15): p. 5598-603. 77. Wang, Y.L., et al., Structural and functional insights into 5 '-ppp RNA pattern recognition by the innate immune receptor RIG-I. Nature Structural & Molecular Biology, 2010. 17(7): p. 781-U19. 78. Lo, M.S., R.M. Brazas, and M.J. Holtzman, Respiratory syncytial virus nonstructural proteins NS1 and NS2 mediate inhibition of Stat2 expression and alpha/beta interferon responsiveness. J Virol, 2005. 79(14): p. 9315-9. 79. Oshansky, C.M., et al., Respiratory syncytial virus proteins modulate suppressors of cytokine signaling 1 and 3 and the type I interferon response to infection by a toll-like receptor pathway. Viral Immunol, 2009. 22(3): p. 147-61. 80. Hu, J. and J.L. Robinson, Treatment of respiratory syncytial virus with palivizumab: a systematic review. World J Pediatr, 2010. 6(4): p. 296-300. 81. Hall, C.B., et al., Use of Ribavirin in the Treatment of Respiratory Syncytial Virus- Infection. Pediatrics, 1993. 92(3): p. 501-504. 82. Stein, D.S., et al., Oral ribavirin treatment of influenza A and B. Antimicrob Agents Chemother, 1987. 31(8): p. 1285-7. 83. Casey, J., et al., Oral ribavirin for treatment of respiratory syncitial virus and parainfluenza 3 virus infections post allogeneic haematopoietic stem cell transplantation. Bone Marrow Transplant, 2013. 48(12): p. 1558-61. 84. Gavin, P.J. and B.Z. Katz, Intravenous ribavirin treatment for severe adenovirus disease in immunocompromised children. Pediatrics, 2002. 110(1 Pt 1): p. e9. 85. Pal, G., Effects of ribavirin on measles. J Indian Med Assoc, 2011. 109(9): p. 666-7. 86. Eberhardt, K.A., et al., Ribavirin for the treatment of Lassa fever: A systematic review and meta-analysis. Int J Infect Dis, 2019. 87: p. 15-20. 87. Moreli, M.L., et al., Effectiveness of the ribavirin in treatment of hantavirus infections in the Americas and Eurasia: a meta-analysis. Virusdisease, 2014. 25(3): p. 385-9. 88. Simoes, E.A., et al., Challenges and opportunities in developing respiratory syncytial virus therapeutics. J Infect Dis, 2015. 211 Suppl 1: p. S1-S20. 89. Acosta, P.L., M.T. Caballero, and F.P. Polack, Brief History and Characterization of Enhanced Respiratory Syncytial Virus Disease. Clin Vaccine Immunol, 2015. 23(3): p. 189-95. 90. Yim, K.C., et al., Human metapneumovirus: enhanced pulmonary disease in cotton rats immunized with formalin-inactivated virus vaccine and challenged. Vaccine, 2007. 25(27): p. 5034-40. 246

91. Ottolini, M.G., et al., Enhanced pulmonary pathology in cotton rats upon challenge after immunization with inactivated parainfluenza virus 3 vaccines. Viral Immunology, 2000. 13(2): p. 231-236. 92. Friedewald, W.T., et al., Low-temperature-grown RS virus in adult volunteers. JAMA, 1968. 204(8): p. 690-4. 93. Kim, H.W., et al., Clinical and immunological response of infants and children to administration of low-temperature adapted respiratory syncytial virus. Pediatrics, 1971. 48(5): p. 745-55. 94. Collins, P.L., et al., Production of infectious human respiratory syncytial virus from cloned cDNA confirms an essential role for the transcription elongation factor from the 5' proximal open reading frame of the M2 mRNA in gene expression and provides a capability for vaccine development. Proc Natl Acad Sci U S A, 1995. 92(25): p. 11563-7. 95. Karron, R.A., U.J. Buchholz, and P.L. Collins, Live-attenuated respiratory syncytial virus vaccines. Curr Top Microbiol Immunol, 2013. 372: p. 259-84. 96. Teng, M.N., et al., Recombinant respiratory syncytial virus that does not express the NS1 or M2-2 protein is highly attenuated and immunogenic in chimpanzees. Journal of Virology, 2000. 74(19): p. 9317-9321. 97. Whitehead, S.S., et al., Recombinant respiratory syncytial virus bearing a deletion of either the NS2 or SH gene is attenuated in chimpanzees. Journal of Virology, 1999. 73(4): p. 3438-3442. 98. Spann, K.M., K.C. Tran, and P.L. Collins, Effects of nonstructural proteins NS1 and NS2 of human respiratory syncytial virus on interferon regulatory factor 3, NF-kappaB, and proinflammatory cytokines. J Virol, 2005. 79(9): p. 5353-62. 99. Belshe, R.B., E.L. Anderson, and E.E. Walsh, Immunogenicity of purified F glycoprotein of respiratory syncytial virus: clinical and immune responses to subsequent natural infection in children. J Infect Dis, 1993. 168(4): p. 1024-9. 100. Langley, J.M., et al., A dose-ranging study of a subunit Respiratory Syncytial Virus subtype A vaccine with and without aluminum phosphate adjuvantation in adults > or =65 years of age. Vaccine, 2009. 27(42): p. 5913-9. 101. Power, U.F., et al., The immunogenicity, protective efficacy and safety of BBG2Na, a subunit respiratory syncytial virus (RSV) vaccine candidate, against RSV-B. Vaccine, 2003. 22(2): p. 168-76. 102. Murawski, M.R., et al., Newcastle disease virus-like particles containing respiratory syncytial virus G protein induced protection in BALB/c mice, with no evidence of immunopathology. J Virol, 2010. 84(2): p. 1110-23. 103. Kim, A.R., et al., Protection induced by virus-like particle vaccine containing tandem repeat gene of respiratory syncytial virus G protein. PLoS One, 2018. 13(1): p. e0191277. 104. Li, X., et al., Protection against respiratory syncytial virus infection by DNA immunization. J Exp Med, 1998. 188(4): p. 681-8. 105. Smith, T.R.F., et al., Development of an intradermal DNA vaccine delivery strategy to achieve single-dose immunity against respiratory syncytial virus. Vaccine, 2017. 35(21): p. 2840-2847. 106. Geall, A.J., et al., Nonviral delivery of self-amplifying RNA vaccines. Proc Natl Acad Sci U S A, 2012. 109(36): p. 14604-9. 247

107. Cano, F., et al., Partial protection to respiratory syncytial virus (RSV) elicited in mice by intranasal immunization using live staphylococci with surface-displayed RSV-peptides. Vaccine, 2000. 18(24): p. 2743-52. 108. Rezaee, F., et al., Ongoing developments in RSV prophylaxis: a clinician's analysis. Curr Opin Virol, 2017. 24: p. 70-78. 109. Widjojoatmodjo, M.N., et al., Recombinant low-seroprevalent adenoviral vectors Ad26 and Ad35 expressing the respiratory syncytial virus (RSV) fusion protein induce protective immunity against RSV infection in cotton rats. Vaccine, 2015. 33(41): p. 5406-5414. 110. Samy, N., et al., Safety and immunogenicity of novel modified vaccinia Ankara-vectored RSV vaccine: A randomized phase I clinical trial. Vaccine, 2020. 38(11): p. 2608-2619. 111. Rana, A.K. and S. Ankri, Reviving the RNAWorld: An Insight into the Appearance of RNA Methyltransferases. Frontiers in Genetics, 2016. 7. 112. Covelo-Molares, H., M. Bartosovic, and S. Vanacova, RNA methylation in nuclear pre- mRNA processing. Wiley Interdiscip Rev RNA, 2018. 9(6): p. e1489. 113. Desrosiers, R., K. Friderici, and F. Rottman, Identification of methylated nucleosides in messenger RNA from Novikoff hepatoma cells. Proc Natl Acad Sci U S A, 1974. 71(10): p. 3971-5. 114. Dominissini, D., et al., Transcriptome-wide mapping of N(6)-methyladenosine by m(6)A- seq based on immunocapturing and massively parallel sequencing. Nat Protoc, 2013. 8(1): p. 176-89. 115. Yang, Y., et al., Dynamic transcriptomic m(6)A decoration: writers, erasers, readers and functions in RNA metabolism. Cell Res, 2018. 28(6): p. 616-624. 116. Balacco, D.L. and M. Soller, The m(6)A Writer: Rise of a Machine for Growing Tasks. Biochemistry, 2019. 58(5): p. 363-378. 117. Lin, S., et al., The m(6)A Methyltransferase METTL3 Promotes Translation in Human Cancer Cells. Mol Cell, 2016. 62(3): p. 335-345. 118. Wang, H., et al., Mettl3-mediated mRNA m(6)A methylation promotes dendritic cell activation. Nat Commun, 2019. 10(1): p. 1898. 119. Geula, S., et al., Stem cells. m6A mRNA methylation facilitates resolution of naive pluripotency toward differentiation. Science, 2015. 347(6225): p. 1002-6. 120. Winkler, R., et al., m(6)A modification controls the innate immune response to infection by targeting type I interferons. Nat Immunol, 2019. 20(2): p. 173-182. 121. Meng, T.G., et al., Mettl14 is required for mouse postimplantation development by facilitating epiblast maturation. FASEB J, 2019. 33(1): p. 1179-1187. 122. Jia, G., et al., N6-methyladenosine in nuclear RNA is a major substrate of the obesity- associated FTO. Nat Chem Biol, 2011. 7(12): p. 885-7. 123. Zheng, G., et al., ALKBH5 is a mammalian RNA demethylase that impacts RNA metabolism and mouse fertility. Mol Cell, 2013. 49(1): p. 18-29. 124. Niu, Y., et al., RNA N6-methyladenosine demethylase FTO promotes breast tumor progression through inhibiting BNIP3. Mol Cancer, 2019. 18(1): p. 46. 125. Rubio, R.M., et al., RNA m(6) A modification enzymes shape innate responses to DNA by regulating interferon beta. Genes Dev, 2018. 32(23-24): p. 1472-1484. 126. Xiao, W., et al., Nuclear m(6)A Reader YTHDC1 Regulates mRNA Splicing. Mol Cell, 2016. 61(4): p. 507-519. 248

127. Roundtree, I.A., et al., YTHDC1 mediates nuclear export of N(6)-methyladenosine methylated mRNAs. Elife, 2017. 6. 128. Alarcon, C.R., et al., HNRNPA2B1 Is a Mediator of m(6)A-Dependent Nuclear RNA Processing Events. Cell, 2015. 162(6): p. 1299-308. 129. Wu, B., et al., Molecular basis for the specific and multivariant recognitions of RNA substrates by human hnRNP A2/B1. Nat Commun, 2018. 9(1): p. 420. 130. Wang, X., et al., N(6)-methyladenosine Modulates Messenger RNA Translation Efficiency. Cell, 2015. 161(6): p. 1388-99. 131. Du, H., et al., YTHDF2 destabilizes m(6)A-containing RNA through direct recruitment of the CCR4-NOT deadenylase complex. Nat Commun, 2016. 7: p. 12626. 132. Li, A., et al., Cytoplasmic m(6)A reader YTHDF3 promotes mRNA translation. Cell Res, 2017. 27(3): p. 444-447. 133. Shi, H., et al., YTHDF3 facilitates translation and decay of N(6)-methyladenosine-modified RNA. Cell Res, 2017. 27(3): p. 315-328. 134. Soh, Y.Q.S., et al., Meioc maintains an extended meiotic prophase I in mice. PLoS Genet, 2017. 13(4): p. e1006704. 135. Patil, D.P., B.F. Pickering, and S.R. Jaffrey, Reading m(6)A in the Transcriptome: m(6)A- Binding Proteins. Trends Cell Biol, 2018. 28(2): p. 113-127. 136. Lavi, S. and A.J. Shatkin, Methylated simian virus 40-specific RNA from nuclei and cytoplasm of infected BSC-1 cells. Proc Natl Acad Sci U S A, 1975. 72(6): p. 2012-6. 137. Sommer, S., et al., The methylation of adenovirus-specific nuclear and cytoplasmic RNA. Nucleic Acids Res, 1976. 3(3): p. 749-65. 138. Moss, B., et al., 5'-Terminal and internal methylated nucleosides in herpes simplex virus type 1 mRNA. J Virol, 1977. 23(2): p. 234-9. 139. Kane, S.E. and K. Beemon, Precise localization of m6A in Rous sarcoma virus RNA reveals clustering of methylation sites: implications for RNA processing. Mol Cell Biol, 1985. 5(9): p. 2298-306. 140. Narayan, P., et al., Unequal distribution of N6-methyladenosine in influenza virus mRNAs. Mol Cell Biol, 1987. 7(4): p. 1572-5. 141. Tirumuru, N., et al., N(6)-methyladenosine of HIV-1 RNA regulates viral infection and HIV-1 Gag protein expression. Elife, 2016. 5. 142. Lichinchi, G., et al., Dynamics of the human and viral m(6)A RNA methylomes during HIV- 1 infection of T cells. Nat Microbiol, 2016. 1: p. 16011. 143. Kennedy, E.M., et al., Posttranscriptional m(6)A Editing of HIV-1 mRNAs Enhances Viral Gene Expression. Cell Host Microbe, 2017. 22(6): p. 830. 144. Hao, H., et al., N6-methyladenosine modification and METTL3 modulate enterovirus 71 replication. Nucleic Acids Res, 2019. 47(1): p. 362-374. 145. Courtney, D.G., et al., Epitranscriptomic Enhancement of Gene Expression and Replication. Cell Host Microbe, 2017. 22(3): p. 377-386 e5. 146. Ye, F., E.R. Chen, and T.W. Nilsen, Kaposi's Sarcoma-Associated Herpesvirus Utilizes and Manipulates RNA N(6)-Adenosine Methylation To Promote Lytic Replication. J Virol, 2017. 91(16).

249

147. Hesser, C.R., et al., N6-methyladenosine modification and the YTHDF2 reader protein play cell type specific roles in lytic viral gene expression during Kaposi's sarcoma- associated herpesvirus infection. PLoS Pathog, 2018. 14(4): p. e1006995. 148. Imam, H., et al., N6-methyladenosine modification of RNA differentially regulates the viral life cycle. Proc Natl Acad Sci U S A, 2018. 115(35): p. 8829-8834. 149. Gokhale, N.S., et al., N6-Methyladenosine in Flaviviridae Viral RNA Genomes Regulates Infection. Cell Host Microbe, 2016. 20(5): p. 654-665. 150. Lichinchi, G., et al., Dynamics of Human and Viral RNA Methylation during Zika Virus Infection. Cell Host Microbe, 2016. 20(5): p. 666-673. 151. Tsai, K., D.G. Courtney, and B.R. Cullen, Addition of m6A to SV40 late mRNAs enhances viral structural gene expression and replication. PLoS Pathog, 2018. 14(2): p. e1006919. 152. Lang, F., et al., EBV epitranscriptome reprogramming by METTL14 is critical for viral- associated tumorigenesis. PLoS Pathog, 2019. 15(6): p. e1007796. 153. Lu, W., et al., N(6)-Methyladenosine-binding proteins suppress HIV-1 infectivity and viral production. J Biol Chem, 2018. 293(34): p. 12992-13005. 154. Tan, B., et al., Viral and cellular N(6)-methyladenosine and N(6),2'-O-dimethyladenosine epitranscriptomes in the KSHV life cycle. Nat Microbiol, 2018. 3(1): p. 108-120. 155. Buchholz, U.J., et al., Live Respiratory Syncytial Virus (RSV) Vaccine Candidate Containing Stabilized Temperature-Sensitivity Mutations Is Highly Attenuated in RSV- Seronegative Infants and Children. J Infect Dis, 2018. 217(9): p. 1338-1346. 156. McFarland, E.J., et al., Live-Attenuated Respiratory Syncytial Virus Vaccine Candidate With Deletion of RNA Synthesis Regulatory Protein M2-2 is Highly Immunogenic in Children. J Infect Dis, 2018. 217(9): p. 1347-1355. 157. Karron, R.A., et al., A gene deletion that up-regulates viral gene expression yields an attenuated RSV vaccine with improved antibody responses in children. Sci Transl Med, 2015. 7(312): p. 312ra175. 158. Abbas, Y.M., et al., Structure of human IFIT1 with capped RNA reveals adaptable mRNA binding and mechanisms for sensing N1 and N2 ribose 2'-O methylations. Proc Natl Acad Sci U S A, 2017. 114(11): p. E2106-E2115. 159. Li, J., E.C. Fontaine-Rodriguez, and S.P. Whelan, Amino acid residues within conserved domain VI of the vesicular stomatitis virus large polymerase protein essential for mRNA cap methyltransferase activity. J Virol, 2005. 79(21): p. 13373-84. 160. Grdzelishvili, V.Z., et al., A single amino acid change in the L-polymerase protein of vesicular stomatitis virus completely abolishes viral mRNA cap methylation. J Virol, 2005. 79(12): p. 7327-37. 161. Li, J., J.T. Wang, and S.P. Whelan, A unique strategy for mRNA cap methylation used by vesicular stomatitis virus. Proc Natl Acad Sci U S A, 2006. 103(22): p. 8493-8. 162. Sun, J., et al., Methyltransferase-defective avian metapneumovirus vaccines provide complete protection against challenge with the homologous Colorado strain and the heterologous Minnesota strain. J Virol, 2014. 88(21): p. 12348-63. 163. Wang, Y., et al., Enhancement of safety and immunogenicity of the Chinese Hu191 measles virus vaccine by alteration of the S-adenosylmethionine (SAM) binding site in the large polymerase protein. Virology, 2018. 518: p. 210-220.

250

164. Johnson, S.M., et al., Respiratory Syncytial Virus Uses CX3CR1 as a Receptor on Primary Human Airway Epithelial Cultures. PLoS Pathog, 2015. 11(12): p. e1005318. 165. Buchholz, U.J., et al., Live vaccines for human metapneumovirus designed by reverse genetics. Expert Rev Vaccines, 2006. 5(5): p. 695-706. 166. Collins, P.L. and B.S. Graham, Viral and host factors in human respiratory syncytial virus pathogenesis. J Virol, 2008. 82(5): p. 2040-55. 167. Teng, M.N., et al., Recombinant respiratory syncytial virus that does not express the NS1 or M2-2 protein is highly attenuated and immunogenic in chimpanzees. J Virol, 2000. 74(19): p. 9317-21. 168. Littauer, E.Q., et al., Stable incorporation of GM-CSF into dissolvable microneedle patch improves skin vaccination against influenza. J Control Release, 2018. 276: p. 1-16. 169. Krakowski, M., et al., Granulocyte macrophage-colony stimulating factor (GM-CSF) recruits immune cells to the pancreas and delays STZ-induced diabetes. J Pathol, 2002. 196(1): p. 103-12. 170. Pashine, A., N.M. Valiante, and J.B. Ulmer, Targeting the innate immune response with improved vaccine adjuvants. Nat Med, 2005. 11(4 Suppl): p. S63-8. 171. Pulendran, B., Division of labor and cooperation between dendritic cells. Nat Immunol, 2006. 7(7): p. 699-700. 172. Fallet, B., et al., Interferon-driven deletion of antiviral B cells at the onset of chronic infection. Sci Immunol, 2016. 1(4). 173. Moseman, E.A., et al., Type I interferon suppresses virus-specific B cell responses by modulating CD8(+) T cell differentiation. Sci Immunol, 2016. 1(4). 174. Cowton, V.M., D.R. McGivern, and R. Fearns, Unravelling the complexities of respiratory syncytial virus RNA synthesis. J Gen Virol, 2006. 87(Pt 7): p. 1805-21. 175. Whelan, S.P., J.N. Barr, and G.W. Wertz, Transcription and replication of nonsegmented negative-strand RNA viruses. Curr Top Microbiol Immunol, 2004. 283: p. 61-119. 176. Duprex, W.P., F.M. Collins, and B.K. Rima, Modulating the function of the measles virus RNA-dependent RNA polymerase by insertion of green fluorescent protein into the open reading frame. J Virol, 2002. 76(14): p. 7322-8. 177. Brown, D.D., et al., Rational attenuation of a by modulating the activity of the RNA-dependent RNA polymerase. J Virol, 2005. 79(22): p. 14330-8. 178. Silin, D., et al., Development of a challenge-protective vaccine concept by modification of the viral RNA-dependent RNA polymerase of canine distemper virus. J Virol, 2007. 81(24): p. 13649-58. 179. Ruedas, J.B. and J. Perrault, Insertion of enhanced green fluorescent protein in a hinge region of vesicular stomatitis virus L polymerase protein creates a temperature-sensitive virus that displays no virion-associated polymerase activity in vitro. J Virol, 2009. 83(23): p. 12241-52. 180. Delgado, M.F., et al., Lack of antibody affinity maturation due to poor Toll-like receptor stimulation leads to enhanced respiratory syncytial virus disease. Nat Med, 2009. 15(1): p. 34-41. 181. Shaw, C.A., et al., The role of non-viral antigens in the cotton rat model of respiratory syncytial virus vaccine-enhanced disease. Vaccine, 2013. 31(2): p. 306-12.

251

182. Grdzelishvili, V.Z., et al., Identification of a new region in the vesicular stomatitis virus L polymerase protein which is essential for mRNA cap methylation. Virology, 2006. 350(2): p. 394-405. 183. Li, J.R., E.C. Fontaine-Rodriguez, and S.P.J. Whelan, Amino acid residues within conserved domain VI of the vesicular stomatitis virus large polymerase protein essential for mRNA cap methyltransferase activity. Journal of Virology, 2005. 79(21): p. 13373- 13384. 184. Li, J.R., J.T. Wang, and S.P.J. Whelan, A unique strategy for mRNA cap methylation used by vesicular stomatitis virus. Proceedings of the National Academy of Sciences of the United States of America, 2006. 103(22): p. 8493-8498. 185. Crowe, J.E., et al., A Further Attenuated Derivative of a Cold-Passaged Temperature- Sensitive Mutant of Human Respiratory Syncytial Virus Retains Immunogenicity and Protective Efficacy against Wild-Type Challenge in Seronegative Chimpanzees. Vaccine, 1994. 12(9): p. 783-790. 186. Crowe, J.E., et al., Satisfactorily Attenuated and Protective Mutants Derived from a Partially Attenuated Cold-Passaged Respiratory Syncytial Virus Mutant by Introduction of Additional Attenuating Mutations during Chemical Mutagenesis. Vaccine, 1994. 12(8): p. 691-699. 187. Crowe, J.E., et al., Cold Passaged, Temperature-Sensitive Mutants of Human Respiratory Syncytial Virus (Rsv) Are Highly Attenuated, Immunogenic, and Protective in Seronegative Chimpanzees, Even When Rsv Antibodies Are Infused Shortly before Immunization. Vaccine, 1995. 13(9): p. 847-855. 188. Karron, R.A., et al., Evaluation of two live, cold-passaged, temperature-sensitive respiratory syncytial virus vaccines in chimpanzees and in human adults, infants, and children. J Infect Dis, 1997. 176(6): p. 1428-36. 189. Wright, P.F., et al., Evaluation of a live, cold-passaged, temperature-sensitive, respiratory syncytial virus vaccine candidate in infancy. Journal of Infectious Diseases, 2000. 182(5): p. 1331-1342. 190. Wright, P.F., et al., Evaluation of a Live, Attenuated Respiratory Syncytial Virus-Vaccine in Infants. Journal of Pediatrics, 1976. 88(6): p. 931-936. 191. Connors, M., et al., A cold-passaged, attenuated strain of human respiratory syncytial virus contains mutations in the F and L genes. Virology, 1995. 208(2): p. 478-84. 192. Luongo, C., et al., Respiratory syncytial virus modified by deletions of the NS2 gene and amino acid S1313 of the L polymerase protein is a temperature-sensitive, live-attenuated vaccine candidate that is phenotypically stable at physiological temperature. J Virol, 2013. 87(4): p. 1985-96. 193. Tang, R.S., et al., Clustered charge-to-alanine mutagenesis of human respiratory syncytial virus L polymerase generates temperature-sensitive viruses. Virology, 2002. 302(1): p. 207-16. 194. Murphy, A.M. and V.Z. Grdzelishvili, Identification of sendai virus L protein amino acid residues affecting viral mRNA cap methylation. J Virol, 2009. 83(4): p. 1669-81. 195. Karron, R.A., et al., Identification of a recombinant live attenuated respiratory syncytial virus vaccine candidate that is highly attenuated in infants. J Infect Dis, 2005. 191(7): p. 1093-104. 252

196. Lin, Y.H., et al., Genetic stability determinants of temperature sensitive, live attenuated respiratory syncytial virus vaccine candidates. Virus Res, 2006. 115(1): p. 9-15. 197. Luongo, C., et al., Increased genetic and phenotypic stability of a promising live-attenuated respiratory syncytial virus vaccine candidate by reverse genetics. J Virol, 2012. 86(19): p. 10792-804. 198. Yue, Y., J. Liu, and C. He, RNA N6-methyladenosine methylation in post-transcriptional gene expression regulation. Genes Dev, 2015. 29(13): p. 1343-55. 199. Roundtree, I.A., et al., Dynamic RNA Modifications in Gene Expression Regulation. Cell, 2017. 169(7): p. 1187-1200. 200. Perry, R.P., et al., The methylated constituents of L cell messenger RNA: evidence for an unusual cluster at the 5' terminus. Cell, 1975. 4(4): p. 387-94. 201. Wei, C.M., A. Gershowitz, and B. Moss, Methylated nucleotides block 5' terminus of HeLa cell messenger RNA. Cell, 1975. 4(4): p. 379-86. 202. Darnell, R.B., S. Ke, and J.E. Darnell, Jr., Pre-mRNA processing includes N(6) methylation of adenosine residues that are retained in mRNA exons and the fallacy of "RNA epigenetics". RNA, 2018. 24(3): p. 262-267. 203. Meyer, K.D., et al., Comprehensive analysis of mRNA methylation reveals enrichment in 3' UTRs and near stop codons. Cell, 2012. 149(7): p. 1635-46. 204. Linder, B., et al., Single-nucleotide-resolution mapping of m6A and m6Am throughout the transcriptome. Nat Methods, 2015. 12(8): p. 767-72. 205. Meyer, K.D. and S.R. Jaffrey, The dynamic epitranscriptome: N6-methyladenosine and gene expression control. Nat Rev Mol Cell Biol, 2014. 15(5): p. 313-26. 206. Vu, L.P., et al., The N6-methyladenosine (m6A)-forming enzyme METTL3 controls myeloid differentiation of normal hematopoietic and leukemia cells. Nat Med, 2017. 207. Jaffrey, S.R. and M.G. Kharas, Emerging links between m6A and misregulated mRNA methylation in cancer. Genome Med, 2017. 9(1): p. 2. 208. Zhao, B.S., et al., m6A-dependent maternal mRNA clearance facilitates zebrafish maternal-to-zygotic transition. Nature, 2017. 542(7642): p. 475-478. 209. Canaani, D., et al., Identification and mapping of N6-methyladenosine containing sequences in simian virus 40 RNA. Nucleic Acids Res, 1979. 6(8): p. 2879-99. 210. Furuichi, Y., et al., Blocked, methylated 5'-terminal sequence in avian sarcoma virus RNA. Nature, 1975. 257(5527): p. 618-20. 211. Krug, R.M., M.A. Morgan, and A.J. Shatkin, Influenza viral mRNA contains internal N6- methyladenosine and 5'-terminal 7-methylguanosine in cap structures. J Virol, 1976. 20(1): p. 45-53. 212. Kennedy, E.M., et al., Posttranscriptional m(6)A Editing of HIV-1 mRNAs Enhances Viral Gene Expression. Cell Host Microbe, 2016. 19(5): p. 675-85. 213. Fearns, R., M.E. Peeples, and P.L. Collins, Increased expression of the N protein of respiratory syncytial virus stimulates minigenome replication but does not alter the balance between the synthesis of mRNA and antigenome. Virology, 1997. 236(1): p. 188- 201. 214. Ruigrok, R.W., T. Crepin, and D. Kolakofsky, Nucleoproteins and nucleocapsids of negative-strand RNA viruses. Curr Opin Microbiol, 2011. 14(4): p. 504-10.

253

215. Afonso, C.L., et al., Taxonomy of the order Mononegavirales: update 2016. Arch Virol, 2016. 161(8): p. 2351-60. 216. Dominissini, D., et al., Topology of the human and mouse m6A RNA methylomes revealed by m6A-seq. Nature, 2012. 485(7397): p. 201-6. 217. Kim, D., B. Langmead, and S.L. Salzberg, HISAT: a fast spliced aligner with low memory requirements. Nat Methods, 2015. 12(4): p. 357-60. 218. Ma, L., et al., Evolution of transcript modification by N6-methyladenosine in primates. Genome Res, 2017. 27(3): p. 385-392. 219. Cui, X., et al., Guitar: An R/Bioconductor Package for Gene Annotation Guided Transcriptomic Analysis of RNA-Related Genomic Features. Biomed Res Int, 2016. 2016: p. 8367534. 220. Liu, L., et al., QNB: differential RNA methylation analysis for count-based small-sample sequencing data with a quad-negative binomial model. BMC Bioinformatics, 2017. 18(1): p. 387. 221. Kolakofsky, D. and A. Bruschi, Antigenomes in Sendai virions and Sendai virus-infected cells. Virology, 1975. 66(1): p. 185-91. 222. Liu, J., et al., A METTL3-METTL14 complex mediates mammalian nuclear RNA N6- adenosine methylation. Nat Chem Biol, 2014. 10(2): p. 93-5. 223. Boyoglu-Barnum, S., et al., Mutating the CX3C Motif in the G Protein Should Make a Live Respiratory Syncytial Virus Vaccine Safer and More Effective. J Virol, 2017. 91(10). 224. Teng, M.N. and P.L. Collins, The central conserved cystine noose of the attachment G protein of human respiratory syncytial virus is not required for efficient viral infection in vitro or in vivo. J Virol, 2002. 76(12): p. 6164-71. 225. Teng, M.N., S.S. Whitehead, and P.L. Collins, Contribution of the respiratory syncytial virus G glycoprotein and its secreted and membrane-bound forms to virus replication in vitro and in vivo. Virology, 2001. 289(2): p. 283-96. 226. Vanover, D., et al., RSV glycoprotein and genomic RNA dynamics reveal filament assembly prior to the plasma membrane. Nat Commun, 2017. 8(1): p. 667. 227. Tripp, R.A., et al., CX3C chemokine mimicry by respiratory syncytial virus G glycoprotein. Nat Immunol, 2001. 2(8): p. 732-8. 228. Gumpper, R.H., W. Li, and M. Luo, Constraints of Viral RNA Synthesis on Codon Usage of Negative-Strand RNA Virus. J Virol, 2019. 93(5). 229. Durbin, A.F., et al., RNAs Containing Modified Nucleotides Fail To Trigger RIG-I Conformational Changes for Innate Immune Signaling. mBio, 2016. 7(5). 230. Russell, C.D., et al., The Human Immune Response to Respiratory Syncytial Virus Infection. Clin Microbiol Rev, 2017. 30(2): p. 481-502. 231. Karron, R.A., et al., Respiratory syncytial virus (RSV) SH and G proteins are not essential for viral replication in vitro: clinical evaluation and molecular characterization of a cold- passaged, attenuated RSV subgroup B mutant. Proc Natl Acad Sci U S A, 1997. 94(25): p. 13961-6. 232. Hastie, M.L., et al., The human respiratory syncytial virus nonstructural protein 1 regulates type I and type II interferon pathways. Mol Cell Proteomics, 2012. 11(5): p. 108- 27.

254

233. Kariko, K., et al., Suppression of RNA recognition by Toll-like receptors: the impact of nucleoside modification and the evolutionary origin of RNA. Immunity, 2005. 23(2): p. 165-75. 234. Lu, M., et al., N(6)-methyladenosine modification enables viral RNA to escape recognition by RNA sensor RIG-I. Nat Microbiol, 2020. 235. Cadena, C., et al., Ubiquitin-Dependent and -Independent Roles of E3 Ligase RIPLET in Innate Immunity. Cell, 2019. 177(5): p. 1187-1200 e16. 236. Weber, M., et al., Incoming RNA virus nucleocapsids containing a 5'-triphosphorylated genome activate RIG-I and antiviral signaling. Cell Host Microbe, 2013. 13(3): p. 336-46. 237. Taylor, G., Animal models of respiratory syncytial virus infection. Vaccine, 2017. 35(3): p. 469-480. 238. Zhang, Z., et al., Single-base mapping of m(6)A by an antibody-independent method. Sci Adv, 2019. 5(7): p. eaax0250. 239. Kotelkin, A., et al., Respiratory syncytial virus infection sensitizes cells to apoptosis mediated by tumor necrosis factor-related apoptosis-inducing ligand. J Virol, 2003. 77(17): p. 9156-72. 240. Wang, H., B. Xu, and J. Shi, N6-methyladenosine METTL3 promotes the breast cancer progression via targeting Bcl-2. Gene, 2020. 722: p. 144076.

255