Insights in Pro-Influenza Virus Activity of ANP32 Proteins Ecco Staller

Insights in pro-influenza virus activity of ANP32 proteins Ecco Staller

Imperial College London Department of Infectious Disease

Submitted for the Degree of Doctor of Philosophy

Abstract

All viruses usurp the host machinery to assist their replication and influenza virus is no exception. One such host factor is the acidic nuclear phosphoprotein of 32 kilodaltons ANP32A and its closely related paralogue ANP32B. Taking a CRISPR/Cas9 genome editing approach (Chapter III) it was demonstrated that human ANP32A and ANP32B are a pair of functionally redundant essential host factors for influenza A and B virus polymerase (FluPol) activity and influenza A virus replication in human cells (Chapter IV). Ablation of either ANP32A or ANP32B has a minor effect on FluPol activity, but ablation of both paralogues leads to complete abrogation of FluPol activity and virus replication. Using these double knockout (dKO) cells it was shown that mouse ANP32A lacks proviral function due to a single amino acid substitution at position 130 (Chapter IV). Natural variation in the genes encoding ANP32A and ANP32B is investigated next (Chapter V). A missense single nucleotide variant (SNV) in the Anp32B gene codes for a mutant protein with alanine at position 130 instead of the wildtype aspartic acid (ANP32B-D130A). This variant is relatively common in carriers of Hispanic/Latino descent and it was hypothesised that carriers of this SNV may have some natural genetic protection against influenza virus. CRISPR/Cas9 editing in human cells recapitalised the homozygous mutant genotype and it was found that FluPol activity and virus replication were compromised in the presence of ANP32B- D130A. Crucially, ANP32B-D130A exerted a dominant-negative effect over wildtype ANP32B and moreover interfered with the functionally redundant paralogue ANP32A (Chapter V). Finally in Chapter VI mutational analysis was carried out in order to map the proviral activity of ANP32 proteins to further structural elements and domains.

Acknowledgements

First and foremost, I would like to thank my supervisor Prof Wendy Barclay for giving me the opportunity to become a real scientist. Wendy has taught me to focus on that one killer experiment. Thank you Wendy!

I also want to thank my co-supervisor Prof Paul Farrell, my advisors Profs Robin Shattock and Peter O’Hare, and Dr Vanessa Sancho-Shimizu.

Dr Carol Sheppard was highly valuable as a mentor in the lab until she decided she would rather have another baby than guide me through the final year of my PhD. Thanks Carol!

Other Barclay lab members have been seminal to my development as a scientist and as a person, including Rebecca Frise, Dr Thomas Peacock, Dr Daniel Goldhill, Dr Jason Long, Dr Jie Zhou, Dr Ruthiran Kugathasan, Dr Jonathan Brown, Dr Bhakti Mistry and Laury Baillon.

I wish to thank my family – Ray, Janneke, Rembrandt, Jacques, Hélène, Stéphane, François, Yvette, Jacques, and Jane – for supporting me throughout this 10-year ordeal. Few of you thought I would make it but thanks to you I did.

One person who did think I would make it is my wife Virginie, to whom I shall be forever grateful. This thesis is dedicated to her and to my son Niels, who was temporarily abandoned by his father in the name of science.

Thank you Virginie

Thank you Niels

Contents Abstract...... 2 Acknowledgements ...... 3 List of Figures ...... 7 List of Tables ...... 9 Statement of originality ...... 10 Copyright Declaration...... 11 I INTRODUCTION ...... 12 1.1 Influenza virus requires host factors ...... 12 1.2 Nuclear import ...... 16 1.2.1 Prior to nuclear import ...... 16 1.2.3 The classical nuclear import pathway ...... 18 1.2.4 Nuclear import of vRNPs ...... 18 1.2.5 Non-canonical roles of alpha importins ...... 19 1.2.6 IMPα-independent import of PB1 and PA ...... 20 1.3 Transcription ...... 20 1.3.1 The role of host RNA polymerase II ...... 21 1.3.2 A role for CMTR1 in cap-snatching ...... 23 1.3.3 Influenza mRNA nuclear export ...... 23 1.4 Replication ...... 24 1.4.1 MCM ...... 25 1.4.2 Hsp90 and UAP56 ...... 25 1.4.3 ANP32 proteins ...... 27 1.4.4. Non-proteinaceous host factors support replication ...... 32 1.5 Nuclear export and beyond ...... 34 1.5.1 Involvement of the Raf/MEK/ERK signalling pathway ...... 34 1.5.2 Involvement of the CRM1 pathway ...... 36 1.5.3 The involvement of CLUH ...... 37 1.5.4 Progeny vRNP transport to the plasma membrane ...... 37 1.6 Discussion ...... 40 II MATERIALS AND METHODS ...... 41 2.1 Cells and cell culture ...... 41 2.2 Plasmids and cloning ...... 41 2.3 CRISPR/Cas9 genome editing ...... 44 2.4 Immunoblot analysis ...... 45 2.5 Minireplicon assays ...... 46

2.6 Fluorescence microscopy ...... 46 2.7 Influenza virus infection ...... 47 2.8 Viral RNA quantitation ...... 48 2.9 Split luciferase complementation assay ...... 48 2.11 Structural modelling ...... 49 2.12 Safety/biosecurity ...... 50 III GENERATING HUMAN CELL LINES LACKING ANP32 EXPRESSION USING CRISPR/CAS9 TECHNOLOGY ...... 51 3.1 CRISPR/Cas9 technology ...... 51 3.2 Cell lines ...... 55 3.3 Guide RNA design ...... 58 3.4 Generating KO cells ...... 65 3.5 Generating cells that lack expression of both ANP32A and ANP32B ...... 72 3.6 Discussion ...... 73 IV ANP32 PROTEINS ARE FUNCTIONALLY REDUNDANT ESSENTIAL HOST FACTORS FOR INFLUENZA VIRUS REPLICATION IN HUMAN CELLS ...... 76 4.1 Influenza virus polymerase activity in ANP32 knockout cell lines ...... 76 4.2 IAV and IBV FluPol activity is abrogated in dKO cells ...... 77 4.3 IAV replication is abrogated in dKO cells ...... 83 4.4 HPAI H5N1 virus can replicate to low titres in dKO cells ...... 87 4.5 Influenza virus RNA does not accumulate in dKO cells ...... 88 4.6 Mouse ANP32A does not support FluPol activity ...... 93 4.7 Murine ANP32A binds IBV polymerase more strongly than IAV polymerase ...... 97 4.8 Discussion ...... 101 V NATURAL VARIATION IN ANP32 PROTEINS ...... 104 5.1 Host genetics of influenza virus disease ...... 104 5.2 Variation in ANP32 proteins ...... 106 5.3 Natural variants of ANP32 proteins affect support of FluPol activity ...... 109 5.4 ANP32 position 130 mutants show impaired binding to FluPol ...... 112 5.5 ANP32 position 130 mutants exert dominant-negative effects ...... 114 5.6 Generating CRISPR-edited cells expressing wildtype ANP32A and mutant ANP32B- D130A ...... 117 5.7 IAV polymerase activity and replication are attenuated in ANP32B-D130A mutant cells ...... 121 5.8 Discussion ...... 123 VI MUTATIONAL ANALYSIS OF ANP32 PROTEINS ...... 127 6.2 Generation of ANP32B LCAR truncations ...... 128 6.3 Uncoupling pro-viral and cellular function ...... 133

6.4 ANP32B nuclear localisation is required for pro-influenza virus function ...... 139 6.5 Inability of ANP32E to support FluPol activity maps to E129 ...... 143 6.6 Discussion ...... 144 VII DISCUSSION ...... 150 VIII REFERENCES ...... 156

List of Figures

Figure 1.1 The influenza virus life cycle (p. 14) Figure 1.2 The influenza virus ribonucleoprotein (p. 15) Figure 1.3 Nuclear import of vRNPs and viral translation products (p. 17) Figure 1.4 Transcription and export of viral mRNA (p. 21) Figure 1.5 Replication and RNP assembly (p. 24) Figure 1.6 Structures of the ANP32 protein LRR domains (p. 28) Figure 1.7 Nuclear export of progeny vRNPs (p. 35) Figure 1.8 Cytoplasmic trafficking of progeny vRNPs (p. 38) Figure 3.1 Bacterial CRISPR-mediated immunity and its adaptation to mammalian genome engineering (p. 52) Figure 3.2 Possible outcomes of DSB repair (p. 53) Figure 3.3 Double-nickase strategy – using a pair of RuvC-D10A Cas9 nickases to generate a DSB (p. 54) Figure 3.4 ANP32 proteins are expressed in HEK293T, A549, HAP1 and eHAP cells (p. 55) Figure 3.5 On average, HEK293T cells contain 4-5 copies of Anp32A and 3-4 copies of Anp32B (p. 56) Figure 3.6 eHAP cells are derived from HAP1 cells, which in turn derive from KBM-7 cells (p. 57) Figure 3.7 Guide RNA cloning strategy (p. 58) Figure 3.8 Double nickase strategy to obtain ANP32A and ANP32B knockout cells (p. 59) Figure 3.9 Genomic cleavage validation of CRISPR clones (p. 60) Figure 3.10 Single plasmid transfection optimization in A549 and HAP1 cells (p. 61) Figure 3.11 The minigenome reporter assay (p. 62) Figure 3.12 Minigenome reporter assays with blue fluorescent protein (BFP) reporter in A549, HAP1 and eHAP cells (p. 63) Figure 3.13 Influenza virus replicates in A549 and HAP1 cells (p. 64) Figure 3.14 Influenza B virus polymerase is active in HAP1 and eHAP cells (p. 64) Figure 3.15 Next-generation sequencing strategy for screening CRISPR clones (p.65) Figure 3.16 Each CRISPR clone has a unique combination of barcode and NGS index (p. 66) Figure 3.17 Genotype of CRISPR knockout cells (p. 67) Figure 3.18 Validation of NGS data by western blotting analysis (p. 67) Figure 3.19 Influenza virus polymerase is active in ANP32A and B single knockout cells (p. 68) Figure 3.20 H1N1 PR8 virus replicates in control, AKO and BKO cells (p. 69) Figure 3.21 Sanger sequencing of selected eHAP knockout cells (p. 70) Figure 3.22 ANP32A is expressed in BKO cells and vice versa (p. 72) Figure 3.23 ANP32A expression is lost in double knockout cells (p. 73)

Figure 3.24 Influenza polymerase activity is abrogated in ANP32A and B double knockout cells (p. 73) Figure 4.1 ANP32 expression profiles of control, AKO, BKO and dKO cells (p. 77) Figure 4.2 AKO and BKO cells support FluPol activity but dKO cells do not (p. 78) Figure 4.3 Optimisation by titration of FLAG-tagged ANP32 expression plasmids (p. 80) Figure 4.4 Exogenous expression of ANP32 proteins recovers FluPol activity in dKO cells (p. 82) Figure 4.5 On a single-cell level FluPol is active only in dKO cells co-expressing ANP32 proteins (p. 83) Figure 4.6 H1N1 PR8 virus replication is abrogated in dKO cells (p. 84) Figure 4.7 Nucleoprotein does not accumulate in dKO cells (p. 85) Figure 4.8 IAV replication is abrogated in dKO cells (p. 87) Figure 4.9 H5N1 Tky/05 can replicate to low titre in dKO cells (p. 88) Figure 4.10 Tagged strand-specific RT primers eliminate non-specific PCR products (p. 89) Figure 4.11 Tagged RT primers eliminate accumulation of non-specific products during RT-qPCR analysis (p. 90) Figure 4.12 RNA synthesis is abrogated in dKO cells (p. 92) Figure 4.13 Murine ANP32A does not support IAV FluPol activity (p. 93) Figure 4.14 IAV polymerase activity is supported by murine ANP32B (p. 94) Figure 4.15 IBV polymerase gains activity in the presence of MusA (p. 96) Figure 4.16 Single residue substitutions at position 130 of ANP32A affect FluPol activity (p. 97) Figure 4.17 Gluc-tagged PB1 and ANP32 constructs retain function (p. 99) Figure 4.18 Murine ANP32A binds IBV polymerase more strongly than IAV polymerase (p. 100) Figure 5.1 Selected naturally occurring missense SNVs in Anp32A and Anp32B (p. 109) Figure 5.2 ANP32A and B mutant proteins show variable capacity to rescue IAV FluPol activity (p. 112) Figure 5.3 ANP32 position 130 mutants show impaired binding to FluPol (p. 113) Figure 5.4 ANP32B-D130A exerts dominant-negative effects on pro-viral function of wildtype ANP32B and ANP32A (p. 116) Figure 5.5 Strategy to generate and screen CRISPR-edited ANP32B-D130A cells (p. 119) Figure 5.6 Screening and validation of CRISPR clones (p. 120) Figure 5.7 FluPol activity is attenuated in ANP32B-D130A mutant cells (p. 122) Figure 5.8 IAV replication attenuated in ANP32B-D130A mutant cells (p. 122) Figure 5.9 Cryo-EM structure of a dimer of ICV FluPol heterotrimers in complex with ANP32A (p. 124) Figure 5.10 Mutation at ANP32 position 130 results in impaired formation of replication-competent FluPol dimer (p. 125) Figure 6.1 ANP32 protein domains and motifs (p. 129) Figure 6.2 ANP32 constructs primary structures (p. 130) Figure 6.3 Human ANP32B LCAR and central region are required for pro-viral function (p. 131)

Figure 6.4 The LCAR domain of human ANP32B is required for FluPol activity, unlike the LCAR of chicken ANP32A (p. 132) Figure 6.5 ANP32 binding to core histones is mediated by the LCAR (p. 134) Figure 6.6 ANP32B-LOA does not support FluPol activity despite having an intact LCAR (p. 135) Figure 6.7 ANP32B-LOA interacts with FluPol despite lacking pro-viral activity (p. 136) Figure 6.8 Human ANP32 proteins lacking the LCAR domain cannot support FluPol activity (p. 138) Figure 6.9 ANP32B NLS mutant does not affect nuclear localisation or FluPol activity (p. 140) Figure 6.10 Substituting the ANP32B NLS for a canonical NES affects FluPol activity (p. 142) Figure 6.11 ANP32E has potential pro-influenza virus activity (p. 144) Figure 6.12 Primary structures of ΔLCAR constructs (p. 146) Figure 6.13 ANP32A LRR bridges asymmetric FluPol dimer (p. 147) Figure 6.14 The 176-183 motif is important for proviral function (p. 148)

List of Tables

Table 3.1 Indel mutations found in ANP32A and ANP32B eHAP knockout cells by NGS and Sanger sequencing (p. 71) Table 4.1 Genotypes of eHAP knockout cells (p. 76) Table 6.1 Interaction and functional profile of ΔLCAR constructs (p. 145)

Statement of originality

I confirm that all the work presented in this thesis is my own, unless otherwise stated. Any material from other sources has been properly acknowledged.

The copyright of this thesis rests with the author. Unless otherwise indicated, its contents are licensed under a Creative Commons Attribution-Non Commercial 4.0 International Licence (CC BY-NC). Under this licence, you may copy and redistribute the material in any medium or format. You may also create and distribute modified versions of the work. This is on the condition that: you credit the author and do not use it, or any derivative works, for a commercial purpose. When reusing or sharing this work, ensure you make the licence terms clear to others by naming the licence and linking to the licence text. Where a work has been adapted, you should indicate that the work has been changed and describe those changes. Please seek permission from the copyright holder for uses of this work that are not included in this licence or permitted under UK Copyright Law.

I INTRODUCTION

1.1 Influenza virus requires host factors

All viruses usurp the host machinery to assist their replication and influenza virus is no exception. There are four distinct influenza virus types, namely A, B, C and D. Respiratory disease in humans is caused predominantly by influenza A and B viruses (IAV and IBV), while influenza C virus (ICV) is known to cause sporadic disease in children. Influenza D virus infects cattle. Two IAV strains (subtypes H3N2 and H1N1, based on antigenic diversity of the HA and NA glycoproteins) and two IBV strains (Victoria and Yamagata lineage) currently circulate in humans, causing seasonal epidemics that affect millions (Krammer et al., 2018). The single- stranded negative-sense RNA genome of influenza A viruses (vRNA) consists of eight segments encoding ten core polyproteins: the RNA-dependent RNA polymerase (RdRp) subunits polymerase basic 1 (PB1), polymerase basic 2 (PB2) and polymerase acidic (PA), the glycoproteins haemagglutinin (HA) and neuraminidase (NA), the viral nucleoprotein NP, the matrix protein M1, the M2 ion channel, the non-structural protein NS1 and finally the nuclear export protein NEP (previously known as NS2). An additional suite of virus strain- dependent non-essential accessory proteins have been described (Vasin et al., 2014).

Eighteen HA subtypes and 11 NA subtypes have been identified in IAV to date, all of which present in avian species except H17N10 and H18N11, which infect bats (Tong et al., 2012; Tong et al., 2013). A large and antigenically diverse reservoir of IAV subtypes is present in wild waterfowl from whence viruses transfer into mammalian host species including pigs, horses, dogs and humans (Lycett et al., 2019). Although most of these zoonoses will be dead end infections, zoonotic events can lead to pandemics when the virus accumulates sufficient mutations to allow for effective transmission among members of the new host species (Long et al., 2019b). Recent species jumps of H5 and H7 subtypes, lethal in poultry, have led to substantial worry among global health experts. Influenza pandemics occur every few decades, most recently in 2009 (H1N1 swine flu) but also in 1968 (H3N2 Hong Kong flu), 1957 (H2N2 Asian flu) and most notoriously in 1918 (H1N1 Spanish flu), the latter killing at least 50 million people.

Mammalian adaptive mutations are required by avian influenza strains before they can cause a pandemic. These include changes in HA, NA and the heterotrimeric RdRp and NP (reviewed in (Long et al., 2019b)). The PB2 subunit of the polymerase in particular is well known to undergo mammalian adaptions. Substitution of the avian-signature glutamate at position 627 and aspartate 701 to mammalian-signature 627K and 701N are common in zoonotic and

12 human adapted strains and adapt polymerase to use mammalian host factors as discussed in this review. Additional adaptations are believed to counteract mammalian antiviral immune pathways. The retinoic acid-inducible gene 1 protein RIG-I recognises influenza viral RNA on the basis of a 5’-terminal triphosphate moiety and an adjacent patch of double-stranded RNA that is the viral promoter binding the RdRp (reviewed in (Fodor and Te Velthuis, 2019; Wandzik et al., 2020). During infection, the viral NS1 protein counteracts RIG-I which would otherwise result in the induction of interferon and interferon-stimulated gene (ISG) expression, such as MX1, PKR and IFITM3 (Pichlmair et al., 2006; Rajsbaum et al., 2012; Rehwinkel et al., 2010). It has also been suggested that the PB2 host-adapting mutation E627K masks the pathogen- associated molecular pattern (PAMP) on the incoming genome from detection by RIG-I (Weber et al., 2015).

Antivirals against influenza virus are in limited supply. The M2 ion channel inhibitors amantadine and rimantadine have been discontinued as all circulating influenza viruses have developed resistance in the form of an S31N substitution in the RNA segment encoding M2. Oseltamivir (Tamiflu) is an NA inhibitor targeting the enzymatic sialidase activity of NA, but NA inhibitors are also compromised by resistance-conferring mutations such as NA-H275Y that spread through the seasonal H1N1 viral population. Recently a variety of compounds that target the replication machinery has come to the fore (reviewed in (Takashita, 2020)). Active sites for RNA binding, cleavage and elongation are highly conserved among influenza viruses, so the RdRp is a potentially promising target for antiviral agents. The PB1-targeting purine analogue favipiravir acts as a chain terminator and has been approved for use in Japan. Resistance mutations have, however, been described, in particular a K229R substitution in the PB1 subunit alongside a fitness compensating PA-P653L mutation (Goldhill et al., 2018). Pimodivir occupies the m7G cap binding domain of the PB2 subunit which is required for cap- snatching and therefore transcription of the vRNA. Pimidovir-resistant H1N1 and H3N2 viruses are routinely detected in clinical trials with this compound (Finberg et al., 2019; Trevejo et al., 2018). Finally, Baloxavir Marboxil binds to the PA endonuclease domain, also required for cap-snatching (Omoto et al., 2018). Approved in Japan, an I38T substitution in PA seems to confer resistance against the drug in circulating H1N1 and H3N2 viruses (Hayden et al., 2018; Hirotsu et al., 2019; Noshi et al., 2018; Uehara et al., 2020).

Given that influenza virus acquires resistance to antivirals targeting virus-encoded proteins rapidly and consistently, an alternative strategy may be called for. Host-directed therapy targets host-encoded rather than virus-encoded factors and as such aims to sidestep rapid accumulation of resistance mutations on the premise that viruses cannot directly control the appearance of host proteins. The strategy is to interfere with host cell factors that are required by a pathogen for replication or persistence (reviewed in (Kaufmann et al., 2018)).

The viral life-cycle involves the following steps, all of which require host factors (reviewed in Staller and Barclay, 2020, in press (Eisfeld et al., 2015; Peacock et al., 2019)) (Figure 1.1). Upon virus binding and endocytosis, the vRNPs (Figure 1.2) are released into the cytoplasm and imported into the host cell nucleus. Once inside the nucleus the negative-sense RNA is transcribed into mRNA which is transferred to the cytoplasm for translation (reviewed in (Wandzik et al., 2020)). Newly generated HA, NA, M1 and M2 traffic to lipid rafts in the plasma membrane while PB1, PB2, PA, NP, NEP and a subset of M1 are recycled into the nucleus where they are required for genome replication and the nuclear export of progeny vRNPs. Replication takes place in two steps, namely synthesis of an intermediate positive-sense complementary RNA (cRNA), which in turn is used as a template for the synthesis of additional vRNA for secondary transcription and assembly into vRNPs for packaging in progeny virions (reviewed in (Fodor and Te Velthuis, 2019)). The final steps of the life cycle are vRNP transport from the perinuclear region to the plasma membrane, followed by particle assembly.

Figure 1.1 The influenza virus life cycle (from te Velthuis and Fodor, Nat Rev Microbiol 2016) Upon virus entry acidification of the endosome and the virion itself, via the M2 ion channel, triggers release of the vRNPs into the cytoplasm. The vRNPs are transported into the cell nucleus, where they are

14 transcribed into viral mRNAs (primary transcription) that are exported to the cytoplasm for translation. vRNP components PB1, PB2, PA and NP cycle back into the nucleus as they are required for replication of the negative-sense vRNA into a complementary positive-sense cRNA. The cRNPs are then used as templates to produce additional vRNPs for secondary transcription and accumulation of progeny genomes. When sufficient glycoproteins (HA and NA) have accumulated in lipid rafts at the plasma membrane, progeny vRNP nuclear export is triggered. The vRNPs traffic to the plasma membrane where they are assembled into virions for egress.

Figure 1.2 The influenza virus ribonucleoprotein (from Eisfeld et al Nat Rev Microbiol 2015) The negative-sense single-stranded genomic RNA (or, in the case of the cRNP, positive-sense complementary RNA) is laid out in a double helical loop structure bound along its length by nucleoprotein. The conserved 3’ and 5’ ends, constituting the promoter, are bound by a trimeric RNA- dependent RNP polymerase (RdRp) complex, which consists of polymerase basic 1 (PB1), polymerase basic 2 (PB2), and polymerase acidic (PA). This incumbent RdRp transcribes and replicates the genome segment in cis.

This chapter examines the host factors required for nuclear import of the vRNPs and vRNP subunits, transcription and replication of the genome, nuclear export of progeny vRNPs and trafficking of vRNPs for egress. The main focus will be on ANP32 proteins, which are essential host factors for influenza virus replication, and perhaps constitute the most promising target for novel antiviral host-directed therapy.

The bulk of this chapter appears as a review article and book chapter in the Cold Spring Harbor Laboratory Press Perspectives in Medicine series Influenza: The Cutting Edge (Staller and Barclay 2020, in press).

1.2 Nuclear import

Unlike most RNA viruses that infect vertebrates, influenza virus transcribes and replicates its genome in the cell nucleus. The advantages of nuclear over cytoplasmic replication are at least fourfold (Hutchinson and Fodor, 2012). First, access to the nucleus facilitates cap- snatching, the process by which the viral RdRp obtains 5’cap structures from cellular RNA species to act as primers for transcription of the viral mRNA. Although cap-snatching is found also in some cytoplasmic replicating viruses including the bunyaviridae and arenaviridae families (Mir et al., 2008), nuclear cap snatching is deemed more efficient and provides direct access to cellular pathways of cotranscriptional mRNA processing and nuclear export (Engelhardt and Fodor, 2006). Second, nuclear replication allows access to the splicing machinery, enabling expansion of the coding capacity of the small genome (about 13.5 kb). Indeed, spliced segments 7 and 8 encode the ion channel M2 and the nuclear export protein NEP, respectively, whereas the full-length transcripts encode the M1 matrix protein and the multifunctional innate immune antagonist NS1.Third, direct interaction with the carboxy- terminal domain (CTD) of the cellular RNA polymerase II (RNA Pol II) via the PA subunit of the RdRp leads to inhibition of RNA Pol II elongation and causes it to be degraded, leading to efficient host shutoff (Rodriguez et al., 2007; Vreede et al., 2010; Vreede and Fodor, 2010). Fourth, one of the main mechanisms by which influenza virus is detected by the host is via the cytoplasmic pathogen recognition receptor (PRR) RIG-I. Encapsidating newly synthesised RNA within RNPs in the nucleus may partially mask the RNA molecular patterns capable of triggering an antiviral response (Rehwinkel et al., 2010).

1.2.1 Prior to nuclear import

Upon entry into the target cell, vRNPs are dissociated from the viral matrix protein M1 in the low pH environment of the endocytosed virion when M2 ion channels open within the acidified endosome (Bui et al., 1996; Martin and Helenius, 1991a). The vRNPs are then actively transported into the nucleus where transcription and replication take place (Kemler et al., 1994; Martin and Helenius, 1991b) (Figure 1.3). Recently, TNPO1 (transportin 1), a member of the importin β family, was implicated in the removal of M1 matrix protein and debundling of vRNPs just after release of the viral cores into the cytoplasm (Miyake et al., 2019). During virus entry, acidification triggers a conformational change in M1 that exposes a non-canonical

16 proline-tyrosine nuclear localization signal (PY-NLS) close to the N terminus (Lee et al., 2006). TNPO1 then associates with the PY-NLS exposed in the primed M1, stripping M1 from vRNPs and allowing dissociation of the eight vRNPs from each other. vRNPs are subsequently transferred to alpha importins for nuclear import. Influenza A viruses thus use different alpha and beta importins for distinct steps in host cell entry.

Figure 1.3 Nuclear import of vRNPs and viral translation products Reproduced from Staller & Barclay, Cold Spring Harb Perspect Med, 2020 Following fusion of the viral and endosomal membranes, incoming viral ribonucleoproteins (vRNPs) associated with M1 matrix protein enter the cytoplasm. The importin β family member TNPO1 / transportin 1 strips M1 from the vRNPs thus allowing dissociation of the eight vRNPs from each other and releasing them for interactions with the classical nuclear import machinery. vRNPs interact with importin α (IMPα) adaptor proteins that recognise a non- classical nuclear localisation signal (NLS) at the surface-exposed N-terminus of the nucleoprotein (NP). Direct interaction between IMPα and the PB2 subunit of the viral RdRp has also been described, but the biological relevance of this is not known. IMPα is complexed with the importin β (IMPβ) transport receptor, which mediates translocation into the nucleus by interactions with the nuclear pore complex (NPC). IMPβ is dissociated from the cargo in the nucleus by Ran-GTP associated with the NPC. IMPα is removed separately by CSE1L, although some IMPα may remain associated with the vRNP and support viral replication via unknown mechanisms. The importins subsequently cycle back into the cytoplasm.

After primary transcription and translation of viral proteins, NP and RdRp components are transported back into the nucleus to support replication and synthesis of progeny vRNPs. PB2 and NP are imported on their own via classical nuclear import, the former in conjunction with its chaperone Hsp90. PB1 and PA are imported as heterodimers in a non-classical manner, through direct interaction with the beta importin RanBP5. PB1 and PA are maintained in a stable conformation and protected against degradation by the non-proteinaceous chaperones long non-coding RNA IPAN (lncRNA-IPAN) and lncRNA-PAAN, respectively. A subset of newly synthesised M1, required for the vRNP nuclear export complex later in the viral life cycle, is believed to associate with the mitochondrial host protein CLUH for nuclear translocation.

1.2.3 The classical nuclear import pathway

The classical pathway of cellular protein nuclear import is mediated by importin α (IMPα) adaptor proteins, which recognise basic nuclear localisation signals (NLSs) on cargo proteins. IMPα isoforms, of which there are seven in humans, are in turn recognised by the importin β (IMPβ) transport receptor via the N-terminal importin β-binding (IBB) domain, which is a functional NLS in itself and autoinhibits IMPα so that recognition of NLSs occurs only when IMPβ is available. The ternary IMPα- IMPβ-cargo complex subsequently traverses the nuclear pore complex (NPC) through interaction of IMPβ with phenylalanine-glycine (FG) repeats on the nucleoporins that compose the NPC. Once inside the nucleus IMPβ is dissociated by Ran- GTP associated with the NPC. IMPα is dissociated from the cargo by a separate mechanism involving the cellular CSE1L protein (Kutay et al., 1997). Subtle differences in NLS affinity, conformational flexibility and IBB autoinhibition between IMPα isoforms leads to a flexible and functionally redundant transport machinery. It has been proposed that importin α subfamilies α2 and α3 provide a fast track for nuclear import of important cellular cargos, and that these subfamilies are specifically targeted by influenza vRNPs (Pumroy et al., 2015).

1.2.4 Nuclear import of vRNPs

Influenza vRNPs are transported into the nucleus by the interaction between the NP subunits associated with the genome and IMPα (O'Neill et al., 1995). IAV NP carries a ‘non-classical’ NLS (ncNLS) at the N-terminus which is surface-exposed and thus available for IMPα binding (Arranz et al., 2012; Moeller et al., 2012). This motif binds IMPα isoforms and is vital for vRNP nuclear import in digitonin-permeabilised cells and for influenza virus replication (Cros et al., 2005; Neumann et al., 1997; Wang et al., 1997). Ultrastructural analysis of the surface

18 availability of the ncNLS on purified vRNPs revealed multiple exposed motifs along the vRNP filament (Wu et al., 2007). IMPα 5 binding prevents the oligomerisation of NP in vitro (Boulo et al., 2011), suggesting an additional role for alpha importins as NP chaperones. It is highly likely that the same IMPα isoform also imports newly synthesized NP into the nucleus during replication.

Once the vRNP has entered the nucleus and primary transcription given rise to new viral proteins, newly synthesized NP and polymerase subunits must be imported into the nucleus to support replication of cRNA and of new vRNA genomes. PB2 has been shown to interact with multiple IMPα isoforms (Gabriel et al., 2008; Gabriel et al., 2011; Tarendeau et al., 2007), but NP is sufficient to traffic vRNA into the nucleus. Thus the PB2-IMPα interaction is likely more important for trafficking newly synthesized PB2 into the nucleus to support and transactivate replication (Cros et al., 2005). The mammalian-adaptive D701N substitution in the PB2 subunit has been suggested to specifically bind IMPα isoforms in different host species (Boivin and Hart, 2011; Gabriel et al., 2008; Gabriel et al., 2011), and was shown to promote nuclear import of vRNPs in mammalian cells (Sediri et al., 2015). The latter observation implies some role for PB2 in import of vRNPs but this has not been entirely resolved.

The differential use of different IMPα isoforms has been associated with host range restriction of avian origin polymerases in mammalian cells. Mammalian-adapted viruses are affected most by silencing of importins α1 and α7, whereas avian-origin viruses require importins α1 and α3 (Gabriel et al., 2011). IMPα 7 knockout mice are less susceptible to influenza infection, and specifically show reduced tropism of virus in the lung (Resa-Infante et al., 2014). Viral mutants that can regain infectivity in this system acquire mutations allowing binding of NP to different isoforms (Resa-Infante et al., 2019; Resa-Infante et al., 2015). Adaptation of an avian- signature H7N7 virus by serial lung passage in mice included key adaptations in PB2 (D701N) and NP (N319K) that increased affinity to mammalian IMPα 1.

1.2.5 Non-canonical roles of alpha importins

An additional role for IMPα isoforms 1 and 7 as positive regulators of mammalian-signature but not avian-signature viral polymerase activity in mammalian cells has been suggested, independent of nuclear import functions (Hudjetz and Gabriel, 2012; Resa-Infante et al., 2008). Importins may remain associated with the viral polymerase even after nuclear import suggesting they play a role in polymerase function. Mutation of the PB2 NLS resulted in reduced RdRp activity in a minireplicon assay, despite proper formation of polymerase

19 complexes (Resa-Infante et al., 2008). Moreover, replacing the PB2 NLS with an ectopic SV40 NLS enabled PB2 nuclear import but abrogated polymerase function, again implying that specific IMPα association with the PB2 subunit in the nucleus is required for RdRp activity. RNAi-mediated silencing of IMPα 1 and 7 in human 293T cells resulted in reduced activity of human-adapted PB2-627K polymerase as measured by minireplicon assay, but the avian- signature 627E polymerase was unaffected (Hudjetz and Gabriel, 2012). The exact role of this interaction of PB2 with specific IMPα isoforms is not yet clear.

1.2.6 IMPα-independent import of PB1 and PA

Whilst newly synthesised PB2 and NP are imported into the nucleus on their own via the classical import pathway, PA-PB1 is imported as a dimeric subcomplex (Huet et al., 2010; Hutchinson et al., 2011; Loucaides et al., 2009). PB1 contains a bipartite NLS that directly interacts with the β-importin Ran binding protein 5 (RanBP5) (Deng et al., 2006). PA contains NLSs between residues 124-139 and 186-247. PA-PB1 forms a stable heterodimeric submodule that strongly interacts with 5’ vRNA, and moreover forms a stable and stoichiometric 1:1:1 complex with RanBP5 as modelled by small angle X- ray scattering (SAXS) (Swale et al., 2016). Interestingly, when bound to RanBP5 the PB1-PA dimer loses its capacity to bind the viral RNA promoter, suggesting that PB1-PA will pick up the vRNA once it dissociates from RanBP5 after nuclear entry. A recent X-ray crystallography study identified the residues of RanBP5 required for the interaction with influenza virus PA-PB1 heterodimer (Swale et al., 2020). The classical IMPα- IMPβ import pathway is thus not required for nuclear accumulation of PB1 and PA. 1.3 Transcription

Upon entry into the nucleus the incoming viral RdRp associated with the vRNPs transcribes the negative-sense genomic RNA segments into positive-strand mRNA in cis (Figure 1.4). mRNA molecules are capped at the 5’ end by an N7-methyl guanosine (m7G) cap, derived from capped nascent host RNA Pol II transcripts, in particular small nucleolar RNAs (snoRNAs) and small nuclear RNAs (snRNAs) (Gu et al., 2015; Koppstein et al., 2015). Cap- snatching is carried out by the RdRp utilising the PB2 cap-binding domain (residues 320-485) and the PA endonuclease (residues 1-196). The 10-15 nt capped host RNA fragments are then used as primers for transcription of viral mRNA (reviewed in (Wandzik et al., 2020) and (Fodor and Te Velthuis, 2019)).

Figure 1.4 Transcription and export of viral mRNA Reproduced from Staller & Barclay, Cold Spring Harb Perspect Med, 2020 Influenza virus acquires 5’ m7G-capped host transcripts from actively transcribing host RNA polymerase II (RNA Pol II), which it subsequently uses to prime viral transcription. To facilitate cap-snatching the viral RdRp associates with the carboxy-terminal domain (CTD) of the large subunit of RNA Pol II through interaction between the PA subunit and the serine 5 phosphorylated (Ser5P) CTD. The methyltransferase CMTR1 appears to have a role in cap-snatching but the mechanism is not known. Influenza mRNAs are exported via the NXF1/TAP pathway through interaction with the mRNA export receptor NXF1-NXT1, which mediates translocation across the NPC via interaction with nucleoporins. Later in infection, as viral proteins accumulate, NXF1-NXT1 complexes are targeted by the virus-encoded virulence factor non-structural protein 1 (NS1) in order to block host mRNA export, in particular antiviral transcripts.

1.3.1 The role of host RNA polymerase II

Cap-snatching is enabled by direct interaction of the large PA carboxy-terminal domain (residues 258-713) with the carboxy-terminal domain (CTD) of host RNA polymerase II (reviewed in (Walker and Fodor, 2019)). Human cellular RNA Pol II is a complex composed of 12 core subunits; the large subunit RPB1 has an unstructured flexible C-terminal domain consisting of 52 heptad repeats with the consensus sequence Tyr1-Ser2-Pro3-Thr4-Ser5-

Pro6-Ser7 (Y1S2P3T4S5P6S7). Post-translational modification of the CTD regulates

21 transcription and cotranscriptional processing of host RNAs (Zaborowska et al., 2016). Early in cellular transcription Ser5 on the CTD is phosphorylated as a trigger for mRNA capping enzymes to associate with the CTD. Throughout elongation CTD Ser5 is gradually dephosphorylated and Ser2 becomes phosphorylated instead. The switch between Ser5P and Ser2P is coupled with the release of 5’end processing factors and recruitment of 3’ end processing factors. Influenza RdRp interacts directly with Pol II by binding to SerP5 CTD at transcription start sites, which puts it in close proximity to the capped host transcripts it requires for initiating its own transcription (Engelhardt et al., 2005). Fluorescence recovery after photobleaching (FRAP) analysis was harnessed to analyse the intranuclear dynamics of GFP-tagged RdRp in living cells. Treatment of the cells with inhibitors of RNA Pol II actinomycin D and α-amanitin led to mobility shifts of the RdRp, confirming the influence the cellular Pol II transcription complex has on the viral RNP (Loucaides et al., 2009).

RdRp does not associate with promoter regions of pol I or Pol III genes, and furthermore shows no affinity with Ser2P CTD associated with elongating Pol II. Pol II activity is required for viral mRNA synthesis but not cRNA or vRNA synthesis. Moreover, viral infection led to inhibition of polymerase elongation; fewer Pol II units were found associated with internal regions of host genes in infected cells (Chan et al., 2006). This elongation inhibition is likely due to competitive binding of the viral RdRp to the CTD, blocking its normal interface with host transcriptional machinery. Alternatively or in addition, the removal of 5’ caps from nascent host transcripts by the RdRp may lead to degradation of the remaining transcript by Xrn2 exonuclease, leading to early termination (Brannan et al., 2012).

Interaction of influenza virus with RNA Pol II results in profound changes to the host cell transcripts. A recent study using mammalian native elongating transcript sequencing (mNET- seq) confirmed there is decreased RNA Pol II occupancy downstream of transcription start sites in influenza virus infected cells. The authors also detected generation of polyadenylated downstream-of-gene transcripts (DoGs) that extend beyond normal poly(A) sites due to catastrophically altered termination (CAT), reminiscent of a cellular reaction to severe stress like osmotic shock or heat shock (Bauer et al., 2018).

Recent structural work has provided insight in the molecular details of RdRp PA subunit interaction with the RNA Pol II CTD and how influenza A B and C polymerases interaction compare to one another (Lukarska et al., 2017; Serna Martin et al., 2018). Ser5P Pol II CTD binding, in combination with vRNA promoter binding to the RdRp is believed to promote a transcription preinitiation conformation in the RdRp in order to facilitate cap snatching; in other words association with RNA Pol II activates the transcriptase conformation of the viral polymerase. As a result of RdRp interacting with host RNA Pol II it has been estimated that

22 eight hours post-infection over half of all the mRNA in an influenza virus-infected cell is viral mRNA (Bercovich-Kinori et al., 2016).

1.3.2 A role for CMTR1 in cap-snatching

CMTR1 is the human 2’-O-ribose cap methyltransferase that adds a methyl group to the 5’-7- methylguanosine cap of eukaryotic mRNA to form the cap1 structure (Smietanski et al., 2014). As influenza virus requires 2’-O-methylation of the mRNA cap for efficient recognition and cleavage by PA (Bouloy et al., 1980; Wakai et al., 2011), it was interesting to find CMTR1 as a major hit in a pooled genome-wide CRISPR/Cas9 screen (Li et al., 2020). Infection of CMTR1 KO A549 cells showed that loss of this protein conferred protection against a variety of H1N1 and H3N2 virus strains. Immunoprecipitation of capped viral and host mRNA using an anti-eIF4E antibody revealed that there was less cap1-associated viral mRNA in CMTR1 knockout cells, presumably because cap-snatching was impaired. Furthermore, loss of CMTR1 resulted in an increased antiviral IFN response in infected cells and conferred synergistic protection against IAV infection with the endonuclease inhibitor Baloxavir Marboxil. How exactly CMTR1 facilitates cap-snatching remains to be elucidated.

1.3.3 Influenza mRNA nuclear export

Cellular mRNAs are predominantly exported to the cytoplasm via the NXF1/TAP pathway through acquisition of the principal mRNA export receptor NXF1-NXT1. NXF1-NXT1 interacts with the FG repeats of nucleoporins, which is required for translocation of mRNAs through the NPC (Stewart, 2010). The influenza A virus non-structural protein NS1 is known to inhibit nuclear export of host mRNA, thus contributing to NS1-mediated inhibition of host immune responses (Satterly et al., 2007). Recently the structural basis for this block was elucidated by co-crystallization of NS1 in complex with NXF1-NXT1, revealing that NS1 acts by preventing binding of NXF1-NXT1 to nucleoporins by occupying the nucleoporin binding domain of NXF1 (Zhang et al., 2019b). Transcripts of interferon- or immune-regulated genes were among the top transcripts retained in the host nucleus following IAV infection. A mutant virus unable to inhibit NXF1-NXT1 was attenuated.

On the other hand, NXF1 has been routinely detected as an important host factor for influenza replication in genome-wide screens (Hao et al., 2008; Karlas et al., 2010; Shapira et al., 2009; Watanabe et al., 2010) and was also shown to interact with RdRp subunits using a novel split luciferase complementation screen (Munier et al., 2013). Influenza mRNAs are transported from the nucleus by different host factors including the NXF1 and CRM1 pathways (Carmody and Wente, 2009; Wang et al., 2008). RNAi-mediated knockdown of NXF1 showed that

23 segment 4 (HA), 7 (M) and 8 (NS1) mRNAs were the most reliant on this pathway (Read and Digard, 2010) and this explains the attenuation of virus replication in cells lacking this host factor. Thus there is an interesting and unresolved dichotomy between viral reliance on the NXF1 pathway for nuclear export of viral mRNAs and its shutoff via NS1 to avoid expression of antiviral genes.

1.4 Replication

Replication of the influenza virus genome occurs in two steps (reviewed in (Fodor and Te Velthuis, 2019; Wandzik et al., 2020). First the RdRp synthesises cRNA, a process that requires newly synthesised RdRp components and NP (Jorba et al., 2009; York et al., 2013). These cRNPs are then used as templates for the synthesis of additional vRNPs for secondary transcription and later for export across the nuclear membrane and to the plasma membrane for incorporation into progeny virions. cRNA synthesis on the vRNA template is subtly distinct from vRNA synthesis on a cRNA template, which requires primer realignment (reviewed in (Te Velthuis and Fodor, 2016)). Host factors have been postulated to play supportive roles in all steps of the replication cycle (Figure 1.5). Oligomerisation of trimeric RdRp molecules into dimers, trimers or even higher order complexes is now believed to be required for efficient replication (Carrique et al, 2020, in press) (Chang et al., 2015; Chen et al., 2019; Fan et al., 2019; Peng et al., 2019).

Figure 1.5 Replication and RNP assembly Reproduced from Staller & Barclay, Cold Spring Harb Perspect Med, 2020 As newly synthesized NP and RdRp components accumulate in the nucleus the

24 genome is replicated into a positive-sense complementary genome (cRNA) which is packaged into a cRNP by newly synthesised RdRp and NP. cRNA synthesis from a vRNA template is believed to require the minichromosome maintenance complex (MCM) which stabilises the transition from initiation to elongation. Functional RdRp complexes are assembled from PA-PB1 dimers and PB2 as their respective chaperones lncRNA-PAAN, lncRNA-IPAN and Hsp90 dissociate. NP assembles with UAP56 after nuclear import, which shields it from aggregation and stores it as trimeric units. UAP56 is subsequently competed off the NP as the latter encapsidates newly synthesised viral RNA. ANP32 proteins are shown as stabilising a trans-activating polymerase required for cRNA-to-vRNA replication or stabilising a trimeric RdRp associating with the 3’ end of nascent vRNA1

1.4.1 MCM

The minichromosome maintenance (MCM) complex was identified as a nuclear factor supporting de novo replication initiation of cRNA on a vRNA template (Kawaguchi and Nagata, 2007). MCM was shown to interact with the PA subunit of the RdRp, and siRNA-mediated knockdown of subunits MCM2 and MCM3 led to reduced viral RNA accumulation in infected human cells. The authors propose a model in which MCM stabilises replication elongation complexes during the transition from replication initiation to elongation, preventing abortive replication and allowing the synthesis of full-length cRNA.

1.4.2 Hsp90 and UAP56

Host factors stimulating influenza virus polymerase in vitro were isolated from nuclear extract of uninfected HeLa cells (Momose et al., 1996; Shimizu et al., 1994). Initially called RNA polymerase activating factor (RAF) 1 and RAF2, these distinct fractions were subsequently identified as Hsp90 and UAP56, respectively (Momose et al., 2001; Momose et al., 2002).

Hsp90 is an abundant heat shock protein that functions as a molecular chaperone and contributes to the correct folding, activation and assembly of a wide variety of transcription factors, steroid receptors and protein kinases (Pearl and Prodromou, 2006; Picard, 2002). Hsp90 was found to interact with PB2 in absence or presence of vRNA, but its RdRp

1 This figure was prepared prior to publication of a key structural paper (Carrique et al. 2020, in press), which shows the ANP32A LRR is actually used by influenza virus to stabilise a replication-competent RdRp dimer. This model will be described in detail in Chapters V and VI of this thesis.

25 stimulatory activity was mapped to the acidic middle region of Hsp90, not the N-terminal chaperone domain. Normally exclusively cytoplasmic, Hsp90 relocalised to the nucleus in influenza virus-infected cells. Exogenous expression of PB2 or the heterotrimeric RdRp in HeLa cells also resulted in nuclear relocalisation of Hsp90 (Naito et al., 2007). Hsp90 was also found to interact with PB1. Treatment of cells with Hsp90 inhibitors resulted in impaired viral growth and reduced the nuclear levels of RNPs, due presumably to selective degradation of PB1 and PB2 in the absence of functional Hsp90 (Chase et al., 2008).

RAF2/UAP56 is a member of the DExD/H-box RNA helicase family and plays critical roles in healthy cells in pre-mRNA splicing and mRNA nuclear export (Luo et al., 2001; Shen, 2009). The ATPase and unwinding activities of UAP56 are required for spliceosome assembly and maturation (Shen et al., 2008; Shen et al., 2007). Furthermore, UAP56 forms part of the transcript export (TREX) complex for efficient export of spliced mRNAs to the cytoplasm (MacMorris et al., 2003).

UAP56 was shown to interact with H1N1 PR8 NP by yeast two-hybrid assay. UAP56 stimulated vRNA synthesis in in vitro assays; interestingly the helicase activity was not be required for this. Furthermore, although interaction of UAP56 with free NP was demonstrated, it did not bind RNA-associated NP. The N-terminal 20 amino acids of NP were found to be sufficient for UAP56 binding. As this is also where the NLS of NP is located it is possible that IMPα and UAP56 bind competitively to the same motif on NP. The proposal is that UAP56 acts as a chaperone for NP, facilitating formation of NP-RNA complexes. Indeed, chaperones are important for nucleic acid binding basic proteins such as histones, which tend to aggregate and become inactive in the absence of appropriate substrates (Momose et al., 2001).

UAP56 stimulates the elongation activity of the viral polymerase, possibly by facilitating the encapsidation of nascent cRNA, which is degraded by host cellular nucleases unless it is stabilised by newly synthesised RdRp and NP (Vreede et al., 2004) (Kawaguchi et al., 2011). In doing this it might work in conjunction with MCM (Kawaguchi and Nagata, 2007). Accumulation of vRNA and cRNA was reduced and delayed in UAP56 knockdown cells. Thus UAP56 facilitates replication reaction-coupled encapsidation of the nascent viral RNA as an NP molecular chaperone (Kawaguchi et al., 2011). Again, the ATP-dependent RNA helicase activity of UAP56 was not required for the encapsidating function.

Binding stoichiometry analysis showed that the NP-UAP56 complex consists of NP and UAP56 at a 3:1 molar ratio, indicating that UAP56 interacts predominantly with trimeric NP. The molecular weight of the complex on the gel filtration column – around 440 kDa – suggested that the NP-UAP56 complex consists of 6 molecules of NP and 2 molecules of UAP56 (Hu et al., 2017). This suggests a model in which a UAP56 dimer binds to two trimeric

NP complexes, stimulating transfer of NP trimers onto viral RNA as a molecular chaperone. Indeed, trimeric NP has higher RNA binding activity than monomeric NP (Tarus et al., 2012). Atomic force microscopy (AFM) analysis of NP-UAP56 showed a dumbbell-shaped complex considered as two trimeric NP units connected by dimeric UAP56.

Although not required for its chaperone function, the helicase function of UAP56 (and its paralogue URH49) may serve the purpose of preventing double-stranded RNA formation in the cell during influenza virus infection, which would be favourable for the virus to avoid innate immune sensing by PRRs. Indeed, virus infection of UAP56-depleted cells led to the rapid accumulation of dsRNA in the perinuclear region as well as activation of the dsRNA-dependent protein kinase R (PKR) which is part of the antiviral response. Although it cannot be ruled out that accumulation of viral dsRNA occurred due to aggregation of NP in the absence of its chaperone UAP56, UAP56 may thus be utilised directly by influenza virus to prevent the formation of dsRNA (Wisskirchen et al., 2011). UAP56 was also found to interact with NS1 but the biological significance of this association remains unclear (Chiba et al., 2018). URH49 shows an identity of 90% at the amino acid level with UAP56 and also interacts with NP. Interestingly, depletion of either URH49 or UAP56 by RNAi led to increased expression levels of the paralogue, and URH49, like UAP56, is involved in the nuclear export of viral mRNA (Wisskirchen et al., 2011).

1.4.3 ANP32 proteins

Acidic (leucine-rich) nuclear phosphoproteins of 32 kDa, a small family comprising ANP32A, ANP32B and ANP32E, fulfil many cellular functions including the regulation of gene expression, cell death and intracellular transport (reviewed in (Reilly et al., 2014)). All of the ANP32 proteins associate with histones, albeit in different ways (Kleiner et al., 2018; Obri et al., 2014; Saavedra et al., 2017; Tochio et al., 2010) and they have been linked to the replication cycles of a variety of viruses including paramyxoviruses and retroviruses (Bodem et al., 2011; Günther et al., 2020; Wang et al., 2019b). ANP32 proteins have an N-terminal leucine-rich repeat (LRR) domain and a low-complexity acidic region (LCAR) (Huyton and Wolberger, 2007; Tochio et al., 2010) (Figure 1.6) and are believed to have histone acetyl transferase (HAT) inhibitory activity, placing them at the chromatin within the nucleus.

Figure 1.6 Structures of the ANP32 protein LRR domains (A) Structure of the ANP32A leucine rich repeat (LRR) domain (residues 1-149) obtained by X-ray crystallography (PDB 2JE0; from Huyton and Wolberger, Protein Science 2007) (B) Structure of the ANP32B LRR domain (residues 1-161) obtained by nuclear magnetic resonance (NMR) spectroscopy (PDB 2RR6; from Tochio et al. Journal of Mol Biol 2010). The close structural resemblance between the paralogues is evident, as is the characteristic concave solenoid structure with five parallel beta sheets flanked by alpha helical cap structures.

ANP32A and ANP32B proteins have been identified as host factors for influenza replication in proteomics, RNAi and CRISPR and functional screens (Bradel-Tretheway et al., 2011; Li et al., 2020; Watanabe et al., 2014). ANP32A (pp32) and ANP32B (APRIL) were purified from nuclear extract of uninfected HeLa cells and shown to support in vitro synthesis of vRNA on a cRNA template (Sugiyama et al., 2015), suggesting they might have a role in vRNA synthesis on a cRNA template.

ANP32 proteins have been implicated as key host factors that determine host range restriction of avian influenza viruses in mammalian cells. A functional screen led to the discovery that avian ANP32A, which carries an additional 33 amino acids between the LRR and the LCAR domains, is able to rescue activity of avian-signature (PB2-627E) RdRp in human cells (Long et al., 2016). This finding addressed a long-standing debate about whether restriction of avian influenza polymerase in mammalian cells was due to a positive factor in avian cells (Moncorgé et al., 2010) or a restriction factor in mammalian cells (Mehle and Doudna, 2008; Weber et al., 2015). The E627K mutation often seen in adaptation of avian influenza viruses to replication in mammalian cells allows the RdRp to co-opt the shorter mammalian ANP32 proteins, which lack the 33-amino acid insertion. Artificial insertion of the avian ANP32A-specific 33 amino acids into human ANP32A or ANP32B enabled avian RdRp activity when co-expressed in human cells. ANP32A has since been shown to interact directly with influenza RdRp and localise to the cell nucleus (Baker et al., 2018; Domingues and Hale, 2017; Long et al., 2019a;

Mistry et al., 2020). In vitro binding assays with purified RdRp components showed that this interaction maps to the 627 domain of PB2 but is independent of the nature of the amino acid at position 627, thus interaction alone does not explain the host specificity phenotype (Baker et al., 2018).

Twenty-seven of the thirty-three additional amino acid residues in chicken ANP32A derive from a gene duplication event in the flighted bird ancestor, but the six remaining amino acids are unique to avian ANP32A and contain a SUMO interaction motif-like sequence (SIM) (Domingues and Hale, 2017). The SIM confers increased binding to the RdRp complex, and is key for the ability to complement avian RdRp. Different bird species encode one or more splice variants of ANP32A, containing either 33, 29 or no additional amino acids. ANP32A lacking the SIM (ANP32A29) was shown to support avian RdRp but to a lesser extent than the variant containing the full 33 amino acids (Baker et al., 2018). Interestingly, ratites like the ostrich lack the gene duplication and thus the additional 33 amino-acid sequence. In support of the notion that the PB2-E627K adaptation facilitates use of shorter mammalian ANP32 proteins, viral replication in ostriches selected for this or other humanizing PB2 mutations (Shinya et al., 2009).

The ratios of ANP32A splice variants (full-length including the 33 amino-acid insertion, 29 amino acids lacking the putative SIM, or mammalian-like lacking the whole 33 amino acids) vary among avian species with passerine species like swallows, blackbirds and magpies encoding a much larger proportion of the shorter (mammalian-like) splice variants than the migratory reservoir species like ducks and geese normally considered IAV carriers. Competitive replication assays in the presence of mixtures of ANP32A splice variants suggested different splice variants can drive mammalian adaptations in the PB2 gene, and mathematical modelling supported the concept that ANP32A splice ratios in birds could be harnessed to track and predict acquisition and maintenance of mammalian RdRp adaptations (Domingues et al., 2019).

A split luciferase complementation assay was used to demonstrate robust interaction between chicken (chA) or human ANP32A (huA) with IAV RdRp and it was confirmed that chA interacts much more strongly with the RdRp than huA (Mistry et al., 2020). Bimolecular fluorescence complementation (BiFC) assays showed the interaction between the RdRp and ANP32A occurs in the cell nucleus, placing the host factor at the site of influenza replication. Removal of the PB2 627 domain resulted in decreased binding of chA but not huA, suggesting the sequences within the 33 amino-acid insertion in chA are responsible for the increased binding to the RdRp. Indeed, removal of the 33 amino acids, as well as removal of the first four amino acids containing the putative SIM, strongly reduced the interaction and likewise abolished

29 avian-signature RdRp activity. Binding of ANP32A to RdRP was affected by the presence of viral-like RNA molecules of various lengths, binding of which by the RdRp is known to result in major conformational shifts (Thierry et al., 2016). Binding affinity to ANP32A was increased if the RdRp was inactive, but binding of active RdRp to ANP32A decreased in presence of RNA template, suggesting that ANP32A does not interact with actively synthesising RdRp. Mutations in the vRNA promoter have been shown to enable avian-origin RdRp activity in human cells, in the absence of mammalian adaptations in PB2 (Crescenzo-Chaigne et al., 2002; Neumann and Hobom, 1995). ANP32 proteins were still essential for replication of this altered RNA however (Mistry et al., 2020).

Solution state NMR spectroscopy combined with quantitative ensemble analysis was used to analyse complexes formed between the avian and human ANP32A LCAR domain and the avian (PB2-627E) or human-adapted (PB2-627K) IAV polymerase. It was found that the unique avian hexapeptide containing the SIM-like motif likely interacts specifically with the linker between the 627 and NLS domains of PB2. In this study the E627K mutation did confer tighter binding with the negatively charged human ANP32A LCAR, as it completes a continuous ridge of solvent-exposed positively charged residues available for interaction with the ANP32A LCAR. Bearing in mind that in other studies, binding differences between 627E and 627K RdRp to human ANP32 proteins were not observed, it is still not clear whether interaction alone explains the host range mutation or whether the differences in binding can only be observed using methodologies. Interestingly, the authors observed an electrostatic interaction between the known pro-viral aspartate at position 130 of ANP32A and the basic arginine residue R646 in the 627 domain of PB2 (Camacho-Zarco et al., 2020). Overall the interactions were shown to be highly flexible and polyvalent. Any role of the LRR domain in binding the RdRp remains unclear2.

CRISPR/Cas9-mediated knockout of human ANP32A and ANP32B carried out in 293T cells or low-ploidy eHAP cells demonstrated that ANP32A and ANP32B are functionally redundant essential host factors for influenza virus polymerase activity and replication in human cells (Staller et al., 2019; Zhang et al., 2019a). Deletion of either ANP32A or ANP32B had negligible effects on virus replication, but replication was severely restricted in cells lacking both paralogues (double knockout). Residues 129 and 130 at the C-terminal end of the LRR have been heavily implicated in pro-influenza virus function (Long et al., 2019a; Staller et al., 2019; Zhang et al., 2019a). Chicken ANP32B, mouse ANP32A and human ANP32E lack pro- influenza viral function due to amino acid substitutions in the proviral dyad 129N-130D. Thus

2 As mentioned in footnote 1, a key structural paper currently in press (Carrique et al. Nature 2020) suggests a potential role for the LRR domain. The details and ramifications of this paper will be discussed in detail in chapter 5.

30 the redundancy for proviral activity of ANP32 proteins seen in human cells is not present in mice or chickens. Indeed a conditional ANP32B knockout mouse model was shown to resist IAV-induced morbidity and mortality (Beck et al., 2020), despite the presence of wildtype ANP32A in the animals. A recent preprint identified both homozygous and heterozygous individuals for the single nucleotide variant (SNV) ANP32B-D130A, in the Latino population of the gnomAD database (Staller et al., 2020b). This variant might exert a dominant negative effect over both the wildtype ANP32B allele and the functionally redundant wildtype ANP32A alleles, which would mean carriers of the SNV have some protection against influenza virus infection.

It has recently been shown that swine ANP32A (swA), uniquely among mammalian ANP32 proteins, is somewhat capable of supporting avian-signature polymerases in mammalian cells (Peacock et al., 2020b; Zhang et al., 2020). This observation supports the notion of pigs as a ‘mixing vessel’ or intermediate host in the generation of pandemic viruses: avian influenza viruses may be capable of just enough replication in the pig host to obtain the necessary mutations for successful zoonosis into other mammalian species. Overexpression experiments in human cells lacking ANP32A and ANP32B expression (dKO) with ANP32A and ANP32B proteins from a variety of natural mammalian host species including human, swine, dog, horse, seal and bat showed that swA had a unique capacity to rescue polymerase activity and viral replication of a wide variety of avian IAV subtypes (H7N9, H7N2, H5N1, H1N1), as well as swine-origin 2009 pH1N1 and Eurasian avian-like isolates from pigs (Peacock et al., 2020b; Zhang et al., 2020). Reciprocal mutations in human and swine ANP32A showed that pro-avian polymerase activity mapped to residues 106 and 156, which are valine and serine in swA but isoleucine and proline in other mammalian ANP32A proteins. These substitutions result in increased binding of swA to avian-signature influenza RdRp (Peacock et al., 2020b; Zhang et al., 2020). Although swine ANP32B does have the ability to support RdRp activity, it would appear that in swine cells ANP32A is the dominant proviral factor: CRISPR/Cas9-mediated ablation of swA in pig epithelial PK15 cells resulted in a sharp reduction in IAV H7N9 and H9N2 polymerase activity as well as reduced H9N2 titres. Complementation in PK15 AKO cells with huA-I106V/P156S double mutant increased, while swA-V106I/S156P double mutant reduced H9N2 titres (Zhang et al., 2020). The 2009 pandemic virus had its immediate precursor in swine. A PA-N321K substitution found in second and third wave 2009 pH1N1 isolates was likely selected as a direct adaptation to human ANP32 proteins. Direct comparison of the activity of polymerase constellations from a pair of first- (PA-321N) and third-wave (PA-321K) pH1N1 isolates in human and pig cells revealed that human ANP32A has a greater capacity to boost activity of the latter constellation, compared with swine ANP32A (Peacock et al., 2020b).

Chicken cells lacking ANP32A expression (chA KO) or expressing a mammalian-like ANP32A protein lacking the 33-amino acid insertion resulted in restriction of 627E but not 627K polymerase (Long et al., 2019a; Park et al., 2020). As chicken ANP32B and ANP32E transcripts were readily detected in these cells it is clear that these family members could not support IAV RdRp activity (Long et al., 2019a). The chicken Anp32B gene encodes isoleucine and asparagine at positions 129 and 130 and this is the reason why it lacks pro-influenza virus RdRp function. Split luciferase complementation and co-immunoprecipitation assays show that binding of chB to the viral polymerase is abrogated due to this sequence difference (Long et al., 2019a; Zhang et al., 2019a). In fact the N129I mutation alone is sufficient to break interaction and pro-viral activity of ANP32 proteins from a variety of relevant species including humans and pigs (Peacock et al., 2020b).

Many avian H7N9 and H9N2 viruses acquire the PB2-E627K substitution when passaged in MDCK cells or mice, but some H9N2 isolates do not. Recombinant viruses in an H7N9 background sensitive to 627K acquisition were generated with single genome segments from an H9N2 virus that did not evolve 627K. The gene driving the emergence of PB2-E627K in mice was found to be PA: the H7N9 virus with H9N2 PA did not acquire 627K in a mouse passage experiment while the H7N9 PA conferred low RdRp activity in human cells. This phenomenon was mapped to four highly conserved residues in the N-terminal domain of PA (Liang et al., 2019). H7N9 viruses were further found to replicate poorly in human ANP32A knockout cells and mice, which is surprising as knockout of both ANP32A and ANP32B is normally required in human cells in order to see reduced H1N1, H3N2 and humanized H5N1 titres (Staller et al., 2019; Zhang et al., 2019a), while murine ANP32A has been shown to lack pro-influenza viral activity (Staller et al., 2019; Zhang et al., 2019a).

To sum up, ANP32 proteins are essential proviral factors that support the replication of influenza virus RNAs by the RdRp. Differences in ANP32 sequences in different species play important roles in host range restriction, and these proteins represent perhaps the most promising host targets for future influenza control strategies.

1.4.4. Non-proteinaceous host factors support replication

Long non-coding RNAs (lncRNAs) are transcripts of >200 nt with poor coding capacity that nonetheless fulfil important functions in diverse cellular processes. In addition, many lncRNAs play a role in virus-host interactions (Meng et al., 2017). Hundreds of lncRNAs are up- or downregulated upon IAV infection, the majority of which are associated with immune responses that contribute to antiviral defence (Peng et al., 2010; Winterling et al., 2014). Some

32 interferon-independent lncRNAs promote rather than counteract influenza virus infection (reviewed in (Wang and Cen, 2020)).

In one study, RNA sequencing (RNA_seq) and real-time PCR were used to identify upregulated lncRNAs in influenza virus-infected A549 cells. Repression of the lncRNA PSMB8-AS1 by CRISPR interference (CRISPRi) led to reduced viral protein accumulation in infected cells and a slight but significant drop in virus titres (More et al., 2019). Another influenza A virus-induced interferon-independent lncRNA, lncRNA-ACOD1 (aconitate decarboxylase 1), was shown to promote viral replication indirectly through its effects on cellular metabolism (Wang et al., 2017). Knockdown of lncRNA-ACOD1 by RNAi resulted in lower H1N1 PR8 viral load in A549 cells. The authors identified glutamic-oxaloacetic transaminase (GOT2), a key metabolic enzyme, as a lncRNA-ACOD1-binding cellular protein. Virus-induced metabolic changes in wildtype cells were abolished when lncRNA-ACOD1 was depleted. The authors propose a model in which lncRNA-ACOD1 facilitates virus replication through stimulating GOT2 activity and production of its metabolites, which can subsequently be harnessed by the virus to support its energy-dependent activities.

Using microarray analysis several lncRNAs were identified that were differentially expressed during H1N1 WSN infection of A549 cells (Winterling et al., 2014). One of these, the interferon- independent large intergenic ncRNA (lincRNA) VIN (virus-inducible lincRNA), was upregulated by influenza A viruses of subtypes H7N7, H1N1 and H3N2, but not by IBV. VIN is localised in the cell nucleus and knockdown by RNAi resulted in a ten-fold decrease in WSN titre in A549 cells. However, the mechanism through which VIN supports IAV replication has not been elucidated.

A loss-of-function screen using an endoribonuclease-prepared short interfering RNA (esiRNA) library targeting human lncRNAs was carried out in 293T cells that express Gaussia luciferase (Gluc) upon infection with influenza A virus. This approach yielded two lncRNAs that are hijacked by the viral replication machinery to support efficient viral RNA synthesis (Wang et al., 2018; Wang et al., 2019a). The first of these, PA-associated noncoding RNA (lncRNA- PAAN) was specifically induced by IAV infection in multiple human cell types. Knockdown of lncRNA-PAAN resulted in attenuated viral replication and RdRp activity while also impairing association of the PB1 and PB2 subunits. Native RNA immunoprecipitation (RIP) analysis showed that lncRNA-PAAN interacted with the PA subunit of the polymerase, likely acting as its chaperone while promoting the assembly of the viral polymerase (Wang et al., 2018).

The other lncRNA, IPAN (influenza virus PB1-associated noncoding RNA), is also induced by IAV infection and believed to promote viral replication through its association with PB1, preventing its degradation (Wang et al., 2019a). IAV infection did not only increase the level

33 of IPAN but also recruited IPAN into the nucleus. IPAN silencing led to a strong reduction of NP in IPAN knockdown cells, while progeny virus titres diminished. CRISPR/Cas9-mediated depletion of lncRNA-IPAN led to a significant reduction in viral protein and RNA levels in knockout cells infected with IAV. Reduced RdRp activity in a minireplicon system suggested that IPAN specifically supported IAV RNA replication. Accelerated turnover of PB1 but not PB2 or PA in the absence of IPAN suggested that IPAN plays a role in stabilising PB1 and indeed PB1 co-precipitates with IPAN. The authors suggest a model in which IPAN associates with PB1 in the cytoplasm where it stabilises and protects the viral protein from degradation by the host machinery. PB1 then brings IPAN into the nucleus where it dissociates upon formation of the heterotrimeric RdRp complex.

1.5 Nuclear export and beyond

Once sufficient progeny vRNPs have been generated these must be exported out of the nucleus and toward the plasma membrane for budding into new virus particles. Host factors are heavily involved in both processes (Figures 1.7 and 1.8).

1.5.1 Involvement of the Raf/MEK/ERK signalling pathway

Nuclear export of progeny vRNPs does not occur constitutively but is regulated by cellular signalling pathways to ensure temporal control over vRNP migration to the cytoplasm in the later stages of the viral life cycle. (Ehrhardt et al., 2013; Eisfeld et al., 2011; Nencioni et al., 2009; Pleschka et al., 2001). The Ras-dependent Raf/MEK/ERK mitogen-activated protein (MAP) kinase signalling pathway regulates many cellular functions including proliferation, differentiation, cell metabolism and immune response. Downstream targets of the pathway may be phosphorylated directly by ERK or alternatively by ERK-activated kinases like the p90 ribosomal S6 kinases (RSK1/2) (Cargnello and Roux, 2011; Katz et al., 2007). HA accumulation in the plasma membrane at late stage of the Influenza viral cycle activates the Raf/MEK/ERK pathway triggering nuclear export of vRNPs (Ludwig et al., 2004; Marjuki et al., 2006). This ensures timely regulation of vRNP export late in the viral life cycle when sufficient glycoproteins have accumulated for budding of progeny virions. MEK inhibitors may be promising antivirals against IAV and IBV; no escape mutants were detected after multipassage use (Ludwig et al., 2004). MEK and ERK were identified separately as host factors required for influenza virus replication in a genome-wide RNAi screen (König et al., 2010).

MEK inhibition also alters the interaction of M1 with vRNPs, thus resulting in a block in the assembly of the export complex at the dense chromatin (Schreiber et al., 2020). Two phosphorylation sites on the viral NP, serine 269 and serine 392, remained unphosphorylated

34 in the presence of the MEK inhibitor CI-1040. Phosphorylation of these residues is carried out by RSK1, whose activation is mediated by virus-induced Raf/MEK/ERK pathway activity. Knockdown by siRNA of the RSK1 isoform, as well as inhibition of both RSK isoforms by the compound BI-D1870, led to nuclear retention of vRNPs and reduced virus titres in A549 cells. Knockdown of RSK2, a related kinase activated in the same way increased virus replication, which was not surprising as RSK2 is involved in antiviral response (Kakugawa et al., 2009). Since the RSK inhibitors block both isoforms, the pro-viral function of RSK1 seems dominant over the antiviral function of RSK2. Nevertheless, to gauge the usefulness of RSK inhibitors to block influenza replication, the relative expression levels of the kinase isoforms in target tissue must be carefully investigated.

Figure 1.7 Nuclear export of progeny vRNPs Reproduced from Staller & Barclay, Cold Spring Harb Perspect Med, 2020 Nuclear export of progeny vRNPs is triggered by the Raf/MEK/ERK phosphorylation cascade leading to S269 and S392 phosphorylation of NP by the ERK effector RSK1. Progeny vRNPs are trafficked to the chromatin via SC35 nuclear speckles where they pick up M1 which is held there by the clustered mitochondria protein homologue CLUH. At the chromatin the export complex is assembled proximal to the Ran guanine exchange factor Rcc1, which regenerates Ran-GTP from Ran-GDP. The soluble export receptor CRM1, also known as Exportin 1, also localises to the chromatin. . The vRNP export complex consists of the vRNP coated with M1, which is bound in turn by the viral nuclear export protein (NEP). NEP, in possession of a nuclear export signal (NES), is believed to form a bridge between the vRNP and the CRM1-Ran-GTP complex, which mediates nuclear export through interaction with the NPC.

1.5.2 Involvement of the CRM1 pathway

CRM1 is a soluble adaptor molecule that binds to leucine-rich nuclear export signals (NESs) on cargo proteins in the nucleoplasm. It then forms a trimeric complex with Ran-GTP that mediates export of the substrate (Petosa et al., 2004). Leptomycin B (LMB) specifically inhibits CRM1/Exportin1, a member of the importin β family, by covalently modifying a specific cysteine residue which is thought to interfere with the normal formation of a stable complex with the export substrate (Kudo et al., 1999).

CRM1 was identified as a host factor required for influenza virus replication in a combined Co- IP / LC/MS and RNAi screen (Watanabe et al., 2010). NP and CRM1 were found to interact in vivo and nuclear retention of both RNA-free NP and RNPs is seen in infected cells treated with LMB. Conversely, cotransfection of exogenous CRM1 resulted in a significant increase in cytoplasmic NP. The inhibitory effect of LMB on RNP nuclear export was seen with a variety of IAV strains and in multiple cell lines including primary CEFs, so use of the CRM1 export pathway is a general feature of influenza viruses (Elton et al., 2001; Watanabe et al., 2001). vRNPs are exported in conjunction with M1 and NEP. Indeed NEP interacted directly with CRM1, in a mammalian two-hybrid system, and NEP was crucial for nuclear export of vRNP complexes (Neumann et al., 2000). Although the N-terminal NES of NEP was not required for NEP binding to CRM1, mutations in the NES did lead to nuclear retention of NP, so the NES functions critically in the nuclear export of NP, and, by extension, of vRNPs (Neumann et al., 2000). NEP was also the only viral protein to interact with nucleoporins in a yeast two-hybrid system, and antibodies directed against NEP, injected directly into the cell nucleus, inhibited the cytoplasmic accumulation of free NP and RNPs. The NEP N-terminal effector domain was recognised as a bona fide NES comparable with known NESs of viral and cellular origin, and mapped to amino acid residues 11-23 (O'Neill et al., 1998). A second NES has also been proposed (Huang et al., 2013).

The daisy-chain model of nuclear export of vRNPs postulates that M1 binds directly to vRNPs with NEP acting as a bridge between M1 and CRM1 to facilitate translocation across the NPC. It is believed that the vRNP export complexes assemble at the host chromatin, bringing them in close contact to the Ran guanine exchange factor Rcc1 which is responsible for generating RanGTP and driving CRM1-dependent nuclear export (Chase et al., 2011). Recycled cellular CRM1-RanGTP complexes are thus usurped by influenza vRNPs after nucleotide exchange by Rcc1 on the chromatin. In this way influenza virus gains preferential access to host cell export machinery by locating vRNP cargo proximal to the sites of RanGTP regeneration

(Chase et al., 2011). vRNA but not cRNA was found in the relevant chromatin fraction, alongside proteinaceous RNP components, M1, small amounts of NEP, RCC1 and core histone H3. CRM1 was found to relocalise to the sites on dense chromatin where vRNP export complexes also accumulate (Chase et al., 2011).

A subset of newly synthesised M1 is imported into the nucleus via its NLSs where it interacts with the newly formed vRNPs, in particular NP (Martin and Helenius, 1991a; Whittaker et al., 1996). However, H5N1 virus M1 alone appeared incapable of interacting with the vRNP in the nucleus. Rather vRNP/M1 interaction was found to require the C-terminal domain of NEP (Brunotte et al., 2014). This leads to a slightly different model in which the C-terminal domain of NEP enhances the binding affinity of M1 to the vRNP. All in all it appears that progeny vRNPs associated with viral M1 and NEP are exported to the cytoplasm through the CRM1- dependent nuclear export pathway as vRNP-M1-NEP-CRM1 complexes.

1.5.3 The involvement of CLUH

CLUH is a PB2-binding host protein with a role in subcellular transport of vRNPs (Watanabe et al., 2010). CLUH plays a key role in the subnuclear transport of vRNPs to the assembly site of viral nuclear export complexes via nuclear speckles (Ando et al., 2016). Pandemic 2009 H1N1, seasonal H3N2 and influenza B virus titres were reduced in cells depleted of CLUH by siRNA and nuclear retention of M1, NEP and vRNPs was observed in the CLUH knockdown cells. CLUH normally functions as a mitochondrial protein, but in infected cells it translocated to the nucleoplasm and nuclear matrix where it co-localised with PB2 and M1. CLUH and M1 accumulate in SC32-positive nuclear speckles where they are joined by progeny vRNPs via interaction with PB2. They are then translocated to the chromatin-bound region where the vRNP export complexes are assembled in the presence of NEP and CRM1. Thus CLUH is involved in the subnuclear transport of progeny vRNPs to the assembly site of viral nuclear export complexes at the dense chromatin via SC35-positive speckles.

1.5.4 Progeny vRNP transport to the plasma membrane

Until recently it seemed clear-cut that vRNPs are transported from the nucleus to the plasma membrane on microtubule networks in association with Rab11A-positive recycling endosomes (RE) (Amorim, 2018; Bruce et al., 2010). Rab11A specifically marks recycling endosomes (RE) which sort and transport cargo to the apical cell membrane. Rab11A recruits molecular motors to RE through interaction with the Rab11 family interacting proteins (FIPs). Rab11- FIPs can associate with both actin and microtubules so Rab11A-RE can use multiple cytoskeletal networks for transport.

After nuclear export, progeny vRNPs accumulate near the microtubule organising centre (MTOC) adjacent to the nuclear membrane (Amorim et al., 2011; Momose et al., 2007). Live- cell imaging approaches using fluorescently tagged vRNP components have shown vRNP movement along microtubules with a characteristic intermittent saltatory motion (Amorim et al., 2011; Avilov et al., 2012). Cells treated with microtubule depolymerising agents had altered vRNP distribution and reduced viral growth. Depletion of Rab11A by RNAi and overexpression of a dominant-negative Rab11A mutant (Rab11A-GDP) impaired the association of vRNPs with Rab11A-positive vesicles, disrupted accumulation of vRNPs at the plasma membrane and sharply reduced the output of infectious progeny virus (Eisfeld et al., 2011). Several recent reports, however, have put the question to the accepted model of vRNP trafficking along microtubules in association with Rab11A.

Figure 1.8 Cytoplasmic trafficking of progeny vRNPs Reproduced from Staller & Barclay, Cold Spring Harb Perspect Med, 2020 The nuclear export complex disassembles at the perinuclear region upon Ran-GTP hydrolysis to Ran-GDP by the Ran GTPase-activating protein RanGAP, which is associated with the NPC. The preferred model for what happens next used to be association of the vRNPs with recycling endosomes (RE) through interaction with Ras-related protein Rab11A at the microtubule organising centre (MTOC). Rab11A-positive RE traffic along microtubules upon recruitment of molecular motors. More recently involvement of modified endoplasmic reticulum (ER) structures rather than RE has been proposed as driving vRNP trafficking. This model comes in various incarnations, including irregular coated vesicles (ICVs) adorned with vRNPs and viral inclusions that resemble liquid organelles containing the vRNPs. In each of these cases Rab11A is also associated with

38 the structures which would explain the colocalisation between vRNPs and Rab11A that led to the initial model. All models result in accumulation of vRNPs at the plasma membrane where they are packaged in novel virions for budding.

Discrepant results included the modest reduction in viral titres seen on depolymerisation of microtubules, and decrease in Rab11A binding of FIP adaptors upon infection, suggesting that association between Rab11A and microtubules might be compromised (Vale-Costa et al., 2016). Ultrastructural inspection of cytoplasmic sites positive for Rab11A and vRNPs, using correlative light and electron microscopy (CLEM) showed clustered vesicles of heterogeneous sizes from where coiled coil structures resembling vRNPs protruded. Similar vesicles, renamed irregular coated vesicles (ICVs) were observed more recently by electron microscopy which seemed to extend from an extensively swollen and tubulated endoplasmic reticulum (ER) (de Castro Martin et al., 2017). This suggests that it may be extensions of the ER and not the recycling endosomes that regulate vRNP transport. Rab11A was found associated with the modified ER and with ICVs as well. So this alternative model postulates that once progeny vRNPs reach the cytoplasm they first bind to a modified ER from where Rab11A-coated vesicles loaded with vRNPs are subsequently released and directed to the plasma membrane (Amorim, 2018).

Another report also suggested the existence of a Rab11A-RE independent pathway for cytoplasmic transport of vRNPs, based on the observation that vRNP subcellular location could be uncoupled from Rab11A in the presence of the microtubule depolymerisation agent nocodazole (Nturibi et al., 2017). Further supporting data for involvement of the ER rather than microtubules comes from the detection of viral inclusions forming in the vicinity of ER exit sites (ERES) (Alenquer et al., 2019). These inclusions are not membrane-bound, rather they have properties of liquid organelles and contain both vRNPs and Rab11A. Thus accumulating evidence suggests that Rab11 subcellular localisation is redirected and its function is impaired during IAV infection.

Three-dimensional movement of fluorescently tagged Rab11A-RE and IAV vRNPs was tracked in infected A549 cells using dual-view inverted selective-plane illumination microscopy (diSPIM). Although Rab11A motion was dependent on microtubules in A549 cells, depletion of microtubule filaments by nocodazole treatment had little impact on vRNP movement (Bhagwat et al., 2020). In addition, a large reduction in the amount of dynein, the minus-end directed microtubule motor, associated with Rab11A was observed in IAV-infected cells. This might be the mechanism behind altered Rab11A movement seen during IAV infection. As over half the IAV vRNP puncta moved independently of Rab11A spots and vice versa, this study

39 provides yet more evidence for a microtubule and Rab11A-independent mechanism of cytoplasmic vRNP transport.

A unifying explanation would be that influenza virus has evolved several mechanism of cytoplasmic vRNP trafficking, either involving Rab11A-RE and the microtubule network or the endoplasmic reticulum and alternative vesicles or even liquid-phase organelles.

1.6 Discussion

Cellular factors are involved in all stages of the influenza life cycle. As details of these essential virus-host interactions are unveiled, we may be given the opportunity to target them specifically, leading to novel means of intervention. Understanding how viral pathogens usurp the cell machinery is key for the development of antiviral agents. At the same time, studying how and why viruses subvert cellular proteins and pathways will give us valuable insights in the workings of the host cell itself. As RNA viruses are prone to rapid evolution, compounds aimed at virus-encoded proteins will inexorably result in the rapid emergence of drug resistance. A focus on host-directed therapy may lead to novel longer-lasting therapeutics and antiviral strategies.

II MATERIALS AND METHODS

2.1 Cells and cell culture

Human HAP1 and eHAP cells (Horizon Discovery) were cultured in Iscove’s modified Dulbecco’s medium (IMDM; Thermo Fisher) supplemented with 10% fetal bovine serum (FBS; Labtech), 1% nonessential amino acids (NEAA; Gibco), and 1% penicillin/streptomycin (Invitrogen). Human lung adenocarcinoma epithelial cells (A549) (ATCC), HEK293T cells, and Madin-Darby canine kidney (MDCK) cells (ATCC) were maintained in Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen) supplemented with 10% FBS, 1% NEAAs, and 1% penicillin-streptomycin (Invitrogen). All cells were maintained at 37°C in a 5% CO 2 atmosphere.

2.2 Plasmids and cloning

All constructs were cloned into pCAGGS expression plasmids by overlap extension polymerase chain reaction (PCR) using KOD Hot Start DNA Polymerase (Sigma) (Lee et al., 2010). Touchdown PCR was used to eliminate the need for optimisation (Don et al., 1991; Korbie and Mattick, 2008), using the following thermocycler conditions:

Step T / °C time notes HotStart Polymerase activation 95 2 min Touchdown Denature 95 20 s (12 cycles) Annealing 68-57 10 s -1°C per cycle Extension 70 Extension time depending on amplicon length Amplification Denature 95 20 s (25 cycles) Annealing 50 10 s Extension 70 Extension time depending on amplicon length Final extension 70 3-5 min

PCR products were run on 1-2% agarose in 0.5 x TAE buffer and visualised with Gel Red dye (Biotium). DNA fragments were extracted from the gel using the Monarch gel extraction kit (NEB). Restriction digestion with NotI and XhoI restriction enzymes (NEB) was carried out as per the manufacturer’s instructions and T4 DNA ligase (NEB) was used for cloning into digested pCAGGS vector. Constructs were transformed into 5-alpha chemically competent E.

41 coli (NEB) as per the manufacturer’s instructions. All constructs for this study were prepared by midi- or maxiprep using Qiagen kits and DNA concentration was established by spectrophotometry on a DeNovix DS-11 FX+ spectrophotometer. DNA sequences were verified by GATC Biotech / Eurofins Genomics and analysed in Geneious software. No Kozak sequence was added for enhanced transcription as the CAG promoter is sufficiently strong.

In detail, overlap extension PCR was carried out in three distinct stages (Heckman and Pease, 2007; Horton et al., 1993; Yon and Fried, 1989). In stage 1, starting from a plasmid encoding, for example, wildtype ANP32B cloned between NotI and XhoI restriction sites, the desired fragments were amplified using flanking FWD primers (containing either a NotI restriction site followed by the start codon, or a FLAG-tag sequence followed by two STOP codons and a XhoI restriction site) in combination with elongated REV primers that partly overlapped. In stage 2 the partly overlapping amplicons underwent touchdown PCR in the absence of primers, the overlapping sequence acting to self-prime the PCR reaction. Finally, in stage 3 the flanking primers were added and the new gene amplified for restriction digestion and cloning into its own pCAGGS expression plasmid. Overlap extension PCR was used to generate deletions, insertions, fusion genes and mutagenesis, with the desired insertion or edit incorporated in the REV primers.

Some constructs were ordered as human codon-optimised synthetic gene strings from GeneArt (Thermo Fisher). These include mouse ANP32A and ANP32B (Chapter IV), and the human ANP32B quintuple loss-of-acidity mutant (LOA) (Chapter VI). cDNAs of full-length human codon-optimized murine ANP32A and ANP32B isoforms (this thesis and (Staller et al., 2019)) were generated by gene synthesis (GeneArt, Thermo Fisher) using GenBank sequences NP_033802.2 (mouse Anp32A ) and NP_570959.1 (mouse Anp32B ) and cloned into pCAGGS expression plasmids that included a C-terminal GSG linker followed by a FLAG tag and 2 stop codons. Human pCAGGS ANP32A and ANP32B, chicken ANP32A, pig ANP32A, and duck ANP32A expression plasmids were cloned by Jason Long and have been described (Long et al., 2016; Peacock et al., 2020b).

The chimeric mouse construct and human ANP32A point mutants (Chapter IV) were cloned by overlapping PCR, using primers CCAACCTGAATGCCTACCGCGAGAAC and GTTCTCGCGGTAG GCATTCAGGTTGG (huANP32A-D130A), and GGTCACTTCGCAGTTAAACAAATCCAG, GTTTAACTGCGAAGTGACCAACAGAAGC, GCCCTCCACGTCGCTGTCAGGGGCCTC, and GACAGCGACGTGGAGGGCTACGTGGAG (mouse Anp32A/Anp32B domain swap). Mouse ANP32A mutant A130D was generated by overlapping PCR using primers

GGTAACCAACCTTAATGATTACCGGGAGAACGTC and GACGTTCTCCCGGTAATCATTAAGGTTGGTTACC. pCAGGS expression plasmids encoding each RNP component for H3N2 Vic/75, H5N1 (PB2-627K) 50-92, pH1N1 England 195, and IBV Florida 06 have been described (Jackson et al., 2002; Moncorgé et al., 2013). pPolI reporter plasmids containing firefly luciferase, blue fluorescent protein (BFP), or DsRed, flanked by IAV- or IBV-specific promoters were cloned by Olivier Moncorge and Hayley Lay of the Barclay Laboratory. All plasmid constructs were verified by Sanger sequencing and analysed manually in Geneious v6.

Human FLAG-tagged pCAGGS-ANP32A and ANP32B expression plasmids were cloned by Jason Long and have been described (Staller et al., 2019). Natural variant proteins (Chapter V) were cloned from these plasmids by overlapping touchdown PCR, using primers CCAACCTGAATAACTACCGCGAGAAC and GTTCTCGCGGTAGTTATTCAGGTTGG (ANP32A-D130N), GAATGACTACCAAGAGAACGTGTTC and GAACACGTTCTCTTGGTAGTCATTC (ANP32A-R132Q), GAGGCCCCTGATGCTGACGCCGAGG and CCTCGGCGTCAGCATCAGGGGCCTC (ANP32A-S158A), GAGGCCCCTGATACTGACGCCGAGGGC and GCCCTCGGCGTCAGTATCAGGGGCCTC (ANP32A-S158T), GTGACAAACGTGAATGACTATCGG and CCGATAGTCATTCACGTTTGTCAC (ANP32B- L128V), CAAACCTGAATGCCTATCGGGAGAGC and GCTCTCCCGATAGGCATTCAGGTTTG (ANP32B-D130A), GACTATCGGCAGAGCGTGTTTAAG and CTTAAACACGCTCTGCCGATAGTC (ANP32B- E133Q), GAGAGCGTGTTTAAGCACCTGCCACAGCTG and CAGCTGTGGCAGGTGCTTAAACACGCTCTC (ANP32B-L138H), GTTGCTGCCACAGTTTACTTATCTCGA and TCGAGATAAGTAAACTGTGGCAGCAAC (ANP32B-L142F).

The ΔLCAR and NLS constructs described in Chapter VI were cloned using primers GCCTCTCTCGTCGTAGCCGTCCAGGTAGGTCAG and GGCTACGACGAGAGAGGCCAGAAGCGC (human ANP32A ΔINHAT 1+2; construct 6), GCCTCTCTCTTCATCTTCCACCACCTG and GAAGATGAAGAGAGAGGCCAGAAGCGC (human ANP32A ΔINHAT 2; construct 5), CTCAAGAAGCAGCTGAGCACCCTGTACCTGGAAACCGACGACGAG and CAGGTACAGGGTGCTCAGCTGCTTCTTGAGCTCGCCCTTCCCACC (human ANP32B NES; construct 14), TACACCCTGCTGAGCCTGCAGAAGCTGAAGGAAACCGACGACGAG and CTTCAGCTTCTGCAGGCTCAGCAGGGTGTACTCGCCCTTCCCACC (human ANP32B NESscr; construct 15) ,

AAAAAAAAGCGGCCGCATGGAGATGAAGAAGAAG, GGGGGGGGCTCGAGTCATCACTTGTCATCGTCATCC, CTTTCTCTATAATCATTCAGGTTTGTGATCTC and GAGATCACAAACCTGAATGATTATAGAGAAAG (human ANP32E-E129N).

2.3 CRISPR/Cas9 genome editing

Pairs of guide RNAs against exon 2 of human Anp32A (GTCAGGTGAAAGAACTTGTCC and GAAGGCCCGACCGTGTGAGCG) and Anp32B (GAGCCTACATTTATTAAACTG and GCAAGCTGCCTAAATTGAAAA) were designed with the aid of the CRISPR design tool at www.crispr.mit.edu (Feng Zhang Lab) (Chapter III). The nontargeting guide RNA pair was GTATTACTGATATTGGTGGG and GAACTCAACCAGAGGGCCAA. The guides were cloned into plasmid pSpCas9n(BB)-2A-Puro (PX462) v2.0 (Feng Zhang Lab), obtained via Addgene, and equimolar amounts of plasmids were transfected using Lipofectamine 3000 (Thermo Fisher). Cells harbouring at least one plasmid were enriched by selection with puromycin at 1.5 µg/ml for 3-5 days and single-cell sorted into 96-well plates containing growth medium, using a fluorescence-activated cell sorter (FACS) Aria IIIU (BD Biosciences) with an 85-µm nozzle. Single cells were grown out into clonal populations over a period of 10 to 14 days. Genetic loci harbouring insertion/deletion mutations (indels) were amplified by PCR using barcoded primers (AGTGACGGAGTGACTGACTG and GAGGTGAGGCCTACGTTGAT for Anp32A; TGTCTTGGACAATTGCAAATCAA and CCATGTGCTTTCTGCTACACT for Anp32B). A total of 268 barcoded amplicons were then prepared for next-generation sequencing (NGS) using the NEBNext Ultra II kit (NEB) and sequenced using 150-bp paired- end reads on an Illumina MiSeq instrument. Reads were mapped using BurrowsWheeler Aligner (BWA) v0.7.5. Indels occurring above a cutoff of 2.5% of reads were detected using an R script (https://github.com/Flu1/CRISPR). DNA sequences were analysed in Geneious v6. NGS reads were deposited in the European Nucleotide Archive and NCBI BioProject database under project accession number PRJEB31093.

Guide RNA GCACTCTCTCGGTAGTCATTC was designed manually against the protospacer sequence in exon 4 of Anp32B to target DNA endonuclease SpCas9, expressed from Addgene plasmid # 62988 (PX459), to the target nucleotide (Chapter V). The guide RNA itself was cloned into Addgene plasmid # 80457 (pmCherry_gRNA). A custom-designed 88 base ssODN (single strand DNA) homology-directed repair template, harbouring the point mutation (cytosine in bold) and a silent PAM mutation (thymidine in italic script) – GAAAAGCCTGGACCTCTTTAACTGTGAGGTTACCAATCTGAATGCCTACCGAGAGAGTG TCTTCAAGCTCCTGCCCCAGCTTACCTAC – was obtained from Integrated DNA

Technologies (IDT). Equal amounts of PX459 and pmCherry_gRNA (total 1 µg), and 1.0 µl 10 µM ssODN template, were transfected by electroporation into approximately 400,000 eHAP cells using the NeonTM transfection system (Invitrogen). Cells and DNA were mixed into a 10 µl volume of suspension buffer, which was subjected to a single 1,200 Volt pulse for a duration of 40 ms. Cells were incubated at 37⁰C for 24 hours in IMDM growth medium without antibiotics. A fluorescence-activated cell sorter (FACS) Aria IIIU (BD Biosciences) with an 85- µm nozzle was used to sort cells expressing mCherry (550-650 nm emission) into 96-well plates containing growth medium. Single cells were grown out into monoclonal populations over a period of 10 to 14 days. Total genomic DNA was extracted using the PurelinkTM Genomic DNA Mini Kit (Invitrogen) and amplified by touchdown PCR to generate a 1,561- base pair fragment of the edited locus (primers TACCTCTGCCCTCTCAATCTCT and ACGCACACAAACACACACTATT). PCR products were then incubated at 65⁰C for 30 minutes in the presence of BsmI restriction enzyme (NEB). The resulting DNA fragments were separated by 1.5% agarose gel electrophoresis. Potentially successfully edited clones were verified by Sanger sequencing (primers TAAAGACCGCTTGATACCCAGG and TGAGGCTGAGTGGGTAGTGG) and analysed in Geneious prime 2019.

2.4 Immunoblot analysis

At least 250,000 cells were lysed in buffer containing 50 mM Tris-HCl (pH 7.8; Sigma-Aldrich), 100 mM NaCl, 50 mM KCl, and 0.5% Triton X-100 (Sigma-Aldrich), supplemented with a cOmplete EDTA-free protease inhibitor cocktail tablet (Roche) and clarified by centrifugation at 16,000 x g in a refrigerated centrifuge (4⁰C). Total protein concentration was established by spectrophotometry (DeNovix DS-11 FX+ spectrophotometer), and equal amounts (20-60 µg total protein per well) prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol (100 µl 14.3 M βME added to 900 µl 4x Laemmli buffer) and boiled at 95⁰C for 5 minutes to denature the proteins. Lysates were resolved by sodium dodecyl sulphate (SDS) – polyacrylamide gel electrophoresis (PAGE) on 4-20% precast gels (Bio-Rad) for 40 minutes at 200 V. Immunoblotting by semi-dry transfer (Bio-Rad Trans-Blot SD semidry transfer cell) onto nitrocellulose membranes (Amersham Protran Premium 0.2 µm NC; GE Healthcare) was carried out using the following primary antibodies: rabbit α-vinculin (catalogue number ab129002, 1/2,000; Abcam), rabbit α-ANP32A (catalogue number ab51013, 1/500; Abcam), rabbit α-ANP32B (10843-1-AP, 1/1,000; Proteintech), mouse α-FLAG (catalogue number F1804, 1/500; Sigma-Aldrich), mouse α-IAV NP (catalogue number ab128193, 1/1,000; Abcam), mouse α-IBV NP (catalog number ab20711, 1/1,000; Abcam), rabbit α-PB1 (catalogue number GTX125923, 1/500; GeneTex), rabbit α-PA (catalogue number GTX118991, 1/500; GeneTex), and rabbit α-IAV PB2 (catalogue number GTX125926,

1/2,000; GeneTex). The following secondary antibodies were used: sheep α-rabbit horseradish peroxidase (HRP) (catalogue number AP510P, 1/10,000; Merck) and goat α- mouse HRP (STAR117P, 1/5,000; AbD Serotec). Protein bands were visualized by chemiluminescence using SuperSignalTM West Femto substrate (Thermofisher Scientific) on a Fusion-FX imaging system (Vilber Lourmat).

2.5 Minireplicon assays

In order to measure influenza virus polymerase activity, pCAGGS expression plasmids encoding H3N2 Vic/75, pH1N1 Eng/195, H5N1 5092, or IBV Florida 06 PB1 (0.02 µg), PB2 (0.02 µg), PA (0.01 µg), and NP (0.04 µg) were transfected into 100,000-200,000 eHAP cells using Lipofectamine 3000 (Thermo Fisher) at ratios of 2 µl P3000 reagent and 3 µl Lipofectamine 3000 reagent per µg plasmid DNA. As reporter constructs, we transfected 0.02 µg pPolI plasmid, which encodes a negative-sense minigenome containing a bioluminescent (firefly luciferase) or fluorescent (BFP; DsRed) protein reporter flanked by conserved influenza A or B virus promoter sequences. pCAGGS-Renilla luciferase (0.02 µg) was co-transfected as a transfection and toxicity control. Amounts of co-transfected ANP32-FLAG constructs were 0.04 ug (equal to pCAGGS-NP) unless otherwise specified in the Figure legends. Twenty-four hours after transfection, cells were lysed in 50 µl passive lysis buffer (Promega) for 30 minutes at room temperature with gentle shaking. Bioluminescence generated by firefly and Renilla luciferases was measured using the dual-luciferase system (Promega) on a FLUOstar Omega plate reader (BMG Labtech).

2.6 Fluorescence microscopy

Approximately 100,000 eHAP cells were cultured on sterilised glass coverslips and transfected as per minigenome reporter assay protocol or infected as described in the Figure legend. Twenty-four hours after transfection, cells were fixed in 4% paraformaldehyde and permeabilized in 0.2% Triton X-100. FLAG-tagged ANP32 constructs were visualised with primary antibody mouse α-FLAG (F1804; 1/200; Sigma) for 2 hours at 37⁰C in a humidified chamber. Cells were incubated with secondary antibody goat α-mouse Alexa Fluor-568 (1/200; Life Technologies) for 1 hour at 37⁰C in a humidified chamber, and counterstained with DAPI. Coverslips were mounted on glass slides using Vectashield mounting medium (H-1000- 10; Vector Laboratories). Cells were imaged with a Zeiss Cell Observer widefield microscope with ZEN Blue software, using a Plan-Apochromat x100 1.40-numerical aperture oil objective (Zeiss), an Orca-Flash 4.0 complementary metal-oxide semiconductor (CMOS) camera

(frame, 2,048 x 2,048 pixels; Hamamatsu), giving a pixel size of 65 nm, and a Colibri 7 light source (Zeiss). Channels acquired and filters for excitation and emission were 4’,6-diamidino- 2- phenylindole (DAPI) (excitation [ex], 365/12 nm, emission [em] 447/60 nm), and TexasRed (ex 562/40 nm, em 624/40 nm). All images were analysed and prepared with Fiji software.

2.7 Influenza virus infection

Cells were infected with virus diluted in serum-free IMDM or DMEM at 37°C (MOI as indicated in the text or relevant figure legends) and replaced with serum-free cell culture medium supplemented with 1.0 µg/ml L-1-tosylamide-2-phenylethyl chloromethyl ketone (TPCK) trypsin (Worthington-Biochemical) after 1 to 2 h. Cell supernatants were harvested at indicated time points post-infection. Infectious titres were determined by plaque assay on MDCK cells. All the viruses used in this work – H1N1 A/Puerto Rico/8/34 (PR8); 2009 pandemic H1N1 A/England/195/2009 (Eng/195); H5N1 A/turkey/England/50-92/91 (5092); H3N2 A/Victoria/3/75 (Vic/75); B/Florida/4/2006 (IBV Florida/06); H5N1 A/turkey/Turkey/1/2005 (Tky/05) – were available in the Barclay laboratory.

In detail, inoculum is left on the cells for one hour to allow the virions to adsorb and enter the cells. Cells are washed in PBS after inoculation and an acid wash step (PBS-HCl at pH 4.0) is included to inactivate any leftover infectious virus. Failure to include an acid wash can lead to false positives where infectious virus is measured at the first time point post-infection, where in reality this is infectious virus that was left over from the inoculum. Virus replication is allowed to proceed as cells are incubated at a relevant temperature depending on the provenance of the virus strain, usually either 33⁰C (the temperature of the upper respiratory tract (URT) of humans), 37⁰C (human lower respiratory tract (LRT), or 42⁰C, the temperature of the avian gastrointestinal tract where avian influenza viruses normally replicate.

Aliquots of supernatant containing free infectious virus are taken at various time points post- infection. Most influenza viruses peak between 24 and 48 hours, after which the titre generally plateaus or diminishes slightly. Titres (measured as PFU per ml) are obtained by plaque assay on MDCK cells, which are used for their susceptibility to influenza viruses, fast growth, formation of tight monolayers, and propensity to form clear plaques that are easily counted.

After inoculation of MDCK cells the inoculum is aspirated and a nutritious overlay containing agarose is added to the wells instead of liquid growth medium. The agarose will set and this means the virus can only spread laterally through the cell monolayer, forming distinct plaques. After 72 hours the overlay is removed and surviving cells are stained with crystal violet dye.

Foci of dead cells (plaques) will be highlighted as round white circles surrounded by a purple monolayer.

2.8 Viral RNA quantitation

Total RNA from ~200,000 PR8-infected eHAP cells was extracted using the RNeasy minikit (Qiagen), with 30 minutes of on-column DNase I treatment (Qiagen). RNA concentrations were established by spectrophotometry (DeNovix DS-11 FX+), and equal amounts (500 ng) were subjected to cDNA synthesis using RevertAid reverse transcriptase (Thermo Scientific). PR8 segment 6 (neuraminidase [NA]) RNA species (vRNA, cRNA, and mRNA) were isolated using 5’-tagged primers GGCCGTCATGGTGGCGAATGAAACCATAAAAAGTTGGAGGAAG, GCTAGCTTCAGCTAGGCATCAGTAGAAACAAGGAGTTTTTTGAAC, and CCAGATCGTTCGAGTCGTTTTTTTTTTTTTTTTTGAACAGACTAC, respectively (tags underlined). Unique fragments of the NA gene were then amplified by real-time quantitative PCR using Fast SYBR green master mix (Thermo Scientific), using the following primers: GGCCGTCATGGTGGCGAAT and CCTTCCCCTTTTCGATCTTG (vRNA,148 bp), CTTTTTGTGGCGTGA ATAGTG and GCTAGCTTCAGCTAGGCATC (cRNA, 108 bp), or CTTTTTGTGGCGTGAATAGTG and CCAGATC GTTCGAGTCGT (mRNA, 87 bp) Quantitative PCR analysis was carried out on a Viia 7 real-time PCR system (Thermo Fisher). Gene expression was calculated by normalizing target gene expression to threshold cycle (CT) values obtained in the mock-infected condition.

2.9 Split luciferase complementation assay pCAGGS expression plasmids encoding H5N1 5092, IBV Florida/06, or H3N2 Vic/75 PB1- luc1, PB2, PA, and the indicated ANP32-luc2 construct were transfected into ~100,000 293T cells at a ratio of 1:1:1:1 (15 ng per well). 5092 PB1-luc1 and Vic/75 PB1-luc1 constructs were cloned by Bhakti Mistry. IBV Florida/06 PB1-luc1 (Chapter IV) was cloned for this thesis using primers AAAAAAAAGCGGCCGCATGAATATAAATCCTTATTTTCTC, CGCGGCTGCTGTGTACCCAATCTCACC, and GGGTACACAGCAGCCGCGGGCGGGGGAGGC. Control conditions contained pCAGGS- luc1 and untagged PB1, or pCAGGS-luc2 and untagged ANP32A, respectively, with all other components remaining constant. pCAGGS-luc1 and pCAGGS-luc2 were cloned by Bhakti

Mistry. Expression plasmids pCAGGS-ANP32Bluc2, ANP32A-D130Nluc2, ANP32B-L128Vluc2, and ANP32B-D130Aluc2 were cloned by overlapping touchdown PCR for this thesis (Chapter V), using primers CGCGGCTGCGTCATCCTCTCCCTCGTCGTC and GAGGATGACGCAGCCGCGGGCGGGGGAGGC (ANP32B), or

CGCGGCTGCGTCGTCATCCTCGCCCTCGTC and GATGACGACGCAGCCGCGGGCGGGGGAGGC (ANP32A).

The mouse ANP32-luc2 constructs (Chapter IV) were cloned for this thesis using primers CGCGGCTGCATCGTCCTCTTCGCCTTCATC and GAGGACGATGCAGCCGCGGGCGGGGGAGGC (MusA-luc2), CGCGGCTGCGTCATCTTCCCCTTCGTCATC and GAAGATGACGCAGCCGCGGGCGGGGGAGGC (MusB-luc2), and CGCGGCTGCATCGTCCTCTTCGCCTTCATC and GAGGACGATGCAGCCGCGGGCGGGGGAGGC (Mus chimera-luc2).

Core histones H3.1 and H4 were ordered as synthetic constructs (GeneArt, Thermo Fisher) and have been described, along with the pCAGGS H4-luc1 construct cloned by Jason Long (Long et al., 2019a). The pCAGGS H3.1-luc1 construct was cloned for this thesis using primers AAAAAAGCGGCCGCGCCACCATGGCTCGTACG, CGCGGCTGCTGCCCTTTCCCCACGGATGCG, GAAAGGGCAGCAGCCGCGGGCGGGGGAGGC and CCCCCCCTCGAGTTATCAGCCTATGCCGCCCTG.

Empty pCAGGS plasmid was used to ensure total transfected DNA was equal across conditions. Twenty-four hours after transfection, cells were lysed in 50 µl Renilla lysis buffer (Promega) for 1 h at room temperature with gentle shaking (Gaussia and Renilla luciferase share the same substrate). Bioluminescence generated by Gaussia luciferase was measured using the Renilla luciferase kit (Promega) on a FLUOstar Omega plate reader (BMG Labtech). Normalized luminescence ratios (NLR) were calculated by dividing the signal from the potential interacting partners by the sum of the two controls, as described in the main text.

2.11 Structural modelling

To illustrate a chimeric construct with the LRR5 from murine ANP32B in murine ANP32A (Figure 4.14 D) we created a homology model of MusB obtained using iTASSER structural prediction software (based primarily on huANP32B [GenBank accession number 2RR6A], huANP32A [accession number 2JQDA], and 2JEOA). The three-dimensional structural model was visualized and created in UCSF Chimera; the LRR is shown in dark grey and the structurally unresolved LCAR in semi-transparent grey. Amino acid residues 128 to 153 are highlighted in blue and residue 130 is in red stick format. Structural models of ANP32A and B (Figure 5.1 B and C) were created using iTASSER structural prediction software (based primarily on huANP32B [GenBank accession number 2RR6A] and huANP32A [accession number 2JQDA], and 2JEOA). The three-dimensional structural models were visualized and

49 created in UCSF Chimera. Amino acid residues affected by selected SNVs are highlighted in purple (ANP32A) or blue (ANP32B) stick format.

2.12 Safety/biosecurity All work with infectious agents was conducted in biosafety level 2 facilities, approved by the Health and Safety Executive of the United Kingdom and in accordance with local rules, at Imperial College London, United Kingdom.

III GENERATING HUMAN CELL LINES LACKING ANP32 EXPRESSION USING CRISPR/CAS9 TECHNOLOGY

Previous work in our laboratory showed that chicken ANP32A was a host factor that could support avian signature FluPol activity in human cells (Long et al., 2016). The logical next step was to ask whether the human orthologues ANP32A and ANP32B were required for influenza virus replication in human cells. Long et al. had already used an RNA interference approach to reduce ANP32 expression and shown a reduction in FluPol activity and virus replication in cells expressing less ANP32A, ANP32B, or both. In knockdown cells residual protein may still carry out its function, and indeed FluPol was still somewhat active in knockdown cells (Long et al., 2016). Complete ablation of expression using knockout by CRISPR/Cas9 genome engineering would be a more robust method to look for a phenotype in cells lacking these proteins.

3.1 CRISPR/Cas9 technology CRISPR-Cas (clustered regularly interspaced palindromic repeats – CRISPR-associated protein) systems provide adaptive immunity in many bacteria and archaea, which are under constant threat of bacteriophage infection (Hampton et al., 2020; Puschnik et al., 2017). In this natural system, fragments of phage DNA are integrated as spacers into a CRISPR array on the bacterial chromosome, a process called acquisition. The spacers are interspersed by repeat sequences and form a memory of invading foreign genetic elements. In the expression step a CRISPR RNA (crRNA) containing the unique spacer is transcribed alongside a transactivating RNA (tracrRNA) and one or several Cas DNA endonucleases. The crRNA forms a hybrid with the tracrRNA in the interference stage (i.e. upon reinfection with the same pathogen) and this hybrid guides the Cas endonuclease(s) to the invading DNA through complementarity between the spacer in the crRNA and the phage DNA. This target sequence is the protospacer, which is cleaved only in the presence of a protospacer-adjacent motif (PAM) in the phage DNA. This sequence – NGG for Cas9 – is essential for recognition and cleavage of the foreign DNA and for distinguishing self from non-self nucleic acid (Figure 3.1).

Figure 3.1. Bacterial CRISPR-mediated immunity and its adaptation to mammalian genome engineering (from Puschnik et al. 2017 Nat Rev Microbiol) The bacterial CRISPR adaptive immunity system consists of three distinct steps: acquisition, expression, and interference. Invading phage DNA is stored as memory, and if the bacterium survives the attack and the same phage attempts to infect it again, the crRNA, which is complementary to the phage DNA, is expressed together with a tracrRNA and Cas endonucleases. These three components form a complex that targets and cleaves the phage DNA with high specificity. The inset shows how the crRNA-tracrRNA hybrid has been modified as a single guide (sgRNA), which targets a human codon-optimized Cas9 protein bearing a nuclear localisation signal (NLS) to the genomic DNA. Presence of a protospacer-adjacent motif (PAM) and design of a 20-nucleotide complementary RNA are sufficient for Cas9 to generate a blunt double- stranded break (DSB) in the target DNA, three base pairs upstream of the PAM. Cellular repair of the damaged DNA leads to either gene knockout or precise edits.

The potential of this system as an RNA-programmable genome editing tool was first recognized at the University of California Berkeley, where a single guide RNA (sgRNA) was engineered from the crRNA-tracrRNA hybrid to demonstrate site-specific (plasmid) DNA cleavage (Jinek et al., 2012). It was further found that the Cas9 endonuclease isolated from the bacterium Streptococcus pyogenes possesses two distinct endonuclease domains. The HNH-like domain cleaves the DNA strand that is complementary to the guide RNA while the RuvC-like domain cleaves the noncomplementary strand. This leads to a blunt double- stranded break in the target DNA, three base pairs upstream of the PAM. In a follow-up paper

52 the same authors showed that expression of a human codon-optimized version of Cas9 and sgRNA can lead to editing of a human gene in HEK293T cells (Jinek et al., 2013).

The cell will react to the DNA damage in one of two main ways. If the cell is undergoing mitosis (M phase of the cell cycle) it can repair the DSB by the high-fidelity homologous recombination (HR) pathway, using the sister chromatid as the repair template (Moynahan and Jasin, 2010). This pathway can be exploited in the laboratory by providing an exogenous repair template carrying a desired edit. Alternatively, cells can use the non-homologous end joining (NHEJ) pathway for DSB repair. This pathway is active throughout the cell cycle and is more flexible, leading to frequent insertion and deletion mutations (indels) at the repaired DNA junctions (Chang et al., 2017). Inside coding regions this can randomly lead to a shifted reading frame and premature termination codon (PTC), which can lead to gene knockout (KO). Thus, one can aim for precise edits or gene KO through provision (or omission) of an exogenous repair template alongside the sgRNA/Cas9 complex (Figure 3.2).

Figure 3.2 Possible outcomes of DSB repair (from Ran et al. 2013 Nature Protocols) The DNA damage wrought by Cas9 can be repaired by the non-homologous end joining (NHEJ) pathway (left) or by homology-directed repair (HDR; right). The former will lead to frequent indel mutations in the target DNA and gene knockout; the latter can be used for precise editing through provision of a repair template bearing the desired edits.

The main issue with RNA-guided genome editing is specificity. It has become clear that a certain level of mismatches (up to five or six in some cases) between guide RNA and target DNA are tolerated, leading to potential off-target cleavage (Hsu et al., 2013). This is a problem especially for potential clinical uses of CRISPR editing, for example to repair genetic lesions in embryos. One way to tackle off-target effects is to use Cas9 nickase (Cas9n) proteins with

53 inactivating mutations in one of the endonuclease domains (for example a D10A mutation in the RuvC domain). A pair of nickases is guided to opposite strands of the locus of interest, slightly offset from each other. Each nickase will cleave a single DNA strand only, but because the nicks are proximal the cell will treat this as a single DSB with 5’overhangs (Figure 3.3). Specificity is increased 50 – 1,500-fold in this manner as two sgRNAs are used, both of which must find their respective target for editing to occur (Ran et al., 2013a). Most importantly, off- target issues are all but eliminated, as off-target nicks are repaired by base-excision repair (BER), which is a high-fidelity repair mechanism (Wallace, 2014). For these reasons we employed a double nickase approach to target ANP32A and ANP32B in the current study.

Figure 3.3. Double-nickase strategy – using a pair of RuvC D10A Cas9 nickases to generate a DSB (adapted from Ran et al. Cell 2013) The nickases are guided to opposite sides of the target DNA and will simultaneously cleave the strand complementary to the sgRNA. This will lead to a DSB with 5’ overhangs.

By 2014 a variety of cell types and organisms had been modified by CRISPR/Cas9, including HEK293 cells and its derivatives, induced pluripotent stem cells (iPSCs), bacteria, yeast, mammalian, fish, insect, and plant embryos (Sander and Joung, 2014). In 2017 a genetic lesion in MYBPC3, which leads to myocardial disease, was corrected by CRISPR editing in human preimplantation embryos (Ma et al., 2017).

We began the studies described here by targeting ANP32A and ANP32B in the HEK293 cell line, a workhorse for polymerase studies in our laboratory. In parallel we used the same approach to target A549 cells as these are derived from human lung tissue and therefore potentially more relevant to influenza virus. Finally, we also targeted the ANP32 genes in haploid cells since we reasoned that editing of a single copy of each gene might be more readily achieved.

3.2 Cell lines Four cell lines were used for generating ANP32A and ANP32B knockouts, namely HEK293T (ATCC CRL-3216), A549 (ATCC CCL-185), HAP1 (Horizon Discovery) and eHAP cells (Horizon Discovery). ANP32A and B proteins are widely expressed in fibroblast-like laboratory cell lines (Consortium, 2013), which was confirmed by western blotting analysis with commercially available antibodies ab51013 (Abcam; α-ANP32A) and 10843-1-AP (Proteintech; α-ANP32B) (Figure 3.4).

Figure 3.4 ANP32 proteins are expressed in HEK293T, A549, HAP1 and eHAP cells Western blotting analysis showing ANP32A (top panels) and ANP32B (bottom panels) expression in HEK293T (A), A549 (B), HAP1 (C) and eHAP cells (D), alongside vinculin loading control. 30 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol. Lysates were resolved by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/1,000; Abcam), rabbit α-ANP32A (ab51013, 1/500, Abcam) and rabbit α-ANP32B (10843-1-AP, 1/1,000, Proteintech), and secondary antibody HRP-conjugated sheep α-rabbit IgG (AP510P, 1/10,000; Merck), and developed on a Fusion- FX imaging system.

The 239T cells were selected for ease of use and availability, popularity with other groups performing CRISPR studies, fast growth and high transfectibility, which is important because one of the key downstream assays is a transfection-based minireplicon reporter assay that tests for FluPol activity. On the downside, 293T cells do not support efficient multicycle replication of most naturally occurring influenza virus strains, so live virus work would be restricted to the mouse-adapted influenza virus strain (H1N1) A/WSN/1933 (Choppin, 1969; Goto and Kawaoka, 1998). Moreover, polyploidy of 293T cells means that multiple copies of the ANP32 genes are present in each cell. Originally derived from the kidney of an aborted human embryo in 1973, HEK293 cells were transformed with sheared Adenovirus 5 DNA

(Graham et al., 1977). The cells were later re-transformed with the SV40 T antigen to become the 293T line (Rio et al., 1985). Although the cells were known to be hypotriploid, harbouring 62-70 chromosomes instead of the normal 46 in diploid cells, the 293T genome was thoroughly analysed by deep sequencing in 2014 and found to be variable and dynamic, characterised by duplications and translocations (Lin et al., 2014). Indeed, studying the chromosomal locations of Anp32A (15q23) and Anp32B (9q22) more closely, we found that 4-5 alleles of Anp32A and 3-4 alleles of Anp32B are expected in any given HEK293T cell (Figure 3.5).

Figure 3.5 On average, HEK293T cells contain 4-5 copies of Anp32A and 3-4 copies of Anp32B (adapted from Lin et al Nat Comm 2014) Anp32A, on the long arm of Chromosome 15 (arrow), is present in the HEK293 genome in four or five copies. This is accounted for by the hypotriploid state of the cells themselves, combined with duplications and translocations where fragments of Chromosome 15 were located elsewhere in the genome. Most cells have either four or five copies (thick horizontal red lines) according to the ploidy numbers on the right y-axis. The left y-axis shows the depth of the sequencing; the x-axis the location on the Chromosome. The Anp32B gene sits in a region on the long arm of Chromosome 9 and has a ploidy of three to four.

The A549 cell line was first initiated from the alveolar cell carcinoma of a 58-year-old male in 1972 (Giard et al., 1973), and further characterised as fibroblast-like adherent hypotriploid (64- 68 Chromosomes) cell lines with a doubling time of 48 hours (Lieber et al., 1976). These cells are believed to have some properties typical of type II alveolar epithelial cells, and may therefore be more relevant to influenza virus work, as this virus typically infects respiratory epithelium in vivo. A549 cells do support multicycle infection by some strains of influenza virus,

56 and have been successfully harnessed for CRISPR knockout screens in an influenza study (Li et al., 2020). On the downside, A549 cells grow more slowly than 293T cells, are multiploid, and transfect poorly.

Some rare human tumours and leukaemias naturally have near-haploid karyotypes. A stable near-haploid cell line with only a disomy of Chromosome 8 was isolated from the heterogeneous human (39M) leukaemia cell line KBM-7 (Andersson et al., 1987; Kotecki et al., 1999). HAP1 cells were obtained serendipitously from KBM-7 cells in a failed attempt to induce pluripotency by expression of Yamanaka factors (Carette et al., 2011). Not only had HAP1 cells lost diploidy for Chromosome 8, they also grew adherently and no longer expressed haematopoietic markers. HAP1 cells, however, are not fully haploid, as a 30- megabase fragment of the long arm of Chromosome 15, encompassing 330 genes including Anp32A, is retained integrated on the long arm of Chromosome 19 (Essletzbichler et al., 2014). This fragment was excised by CRISPR/Cas9, yielding a fully haploid human cell line named eHAP (Figure 3.6).

It should be noted that upon passaging haploid cells tend to revert to diploidy by endoreplication, which is promoted by genome instability (Fox and Duronio, 2013). Furthermore, maintaining HAP1 and eHAP cells at higher-than-ideal confluency (>70-80%) will lead to reversion to diploidy in a subset of cells (Jan Carette, pers comm). Nevertheless, for CRISPR/Cas9 editing purposes a diploid cell line is still more straightforward than polyploid cells like 293T and A549.

Figure 3.6 eHAP cells are derived from HAP1 cells, which in turn derive from KBM-7 cells (adapted from Kotecki et al Exp Cell Res 1999 and Essletzbichler et al Genome Research 2014). Myeloid leukaemia KBM-7 cells are haploid for all chromosomes except Chromosome 8. This disomy was lost when HAP1 cells were derived from KBM-7 cells, but a 30-megabase fragment of Chromosome 15 carrying 330 genes including Anp32A was found fused to Chromosome 19. This fragment is eliminated in the fully haploid eHAP cells.

3.3 Guide RNA design The Anp32A gene spans 42,388 nucleotides on the long arm of Chromosome 15 (region 2 band 3), between nucleotides 68,778,535 and 68,820,922, and is transcribed in the reverse direction. The coding sequence covers 747 nucleotides (excluding the termination codon) spread over seven exons. As exon 1 is very small, two pairs of sgRNAs were designed against exon 2, with an additional pair against exon 4 as a backup. The Anp32B gene spans 32,737 nucleotides on the long arm of Chromosome 9 (region 2 band 2), between nucleotides 97,983,207 and 98,015,943, but is read in the forward direction. The coding sequence of 753 nucleotides is divided into 7 exons, and pairs of sgRNAs were designed against exons 2 and 4. Twenty-nucleotide single-strand DNA molecules (primers, basically) were designed against opposite strands of the target DNA with the aid of the CRISPR design tool at www.crispr.mit.edu (Feng Zhang lab). None of the pairs had any genic off-target cleavage site according to the algorithm on the website. As negative controls, we selected known non- targeting gRNAs with no target in the human genome (Doench et al., 2016). Given that the expression plasmid we used, pSpCas9n(BB)-2A-Puro (PX462) v2.0 (Feng Zhang lab), expresses the sgRNA under a human U6 promoter, a guanine was added to the 5’ end of the guide if it did not already start with G. PX462 v2.0 already contains the guide RNA scaffold that interacts with Cas9, preceded by two back-to-back BbsI restriction sites to allow insertion of the part of the sgRNA that is complementary with the target DNA (Ran et al., 2013b). The 9,175 base-pair plasmid further encodes a D10A mutant Cas9 nickase, as well as ampicillin and puromycin resistance genes for selection in prokaryotes and eukaryotes, respectively. Digestion of the plasmid with BbsI leaves short overhangs on both sides for integration of the unique sgRNA-encoding duplex (Figure 3.7). Our overall CRISPR/Cas9 strategy to obtain ANP32 knockout cell lines is illustrated in Figure 3.8.

Figure 3.7 Guide RNA cloning strategy (A) CRISPR plasmid PX462 v2.0 was digested with the restriction enzyme BbsI, which cleaves at the red arrowheads and leaves 4-nt overhangs into the hU6

58 promoter (green) and the gRNA scaffold (blue), linearizing the plasmid as shown in (B). The unique gRNA-encoding duplex is then ligated into PX462 (C). The part of the gRNA that is complementary to the target DNA (golden 20N) is preceded by a guanine (green) required by the U6 promoter. PX462 sequence obtained from Addgene.

Figure 3.8 Double nickase strategy to obtain ANP32A and ANP32B knockout cells Exons 2 (e2) of both Anp32A (left) and Anp32B (right) were targeted by a pair of Cas9 nickases (Cas9n) in complex with unique guide RNAs complementary to opposite strands of the target locus. Successful simultaneous on-target cleavage resulted in double-stranded breaks with a 37-base pair (Anp32A) or 56-base pair (Anp32B) 5’ overhang, respectively, well within the ≤ 100 bp range for successful editing (Ran et al., 2013a)

An initial round of CRISPR was set up in HEK293T cells, using pairs of gRNAs against exons 2 of both genes. Closely following a protocol from the Feng Zhang lab at MIT, we transfected pairs of sequence-verified PX462 plasmids bearing the specific guide RNAs into heterogeneous cell populations and then enriched transfected cells by Puromycin selection (Ran et al., 2013b). Clonal populations were obtained by single cell sorting into 96-well plates using a fluorescence-activated cell sorter (FACS). The clones were validated by a genomic cleavage assay (GeneArt) and by western blotting analysis.

The genomic cleavage assay entails initial PCR amplification of the target locus, followed by melting and random re-annealing of the PCR amplicons. Any mismatched amplicons resulting from indel mutations at the Cas9 cleavage site are then cleaved by T7 endonuclease I (T7E1),

59 an enzyme that recognizes faulty base pairing. The assay thus suggests CRISPR editing at one or several alleles of the clones, but it cannot assure the user that all alleles have been affected (Figure 3.9).

Figure 3.9 Genomic cleavage validation of CRISPR clones 2% agarose gels showing Anp32A (400 bp) and Anp32B (412 bp) PCR amplicons before (top) and after melting / annealing reaction + T7E1 cleavage (bottom). Before cleavage indel mutations between clones may be observed if they are large enough to show on the gel (for example compare Anp32B lanes 4 and 5), or within clones if distinct alleles have been affected differently (Anp32B lanes 1 and 2). After cleavage one will be able to tell whether indel mutations are present in alleles of similar size, as in all the clones with the exception of Anp32B lane 6, and the negative control.

Western blotting analysis was then carried out to confirm these results. If all alleles have indel mutations leading to frameshifts no full-length protein should be expressed in the cells. Messenger RNAs harbouring premature termination codons (PTCs) are targeted by the nonsense-mediated decay (NMD) pathway as translation stalls on the ribosome. Both the aberrant mRNA and the truncated polypeptide are subsequently targeted for degradation (Hug et al., 2016; Kervestin and Jacobson, 2012). It is, however, difficult to prove the complete absence of protein expression by western blotting alone. Among other things this depends on the quality of the antibody. Still, western blotting analysis suggested successful knockout of either ANP32A or ANP32B, and these clones were taken forward for Sanger sequencing.

Unfortunately, all of the clonal lines for which indels had clearly been present in at least some alleles, according to the genomic cleavage assay, and for which protein expression had been non-detectable by western blotting, still had one or more alleles that were either wildtype or repaired in frame. As an example, one putative Anp32A KO clone had five alleles: one

60 wildtype; one with a 3-nt in-frame deletion of valine 23, two distinct single nucleotide deletions leading to PTCs at amino acid residues 30 and 32, respectively, and finally a single-nucleotide insertion introducing a PTC at residue 48.

At this point we decided to use A549, HAP1 and eHAP cells instead. Having already established that both ANP32A and B proteins are expressed in these cell lines previously by western blotting (Figure 3.4), we first optimised transfection protocols for these cell lines, comparing different transfection reagents at a variety of reagent-to-plasmid DNA ratios. We found that single eGFP plasmid transfection worked best in A549 with XtremeGene 9 at a ratio of 2 µl reagent to 1 µg plasmid DNA. HAP1 cells transfected best with Lipofectamine 3,000 at a 3:1 ratio (Figure 3.10).

Figure 3.10 Single plasmid transfection optimization in A549 and HAP1 cells Three common commercially available transfection reagents – Lipofectamine 3,000, Turbofectin 8.0 and XtremeGene 9 – were assayed in various µl reagent to µg pCAGGS-GFP plasmid DNA ratios (2:1, 3:1, 6:1 and 3:2; not all shown here). A549 transfection was most efficient with XtremeGene 9 at a 2:1 ratio (bottom left panel), while HAP1 cells were transfected best with Lipofectamine 3,000 at a 3:1 ratio (top right panel).

Images were taken 48 hours post-transfection using the GFP channel (ex 470/40 nm, em 525/50 nm) on an Axiovert 40 CFL microscope.

Next, we tested whether the cell lines were suitable for minigenome reporter assays, a key assay in the lab, which is based on simultaneous transfection of at least five plasmids (Figure 3.11). Comparing A549 cells with HAP1 cells and eHAP cells, with 293T cells as a positive control, we found that A549 cells gave poor but detectable signals with both XtremeGene 9 and Lipofectamine 3,000, while HAP1 and eHAP cells gave relatively robust signals with Lipofectamine 3,000 (Figure 3.12).

Figure 3.11 The minigenome reporter assay Minigenome reporter assays are based on exogenous expression of the components of the viral ribonucleoprotein (vRNP), which is the minimal influenza virus infectious unit. The vRNP is reconstituted through expression of PB1, PB2 and PA (together forming the heterotrimeric RNA-dependent RNA polymerase), nucleoprotein NP, and a reporter plasmid with an RNA Polymerase I promoter. The reporter is transcribed in situ by cellular RNA Polymerase I into a virus-like negative-sense RNA molecule with a reporter gene (luciferase or a fluorescent protein) between conserved viral promoter regions that bind the RdRp. A luciferase or fluorescent signal ensues when the reporter gene is replicated and transcribed by the reconstituted viral polymerase. When the readout is bioluminescence rather than fluorescence, a cellular transcription control plasmid pCAGGS- Renilla luciferase is co-transfected, and the firefly to Renilla ratio is the FluPol activity readout.

Figure 3.12 Minigenome reporter assays with blue fluorescent protein (BFP) reporter in A549, HAP1 and eHAP cells Minigenome reporter assays in 293T, A549, HAP1 and eHAP cells with pCAGGS plasmids expressing RNP components PB1, PB2, PA and NP of the mammalian-adapted (PB2-E627K) H5N1 strain A/Turkey/England/50-92/91 (hereafter ‘5092K’), and a pPolI-BFP-NLS minigenome reporter, in a ratio 2:2:1:4:2, with 100 ng PB1 per well in a 12-well plate (~400,000 cells). Fluorescence was observed 24 hours post-transfection using the DAPI channel (ex 365/12 nm, em 447/60 nm) on an Axiovert 40 CFL microscope.

Next, we infected A549 and HAP1 cells with the laboratory-adapted H1N1 strain A/Puerto Rico/8/1934 (PR8) at a multiplicity of infection (MOI) of 0.001, in order to measure multicycle growth dynamics (Figure 3.13). We found that influenza virus replicates in both cell lines, although infectious virus accumulation plateaued earlier in HAP1 cells than in A549 cells. Still, the obtained titres were more than sufficient to proceed with these cell lines.

Figure 3.13 Influenza virus replicates in A549 and HAP1 cells A549 (circles) and HAP1 cells were infected with H1N1 PR8 virus at MOI 0.001 and incubated at 37⁰C in the presence of 1 µg/ml trypsin to allow multicycle replication. Supernatants were harvested at the indicated time points post-infection and PFU/ml established by plaque assay on MDCK cells. Representative data is shown from one of two independent triplicate experiments.

Two lineages of influenza B virus (IBV) currently circulate in humans. We carried out minigenome assays with plasmids that reconstituted the polymerase from the Yamagata lineage strain B/Florida/4/2006 (Florida 06), in the form of vRNP components with either a bioluminescent firefly luciferase or a fluorescent DsRed reporter gene, flanked by IBV promoter sequences (Figure 3.14). We found that IBV polymerase replicates potently in HAP1 and eHAP cells, and in 293T cells.

Figure 3.14 Influenza B virus polymerase is active in HAP1 and eHAP cells Minigenome assays in 293T (A), HAP1 (B) and eHAP cells (C). Cells were transfected with IBV Florida 06 RNP components

PB1, PB2, PA and NP, alongside an IBV minigenome reporter carrying a firefly luciferase (A) or DsRed- NLS (B and C) reporter gene flanked by IBV promoters, at a ratio 2:2:1:4:2 (40 ng PB1 in ~200,000 cells). In (A) a pCAGGS-Renilla luciferase expression plasmid (i.e. under RNA Pol II reporter) was cotransfected as a cellular transcription and transfection control, at the same amount as the firefly reporter. Data shown in (A) are firefly activity normalised to Renilla 24 hours post-transfection, plotted as mean (SD) from a representative triplicate (technical) repeat (n = 2 biological repeat experiments). Fluorescence in (B) and (C) was observed 24 hours post-transfection using the TexasRed channel (ex 562/40 nm, em 624/40 nm) of an Axiovert 40 CFL microscope.

Having confirmed that A549, HAP1 and eHAP cells would be suitable model cell lines in which to perform transfection-based assays as well as virus infection, we targeted the ANP32A and B genes for CRISPR editing in these cell lines.

3.4 Generating KO cells

We carried out CRISPR editing as before, employing nickase pair transfection, puromycin selection, single cell sorting, and growth in 96-well plates. However, we employed a more efficient validation strategy, namely screening all the clones simultaneously by next- generation sequencing (NGS) of barcoded amplicons (Figure 3.15). This enabled high throughput simultaneous screening and was less prone to false positives than combining genomic cleavage assays, western blotting and Sanger sequencing.

Figure 3.15 Next-generation sequencing strategy for screening CRISPR clones DNA was extracted from putative ANP32 knockout clones and the CRISPR locus was PCR amplified with barcoded primers. Sequencing adaptors and unique NGS indexes were added to the amplicons by PCR, the samples were pooled and NGS with paired-end reads was carried out on an Illumina MiSeq

65 machine. Reads were then aligned and indel mutations detected in an R script written in-house for this purpose by Dr Daniel H. Goldhill. The sequences were then transferred to Geneious for further analysis.

We designed 20 unique forward primers for PCR amplification of the Anp32A and Anp32B CRISPR-targeted loci (10 per gene), which were tagged with 6-base barcodes based on the Hamming code quaternary format and optimised for redundancy and GC content (Bystrykh, 2012). Eventually a total of 268 amplicons (each derived from a clonal CRISPR cell line) were divided into 14 pools of up to 24 amplicons with unique barcodes. Using the NEBNext Ultra II kit (NEB) sequencing adaptors, one of 24 unique sequencing indexes were added to the amplicons (Figure 3.16). Each sample now had a unique barcode / NGS index combination, and each of the 24 sub-pools were pooled into a single final pool.

Barcode Barcode A1 A2 A3 A4 A5 A10 A7 B1 B2 B3 B4 B5 B8 B9

Pool sample ul sample ul sample ul sample ul sample ul sample ul sample ul Pool sample ul sample ul sample ul sample ul sample ul sample ul sample ul 1 AA1 1.4 HA1 2.9 HA25 1.4 eHA1 1.8 eHA25 1.6 AA25 8.6 eHA49 1.8 1 AB1 1.7 HB1 1.4 eHB1 3.1 eHB24 1.7 AB49 2.7 AnT1 2.4 2 AA2 1.2 HA2 1.7 HA26 1.4 eHA2 1.0 eHA26 1.6 AA26 2.2 eHA50 1.8 2 AB2 1.3 AB26 2.6 HB2 1.6 eHB2 3.3 eHB25 3.2 AB50 1.6 3 AA3 1.1 HA3 1.4 HA27 1.1 eHA3 1.4 eHA27 1.2 AA27 1.6 AnT1 1.7 3 AB3 1.5 AB27 1.4 HB3 1.4 eHB3 2.4 eHB26 1.8 HB25 2.0 AnT3 10.1 4 AA4 1.1 HA4 2.5 HA28 1.3 eHA4 1.3 eHA28 1.5 AA28 1.7 AnT2 2.4 4 AB4 1.3 AB28 1.3 HB4 1.5 eHB4 3.3 eHB27 7.1 HB26 2.3 HnT1 1.6 5 AA5 1.6 HA5 1.3 HA29 1.3 eHA5 1.2 eHA29 1.4 AA29 2.5 AnT3 2.6 5 AB5 1.4 AB29 8.9 HB5 1.5 eHB5 1.6 eHB28 2.2 HB27 2.4 HnT2 1.7 6 AA6 0.9 HA6 1.9 HA30 1.1 eHA6 1.5 eHA30 1.5 AA30 2.6 HnT1 1.3 6 AB6 1.3 AB30 1.5 HB6 1.4 eHB5(2) 2.1 HB28 3.3 HnT3 1.6 7 AA7 1.0 HA7 2.4 HA31 1.3 eHA7 1.2 eHA31 1.2 AA31 2.3 HnT2 1.3 7 AB7 1.6 AB31 1.6 HB7 1.6 eHB6 2.4 HB29 2.1 eHnT1 1.8 8 AA8 1.4 HA8 2.3 HA32 1.5 eHA8 1.0 AA32 6.2 HnT3 1.6 8 AB8 1.8 AB32 1.9 HB8 1.3 eHB7 2.7 HB30 2.4 eHnT2 2.3 9 AA9 1.1 HA9 2.8 HA33 1.1 eHA9 1.5 eHA33 1.2 AA33 1.8 eHnT1 1.5 9 AB9 1.5 HB9 5.0 eHB8 1.7 eHB32 3.3 eHnT3 2.7 10 AA10 2.0 HA10 2.7 HA34 1.7 eHA10 1.4 AA34 2.1 eHnT2 1.9 10 AB10 1.7 AB34 1.4 HB10 3.9 eHB9 1.9 eHB33 5.2 11 AA11 1.4 HA11 2.7 HA35 1.6 eHA11 1.4 eHA35 1.4 AA35 3.8 eHnT3 2.2 11 AB11 1.3 AB35 2.1 HB11 2.4 eHB10 2.6 eHB50 5.2 12 AA12 1.4 HA12 2.2 HA36 1.6 eHA12 1.6 eHA36 2.3 AA36 3.0 12 AB12 1.4 AB36 1.7 HB12 1.6 eHB11 3.5 13 AA13 1.2 HA13 2.8 HA37 1.5 eHA13 1.5 eHA37 1.6 AA37 2.3 13 AB13 1.6 AB37 1.5 HB13 1.8 eHB12 6.6 eHB36 6.6 14 AA14 1.4 HA14 2.4 HA38 1.7 eHA14 1.6 eHA38 1.4 AA38 2.3 14 AB14 2.8 AB38 1.7 HB14 4.2 eHB13 2.4 15 AA15 1.4 HA15 3.8 HA39 1.5 eHA15 1.3 eHA39 1.5 AA39 3.2 15 AB15 3.7 AB39 2.0 HB15 5.0 eHB14 5.8 16 AA16 1.3 HA16 3.3 HA40 1.6 eHA16 1.4 eHA40 1.7 AA40 4.9 16 AB16 1.7 AB40 1.9 HB16 2.0 eHB15 3.1 eHB39 4.5 17 AA17 1.5 HA17 3.9 HA41 1.5 eHA17 1.9 eHA41 1.6 AA41 2.1 17 AB17 1.8 AB41 5.1 HB17 1.1 eHB16 2.3 eHB40 5.7 18 AA18 1.1 HA18 2.7 HA42 1.6 eHA18 1.6 eHA42 1.6 AA42 3.1 18 AB18 1.5 AB42 2.2 HB18 1.7 eHB17 2.1 eHB41 4.9 19 AA19 1.5 HA19 2.2 HA43 1.7 eHA43 2.8 AA43 4.6 19 AB43 1.8 HB19 2.0 eHB18 7.2 20 AA20 1.3 HA20 2.8 HA44 1.5 eHA20 1.7 eHA44 1.8 HA49 2.4 20 AB44 1.9 HB20 8.7 eHB19 2.4 21 AA21 1.3 HA21 2.6 HA45 1.5 eHA21 1.4 eHA45 2.6 HA50 2.4 21 AB21 2.2 AB45 2.0 HB21 1.6 eHB20 2.5 22 AA22 1.5 HA22 2.5 HA46 2.5 eHA22 1.9 eHA46 4.0 22 AB22 2.0 AB46 6.8 HB22 2.6 eHB21 2.5 23 AA23 2.0 HA23 3.4 HA47 2.2 eHA23 4.6 eHA47 2.2 23 AB23 2.1 AB47 2.7 HB23 2.2 24 AA24 1.7 HA24 2.4 HA48 1.4 eHA24 1.6 eHA48 1.7 24 AB24 1.7 AB48 2.5 HB24 2.1 eHB23 3.2

Figure 3.16 Each CRISPR clone has a unique combination of barcode and NGS index Next generation sequencing with barcoded primers approach labelled PCR amplicons of each CRISPR- modified locus with a unique sequence of nucleotides consisting of 14 barcodes in combination with 24 NGS indexes. All the samples can thus be sequenced in a single pool. AA = A549 ANP32A KO; HA1 = HAP1 ANP32A KO; eHA = eHAP ANP32A KO; AB = A549 ANP32B KO; HB = HAP1 ANP32B KO; eHB = eHAP ANP32B KO; nT = non-targeting guide RNA negative control.

The amplicons were sequenced using 150-bp paired end reads on an Illumina MiSeq instrument, and reads were mapped using Burrows-Wheeler Aligner (BWA) v0.7.5. An R script (https://github.com/Flu1/CRISPR) was developed in-house by Dr Daniel Goldhill to detect indel mutations above a cut-off of 2.5% of reads. Selecting indel mutations inducing a shift in the reading frame we obtained ANP32A knockout in 17/43 A549 clones (39.5%), 19/50 HAP1

66 clones (38.0%) and 24/50 eHAP clones (48.0%). We obtained 5/50 ANP32B knockouts in A549 cells (10.0%), 8/30 HAP1 clones (26.7%), and 13/52 eHAP clones (25.0%) (Figure 3.17). So, overall, targeting the Anp32A gene proved more efficient than Anp32B (60/140 (42.9%) versus 26/111 (23.4%), respectively), and knockout was most successful in eHAP cells. Moreover, eHAP cells seemed to have a more stable genome, with about half the clones retaining haploidy. In contrast, we observed HAP1 clones with up to four alleles, although this might also be explained by non-clonality.

Figure 3.17 Genotype of CRISPR knockout cells The NGS data were analysed using a custom R script that detected insertion/deletion mutations in the sequencing reads. From these we tabulated clones for which each allele had a mutation leading to a shift in the reading frame.

The output sequences were transferred to Geneious v6 for further analysis and the results were validated by western blotting (Figure 3.18).

Figure 3.18 Validation of NGS data by western blotting analysis Expression of ANP32A (A) or ANP32B (B) was analysed by western blotting analysis in putative ANP32A and ANP32B knockout cell lines, alongside vinculin loading control. Green arrows indicate clones that were taken forward for

67 further phenotypic analysis. These blots show panels of eHAP lines corresponding to the numbers in Figure 3.17. 30 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β- mercaptoethanol. Lysates were resolved by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/1,000; Abcam), rabbit α-ANP32A (ab51013, 1/500, Abcam) and rabbit α-ANP32B (10843- 1-AP, 1/1,000, Proteintech), and secondary antibody HRP-conjugated sheep α-rabbit IgG (AP510P, 1/10,000; Merck), and developed on a Fusion-FX imaging system.

Next, minigenome assays were carried out in selected eHAP and A549 ANP32 knockout lines

(Figure 3.19) with reconstituted polymerases of the seasonal H3N2 strain A/Victoria/3/75 (Vic/75) and the 2009 pandemic H1N1 isolate A/England/195/2009 (Eng/195). No significant differences in FluPol activity was observed. However, an interesting pattern was observed for IBV Florida 06 polymerase and for H5N1 50-92 polymerase with the mammalian PB2 adaptation E627K. In both eHAP and A549 cells, these polymerases were more active in the absence of ANP32A (AKO), while activity was generally reduced in the absence of ANP32B (BKO). Certain polymerase constellations thus seem to prefer ANP32B to support their replication. Such preferences are investigated detail in recent reports (Peacock et al., 2020a; Peacock et al., 2020b).

Figure 3.19 Influenza virus polymerase is active in ANP32A and B single knockout cells Minigenome assays in eHAP (A-D) and A549 (E-F) cells lacking either ANP32A (AKO) or ANP32B (BKO) expression. Control cells are clonal cells targeted with non-targeting guide RNAs. Cells were transfected with pCAGGS expression plasmids encoding PB1, PB2, PA and NP, reconstituting RNPs

68 from IAV Vic/75 (A), 5092K (B and E), Eng/195 (C) or IBV Florida 06 (D and F), along with IAV or IBV firefly luciferase minigenome reporter and Renilla luciferase control, at a ratio 2:2:1:4:2:2 (40 ng PB1 in ~200,000 cells) Data shown are firefly activity normalized to Renilla, plotted as mean (SD) obtained by one-way analysis of variance (ANOVA) from one representative triplicate (technical) repeat (n=2 biological repeat experiments) ns, not significant; *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001.

Next, we infected eHAP AKO and BKO cells with H1N1 PR8 virus at MOI 0.001 (Figure 3.20). There was no significant difference in virus replication 24 hours post-infection, as established by one-way ANOVA, although replication in BKO cells trended lower than in AKO cells.

Figure 3.20 H1N1 PR8 virus replicates in control, AKO and BKO cells eHAP cells were infected with H1N1 PR8 virus at MOI 0.001 and incubated at 37⁰C degrees in the presence of 1 µg/ml trypsin. PFU/ml of supernatants harvested 24 hours post-infection were established by plaque assay on MDCK cells. Data shown are mean PFU/ml (SD) from a single triplicate experiment; statistical analysis by one- way ANOVA. LOD (dotted line) denotes the limit of detection based on the dilution factor in plaque assays. ns, not significant

In case the phenotypes we were seeing were due to there being remaining intact alleles in these clones, in spite of the NGS data and western blotting results, we decided to validate the NGS data by Sanger sequencing. Selecting three eHAP AKO and three eHAP BKO clones,

69 we amplified the relevant loci from fresh genomic DNA using PCR (Figure 3.21), cloned the amplicons into TOPO blunt vectors and sent twelve colonies per clone for sequencing. We did not find wildtype alleles in any of the clones, although Sanger sequencing did identify some indel mutations that had been missed by the R script (Table 3.1)

Figure 3.21 Sanger sequencing of selected eHAP knockout cells (A) Anp32A and Anp32B loci were amplified by touchdown PCR from fresh genomic DNA and run on a 1.5% agarose gel. Where possible, separate alleles were gel extracted and cloned into TOPO vectors. Cells that had been cloned after CRISPR using non-targeting primers (nT) show the size of the wildtype locus, which is identical in clones in which the other gene was targeted (Anp32A primers in clone eHB2 and Anp32B primers in clone eHA8) (B) A typical example of Cas9-targeted alleles as shown in Geneious v6, with two distinct deletions clearly visible and premature termination codons downstream of the deletions in black. This is the diploid clone eHA18 (eHAP ANP32A KO # 18), which has one allele with a 20-bp deletion and one allele with a 25-bp deletion, as confirmed by both NGS and Sanger sequencing (see also Table 3.1)

NGS Sanger clone insertion deletion insertion deletion eHA8 1 bp 238 bp 1 bp eHA18 20 bp 20 bp 25 bp 25 bp eHA21 5 bp 90 bp 23 bp 5 bp 23 bp eHB2 34 bp 34 bp 85 bp 85 bp eHB6 2 bp 2 bp 11 bp eHB20 17 bp 17 bp 26 bp 26 bp

Table 3.1 Indel mutations found in ANP32A and ANP32B eHAP knockout cells by NGS and Sanger sequencing Insertions and deletions detected in three separate clonal eHAP ANP32A and ANP32B knockout cell lines by next-generation sequencing and Sanger sequencing analysis. The large 238-base pair (bp) insertion in the eHA8 clone (eHAP ANP32A KO # 8) was missed by the NGS analytical R script, as was a large 90-bp insertion in eHA21. Both were present in the same allele as the 1-bp deletion and the 5-bp insertion, respectively. An 11-bp deletion in eHB6 flagged up by the algorithm in the R script was not found by Sanger sequencing.

We also used fresh lysates to validate western blots and to ensure the non-targeted paralogue was expressed. Although the guide RNA pairs were not predicted to have any off-target effects, the Anp32A and Anp32B genes are similar enough to warrant caution. The blots showed that ANP32A was indeed expressed in eHAP BKO cells, and that ANP32B was expressed in eHAP AKO cells (Figure 3.22).

Figure 3.22 ANP32A is expressed in BKO cells and vice versa Western blots showing ANP32A expression (A) in BKO and control cells but not in AKO cells, and (B) ANP32B expression in AKO and control cells, but not in BKO cells. Vinculin was included as a loading control. 30 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol. Lysates were resolved by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/1,000; Abcam), rabbit α-ANP32A (ab51013, 1/500, Abcam) and rabbit α-ANP32B (10843-1-AP, 1/1,000, Proteintech), and secondary antibody HRP-conjugated sheep α-rabbit IgG (AP510P, 1/10,000; Merck), and developed on a Fusion- FX imaging system.

3.5 Generating cells that lack expression of both ANP32A and ANP32B

The loss of expression of either ANP32A or ANP32B showed little phenotypic efect on influenza virus polymerase activity or replication, suggesting that ANP32A and ANP32B might be functionally redundant in their support for FluPol. We therefore set out to generate eHAP double knockout cells lacking both ANP32A and ANP32B. We set up CRISPR reactions targetting the Anp32A gene in BKO clone eHB20, and the Anp32B gene in AKO clone eHA8, under the expectation that targeting Anp32A in the BKO clone might be more efficient, as it had been when generating the single knockout cells.

Ablation of ANP32A protein expression in three distinct clonal lines derived from BKO clone eHB20 was confirmed by western blotting (Figure 3.23) and Sanger sequencing. Double knockout (dKO) clone 26 had a 5-bp and a 29-bp deletion in each Anp32A allele, dKO clone 44 had a 5-bp insertion and a 20-bp deletion, and dKO clone 45 had a 13-bp and a 25-bp deletion. All had retained the 17-bp and 26-bp deletions in the Anp32B alleles that were already present in the parent clone, eHB20.

Figure 3.23 ANP32A expression is lost in double knockout cells Western blots showing absence of ANP32A expression in dKO clones derived from BKO clone eHB20; vinculin was used as a loading control. Green arrows show dKO clones taken forward for further phenotypic analysis. 30 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol. Lysates were resolved by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/1,000; Abcam), rabbit α-ANP32A (ab51013, 1/500, Abcam) and rabbit α-ANP32B (10843-1-AP, 1/1,000, Proteintech), and secondary antibody HRP-conjugated sheep α-rabbit IgG (AP510P, 1/10,000; Merck). Membranes were incubated with SuperSignalTM West Femto substrate and developed on a Fusion-FX imaging system.

Next, minigenome assays were carried out in all three double knockout clones. H3N2 Vic/75, pH1N1 Eng/195 and H5N1 5092K polymerase activity was completely abrogated in each of the three dKO lines, but retained in the non-targetted control cells (Figure 3.24).

Figure 3.24 Influenza polymerase activity is abrogated in ANP32A and B double knockout cells Minigenome assays in three distinct eHAP clonal lines lacking expression of ANP32A and ANP32B. pCAGGS expression plasmids encoding RNP components PB1, PB2, PA and NP from H3N2 Vic/75 (A), pH1N1 Eng/195 (B) or H5N1 5092K (C) were transfected along with a pPolI-firefly luciferase minigenome reporter and Renilla control, in a 2:2:1:4:2:2 ratio (40 ng PB1 in ~200,000 cells). Data shown are firefly activity normalized to Renilla, plotted as mean (SD) analysed by one-way analysis of variance (ANOVA) from one representative triplicate (technical) repeat (n=3 biological repeat experiments) ****, P<0.0001.

3.6 Discussion

This chapter describes the generation of ANP32A and ANP32B knockout cells by CRISPR/Cas9 genome editing. Initially the genes were targetted in HEK293T cells, however the polyploidy of these cells hindered progress. Indeed, an early putative HEK293T ANP32A knockout clone was found to carry five alleles of the gene, as with hindsight might have been predicted from the available literature (Lin et al., 2014).Bearing in mind the availability of human haploid cell lines and their obvious suitability for genome editing, we turned to these cells instead.

Because the human genome is so large, and guide RNAs are only small and complementarity to the target DNA is flexible, we opted for a double nickase CRISPR strategy. Any 20- nucleotide guide RNA will potentially have dozens of off-target sequences in the genome, including in coding regions, but a pair of them normally will not. The two guide RNAs against Anp32A exon 2, for example, individually had 478 and 134 offtargets, 29 and 16 of which were inside genes, respectively. The pair of them together, in contrast, had zero offtargets in the entire genome. Especially when targeting a gene in a family of similar orthologues, a double nickase strategy is key.

In the second round of CRISPR we designed a screening strategy with next-generation sequencing followed by western blotting, rather than western blotting (and genomic cleavage assays) followed by sequencing. NGS can be scaled up to hundreds of clones relatively easily, and preparing amplicons with barcoded primers allows a single round of sequencing. The most promising clones were then be validated by western blotting and Sanger sequencing. Genomic cleavage assays are not particularly useful on the single clone level, because it remains unclear whether any wildtype alleles remain (small indel mutations will be similar in size to wildtype. Such assays are useful to check the heterogeneous cell population, before single cell sorting, to give the user an indication that CRISPR/Cas9 editing has worked. Moreover rather than keeping all potential editted clones in culture, it was found more efficient to amplify the cells in 96-well plates and transfer directly to 6-well plates to harvest cells for genomic DNA extraction (15%), western blotting lysate (70%) and transfer 15% of the cells to a new 6-well plate for further amplification before freezing down as a single aliquot. The clones never failed to resume growth upon thawing.

We established that eHAP cells are a useful cellular model for influenza virology. It may seem inappropriate to use haploid cells originally derived from lymphocyte blasts as a model for a respiratory pathogen, but many other cancerous or artificially transformed fibroblast-like cells such as 293T or A549 cells are also not perfect models. eHAP cells are easy to propagate, have low doubling times, are relatively easy to transfect, and are susceptible to most influenza virus strains. The more relevant cell models, like human airway epithelium (HAE) cells grown

74 at air-liquid interface and commercially available normal human bronchial epithelium cells (NHBC) are much more difficult to manipulate and more expensive.

Surprisingly given the importance of ANP32 proteins in a multitude of cellular processes, cells lacking either ANP32A or ANP32B, and even the double knockout cells lacking both paralogues, were viable and showed growth patterns identical to control cells. Mutant mouse strains carrying loss-of-function alleles have been described (reviewed in (Reilly et al., 2014)). In general, mice lacking Anp32A or Anp32E show no deleterious phenotypes, but ablation of Anp32B is lethal, suggesting a role for ANP32B in normal development (Reilly et al., 2011). A conditional ANP32B knockout mouse model was described recently (Beck et al., 2020), in which the absence of ANP32B can be studied.

There was no virus or polymerase phenotype in the single knockout cells. This is in line with the fact that ANP32 proteins had not been detected previoulsy as influenza cofactors using RNA interference or CRISPR screens (Han et al., 2018; Karlas et al., 2010; König et al., 2010; Li et al., 2020; Watanabe et al., 2014). Such approaches target a single gene per cell population. We now understand that this is because ANP32A and ANP32B form a pair of essential host factors that are functionally redundant in their support for influenza virus polymerase. The phenotype in the double knockout cells is striking; FluPol activity does not reach above background levels and there was no PR8 virus replication. In the next chapter we validate and use the dKO cells to further understand the role of these proteins in influenza virus replication.

IV ANP32 PROTEINS ARE FUNCTIONALLY REDUNDANT ESSENTIAL HOST FACTORS FOR INFLUENZA VIRUS REPLICATION IN HUMAN CELLS

This chapter will describe how the eHAP double knockout cells were used to better understand how ANP32 host factors support of influenza virus polymerase. From the results in the previous chapter we concluded that ANP32A and ANP32B are functionally redundant but essential host factors for influenza virus polymerase activity. In this chapter we build on and expand these findings, which would eventually culminate in a publication in the Journal of Virology (Staller et al., 2019). Figures from the paper are reproduced in this Chapter.

4.1 Influenza virus polymerase activity in ANP32 knockout cell lines For the work described in this chapter single cell lines representing ANP32A knockout (AKO), ANP32B knockout (BKO), double knockout (dKO), and a non-targeting negative control cell line (control) were taken forward. The AKO line is diploid with 20-bp and 25-bp deletions in the Anp32A alleles, the diploid BKO line bears 17-bp and 26-bp deletions in Anp32B, and is the parental line of the dKO cell line, which therefore has the same genotype at the Anp32B locus, with additional 5-bp and 29-bp deletions at the Anp32A locus (Table 4.1). Western blotting analysis was carried out to confirm the expression phenotypes (Figure 4.1).

Table 4.1 Genotypes of eHAP knockout cells A single clonal cell line of each kind (AKO, BKO and dKO) was taken forward for the work described in this chapter, along with a non-targeting negative control cell line with wildtype Anp32A and Anp32B genes.

Figure 4.1 ANP32 expression profile of control, AKO, BKO and dKO cells. From Staller et al J Virol 2019 Western blots showing expression of ANP32A in control and BKO cells, but not in AKO or dKO cells. ANP32B is expressed in control and AKO cells but not in BKO or dKO cells. Vinculin was used as a loading control. 50 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol. Lysates were resolved by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/1,000; Abcam), rabbit α-ANP32A (ab51013, 1/500, Abcam) and rabbit α-ANP32B (10843- 1-AP, 1/1,000, Proteintech), and secondary antibody HRP-conjugated sheep α-rabbit IgG (AP510P, 1/10,000; Merck), and developed on a Fusion-FX imaging system.

4.2 IAV and IBV FluPol activity is abrogated in dKO cells Minigenome assays were conducted with a panel of viruses to compare FluPol activity in the four cell lines (Figure 4.2). Reconstituted RNPs were used from a seasonal H3N2 virus (Vic/75), a mammalian-adapted avian H5N1 strain (5092K), a 2009 pandemic H1N1 virus (Eng/195), and a seasonal influenza B virus from the Yamagata lineage (Florida/06), all of which were introduced in the previous chapter. FluPol activity for polymerases reconstituted from Eng/195 or Florida/06 viruses was not reduced in AKO cells. In fact activity of polymerases from Vic/75 and 5092K was even enhanced in absence of ANP32A. In BKO cells there was a clear reduction in 5092K and Florida/06 FluPol activity, but not in Vic/75 or Eng/195 FluPol activity. The phenotype in dKO cells, however, was unequivocal: none of the influenza polymerases tested showed any activity. These phenotypes were not due to differential expression of vRNP components PB2 and NP, confirmed by western blot in lower panels. These data suggest functional redundancy in the support of FluPol in human cells by ANP32A and ANP32B. These striking results were described also in a separate study using minireplicon assays in sets of HEK239T knockout cells (Zhang et al., 2019a).

Figure 4.2 AKO and BKO cells support FluPol activity but dKO cells do not From Staller et al J Virol 2019 Minigenome assays in eHAP control, AKO, BKO and dKO cells. Cells were transfected with RNP components PB1, PB2, PA, NP and pPolI plasmids expressing IAV or IBV firefly luciferase minigenome reporter, as well as Renilla luciferase control, in a 2:2:1:4:2:2 ratio (40 ng PB1 in ~200,000 cells). Reconstituted vRNPs are from H3N2 Vic/75 (A), H5N1 5092K (B), pH1N1 Eng/195 (C), or IBV Florida/06 (D). Data shown are firefly activity normalised to Renilla, plotted as mean (SD) and analysed by one-way analysis of variance (ANOVA) from one representative (technical) triplicate repeat (n = 3 independent biological repeat experiments). ns, not significant; *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001. Accompanying western blots show expression of respective RNP components in each cell type (representative of one minigenome assay). 30 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol. Lysates were resolved by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/1,000; Abcam), mouse α-IAV NP (ab128193, 1/1,000; Abcam), mouse α-IBV NP (ab20711, 1/1,000; Abcam) and rabbit α-IAV PB2 (GTX125926, 1/2,000; GeneTex) and with HRP-conjugated secondary antibodies sheep α-rabbit IgG (AP510P, 1/10,000;

Merck) and goat α-mouse IgG (STAR117P, 1/5,000; AbD Serotec), and developed on a Fusion-FX imaging system.

In order to confirm functional redundancy, and to prove that the phenotype observed in the dKO cells was indeed due to the absence of ANP32A and ANP32B, and not some inadvertent CRISPR-induced change in cellular physiology, exogenous pCAGGS plasmid-expressed FLAG-tagged ANP32A, ANP32B, or both were used to complement absence of endogenous proteins. Titrations were performed to establish the amount of plasmid DNA required for robust rescue of the phenotype. The standard amounts of plasmid DNA transfected for minigenome assays in our laboratory, in a 24-well plate, are 40 ng PB1 and PB2, 20 ng PA, 80 ng NP, 20 ng reporter gene and 20 ng Renilla luciferase cellular transcription control per well. This 2:2:1:4:2:2 ratio is kept constant across well sizes and cell types. The rationale behind transfecting less PA is to minimize the effects of expression of the powerful endonuclease PA- X, accessible via ribosomal frameshifting on the PA gene segment, which targets cellular gene expression and is therefore toxic to the cells (Jagger et al., 2012). The amount of NP is doubled as more of it is needed to bind the viral RNA along its entire length. Broad titration of either 5 ng, 50 ng, or 500 ng ANP32 protein showed 50 ng ANP32 to be close to optimal (Figure 4.3). Previous experiments to optimise the amount of complementing ANP32 expression plasmid in HEK293T cells showed that small amounts were sufficient to support influenza polymerase, while larger amounts were deleterious for FluPol activity (Long et al., 2016). Further refinement suggested that the same amount of ANP32 plasmid as NP provided a consistently robust complementation of FluPol activity in dKO cells (i.e. 80-100 ng per well in a 24-well plate, which translates to approximately 2 x 105 cells). Complementation of 200 ng ANP32B in support of H3N2 Vic/75 polymerase led to significantly higher FluPol activity than the same amount of ANP32A (P<0.001) or 100 ng ANP32A and 100 ng ANP32B (P<0.001) (Figure 4.3 B). Complementation of 40 ng ANP32B rather than 20 ng significantly increased H3N2 Vic/75 polymerase activity (P = 0.003) but adding 80 ng ANP32B did not increase FluPol activity further (P = 0.8570) (Figure 4.3 C). These statistical analyses were carried out by one-way analysis of variance (ANOVA), excluding the ‘negative’ condition (eHAP cells expressing ANP32A and ANP32B) but including the double knockout condition (eHAP dKO cells complemented with Empty pCAGGS plasmid). Although some of the differences were highly significant when analysed in this way, absolute changes in FluPol activity were minor. Complementation data from minigenome assays with H5N1 5092K, pH1N1 Eng/195 and IBV Florida/06 were similar to the data shown in Figure 4.3. The reason for incomplete rescue, that is to say FluPol activity was never equal to control cells, is likely transfection efficiency: whereas all the control cells express ANP32 proteins, only a proportion of dKO cells (no more

79 than 40%) express ANP32 alongside the five minigenome assay plasmids. This was the reason why the control cells were excluded from statistical analysis by ANOVA.

Figure 4.3 Optimisation by titration of FLAG-tagged ANP32 expression plasmids with H3N2 Victoria polymerase Minigenome assays in dKO cells (24-well plates) complemented with FLAG- tagged ANP32 proteins. (A) Vic/75 polymerase activity when 5 ng, 50 ng, or 500 ng FLAG-tagged ANP32A, ANP32B, or equal amounts of ANP32A and ANP32B expression plasmids were co- transfected in dKO cells alongside H3N2 Vic/75 RNP components PB1, PB2, PA, NP, a firefly luciferase minigenome and Renilla control. Data shown are firefly activity normalised to Renilla, plotted as mean (SD) analysed by one-way analysis of variance (ANOVA) (n = 2 biological replicate experiments). ns, not significant; **, P<0.01. (B) Vic/75 polymerase activity when 200 ng empty pCAGGS, ANP32A, ANP32B, or equal amounts of both expression plasmids were co-transfected with H3N2 Vic/75 RNP components. (C) Vic/75 polymerase activity when increasing amounts of pCAGGS-ANP32B (0 – 20 – 40 – 80 ng) were co-transfected with H3N2 Vic/75 RNP components. Statistical interpretation of the assays in (B) and (C) is discussed in the main text.

Using the optimised complementation protocol, activity of a panel of FluPol constellations – Vic/75, 5092K, Eng/195 and Florida/06 – was measured in dKO cells, complemented with either ANP32A only, ANP32B only, or both proteins in equal amounts (Figure 4.4). The ratio of RNP components and ANP32 expression plasmids (PB1:PB2:PA:NP:reporter:Renilla:ANP32) was 2:2:1:4:2:2:5, unless stated otherwise in Figure legends.

All the FluPol constellations were rescued by exogenous transient expression of ANP32 proteins, regardless of whether both proteins were present or just one of them. In concordance with Figure 4.2 some polymerases displayed a preference for complementation with ANP32B over ANP32A: 5092K and Florida/06 polymerases were 2-3 fold more active in dKO cells in the presence of ANP32B than ANP32A, while the presence of both ANP32A and ANP32B reduced FluPol activity rather than enhance it further. Using an anti-FLAG antibody to detect ANP32 protein expression, western blotting analysis showed these phenotypic effects were not due to differential ANP32 expression.

Figure 4.4 Exogenous expression of ANP32 proteins recovers FluPol activity in dKO cells From Staller et al J Virol 2019 Minigenome assays in dKO cells with co-expressed FLAG-tagged ANP32 proteins or empty pCAGGS. H3N2 Vic/75 (A), H5N1 5092K (B), pH1N1 Eng/195 (C), or IBV Florida/06 (D) PB1, PB2, PA, NP, IAV or IBV firefly luciferase minigenome reporter, Renilla control and ANP32- FLAG were transiently expressed in a ratio of 2:2:1:4:2:2:5 (40 ng PB1 in ~200,000 cells). Data shown are firefly activity normalised to Renilla, plotted as mean (SD) analysed by one-way analysis of variance (ANOVA) from one representative (technical) triplicate repeat (n = 3 independent biological repeat experiments). ns, not significant; **, P<0.01; ***, P<0.001; ****, P<0.0001. Accompanying western blots show expression of FLAG-tagged ANP32 constructs alongside respective RNP components, representative of one minigenome assay. 30 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol. Lysates were resolved by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/1,000; Abcam), rabbit α-IAV PB2 (GTX125926, 1/2,000; GeneTex), and mouse α-FLAG (F1804, 1/500; Sigma-Aldrich), followed by HRP-conjugated secondary antibodies sheep α-rabbit IgG (AP510P, 1/10,000; Merck) and goat α-mouse IgG (STAR117P, 1/5,000; AbD Serotec), and developed on a Fusion-FX imaging system.

Minigenome reporter assays involve transfection of a set of 5-6 expression plasmids followed by cell lysis. Bioluminescence is then measured on whole cell lysate. In order to investigate FluPol activity at a single cell level, minigenome assays were carried out in dKO cells with an NLS-conjugated blue fluorescent protein (BFP) IAV-like RNA reporter instead of firefly luciferase (Figure 4.5). FluPol activity was established by fluorescence microscopy rather than bioluminescence, with green fluorescent protein (GFP)-conjugated PA acting as an internal transcription control. H5N1 5092 RNP components PB1, PB2-627K, GFP-tagged PA and NP were transfected alongside mCherry-tagged ANP32A or ANP32B. PA-GFP and ANP32-mCherry expression plasmids had been used in the laboratory previously and were known to still perform their respective functions. A small proportion of dKO cells (5-10%) showed blue fluorescence indicating active influenza polymerase, but only when ANP32 proteins were co-expressed (red fluorescence). Some cells expressed ANP32-mCherry or PA- GFP without showing active polymerase, suggesting not all vRNP components required for transcription of the reporter gene were present. In contrast, all the cells with active reconstituted FluPol expressed both ANP32-mCherry and PA-GFP.

Figure 4.5 On a single-cell level FluPol is active only in dKO cells co-expressing ANP32 proteins From Staller et al J Virol 2019 Expression from IAV minigenome encoding NLS-tagged BFP in eHAP dKO cells exogenously reconstituted with mCherry-tagged ANP32A (A) or ANP32B (B). Cells were transfected with expression plasmids encoding H5N1 5092 PB1, PB2-627K, GFP-tagged PA (green; top right panels), NP, a BFP minigenome reporter (blue; bottom left panels) and mCherry-ANP32 (red; top left panels) in a 2:2:1:4:2:5 ratio. Blue fluorescence indicating active FluPol is seen only in cells expressing ANP32. Approximately 200,000 cells were cultured on glass coverslips in 12-well plates and transfected as described. Cells were fixed in 4% paraformaldehyde and then visualised: coverslips were mounted on glass slides using Vectashield mounting medium (H-1000-10; Vector Laboratories). Cells were imaged with a Zeiss Cell Observer widefield microscope with ZEN Blue software, using a Plan- Apochromat 100 1.40-numerical aperture oil objective (Zeiss), an Orca-Flash 4.0 complementary metal- oxide semiconductor (CMOS) camera (frame, 2,048 2,048 pixels; Hamamatsu), giving a pixel size of 65 nm, and a Colibri 7 light source (Zeiss). Channels acquired and filters for excitation and emission were 4 =,6-diamidino-2- phenylindole (DAPI) (excitation [ex], 365/12 nm, emission [em] 447/60 nm), GFP (ex 470/40 nm, em 525/50 nm), and TexasRed (ex 562/40 nm, em 624/40 nm). Images were analysed and prepared with Fiji software. The detection limit was adjusted individually for each channel (taking care to remain well above control background level). Imaging was carried out by David Gaboriau of the Imperial College Facility for Imaging by Light Microscopy (FILM)

4.3 IAV replication is abrogated in dKO cells

Although minigenome reporter assay are extremely useful and flexible, it was important to understand the effects of ANP32 proteins in the context of infectious virus. Influenza viruses express nuclear export protein (NEP) and non-structural protein 1 (NS1) that are not provided in minigenome assays, but that are known to affect virus replication (Hale et al., 2008; Manz

83 et al., 2012). It is conceivable that these or other mechanisms have the capacity to overcome the block in replication imposed by the absence of ANP32 proteins in dKO cells.

Virus infection assays in our laboratory are carried out as follows. First, cells are infected at a multiplicity of infection (MOI, the number of infectious virions per cell) of choice. For multicycle growth curves, when virus is allowed to multiply over the course of several days and infect additional uninfected cells in the process, this MOI is generally 0.001 – 0.005. Such low numbers are believed to be relevant at early time points during influenza infection in vivo (Frise et al., 2016; Nikitin et al., 2014). To support multicycle growth in cell culture, trypsin is added to the growth medium. This protease is required to posttranslationally cleave the haemagglutinin (HA), activating the membrane fusion potential of the glycoprotein. In contrast, serum is left out of the growth medium because it inactivates trypsin.

The laboratory-adapted influenza A virus strain PR8 was used for multicycle growth curves on control, AKO, BKO and dKO cells, at an MOI of 0.001, i.e. one infectious virion per 1,000 cells. Additional growth curves were set up in dKO cells complemented by transient transfection with ANP32A and ANP32B expression plasmids six hours prior to infection with H1N1 PR8 (Figure 4.6). Virus replicated to similar titres in control, AKO and BKO cells, albeit trending slightly lower in BKO cells at early time points. Strikingly, no virus replication was observed in the dKO cells. In dKO cells complemented with ANP32 proteins, however, virus growth recovered. ANP32 proteins were thus essential for H1N1 PR/8 influenza virus replication in human eHAP cells.

Figure 4.6 H1N1 PR8 virus replication is abrogated in dKO cells From Staller et al J Virol 2019 (A) Control (black), AKO (red), BKO (blue) and dKO (purple) eHAP cells were infected with H1N1 PR8/1934 virus at MOI = 0.001 and incubated at 37⁰C in the presence of 1 µg/ml trypsin to allow multicycle replication. Supernatants were harvested at the indicated time points post-infection and PFU/ml established by plaque assay on MDCK cells. (B) dKO cells were transfected with equal amounts of FLAG-tagged ANP32A and ANP32B expression plasmids six hours prior to infection with H1N1 PR8/1934 virus at MOI = 0.001. Titres in dKO cells (purple) and dKO cells complemented with ANP32

84 proteins (green) were established as in (A). Data shown are mean (SD) PFU/ml measured by plaque assay on MDCK cells. LOD (dotted line) denotes the limit of detection based on the dilution factor in the plaque assays. Graphs shown are one representative (technical) triplicate assay (n = 3 independent biological replicate experiments). Statistical significance was calculated per time point by t test, comparing against control cells in (A) and against dKO cells in (B). ns, not significant; *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001.

In order to visualise virus infection at a single cell level, control and dKO cells were infected with H1N1 PR8 virus at MOI 0.2 for five hours. Cells were then fixed in 4% paraformaldehyde (PFA), permeabilized with 0.2% Triton and incubated with a primary antibody against IAV nucleoprotein (NP), the most abundant of the viral proteins. A fluorescent secondary antibody was used to measure NP accumulation in the cells (Figure 4.7). Most individual cells were infected at this relatively high MOI, but NP only accumulated in control cells, i.e. where ANP32 proteins were present. In dKO cells no accumulation of NP was seen, as in mock-infected dKO cells. This assay is thus not sensitive enough to detect incoming NP present in the vRNPs of the infecting virus, or primary transcription products. Presumably replication and secondary transcription is required for the signal to be captured.

Figure 4.7 Nucleoprotein does not accumulate in dKO cells From Staller et al J Virol 2019 Immunofluorescence analysis of NP expression in H1N1 PR8/33 infected eHAP control cells, dKO cells, or mock-infected dKO cells. ~250,000 cells were cultured on glass coverslips and infected with H1N1 PR8/33 virus (MOI 0.2) five hours prior to fixing in 4% paraformaldehyde (PFA) and permeabilisation in 0.2% Triton X-100 detergent. Cells were incubated with primary mouse α-IAV NP antibody (ab128193, 1/200; Abcam) followed by goat α-mouse Alexa Fluor-568 conjugated secondary antibody (1/200; Life

Technologies) in phosphate-buffered saline (PBS) buffer complemented with 1% bovine serum albumin (BSA) and 0.1% Tween 20 (blocking for 30 minutes at 37⁰C; antibody incubations for 1 hour each at 37⁰C in a humidified chamber). Cells were counterstained for 5 minutes at 37⁰C with 4’,6-diamidino-2- phenylindole (DAPI, 1/1,000) to stain chromosomal DNA. Coverslips were mounted on microscopy slides using Vectashield mounting medium. Imaging was carried out by David Gaboriau of the Imperial College Facility for Imaging by Light Microscopy (FILM) on a Zeiss Cell Observer widefield microscope using the DAPI (ex 365/12 nm, em 447/60 nm) and TexasRed (ex 562/40 nm, em 624/40 nm) channels.

PR8 is a lab-adapted virus that is pathogenic in inbred mice and has been used for vaccine production as it is attenuated in humans (Burnet, 1937; Francis et al., 1945; Hannoun, 2013). Since its isolation in 1934, the strain has been extensively passaged in cell culture, mice and eggs, losing its ability to transmit between humans in the process. Multicycle growth curves in control, AKO, BKO and dKO cells were set up with the more relevant influenza virus strains whose RNP-encoding genes had been used in minigenome assays in Figures 4.2 and 4.4. These strains included a H3N2 Vic/75 6:2 reassortant virus with the external genes haemagglutinin (HA) and neuraminidase (NA) from PR8, a H5N1 5092K 5:3 reassortant with HA, NA and matrix (M) genes from PR8, and a wildtype pH1N1 Eng/195 virus with all its own genes. Using reassortant viruses with PR8 HA and NA genes is a safety measure so that pathogenic viruses can be used at biosafety level 2, as PR8 virus does not transmit among humans. On the other hand, PR8 is known to infect most cultured cells very well, and the replication phenotype under study here will not be affected by the glycoproteins. Upon low- MOI infection, all three IAV strains replicated to high titres in control, AKO and BKO cells, with slight attenuation in BKO cells at early time points. As with PR8 virus, replication was completely abrogated in dKO cells (Figure 4.8).

None of the IAV viruses so far tested achieved multicycle replication in cells lacking ANP32A and ANP32B. It was thus concluded that IAV replication was dependent on ANP32 proteins, and that ANP32A and ANP32B were functionally redundant in their support for influenza A virus replication. Infection with IBV Florida 06 had been unsuccessful – no plaques were recovered even from control cells, suggesting that eHAP cells are not susceptible to IBV, at least not to the same extent as to IAV.

Figure 4.8 IAV replication is abrogated in dKO cells From Staller et al J Virol 2019 Control (black), AKO (red), BKO (blue) and dKO (purple) eHAP cells were infected with H3N2 Vic/75 6:2 reassortant virus with PR8 HA and NA (A), H5N1 5092K 5:3 reassortant virus with PR8 HA, NA and M genes (B), or pH1N1 Eng/195 (C) at MOI = 0.005 and incubated at 37⁰C in the presence of 1 µg/ml trypsin to allow multicycle replication. Supernatants were harvested at the indicated time points post-infection and PFU/ml established by plaque assay on MDCK cells. Data shown are mean (SD) PFU/ml measured by plaque assay on MDCK cells. LOD (dotted line) denotes the limit of detection based on the dilution factor in the plaque assays. Graphs shown are one representative technical triplicate assay (n = 3 independent biological repeat experiments). Statistical significance was calculated per time point by t test, comparing against control cells. ns, not significant; *, P<0.05; **, P<0.01; ***, P<0.001.

4.4 HPAI H5N1 virus can replicate to low titres in dKO cells A/turkey/Turkey/1/2005 (H5N1) is a highly pathogenic avian influenza (HPAI) virus that caused widespread morbidity and mortality in 2005 - 06, not to mention worldwide anxiety among public health experts due to its pandemic potential (Oner et al., 2006). This avian virus encodes a lysine residue at position 627 of the PB2 subunit, rather than the glutamate normally seen in avian-adapted FluPol, as a result of selection for this genotype in an unknown mammalian host and reintroduction into birds. This mammalian adaptation of the viral polymerase of this set of H5N1 viruses increases their potential pandemic threat.

Control and dKO eHAP cells were infected with a H5N1 Turkey/05 6:2 reassortant virus with PR8 HA and NA to test if it could replicate in the absence of ANP32 proteins. H5N1 Tky/05 replicated to high titres in eHAP control cells. Interestingly, a few plaques were consistently observed on MDCK cells inoculated with supernatant derived at late time points (48-72 hpi) from the dKO cells (Figure 4.9). To date, Turkey/05 has been the only influenza A virus that has consistently shown replication in cells lacking ANP32A and ANP32B. Still, replication in dKO cells was significantly lower than in control cells. Work is currently underway to understand why this virus strain seems to retain some capacity to replicate in the absence of ANP32 proteins.

Figure 4.9 H5N1 Tky/05 can replicate to low titre in dKO cells Control (black) and dKO (purple) eHAP cells were infected with H5N1 Tky/05 6:2 reassortant virus with PR8 HA and NA at MOI = 0.005 and incubated at 37⁰C in the presence of 1 µg/ml trypsin to allow multicycle replication. Supernatants were harvested at the indicated time points post-infection and PFU/ml established by plaque assay on MDCK cells. Data shown are mean (SD) PFU/ml measured by plaque assay on MDCK cells. LOD (dotted line) denotes the limit of detection based on the dilution factor in the plaque assays. Graphs shown are one representative technical triplicate assay (n = 3 independent biological replicate experiments). Statistical significance was calculated per time point by t test. ***, P<0.001; ****, P<0.0001.

4.5 Influenza virus RNA does not accumulate in dKO cells FluPol is both a transcriptase – synthesising positive sense mRNA from the negative sense genomic RNA (vRNA) – and a replicase synthesising novel virus genomes via an intermediate positive sense complementary RNA (cRNA). Accumulation of these three influenza virus RNA species (vRNA, cRNA, and mRNA) can be measured by various means, including radiolabelled primer extension assays (Robb et al., 2009) and reverse transcription of the RNA to complementary DNA (cDNA) followed by quantitative PCR (RT-qPCR). In the reverse transcription step, primers specific to each RNA species are used (Kawakami et al., 2011). RT-qPCR was used to quantify RNA accumulation. A potential issue with this method was accumulation of non-specific products, even in negative control samples with no viral RNA present, possibly due to self-priming activity of the RT primers (Prof RAM Fouchier, personal communication). This problem was eliminated through the design of strand-specific RT primers with a unique 5’ tag sequence, followed by PCR using the tag sequence as the forward primer (Kawakami et al., 2011). The tagged RT primers were designed against PR8 neuraminidase (NA) to enable quantification of all recombinant viruses that varied in their internal gene constellation. cDNA synthesis with tagged RT primers followed by PCR yielded

88 a single product of the right size, with no non-specific products visible on agarose gel electrophoresis of the PCR amplicons (Figure 4.10), suggesting amplification of a single PCR product without non-specific products derived from self-priming.

Figure 4.10 Tagged strand-specific RT primers eliminate non-specific PCR products (A) Cartoon illustrating reverse transcription primers specific to PR8 neuraminidase (segment 6) vRNA, cRNA or mRNA, each with an 18-20 nucleotide tag (blue, orange and green, respectively) at the 5’ end that was unrelated to influenza virus (Kawakami et al., 2011). (B) PCR with forward primers identical to the 5’ tag yields a single product of the right size. ~500,000 MDCK cells were infected with H1N1 PR8 virus at an MOI of 1. 24 hours post-infection total RNA was extracted using the RNeasy minikit with on- column DNase I treatment for 1 hour to degrade genomic DNA. Total RNA concentration was obtained by spectrophotometry (NanoDrop) and cDNA was synthesised from equal amounts of total RNA (500 ng) using 5’-tagged RT primers. In order to mirror the qPCR protocol, 5 ng of the cDNA was amplified by PCR using Fast SYBR green reagents and accompanying thermocycler settings recommended by the protocol (40 cycles of denaturing at 95⁰C for 5 seconds; annealing at 58⁰ for 15 seconds; extension at 72⁰C for 10 seconds). GAPDH was used as a cellular transcription control amplified via RT with a poly (dT) primer. PCR amplicons were resolved by 2% agarose gel electrophoresis for 30 minutes at 130V.

Direct comparison of cDNA synthesis with the tagged RT primers versus the previous untagged set, followed by qPCR amplification of the RT products, was carried out by Miss OC Swann in our laboratory (Figure 4.11). The results showed that non-specific products

89 accumulated using the untagged RT primers to specifically amplify either vRNA, cRNA, or mRNA, but not using the novel tagged set.

Figure 4.11 Tagged RT primers eliminate accumulation of non-specific products during RT- qPCR analysis. eHAP control cells were infected with H1N1 PR8 virus at an MOI of 3, and total RNA extracted 6 hours post-infection using the RNeasy minikit with on-column DNase I treatment for 1 hour to degrade genomic DNA. Total RNA concentration was obtained by spectrophotometry (NanoDrop) and equal amounts of RNA (500 ng) were subjected to cDNA synthesis using RevertAid reverse transcriptase and either untagged (left) or tagged (right) primers against NA vRNA. Unique fragments of the NA gene were then amplified by real-time quantitative PCR using Fast SYBR green master mix.

Data show mean (SD) 40-CT values of one representative experiment in technical triplicate (n = 2 independent biological repeat experiments). Only the vRNA experiment is shown here but results were similar with cRNA and mRNA. This analysis was carried out by Miss OC Swann.

The incoming FluPol molecules associated with each vRNA segment transcribe the vRNA into mRNA in cis, which is then translated into viral proteins by the host cell machinery. RNP components PB1, PB2, PA and NP then shuttle back into the cell nucleus where they stabilise nascent cRNA replicated from the incoming vRNA (Figure 4.12 A). Thus, de novo production of FluPol is required to avoid degradation of the cRNA. In addition, the newly synthesized FluPol proteins are required to support vRNA synthesis from a cRNA template, which will only proceed in the presence of trans-activating or trans-acting FluPol (Jorba et al., 2009; Vreede et al., 2004; York et al., 2013). Once the cRNA has been replicated back into more vRNA, secondary transcription can start and vRNPs can become incorporated in new virions.

It has been suggested that ANP32A and ANP32B specifically support the synthesis of vRNA from a cRNA template (Sugiyama et al., 2015), although the evidence is not conclusive. To test what viral RNA species were synthesized in the absence of ANP32 proteins we set up the following eHAP control and dKO cells transfected transiently with a catalytically inactive polymerase complex composed of H5N1 5092 PB1-D446Y (Biswas and Nayak, 1994), PB2-

627K and PA in a 1:1:1 ratio. This FluPol constellation allowed stabilisation of cRNA without contributing to replication. Twenty hours after transfection the cells were infected with H1N1 PR8 virus at MOI 10 for four hours, prior to RNA extraction and RT-qPCR. Upon infection, cells in some conditions were treated with the translation inhibitor cycloheximide (CHX), so that transcription of the vRNA by the incoming FluPol did not lead to synthesis of catalytically active trans-activating FluPol. If ANP32 proteins are indeed required for cRNA-to-vRNA replication, accumulation of vRNA above the incoming level should be observed only in control cells expressing ANP32 proteins, and only in the absence of CHX.

Indeed, vRNA levels in control cells were significantly higher in the absence of CHX, which was not the case in dKO cells. In addition, mRNA levels exceeded primary transcription levels in control cells, while there was a reduction in mRNA in the dKO cells in the absence of CHX. More cRNA was measured in control cells in the absence of CHX, compared to dKO cells, presumably because additional cRNA was synthesised from newly synthesised vRNA. The fact that cRNA was observed in dKO suggests cRNA synthesis on a vRNA template (by the cis-acting incoming polymerase) is not affected in the absence of ANP32 proteins3.

3 Although this Figure was published as part of Staller et al. J Virol 2019, we no longer believe our conclusions are supported by the data. The figure lacks earlier time points that may illuminate the accumulation of incoming vRNA versus primary transcription products versus secondary transcription products. Subsequent work by Miss OC Swann and Dr CM Sheppard of the Barclay laboratory supports the notion that ANP32 proteins are required for vRNA-to-cRNA replication as well as cRNA-to-vRNA replication. Further work is required to differentiate between the replication steps dependent on ANP32 proteins.

Figure 4.12 RNA synthesis is abrogated in dKO cells (A) Cartoon of the experimental setup. Catalytically inactive (PB1-D446Y) FluPol is pre-expressed prior to infection with influenza virus. The incoming cis-acting FluPol associated with the viral genome will synthesise mRNA (primary transcription) which is then translated into novel (trans-activating and/or trans-acting) FluPol, which cycles back into the cell nucleus where it stabilises the cRNA synthesised from the incoming vRNA during the first replication step. This cycle is broken by cycloheximide treatment which blocks translation of new virus proteins. In the absence of trans-activating FluPol, the pre-expressed catalytically inactive FluPol will stabilise the cRNA without synthesising additional genomic vRNA. In cells that are not treated with CHX, catalytically active FluPol will get synthesised and vRNA synthesis and secondary transcription can proceed. (B-D) From Staller et al J Virol 2019 RT-qPCR analysis demonstrating accumulation of PR8 segment 6 (NA) vRNA (B), cRNA (C) and mRNA (D) in the absence or presence of 100 µg/ml cycloheximide (CHX). ~250,000 eHAP control or dKO cells were transfected with H5N1 5092 polymerase subunits PB1-D446Y (catalytically dead), PB2-627K and PA in a 1:1:1 ratio twenty hours before infection with H1N1 PR8 virus at an MOI of 10 to ensure infection of all the cells. Total RNA was extracted 4 hours post-infection using the RNeasy minikit, with 30 minutes on-column DNase I treatment to degrade genomic DNA. Total RNA concentration was obtained by spectrophotometry (NanoDrop) and equal amounts of RNA (500 ng) was subjected to cDNA synthesis using RevertAid reverse transcriptase and RNA species-specific tagged primers. Unique fragments of the NA gene were then amplified by real-time quantitative PCR using Fast SYBR green master mix and primers against

92 the 5’ tag of the RT primers. Data show mean (SD) 40-CT values normalised to mean mock-infected levels, analysed per cell type by one-way analysis of variance (ANOVA). Experimental data are representative of three independent biological repeats carried out in technical triplicates. ***, P<0.001; ****, P<0.0001.

4.6 Mouse ANP32A does not support FluPol activity

Next, the double knockout cells were used to test pro-viral function of ANP32 proteins from non-human animals that are relevant either as reservoirs, intermediary hosts or animal models of influenza viruses. For example, minigenome assays in dKO cells have helped establish why avian ANP32B cannot be co-opted by FluPol (Long et al., 2019a; Zhang et al., 2019a) and that ANP32A from swine can support avian signature FluPol to some extent (Peacock et al., 2020b; Zhang et al., 2020).

Minigenome reporter assays were carried out in dKO cells complementing H3N2 Vic/75 and H5N1 5092K vRNPs with ANP32A proteins of swine, mouse, chicken and duck origin (Figure 4.13). Whereas chicken, swine and duck ANP32A all supported FluPol activity in dKO cells, murine ANP32A (MusA) did not. This was not due to lower expression of MusA, as analysed by immunoblotting. Primary sequence alignment showed differences between pro-viral human ANP32A and murine ANP32A, and between murine ANP32A and ANP32B (MusB) in surface- exposed amino acid residues at the C-terminal end of the LRR domain. In particular, amino acid residue 130 of MusA was an alanine instead of the canonical aspartate.

Figure 4.13 Murine ANP32A does not support IAV FluPol activity From Staller et al J Virol 2019 (A) Minigenome reporter assay in eHAP dKO cells with co-transfected FLAG-tagged ANP32A proteins from swine (SusA), mouse (MusA), duck (AnasA) and chicken (GallusA) with H3N2 Victoria RNP components PB1, PB2, PA and NP, pPolI-firefly luciferase minigenome reporter and Renilla control, in a ratio 2:2:1:4:2:2. Data show mean (SD) firefly activity normalised to Renilla and analysed by one-way

ANOVA from one representative technical triplicate repeat (n = 2 independent biological repeat experiments). ns, not significant; ****, P<0.0001. Accompanying western blots show expression of vRNP component PB2 and FLAG-tagged ANP32 constructs, alongside vinculin loading control. 30 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol. Lysates were resolved by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/1,000; Abcam), rabbit α-IAV PB2 (GTX125926, 1/2,000; GeneTex), and mouse α-FLAG (F1804, 1/500; Sigma-Aldrich), followed by HRP-conjugated secondary antibodies sheep α-rabbit IgG (AP510P, 1/10,000; Merck) and goat α-mouse IgG (STAR117P, 1/5,000; AbD Serotec), and developed on a Fusion-FX imaging system.

We hypothesised that in mice, IAV FluPol might rely solely on MusB to support its replication. Indeed, complementation of FluPol with MusB recovered FluPol activity in dKO cells (Figure 4.14). In order to test whether the differences in LRR 5 were responsible for this phenotype, a FLAG-tagged chimera of MusA and MusB (MusA128-153) was generated by substituting amino acid residues 128-153 of MusB into MusA. Insertion of this MusB segment conferred gain of function on MusA; the chimera could recover FluPol activity in dKO cells, albeit not to the extent shown by MusB. These phenotypes could not be explained by differences in expression or nuclear localisation of the protein constructs.

Figure 4.14 IAV polymerase is supported by murine ANP32B From Staller et al J Virol 2019 (A-C) Minigenome reporter assays in eHAP dKO cells showing activity of reconstituted H3N2 Vic/75 (A),

H5N1 5092K (B) and pH1N1 Eng/195 FluPol complemented with FLAG-tagged murine ANP32A (MusA), ANP32B (MusB) or ANP32A128-153 chimeric construct (MusA128-153). Data show mean (SD) firefly activity normalised to Renilla analysed by one-way ANOVA from one representative technical triplicate repeat (n = 2 independent biological repeat experiments). ns, not significant; *, P<0.05; ****, P<0.0001. Accompanying western blots show expression of vRNP component PB2 and co-expressed FLAG-tagged ANP32 constructs alongside vinculin loading control. 30 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol. Lysates were resolved by SDS- PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/1,000; Abcam), rabbit α-IAV PB2 (GTX125926, 1/2,000; GeneTex), and mouse α-FLAG (F1804, 1/500; Sigma-Aldrich), followed by HRP- conjugated secondary antibodies sheep α-rabbit IgG (AP510P, 1/10,000; Merck) and goat α-mouse IgG (STAR117P, 1/5,000; AbD Serotec), and developed on a Fusion-FX imaging system. (D) Structural model of murine ANP32A/ANP32B chimera ANP32A128-153 highlighting the swapped domain in blue with amino acid residue 130D in red stick format. Homology model was created in iTASSER under supervision of Dr CM Sheppard, using PDB submissions 2RR6A (human ANP32B), 2JQDA (human ANP32A) and 2JEOA (human ANP32A) as templates. The LRR and N-terminal region of the LCAR are shaded dark grey, corresponding to the structurally resolved N-terminal region (amino acids 1-161) of human ANP32B (Tochio et al., 2010). The intrinsically disordered LCAR is in semi-transparent grey. (E) Immunofluorescence analysis showing nuclear localisation of MusA, MusB and MusA128-153 detected with mouse anti-FLAG primary antibody (F1804, 1/200; Sigma-Aldrich) and Alexa Fluor-568 goat anti- mouse conjugate (1/200; Invitrogen). ~200,000 eHAP dKO cells were cultured on glass coverslips and transfected with 200 ng FLAG-tagged ANP32 pCAGGS expression plasmid 24 hours prior to fixing in 4% PFA and permeabilisation in 0.2% Triton X-100. Blocking and staining with primary and secondary antibodies was carried out in PBS supplemented with 1% bovine serum albumin (BSA) and 0.1% Tween 20 (blocking for 30 minutes at 37⁰C; antibody incubations for 1 hour each at 37⁰C in a humidified chamber). Chromosomal DNA was stained with DAPI for 5 minutes at 37⁰C. Coverslips were mounted on microscopy slides using Vectashield mounting medium. Imaging was carried out by David Gaboriau of the Imperial College Facility for Imaging by Light Microscopy (FILM) on a Zeiss Cell Observer widefield microscope using the DAPI (ex 365/12 nm, em 447/60 nm) and TexasRed (ex 562/40 nm, em 624/40 nm) channels.

Next, minigenome reporter assays were carried out in eHAP dKO cells with MusA or MusB in support of reconstituted IBV Florida/06 polymerase. Surprisingly, Florida/06 FluPol recovered some activity in dKO cells in the presence of MusA, although MusB remained the more potent factor (Figure 4.15). Although expression of MusA compared with Empty vector did not reach statistical significance when analysed by one-way ANOVA, comparison of the two conditions by t test gave a P value of 0.0003. Similarly, direct comparison of luminescence by each FluPol constellation in dKO cells in the presence or absence of MusA confirmed that IBV Florida/06 polymerase gained activity when MusA was co-expressed, unlike any of the IAV polymerases tested (Figure 4.15B). Intriguingly, IBV polymerase activity reached even higher levels in the

95 presence of the chimeric construct MusA128-153 compared to MusB alone. These differences could not be explained by differences in expression as measured by western blotting analysis. These findings may suggest the exact residues in ANP32 presented for support of polymerase by ANP32 proteins are different between IAV and IBV polymerase.

Figure 4.15 IBV polymerase gains activity in the presence of MusA (A) From Staller et al J Virol 2019 Minigenome reporter assay in eHAP dKO cells showing activity of reconstituted IBV Florida/06 FluPol complemented with FLAG-tagged murine ANP32A (MusA), ANP32B (MusB) or ANP32A128-153 chimeric construct (MusA128-153). Data show mean (SD) firefly activity normalised to Renilla analysed by one-way ANOVA from one representative technical triplicate repeat (n = 2 independent biological repeat experiments). ****, P<0.0001. Accompanying western blots show expression of FLAG-tagged ANP32 constructs and vinculin loading control. 30 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol. Lysates were resolved by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/1,000; Abcam) and mouse α-FLAG (F1804, 1/500; Sigma-Aldrich), followed by HRP-conjugated secondary antibodies sheep α-rabbit IgG (AP510P, 1/10,000; Merck) and goat α-mouse IgG (STAR117P, 1/5,000; AbD Serotec), and developed on a Fusion-FX imaging system. (B) Minigenome assay comparing firefly luciferase values in dKO cells expressing IAV Vic/75, 5092K, or Eng/195, or IBV Florida/06 RNP components complemented with either MusA or Empty pCAGGS expression plasmid. Data show mean (SD) firefly activity; statistical analysis by t test comparing each FluPol constellation’s activity in the absence (Empty) or presence (MusA) of murine ANP32A. ns, not significant; ****, P<0.0001.

Finally, a D130A single amino acid substitution was introduced in human ANP32A, as well as the reverse A130D substitution in MusA. Amino acid substitutions at positions 129 and 130 of chicken ANP32 proteins had major phenotypic effects on their capacity to support FluPol activity (Long et al., 2019a). H3N2 Vic/75 activity was reduced significantly in dKO cells complemented with human ANP32A-D130A, compared with wildtype ANP32A, while there

96 was a small but significant gain-of-function in the ability of MusA to support FluPol activity when aspartate was introduced at position 130 (Figure 4.16) Similar results have been presented in another study (Zhang et al., 2019a).

Figure 4.16 Single residue substitutions at position 130 of ANP32A affect FluPol activity. From Staller et al J Virol 2019 Minigenome assay in eHAP dKO cells measuring activity of reconstituted H3N2 Vic/75 polymerase co-expressing wildtype human or mouse ANP32A or position 130 point mutants. Data show mean (SD) firefly activity normalised to Renilla and analysed by t test from one representative technical triplicate repeat (n = 3 independent biological replicate experiments). **, P<0.01; ****, P<0.0001. Accompanying western blots show expression of vRNP component PB2 and co-expressed FLAG-tagged ANP32 constructs. 30 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol. Lysates were resolved by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/1,000; Abcam), rabbit α-IAV PB2 (GTX125926, 1/2,000; GeneTex), and mouse α-FLAG (F1804, 1/500; Sigma-Aldrich), followed by HRP-conjugated secondary antibodies sheep α-rabbit IgG (AP510P, 1/10,000; Merck) and goat α-mouse IgG (STAR117P, 1/5,000; AbD Serotec), and developed on a Fusion-FX imaging system.

4.7 Murine ANP32A binds IBV polymerase more strongly than IAV polymerase One reason why MusA does not support FluPol activity may be that it does not properly interact with IAV FluPol. Work in our laboratory and others had established that ANP32 proteins bind directly to IAV polymerase, using a variety of assays including co-immunoprecipitation (Baker et al., 2018; Domingues and Hale, 2017; Mistry et al., 2020), bimolecular fluorescence complementation assays (BiFC) (Mistry et al., 2020) and split luciferase complementation assays (Long et al., 2019a). Pro-viral function of ANP32 proteins has been correlated with binding to FluPol. Chicken ANP32B, which does not support influenza polymerase, does not interact with H5N1 5092 FluPol either, regardless of the amino acid at position 627 of the PB2

97 subunit (E or K), as measured by split luciferase complementation assay (Long et al., 2019a). We hypothesised that the inability of murine ANP32A to support influenza A virus polymerase activity, might be explained by a lack of interaction with FluPol.

A split luciferase complementation assay to investigate FluPol-ANP32 interactions was developed in the laboratory by Dr B Mistry. The assay is based on Gaussia princeps luciferase, a small 185-amino acid (19.9 kDa) monomeric protein isolated from the marine copepod G. princeps (Remy and Michnick, 2006). The human codon-optimised form of this luciferase is both smaller and brighter than Photinus pyralis (firefly) or Renilla reniformis (Renilla) luciferases. The protein is split into N-terminal (amino acids 18-109; luc1) and C-terminal (amino acids 110-185; luc2) fragments, which are fused to potentially interacting proteins. When the proteins interact, the luciferase is reconstituted and luminescence can be measured by addition of a substrate (Gaussia and Renilla luciferases share the same substrate, coelenterate-luciferin). Split luciferase assays have been used widely, including in influenza virology (Munier et al., 2013). Optimisation experiments in our laboratory showed that ANP32- FluPol interactions are best measured with the luc1 fragment fused to the C-terminus of the PB1 subunit of FluPol (PB1-luc1), and luc2 fused to the C-terminus of ANP32 (ANP32-luc2) (Mistry et al., 2020).

Given that MusA seemed to be capable of supporting IBV polymerase function to some extent, but not IAV FluPol, we hypothesised an interaction between MusA and IBV polymerase may exist, but none between MusA and IAV polymerase. In contrast, as MusB supports both IAV and IBV polymerase activity, we might see interaction between these proteins. These interactions between murine ANP32 proteins and influenza B polymerase were probed using a split luciferase assay. IBV Florida/06 PB1-luc1 was generated by overlap extension PCR, as well as MusA-luc2, MusB-luc2 and MusA128-153-luc2. First, minigenome reporter assays were carried out to test whether the constructs had retained functionality (Figure 4.17), as this will suggest correct folding of the fusion proteins. IBV polymerase with luc1-tagged PB1 retained about 20% of its activity in eHAP control cells (i.e. supported by human ANP32A and ANP32B) compared to untagged PB1, and also in dKO cells complemented with murine ANP32B. MusA, MusB and the chimera all retained full capacity to rescue IBV FluPol in dKO cells when fused to luc2 at the C-terminus. Although retention of function is not strictly necessary for interaction assays, it is important that the tertiary and quaternary structure of the tagged proteins and protein complexes remain intact. Functionality can thus be used as a proxy for correct folding and assembly.

Figure 4.17 Gluc-tagged PB1 and ANP32 constructs retain function (A) Minigenome assays in eHAP control and dKO cells comparing reconstituted IBV FluPol function with or without the PB1 subunit fused at the C-terminus to Gaussia luciferase N-terminal fragment (luc1). Data show are mean (SD) firefly activity normalised to Renilla from a technical triplicate experiment (n = 1 independent biological repeat experiment), analysed by one-way analysis of variance per cell type (control or dKO) (B) Minigenome assay in dKO cells comparing IBV Florida/06 FluPol activity with co-transfected murine ANP32A, ANP32B, or ANP32A128-153 (musAB) with or without Gaussia luciferase C-terminal fragment (luc2). Data show are mean (SD) firefly activity normalised to Renilla from a technical triplicate experiment (n = 1 independent biological repeat experiment), analysed by one-way ANOVA. ns, not significant; **, P<0.01; ****, P<0.0001.

Next, split luciferase complementation assays were carried out using human ANP32A-luc2 as a positive control as it is known to interact with IAV polymerase (Long et al., 2019a; Mistry et al., 2020). Two control conditions were set up per experimental condition: either untagged PB1 with luc2-tagged ANP32 and an unbound luc1 fragment, or luc1-tagged PB1 with untagged ANP32 and the unbound luc2 fragment. The normalised luminescence ratio (NLR) was then calculated by dividing the luminescence obtained in the experimental condition (luc1- tagged PB1 + luc2-tagged ANP32) by the sum of the control conditions.

MusA and MusA128-153 showed low reconstituted Gaussia luciferase signal with IAV polymerase, compared with human ANP32A and MusB, suggesting poor interaction. In contrast, luminescence signals obtained with IBV polymerase were not different between human ANP32A, MusA and MusA128-153. MusB interacted significantly more strongly than MusA (Figure 4.18). These results suggest murine ANP32A interacts very poorly with influenza A virus polymerase. But MusA does interact with influenza B virus polymerase, which it supports to some extent functionally. The split luciferase binding assay thus recapitulates the minigenome assay findings. Strong binding of MusA128-143 to IBV polymerase was not observed, although this construct was most potent in its support for IBV FluPol activity (compare Figure 4.15). None of the differences we observed were due to variable expression of the constructs, as analysed by immunoblotting.

Figure 4.18 Murine ANP32A binds IBV polymerase more strongly than IAV polymerase (A) From Dr B Mistry. Schematic showing the split luciferase complementation assay with the N-terminal fragment of Gaussia luciferase (Luc1) fused to the C-terminus of the PB1 subunit of FluPol and the C-terminal fragment (Luc2) fused to the C-terminus of ANP32. Binding of ANP32 to FluPol reconstitutes an active Gaussia luciferase and a bioluminescence signal is generated upon addition of coelenterate-luciferin, the luciferase substrate. (B-C) Split luciferase complementation assays measuring interaction between luc2-tagged human ANP32A (huA), murine ANP32A (MusA), murine ANP32B (MusB), or the chimera (MusA128-153) and either reconstituted IAV Vic/75 (B) or IBV Florida/06 (C) polymerase with luc1-tagged PB1. Equal amounts (15 ng) of pCAGGS expression plasmids encoding FluPol components PB1-luc1, PB2 and PA, and the indicated ANP32-luc2 were transfected into HEK293T cells. Cells were lysed 24 hours post-transfection and luminescence read on a plate reader. NLR was calculated as explained in the main text and in Chapter II Materials and Methods. (D) Western blotting analysis showing expression of PB1-luc1 and ANP32-luc2 alongside vinculin loading control. 50 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol. Lysates were resolved by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/1,000; Abcam) and rabbit α-Gaussia luciferase (PA1-181, 1/1000; ThermoFisher), followed by HRP-conjugated secondary antibody sheep α-rabbit IgG (AP510P, 1/10,000; Merck), and developed on a Fusion-FX imaging system.

100

4.8 Discussion This chapter has demonstrated that ANP32A and ANP32B are essential host factors for influenza A and B virus polymerase activity, and for IAV replication in human cells. This pair of host factors is functionally redundant in their support of FluPol; ablation of either ANP32A or ANP32B does not lead to abrogation of FluPol activity or virus replication. Ablation of both paralogues, however, leads to complete inactivity of FluPol and abrogation of replication. This redundancy is the likely reason why ANP32 proteins have failed to be identified as essential for influenza virus replication in RNAi or CRISPR screens, as such approaches rely on single- gene knockout followed by a survival (live/dead) screen upon viral challenge (Han et al., 2018; Karlas et al., 2010; Li et al., 2020). Our findings were corroborated by an independent study using a similar CRISPR/Cas9 editing approach in HEK293T cells (Zhang et al., 2019a), lending strength to our conclusions.

Influenza B virus replication in the eHAP cells has not been properly tested, so it is unclear whether these common seasonal viruses also require ANP32 proteins in order to achieve replication, even if minigenome assays suggest they do. A single isolate, H5N1 Tky/05, was described in which we consistently do see some replication in human cells lacking ANP32A and ANP32B, and also in chicken cells lacking ANP32A expression (unpublished data). As chickens have only one functional homologue (chicken ANP32B does not support FluPol activity), chicken cells lacking ANP32A are the equivalent of human dKO cells. The conundrum of H5N1 Tky/05 replication in the absence of ANP32A and ANP32B is being actively pursued in our laboratory. One interesting hypothesis is that IAV Tky/05 FluPol has somehow evolved to utilise human ANP32E (or chicken ANP32B or ANP32E), which are normally not pro-viral. We know that human ANP32E cannot support FluPol under normal circumstances, as it is present in the dKO cells. More about ANP32E will be said in Chapter 6.

It was intriguing to identify an ANP32 orthologue that does not support IAV FluPol activity, namely murine ANP32A. Although mice are not natural influenza virus hosts, they are widely used as model organisms in the laboratory. Using minigenome reporter assays we demonstrated that murine ANP32B is the sole pro-influenza A virus host factor in mice, and that much of the difference in pro-viral function is explained by the identity of the amino acid residue at position 130. This was confirmed independently using a similar approach (Zhang et al., 2019a). Interestingly, murine ANP32A seemed capable of supporting influenza B virus polymerase, albeit less well than murine ANP32B.

An obvious experiment we considered was to obtain ANP32B knockout mice and to compare the outcome of infection with IAV or IBV. We hypothesised that ANP32B KO mice would be protected against influenza A virus challenge but not influenza B virus challenge. A conditional

101

C57BL/6J ANP32B knockout mouse with Anp32B exon 4 flanked by LoxP sites has been developed (Chemnitz et al., 2019). It was important for the knockout to be conditional, as previous work in the literature had shown that systemic ANP32B knockout in mice led to poor results – in a Black 6 background the mice do not survive gestation while BALB/c ANP32B KO mice show runtism and major organ failure (Reilly et al., 2011). We inquired about obtaining these mice but dropped the project when it became clear that influenza groups at the German institute were already doing the experiments we had envisioned. Indeed a paper came out recently describing how ANP32B KO mice, but not ANP32A KO mice, showed reduced viral loads and mortality when infected with H3N2 or H5N1 influenza A virus (Beck et al., 2020). IBV infection, however, had not been tested.

A preference for ANP32B over ANP32A was detected for some of the polymerase constellations in minigenome assays, and to a smaller extent at early time points in virus replication experiments. This effect was particularly evident for influenza B virus Florida/06 and H5N1 5092K polymerases. We also saw a reduction in polymerase activity when both ANP32 orthologues are expressed, compared with ANP32B on its own. It is possible that ANP32B is preferred simply because it is more abundant in tissues normally targeted by influenza viruses. Indeed, in healthy lung tissue there are almost five-fold more ANP32B than ANP32A transcripts, median transcripts per kilobase million (TPM) of 201 and 42, respectively (Lonsdale et al., 2013). Moreover, at a half-life of 28 hours ANP32B is more stable than ANP32A, which has a half-life of 12.5 hours (Fries et al., 2007). Although we do consistently see more ANP32B than ANP32A expressed in our cell lines this could be due to primary antibody quality. A weakness of the work here presented is that we have failed to properly account for gene dosing effects. We have in fact purified high amounts of ANP32A and ANP32B in E. coli – again obtaining more ANP32B than ANP32A – and these lysates of known concentration and purity should be used to establish quantitatively the amount of ANP32 protein endogenously expressed in our cell lines. We will then also be able to normalise exogenously expressed FLAG-tagged versions of the proteins to these quantified naturally expressed amounts.

Abundance and stability, however, would not explain why some polymerases prefer ANP32B over ANP32A but others do not. After all, all influenza viruses have similar tropisms. Another hypothesis to explain ANP32B preference is stronger binding of ANP32B to FluPol. It is conceivable that ANP32A and ANP32B bind competitively to FluPol, such that in the absence of ANP32A FluPol can recruit the preferred ANP32B more easily. This would explain the effect that FluPol activity is reduced when both ANP32A and ANP32B are expressed, compared to ANP32B on its own. Alternatively, ANP32B might be held in heterodimers or larger complexes with ANP32A. Absence of ANP32A might liberate ANP32B for recruitment by FluPol.

102

We described an RT-qPCR approach to quantify accumulation of influenza virus RNA species mRNA, vRNA and cRNA, using tagged RT primers and a PCR primer pair with the tag as the forward primer. This approach has allowed us to differentiate between levels of incoming vRNA versus vRNA replicated from the intermediary cRNA, and also between mRNA generated during primary transcription compared with secondary transcription. We also show that cRNA accumulated in initial replication in cis by the incumbent FluPol associated with the incoming vRNA. It is clear that vRNA accumulates in control cells containing ANP32 proteins, in the absence of cycloheximide, but not in dKO cells. We have not managed, however, to provide conclusive evidence in favour of ANP32 proteins being important for vRNA replication on a cRNA template. Rather, unpublished work suggests that ANP32 proteins may be equally indispensable for cRNA replication on a vRNA template. Further work is required in this area to reach firm conclusions.

103

V NATURAL VARIATION IN ANP32 PROTEINS This chapter investigates natural variation in the genes encoding ANP32A and ANP32B. Because of their importance for influenza virus replication, there might be variants in the human population that hinder this proviral activity and thus bestow a certain level of protection against influenza virus infection on the carrier. Other variations might exacerbate symptoms and lead to severe outcomes of infection in patients. In this chapter, publicly available databases were used to screen for natural variation in ANP32 proteins in the human population. Several missense single nucleotide variants (SNVs) in both genes were found that had effects on FluPol activity when the protein products were co-expressed with IAV polymerase in human eHAP dKO cells4. One SNV was of particular interest as it led to a mutant ANP32B-D130A protein, and this variant was prevalent in a much higher proportion of the population than any other variant in either gene. Moreover, it was the only variant for which homozygous carriers had been described. The homozygous genotype wildtype Anp32A / mutant Anp32B was recapitulated in eHAP cells by CRISPR/Cas9 genome editing, and FluPol activity and IAV replication were attenuated in these cells. To our knowledge this was the first description of a SNV in the coding region of a human gene that may confer some protection against influenza virus. The work described in this chapter has culminated in a manuscript available on the bioRxiv server (Staller et al., 2020b).

5.1 Host genetics of influenza virus disease

Rare cases of life-threatening influenza pneumonitis or encephalitis can result from genetic deficiencies. In such individuals infection with seasonal or pandemic IAV can cause fatal acute respiratory distress syndrome (ARDS) or viral encephalitis without respiratory manifestations (Ciancanelli et al., 2016; Zhang, 2020). Monogenic defects of innate immunity leading to inborn errors of immunity (IEI) have been identified in recent years as the culprits behind severe influenza cases. Autosomal dominant (AD) deficiency of the endothelial transcriptional activator GATA2 leads to a pleiotropic syndromic disorder characterized by complex haematopoietic and immunological defects. This syndrome has resulted in at least four deaths from severe IAV infection (Sologuren et al., 2018). Isolated cases of influenza ARDS have also been reported in patients with autosomal recessive (AR) deficiencies in interferon regulatory factor (IRF) 7 and IRF9 (Ciancanelli et al., 2015). Defects in these proteins interrupt

4 The difference between a single-nucleotide polymorphism (SNP) and a single-nucleotide variant (SNV) is that the formers occurs at a fairly high frequency in the population, usually ≥ 0.005. SNPs are thus relatively common, while SNVs may occur at any frequency.

104 both type I and type III interferon (IFN) signalling, which is indispensable for host defence against influenza virus. Mutations in the IRF9 gene leading to aberrant splicing have been described in children with severe pulmonary influenza infection or increased susceptibility to viral infection (Bravo Garcia-Morato et al., 2019; Hernandez et al., 2018). The endosomal membrane-associated pattern recognition receptor (PRR) Toll-like receptor (TLR) 3 recognizes influenza dsRNA leading to type I and III IFN responses against IAV. TLR3 is essential for viral sensing in fibroblasts and endothelial cells, highlighting the importance of immune sensing in target tissues. Severe influenza pneumonitis in children with TLR3 deficiency has been reported (Lim et al., 2019). Furthermore, variants in the TLR3 gene have been described in patients with IAV-associated encephalopathy (IAE) and pneumonia (Esposito et al., 2012; Hidaka et al., 2006).

Mx1 deficiency in inbred laboratory strains of mice leads to profound susceptibility to influenza virus in a species that is not naturally susceptible to the virus (Staeheli et al., 1988). The human interferon-stimulated gene (ISG) MX1 encodes the dynamin-like GTPase MxA, a restriction factor that binds influenza virus nucleoprotein NP, thus inhibiting the nuclear import of incoming vRNPs and retaining newly synthesised RNPs in the cell nucleus (Ciancanelli et al., 2016; Graf et al., 2018; Haller et al., 2015). Although inherited deficiency in MX1 has not been described in humans, non-synonymous natural variation in the gene has been shown to lead to loss of antiviral activity in some cases. These include amino acid substitutions in the G interface and stalk region of MxA, resulting in defective GTPase activity and impaired oligomerization and viral target recognition, respectively (Graf et al., 2018).

Interferon-induced transmembrane protein 3 (IFITM3) was first described as a restriction factor for influenza virus in RNAi and proteomic screens (Brass et al., 2009; Yount et al., 2010). Upon post-translational modification IFITM3 localises to the endosomal compartment where it blocks the virus fusion pore and thus delivery of vRNPs into the cytosol. A SNP that leads to an N-terminal 21-amino acid truncation, through alteration of the first splice acceptor site, is perhaps the best-known genetic lesion affecting susceptibility to influenza virus. The minority homozygous (CC) allele (SNP rs12252) was over-represented in 53 UK patients that required hospitalisation upon infection with pH1N1 2009 IAV, and in fatal cases among Chinese patients suffering from infection with 2009 pH1N1 and H7N9 viruses (Everitt et al., 2012; Lee et al., 2017). Although subsequent studies with larger sample sizes failed to find an association with rs12252 in their cohorts (Mills et al., 2014; Randolph et al., 2017), a meta-analysis including 12 studies and over 16,000 subjects confirmed the association between the IFITM3 CC minority allele and influenza severity (Prabhu et al., 2018). The homozygous CC allele is rare in Europe but common in otherwise healthy Asian populations (Wellington et al., 2019). Another important SNP associated with severe influenza is located in the 5’ UTR of IFITM3

105 and leads to transcriptional repression through increased CTCF binding to the promoter (Allen et al., 2017).

Complement decay-accelerating factor CD55 (also known as DAF) protects cells from complement attack by blocking C3 and C5 activation. Two SNPs in CD55 have been associated with severe IAV infection in Greek and Chinese cohorts (Chatzopoulou et al., 2019; Lee et al., 2017; Zhou et al., 2012). Mutations in the pulmonary surfactant protein A2 (SFTPA2) leading to single amino acid substitutions have been associated with ARDS after infection with pH1N1 2009 virus in a Spanish population, while increased expression of minor allele variants of the protease TMPRSS2, which activates IAV haemagglutinin, is a risk factor for severe pH1N1 influenza (Cheng et al., 2015; Herrera-Ramos et al., 2014).

Almost all the host genetic variation described in the literature is associated with exacerbating influenza virus infection, rather than alleviating it. This bias is explained by the opportunity to perform whole genome (WGS) or whole exome sequencing (WES) on patients who present at the hospital with severe symptoms. Carriers of variants that may confer protection to influenza virus tend to stay off the radar, but surely such variants must exist. So far, besides a pair of polymorphisms in the non-coding region (NCR) of LGALS1, a gene encoding the soluble PRR galectin-1, no SNVs protective against influenza virus have been described. Galectin-1 is an S-type lectin that recognises galactose-containing oligosaccharides present in cell and viral membranes. Functional variants of the gene may confer some protection against H7N9 influenza in Chinese poultry workers (Chen et al., 2015; Nogales and M, 2019). Perhaps the best-known mutation conferring innate genetic resistance to viral infection is the Δ32 variant of the HIV-1 co-receptor C-C chemokine receptor 5 (CCR5) (Carrington et al., 1999). Single nucleotide variants (SNVs) that reduce susceptibility to virus infection include a P424A substitution in the filovirus endosomal fusion receptor Niemann-PickC1 (NPC1) and a G428A mutation in the fucosyltransferase FUT2 that renders homozygous carriers resistant to norovirus infection (Kondoh et al., 2018; Lindesmith et al., 2003).

5.2 Variation in ANP32 proteins

The question of natural variation in ANP32 proteins was investigated bearing in mind the accrued knowledge about the importance of LRR5, and amino acid residues 129 and 130 in particular, for support of influenza polymerase (Long et al., 2019a; Staller et al., 2019; Zhang et al., 2019a). Publicly available databases were screened for rare missense SNVs in the human Anp32A and Anp32B genes that were predicted, on the basis of our previous work, to have potential functional impact on the pro-influenza virus activity of the proteins. Changes in charge, bulk, polarity and solvent exposure of the amino acid substitutions encoded by the

106

SNVs were considered. We predominantly used the genome aggregation database gnomAD v2 (previously known as ExAC), which holds 125,748 whole exomes and 15,708 genomes from unrelated individuals sequenced in disease-specific or population genetic studies, aligned against the GRCh37/hg19 reference, and v3 which holds an additional largely non- overlapping set of 71,702 genomes aligned against GRCh38 (Karczewski et al., 2019). The gnomAD database identifies 54 missense SNVs in Anp32A and 82 in Anp32B (observed/expected score of 0.39 vs 0.68), suggesting that the former may be more conserved. It is not currently possible to obtain specific phenotypic information about individual carriers of a particular SNV from gnomAD.

The National Center for Biotechnology Information (NCBI) database dbSNP records 136 minor variants in Anp32B and 82 in Anp32A (Sherry et al., 2001). This difference stems from the intricacies of the variant calling process and quality control, which is more stringent at gnomAD (Collins et al., 2019). An additional large database is Trans-omics for precision medicine (TOPMed), which currently holds genomic information from over 149,000 individuals (Taliun et al., 2019). Smaller databases were consulted, including the Grand Opportunity Sequencing Project (GO-ESP), the Avon Longitudinal Study of Parents and Children (ALSPAC) and the 1000 genomes project (Fu et al., 2013; Genomes Project et al., 2015; Walter et al., 2015). The data from different databases were mostly overlapping – TOPMed, for example, is cited as a contributing project on the gnomAD website – but not always. For instance, one SNV in Anp32A was described in the ALSPAC study, but not in the much larger gnomAD database. It was, however, cited in dbSNP.

Individuals in the gnomAD database are sorted by age, gender and ethnicity, with cohorts of African/African American, Latino, Amish, Non-Finnish European, Finnish, South Asian, Ashkenazi Jewish, East Asian, or ‘Other’ descent. Individuals in the ‘other’ category did not unambiguously cluster with major populations in a principal component analysis (PCA). For each variant the total number of mutant alleles, the number of homozygous carriers and the global minor allele frequency (MAF) is given.

Four SNVs in Anp32A and five in Anp32B were selected for investigation (Figure 5.1). All the variants in Anp32A were exceedingly rare (MAF<0.001) and only heterozygous carriers were identified. The SNV responsible for a mutant ANP32A-D130N protein (rs771143708) was found in a single heterozygous individual out of 3,854 in the ALSPAC cohort, rs751239680 (ANP32A-R132Q) was found in a single heterozygous individual out of 10,052 in the African cohort of the gnomAD database, rs772530468 (ANP32A-S158T) was present in a single South Asian male in 23,070 in the gnomAD database, as well as two separate heterozygous individuals in the TOPMed database (one of whom may or may not overlap with the individual

107 in the gnomAD database). Finally, a single African male in 5,032 in the gnomAD database was a heterozygous carrier of rs772530468 (ANP32A-S158A).

Most variants in Anp32B were also low-frequency (MAF<0.0001): rs377406514 (ANP32B- L128V) was found in a single African female out of 10,066 in the gnomAD database; rs771977254 (ANP32B-E133Q) occurred in two out of 4,900 Ashkenazi Jewish females in the gnomAD database; rs761932651 (ANP32B-L138H) was found in one South Asian male in 23,070 in the gnomAD database; rs770020996 (ANP32B-L142F) occurred in one female in 7,544 and one male in 23,072 from the South Asian cohort of the gnomAD database. There was, however, an interesting exception. SNV rs182096718, encoding ANP32B-D130A, was relatively common in the Hispanic / Latino cohort of the gnomAD database v2, where in a total pool of 35,420 alleles 1,209 minority alleles were identified, as well as 25 homozygous carriers.

In the previous chapter the importance of amino acid residue 130 for pro-influenza virus function of ANP32A was described. Murine ANP32A naturally harbours alanine at position 130 and does not support IAV FluPol activity in minigenome assays, and substituting 130A in human ANP32A greatly reduces FluPol activity (Staller et al., 2019; Zhang et al., 2019a). The D130A substitution in ANP32B was present in 3.41% of the Latino cohort, which, compared with all other SNVs in either Anp32A or Anp32B is a relatively high frequency. This variant may therefore be listed as a polymorphism, but only within the Latino cohort. Globally, the minority C allele was found at a frequency of 0.004363 in gnomAD v2, 0.000308 in GoESP, 0.001744 in TOPMed and 0.001797 in 1000 Genomes, with the caveat that almost all carriers are found in a single subpopulation. Attempts to obtain more information about carriers of this SNV from the gnomAD curators at MIT were unsuccessful. Nevertheless, intelligence obtained from collaborators of a colleague, Dr V Sancho-Shimizu, confirmed that three of the homozygous carriers were Mexican Americans residing in California.

108

Figure 5.1 Selected naturally occurring missense SNVs in Anp32A and Anp32B From Staller et al bioRxiv 2020 (A) Amino acid substitutions (variant), NCBI dbSNP Reference IDs, number of minority alleles, global minor allele frequency (MAF), number of homozygous carriers (homs), predominant population cohort in which the variant is found, and the main source database for each variant (B) Structural model of ANP32A showing amino acids affected by selected SNVs (C) Structural model of ANP32B showing amino acids affected by selected SNVs. Structural models were created using iTASSER structural prediction software (based primarily on huANP32B [GenBank accession number 2RR6A] and huANP32A [accession number 2JQDA], and 2JEOA). The three- dimensional structural models were visualized and created in UCSF Chimera, under supervision of Dr CM Sheppard.

5.3 Natural variants of ANP32 proteins affect support of FluPol activity

A method to measure the capacity of mutated ANP32 proteins to act as pro-viral factors for influenza polymerase activity was described in the previous chapter. This is achieved by exogenous expression of the cloned mutants in human eHAP cells lacking ANP32A and ANP32B (dKO), in which influenza polymerase is unable to function in absence of

109 complementing ANP32 (Staller et al. 2019; Peacock et al. 2020). Here we expressed each natural ANP32A or ANP32B variant and tested its ability to support reconstituted polymerases from the 2009 pandemic H1N1 isolate Eng/195 or the seasonal H3N2 virus Vic/75 in minigenome reporter assays (Figure 5.2).

Compared with wildtype ANP32A (WTA), ANP32A-D130N did not support Eng/195 polymerase activity at all, while R132Q and S158A substitutions had significantly reduced capacity to support FluPol activity (Figure 5.2 A). An artificial mutant with the phosphomimic S158E, also had reduced capacity to support Eng/195 polymerase. In contrast, ANP32A- S158T supported FluPol to an extent similar to wildtype ANP32A. Compared with wildtype ANP32B (WTB), ANP32B-D130A had a large deleterious effect on the support for Eng/195 polymerase activity, as did the leucine to valine substitution at position 128 (ANP32B-L128V) (Figure 5.2 B). ANP32B-E133Q was also significantly less able to support Eng/195 FluPol activity. Substitution of the leucines at positions 138 and 142 to histidine (L138H) and phenylalanine (L142F), respectively, did not compromise the ability of the mutant ANP32B proteins to support Eng/195 polymerase activity.

In general, the effects of natural variation in ANP32 proteins was similar for Vic/75 polymerase. ANP32A-D130N did not support Vic/75 polymerase activity, compared with WTA (Figure 5.2 C), but R132Q and S158A substitutions had a smaller effect on Vic/75 FluPol activity than on Eng/195 FluPol activity. The S158T substitution and the phosphomimic S158E were as capable of supporting Vic/75 FluPol activity as WTA. As seen with Eng/195 polymerase, the ANP32B-L128V substitution was unable to rescue Vic/75 polymerase activity, but the D130A mutation had less detrimental effect and resulted in only a <2-fold reduction in Vic/75 polymerase activity (Figure 5.2 D). This is in contrast to Eng/195 polymerase activity complemented with ANP32B-D130A, where a reduction of >17- fold was observed (compare Figures 5.2 B and D). All in all, Eng/195 polymerase, which has circulated in humans since 2009, was more sensitive to alterations in ANP32 than Vic/75 polymerase, which has replicated in human cells for over 40 years. The observed differences in FluPol activity were not explained by differences in expression of the FLAG- tagged ANP32 constructs (western blots accompanying Figures 5.2 A-D), nor by impaired nuclear localisation of the mutant proteins (Figures 5.2 E and F).

110

111

Figure 5.2 ANP32A and B mutant proteins show variable capacity to rescue IAV FluPol activity From Staller et al bioRxiv 2020 (A-D) Minigenome reporter assays in eHAP dKO cells with co- transfected FLAG-tagged ANP32A (A and C) or ANP32B variants (B and D) with either pH1N1 Eng/195 (A and B) or H3N2 Vic/75 (C and D) RNP components PB1, PB2, PA and NP, pPolI-firefly luciferase minigenome reporter, and Renilla luciferase control in a 2:2:1:4:2:2 ratio. Data show mean (SD) of firefly activity normalized to Renilla and analysed by t test from one representative technical triplicate repeat (n = 3 independent biological replicate experiments). Accompanying western blots show expression of the FLAG-tagged ANP32 constructs alongside selected RNP components and endogenous vinculin loading control. 20-60 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol. Lysates were resolved by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/1,000; Abcam), rabbit α-IAV PB2 (GTX125926, 1/2,000; GeneTex), mouse α-NP (catalogue number ab128193, 1/1,000; Abcam), rabbit α-PA (catalogue number GTX118991, 1/500; GeneTex), and mouse α-FLAG (F1804, 1/500; Sigma-Aldrich) and secondary antibodies, HRP – conjugated sheep α-rabbit IgG (AP510P, 1/10,000; Merck) and goat α- mouse IgG (STAR117P, 1/5,000; AbD Serotec), and developed on a Fusion-FX imaging system. (E- F) Immunofluorescence analysis showing nuclear localisation of FLAG-tagged ANP32A (E) or ANP32B (F) variants, detected with anti-FLAG primary antibody and Alexa Fluor-568 anti-mouse conjugate and counterstained with DAPI. ~250,000 eHAP dKO cells were cultured on glass coverslips and transfected with 200 ng pCAGGS-ANP32 expression plasmids 24 hours prior to fixing in 4% paraformaldehyde (PFA) and permeabilisation in 0.2% Triton X-100 detergent. Cells were incubated with primary antibody mouse α-FLAG (F1804, 1/200; Sigma-Aldrich) followed by goat α-mouse IgG Alexa Fluor-568 conjugated secondary antibody (1/200; Life Technologies) in phosphate-buffered saline (PBS) buffer complemented with 1% bovine serum albumin (BSA) and 0.1% Tween 20 (blocking for 30 minutes at 37⁰C; antibody incubations for 1 hour each at 37⁰C, all in humidified chamber). Chromosomal DNA was co-stained with 4’,6-diamidino-2-phenylindole (DAPI, 1/1,000) during incubation with the secondary antibody. Coverslips were mounted on microscopy slides using Vectashield mounting medium. Imaging was carried out by David Gaboriau of the Imperial College Facility for Imaging by Light Microscopy (FILM) on a Zeiss Cell Observer widefield microscope using the DAPI (ex 365/12 nm, em 447/60 nm) and TexasRed (ex 562/40 nm, em 624/40 nm) channels.

5.4 ANP32 position 130 mutants show impaired binding to FluPol

Using the split luciferase complementation assay described in the previous chapter we next assessed whether mutations at amino acid 130 of the human ANP32 paralogues affected interaction with FluPol. To this purpose, carboxy-terminal luc2-tagged versions of ANP32A- D130N and ANP32B-D130A were cloned in pCAGGS expression vectors by overlap PCR. The naturally occurring human ANP32 variants ANP32A-D130N and ANP32B-D130A indeed showed reduced binding to FluPol (Figure 5.3). The luciferase signal indicating binding of

112

ANP32A-D130N to FluPol was reduced almost 3-fold, relative to wildtype ANP32A (P=0.0012) and the signal indicating FluPol interaction of ANP32B-D130A was reduced >2- fold compared with wild type ANP32B (P=0.0008). Western blotting analysis showed the differences in binding affinity were not due to differential expression of the Vic/75 PB1-luc1 and ANP32-luc2 constructs.

Figure 5.3 ANP32 position 130 mutants show impaired binding to FluPol From Staller et al bioRxiv 2020 Split luciferase complementation assays showing reduced interaction between naturally occurring human ANP32 variants ANP32A-D130N (A) and ANP32B-D130A (B) and Vic/75 FluPol. Split luciferase complementation assays were carried by transfecting equal amounts (15 ng) of pCAGGS expression plasmids encoding Vic/75 polymerase components PB1-luc1, PB2 and PA, as well as the indicated ANP32-luc2 construct into 293T cells. 24 hours post-transfection cells were lysed and luminescence was measured. Total amount of plasmid (in ng) in experimental and control conditions was kept constant by using empty pCAGGS. Normalised luminescence ratio (NLR) was obtained as described in Chapter II Materials and Methods, by dividing luminescence in the experimental condition (tagged PB1 + tagged ANP32) by the sum of the luminescence measured in the control conditions (i.e. background interaction of unbound luc1 with ANP32-luc2, and unbound luc2 with PB1-luc1, respectively). Data shown are mean (SD) representative of 3 independent biological repeat experiments, each carried out in technical triplicates; statistical analysis by t-test. (C) Western blot showing expression of luc1-tagged H3N2 Vic/75 PB1 and luc2-tagged ANP32 constructs alongside vinculin loading control.

113

50 µg total protein per well was prepared in 4x Laemmli buffer supplemented with β-mercaptoethanol. Lysates were resolved by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes by semidry transfer. Membranes were incubated with primary antibodies rabbit α-vinculin (ab129002, 1/2000; Abcam) and rabbit α-Gaussia luciferase (PA1-181, 1/1000; ThermoFisher) and secondary antibody, HRP – conjugated sheep α-rabbit IgG (AP510P, 1/10,000; Merck), and developed on a Fusion-FX imaging system.

5.5 ANP32 position 130 mutants exert dominant-negative effects

Since ANP32A and B serve redundant roles in supporting FluPol, individual SNVs in either gene would only have significance only if they exerted dominant negative effects over the wildtype proteins. A heterozygous carrier of an ANP32B SNV will express one copy of wildtype ANP32B as well as wildtype ANP32A (genotype Anp32B +/-; Anp32A +/+, where minus stands for presence of the SNV). A homozygous carrier will express two copies of mutant ANP32B alongside wildtype ANP32A (genotype Anp32B -/-; Anp32A +/+). As functional redundancy in pro-viral function between ANP32A and ANP32B has been established (Staller et al., 2019; Zhang et al., 2019a), the variant ANP32B-D130A will only have an effect if it exerted a dominant-negative effect on the wildtype version of ANP32B (in heterozygous carriers) and paralogue interference (Kaltenegger and Ober, 2015) over wildtype ANP32A (in heterozygous and homozygous carriers).

To investigate the functional consequence of the ANP32B- D130A variant, we recapitulated heterozygous or homozygous ANP32B-D130A variant genotypes by exogenous expression of the mutated proteins in CRISPR-edited eHAP cells that lack ANP32B expression (BKO; described in previous chapters) (Figure 5.4 A). Minigenome assays were performed with reconstituted pH1N1 Eng/195 polymerase complemented by transient transfection with increasing amounts of either wildtype ANP32B (homozygous wildtype genotype; blue bars), ANP32B-D130A (homozygous mutant genotype; red bars) or a 1:1 ratio of both (heterozygous mutant genotype; purple bars). We found that adding increasing amounts of wildtype ANP32B, essentially mimicking the homozygous wildtype genotype (blue bars), had no effect on FluPol activity compared with FluPol supported by ANP32A alone (grey bar). This is presumably due to the presence of wildtype ANP32A in the BKO cells, since ANP32A and ANP32B serve redundant roles in supporting FluPol. In contrast, polymerase activity decreased significantly when ANP32B-D130A rather than wildtype ANP32B was co- transfected, and this effect was more dramatic the more variant was expressed (red bars). This suggests paralogue interference of ANP32B-D130A over wildtype ANP32A, which would otherwise support polymerase activity effectively (grey bar). Co-transfecting

114 increasing amounts of wildtype and mutant ANP32B in a 1:1 ratio, recapitulating the heterozygous mutant genotype (purple bars), also resulted in a significant drop in polymerase activity. This suggests that ANP32B-D130A also exerts a dominant-negative effect over wildtype ANP32B. These observations were unrelated to expression levels of the FLAG-tagged ANP32 constructs (accompanying western blots). Statistical significance of the differences in Eng/195 polymerase activity between conditions is shown in the accompanying Table.

We next investigated whether similar dominant-negative effects were exerted by the rare ANP32A-D130N variant (Figure 5.4 B). In eHAP cells lacking ANP32A expression (AKO) pH1N1 Eng/195 polymerase activity increased significantly when wildtype ANP32A was provided by transient co-transfection (blue bars), compared with FluPol activity supported by ANP32B alone (grey bar). In contrast, decreased polymerase activity was evident when either ANP32A-D130N alone (red bars) or wildtype and mutant ANP32A in a 1:1 ratio (purple bars) were co-transfected. This suggests that, like ANP32B-D130A, ANP32A-D130N is dominant-negative over wildtype ANP32A, and exerts paralogue interference over wildtype ANP32B. Together these data show that different substitutions of the canonical aspartate at position 130 of both pro-influenza viral human ANP32 proteins can have similar phenotypic effects and impact influenza proviral activity even in the heterozygote genotypes.

115

Figure 5.4 ANP32B-D130A exerts dominant-negative effects on pro-viral function of wildtype ANP32B and ANP32A From Staller et al bioRxiv 2020 (A) Minigenome reporter assay in eHAP cells

116 lacking ANP32B expression (BKO) with 2009 pH1N1 Eng/195 RNP components PB1, PB2, PA and NP, pPolI-firefly luciferase minigenome reporter, and Renilla luciferase control, in a 2:2:1:4:2:2 transfection ratio with 20 ng pCAGGS-PB1 and 10 ng pCAGGS-PA per well (~1.0 x 105 cells). Complementation by co-transfection of indicated combinations of wildtype ANP32B, ANP32B-D130A, or empty pCAGGS plasmid (in ng) as shown. Total amounts of transfected plasmid DNA were kept constant across conditions by supplementing with empty pCAGGS plasmid. Data shown are mean (SD) of firefly activity normalized to Renilla from one representative biological repeat (n = 3 independent triplicate experiments); statistical analysis in the accompanying table was carried out by t test. Accompanying western blot shows expression of the FLAG-tagged ANP32 constructs alongside RNP components PB2 and NP, and vinculin loading control. Western blots were carried out as before. Cartoon figures show the wildtype and mutant genotypes found in carriers of SNV rs182096718. All three genotypes have two alleles of wildtype Anp32A (yellow). Homozygous ANP32B wildtype individuals carry two wildtype Anp32B alleles (blue); homozygous mutant carriers have two alleles encoding ANP32B-D130A (red), and finally heterozygous mutant carriers have one wildtype and one variant Anp32B allele. (B) Minigenome reporter assay in eHAP cells lacking ANP32A expression (AKO) with 2009 pH1N1 Eng/195 RNP components PB1, PB2, PA and NP, pPolI-firefly luciferase minigenome reporter, and Renilla luciferase control, in a 2:2:1:4:2:2 transfection ratio. Co-transfected wildtype ANP32A, ANP32A-D130N and Empty pCAGGS plasmid are indicated in the Table, as in (A). Accompanying western blot shows expression of the FLAG-tagged ANP32 constructs alongside RNP component PB2 and a vinculin loading control. Western blots were carried out as before. Cartoon figures show the wildtype and mutant genotypes found in carriers of SNV rs771143708. All three genotypes have two alleles of wildtype Anp32B (red). Homozygous ANP32A wildtype individuals carry two wildtype Anp32A alleles (yellow); hypothetical homozygous mutant carriers have two alleles encoding ANP32A-D130N (pink), and finally heterozygous mutant carriers have one wildtype and one variant Anp32A allele.

5.6 Generating CRISPR-edited cells expressing wildtype ANP32A and mutant ANP32B- D130A

In order to further probe the potential significance of the relatively common ANP32B-D130A variant CRISPR/Cas9 genome editing was used to generate eHAP cells recapitulating the homozygous genotype wildtype Anp32A / mutant Anp32B. Two appropriate protospacer- adjacent motifs (PAM; AGG and TGG) were selected in close proximity to the exon 4 target nucleotide, on the opposite DNA strand. Wildtype SpCas9 cleavage at these sites would lead to double stranded breaks (DSB) in the DNA, either two or seven nucleotides upstream of the target adenine. Guide RNAs were designed against the corresponding protospacers and cloned into plasmid mCherry_gRNA (Addgene), which expresses the red fluorescent protein mCherry for selection purposes. Selection by fluorescence rather than antibiotic was chosen

117 because this would be less stressful for the cells after having undergone transfection by electroporation. Wildtype SpCas9 (a DNA endonuclease with two distinct endonuclease domains, leading to blunt DSBs in the target DNA) was expressed from the plasmid PX459 (Addgene). In an alternative experiment , a high-fidelity version of SpCas9, carrying mutations that lower its affinity for DNA, meant to reduce off-target effects and improve specificity (Kleinstiver et al., 2016), was used. SpCas9-HF1 was expressed from Addgene plasmid PX330.

Precise editing requires a repair template which serves to mimic the sister chromatid that is used by the cell in homology-directed DNA repair (as explained in Chapter III). These repair templates were ordered from IDT as 88-base single-stranded DNA molecules (ssODN), with a cytosine in place of adenine at the target site. The length was required because it is the extensive homology arms (40-50 bases) on either side of the edit that allow the ssODN to act as a repair template. In addition to the A>C mutation a silent PAM mutation was introduced in the ssODN (also known as Cas9-blocking mutation) to avoid re-cleavage of the target site upon successful repair (Okamoto et al., 2019). Four separate CRISPR conditions were set up, namely wildtype SpCas9 or SpCas9-HF1 with either gRNA1 or gRNA2 and their respective ssODNs. Serendipitously, a successful edit would introduce a BsmI restriction site (GAATGC) in the target DNA, where the original locus reads GAATGA. No additional BsmI restriction sites were found in the vicinity. This suggested a straightforward means of screening CRISPR clones, namely by PCR amplification of the locus followed by digestion with BsmI. The amplicon should be cleaved only in successfully edited clones. In order to test this strategy, the relevant locus was amplified from wildtype eHAP genomic DNA (without BsmI restriction site), and from the ANP32B-D130A-luc2 plasmid construct (which has the BsmI restriction site). The gel-extracted amplicons were then incubated for 30 minutes at 65⁰C in the presence of BsmI restriction enzyme (Figure 5.5). DNA fragments were amplified from the genomic DNA and from the plasmid. Incubation with BsmI led to cleavage of the plasmid-derived amplicon into fragments of expected size, while the wildtype genomic DNA remained uncleaved. To control for amplification of the correct locus, amplicons were digested in parallel by restriction enzyme BbsI, yielding products of expected size.

118

Figure 5.5 Strategy to generate and screen CRISPR-edited ANP32B-D130A cells (A) Schematic showing the CRISPR editing strategy. Two sets of gRNAs and ssODNs were designed, but for clarity only one is depicted here. Anp32B is located on Chromosome 9, with the codon encoding 130D in exon 4 of the gene. The red arrow indicates the reading direction. A protospacer-adjacent motif (PAM) was identified near the target adenine on the opposite DNA strand, and a guide RNA was designed against the protospacer (underlined). SpCas9 (either wildtype or HF1) cut two base pairs upstream of the target nucleotide (dark blue arrowheads). An 88-base single-stranded DNA (ssODN) homology-directed repair (HDR) template was designed bearing the missense A>C substitution as well as a silent Cas9-blocking mutation between homology arms. A successful edit introduced a BsmI restriction site (GAATGC) not present in the wildtype allele, enabling screening by PCR amplification of a 1,561 base pair genome fragment followed by restriction digestion of the PCR product. The PCR amplicon derived from the plasmid is not of the same size because a different primer pair was used (B-C) The screening strategy was tested through PCR amplification of either genomic DNA (gDNA) from wildtype eHAP cells (which don’t have the edit) and a pCAGGS expression plasmid encoding luc2-tagged human ANP32B-D130A (used because it is larger and the cut site is skewed away from the centre of the amplicon). The amplicons were then digested with BsmI resulting in cleavage of the plasmid-derived amplicon bearing 130A, but not the gDNA-derived amplicon.

For transfection of pairs of plasmids (mCherry_gRNA and SpCas9) in combination with the ssODNs, electroporation rather than lipofection is recommended (Ran et al., 2013b). First, the electroporation protocol on a NeonTM Transfection System (Invitrogen) was optimised for eHAP cells. A single 1,200 V pulse for 40 ms gave a transfection efficiency of approximately 60% (as measured by mCherry expression 24 hours post-transfection) and 100% cell viability. Transfection mixes were prepared and optimised settings were used for electroporation. Cells were allowed to recover in growth medium without antibiotics for 48 hours, after which they

119 were single-cell sorted into 96-well plates by fluorescence-activated cell sorting (FACS), exploiting mCherry emission at 561-610 nm.

31 clones in the SpCas9-HF / gRNA1 condition were screened, as well as 33 clones in the SpCas9-HF / gRNA2 condition, 41 clones in the wildtype SpCas9 / gRNA1 condition and 29 clones in the wildtype SpCas9 / gRNA2 condition. The latter condition was the only one that yielded positive results (2 clones), suggesting no advantage of SpCas9-HF1 in this set-up and that gRNA1 did not function as well as gRNA2. The two potential positive clones, as well as three negative ones, were Sanger sequenced to confirm genotypes, the latter to function as negative control lines. One positive clone had the expected A>C edit at the target nucleotide, resulting in the desired D130A substitution (Figure 5.6). Interestingly, analysing the sequencing chromatograms of the putative negative control clones, one was homozygous for an in-frame nine-base pair deletion that translated to an ANP32B protein lacking the functionally important triad 128L-129N-130D (ANP32BΔ128-130). Western blotting analysis confirmed expression of ANP32B-D130A and ANP32B Δ128-130, as well as unaltered ANP32A in each of the clones.

Figure 5.6 Screening and validation of CRISPR clones (A) Screening of eleven CRISPR clones by PCR amplification followed by BsmI restriction digestion shows a positive clone in lane 11. The digestion is incomplete because incubation was short (30 minutes). (B) Sanger sequencing of the clone in lane 11 in (A) revealed the D130A substitution and western blotting confirmed expression of ANP32A and ANP32B-D130A in the mutant clone and in an unsuccessfully edited control clone. (C) A

120 homozygous in-frame 9-base pair deletion was found in one of the uncleaved negative clones, leading to a protein lacking amino acid residues 128-130 (ANP32B Δ128-130). Accompanying western blot shows the mutant protein is expressed at levels similar to the wildtype (the blots in (B) and (C) derive from the same experiment), and ANP32A expression is not compromised. Western blots were carried out as described.

5.7 IAV polymerase activity and replication are attenuated in ANP32B-D130A mutant cells

Using minigenome reporter assays, we found that reconstituted IAV polymerases from Eng/195 or Vic/75 were significantly less active in edited ANP32B-D130A and Δ128-130 cells, compared with control cells (Figure 5.7). As described in Chapter IV polymerase activity was not attenuated in BKO cells, suggesting paralogue interference of mutant ANP32B over wildtype ANP32A. In other words, IAV FluPol activity was higher in cells completely lacking ANP32B expression than in cells expressing the D130A mutant. No polymerase activity was seen in cells lacking both ANP32A and ANP32B (dKO), nor in the negative control conditions where the polymerase subunit PB2 was lacking (2P). Western blotting analysis showed the observed effects were not due to differential expression of transfected RNP components. A separate western blot showed expression profiles of ANP32A and ANP32B in each cell line.

121

Figure 5.7 FluPol activity is attenuated in ANP32B-D130A mutant cells From Staller et al bioRxiv 2020 Minigenome reporter assays in CRISPR-edited cell lines with pH1N1 Eng/195 (A) or H3N2 Vic/75 (B) RNP components PB1, PB2, PA and NP, pPolI-firefly luciferase minigenome reporter, and Renilla luciferase control, in a 2:2:1:4:2:2 ratio. Data show mean (SD) of firefly activity normalized to Renilla and analysed by t-test from one representative technical triplicate repeat (n = 3 independent biological repeat experiments). Accompanying western blots show expression of RNP components PB1, PA and NP alongside vinculin loading control. 3P, FluPol subunits PB1, PB2 and PA; 2P, negative control lacking FluPol subunit PB2. (C) Western blotting analysis showing expression of endogenous ANP32A and ANP32B, alongside vinculin loading control, in CRISPR-edited monoclonal control, ANP32B-D130A and ANP32B-Δ128-130 cell lines, as well as the previously established ANP32B knockout (BKO) and ANP32A and B double knockout (dKO) cell lines. Western blotting was carried out as described.

Next we infected mutant (ANP32B-D130A), ANP32B knockout, or control cells at low-MOI (0.005) with 6:2 recombinant influenza A viruses containing internal genes from pH1N1 Eng/195 and H3N2 Vic/75, with the neuraminidase (NA) and haemagglutinin (HA) external genes of the laboratory-adapted H1N1 PR/8 virus. Eng/195 6:2 virus replication was severely attenuated in mutant D130A cells, in comparison with either control or BKO cells (Figure 5.8). Plaques were below the level of detection (LOD) 16 and 24 hours post- infection and still 3 logs lower than control cells 48 hours after infection (P=0.0012). Vic/75 6:2 infectious virus production in mutant cells was also lower than in control cells but less affected than Eng/195. Both viruses replicated to higher titres in cells that completely lacked ANP32B (BKO) than in cells expressing the ANP32B-D130A mutant, suggesting paralogue interference of ANP32B-D130A over wildtype ANP32A also in an infectious virus context.

Figure 5.8 IAV replication is attenuated in ANP32B-D130A mutant cells From Staller et al bioRxiv 2020 CRISPR-edited control (wildtype; black), mutant ANP32B-D130A (D130A; red), and ANP32B knockout (blue) monoclonal cell lines were infected with either England/195 (A) or Victoria/75 (B) 6:2

122 reassortant viruses with PR8 hemagglutinin (HA) and neuraminidase (NA) external genes at MOI = 0.005, and incubated at 37⁰C in the presence of 1 µg/ml trypsin to allow multicycle replication. Supernatants were harvested at the indicated times post-infection and pfu/ml established by plaque assay on MDCK cells. LOD (dotted line) denotes the limit of detection in the plaque assays. Shown are representative data from one of two independent infection experiments carried out in triplicate. P values were calculated per time point by t test and represent differences between control and mutant cell lines. Plaque assays were carried out by Miss L Baillon.

5.8 Discussion

Here, we used publicly available databases to perform a biased screen for naturally occurring single nucleotide variants in human Anp32A and Anp32B, under the hypothesis that some mutations may affect susceptibility to influenza virus infection. One of the mutations, which translates to an aberrant ANP32B-D130A protein, was highly enriched in carriers of Latino descent. Although most carriers of this SNV (>1,200) are heterozygous for the mutation, 25 homozygotes have been reported so far in the gnomAD database. Almost all carriers occur in the Hispanic/Latino subpopulation. We generated this naturally occurring homozygous genotype by CRISPR/Cas9 genome editing in low-ploidy human eHAP cells and found that IAV FluPol activity and virus replication were significantly attenuated in the mutant cell line, compared to wildtype control cells or cells lacking ANP32B. This effect was most pronounced with the 2009 pandemic H1N1 strain Eng/195. Importantly, we demonstrate that the mutant ANP32B-D130A protein exerts a dominant-negative effect over wildtype ANP32B, as well as interference over the functionally redundant paralogue ANP32A. This suggests that despite the redundancy in pro-influenza viral function of ANP32A and B, even in heterozygotes there may be some consequence of this genetic variation to outcome of virus infection. We provide the first example of a single nucleotide variant in the coding region of a human gene that might confer some protection against influenza virus infection.

Although we and others have previously reported redundancy in reconstituted systems for ANP32A and B in their support of FluPol activity (Staller et al., 2019; Zhang et al., 2019a), the data presented here imply that deleterious mutations in one ANP32 protein might exert dominant-negative effects on the pro-viral activity of the other, a phenomenon known as paralogue interference. Poor interaction with FluPol of one paralogue might be expected to make no difference when both paralogues are present in the cell, but clearly this does not seem to be the case.

123

Recent structural data suggest that ANP32 proteins assist influenza polymerase by stabilizing a dimer between the replicating FluPol trimer, FluPolR, and a second trimer that might be recruited as an encapsidating complex, FluPolE (Carrique et al. 2020, in press) (Figure 5.9). To explain the dominant-negative effects we observe here, as well as the compromised replicative activity, we propose a model (Figure 5.10) in which the mutant ANP32B-D130A binds competitively to FluPolR, via the unaltered N-terminal portion of the LRR domain, but is unable to recruit FluPolE due to the lack of acidity at position 130. Consequently, nascent viral RNA generated by FluPolR is not transferred effectively to FluPolE, and replication is compromised. Furthermore, recruitment of ANP32B-D130A to FluPolR occupies the binding site for wildtype ANP32A or ANP32B proteins, explaining the paralogue interference and dominant-negative effects we observe, respectively.

Figure 5.9 Cryo-EM structure of a dimer of ICV FluPol heterotrimers in complex with ANP32A (from Carrique et al 2020, in press) The ANP32A LRR domain in shown bridging a dimer of influenza C virus FluPol heterotrimers. The ‘replicating’ FluPol (FluPolR; blue), bound to the viral promoter, is proposed to replicate the RNA and transfer the nascent RNA to the ‘encapsidating’ FluPol (FluPolE; green) in order to avoid exposing the RNA. ANP32A facilitates this process by stabilising the dimer. Cryo-EM density map has been deposited in the Electron Microscopy Data Bank (EMDB) with accession code 10665, and in the Protein Data Bank (PDB) with accession code 6XZQ

124

Figure 5.10 Mutation at ANP32 position 130 results in impaired formation of replication- competent FluPol dimer From Staller et al bioRxiv 2020 Cartoon explaining why FluPol activity is impaired in cells harbouring ANP32B-D130A. Wildtype (130D) ANP32B (or ANP32A) acts by binding to a viral promoter-bound replicating FluPol (FluPolR, in blue) via the N-terminal region of the LRR domain, while the C-terminal portion of the LRR, harbouring the surface-exposed acidic residue 130D, is used to recruit the encapsidating FluPol (FluPolE, in green), via direct interaction with basic residues on the PB1 subunit of FluPolE (A). In the context of a properly formed FluPolR-ANP32-FluPolE complex the nascent RNA product replicated by FluPolR (in pink) transfers to FluPolE and replication proceeds efficiently. The mutant ANP32B-D130A can still bind efficiently – and therefore competitively – to FluPolR, as the N-terminal portion of the LRR is unaffected (B), but because there is now an alanine in place of the acidic aspartate at position 130, recruitment of FluPolE is impaired. Furthermore, binding of ANP32B-D130A to FluPolR prevents recruitment of wildtype ANP32A or ANP32B, explaining the paralogue interference and dominant-negative effects, respectively. Thus, in the presence of ANP32B-D130A, the replication-competent FluPolR-ANP32-FluPolE complex is not formed as efficiently, and therefore replication is compromised.

A caveat of the CRISPR/Cas9 editing approach employed here is the use of haploid eHAP cells, where a more relevant diploid cell type may have been preferable. In particular,

125 heterozygous genotypes cannot easily be recapitulated in a haploid cell line, even if in practice eHAP cells often revert to diploidy. Certainly, in future the experiments here described should be repeated in diploid human cells pertinent to influenza virus replication, such as normal human bronchial epithelium (NHBC) or human airway epithelium (HAE) cells that are grown at air-liquid interface (ALI). Another approach would be using induced pluripotent stem (iPS) cells derived from (patient) blood monocytes. CRISPR editing to introduce SNVs can then be performed in the iPS stage, and the cells can be grown out into lung epithelium cells. Novel CRISPR platforms like prime editing have been described recently that may make simultaneous generation of a panel of SNVs in primary cells more straightforward than using standard HDR CRISPR or base editing methods (Anzalone et al., 2019). The prime editing method introduces edits without the need for DSBs and repair templates, potentially increasing efficiency while reducing off-target effects and unwanted indel mutations at the target locus.

We can only speculate as to why the ANP32B-D130A variant is enriched in the Latino cohort compared with other populations. In general, enrichment of a trait can be due to founder effects or selective sweeps. Such a selective sweep may have occurred in indigenous Latino populations, possibly under pressure of a pathogen. In this vein, it would be interesting to test the 1918 pandemic influenza virus in the mutant cell line, as well as pathogens that are known to have decimated indigenous populations in the deeper past. Starting in the 15th century, waves of respiratory infections introduced by European settlers, including smallpox, influenza, tuberculosis and measles, decimated indigenous populations of the New World. Smallpox alone has been estimated to kill off over 90% of indigenous peoples. It is tempting to speculate that genetic relics of what has been called The Great Dying might persist to this day, and that the ANP32B-D130A substitution might be one of those relics.

126

VI MUTATIONAL ANALYSIS OF ANP32 PROTEINS

This chapter describes attempts to map pro-influenza virus function of ANP32 proteins to structural elements and domains. Particular interest lay in assigning function to the unstructured LCAR, a question that was approached by generating a series of truncations in ANP32A and ANP32B, and comparing them with chicken ANP32A truncations already described (Long et al., 2019a; Mistry et al., 2020). Where the focus in those manuscripts lay predominantly with chicken and human ANP32A, a preference for ANP32B by at least some FluPol constellations had been established, as described in previous chapters of this thesis. ANP32B being the more potent host factor in humans (Peacock et al., 2020a; Peacock et al., 2020b), the focus was predominantly on this protein.

Another aim was to probe whether an endogenous function of ANP32 proteins – all human paralogues are involved in histone binding / transcriptional regulation – was related to or distinct from its pro-influenza virus activity. Uncoupling cellular and pro-viral function is paramount to informing potential future design of small molecule inhibitors against ANP32 proteins. Host-directed therapy (HDT) targets host-encoded rather than virus-encoded factors and as such aims to sidestep rapid accumulation of resistance mutations on the premise that viruses cannot directly control the physical appearance of host proteins. The strategy is to interfere with host cell factors that are required by a pathogen for replication or persistence (reviewed in (Kaufmann et al., 2018)). Ideal host-directed antiviral agents should not disrupt vital cellular functions.

RNA virus-encoded factors are hard to target therapeutically because of the low fidelity and absence of proof-reading mechanisms of virus-encoded RNA-dependent RNA polymerases (RdRp), which leads to frequent mutation. Combined with the high number of progeny genome copies in any infected cell, traditional drug-resistant mutants evolve very rapidly, as discussed in the Introduction of this thesis. As an illustration, upon minimal passaging of an influenza A virus in human cells, a study from our laboratory showed rapid evolution of resistance to the nucleoside analogue favipiravir, which targets the virus-encoded RdRp (Goldhill et al., 2018). On the other hand, a drawback of HDT may be the potential evolution of more virulent virus strains that no longer require the host factor in question, as illustrated by the evolution of a pH1N1 escape mutant in mice that no longer required importin-α7 (Resa-Infante et al., 2015).

127

Another reason to understand which motifs of ANP32 can be targeted without affecting the usual host-related function is that the design of influenza virus-resistant gene-edited chickens is currently ongoing, initially targeting the 129N-130D dyad of chicken ANP32A. Such an approach would only work if the edited gene does not compromise embryonic development and leads to an otherwise healthy animal. From mouse studies it is known that ablation of Anp32B leads to fully penetrant lethality or severe developmental issues, as discussed (Reilly et al., 2011). On the other hand, studies in pigs have shown how a host gene editing approach can work against a viral pathogen. Deletion of exon 7 of the scavenger receptor CD163 rendered pig macrophages, as well as whole animals, fully resistant to Porcine Reproductive and Respiratory Syndrome Virus (PRRSV) (Burkard et al., 2017; Burkard et al., 2018).

It is assumed that the pro-viral function of ANP32 proteins is exerted in the cell nucleus, as this is the site of influenza replication, but this has not strictly been proven. Even though we and others have demonstrated nuclear localisation of ANP32 proteins by immunofluorescence microscopy (Long et al., 2019a; Staller et al., 2019), and nuclear FluPol binding by bimolecular fluorescence complementation (BiFC) (Mistry et al., 2020), this does not strictly prove nuclear pro-viral activity, merely nuclear presence of ANP32 proteins. In an attempt to determine the site of pro-viral function, a panel of nuclear localisation signal (NLS) mutants was generated in ANP32B.

Finally, potential pro-influenza virus activity of the ANP32E paralogue was investigated. As ANP32E is expressed in the double ANP32A/B knockout cells described in previous chapters, and viral polymerases are not active in those cells, it was initially concluded that ANP32E can be of little import to influenza virus. Here, we investigate why this is the case.

6.2 Generation of ANP32B LCAR truncations

Published work from our laboratory and others has demonstrated the functional significance of the LCAR and central regions of chicken and human ANP32A (Baker et al., 2018; Domingues et al., 2019; Domingues and Hale, 2017; Long et al., 2016; Long et al., 2019a; Mistry et al., 2020; Zhang et al., 2019a). Very little work has been done on human ANP32B however, although the work described in this thesis and elsewhere suggests ANP32B is the more potent host factor in humans (Peacock et al., 2020a; Peacock et al., 2020b; Staller et al., 2019). The primary structures of human and chicken ANP32A and human ANP32B are highly similar, with an N-terminal leucine-rich repeat region (LRR) comprising amino acid residues 1-149, a 27-amino acid central region that is duplicated in avian ANP32A, and an unstructured C-terminal low complexity acidic region (LCAR) of about 75 mostly acidic amino acid residues (Figure 6.1). In addition to the 27 amino-acid duplication, avian ANP32A has

128 six unique amino acids immediately preceding the duplication, namely 176-VLSLVK-181. This unique motif is believed to contain a SUMO interacting motif (SIM)-like sequence, as discussed in the Introduction of this thesis (Domingues and Hale, 2017).

Figure 6.1 ANP32 protein domains and motifs Human ANP32B (top), human ANP32A (middle) and chicken ANP32A (bottom) share an N-terminal LRR, a 27-amino acid central region that is duplicated in avian ANP32A, and a C-terminal LCAR. A canonical basic nuclear localisation signal (indicated in human ANP32B) is found near the carboxy terminus. Pro-influenza virus residues 128-130 (128-130 loop) are located in LRR5. The inhibitor of histone acetyl transferase (INHAT) domains depicted in human ANP32A will be explained later. Chicken ANP32A harbours a unique 176-VLSLVK-181 SUMO interaction motif (SIM)-like sequence, which is discussed in the Introduction.

Three ANP32B constructs lacking increasing portions of the LCAR and central domains were generated: huB-175 lacked the entire LCAR, huB-168 and 155 lacked the LCAR and progressive portions of the central region (Figure 6.2 constructs 9-11). Strictly speaking, these constructs were not truncations but rather large deletions, as the final 19 amino acid residues at the C-terminus that contained the NLS were retained (denoted ‘Cter’ in Figure 6.1).

129

Figure 6.2 ANP32 constructs primary structures Full primary structure of all ANP32 constructs described in this chapter, with the LRR domain (residues 1-148) in blue, the central region (149-175) in brown, the avian ANP32A 33-amino acid insertion (176-208) in yellow, and the LCAR domain in green. All constructs are FLAG-tagged at the carboxy terminus (purple). The proviral leucine 128 – asparagine 129 – aspartate 130 in bold and underlined in the full-length (FL) wildtype chicken ANP32A, human ANP32A and human ANP32B. Functional and structural details of each construct will be explained throughout the main text.

The ability of the ANP32B LCAR deletion constructs to support Vic/75 FluPol activity in dKO cells was tested (Figure 6.3). FluPol retained some activity compared with full-length wildtype ANP32B when huB ΔLCAR was co-expressed, but activity was not supported by the huB 168 and 155 constructs, lacking increasing parts of the central region as well as the LCAR. These phenotypes were not due to altered expression, as analysed by immunoblotting analysis. These results implied that the LCAR, and the central region of human ANP32B in particular, were required for pro-influenza virus function, in contrast to what had been shown for chicken

130

ANP32A, which retained its proviral activity without the LCAR (residues 1-208; Figure 6.2 construct 2) (Domingues and Hale, 2017; Mistry et al., 2020).

Figure 6.3 Human ANP32B LCAR and central region are required for pro-viral function Minigenome reporter assays in eHAP dKO cells with co-expressed wildtype ANP32B (huB), huB ΔLCAR, huB 168 or huB 155 deletant pCAGGS expression plasmids (100 ng) in support of H3N2 Vic/75 FluPol. RNP components PB1, PB2, PA and NP, a pPolI firefly luciferase minigenome and Renilla luciferase control were transfected in 2 x 105 cells in a 2:2:1:4:2:2 ratio (40 ng PB1) and bioluminescence measured 24 hours post-transfection. Data shown are mean (SD) firefly activity normalized to Renilla and analysed by one-way ANOVA from one representative technical triplicate repeat (n=3 independent biological repeat experiments) ****, P<0.0001. Accompanying western blots show expression of the ANP32B constructs and Vic/75 PB2 alongside Vinculin loading control. Western blotting was carried out as described.

The apparent difference in requirement for LCAR for proviral activity between human and chicken ANP32 proteins was probed further: H5N1 50-92 (PB2-627K or 627E) FluPol activity was compared in dKO cells complemented with either full-length or truncated chicken ANP32A (chA; Figure 6.2 constructs 1 and 2), or human ANP32B (huB; constructs 8 and 10) (Figure 6.4). It was confirmed that whereas chicken ANP32A lacking its LCAR (chA-208) could still support FluPol activity, huB 168 lacking the LCAR and C-terminal part of the central region could not. Surprisingly, provision of the truncated version of chicken ANP32A led to higher

131

PB2-627K FluPol activity that the full-length wildtype version. These differences could not be explained by differences in expression of the constructs, as obtained by western blotting analysis, or nuclear localisation of the new huB-168 construct analysed by immunofluorescence microscopy. The FLAG-tagged chicken ANP32 constructs as well as wildtype ANP32B were already known to localise to the nucleus (Long et al., 2019a; Mistry et al., 2020; Staller et al., 2019).

Figure 6.4 The LCAR domain of human ANP32B is required for FluPol activity, unlike the LCAR of chicken ANP32A (A) Minigenome reporter assays in eHAP dKO cells with co-expressed wildtype or LCAR truncated ANP32B (huB and huB-168, respectively; 50 ng) or chicken ANP32A (chA and chA- 208; 50 ng) in support of reconstituted H5N1 50-92 FluPol with either avian signature (PB2-627E) or human-adapted PB2-627K. Expression plasmids encoding RNP components PB1, PB2 (either 627K (Kpol) or 627E (Epol)), PA and NP, a pPolI firefly luciferase minigenome and Renilla luciferase control were transfected in 1 x 105 cells (48-well plates), in a ratio 2:2:1:4:2:2 (20 ng PB1). Data shown are mean (SD) firefly activity normalized to Renilla and analysed by one-way ANOVA from one representative technical triplicate repeat (n=3 independent biological repeat experiments). ns, not significant; ****, P<0.0001. Accompanying western blots show expression of the FLAG-tagged ANP32 constructs and 50-92 PB2 alongside Vinculin loading control. Western blotting analysis was carried out as described (B) Immunofluorescence microscopy analysis showing the new FLAG-tagged construct, huB-168, localised to the cell nucleus. ~250,000 eHAP dKO cells were cultured on glass coverslips and transfected with 200 ng FLAG-tagged pCAGGS-ANP32B-168 plasmid 24 hours prior to fixing in 4% paraformaldehyde (PFA) and permeabilisation in 0.2% Triton X-100 detergent. Cells were incubated with primary antibody mouse α-FLAG (F1804, 1/200; Sigma-Aldrich) followed by goat α-mouse Alexa Fluor-568 conjugated secondary antibody (1/200; Life Technologies) in phosphate-buffered saline (PBS) buffer complemented with 1% bovine serum albumin (BSA) and 0.1% Tween 20 (blocking for 30 minutes at 37⁰C; antibody incubations for 1 hour each at 37⁰C in a humidified chamber). Chromosomal DNA was stained at 37⁰C with 4’,6-diamidino-2-phenylindole (DAPI, 1/1,000) during incubation with the

132 secondary antibody. Coverslips were mounted on microscopy slides using Vectashield mounting medium. Imaging was carried out by David Gaboriau of the Imperial College Facility for Imaging by Light Microscopy (FILM) on a Zeiss Cell Observer widefield microscope using the DAPI (ex 365/12 nm, em 447/60 nm) and TexasRed (ex 562/40 nm, em 624/40 nm) channels.

6.3 Uncoupling pro-viral and cellular function Both ANP32A and ANP32B are involved in cellular transcription regulation and chromatin architecture through direct association with core histones H3 and H4 (reviewed in (Reilly et al., 2014)). These interactions are generally believed to be mediated primarily by the acidic LCAR domain, although it has also been suggested that ANP32B binds histones with the concave face of the LRR domain (Tochio et al., 2010). As a member of the inhibitor of histone acetyl transferase (INHAT) complex, ANP32A blocks histone H3 tail acetylation at lysines 9, 14, 18 and 23, by physically associating with it (Fan et al., 2006; Schneider et al., 2004; Seo et al., 2002; Seo et al., 2001). ANP32A also blocks phosphorylation of serines in the histone H3 tail (Fan et al., 2006). ANP32B is not part of the INHAT complex, but it is believed to inhibit histone acetylation through similar mechanisms (Munemasa et al., 2008). It has been proposed that the ANP32A domain responsible for HAT inhibition comprises amino acid residues 151-180 (Seo et al., 2002), thus mapping largely to the central domain and into the N-terminal portion of the LCAR domain. This is indicated as INHAT domain I in Figure 6.1. Rather confusingly, the same publication describes INHAT domain II (residues 191-249, comprising most of the LCAR), which paradoxically has no effect on INHAT activity of ANP32A.

C-terminal FLAG-tagged, as well as luc2-tagged, ANP32A constructs were generated lacking either INHAT domain II only (huA-189; construct 5 in Figure 6.2) or INHAT domains I and II (huA-148; construct 6), recapitulating published constructs (Seo et al., 2002), but including the C-terminal 18 residues that contain the NLS. Furthermore, an ANP32B quintuple loss-of- acidity mutant (ANP32B-LOA; construct 12 in Figure 6.2, with alanines highlighted in red) described in the literature were generated (Tochio et al., 2010), with either a FLAG or luc2 C- terminal tag. This ANP32B variant, ordered as a synthetic gene string, had five alanine residues in the LRR domain substituted at positions D25, E45, E70, E73 and D119, and was shown to lose its binding capacity to core histone H3-H4 tetramers, as measured by isothermal titration calorimetry (ITC). The published LOA mutant consisted of the N-terminal resolved portion of ANP32B alone (amino acid residues 1-161), while the construct generated in this study was the full-length protein. Lastly, a core histone 3.1 construct fused at its C-terminus to the luc1 fragment was cloned, in order to analyse binding to the ANP32 constructs. Histone H4-luc1 had been generated previously (Long et al., 2019a).

133

First, using the split luciferase complementation assay, interaction of core histones H3 and H4 was tested with some of the constructs, namely huA-148 (ΔINHAT I + II), huB-168, huB-LOA, and chA-208, compared to the full-length equivalents (Figure 6.5). All constructs lacking the LCAR interacted poorly with core histones H3 and H4, suggesting the LCAR and not the LRR mediates histone binding. Surprisingly, the full-length version of huB-LOA interacted with H3 and H4 significantly better than wildtype ANP32B. Thus this construct does not seem to affect core histone binding as suggested in the literature (Tochio et al., 2010), likely because their construct lacks the LCAR.

Figure 6.5 ANP32 binding to core histones is mediated by the LCAR Split luciferase complementation assays in HEK 293T cells showing core histones H3 (A-D) and H4 (E-H) binding by chicken ANP32A (full-length wildtype (FL) and chA-208 (ΔLCAR); A and E), human ANP32A (wildtype and huA-149 (ΔLCAR); B and F), human ANP32B (wildtype and huB-168 (ΔLCAR); C and G) and the quintuple loss-of-acidity human ANP32B mutant (LOA; D and H). pCAGGS expression plasmids encoding luc2-tagged ANP32 proteins (25 ng) were transfected in 1 x 105 HEK293T cells along with equal amounts of luc1-tagged histone constructs. Cells were lysed and luciferase substrate added 24 hours post-transfection. Data shown are mean (SD) normalised luminescence ratio (NLR) of one representative technical triplicate repeat experiment (n = 2 independent biological repeat experiments). Statistical significance was obtained by t test. **, P<0.01; ***, P<0.001; ****, P<0.0001.

Next, minigenome reporter assays were carried out in eHAP dKO cells to test whether the constructs could support FluPol activity (Figure 6.6). The LCAR of the human ANP32A and ANP32B proteins was required for pro-influenza virus function, but the avian LCAR was not. The huB LOA construct also did not support H3N2 Vic/75 or H5N1 50-92 (PB2-627K) FluPol activity despite an unaltered LCAR domain, suggesting that the LCAR alone is not sufficient for proviral activity. This was not due to poor expression, as analysed by western blotting analysis, or aberrant nuclear localisation measured by immunofluorescence microscopy.

134

Human ANP32A lacking INHAT domain II only (huA-189) retained some capacity to support FluPol activity, but human ANP32A lacking both INHAT domains (huA-148) lost all FluPol supportive activity, although poor expression of this construct perhaps explained the observed phenotype, at least in part. When tested for ability to support H3N2 Vic/75 FluPol, chicken ANP32A lacking its LCAR (chA-208) was significantly less able to support FluPol activity, in contrast with the avian-origin H5N1 FluPol constellation with the mammalian-adaptive PB2 mutation E627K.

Figure 6.6 ANP32B-LOA does not support FluPol activity despite having an intact LCAR (A and B) Minigenome assays in eHAP dKO cells with co-expressed wildtype or LCAR truncated human ANP32A (huA-189 and 148), human ANP32B (huB-168 and 155), or chicken ANP32A (chA-208), as well as the human ANP32B quintuple loss-of-acidity LRR mutant (huB-LOA) in support of H3N2 Vic/75 (A) or H5N1 50-92 (PB2-627K) (B) FluPol. RNP components PB1, PB2, PA and NP, a pPolI firefly luciferase minigenome and Renilla luciferase control, and the indicated ANP32 construct or Empty pCAGGS were transfected in 1 x 105 cells (48-well plates) in a ratio 2:2:1:4:2:2:5 (20 ng PB1). Data shown are mean (SD) firefly activity normalized to Renilla and analysed by one-way ANOVA from one representative technical triplicate repeat (n=2 independent biological repeat experiments). ns, not significant; ****, P<0.0001. (C) Western blotting analysis showing expression levels in eHAP dKO cells of FLAG-tagged ANP32A (top) and ANP32B (bottom) constructs used for the minigenome assays in (A) and (B). Western blotting analysis was carried out as described (D) Immunofluorescence microscopy analysis showing FLAG-tagged ANP32A (top) and ANP32B (bottom) constructs localised to the eHAP dKO cell nucleus. The analysis was carried out as in the legend to Figure 6.4, culturing ~250,000 eHAP dKO cells on glass coverslips and transfecting with 200 ng FLAG-tagged pCAGGS- ANP32 constructs 24 hours prior to fixing. Cells were incubated with primary antibody mouse α-FLAG (F1804, 1/200; Sigma-Aldrich) followed by goat α-mouse Alexa Fluor-568 conjugated secondary

135 antibody (1/200; Life Technologies). Chromosomal DNA was stained DAPI (1/1,000) during incubation with the secondary antibody. Imaging was carried out by David Gaboriau of the Imperial College Facility for Imaging by Light Microscopy (FILM) on a Zeiss Cell Observer widefield microscope using the DAPI (ex 365/12 nm, em 447/60 nm) and TexasRed (ex 562/40 nm, em 624/40 nm) channels.

Next, using the split luciferase complementation assay, binding of the ANP32 constructs to FluPol was investigated. As in co-immunoprecipitation experiments previously (Mistry et al., 2020), it was confirmed that the LCAR of chicken ANP32A is required for binding to H5N1 50- 92 FluPol with either lysine or glutamate at position 627 of the polymerase subunit PB2 (Figure 6.7). This was unexpected bearing in mind that this construct was able to support FluPol activity (Figure 6.6). FluPol binding of the human ANP32A and ANP32B constructs lacking the LCAR (huA-148 and huB-168, respectively) did correlate with the functional data – neither construct was able to interact with H5N1 50-92 FluPol, with or without the mammalian adaptive PB2-E627K substitution. Although this had been observed previously (Mistry et al., 2020), it is interesting to highlight that although avian-signature PB2-627E polymerase activity cannot be supported by human ANP32 proteins in human cells, the human proteins interacted with either FluPol version. Intriguingly, the huB-LOA mutant did bind avian- and mammalian- signature H5N1 50-92 FluPol, albeit not as well as wildtype human ANP32B, despite showing no support for FluPol activity.

Figure 6.7 ANP32B-LOA interacts with FluPol despite lacking pro-viral activity Split luciferase complementation assays in HEK 293T cells showing binding of chicken ANP32A (full-length wildtype (FL) or chA-208 (ΔLCAR); A and E), human ANP32B (wildtype (FL) or huB-168 (ΔLCAR); B and F), human ANP32A (wildtype or huA-148; C and G) and huB-LOA (D and H) to mammalian-adapted (PB2- 627K; A-D) or avian (PB2-627E; E-H) H5N1 50-92 FluPol. Luc2-tagged ANP32 constructs (15 ng) were transfected in 1 x 105 HEK 293T cells (48-well plates) along with equal amounts of FluPol subunits PB1- luc1, PB2 (627E or K) and PA (all 15 ng). Cells were lysed and luciferase substrate added 24 hours post-transfection. Data shown are mean (SD) normalised luminescence ratio (NLR) of one

136 representative technical triplicate repeat experiment (n = 2 independent biological repeat experiments). Statistical significance was obtained by t test. *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001.

A caveat of the data shown above was that the LCAR truncations in the ANP32A and ANP32B constructs were of unequal length. The human ANP32A-ΔLCAR construct (huA-148) lacked the central region as well as the LCAR (it was initially designed to lack the INHAT domains described in the literature (Seo et al., 2002)), while the ANP32B-ΔLCAR construct used in the assays above (huB-168) also lacked seven amino acid residues of the central region. The reason for this was that in our initial experiments (Figure 6.3) the huB-175 mutant (which lacked the LCAR but retained the central region) had some capacity to support FluPol activity. In order to resolve these discrepancies, an additional human ANP32A LCAR deletion (Figure 6.2 construct 4) was generated that was equivalent to the chicken ANP32A ΔLCAR construct (Figure 6.2 construct 2) which had been published, and to huB ΔLCAR (construct 9). The equivalent constructs huA ΔLCAR and huB ΔLCAR were then tested for function and interaction with FluPol.

First, minigenome reporter assays were carried out in eHAP dKO cells co-expressing the set of ANP32 constructs with or without LCAR domain (chicken ANP32A wildtype or chA ΔLCAR; human ANP32A wildtype or huA ΔLCAR; human ANP32B wildtype or huB ΔLCAR) with a panel of FluPol constellations. Identical patterns of H3N2 Vic/75 and H5N1 50-92 (PB2-627K) FluPol rescue were found when the new constructs (huA ΔLCAR and huB ΔLCAR) were transfected, compared with the slightly shorter constructs used previously: neither human construct lacking the LCAR supported FluPol activity, whilst, as shown before, the truncated chicken orthologue did.

Next, additional analysis of FluPol constellations derived from pH1N1 Eng/195 and influenza B virus Florida/06 was carried out (Figure 6.8). In line with the results obtained with the other influenza A virus polymerases, the 2009 pandemic strain polymerase was not supported by huA ΔLCAR or huB ΔLCAR but retained its activity in the presence of ch ΔLCAR, albeit activity was lower than with full-length chicken ANP32A. The ANP32B-LOA construct was unable to support Eng/195 FluPol activity. The chicken ANP32A orthologue did not support IBV polymerase very well. This has also been shown in published work from our laboratory (Peacock et al., 2020b). Nevertheless, the wildtype protein was capable of supporting IBV FluPol to some extent, a capacity that was lost when the truncated chA-208 construct was co- expressed. This loss of function was in contrast with chA-208 support for IAV polymerase. Human ANP32B is the more efficient host factor for support of IBV polymerase (Peacock et al., 2019; Peacock et al., 2020b; Staller et al., 2020b; Staller et al., 2019). This was reflected in these assays as well, but IBV FluPol activity was not supported by co-expressing ΔLCAR versions of either ANP32A or ANP32B. The huB-LOA construct did not rescue IBV polymerase

137 activity either. None of these phenotypes were due to differences in expression of the novel constructs, as analysed by immunoblotting.

Split luciferase complementation assays were also carried out with the novel (huA ΔLCAR and huB ΔLCAR) and existing (chA ΔLCAR and huB-LOA) ANP32 constructs, in order to compare their interaction with H5N1 50-92 (PB2-627K) FluPol, compared to the respective full-length proteins. The LCAR was required for chicken and human ANP32 interaction with FluPol, while the huB-LOA construct bound to FluPol despite not being able to support its replicative activity, albeit less well than wildtype ANP32B.

Figure 6.8 Human ANP32 proteins lacking the LCAR domain cannot support FluPol activity (A and B) Minigenome reporter assays in 1 x 105 eHAP dKO cells (48-well plates) with co-expressed FLAG-tagged chicken ANP32A (chA), human ANP32A (huA) or human ANP32B (huB) with (FL) or without (ΔLCAR) the LCAR domain (50 ng), in support of pH1N1 Eng/195 (A) or IBV Florida/06 (B) polymerases reconstituted by transfection of RNP components PB1, PB2, PA and NP, a pPolI firefly luciferase minigenome and Renilla luciferase control, in a ratio 2:2:1:4:2:2 (20 ng PB1). Data shown are mean (SD) firefly activity normalised to Renilla, from one representative technical triplicate experiment (n = 2 independent biological repeat experiments), analysed by one-way ANOVA. Statistical significance asterisks indicate, from left to right, full-length wildtype chicken ANP32A (FL) versus chA ΔLCAR; full-length wildtype human ANP32A (FL) versus huA ΔLCAR; full-length wildtype human ANP32B (FL) versus huB ΔLCAR; and full-length wildtype human ANP32B (FL) versus huB-LOA,

138 obtained by t test. GFP, which is similar in size to ANP32 proteins but does not support FluPol activity, was used as negative control. (C) Split luciferase complementation assay in HEK 293T cells showing binding of chicken ANP32A (chA), human ANP32A (huA) or human ANP32B (huB) with (FL) or without (ΔLCAR) the LCAR domain, as well as huB-LOA to mammalian-adapted (PB2-627K) H5N1 50-92 FluPol. Luc2-tagged ANP32 constructs (15 ng) were transfected in 1 x 105 HEK293T cells (48-well plates) along with equal amounts of FluPol subunits PB1-luc1, PB2 (627K) and PA (15 ng each). Cells were lysed and luciferase substrate added 24 hours post-transfection. Data shown are mean (SD) normalised luminescence ratio (NLR) of one representative technical triplicate repeat experiment (n = 2 independent biological repeat experiments). Statistical significance was obtained by t test comparing, from left to right, full-length wildtype chicken ANP32A (FL) versus chA ΔLCAR; full-length wildtype human ANP32A (FL) versus huA ΔLCAR; full-length wildtype human ANP32B (FL) versus huB ΔLCAR; and full-length wildtype human ANP32B (FL) versus huB-LOA. **, P<0.01; ***, P<0.001; ****, P<0.0001. (D) Western blotting analysis showing expression of the novel human ANP32A and ANP32B LCAR mutants, compared with the wildtype equivalents and chA ΔLCAR. Western blotting analysis was carried out as described.

6.4 ANP32B nuclear localisation is required for pro-influenza virus function Although we and others have shown by immunofluorescence microscopy that human ANP32B localises predominantly to the cell nucleus, and by bimolecular fluorescence complementation (BiFC) that chicken and human ANP32A interact with FluPol inside the nucleus (Mistry et al., 2020), it has been suggested that ANP32 proteins exert their pro-influenza virus function predominantly in the cytoplasm, perhaps by shuttling FluPol components into the nucleus.

To assess this, we attempted to generate ANP32 mutants that would not reach the nucleus. ANP32B possesses a classical monopartite nuclear localisation signal (NLS) near the C- terminus, namely 239-KRKR-242 (Fries et al., 2007). This basic Lys-Arg-Lys-Arg motif is recognized by the adaptor protein importin-α, which in turn forms a heterodimer with the karyopherin importin-β. The latter mediates active nuclear import through interaction with the nuclear pore complex (NPC) (Lange et al., 2007). As ANP32B is known to shuttle between the cytoplasm and nucleus, it must also contain a nuclear export signal (NES) recognizable by the nuclear exportin CRM1 (Fung et al., 2017). ANP32B actually functions as an adaptor protein facilitating CRM1-mediated nuclear export of HuR as part of the heat shock response (Brennan et al., 2000; Gallouzi et al., 2001). The ANP32B NES was found to be the canonical leucine-rich HIV-1 Rev-like motif 109-LKKLECLKSLDL-120 (Fries et al., 2007). An ANP32B construct fused to the cytoplasmic enzyme β-galactosidase with the NLS replaced by alanines was predominantly localised in the cytoplasm of transfected HeLa cells (Fries et al., 2007), as measured by immunofluorescence microscopy with a primary antibody against βGal. A caveat

139 of this approach is that the larger fusion protein would be more dependent on active transport. On the other hand, exogenously expressed FLAG-tagged chicken ANP32A constructs lacking the LCAR and therefore also the NLS, but not the putative NES, have been shown to remain predominantly nuclear in MRC5 cells (Domingues and Hale, 2017).

Using overlap extension PCR, a FLAG-tagged ANP32B NLS mutant was generated, with K241 and R242 replaced by alanine residues (i.e. KRKR  KRAA; construct 13 in Figure 6.2). Minigenome reporter assays showed that this construct could support H3N2 Vic/75 FluPol as well as wildtype ANP32B (Figure 6.9). Expression of the NLS mutant construct was not affected, as analysed by immunoblotting, while immunofluorescence microscopy demonstrated no difference in nuclear localisation compared with the wildtype protein.

Figure 6.9 ANP32B NLS mutant does not affect nuclear localisation or FluPol activity (A) Cartoon of ANP32B with the C-terminal nuclear localisation signal (NLS) highlighted in the box. The two C- terminal residues of the underlined 239-KRKR-242 motif (NLS), which is recognized by importin-α, were replaced by alanines (KRAA). (B) Minigenome reporter assay in 2 x 105 eHAP dKO cells (24-well plate) with co-expressed wildtype ANP32B (WT) or NLS mutant (KRAA) (100 ng) in support of H3N2 Vic/75 FluPol reconstituted from vRNP components PB1, PB2, PA and NP, a pPolI firefly luciferase minigenome reporter and Renilla luciferase control, in a ratio 2:2:1:4:2:2 (40 ng PB1). Data shown are mean (SD) firefly activity normalized to Renilla with statistical analysis by one-way ANOVA from one representative technical triplicate repeat (n=2 independent biological repeat experiments). ns = not significant. Accompanying western blot shows expression of the FLAG-tagged wildtype and mutant ANP32B constructs alongside Vic/75 PB2 and cellular vinculin loading control. (C) Immunofluorescence analysis showing nuclear localisation of the ANP32B NLS mutant. RNP components were co-expressed to control for binding of ANP32B to newly synthesised FluPol components in the cytoplasm, as an alternative route into the nucleus. Expression and nuclear localisation of RNP components was controlled by using GFP-tagged PA of H5N1 5092K FluPol. ~250,000 eHAP dKO cells were cultured

140 on glass coverslips in 12-well plates and transfected with 50 ng FLAG-tagged ANP32B-KRAA construct and FluPol components (20 ng PB1, PB2-627K, and PA-GFP) 24 hours prior to fixing. Cells were incubated with primary antibody mouse α-FLAG (F1804, 1/200; Sigma-Aldrich) followed by goat α- mouse Alexa Fluor-568 conjugated secondary antibody (1/200; Life Technologies). Chromosomal DNA was stained with DAPI (1/1,000) during incubation with the secondary antibody. Imaging was carried out by David Gaboriau of the Imperial College Facility for Imaging by Light Microscopy (FILM) on a Zeiss Cell Observer widefield microscope using the DAPI (ex 365/12 nm, em 447/60 nm), GFP (ex 470/40 nm, em 525/50 nm) and TexasRed (ex 562/40 nm, em 624/40 nm) channels.

The pattern recognition receptor RIG-I is generally believed to be cytoplasmic, but nuclear localisation has been shown (Liu et al., 2018). In the nucleus, it could potentially detect influenza virus dsRNA intermediate replication products. Stable expressed was set up of RIG- I fused to a well-established NES derived from the transcription factor Nrf2 (Li et al., 2005), or a scrambled version thereof, in RIG-I knockout A549 cells. Nuclear localisation of RIG-I was no longer evident as measured by immunofluorescence microscopy when it was fused to the genuine NES, compared with the scrambled version.

Using overlap extension PCR, the same NES sequence – LKKQLSTLYL – was cloned in place of the C-terminal NLS of ANP32B to generate a FLAG-tagged ANP32B construct with an additional NES but no NLS (ANP32B-NES; Figure 6.2 construct 14). An additional ANP32B construct with a scrambled version of the NES – YTLLSLQKLK – was also generated (ANP32B-NESscr; Figure 6.2 construct 15). It was hypothesised that because ANP32B is small enough to diffuse through the nuclear pore (32 kDa) it could not be biased toward the cytoplasm solely by eliminating the NLS. But by introducing an additional well-established NES it might be rendered predominantly cytoplasmic. This might lead to a difference in its capacity to support FluPol activity in the nucleus. The authors of the studies described above did not have issues with potential diffusion into the nucleus because they used large fusion partners: β-galactosidase is 82 kDa and RIG-I is 107 kDa, well above the 60 kDa threshold for diffusion across the nuclear pore (Wang and Brattain, 2007).

Minigenome reporter assays were carried out in eHAP dKO cells with H3N2 Vic/75 and H5N1 5092K polymerase constellations in the presence of the NLS mutants. FluPol activity was attenuated about two-fold when the ANP32B construct containing the canonical Nrf2 NES in place of the NLS was co-expressed, compared to the wildtype and KRAA constructs. Importantly, FluPol activity was not affected in the presence of the ANP32B construct with a scrambled version of the NES (Figure 6.10). Western blotting analysis suggested that this was not due to poor expression of the constructs, although a high-quality western blot was not obtained.

141

Thus, the C-terminal NLS of ANP32B did not seem required for nuclear localisation through the importin pathway; ANP32B is small enough to diffuse into the nucleus or otherwise it might enter in complex with other proteins. Some other function may be ascribed to the NLS in future work, or additional motifs might be used for nuclear import. On the other hand, actively cycling of ANP32B out of the nucleus through addition of an additional NES had a clearly negative effect on FluPol activity, suggesting ANP32B is required inside the nucleus to exert its pro- influenza virus function. Immunofluorescence analysis of the ANP32B-NES constructs was of insufficient quality to observe differences in cytoplasmic localisation.

Figure 6.10 Substituting the ANP32B NLS for a canonical NES affects FluPol activity (A and B) Minigenome reporter assays in 2 x 105 (24-well plates) eHAP dKO cells with co-expressed FLAG- tagged wildtype ANP32B (WT), NLS mutant (KRAA), NLS  NES mutant (NES), or a scrambled version of the NES (NESscr) (100 ng), in support of either H5N1 50-92 (PB2-627K) (A) or H3N2 Vic/75 FluPol (B) reconstituted from vRNP components PB1, PB2, PA and NP, a pPolI firefly luciferase minigenome reporter and Renilla luciferase control, in a ratio 2:2:1:4:2:2 (40 ng PB1). Data shown are mean (SD) firefly activity normalized to Renilla with statistical analysis by one-way ANOVA from one representative technical triplicate repeat (n=2 independent biological repeat experiments). ns, not significant; ****, P<0.0001. (C) Western blot showing expression of FLAG-tagged wildtype ANP32B and NES mutant constructs in eHAP dKO cells. (D) Immunofluorescence microscopy analysis suggesting predominantly nuclear localisation of FLAG-tagged wildtype ANP32B and NES mutants. ~250,000 eHAP dKO cells were cultured on glass coverslips in 12-well plates and transfected with 200 ng FLAG-tagged pCAGGS- ANP32B constructs 24 hours prior to fixing. Cells were incubated with primary antibody mouse α-FLAG (F1804, 1/200; Sigma-Aldrich) followed by goat α-mouse Alexa Fluor-568 conjugated secondary antibody (1/200; Life Technologies). Chromosomal DNA was stained with DAPI (1/1,000) during incubation with the secondary antibody. Imaging was carried out by David Gaboriau of the Imperial

142

College Facility for Imaging by Light Microscopy (FILM) on a Zeiss Cell Observer widefield microscope using the DAPI (ex 365/12 nm, em 447/60 nm) and TexasRed (ex 562/40 nm, em 624/40 nm) channels.

6.5 Inability of ANP32E to support FluPol activity maps to E129 ANP32E is a multifunctional protein with roles in cancer, apoptosome activation and protein phosphatase PP2A inhibition (Reilly et al., 2014). Like its ANP32A and ANP32B paralogues it is also a histone chaperone, specifically removing histone H2A.Z from DNA, possibly leading to transcriptional activation (Mao et al., 2014; Obri et al., 2014). ANP32E interacts with histone H2A.Z through a unique conserved sequence in the LCAR not found in ANP32A or ANP32B. Containing a small N-terminal alpha helix, the H2A.Z interacting domain (ZID) spans amino acids 215-240; ANP32E (268 amino acid residues) is longer than ANP32A (249) or ANP32B (251). The primary structure of ANP32E is depicted in Figure 6.2, construct 16, with the ZID domain underlined.

Western blotting analysis showed that ANP32E was endogenously expressed in eHAP control and dKO cells (Figure 6.11). As the latter did not support FluPol activity or virus replication, ANP32E clearly does not possess pro-influenza virus activity. Alignment of the primary structures of all three human ANP32 paralogues showed that ANP32E has a glutamic acid residue at position 129, instead of the canonical asparagine (compare Figure 6.2 constructs 3 (ANP32A), 8 (ANP32B) and 16 (ANP32E)). An asparagine residue was substituted for the glutamate at position 129 in order to investigate whether gain-of-function mutation could be obtained in ANP32E. Interestingly, ANP32E-E129N could support IAV polymerase constellations derived from H3N2 Vic/75, H5N1 50-92 (PB2-627K) or pH1N1 Eng/195 to some extent, compared with GFP negative control or with wildtype ANP32E (Figure 6.11), albeit not as well as wildtype ANP32A or ANP32B.

Strikingly, ANP32E-E129N was as potent as wildtype ANP32B, and more potent than wildtype ANP32A, in its support for IBV polymerase, while wildtype ANP32E did not rescue IBV polymerase activity. Western blotting analysis showed that these phenotypes were unrelated to expression of the constructs. These results lend additional strength to the observation that the 129-130 dyad in ANP32 proteins is essential for pro-influenza virus activity, as well as suggesting potential differences in the sequences of ANP32 proteins required to support influenza B virus polymerase, compared with influenza A virus polymerase.

143

Figure 6.11 ANP32E has potential pro-influenza virus activity (A) Western blotting analysis using a polyclonal anti-ANP32E antibody raised in rabbit (ab5993, 1/1,000; Abcam) showing ANP32E is expressed in wildtype and dKO eHAP cells. Western blotting was carried out as described (B-E) Minigenome reporter assays in eHAP dKO cells (~1 x 105) with 50 ng co-expressed FLAG-tagged ANP32A, ANP32B, ANP32E, ANP32E-E129N mutant, or GFP negative control in support of polymerases from H3N2 Vic/75 (B), pH1N1 Eng/195 (C), H5N1 50-92 (PB2-627K) (D) or IBV Florida/06 (E) reconstituted from vRNP components PB1, PB2, PA and NP, a pPolI firefly luciferase reporter and Renilla luciferase control, in a ratio 2:2:1:4:2:2 (20 ng PB1). Data shown are mean (SD) firefly activity normalised to Renilla, from one representative technical triplicate experiment (n = 2 independent biological repeat experiments), analysed by t test. ns, not significant; ***, P<0.001; ****, P<0.0001. (F) Western blotting analysis showing expression of FLAG-tagged wildtype and mutant ANP32E in eHAP dKO cells.

6.6 Discussion

This chapter has shown that the acidic LCAR domain of human ANP32A and ANP32B are required to support (mammalian-adapted PB2-627K) influenza virus polymerase activity, whereas the LCAR of chicken ANP32A is not – chicken ANP32A ΔLCAR supports both avian (PB2-627E) and mammalian-adapted IAV FluPol. It was further shown that interaction between human and chicken ANP32 proteins with FluPol, and with core histones H3 and H4, is mediated by the LCAR domain. So, in the chicken ANP32A-ΔLCAR construct, comprising

144 amino acid residues 1-208, we identified an ANP32 orthologue that supports FluPol activity seemingly without physically associating with it, as measured both by co-immunoprecipitation (Mistry et al., 2020) and split Gaussia luciferase complementation assays. Given the general positive correlation found between binding and pro-influenza viral function of ANP32 proteins, this is puzzling. In contrast, the ANP32B loss-of-acidity (LOA) mutant, which lacks five acidic residues on the concave face of the LRR domain, but retains a full-length LCAR domain, could not support FluPol activity despite interacting with it, as measured by split luciferase complementation assay. ANP32B-LOA also interacted with core histones H3 and H4. So, in this construct we seem to have uncoupled an endogenous function of ANP32B, namely as a histone chaperone, from its pro-influenza virus polymerase function. Table 6.1 summarizes these findings. Full alignment of the primary structures of the constructs as presented in Figure 6.2 are reproduced in Figure 6.12.

Construct Histone binding FluPolE/K binding FluPolE function FluPolK function 1 chA FL Y Y Y Y 2 chA ΔLCAR N N Y Y 3 huA FL Y Y N Y 4 huA ΔLCAR N N N N 8 huB FL Y Y N Y 9 huB ΔLCAR N N N N 12 huB LOA Y Y N N

Table 6.1 Interaction and functional profile of ΔLCAR constructs Interaction with core histones H3.1 and H4, and with reconstituted IAV H5N1 5092 polymerase was measured by split luciferase complementation assay, with luc1 fused to the carboxy terminal of PB1 and luc2 to the carboxy terminal of ANP32 proteins. 18-19 amino acid residues including the canonical KRKR NLS were retained in the ΔLCAR constructs. Capacity to support IAV and IBV FluPol replication in dKO cells was measured with minireplicon assays. Avian-signature H5N1 5092 (PB2-627E) polymerase activity was probed only with the chicken constructs, as shorter mammalian ANP32 proteins cannot support avian FluPol activity. Capacity to bind histones and FluPol usually correlates with proviral function and requires the LCAR domain, but this correlation breaks down for the constructs in bold script: Chicken ANP32A lacking the LCAR loses its capacity to interact with FluPol and histones but can still support FluPol activity; the huB LOA construct retains its binding capacity to both FluPol and histones, but is unable to support FluPol replication.

145

Figure 6.12 Primary structures of ΔLCAR constructs Alignment of chicken ANP32A, human ANP32A, and human ANP32B full-length (FL) and ΔLCAR constructs, as well as the human ANP32B loss-of-acidity construct, as first presented in Figure 6.2.

In order to explain the imperfect correlation between binding to FluPol and pro-viral functionality of ANP32 proteins, it is helpful to turn to the unpublished structural work introduced in the previous chapter, where chicken and human ANP32A were captured bridging a dimer of a ‘replicating’ and an ‘encapsidating’ influenza C virus (ICV) heterotrimeric polymerase (Figure 6.13). The replicating polymerase (FluPolR) interacts predominantly with the N-terminal portion of the LRR domain of ANP32A, while the pro-viral 129-130 dyad at the C-terminal end of the LRR interacts directly with residues of the encapsidating polymerase

(FluPolE). Specifically, ANP32A-129N forms a hydrogen bond with lysine 391 of the P3 subunit of the polymerase (P3 in ICV is the equivalent of PA in IAV and IBV), while 130D bonds with P3-608K. Furthermore, ANP32A-119D, one of the acidic residues that were replaced by alanine in the quintuple ANP32B loss-of-acidity (LOA) mutant, forms a hydrogen bond with

P3-543R of FluPolE. The other four acidic residues substituted by alanine in ANP32B-LOA,

D25, E45, E70 and E73, map to FluPolR, although none of these were shown to interact directly with the polymerase. These observations taken together suggest that ANP32B-LOA may still interact with mammalian-adapted PB2-627K FluPol through its LCAR domain, in particular via the acidic motif 176-DEEDEDDE-183, and that this is the interaction measured in the split luciferase assay. It’s proviral function is lost, however, as it can no longer interact with the FluPol dimer via its LRR. The 176-183 motif has been suggested to pack against the PB2 domains of the replication-competent FluPol dimer (Figure 6.13 C; Carrique et al. Nature 2020, in press). This motif bears a mixture of acidic and basic residues in chicken ANP32A, but only acidic residues in the human orthologues (Figure 6.13 D). This might be the reason why chicken ANP32A but not human ANP32 proteins, is able to support 627E as well as 627K polymerase, with 627E potentially acting to repel the acidic human motifs. Indeed a chimeric

146 human ANP32A construct with the chicken ANP32A 176-183 motif inserted was able to support avian-signature PB2-627E polymerase (Carrique et al 2020, in press).

Figure 6.13 ANP32A LRR bridges asymmetric FluPol dimer (Figures adapted from Carrique et al. 2020, in press) (A) The human ANP32A LRR domain (orange) in complex with a dimer of ICV polymerases. The replicating polymerase (FluPolR; in blue) is bound by a 47-nucleotide vRNA promoter; the encapsidating polymerase (FluPolE) is shown in green. The PB2 domains are juxtaposed. The N- terminal portion of ANP32A interacts with FluPolR, while the C-terminal half including residues D119, N129 and D130 maps to FluPolE. EMDB 10665 and PDB 6XZQ (B) Close up of the pro-influenza virus dyad 129N-130D forming hydrogen bonds with residues K391 and K608, respectively. (C) Close-up of the unique part of the chicken ANP32A central region that packs against the PB2 domain of FluPolR containing the species-specific lysine 649 (the ICV equivalent of IAV and IBV PB2-627). EMDB 10666 and PDB 6XZR (D) Alignment of amino acids 176-183 of chicken ANP32A (chA), human ANP32A (huA) and human ANP32B (huB), which maps to the central region of chicken ANP32A but to the N-terminal portion of the LCAR of the human proteins. These amino acids are believed to pack against the juxtaposed PB2 subunits of the asymmetric FluPol dimer in (A) and (C). Basic residues are shown in blue; acidic residues in red.

Cryo-EM density maps with the corresponding atomic coordinates have been deposited in the Electron Microscopy Data Bank with accession codes EMD-10665 (FluPolC–huANP32A subclass 1), EMD-10667 (FluPolC–huANP32A subclass 2), EMD-10666 (FluPolC–chANP32A subclass 1), EMD- 10659 (FluPolC–chANP32A subclass 2), EMD-10662 (FluPolC–chANP32A subclass 3) and EMD-10664 (FluPolC–chANP32A subclass 4), and the Protein Data Bank with accession codes 6XZQ (FluPolC– huANP32A subclass 1), 6Y0C (FluPolC-huANP32A subclass 2), 6XZR (FluPolC– chANP32A subclass 1),

147

6XZD (FluPolC–chANP32A subclass 2), 6XZG (FluPolC–chANP32A subclass 3) and 6XZP (FluPolC– chANP32A subclass 4).

Alignment of the amino-terminal portion of the ANP32 LCAR of each of the constructs shows this in more detail (Figure 6.14). The chicken ANP32A ΔLCAR construct retains the crucial mixture of acidic and basic residues within the 176-183 motif and is therefore still able to support FluPol function. The reason why binding to the FluPol dimer is not observed is currently unknown and requires further investigation. Interestingly, the human ANP32A and ANP32B ΔLCAR constructs harbour a mixture of basic and acidic residues within the crucial motif due to retention of the basic C-terminal NLS. It may therefore be interesting to test whether these shortened proteins can support avian FluPol, even if interaction with mammalian-adapted FluPol was not observed. It should also be noted that the split luciferase complementation assay is imperfect, as luc2 is fused to the LCAR which may alter its mobility and function to an extent.

Figure 6.14 The 176-183 motif is important for proviral function The 176-183 motif at the N- terminal end of the ANP32 LCAR domain is believed to pack against the 627 domains of either one or the other FluPol heterotrimer that make up the replication-competent dimer. This domain harbours the important PB2-627 residue that is acidic in avian FluPol but basic in most mammalian-adapted polymerases. The motif contains a mixture of basic and acidic residues in chicken ANP32A (176- VLSLVKDR-183) but is wholly acidic in the human orthologues. Substitution of the chicken motif in human ANP32A enabled support of avian-signature FluPol by the latter. Interestingly, as the basic NLS

148 was retained in the human ΔLCAR constructs they now harbour basic residues at the crucial site, so it may be interesting to test whether human ΔLCAR constructs can support avian FluPol activity.

Through mutational analysis of the classical monopartite nuclear localisation signal (NLS) at the C-terminus of ANP32B, we showed that nuclear localisation may be required for support of FluPol activity by ANP32 proteins. But because of their small size it has proved extremely difficult to expel ANP32 proteins from the nucleus. Follow-up work might investigate the involvement of a putative alternative nuclear localisation sequence described in the literature. Most leucine rich repeat (LRR) domains, including the LRRs of ANP32 proteins, have capping motifs at the N-terminal and C-terminal ends (Figure 1.6) , presumably to shield the hydrophobic leucine-rich core from solvent, preventing aggregation (Dao et al., 2014). These caps have highly conserved primary sequences – the C-terminal cap of ANP32A has a consensus sequence of 131-YRxxφxxxφPxφxxLD-146 (φ represents a hydrophobic residue; x represents any residue), where the flanking tyrosine and aspartate side-chains form a structural hydrogen bond (Ceulemans et al., 1999). Mutation of the C-terminal cap led to nuclear exclusion of the fission yeast protein phosphatase 1 regulatory subunit sds22 (Stone et al., 1993). Mutational analysis of this domain in human ANP32 proteins may further illuminate potential mechanisms of subcellular localisation.

Finally, we demonstrate that the reason why human ANP32E cannot support influenza polymerase is because of a glutamate residue at position 129 rather than the canonical pro- viral asparagine. A single gain-of-function amino acid substitution, ANP32E-E129N, led to some activity of IAV FluPol in dKO cells, as well as full activity of IBV FluPol. Further work is required to understand differences in preference for ANP32 host factors between influenza polymerase constellations derived from different strains of IAV and IBV, as well as from different host species. It is possible that ANP32E is able to support FluPol of certain strains that are able to interact with it despite the poor match at 129. Access to ANP32E in the eHAP double knockout cells that lack ANP32A and ANP32B expression, in combination with its potential to support FluPol activity, might explain the low levels of replication seen by FluPol constellations from H5N1 Tky/05 (this thesis), as well as H1N1 WSN/33 in HEK293T double knockout cells (Zhang et al., 2019a).

149

VII DISCUSSION

Seasonal influenza viruses cause respiratory disease in millions during annually recurring epidemics, claiming 200,000 – 500,000 lives worldwide. In addition, avian influenza viruses can occasionally transfer into humans either directly from birds or via intermediate mammalian hosts such as pigs, potentially causing pandemic influenza (Krammer et al., 2018; Lycett et al., 2019). To infect mammalian cells efficiently, adaptations to the influenza polymerase must occur. The most common of these is a glutamate-to-lysine substitution in the PB2 subunit of the polymerase (Subbarao et al., 1993). Pandemic influenza has the potential to kill millions, as no protective immunity exists in the human population. In order to pre-empt pandemic outbreaks and increase treatment options against seasonal influenza novel therapeutics are required. Traditional antivirals target components of the virus-encoded proteome, but resistance against these is acquired rapidly and consistently. A promising alternative is host- directed therapy, whose aim is to interfere with host cell factors that are required for influenza replication (Kaufmann et al., 2018).

Like all viruses, influenza virus usurps the host machinery to support its life cycle. Many of these are required for transcription and replication of the viral genome, which takes place in the host cell nucleus. Among such host factors ANP32 proteins represent perhaps the most promising targets for future influenza control strategies. Indeed, a collaboration between the Barclay laboratory and researchers at the University of Edinburgh aims to generate influenza- resistant chickens, based on amino acid changes in avian ANP32A. The closely related human paralogues ANP32A and ANP32B were first described to enhance influenza replication in 2015 (Sugiyama et al., 2015). In a seminal 2016 functional screen from the Barclay laboratory it was shown that chicken ANP32A, which contains a 33-amino acid insertion between the LRR and LCAR domains, is required for avian (PB2-627E) FluPol activity (Long et al., 2016). Avian-signature FluPol is inactive in human cells because the human ANP32 orthologues lack the 33-amino acid insertion. Artificial insertion of these 33 amino acids into human ANP32A or ANP32B rescued avian FluPol activity in human cells (Long et al., 2016). This was the starting point of this thesis.

Although human ANP32A and ANP32B had been targeted previously by RNA interference (Long et al., 2016; Sugiyama et al., 2015), showing a phenotype in which influenza polymerase activity and virus replication were reduced but not abrogated, it was decided to attempt ablation of the genes by CRISPR/Cas9 genome editing technology (Ran et al., 2013b). In order to enhance specificity and reduce off-target DNA cleavage a double-nickase strategy (Ran et al., 2013a) was developed against exon 2 of Anp32A and Anp32B. Having been

150 unsuccessful in HEK293T and A549 cells due to polyploidy issues, low-ploidy eHAP cells (Essletzbichler et al., 2014) were selected for generating knockout cell lines. Minigenome reporter assays and virus challenge of ANP32A (AKO) and ANP32B (BKO) knockout cells revealed a mixed phenotype in which small deleterious effects on polymerase activity were seen only for certain polymerase constellations. It was then decided to generate cells ablated for both ANP32A and ANP32B (dKO). Minigenome assays and virus challenge in three independent monoclonal dKO cell lines revealed a striking phenotype in which influenza polymerase activity and virus replication were completely abrogated (Chapter III).

Direct comparison in dKO cells of influenza polymerases derived from a wide variety of sources – a seasonal H3N2 strain (Vic/75), a 2009 pandemic H1N1 isolate (Eng/195), a mammalian-adapted avian H5N1 strain (5092K), and a seasonal influenza B virus (Florida/06) – showed that ANP32 proteins were required for the activity of each. Furthermore, multicycle growth curves in dKO cells showed unequivocally that human ANP32 proteins are required for virus replication. The fact that FluPol was still active when either ANP32A or ANP32B was present but activity was completely abrogated in the absence of both paralogues led to the conclusion that human ANP32A and ANP32B form a pair of functionally redundant essential host factors for influenza virus replication in human cells (Chapter IV). This work culminated in publication in the Journal of Virology in 2019 (Staller et al., 2019). Lending strength to our main conclusion a separate report was published describing similar results in HEK293T double knockout cells (Zhang et al., 2019a).

Future work should address whether influenza B virus replication is equally affected in dKO cells. Although minireplicon assays suggested IBV FluPol activity is indeed dependent on ANP32 proteins, we did not manage to achieve replication of IBV Florida/06, a seasonal strain of the Yamagata lineage, in eHAP cells. It was puzzling to consistently find minimal accumulation of infectious virus in growth curves with the avian zoonotic strain H5N1 Tky/05, whereas titres of all other viruses tested remained below the level of detection. The question of why this isolate should achieve some replication in the absence of ANP32A and ANP32B is currently being actively pursued in the Barclay laboratory.

Pro-influenza viral activity of ANP32 proteins has been mapped to amino acid residues 129 and 130 (Long et al., 2019a; Staller et al., 2019; Zhang et al., 2019a). ANP32 proteins that do not support influenza polymerase – human ANP32E, chicken ANP32B, and mouse ANP32A – invariably encode amino acids that divert from the proviral dyad 129N-130D. It was shown that mouse ANP32A, which naturally harbours alanine at position 130, does not support influenza A virus polymerase, although intriguingly it does support influenza B virus polymerase (Chapter IV). A domain swap introducing mouse ANP32B LRR 5 into mouse

151

ANP32A allowed the chimeric construct to rescue influenza polymerase activity in dKO cells. The same phenotype was obtained by the single amino acid substitution A130D in mouse ANP32A (Staller et al., 2019; Zhang et al., 2019a). The capacity of mouse ANP32A to support IBV FluPol remains unexplained and may be investigated in future, for instance using the recently developed ANP32B knockout mouse model (Beck et al., 2020; Chemnitz et al., 2019).

Using the knowledge about the importance of ANP32 amino acid residues 129 and 130 for influenza virus polymerase activity, we searched public databases for naturally occurring variation in ANP32A and ANP32B in the human population (Chapter V). One particular SNV in the Anp32B gene, rs182096718, was interesting as it encodes the amino acid substitution D130A, the exact same substitution seen in mouse ANP32A and the reason why it lacks proviral activity. Whereas all other SNVs were extremely rare (global minor allele frequency (MAF) < 0.0001), rs182096718 was relatively common at a MAF of 0.0044. Almost all the variation clustered within the Latino subpopulation of the gnomAD database however, where in a total pool of 35,420 alleles 1,209 D130A alleles were identified. Moreover, rs182096718 was the only SNV for which homozygous carriers had been described in the database, again all of them in the Latino cohort.

Given the importance of 130D it was hypothesised that carriers of rs182096718 may have some protection against influenza virus. In order to prove the potential significance of this in heterozygous carriers of the SNV it was paramount to demonstrate that ANP32B-D130A exert a dominant-negative effect over the wildtype allele. Indeed, co-transfection of wildtype ANP32B and ANP32B-D130A in a 1:1 ratio in cells lacking ANP32B expression (BKO) led to reduced FluPol activity compared with provision of wildtype ANP32B alone. As ANP32A and ANP32B are functionally redundant in their support for influenza polymerase, and ANP32A is presumably present in both homozygous and heterozygous carriers of rs182096718, paralogue interference of ANP32B-D130A over wildtype ANP32A had to be demonstrated as well. Indeed, expression of ANP32B-D130A but not wildtype ANP32B in cells naturally expressing ANP32A (BKO) showed reduced FluPol activity.

Human eHAP cells expressing wildtype ANP32A and ANP32B-D130A (the homozygous mutant genotype) were then generated by CRISPR/Cas9 editing. In order to achieve a specific A>C point mutation in exon 4 of Anp32B a homology-directed repair template bearing the mutation was introduced alongside Cas9 DNA endonuclease. A single clone bearing the mutation was obtained, where multiple clones would have been desirable. Minigenome reporter assays and virus challenge in the mutant cells showed reduced pandemic H1N1 virus FluPol activity and replication, compared with control cells as well as BKO cells. Again this suggested paralogue interference of ANP32B-D130A over the wildtype ANP32A present in

152 these cells (Chapter V). Finally, a model is proposed to account for both the reduction in FluPol activity and the dominant-negative effects over the wildtype ANP32 proteins. This model builds on recent structural work showing ANP32A as stabilising a replication-competent dimer of heterotrimeric FluPol units (Carrique et al. 2020 in press). We propose that competitive binding of ANP32B-D130A to the replicating FluPol via the unaltered N-terminal portion of its LRR prevents recruitment of wildtype ANP32B or ANP32A, explaining the dominant-negative effect and the paralogue interference, respectively. Recruitment of the encapsidating polymerase is impaired due to the absence of an acidic residue at position 130, leading to impaired formation of the replication-competent FluPolR-ANP32-FluPolE complex. This compromised recruitment would explain the reduction in replication efficiency. The work on natural genetic variation in ANP32 proteins described in Chapter V has culminated in a manuscript currently available as a preprint on the BioRxiv server (Staller et al., 2020a).

An outstanding question is what would happen to the nascent viral RNA replicated by FluPolR, if indeed replication of the template RNA takes place at all. The nascent RNA is believed to be picked up by FluPolE as soon as it emerges from FluPolR, but if FluPolE recruitment to FluPolR is impaired in the presence of ANP32B-D130A this will not be possible. Formation of polymerase dimers is a common evolutionary strategy through which viruses are believed to avoid detection of viral RNA by the host innate immune system. It would therefore be interesting to test whether IFN expression and interferon-stimulated gene (ISG) activity are upregulated in cells expressing ANP32B-D130A.

One obvious caveat of the work described in Chapter V is the use of a haploid laboratory cell line in which the more common heterozygous genotype with one wildtype and one mutant ANP32B allele could not be recapitulated. In future work the heterozygous genotype should be investigated using diploid lung epithelium cells, the target tissue of influenza virus. These could be derived directly from patients or alternatively differentiated from induced pluripotent stem cells. CRISPR/Cas9 editing may then be carried out at the iPSC stage, a process that has been described (Lee et al., 2020; van Riet et al., 2020). The latter strategy would provide a reproducible high quality bank of isogenic pairs of lung epithelial cells in which to study influenza virus host factors.

Mutational analysis was carried out to map pro-influenza virus function of ANP32 proteins to structural elements and domains, in particular the intrinsically disordered LCAR domain. It was confirmed that chicken ANP32A lacking the LCAR domain (amino acids 1-208) was still able to support both avian (PB2-627E) and mammalian-adapted (PB2-627K) influenza polymerase. Human ANP32A and ANP32B ΔLCAR constructs (amino acids 1-175), in contrast, could not support FluPol activity. This may be explained by a crucial interaction

153 between the 176-183 motif of ANP32 proteins and the 627 domains of the replicating and encapsidating FluPol in the dimeric structure (Carrique et al. 2020, in press). The 176-183 motif is identical in the full-length and ΔLCAR constructs of chicken ANP32A, but different between full-length and ΔLCAR human constructs. Significant charge changes in this motif may result in loss of function. Future work should address the function of the carboxy-terminal portion of the LCAR beyond the 176-183 domain. The acidic nature of this domain suggests a possible role in the recruitment of the basic nucleoprotein to the growing nascent RNA strand as it is being replicated by FluPol, but this idea requires further investigation. Using a split luciferase complementation assay developed in the Barclay laboratory interaction between chicken ANP32A ΔLCAR and FluPol could not be detected. It remains unclear how this construct could support FluPol activity without physically associating with it, so this is likely an artefact of the binding assay. Nevertheless absence of binding of chicken ANP32A ΔLCAR to influenza polymerase has been shown independently also by co-immunoprecipitation (Mistry et al., 2020). Future work should address these puzzling observations.

Of note, the split luciferase complementation assay used throughout this work is carried out in 293T cells, as these are more easily transfected than eHAP cells. A caveat of this approach is the presence of endogenous ANP32 proteins in these cells, which may compromise the observations if the overexpressed ANP32 proteins were to form homo- or heterodimers with their endogenous counterparts. Furthermore, it is unclear whether binding to FluPol dimers or monomers is observed, and if binding to dimers is the case, whether binding to the replicating or the encapsidating FluPol is measured. It is therefore important to repeat these binding assays in eHAP cells lacking at least ANP32A and ANP32B (dKO) or even all three family members. Alternative binding assays like co-immunoprecipitation should also be included where possible to verify the split luciferase binding data.

We attempted to uncouple an endogenous function of ANP32 proteins as histone chaperones from their pro-viral function. The rationale behind this idea is that if ANP32 proteins were to be targeted in a host-directed therapy approach, for instance through the design of small molecule inhibitors targeting the proviral dyad 129N-130D, it is paramount that cellular functions remain intact as much as possible. It was confirmed that the LCAR of human ANP32 proteins is required for core histone binding, but the LCAR was also needed for FluPol binding and support of FluPol function. The exception was the quintuple loss-of-acidity (LOA) mutant in ANP32B, which despite retaining both the LCAR and the 129N-130D proviral dyad, was unable to support FluPol activity. It did, however, interact with histones. The ANP32B-LOA construct thus comes closest to uncoupling endogenous from proviral function.

154

In summary, this study has provided further insights in the pro-influenza virus activity of ANP32 proteins. It has become clear that human ANP32A and ANP32B are a pair of functionally redundant essential host factors for influenza polymerase activity and virus replication in human cells. An additional finding was that the naturally occurring missense single nucleotide variant encoding ANP32B-D130A may result in some protection against influenza virus in carriers. The data presented in this study may lead to novel therapeutic strategies against influenza virus, and has provided additional insights in the fundamentals of virus-host interactions.

155

VIII REFERENCES . Alenquer, M., Vale-Costa, S., Etibor, T.A., Ferreira, F., Sousa, A.L., and Amorim, M.J. (2019). Influenza A virus ribonucleoproteins form liquid organelles at endoplasmic reticulum exit sites. Nat Commun 10, 1629. Allen, E.K., Randolph, A.G., Bhangale, T., Dogra, P., Ohlson, M., Oshansky, C.M., Zamora, A.E., Shannon, J.P., Finkelstein, D., Dressen, A., et al. (2017). SNP-mediated disruption of CTCF binding at the IFITM3 promoter is associated with risk of severe influenza in humans. Nat Med 23, 975-983. Amorim, M.J. (2018). A Comprehensive Review on the Interaction Between the Host GTPase Rab11 and Influenza A Virus. Frontiers in cell and developmental biology 6, 176. Amorim, M.J., Bruce, E.A., Read, E.K., Foeglein, A., Mahen, R., Stuart, A.D., and Digard, P. (2011). A Rab11- and microtubule-dependent mechanism for cytoplasmic transport of influenza A virus viral RNA. Journal of virology 85, 4143-4156. Andersson, B.S., Beran, M., Pathak, S., Goodacre, A., Barlogie, B., and McCredie, K.B. (1987). Ph- positive chronic myeloid leukemia with near-haploid conversion in vivo and establishment of a continuously growing cell line with similar cytogenetic pattern. Cancer genetics and cytogenetics 24, 335-343. Ando, T., Yamayoshi, S., Tomita, Y., Watanabe, S., Watanabe, T., and Kawaoka, Y. (2016). The host protein CLUH participates in the subnuclear transport of influenza virus ribonucleoprotein complexes. Nature microbiology 1, 16062. Anzalone, A.V., Randolph, P.B., Davis, J.R., Sousa, A.A., Koblan, L.W., Levy, J.M., Chen, P.J., Wilson, C., Newby, G.A., Raguram, A., et al. (2019). Search-and-replace genome editing without double-strand breaks or donor DNA. Nature 576, 149-157. Arranz, R., Coloma, R., Chichon, F.J., Conesa, J.J., Carrascosa, J.L., Valpuesta, J.M., Ortin, J., and Martin-Benito, J. (2012). The structure of native influenza virion ribonucleoproteins. Science 338, 1634-1637. Avilov, S.V., Moisy, D., Naffakh, N., and Cusack, S. (2012). Influenza A virus progeny vRNP trafficking in live infected cells studied with the virus-encoded fluorescently tagged PB2 protein. Vaccine 30, 7411-7417. Baker, S.F., Ledwith, M.P., and Mehle, A. (2018). Differential Splicing of ANP32A in Birds Alters Its Ability to Stimulate RNA Synthesis by Restricted Influenza Polymerase. Cell reports 24, 2581- 2588.e2584. Bauer, D.L.V., Tellier, M., Martínez-Alonso, M., Nojima, T., Proudfoot, N.J., Murphy, S., and Fodor, E. (2018). Influenza Virus Mounts a Two-Pronged Attack on Host RNA Polymerase II Transcription. Cell reports 23, 2119-2129.e2113. Beck, S., Zickler, M., Pinho dos Reis, V., Günther, T., Grundhoff, A., Reilly, P.T., Mak, T.W., Stanelle- Bertram, S., and Gabriel, G. (2020). ANP32B Deficiency Protects Mice From Lethal Influenza A Virus Challenge by Dampening the Host Immune Response. Frontiers in Immunology 11. Bercovich-Kinori, A., Tai, J., Gelbart, I.A., Shitrit, A., Ben-Moshe, S., Drori, Y., Itzkovitz, S., Mandelboim, M., and Stern-Ginossar, N. (2016). A systematic view on influenza induced host shutoff. Elife 5. Bhagwat, A.R., Le Sage, V., Nturibi, E., Kulej, K., Jones, J., Guo, M., Tae Kim, E., Garcia, B.A., Weitzman, M.D., Shroff, H., et al. (2020). Quantitative live cell imaging reveals influenza virus manipulation of Rab11A transport through reduced dynein association. Nat Commun 11, 23. Biswas, S.K., and Nayak, D.P. (1994). Mutational analysis of the conserved motifs of influenza A virus polymerase basic protein 1. Journal of virology 68, 1819-1826. Bodem, J., Schied, T., Gabriel, R., Rammling, M., and Rethwilm, A. (2011). Foamy virus nuclear RNA export is distinct from that of other retroviruses. Journal of virology 85, 2333-2341. Boivin, S., and Hart, D.J. (2011). Interaction of the influenza A virus polymerase PB2 C-terminal region with importin alpha isoforms provides insights into host adaptation and polymerase assembly. The Journal of biological chemistry 286, 10439-10448.

156

Boulo, S., Akarsu, H., Lotteau, V., Müller, C.W., Ruigrok, R.W., and Baudin, F. (2011). Human importin alpha and RNA do not compete for binding to influenza A virus nucleoprotein. Virology 409, 84-90. Bouloy, M., Plotch, S.J., and Krug, R.M. (1980). Both the 7-methyl and the 2'-O-methyl groups in the cap of mRNA strongly influence its ability to act as primer for influenza virus RNA transcription. Proceedings of the National Academy of Sciences of the United States of America 77, 3952-3956. Bradel-Tretheway, B.G., Mattiacio, J.L., Krasnoselsky, A., Stevenson, C., Purdy, D., Dewhurst, S., and Katze, M.G. (2011). Comprehensive proteomic analysis of influenza virus polymerase complex reveals a novel association with mitochondrial proteins and RNA polymerase accessory factors. Journal of virology 85, 8569-8581. Brannan, K., Kim, H., Erickson, B., Glover-Cutter, K., Kim, S., Fong, N., Kiemele, L., Hansen, K., Davis, R., Lykke-Andersen, J., et al. (2012). mRNA decapping factors and the exonuclease Xrn2 function in widespread premature termination of RNA polymerase II transcription. Molecular cell 46, 311-324. Brass, A.L., Huang, I.C., Benita, Y., John, S.P., Krishnan, M.N., Feeley, E.M., Ryan, B.J., Weyer, J.L., van der Weyden, L., Fikrig, E., et al. (2009). The IFITM proteins mediate cellular resistance to influenza A H1N1 virus, West Nile virus, and dengue virus. Cell 139, 1243-1254. Bravo Garcia-Morato, M., Calvo Apalategi, A., Bravo-Gallego, L.Y., Blazquez Moreno, A., Simon- Fuentes, M., Garmendia, J.V., Mendez Echevarria, A., Del Rosal Rabes, T., Dominguez-Soto, A., Lopez- Granados, E., et al. (2019). Impaired control of multiple viral infections in a family with complete IRF9 deficiency. The Journal of allergy and clinical immunology 144, 309-312.e310. Brennan, C.M., Gallouzi, I.E., and Steitz, J.A. (2000). Protein ligands to HuR modulate its interaction with target mRNAs in vivo. The Journal of cell biology 151, 1-14. Bruce, E.A., Digard, P., and Stuart, A.D. (2010). The Rab11 pathway is required for influenza A virus budding and filament formation. Journal of virology 84, 5848-5859. Brunotte, L., Flies, J., Bolte, H., Reuther, P., Vreede, F., and Schwemmle, M. (2014). The nuclear export protein of H5N1 influenza A viruses recruits Matrix 1 (M1) protein to the viral ribonucleoprotein to mediate nuclear export. The Journal of biological chemistry 289, 20067-20077. Bui, M., Whittaker, G., and Helenius, A. (1996). Effect of M1 protein and low pH on nuclear transport of influenza virus ribonucleoproteins. Journal of virology 70, 8391-8401. Burkard, C., Lillico, S.G., Reid, E., Jackson, B., Mileham, A.J., Ait-Ali, T., Whitelaw, C.B., and Archibald, A.L. (2017). Precision engineering for PRRSV resistance in pigs: Macrophages from genome edited pigs lacking CD163 SRCR5 domain are fully resistant to both PRRSV genotypes while maintaining biological function. PLoS pathogens 13, e1006206. Burkard, C., Opriessnig, T., Mileham, A.J., Stadejek, T., Ait-Ali, T., Lillico, S.G., Whitelaw, C.B.A., and Archibald, A.L. (2018). Pigs Lacking the Scavenger Receptor Cysteine-Rich Domain 5 of CD163 Are Resistant to Porcine Reproductive and Respiratory Syndrome Virus 1 Infection. Journal of virology 92. Burnet, F.M. (1937). Influenza Virus on the Developing Egg. IV The Pathogenicity and Immunizing Power of Egg Virus for Ferrets and Mice. Br J Exp Pathol 18, 37-43. Bystrykh, L.V. (2012). Generalized DNA barcode design based on Hamming codes. PloS one 7, e36852. Camacho-Zarco, A.R., Kalayil, S., Maurin, D., Salvi, N., Delaforge, E., Milles, S., Jensen, M.R., Hart, D.J., Cusack, S., and Blackledge, M. (2020). Molecular basis of host-adaptation interactions between influenza virus polymerase PB2 subunit and ANP32A. Nat Commun 11, 3656. Carette, J.E., Raaben, M., Wong, A.C., Herbert, A.S., Obernosterer, G., Mulherkar, N., Kuehne, A.I., Kranzusch, P.J., Griffin, A.M., and Ruthel, G. (2011). Ebola virus entry requires the cholesterol transporter Niemann–Pick C1. Nature 477, 340-343. Cargnello, M., and Roux, P.P. (2011). Activation and function of the MAPKs and their substrates, the MAPK-activated protein kinases. Microbiology and molecular biology reviews : MMBR 75, 50-83. Carmody, S.R., and Wente, S.R. (2009). mRNA nuclear export at a glance. Journal of cell science 122, 1933-1937.

157

Carrington, M., Dean, M., Martin, M.P., and O'Brien, S.J. (1999). Genetics of HIV-1 Infection: Chemokine Receptor Ccr5 Polymorphism and Its Consequences. Human Molecular Genetics 8, 1939- 1945. Ceulemans, H., De Maeyer, M., Stalmans, W., and Bollen, M. (1999). A capping domain for LRR protein interaction modules. FEBS letters 456, 349-351. Chan, A.Y., Vreede, F.T., Smith, M., Engelhardt, O.G., and Fodor, E. (2006). Influenza virus inhibits RNA polymerase II elongation. Virology 351, 210-217. Chang, H.H.Y., Pannunzio, N.R., Adachi, N., and Lieber, M.R. (2017). Non-homologous DNA end joining and alternative pathways to double-strand break repair. Nature reviews Molecular cell biology 18, 495-506. Chang, S., Sun, D., Liang, H., Wang, J., Li, J., Guo, L., Wang, X., Guan, C., Boruah, B.M., Yuan, L., et al. (2015). Cryo-EM structure of influenza virus RNA polymerase complex at 4.3 Å resolution. Molecular cell 57, 925-935. Chase, G., Deng, T., Fodor, E., Leung, B.W., Mayer, D., Schwemmle, M., and Brownlee, G. (2008). Hsp90 inhibitors reduce influenza virus replication in cell culture. Virology 377, 431-439. Chase, G.P., Rameix-Welti, M.A., Zvirbliene, A., Zvirblis, G., Götz, V., Wolff, T., Naffakh, N., and Schwemmle, M. (2011). Influenza virus ribonucleoprotein complexes gain preferential access to cellular export machinery through chromatin targeting. PLoS pathogens 7, e1002187. Chatzopoulou, F., Gioula, G., Kioumis, I., Chatzidimitriou, D., and Exindari, M. (2019). Identification of complement-related host genetic risk factors associated with influenza A(H1N1)pdm09 outcome: challenges ahead. Medical microbiology and immunology 208, 631-640. Chemnitz, J., Pieper, D., Stich, L., Schumacher, U., Balabanov, S., Spohn, M., Grundhoff, A., Steinkasserer, A., Hauber, J., and Zinser, E. (2019). The acidic protein rich in leucines Anp32b is an immunomodulator of inflammation in mice. Scientific Reports 9, 4853. Chen, K.Y., Santos Afonso, E.D., Enouf, V., Isel, C., and Naffakh, N. (2019). Influenza virus polymerase subunits co-evolve to ensure proper levels of dimerization of the heterotrimer. PLoS pathogens 15, e1008034. Chen, Y., Zhou, J., Cheng, Z., Yang, S., Chu, H., Fan, Y., Li, C., Wong, B.H., Zheng, S., Zhu, Y., et al. (2015). Functional variants regulating LGALS1 (Galectin 1) expression affect human susceptibility to influenza A(H7N9). Scientific reports 5, 8517. Cheng, Z., Zhou, J., To, K.K., Chu, H., Li, C., Wang, D., Yang, D., Zheng, S., Hao, K., Bosse, Y., et al. (2015). Identification of TMPRSS2 as a Susceptibility Gene for Severe 2009 Pandemic A(H1N1) Influenza and A(H7N9) Influenza. The Journal of infectious diseases 212, 1214-1221. Chiba, S., Hill-Batorski, L., Neumann, G., and Kawaoka, Y. (2018). The Cellular DExD/H-Box RNA Helicase UAP56 Co-localizes With the Influenza A Virus NS1 Protein. Frontiers in microbiology 9, 2192. Choppin, P.W. (1969). Replication of influenza virus in a continuous cell line: high yield of infective virus from cells inoculated at high multiplicity. Virology 39, 130-134. Ciancanelli, M.J., Abel, L., Zhang, S.Y., and Casanova, J.L. (2016). Host genetics of severe influenza: from mouse Mx1 to human IRF7. Curr Opin Immunol 38, 109-120. Ciancanelli, M.J., Huang, S.X., Luthra, P., Garner, H., Itan, Y., Volpi, S., Lafaille, F.G., Trouillet, C., Schmolke, M., Albrecht, R.A., et al. (2015). Infectious disease. Life-threatening influenza and impaired interferon amplification in human IRF7 deficiency. Science (New York, NY) 348, 448-453. Collins, R.L., Brand, H., Karczewski, K.J., Zhao, X., Alföldi, J., Francioli, L.C., Khera, A.V., Lowther, C., Gauthier, L.D., Wang, H., et al. (2019). An open resource of structural variation for medical and population genetics. bioRxiv, 578674. Consortium, G.T. (2013). The Genotype-Tissue Expression (GTEx) project. Nat Genet 45, 580-585. Crescenzo-Chaigne, B., van der Werf, S., and Naffakh, N. (2002). Differential effect of nucleotide substitutions in the 3' arm of the influenza A virus vRNA promoter on transcription/replication by avian and human polymerase complexes is related to the nature of PB2 amino acid 627. Virology 303, 240-252.

158

Cros, J.F., García-Sastre, A., and Palese, P. (2005). An unconventional NLS is critical for the nuclear import of the influenza A virus nucleoprotein and ribonucleoprotein. Traffic (Copenhagen, Denmark) 6, 205-213. Dao, T.P., Majumdar, A., and Barrick, D. (2014). Capping motifs stabilize the leucine-rich repeat protein PP32 and rigidify adjacent repeats. Protein science : a publication of the Protein Society 23, 801-811. de Castro Martin, I.F., Fournier, G., Sachse, M., Pizarro-Cerda, J., Risco, C., and Naffakh, N. (2017). Influenza virus genome reaches the plasma membrane via a modified endoplasmic reticulum and Rab11-dependent vesicles. Nat Commun 8, 1396. Deng, T., Engelhardt, O.G., Thomas, B., Akoulitchev, A.V., Brownlee, G.G., and Fodor, E. (2006). Role of ran binding protein 5 in nuclear import and assembly of the influenza virus RNA polymerase complex. Journal of virology 80, 11911-11919. Doench, J.G., Fusi, N., Sullender, M., Hegde, M., Vaimberg, E.W., Donovan, K.F., Smith, I., Tothova, Z., Wilen, C., Orchard, R., et al. (2016). Optimized sgRNA design to maximize activity and minimize off- target effects of CRISPR-Cas9. Nature biotechnology 34, 184-191. Domingues, P., Eletto, D., Magnus, C., Turkington, H.L., Schmutz, S., Zagordi, O., Lenk, M., Beer, M., Stertz, S., and Hale, B.G. (2019). Profiling host ANP32A splicing landscapes to predict influenza A virus polymerase adaptation. Nature communications 10, 3396-3396. Domingues, P., and Hale, B.G. (2017). Functional Insights into ANP32A-Dependent Influenza A Virus Polymerase Host Restriction. Cell Rep 20, 2538-2546. Don, R.H., Cox, P.T., Wainwright, B.J., Baker, K., and Mattick, J.S. (1991). 'Touchdown' PCR to circumvent spurious priming during gene amplification. Nucleic acids research 19, 4008-4008. Ehrhardt, C., Rückle, A., Hrincius, E.R., Haasbach, E., Anhlan, D., Ahmann, K., Banning, C., Reiling, S.J., Kühn, J., Strobl, S., et al. (2013). The NF-κB inhibitor SC75741 efficiently blocks influenza virus propagation and confers a high barrier for development of viral resistance. Cellular microbiology 15, 1198-1211. Eisfeld, A.J., Kawakami, E., Watanabe, T., Neumann, G., and Kawaoka, Y. (2011). RAB11A is essential for transport of the influenza virus genome to the plasma membrane. Journal of virology 85, 6117- 6126. Eisfeld, A.J., Neumann, G., and Kawaoka, Y. (2015). At the centre: influenza A virus ribonucleoproteins. Nature reviews Microbiology 13, 28-41. Elton, D., Simpson-Holley, M., Archer, K., Medcalf, L., Hallam, R., McCauley, J., and Digard, P. (2001). Interaction of the influenza virus nucleoprotein with the cellular CRM1-mediated nuclear export pathway. Journal of virology 75, 408-419. Engelhardt, O.G., and Fodor, E. (2006). Functional association between viral and cellular transcription during influenza virus infection. Reviews in medical virology 16, 329-345. Engelhardt, O.G., Smith, M., and Fodor, E. (2005). Association of the influenza A virus RNA- dependent RNA polymerase with cellular RNA polymerase II. Journal of virology 79, 5812-5818. Esposito, S., Molteni, C.G., Giliani, S., Mazza, C., Scala, A., Tagliaferri, L., Pelucchi, C., Fossali, E., Plebani, A., and Principi, N. (2012). Toll-like receptor 3 gene polymorphisms and severity of pandemic A/H1N1/2009 influenza in otherwise healthy children. Virology journal 9, 270. Essletzbichler, P., Konopka, T., Santoro, F., Chen, D., Gapp, B.V., Kralovics, R., Brummelkamp, T.R., Nijman, S.M., and Burckstummer, T. (2014). Megabase-scale deletion using CRISPR/Cas9 to generate a fully haploid human cell line. Genome Res 24, 2059-2065. Everitt, A.R., Clare, S., Pertel, T., John, S.P., Wash, R.S., Smith, S.E., Chin, C.R., Feeley, E.M., Sims, J.S., Adams, D.J., et al. (2012). IFITM3 restricts the morbidity and mortality associated with influenza. Nature 484, 519-523. Fan, H., Walker, A.P., Carrique, L., Keown, J.R., Serna Martin, I., Karia, D., Sharps, J., Hengrung, N., Pardon, E., Steyaert, J., et al. (2019). Structures of influenza A virus RNA polymerase offer insight into viral genome replication. Nature 573, 287-290.

159

Fan, Z., Zhang, H., and Zhang, Q. (2006). Tumor suppressor pp32 represses cell growth through inhibition of transcription by blocking acetylation and phosphorylation of histone H3 and initiating its proapoptotic activity. Cell death and differentiation 13, 1485-1494. Finberg, R.W., Lanno, R., Anderson, D., Fleischhackl, R., van Duijnhoven, W., Kauffman, R.S., Kosoglou, T., Vingerhoets, J., and Leopold, L. (2019). Phase 2b Study of Pimodivir (JNJ-63623872) as Monotherapy or in Combination With Oseltamivir for Treatment of Acute Uncomplicated Seasonal Influenza A: TOPAZ Trial. The Journal of infectious diseases 219, 1026-1034. Fodor, E., and Te Velthuis, A.J.W. (2019). Structure and Function of the Influenza Virus Transcription and Replication Machinery. Cold Spring Harbor perspectives in medicine. Fox, D.T., and Duronio, R.J. (2013). Endoreplication and polyploidy: insights into development and disease. Development (Cambridge, England) 140, 3-12. Francis, T., Salk, J.E., Pearson, H.E., and Brown, P.N. (1945). PROTECTIVE EFFECT OF VACCINATION AGAINST INDUCED INFLUENZA A. The Journal of clinical investigation 24, 536-546. Fries, B., Heukeshoven, J., Hauber, I., Gruttner, C., Stocking, C., Kehlenbach, R.H., Hauber, J., and Chemnitz, J. (2007). Analysis of nucleocytoplasmic trafficking of the HuR ligand APRIL and its influence on CD83 expression. The Journal of biological chemistry 282, 4504-4515. Frise, R., Bradley, K., van Doremalen, N., Galiano, M., Elderfield, R.A., Stilwell, P., Ashcroft, J.W., Fernandez-Alonso, M., Miah, S., Lackenby, A., et al. (2016). Contact transmission of influenza virus between ferrets imposes a looser bottleneck than respiratory droplet transmission allowing propagation of antiviral resistance. Sci Rep 6, 29793. Fu, W., O’Connor, T.D., Jun, G., Kang, H.M., Abecasis, G., Leal, S.M., Gabriel, S., Rieder, M.J., Altshuler, D., Shendure, J., et al. (2013). Analysis of 6,515 exomes reveals the recent origin of most human protein-coding variants. Nature 493, 216-220. Fung, H.Y.J., Fu, S.-C., and Chook, Y.M. (2017). Nuclear export receptor CRM1 recognizes diverse conformations in nuclear export signals. eLife 6, e23961. Gabriel, G., Herwig, A., and Klenk, H.D. (2008). Interaction of polymerase subunit PB2 and NP with importin alpha1 is a determinant of host range of influenza A virus. PLoS pathogens 4, e11. Gabriel, G., Klingel, K., Otte, A., Thiele, S., Hudjetz, B., Arman-Kalcek, G., Sauter, M., Shmidt, T., Rother, F., Baumgarte, S., et al. (2011). Differential use of importin-α isoforms governs cell tropism and host adaptation of influenza virus. Nat Commun 2, 156. Gallouzi, I.E., Brennan, C.M., and Steitz, J.A. (2001). Protein ligands mediate the CRM1-dependent export of HuR in response to heat shock. RNA (New York, NY) 7, 1348-1361. Genomes Project, C., Auton, A., Brooks, L.D., Durbin, R.M., Garrison, E.P., Kang, H.M., Korbel, J.O., Marchini, J.L., McCarthy, S., McVean, G.A., et al. (2015). A global reference for human genetic variation. Nature 526, 68-74. Giard, D.J., Aaronson, S.A., Todaro, G.J., Arnstein, P., Kersey, J.H., Dosik, H., and Parks, W.P. (1973). In vitro cultivation of human tumors: establishment of cell lines derived from a series of solid tumors. J Natl Cancer Inst 51, 1417-1423. Goldhill, D.H., Te Velthuis, A.J.W., Fletcher, R.A., Langat, P., Zambon, M., Lackenby, A., and Barclay, W.S. (2018). The mechanism of resistance to favipiravir in influenza. Proc Natl Acad Sci U S A 115, 11613-11618. Goto, H., and Kawaoka, Y. (1998). A novel mechanism for the acquisition of virulence by a human influenza A virus. Proc Natl Acad Sci U S A 95, 10224-10228. Graf, L., Dick, A., Sendker, F., Barth, E., Marz, M., Daumke, O., and Kochs, G. (2018). Effects of allelic variations in the human myxovirus resistance protein A on its antiviral activity. The Journal of biological chemistry 293, 3056-3072. Graham, F.L., Smiley, J., Russell, W.C., and Nairn, R. (1977). Characteristics of a human cell line transformed by DNA from human adenovirus type 5. J Gen Virol 36, 59-74. Gu, W., Gallagher, G.R., Dai, W., Liu, P., Li, R., Trombly, M.I., Gammon, D.B., Mello, C.C., Wang, J.P., and Finberg, R.W. (2015). Influenza A virus preferentially snatches noncoding RNA caps. RNA (New York, NY) 21, 2067-2075.

160

Günther, M., Bauer, A., Müller, M., Zaeck, L., and Finke, S. (2020). Interaction of host cellular factor ANP32B with matrix proteins of different paramyxoviruses. The Journal of general virology 101, 44- 58. Hale, B.G., Randall, R.E., Ortin, J., and Jackson, D. (2008). The multifunctional NS1 protein of influenza A viruses. The Journal of general virology 89, 2359-2376. Haller, O., Staeheli, P., Schwemmle, M., and Kochs, G. (2015). Mx GTPases: dynamin-like antiviral machines of innate immunity. Trends in microbiology 23, 154-163. Hampton, H.G., Watson, B.N.J., and Fineran, P.C. (2020). The arms race between bacteria and their phage foes. Nature 577, 327-336. Han, J., Perez, J.T., Chen, C., Li, Y., Benitez, A., Kandasamy, M., Lee, Y., Andrade, J., tenOever, B., and Manicassamy, B. (2018). Genome-wide CRISPR/Cas9 Screen Identifies Host Factors Essential for Influenza Virus Replication. Cell reports 23, 596-607. Hannoun, C. (2013). The evolving history of influenza viruses and influenza vaccines. Expert review of vaccines 12, 1085-1094. Hao, L., Sakurai, A., Watanabe, T., Sorensen, E., Nidom, C.A., Newton, M.A., Ahlquist, P., and Kawaoka, Y. (2008). Drosophila RNAi screen identifies host genes important for influenza virus replication. Nature 454, 890-893. Hayden, F.G., Sugaya, N., Hirotsu, N., Lee, N., de Jong, M.D., Hurt, A.C., Ishida, T., Sekino, H., Yamada, K., Portsmouth, S., et al. (2018). Baloxavir Marboxil for Uncomplicated Influenza in Adults and Adolescents. The New England journal of medicine 379, 913-923. Heckman, K.L., and Pease, L.R. (2007). Gene splicing and mutagenesis by PCR-driven overlap extension. Nature protocols 2, 924-932. Hernandez, N., Melki, I., Jing, H., Habib, T., Huang, S.S.Y., Danielson, J., Kula, T., Drutman, S., Belkaya, S., Rattina, V., et al. (2018). Life-threatening influenza pneumonitis in a child with inherited IRF9 deficiency. The Journal of experimental medicine 215, 2567-2585. Herrera-Ramos, E., Lopez-Rodriguez, M., Ruiz-Hernandez, J.J., Horcajada, J.P., Borderias, L., Lerma, E., Blanquer, J., Perez-Gonzalez, M.C., Garcia-Laorden, M.I., Florido, Y., et al. (2014). Surfactant protein A genetic variants associate with severe respiratory insufficiency in pandemic influenza A virus infection. Critical care (London, England) 18, R127. Hidaka, F., Matsuo, S., Muta, T., Takeshige, K., Mizukami, T., and Nunoi, H. (2006). A missense mutation of the Toll-like receptor 3 gene in a patient with influenza-associated encephalopathy. Clinical immunology (Orlando, Fla) 119, 188-194. Hirotsu, N., Sakaguchi, H., Sato, C., Ishibashi, T., Baba, K., Omoto, S., Shishido, T., Tsuchiya, K., Hayden, F.G., Uehara, T., et al. (2019). Baloxavir marboxil in Japanese pediatric patients with influenza: safety and clinical and virologic outcomes. Clinical infectious diseases : an official publication of the Infectious Diseases Society of America. Horton, R.M., Ho, S.N., Pullen, J.K., Hunt, H.D., Cai, Z., and Pease, L.R. (1993). Gene splicing by overlap extension. Methods in enzymology 217, 270-279. Hsu, P.D., Scott, D.A., Weinstein, J.A., Ran, F.A., Konermann, S., Agarwala, V., Li, Y., Fine, E.J., Wu, X., Shalem, O., et al. (2013). DNA targeting specificity of RNA-guided Cas9 nucleases. Nature biotechnology 31, 827-832. Hu, Y., Gor, V., Morikawa, K., Nagata, K., and Kawaguchi, A. (2017). Cellular splicing factor UAP56 stimulates trimeric NP formation for assembly of functional influenza viral ribonucleoprotein complexes. Scientific reports 7, 14053. Huang, S., Chen, J., Chen, Q., Wang, H., Yao, Y., Chen, J., and Chen, Z. (2013). A second CRM1- dependent nuclear export signal in the influenza A virus NS2 protein contributes to the nuclear export of viral ribonucleoproteins. Journal of virology 87, 767-778. Hudjetz, B., and Gabriel, G. (2012). Human-like PB2 627K influenza virus polymerase activity is regulated by importin-α1 and -α7. PLoS pathogens 8, e1002488.

161

Huet, S., Avilov, S.V., Ferbitz, L., Daigle, N., Cusack, S., and Ellenberg, J. (2010). Nuclear import and assembly of influenza A virus RNA polymerase studied in live cells by fluorescence cross-correlation spectroscopy. Journal of virology 84, 1254-1264. Hug, N., Longman, D., and Caceres, J.F. (2016). Mechanism and regulation of the nonsense-mediated decay pathway. Nucleic Acids Res 44, 1483-1495. Hutchinson, E.C., and Fodor, E. (2012). Nuclear import of the influenza A virus transcriptional machinery. Vaccine 30, 7353-7358. Hutchinson, E.C., Orr, O.E., Man Liu, S., Engelhardt, O.G., and Fodor, E. (2011). Characterization of the interaction between the influenza A virus polymerase subunit PB1 and the host nuclear import factor Ran-binding protein 5. The Journal of general virology 92, 1859-1869. Huyton, T., and Wolberger, C. (2007). The crystal structure of the tumor suppressor protein pp32 (Anp32a): structural insights into Anp32 family of proteins. Protein Sci 16, 1308-1315. Jackson, D., Cadman, A., Zurcher, T., and Barclay, W.S. (2002). A reverse genetics approach for recovery of recombinant influenza B viruses entirely from cDNA. J Virol 76, 11744-11747. Jagger, B.W., Wise, H.M., Kash, J.C., Walters, K.A., Wills, N.M., Xiao, Y.L., Dunfee, R.L., Schwartzman, L.M., Ozinsky, A., Bell, G.L., et al. (2012). An overlapping protein-coding region in influenza A virus segment 3 modulates the host response. Science (New York, NY) 337, 199-204. Jinek, M., Chylinski, K., Fonfara, I., Hauer, M., Doudna, J.A., and Charpentier, E. (2012). A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science (New York, NY) 337, 816-821. Jinek, M., East, A., Cheng, A., Lin, S., Ma, E., and Doudna, J. (2013). RNA-programmed genome editing in human cells. Elife 2, e00471. Jorba, N., Coloma, R., and Ortin, J. (2009). Genetic trans-complementation establishes a new model for influenza virus RNA transcription and replication. PLoS Pathog 5, e1000462. Kakugawa, S., Shimojima, M., Goto, H., Horimoto, T., Oshimori, N., Neumann, G., Yamamoto, T., and Kawaoka, Y. (2009). Mitogen-activated protein kinase-activated kinase RSK2 plays a role in innate immune responses to influenza virus infection. Journal of virology 83, 2510-2517. Kaltenegger, E., and Ober, D. (2015). Paralogue Interference Affects the Dynamics after Gene Duplication. Trends Plant Sci 20, 814-821. Karczewski, K.J., Francioli, L.C., Tiao, G., Cummings, B.B., Alföldi, J., Wang, Q., Collins, R.L., Laricchia, K.M., Ganna, A., Birnbaum, D.P., et al. (2019). Variation across 141,456 human exomes and genomes reveals the spectrum of loss-of-function intolerance across human protein-coding genes. bioRxiv, 531210. Karlas, A., Machuy, N., Shin, Y., Pleissner, K.P., Artarini, A., Heuer, D., Becker, D., Khalil, H., Ogilvie, L.A., Hess, S., et al. (2010). Genome-wide RNAi screen identifies human host factors crucial for influenza virus replication. Nature 463, 818-822. Katz, M., Amit, I., and Yarden, Y. (2007). Regulation of MAPKs by growth factors and receptor tyrosine kinases. Biochimica et biophysica acta 1773, 1161-1176. Kaufmann, S.H.E., Dorhoi, A., Hotchkiss, R.S., and Bartenschlager, R. (2018). Host-directed therapies for bacterial and viral infections. Nature reviews Drug discovery 17, 35-56. Kawaguchi, A., Momose, F., and Nagata, K. (2011). Replication-coupled and host factor-mediated encapsidation of the influenza virus genome by viral nucleoprotein. Journal of virology 85, 6197- 6204. Kawaguchi, A., and Nagata, K. (2007). De novo replication of the influenza virus RNA genome is regulated by DNA replicative helicase, MCM. The EMBO journal 26, 4566-4575. Kawakami, E., Watanabe, T., Fujii, K., Goto, H., Watanabe, S., Noda, T., and Kawaoka, Y. (2011). Strand-specific real-time RT-PCR for distinguishing influenza vRNA, cRNA, and mRNA. Journal of virological methods 173, 1-6. Kemler, I., Whittaker, G., and Helenius, A. (1994). Nuclear import of microinjected influenza virus ribonucleoproteins. Virology 202, 1028-1033.

162

Kervestin, S., and Jacobson, A. (2012). NMD: a multifaceted response to premature translational termination. Nature reviews Molecular cell biology 13, 700-712. Kleiner, R.E., Hang, L.E., Molloy, K.R., Chait, B.T., and Kapoor, T.M. (2018). A Chemical Proteomics Approach to Reveal Direct Protein-Protein Interactions in Living Cells. Cell chemical biology 25, 110- 120.e113. Kleinstiver, B.P., Pattanayak, V., Prew, M.S., Tsai, S.Q., Nguyen, N.T., Zheng, Z., and Joung, J.K. (2016). High-fidelity CRISPR-Cas9 nucleases with no detectable genome-wide off-target effects. Nature 529, 490-495. Kondoh, T., Letko, M., Munster, V.J., Manzoor, R., Maruyama, J., Furuyama, W., Miyamoto, H., Shigeno, A., Fujikura, D., Takadate, Y., et al. (2018). Single-Nucleotide Polymorphisms in Human NPC1 Influence Filovirus Entry Into Cells. J Infect Dis 218, S397-s402. König, R., Stertz, S., Zhou, Y., Inoue, A., Hoffmann, H.H., Bhattacharyya, S., Alamares, J.G., Tscherne, D.M., Ortigoza, M.B., Liang, Y., et al. (2010). Human host factors required for influenza virus replication. Nature 463, 813-817. Koppstein, D., Ashour, J., and Bartel, D.P. (2015). Sequencing the cap-snatching repertoire of H1N1 influenza provides insight into the mechanism of viral transcription initiation. Nucleic Acids Res 43, 5052-5064. Korbie, D.J., and Mattick, J.S. (2008). Touchdown PCR for increased specificity and sensitivity in PCR amplification. Nature protocols 3, 1452-1456. Kotecki, M., Reddy, P.S., and Cochran, B.H. (1999). Isolation and characterization of a near-haploid human cell line. Exp Cell Res 252, 273-280. Krammer, F., Smith, G.J.D., Fouchier, R.A.M., Peiris, M., Kedzierska, K., Doherty, P.C., Palese, P., Shaw, M.L., Treanor, J., Webster, R.G., et al. (2018). Influenza. Nature reviews Disease primers 4, 3. Kudo, N., Matsumori, N., Taoka, H., Fujiwara, D., Schreiner, E.P., Wolff, B., Yoshida, M., and Horinouchi, S. (1999). Leptomycin B inactivates CRM1/exportin 1 by covalent modification at a cysteine residue in the central conserved region. Proceedings of the National Academy of Sciences of the United States of America 96, 9112-9117. Kutay, U., Bischoff, F.R., Kostka, S., Kraft, R., and Görlich, D. (1997). Export of importin alpha from the nucleus is mediated by a specific nuclear transport factor. Cell 90, 1061-1071. Lange, A., Mills, R.E., Lange, C.J., Stewart, M., Devine, S.E., and Corbett, A.H. (2007). Classical nuclear localization signals: definition, function, and interaction with importin alpha. The Journal of biological chemistry 282, 5101-5105. Lee, B.J., Cansizoglu, A.E., Süel, K.E., Louis, T.H., Zhang, Z., and Chook, Y.M. (2006). Rules for nuclear localization sequence recognition by karyopherin beta 2. Cell 126, 543-558. Lee, H., Flynn, R., Sharma, I., Haberman, E., Carling, P.J., Nicholls, F.J., Stegmann, M., Vowles, J., Haenseler, W., Wade-Martins, R., et al. (2020). LRRK2 Is Recruited to Phagosomes and Co-recruits RAB8 and RAB10 in Human Pluripotent Stem Cell-Derived Macrophages. Stem Cell Reports 14, 940- 955. Lee, J., Shin, M.K., Ryu, D.K., Kim, S., and Ryu, W.S. (2010). Insertion and deletion mutagenesis by overlap extension PCR. Methods Mol Biol 634, 137-146. Lee, N., Cao, B., Ke, C., Lu, H., Hu, Y., Tam, C.H.T., Ma, R.C.W., Guan, D., Zhu, Z., Li, H., et al. (2017). IFITM3, TLR3, and CD55 Gene SNPs and Cumulative Genetic Risks for Severe Outcomes in Chinese Patients With H7N9/H1N1pdm09 Influenza. The Journal of infectious diseases 216, 97-104. Li, B., Clohisey, S.M., Chia, B.S., Wang, B., Cui, A., Eisenhaure, T., Schweitzer, L.D., Hoover, P., Parkinson, N.J., Nachshon, A., et al. (2020). Genome-wide CRISPR screen identifies host dependency factors for influenza A virus infection. Nat Commun 11, 164. Li, W., Jain, M.R., Chen, C., Yue, X., Hebbar, V., Zhou, R., and Kong, A.N. (2005). Nrf2 Possesses a redox-insensitive nuclear export signal overlapping with the leucine zipper motif. The Journal of biological chemistry 280, 28430-28438.

163

Liang, L., Jiang, L., Li, J., Zhao, Q., Wang, J., He, X., Huang, S., Wang, Q., Zhao, Y., Wang, G., et al. (2019). Low Polymerase Activity Attributed to PA Drives the Acquisition of the PB2 E627K Mutation of H7N9 Avian Influenza Virus in Mammals. mBio 10. Lieber, M., Smith, B., Szakal, A., Nelson-Rees, W., and Todaro, G. (1976). A continuous tumor-cell line from a human lung carcinoma with properties of type II alveolar epithelial cells. International journal of cancer 17, 62-70. Lim, H.K., Huang, S.X.L., Chen, J., Kerner, G., Gilliaux, O., Bastard, P., Dobbs, K., Hernandez, N., Goudin, N., Hasek, M.L., et al. (2019). Severe influenza pneumonitis in children with inherited TLR3 deficiency. The Journal of experimental medicine 216, 2038-2056. Lin, Y.C., Boone, M., Meuris, L., Lemmens, I., Van Roy, N., Soete, A., Reumers, J., Moisse, M., Plaisance, S., Drmanac, R., et al. (2014). Genome dynamics of the human embryonic kidney 293 lineage in response to cell biology manipulations. Nat Commun 5, 4767. Lindesmith, L., Moe, C., Marionneau, S., Ruvoen, N., Jiang, X., Lindblad, L., Stewart, P., LePendu, J., and Baric, R. (2003). Human susceptibility and resistance to Norwalk virus infection. Nature Medicine 9, 548-553. Liu, G., Lu, Y., Thulasi Raman, S.N., Xu, F., Wu, Q., Li, Z., Brownlie, R., Liu, Q., and Zhou, Y. (2018). Nuclear-resident RIG-I senses viral replication inducing antiviral immunity. Nature Communications 9, 3199. Long, J.S., Giotis, E.S., Moncorge, O., Frise, R., Mistry, B., James, J., Morisson, M., Iqbal, M., Vignal, A., Skinner, M.A., et al. (2016). Species difference in ANP32A underlies influenza A virus polymerase host restriction. Nature 529, 101-104. Long, J.S., Idoko-Akoh, A., Mistry, B., Goldhill, D., Staller, E., Schreyer, J., Ross, C., Goodbourn, S., Shelton, H., Skinner, M.A., et al. (2019a). Species specific differences in use of ANP32 proteins by influenza A virus. eLife 8, e45066. Long, J.S., Mistry, B., Haslam, S.M., and Barclay, W.S. (2019b). Host and viral determinants of influenza A virus species specificity. Nat Rev Microbiol 17, 67-81. Lonsdale, J., Thomas, J., Salvatore, M., Phillips, R., Lo, E., Shad, S., Hasz, R., Walters, G., Garcia, F., Young, N., et al. (2013). The Genotype-Tissue Expression (GTEx) project. Nat Genet 45, 580-585. Loucaides, E.M., von Kirchbach, J.C., Foeglein, A., Sharps, J., Fodor, E., and Digard, P. (2009). Nuclear dynamics of influenza A virus ribonucleoproteins revealed by live-cell imaging studies. Virology 394, 154-163. Ludwig, S., Wolff, T., Ehrhardt, C., Wurzer, W.J., Reinhardt, J., Planz, O., and Pleschka, S. (2004). MEK inhibition impairs influenza B virus propagation without emergence of resistant variants. FEBS letters 561, 37-43. Lukarska, M., Fournier, G., Pflug, A., Resa-Infante, P., Reich, S., Naffakh, N., and Cusack, S. (2017). Structural basis of an essential interaction between influenza polymerase and Pol II CTD. Nature 541, 117-121. Luo, M.L., Zhou, Z., Magni, K., Christoforides, C., Rappsilber, J., Mann, M., and Reed, R. (2001). Pre- mRNA splicing and mRNA export linked by direct interactions between UAP56 and Aly. Nature 413, 644-647. Lycett, S.J., Duchatel, F., and Digard, P. (2019). A brief history of bird flu. Philosophical transactions of the Royal Society of London Series B, Biological sciences 374, 20180257. Ma, H., Marti-Gutierrez, N., Park, S.W., Wu, J., Lee, Y., Suzuki, K., Koski, A., Ji, D., Hayama, T., Ahmed, R., et al. (2017). Correction of a pathogenic gene mutation in human embryos. Nature 548, 413-419. MacMorris, M., Brocker, C., and Blumenthal, T. (2003). UAP56 levels affect viability and mRNA export in Caenorhabditis elegans. RNA (New York, NY) 9, 847-857. Manz, B., Brunotte, L., Reuther, P., and Schwemmle, M. (2012). Adaptive mutations in NEP compensate for defective H5N1 RNA replication in cultured human cells. Nat Commun 3, 802. Mao, Z., Pan, L., Wang, W., Sun, J., Shan, S., Dong, Q., Liang, X., Dai, L., Ding, X., Chen, S., et al. (2014). Anp32e, a higher eukaryotic histone chaperone directs preferential recognition for H2A.Z. Cell research 24, 389-399.

164

Marjuki, H., Alam, M.I., Ehrhardt, C., Wagner, R., Planz, O., Klenk, H.D., Ludwig, S., and Pleschka, S. (2006). Membrane accumulation of influenza A virus hemagglutinin triggers nuclear export of the viral genome via protein kinase Calpha-mediated activation of ERK signaling. The Journal of biological chemistry 281, 16707-16715. Martin, K., and Helenius, A. (1991a). Nuclear transport of influenza virus ribonucleoproteins: the viral matrix protein (M1) promotes export and inhibits import. Cell 67, 117-130. Martin, K., and Helenius, A. (1991b). Transport of incoming influenza virus nucleocapsids into the nucleus. Journal of virology 65, 232-244. Mehle, A., and Doudna, J.A. (2008). An inhibitory activity in human cells restricts the function of an avian-like influenza virus polymerase. Cell host & microbe 4, 111-122. Meng, X.Y., Luo, Y., Anwar, M.N., Sun, Y., Gao, Y., Zhang, H., Munir, M., and Qiu, H.J. (2017). Long Non-Coding RNAs: Emerging and Versatile Regulators in Host-Virus Interactions. Front Immunol 8, 1663. Mills, T.C., Rautanen, A., Elliott, K.S., Parks, T., Naranbhai, V., Ieven, M.M., Butler, C.C., Little, P., Verheij, T., Garrard, C.S., et al. (2014). IFITM3 and susceptibility to respiratory viral infections in the community. The Journal of infectious diseases 209, 1028-1031. Mir, M.A., Duran, W.A., Hjelle, B.L., Ye, C., and Panganiban, A.T. (2008). Storage of cellular 5' mRNA caps in P bodies for viral cap-snatching. Proceedings of the National Academy of Sciences of the United States of America 105, 19294-19299. Mistry, B., Long, J.S., Schreyer, J., Staller, E., Sanchez-David, R.Y., and Barclay, W.S. (2020). Elucidating the Interactions between Influenza Virus Polymerase and Host Factor ANP32A. Journal of Virology 94, e01353-01319. Miyake, Y., Keusch, J.J., Decamps, L., Ho-Xuan, H., Iketani, S., Gut, H., Kutay, U., Helenius, A., and Yamauchi, Y. (2019). Influenza virus uses transportin 1 for vRNP debundling during cell entry. Nature microbiology 4, 578-586. Moeller, A., Kirchdoerfer, R.N., Potter, C.S., Carragher, B., and Wilson, I.A. (2012). Organization of the influenza virus replication machinery. Science 338, 1631-1634. Momose, F., Basler, C.F., O'Neill, R.E., Iwamatsu, A., Palese, P., and Nagata, K. (2001). Cellular splicing factor RAF-2p48/NPI-5/BAT1/UAP56 interacts with the influenza virus nucleoprotein and enhances viral RNA synthesis. Journal of virology 75, 1899-1908. Momose, F., Handa, H., and Nagata, K. (1996). Identification of host factors that regulate the influenza virus RNA polymerase activity. Biochimie 78, 1103-1108. Momose, F., Kikuchi, Y., Komase, K., and Morikawa, Y. (2007). Visualization of microtubule-mediated transport of influenza viral progeny ribonucleoprotein. Microbes and infection 9, 1422-1433. Momose, F., Naito, T., Yano, K., Sugimoto, S., Morikawa, Y., and Nagata, K. (2002). Identification of Hsp90 as a stimulatory host factor involved in influenza virus RNA synthesis. The Journal of biological chemistry 277, 45306-45314. Moncorgé, O., Long, J.S., Cauldwell, A.V., Zhou, H., Lycett, S.J., and Barclay, W.S. (2013). Investigation of influenza virus polymerase activity in pig cells. J Virol 87, 384-394. Moncorgé, O., Mura, M., and Barclay, W.S. (2010). Evidence for avian and human host cell factors that affect the activity of influenza virus polymerase. Journal of virology 84, 9978-9986. More, S., Zhu, Z., Lin, K., Huang, C., Pushparaj, S., Liang, Y., Sathiaseelan, R., Yang, X., and Liu, L. (2019). Long non-coding RNA PSMB8-AS1 regulates influenza virus replication. RNA biology 16, 340- 353. Moynahan, M.E., and Jasin, M. (2010). Mitotic homologous recombination maintains genomic stability and suppresses tumorigenesis. Nature reviews Molecular cell biology 11, 196-207. Munemasa, Y., Suzuki, T., Aizawa, K., Miyamoto, S., Imai, Y., Matsumura, T., Horikoshi, M., and Nagai, R. (2008). Promoter region-specific histone incorporation by the novel histone chaperone ANP32B and DNA-binding factor KLF5. Molecular and cellular biology 28, 1171-1181.

165

Munier, S., Rolland, T., Diot, C., Jacob, Y., and Naffakh, N. (2013). Exploration of binary virus-host interactions using an infectious protein complementation assay. Molecular & cellular proteomics : MCP 12, 2845-2855. Naito, T., Momose, F., Kawaguchi, A., and Nagata, K. (2007). Involvement of Hsp90 in assembly and nuclear import of influenza virus RNA polymerase subunits. Journal of virology 81, 1339-1349. Nencioni, L., De Chiara, G., Sgarbanti, R., Amatore, D., Aquilano, K., Marcocci, M.E., Serafino, A., Torcia, M., Cozzolino, F., Ciriolo, M.R., et al. (2009). Bcl-2 expression and p38MAPK activity in cells infected with influenza A virus: impact on virally induced apoptosis and viral replication. The Journal of biological chemistry 284, 16004-16015. Neumann, G., Castrucci, M.R., and Kawaoka, Y. (1997). Nuclear import and export of influenza virus nucleoprotein. Journal of virology 71, 9690-9700. Neumann, G., and Hobom, G. (1995). Mutational analysis of influenza virus promoter elements in vivo. The Journal of general virology 76 ( Pt 7), 1709-1717. Neumann, G., Hughes, M.T., and Kawaoka, Y. (2000). Influenza A virus NS2 protein mediates vRNP nuclear export through NES-independent interaction with hCRM1. The EMBO journal 19, 6751-6758. Nikitin, N., Petrova, E., Trifonova, E., and Karpova, O. (2014). Influenza Virus Aerosols in the Air and Their Infectiousness. Advances in Virology 2014, 859090. Nogales, A., and M, L.D. (2019). Host Single Nucleotide Polymorphisms Modulating Influenza A Virus Disease in Humans. Pathogens (Basel, Switzerland) 8. Noshi, T., Kitano, M., Taniguchi, K., Yamamoto, A., Omoto, S., Baba, K., Hashimoto, T., Ishida, K., Kushima, Y., Hattori, K., et al. (2018). In vitro characterization of baloxavir acid, a first-in-class cap- dependent endonuclease inhibitor of the influenza virus polymerase PA subunit. Antiviral research 160, 109-117. Nturibi, E., Bhagwat, A.R., Coburn, S., Myerburg, M.M., and Lakdawala, S.S. (2017). Intracellular Colocalization of Influenza Viral RNA and Rab11A Is Dependent upon Microtubule Filaments. Journal of virology 91. O'Neill, R.E., Jaskunas, R., Blobel, G., Palese, P., and Moroianu, J. (1995). Nuclear import of influenza virus RNA can be mediated by viral nucleoprotein and transport factors required for protein import. The Journal of biological chemistry 270, 22701-22704. O'Neill, R.E., Talon, J., and Palese, P. (1998). The influenza virus NEP (NS2 protein) mediates the nuclear export of viral ribonucleoproteins. The EMBO journal 17, 288-296. Obri, A., Ouararhni, K., Papin, C., Diebold, M.L., Padmanabhan, K., Marek, M., Stoll, I., Roy, L., Reilly, P.T., Mak, T.W., et al. (2014). ANP32E is a histone chaperone that removes H2A.Z from chromatin. Nature 505, 648-653. Okamoto, S., Amaishi, Y., Maki, I., Enoki, T., and Mineno, J. (2019). Highly efficient genome editing for single-base substitutions using optimized ssODNs with Cas9-RNPs. Scientific reports 9, 4811. Omoto, S., Speranzini, V., Hashimoto, T., Noshi, T., Yamaguchi, H., Kawai, M., Kawaguchi, K., Uehara, T., Shishido, T., Naito, A., et al. (2018). Characterization of influenza virus variants induced by treatment with the endonuclease inhibitor baloxavir marboxil. Scientific reports 8, 9633. Oner, A.F., Bay, A., Arslan, S., Akdeniz, H., Sahin, H.A., Cesur, Y., Epcacan, S., Yilmaz, N., Deger, I., Kizilyildiz, B., et al. (2006). Avian influenza A (H5N1) infection in eastern Turkey in 2006. The New England journal of medicine 355, 2179-2185. Park, Y.H., Chungu, K., Lee, S.B., Woo, S.J., Cho, H.Y., Lee, H.J., Rengaraj, D., Lee, J.H., Song, C.S., Lim, J.M., et al. (2020). Host-Specific Restriction of Avian Influenza Virus Caused by Differential Dynamics of ANP32 Family Members. The Journal of infectious diseases 221, 71-80. Peacock, T.P., Sheppard, C.M., Staller, E., and Barclay, W.S. (2019). Host Determinants of Influenza RNA Synthesis. Annual Review of Virology 6, 215-233. Peacock, T.P., Sheppard, C.M., Staller, E., Frise, R., Swann, O.C., Goldhill, D.H., Long, J.S., and Barclay, W.S. (2020a). Mammalian ANP32A and ANP32B proteins drive alternative avian influenza virus polymerase adaptations. bioRxiv, 2020.2009.2003.282384.

166

Peacock, T.P., Swann, O.C., Salvesen, H.A., Staller, E., Leung, P.B., Goldhill, D.H., Zhou, H., Lillico, S.G., Whitelaw, C.B.A., Long, J.S., et al. (2020b). Swine ANP32A Supports Avian Influenza Virus Polymerase. J Virol 94. Pearl, L.H., and Prodromou, C. (2006). Structure and mechanism of the Hsp90 molecular chaperone machinery. Annual review of biochemistry 75, 271-294. Peng, Q., Liu, Y., Peng, R., Wang, M., Yang, W., Song, H., Chen, Y., Liu, S., Han, M., Zhang, X., et al. (2019). Structural insight into RNA synthesis by influenza D polymerase. Nature microbiology 4, 1750-1759. Peng, X., Gralinski, L., Armour, C.D., Ferris, M.T., Thomas, M.J., Proll, S., Bradel-Tretheway, B.G., Korth, M.J., Castle, J.C., Biery, M.C., et al. (2010). Unique signatures of long noncoding RNA expression in response to virus infection and altered innate immune signaling. mBio 1. Petosa, C., Schoehn, G., Askjaer, P., Bauer, U., Moulin, M., Steuerwald, U., Soler-López, M., Baudin, F., Mattaj, I.W., and Müller, C.W. (2004). Architecture of CRM1/Exportin1 suggests how cooperativity is achieved during formation of a nuclear export complex. Molecular cell 16, 761-775. Picard, D. (2002). Heat-shock protein 90, a chaperone for folding and regulation. Cellular and molecular life sciences : CMLS 59, 1640-1648. Pichlmair, A., Schulz, O., Tan, C.P., Näslund, T.I., Liljeström, P., Weber, F., and Reis e Sousa, C. (2006). RIG-I-mediated antiviral responses to single-stranded RNA bearing 5'-phosphates. Science (New York, NY) 314, 997-1001. Pleschka, S., Wolff, T., Ehrhardt, C., Hobom, G., Planz, O., Rapp, U.R., and Ludwig, S. (2001). Influenza virus propagation is impaired by inhibition of the Raf/MEK/ERK signalling cascade. Nature cell biology 3, 301-305. Prabhu, S.S., Chakraborty, T.T., Kumar, N., and Banerjee, I. (2018). Association between IFITM3 rs12252 polymorphism and influenza susceptibility and severity: A meta-analysis. Gene 674, 70-79. Pumroy, R.A., Ke, S., Hart, D.J., Zachariae, U., and Cingolani, G. (2015). Molecular determinants for nuclear import of influenza A PB2 by importin α isoforms 3 and 7. Structure (London, England : 1993) 23, 374-384. Puschnik, A.S., Majzoub, K., Ooi, Y.S., and Carette, J.E. (2017). A CRISPR toolbox to study virus-host interactions. Nature reviews Microbiology 15, 351-364. Rajsbaum, R., Albrecht, R.A., Wang, M.K., Maharaj, N.P., Versteeg, G.A., Nistal-Villán, E., García- Sastre, A., and Gack, M.U. (2012). Species-specific inhibition of RIG-I ubiquitination and IFN induction by the influenza A virus NS1 protein. PLoS pathogens 8, e1003059. Ran, F.A., Hsu, P.D., Lin, C.Y., Gootenberg, J.S., Konermann, S., Trevino, A.E., Scott, D.A., Inoue, A., Matoba, S., Zhang, Y., et al. (2013a). Double nicking by RNA-guided CRISPR Cas9 for enhanced genome editing specificity. Cell 154, 1380-1389. Ran, F.A., Hsu, P.D., Wright, J., Agarwala, V., Scott, D.A., and Zhang, F. (2013b). Genome engineering using the CRISPR-Cas9 system. Nat Protoc 8, 2281-2308. Randolph, A.G., Yip, W.K., Allen, E.K., Rosenberger, C.M., Agan, A.A., Ash, S.A., Zhang, Y., Bhangale, T.R., Finkelstein, D., Cvijanovich, N.Z., et al. (2017). Evaluation of IFITM3 rs12252 Association With Severe Pediatric Influenza Infection. The Journal of infectious diseases 216, 14-21. Read, E.K., and Digard, P. (2010). Individual influenza A virus mRNAs show differential dependence on cellular NXF1/TAP for their nuclear export. The Journal of general virology 91, 1290-1301. Rehwinkel, J., Tan, C.P., Goubau, D., Schulz, O., Pichlmair, A., Bier, K., Robb, N., Vreede, F., Barclay, W., Fodor, E., et al. (2010). RIG-I detects viral genomic RNA during negative-strand RNA virus infection. Cell 140, 397-408. Reilly, P.T., Afzal, S., Gorrini, C., Lui, K., Bukhman, Y.V., Wakeham, A., Haight, J., Ling, T.W., Cheung, C.C., Elia, A.J., et al. (2011). Acidic nuclear phosphoprotein 32kDa (ANP32)B-deficient mouse reveals a hierarchy of ANP32 importance in mammalian development. Proc Natl Acad Sci U S A 108, 10243- 10248. Reilly, P.T., Yu, Y., Hamiche, A., and Wang, L. (2014). Cracking the ANP32 whips: important functions, unequal requirement, and hints at disease implications. Bioessays 36, 1062-1071.

167

Remy, I., and Michnick, S.W. (2006). A highly sensitive protein-protein interaction assay based on Gaussia luciferase. Nature methods 3, 977-979. Resa-Infante, P., Bonet, J., Thiele, S., Alawi, M., Stanelle-Bertram, S., Tuku, B., Beck, S., Oliva, B., and Gabriel, G. (2019). Alternative interaction sites in the influenza A virus nucleoprotein mediate viral escape from the importin-α7 mediated nuclear import pathway. The FEBS journal 286, 3374-3388. Resa-Infante, P., Jorba, N., Zamarreño, N., Fernández, Y., Juárez, S., and Ortín, J. (2008). The host- dependent interaction of alpha-importins with influenza PB2 polymerase subunit is required for virus RNA replication. PloS one 3, e3904. Resa-Infante, P., Paterson, D., Bonet, J., Otte, A., Oliva, B., Fodor, E., and Gabriel, G. (2015). Targeting Importin-α7 as a Therapeutic Approach against Pandemic Influenza Viruses. Journal of virology 89, 9010-9020. Resa-Infante, P., Thieme, R., Ernst, T., Arck, P.C., Ittrich, H., Reimer, R., and Gabriel, G. (2014). Importin-α7 is required for enhanced influenza A virus replication in the alveolar epithelium and severe lung damage in mice. Journal of virology 88, 8166-8179. Rio, D.C., Clark, S.G., and Tjian, R. (1985). A mammalian host-vector system that regulates expression and amplification of transfected genes by temperature induction. Science (New York, NY) 227, 23-28. Robb, N.C., Smith, M., Vreede, F.T., and Fodor, E. (2009). NS2/NEP protein regulates transcription and replication of the influenza virus RNA genome. The Journal of general virology 90, 1398-1407. Rodriguez, A., Pérez-González, A., and Nieto, A. (2007). Influenza virus infection causes specific degradation of the largest subunit of cellular RNA polymerase II. Journal of virology 81, 5315-5324. Saavedra, F., Rivera, C., Rivas, E., Merino, P., Garrido, D., Hernández, S., Forné, I., Vassias, I., Gurard- Levin, Z.A., Alfaro, I.E., et al. (2017). PP32 and SET/TAF-Iβ proteins regulate the acetylation of newly synthesized histone H4. Nucleic Acids Res 45, 11700-11710. Sander, J.D., and Joung, J.K. (2014). CRISPR-Cas systems for editing, regulating and targeting genomes. Nature biotechnology 32, 347-355. Satterly, N., Tsai, P.L., van Deursen, J., Nussenzveig, D.R., Wang, Y., Faria, P.A., Levay, A., Levy, D.E., and Fontoura, B.M. (2007). Influenza virus targets the mRNA export machinery and the nuclear pore complex. Proceedings of the National Academy of Sciences of the United States of America 104, 1853-1858. Schneider, R., Bannister, A.J., Weise, C., and Kouzarides, T. (2004). Direct binding of INHAT to H3 tails disrupted by modifications. The Journal of biological chemistry 279, 23859-23862. Schreiber, A., Boff, L., Anhlan, D., Krischuns, T., Brunotte, L., Schuberth, C., Wedlich-Söldner, R., Drexler, H., and Ludwig, S. (2020). Dissecting the mechanism of signaling-triggered nuclear export of newly synthesized influenza virus ribonucleoprotein complexes. Proceedings of the National Academy of Sciences of the United States of America 117, 16557-16566. Sediri, H., Schwalm, F., Gabriel, G., and Klenk, H.D. (2015). Adaptive mutation PB2 D701N promotes nuclear import of influenza vRNPs in mammalian cells. European journal of cell biology 94, 368-374. Seo, S.B., Macfarlan, T., McNamara, P., Hong, R., Mukai, Y., Heo, S., and Chakravarti, D. (2002). Regulation of histone acetylation and transcription by nuclear protein pp32, a subunit of the INHAT complex. The Journal of biological chemistry 277, 14005-14010. Seo, S.B., McNamara, P., Heo, S., Turner, A., Lane, W.S., and Chakravarti, D. (2001). Regulation of histone acetylation and transcription by INHAT, a human cellular complex containing the set oncoprotein. Cell 104, 119-130. Serna Martin, I., Hengrung, N., Renner, M., Sharps, J., Martínez-Alonso, M., Masiulis, S., Grimes, J.M., and Fodor, E. (2018). A Mechanism for the Activation of the Influenza Virus Transcriptase. Molecular cell 70, 1101-1110.e1104. Shapira, S.D., Gat-Viks, I., Shum, B.O., Dricot, A., de Grace, M.M., Wu, L., Gupta, P.B., Hao, T., Silver, S.J., Root, D.E., et al. (2009). A physical and regulatory map of host-influenza interactions reveals pathways in H1N1 infection. Cell 139, 1255-1267. Shen, H. (2009). UAP56- a key player with surprisingly diverse roles in pre-mRNA splicing and nuclear export. BMB reports 42, 185-188.

168

Shen, H., Zheng, X., Shen, J., Zhang, L., Zhao, R., and Green, M.R. (2008). Distinct activities of the DExD/H-box splicing factor hUAP56 facilitate stepwise assembly of the spliceosome. Genes & development 22, 1796-1803. Shen, J., Zhang, L., and Zhao, R. (2007). Biochemical characterization of the ATPase and helicase activity of UAP56, an essential pre-mRNA splicing and mRNA export factor. The Journal of biological chemistry 282, 22544-22550. Sherry, S.T., Ward, M.H., Kholodov, M., Baker, J., Phan, L., Smigielski, E.M., and Sirotkin, K. (2001). dbSNP: the NCBI database of genetic variation. Nucleic acids research 29, 308-311. Shimizu, K., Handa, H., Nakada, S., and Nagata, K. (1994). Regulation of influenza virus RNA polymerase activity by cellular and viral factors. Nucleic Acids Res 22, 5047-5053. Shinya, K., Makino, A., Ozawa, M., Kim, J.H., Sakai-Tagawa, Y., Ito, M., Le, Q.M., and Kawaoka, Y. (2009). Ostrich involvement in the selection of H5N1 influenza virus possessing mammalian-type amino acids in the PB2 protein. Journal of virology 83, 13015-13018. Smietanski, M., Werner, M., Purta, E., Kaminska, K.H., Stepinski, J., Darzynkiewicz, E., Nowotny, M., and Bujnicki, J.M. (2014). Structural analysis of human 2'-O-ribose methyltransferases involved in mRNA cap structure formation. Nat Commun 5, 3004. Sologuren, I., Martinez-Saavedra, M.T., Sole-Violan, J., de Borges de Oliveira, E., Jr., Betancor, E., Casas, I., Oleaga-Quintas, C., Martinez-Gallo, M., Zhang, S.Y., Pestano, J., et al. (2018). Lethal Influenza in Two Related Adults with Inherited GATA2 Deficiency. Journal of clinical immunology 38, 513-526. Staeheli, P., Grob, R., Meier, E., Sutcliffe, J.G., and Haller, O. (1988). Influenza virus-susceptible mice carry Mx genes with a large deletion or a nonsense mutation. Molecular and cellular biology 8, 4518- 4523. Staller, E., Baillon, L., Frise, R., Peacock, T.P., Sheppard, C.M., Sancho-Shimizu, V., and Barclay, W. (2020a). A rare variant in ANP32B impairs influenza virus replication in human cells. bioRxiv, 2020.2004.2006.027482. Staller, E., Baillon, L., Frise, R., Peacock, T.P., Sheppard, C.M., Sancho-Shimizu, V., and Barclay, W.S. (2020b). A rare variant in Anp32B impairs influenza virus replication in human cells. bioRxiv, 2020.2004.2006.027482. Staller, E., Sheppard, C.M., Neasham, P.J., Mistry, B., Peacock, T.P., Goldhill, D.H., Long, J.S., and Barclay, W.S. (2019). ANP32 Proteins Are Essential for Influenza Virus Replication in Human Cells. Journal of virology 93. Stewart, M. (2010). Nuclear export of mRNA. Trends in biochemical sciences 35, 609-617. Stone, E.M., Yamano, H., Kinoshita, N., and Yanagida, M. (1993). Mitotic regulation of protein phosphatases by the fission yeast sds22 protein. Current Biology 3, 13-26. Subbarao, E.K., London, W., and Murphy, B.R. (1993). A single amino acid in the PB2 gene of influenza A virus is a determinant of host range. J Virol 67, 1761-1764. Sugiyama, K., Kawaguchi, A., Okuwaki, M., and Nagata, K. (2015). pp32 and APRIL are host cell- derived regulators of influenza virus RNA synthesis from cRNA. Elife 4. Swale, C., Da Costa, B., Sedano, L., Garzoni, F., McCarthy, A.A., Berger, I., Bieniossek, C., Ruigrok, R.W.H., Delmas, B., and Crépin, T. (2020). X-ray Structure of the Human Karyopherin RanBP5, an Essential Factor for Influenza Polymerase Nuclear Trafficking. Journal of molecular biology 432, 3353-3359. Swale, C., Monod, A., Tengo, L., Labaronne, A., Garzoni, F., Bourhis, J.M., Cusack, S., Schoehn, G., Berger, I., Ruigrok, R.W., et al. (2016). Structural characterization of recombinant IAV polymerase reveals a stable complex between viral PA-PB1 heterodimer and host RanBP5. Scientific reports 6, 24727. Takashita, E. (2020). Influenza Polymerase Inhibitors: Mechanisms of Action and Resistance. Cold Spring Harbor perspectives in medicine.

169

Taliun, D., Harris, D.N., Kessler, M.D., Carlson, J., Szpiech, Z.A., Torres, R., Taliun, S.A.G., Corvelo, A., Gogarten, S.M., Kang, H.M., et al. (2019). Sequencing of 53,831 diverse genomes from the NHLBI TOPMed Program. bioRxiv, 563866. Tarendeau, F., Boudet, J., Guilligay, D., Mas, P.J., Bougault, C.M., Boulo, S., Baudin, F., Ruigrok, R.W., Daigle, N., Ellenberg, J., et al. (2007). Structure and nuclear import function of the C-terminal domain of influenza virus polymerase PB2 subunit. Nature structural & molecular biology 14, 229-233. Tarus, B., Bakowiez, O., Chenavas, S., Duchemin, L., Estrozi, L.F., Bourdieu, C., Lejal, N., Bernard, J., Moudjou, M., Chevalier, C., et al. (2012). Oligomerization paths of the nucleoprotein of influenza A virus. Biochimie 94, 776-785. Te Velthuis, A.J., and Fodor, E. (2016). Influenza virus RNA polymerase: insights into the mechanisms of viral RNA synthesis. Nat Rev Microbiol 14, 479-493. Thierry, E., Guilligay, D., Kosinski, J., Bock, T., Gaudon, S., Round, A., Pflug, A., Hengrung, N., El Omari, K., Baudin, F., et al. (2016). Influenza Polymerase Can Adopt an Alternative Configuration Involving a Radical Repacking of PB2 Domains. Molecular cell 61, 125-137. Tochio, N., Umehara, T., Munemasa, Y., Suzuki, T., Sato, S., Tsuda, K., Koshiba, S., Kigawa, T., Nagai, R., and Yokoyama, S. (2010). Solution structure of histone chaperone ANP32B: interaction with core histones H3-H4 through its acidic concave domain. J Mol Biol 401, 97-114. Tong, S., Li, Y., Rivailler, P., Conrardy, C., Castillo, D.A., Chen, L.M., Recuenco, S., Ellison, J.A., Davis, C.T., York, I.A., et al. (2012). A distinct lineage of influenza A virus from bats. Proceedings of the National Academy of Sciences of the United States of America 109, 4269-4274. Tong, S., Zhu, X., Li, Y., Shi, M., Zhang, J., Bourgeois, M., Yang, H., Chen, X., Recuenco, S., Gomez, J., et al. (2013). New world bats harbor diverse influenza A viruses. PLoS pathogens 9, e1003657. Trevejo, J.M., Asmal, M., Vingerhoets, J., Polo, R., Robertson, S., Jiang, Y., Kieffer, T.L., and Leopold, L. (2018). Pimodivir treatment in adult volunteers experimentally inoculated with live influenza virus: a Phase IIa, randomized, double-blind, placebo-controlled study. Antiviral therapy 23, 335-344. Uehara, T., Hayden, F.G., Kawaguchi, K., Omoto, S., Hurt, A.C., De Jong, M.D., Hirotsu, N., Sugaya, N., Lee, N., Baba, K., et al. (2020). Treatment-Emergent Influenza Variant Viruses With Reduced Baloxavir Susceptibility: Impact on Clinical and Virologic Outcomes in Uncomplicated Influenza. The Journal of infectious diseases 221, 346-355. Vale-Costa, S., Alenquer, M., Sousa, A.L., Kellen, B., Ramalho, J., Tranfield, E.M., and Amorim, M.J. (2016). Influenza A virus ribonucleoproteins modulate host recycling by competing with Rab11 effectors. Journal of cell science 129, 1697-1710. van Riet, S., Ninaber, D.K., Mikkers, H.M.M., Tetley, T.D., Jost, C.R., Mulder, A.A., Pasman, T., Baptista, D., Poot, A.A., Truckenmüller, R., et al. (2020). In vitro modelling of alveolar repair at the air-liquid interface using alveolar epithelial cells derived from human induced pluripotent stem cells. Sci Rep 10, 5499. Vasin, A.V., Temkina, O.A., Egorov, V.V., Klotchenko, S.A., Plotnikova, M.A., and Kiselev, O.I. (2014). Molecular mechanisms enhancing the proteome of influenza A viruses: an overview of recently discovered proteins. Virus research 185, 53-63. Vreede, F.T., Chan, A.Y., Sharps, J., and Fodor, E. (2010). Mechanisms and functional implications of the degradation of host RNA polymerase II in influenza virus infected cells. Virology 396, 125-134. Vreede, F.T., and Fodor, E. (2010). The role of the influenza virus RNA polymerase in host shut-off. Virulence 1, 436-439. Vreede, F.T., Jung, T.E., and Brownlee, G.G. (2004). Model suggesting that replication of influenza virus is regulated by stabilization of replicative intermediates. Journal of virology 78, 9568-9572. Wakai, C., Iwama, M., Mizumoto, K., and Nagata, K. (2011). Recognition of cap structure by influenza B virus RNA polymerase is less dependent on the methyl residue than recognition by influenza A virus polymerase. Journal of virology 85, 7504-7512. Walker, A.P., and Fodor, E. (2019). Interplay between Influenza Virus and the Host RNA Polymerase II Transcriptional Machinery. Trends in microbiology 27, 398-407.

170

Wallace, S.S. (2014). Base excision repair: a critical player in many games. DNA Repair (Amst) 19, 14- 26. Walter, K., Min, J.L., Huang, J., Crooks, L., Memari, Y., McCarthy, S., Perry, J.R., Xu, C., Futema, M., Lawson, D., et al. (2015). The UK10K project identifies rare variants in health and disease. Nature 526, 82-90. Wandzik, J.M., Kouba, T., and Cusack, S. (2020). Structure and Function of Influenza Polymerase. Cold Spring Harbor perspectives in medicine. Wang, J., and Cen, S. (2020). Roles of lncRNAs in influenza virus infection. Emerging microbes & infections 9, 1407-1414. Wang, J., Wang, Y., Zhou, R., Zhao, J., Zhang, Y., Yi, D., Li, Q., Zhou, J., Guo, F., Liang, C., et al. (2018). Host Long Noncoding RNA lncRNA-PAAN Regulates the Replication of Influenza A Virus. Viruses 10. Wang, J., Zhang, Y., Li, Q., Zhao, J., Yi, D., Ding, J., Zhao, F., Hu, S., Zhou, J., Deng, T., et al. (2019a). Influenza Virus Exploits an Interferon-Independent lncRNA to Preserve Viral RNA Synthesis through Stabilizing Viral RNA Polymerase PB1. Cell reports 27, 3295-3304.e3294. Wang, P., Palese, P., and O'Neill, R.E. (1997). The NPI-1/NPI-3 (karyopherin alpha) binding site on the influenza a virus nucleoprotein NP is a nonconventional nuclear localization signal. Journal of virology 71, 1850-1856. Wang, P., Xu, J., Wang, Y., and Cao, X. (2017). An interferon-independent lncRNA promotes viral replication by modulating cellular metabolism. Science (New York, NY) 358, 1051-1055. Wang, R., and Brattain, M.G. (2007). The maximal size of protein to diffuse through the nuclear pore is larger than 60kDa. FEBS letters 581, 3164-3170. Wang, W., Cui, Z.Q., Han, H., Zhang, Z.P., Wei, H.P., Zhou, Y.F., Chen, Z., and Zhang, X.E. (2008). Imaging and characterizing influenza A virus mRNA transport in living cells. Nucleic Acids Res 36, 4913-4928. Wang, Y., Zhang, H., Na, L., Du, C., Zhang, Z., Zheng, Y.H., and Wang, X. (2019b). ANP32A and ANP32B are key factors in the Rev-dependent CRM1 pathway for nuclear export of HIV-1 unspliced mRNA. The Journal of biological chemistry 294, 15346-15357. Watanabe, K., Takizawa, N., Katoh, M., Hoshida, K., Kobayashi, N., and Nagata, K. (2001). Inhibition of nuclear export of ribonucleoprotein complexes of influenza virus by leptomycin B. Virus research 77, 31-42. Watanabe, T., Kawakami, E., Shoemaker, J.E., Lopes, T.J., Matsuoka, Y., Tomita, Y., Kozuka-Hata, H., Gorai, T., Kuwahara, T., Takeda, E., et al. (2014). Influenza virus-host interactome screen as a platform for antiviral drug development. Cell host & microbe 16, 795-805. Watanabe, T., Watanabe, S., and Kawaoka, Y. (2010). Cellular networks involved in the influenza virus life cycle. Cell host & microbe 7, 427-439. Weber, M., Sediri, H., Felgenhauer, U., Binzen, I., Bänfer, S., Jacob, R., Brunotte, L., García-Sastre, A., Schmid-Burgk, J.L., Schmidt, T., et al. (2015). Influenza virus adaptation PB2-627K modulates nucleocapsid inhibition by the pathogen sensor RIG-I. Cell host & microbe 17, 309-319. Wellington, D., Laurenson-Schafer, H., Abdel-Haq, A., and Dong, T. (2019). IFITM3: How genetics influence influenza infection demographically. Biomedical journal 42, 19-26. Whittaker, G., Bui, M., and Helenius, A. (1996). Nuclear trafficking of influenza virus ribonuleoproteins in heterokaryons. Journal of virology 70, 2743-2756. Winterling, C., Koch, M., Koeppel, M., Garcia-Alcalde, F., Karlas, A., and Meyer, T.F. (2014). Evidence for a crucial role of a host non-coding RNA in influenza A virus replication. RNA biology 11, 66-75. Wisskirchen, C., Ludersdorfer, T.H., Müller, D.A., Moritz, E., and Pavlovic, J. (2011). The cellular RNA helicase UAP56 is required for prevention of double-stranded RNA formation during influenza A virus infection. Journal of virology 85, 8646-8655. Wu, W.W., Weaver, L.L., and Panté, N. (2007). Ultrastructural analysis of the nuclear localization sequences on influenza A ribonucleoprotein complexes. Journal of molecular biology 374, 910-916. Yon, J., and Fried, M. (1989). Precise gene fusion by PCR. Nucleic Acids Research 17, 4895-4895.

171

York, A., Hengrung, N., Vreede, F.T., Huiskonen, J.T., and Fodor, E. (2013). Isolation and characterization of the positive-sense replicative intermediate of a negative-strand RNA virus. Proc Natl Acad Sci U S A 110, E4238-4245. Yount, J.S., Moltedo, B., Yang, Y.Y., Charron, G., Moran, T.M., Lopez, C.B., and Hang, H.C. (2010). Palmitoylome profiling reveals S-palmitoylation-dependent antiviral activity of IFITM3. Nature chemical biology 6, 610-614. Zaborowska, J., Egloff, S., and Murphy, S. (2016). The pol II CTD: new twists in the tail. Nature structural & molecular biology 23, 771-777. Zhang, H., Li, H., Wang, W., Wang, Y., Han, G.Z., Chen, H., and Wang, X. (2020). A unique feature of swine ANP32A provides susceptibility to avian influenza virus infection in pigs. PLoS pathogens 16, e1008330. Zhang, H., Zhang, Z., Wang, Y., Wang, M., Wang, X., Zhang, X., Ji, S., Du, C., Chen, H., and Wang, X. (2019a). Fundamental Contribution and Host Range Determination of ANP32A and ANP32B in Influenza A Virus Polymerase Activity. J Virol 93. Zhang, K., Xie, Y., Muñoz-Moreno, R., Wang, J., Zhang, L., Esparza, M., García-Sastre, A., Fontoura, B.M.A., and Ren, Y. (2019b). Structural basis for influenza virus NS1 protein block of mRNA nuclear export. Nature microbiology 4, 1671-1679. Zhang, Q. (2020). Human genetics of life-threatening influenza pneumonitis. Hum Genet. Zhou, J., To, K.K., Dong, H., Cheng, Z.S., Lau, C.C., Poon, V.K., Fan, Y.H., Song, Y.Q., Tse, H., Chan, K.H., et al. (2012). A functional variation in CD55 increases the severity of 2009 pandemic H1N1 influenza A virus infection. The Journal of infectious diseases 206, 495-503.

172