# EMERGING VIRUSES: HOST IMMUNITY AND NOVEL THERAPEUTIC INTERVENTIONS

EDITED BY : Alan Chen-Yu Hsu, Anna Smed-Sörensen, Ding Yuan Oh and Hiroyuki Oshiumi PUBLISHED IN : Frontiers in Immunology

#### Frontiers Copyright Statement

© Copyright 2007-2019 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use. ISSN 1664-8714 ISBN 978-2-88945-742-7 DOI 10.3389/978-2-88945-742-7

### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

## Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

## Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

Frontiers in Immunology 1 February 2019 | Emerging Viruses

# EMERGING VIRUSES: HOST IMMUNITY AND NOVEL THERAPEUTIC INTERVENTIONS

Topic Editors:

Alan Chen-Yu Hsu, The University of Newcastle and Hunter Medical Research Institute, Australia Anna Smed-Sörensen, Karolinska Institutet, Karolinska University Hospital, Sweden Ding Yuan Oh, Federation University, Australia Hiroyuki Oshiumi, Kumamoto University, Japan

Cover image: Kateryna Kon/Shutterstock.com

We acknowledge the initiation and support of this Research Topic by the International Union of Immunological Societies (IUIS). We hereby state publicly that the IUIS has had no editorial input in articles included in this Research Topic, thus ensuring that all aspects of this Research Topic are evaluated objectively, unbiased by any specific policy or opinion of the IUIS.

Citation: Hsu, A. C.-Y., Smed-Sörensen, A., Oh, D. Y., Oshiumi, H., eds. (2019). Emerging Viruses: Host Immunity and Novel Therapeutic Interventions. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-742-7

# Table of Contents

*05 Editorial: Emerging Viruses: Host Immunity and Novel Therapeutic Interventions*

Anna Smed-Sörensen, Ding Yuan Oh, Hiroyuki Oshiumi and Alan Chen-Yu Hsu

# INFLUENZA VIRUS


Paul S. Foster and Ming Yang

*91 The Hurdles From Bench to Bedside in the Realization and Implementation of a Universal Influenza Vaccine* Sophie A. Valkenburg, Nancy H. L. Leung, Maireid B. Bull, Li-meng Yan,

Athena P. Y. Li, Leo L. M. Poon and Benjamin J. Cowling


# CORONAVIRUS

*154 Pathogenicity and Viral Shedding of MERS-CoV in Immunocompromised Rhesus Macaques*

Joseph Prescott, Darryl Falzarano, Emmie de Wit, Kath Hardcastle, Friederike Feldmann, Elaine Haddock, Dana Scott, Heinz Feldmann and Vincent Jacobus Munster

# FLAVIVIRUS

# *163 Flavivirus Receptors: Diversity, Identity, and Cell Entry*

Mathilde Laureti, Divya Narayanan, Julio Rodriguez-Andres, John K. Fazakerley and Lukasz Kedzierski

*174 Cell-Mediated Immune Responses and Immunopathogenesis of Human Tick-Borne Encephalitis Virus-Infection*

Kim Blom, Angelica Cuapio, J. Tyler Sandberg, Renata Varnaite, Jakob Michaëlsson, Niklas K. Björkström, Johan K. Sandberg, Jonas Klingström, Lars Lindquist, Sara Gredmark Russ and Hans-Gustaf Ljunggren

# EBOLA VIRUS

*184 Advances in Designing and Developing Vaccines, Drugs, and Therapies to Counter Ebola Virus*

Kuldeep Dhama, Kumaragurubaran Karthik, Rekha Khandia, Sandip Chakraborty, Ashok Munjal, Shyma K. Latheef, Deepak Kumar, Muthannan Andavar Ramakrishnan, Yashpal Singh Malik, Rajendra Singh, Satya Veer Singh Malik, Raj Kumar Singh and Wanpen Chaicumpa

# HENTAVIRUS

*211 Neutrophil Activation in Acute Hemorrhagic Fever With Renal Syndrome is Mediated by Hantavirus-Infected Microvascular Endothelial Cells* Tomas Strandin, Satu Mäkelä, Jukka Mustonen and Antti Vaheri

# Editorial: Emerging Viruses: Host Immunity and Novel Therapeutic Interventions

Anna Smed-Sörensen1†, Ding Yuan Oh2†, Hiroyuki Oshiumi 3† and Alan Chen-Yu Hsu4,5 \*

<sup>1</sup> Division of Immunology and Allergy, Department of Medicine Solna, Karolinska Institutet, Karolinska University Hospital, Stockholm, Sweden, <sup>2</sup> School of Health and Life Sciences, Federation University, Gippsland, VIC, Australia, <sup>3</sup> Department of Immunology, Faculty of Life Sciences, Kumamoto University, Kumamoto, Japan, <sup>4</sup> Viruses, Infections/Immunity, Vaccines & Asthma, Hunter Medical Research Institute, Newcastle, NSW, Australia, <sup>5</sup> Faculty of Health & Medicine, Priority Research Centre for Healthy Lungs, The University of Newcastle, Newcastle, NSW, Australia

Keywords: influenza virus, coronavirus, MERS coronavirus, Ebola, Flavivirus, Hantavirus, vaccines, antiviral drugs

**Emerging Viruses: Host Immunity and Novel Therapeutic Interventions**

### **Editorial on the Research Topic**

#### Edited and reviewed by:

Ian Marriott, University of North Carolina at Charlotte, United States

\*Correspondence: Alan Chen-Yu Hsu alan.hsu@newcastle.edu.au

†These authors share first authorship

#### Specialty section:

This article was submitted to Microbial Immunology, a section of the journal Frontiers in Immunology

Received: 16 October 2018 Accepted: 16 November 2018 Published: 30 November 2018

#### Citation:

Smed-Sörensen A, Oh DY, Oshiumi H and Hsu AC-Y (2018) Editorial: Emerging Viruses: Host Immunity and Novel Therapeutic Interventions. Front. Immunol. 9:2828. doi: 10.3389/fimmu.2018.02828 The inevitable emergence of novel infectious viruses and their unpredictable pathogenicity, transmissibility, and pandemic potentials are a major threat to human health. In particular, highly pathogenic influenza A viruses (IAVs), severe acute respiratory syndrome- and Middle East respiratory syndrome-coronavirus (SARS-CoV; MERS-CoV), Ebola virus (EBOV), and mosquitoborne Zika virus (ZIKV; Flavivirus) have attracted the world's attention due to their high pathogenicity, transmission and high mortality. While these viruses are mostly found in animals they can cause diseases and death in humans (zoonosis) when transmitted via close contact. Transmission of these pathogens is likely worsened by globalization and close quarter living in densely populated areas, all of which promotes rapid virus evolution and pandemic potential in humans. Significant research has resulted in deeper understanding of the virology, virus-host interactions, and disease pathology. These investigations have led to the development of vaccines and antiviral drugs, particularly for IAVs, and experimental vaccines for EBOV. Nevertheless, rapid and frequent virus mutations/reassortment often render current therapeutics ineffective, highlighting the limitations of our current therapeutic design strategies. Development of novel prevention and treatment options against these continuously mutating viruses must be explored before the next pandemic occurs.

In this Research Topic, a series of articles provides comprehensive insights on the current view of the virology, innate immune responses, and novel therapeutics to IAVs, MERS-CoV, EBOV, and flavivirus in experimental settings as well as in clinical trials.

In original research articles, Westenius et al. demonstrate that the highly pathogenic avian IAV H5N1 but not H3N2 and H7N9 virus replicate efficiently in primary human dendritic cells (DCs) and macrophages despite the robust induction of antiviral cytokines. This indicates an unusual level of resistance to host antiviral responses by the IAV H5N1 subtype. Prescott et al. show that MERS-CoV infection leads to much higher viral replication in the immuno-compromised rhesus macaque model, although this is accompanied by milder pathology in airways compared with non-immunocompromised control animals. This indicates that MERS-CoV infection in healthy individuals causes severe pathological changes with increased inflammatory response and cellular infiltrates in the airways. This is consistent with the finding that MERS-CoV-infected human patients have increased numbers of neutrophils and macrophages in their bronchial lavage fluid (Prescott et al.). Strandin et al. also show that the levels of neutrophils and the pro-inflammatory cytokine IL-8 are substantially higher in the blood of patients with hantavirus infection-mediated haemorrhagic fever with renal syndrome (HFRS). Neutrophil extracellular trap (NET) activation is evident and is the result of hantavirus-infected microvascular endothelial cells, indicating the importance of neutrophils in the disease pathology driven by both MERS-CoV and hantavirus infections.

This Research Topic also features a number of Review Articles on IAVs, flaviviruses, and EBOV, innate immune responses to infections, and novel therapeutic strategies that are currently in experimental phase or human clinical trials. Dou et al. provide a detailed review on IAV entry, viral replication, viral assembly and budding process, while Horman et al. review the IAV fitness, clinical manifestation of the disease, pathogenesis of highly pathogenic IAV infections. Hsu reviews the essential mutations and IAV virulence factors that are important in IAV transmission, inflammatory cytokine storm and efficient suppression of host antiviral response. Immune cells, such as macrophages and DCs are important in the immediate control of viral replication in the airways and also in establishing appropriate adaptive immune response for efficient clearance of the virus. Vangeti et al. discuss how these immune cells contribute to increased inflammation and severe disease caused by IAV. The prolonged inflammation and severe pathologies in the airways are not only observed with IAV, but also with MERS-CoV and hantavirus infections. Micro-RNAs (miRNAs; miRs) are a novel class of immunoregulators that have been shown to be involved in innate immune responses. Nguyen et al. review several miRNAs that are highly induced by IAVs and directly promotes nuclear-factor-kappa-B (NF-κB)-mediated inflammatory response.

Flavivirus, such as Dengue virus (DENV) and recently emerged ZIKV are mosquito-borne infectious pathogens. Laureti et al. critically review important Flavivirus species, including DENV, Japanese encephalitis virus (JEV), ZIKV, and yellow fever virus (YFV), their binding receptor diversity and virus entry mechanisms. Blom et al.review the innate immune responses and immuno-pathologies induced by these pathogens.

In terms of therapeutics, virus-targeted strategies remain a popular approach. Valkenburg et al. review the criteria, strategies, and obstacles in the development of universal influenza vaccines, and the requirement to increase the strength and duration of vaccine-induced immune responses. In addition to vaccines, Davidson reviews the potentials of several IAV haemagglutinin (HA)-targeted monoclonal antibodies and viral polymerase inhibitory compounds that show promising therapeutic effects in pre-clinical or clinical trials. Lee et al. review DNA vaccines carrying various chimeric fusion proteins of conserved regions of IAV structural proteins and their effectiveness in inducing cross-reactive antibody response to different subtypes of IAVs. While universal IAV vaccines are still in experimental/clinical phase, vaccines targeting one flavivirus species have been shown to induce cross-reactive response against another viral species. Blom et al. review T cell cross-reactivity induced by JEV- and YFV-vaccine to DENV and ZIKA, respectively, indicating the potential use of JEV or YFV vaccine as protective therapeutics against ZIKA.

Viruses, such as IAVs undergo rapid virus mutations that often render vaccines and antiviral drugs less effective. Alternative host-targeted approaches are also extensively reviewed in this Research Topic as strategies to improve anti-viral therapy. HA cleavage and activation by host proteases is a critical step to rendering newly made IAV particles infectious. Yip et al. review a number of clinically used protease inhibitors currently used for diseases, such as liver fibrosis and cancer that also inhibit HA activation in in vitro and/or in vivo models. This highlights the potential of repurposing these compounds as antivirals drugs against IAVs. Hsu also reviews a number of peptide-based small molecules that inhibit HA-mediated viral internalization and reduce infection, and many of which are currently in the experimental phase of testing and in human clinical trials.

As disease pathologies are mostly driven by exaggerated immune responses, immuno-modulatory molecules have also been discussed as potential therapeutics to reduce tissue damage. Nguyen et al. review various miRNA inhibitors shown to directly suppress IAV replication, as well as those that reduce IAVinduced inflammatory cytokine storm in in vivo models. As an exaggerated inflammatory response appears to be a common phenomenon driven by most of the infectious viruses described here, they may also be applicable to other viral infections, such as MERS-CoV as treatments. Antiviral responses are critical in the immediate control of viral replication and are induced by the binding of host pattern recognition receptors to viral RNAs. Viruses, such as IAVs and EBOV produce virulence factors that inhibit the production of antiviral cytokines (Hsu, Dhama). Yong et al. review the use of synthetic virus RNA analogs and small molecule modulators as pan-antiviral drugs and vaccine adjuvants that boost antiviral responses against RNA viruses, such as IAVs or flavivirus.

The recent Ebola epidemics in Africa during 2014–2016 and in 2018 have raised serious concerns of EBOV infection as a global health threat due to its high mortality rate. Dhama et al. not only review the general virology of EBOV and disease progression, but also discuss current progress in the development of virus-, DNA-, and plant-based vaccines and treatment-based therapeutics that are urgently needed to prevent or reduce EBOV-mediated disease and mortality.

Collectively, this Research Topic highlights the ease in which viruses are able to cause severe disease, and the complexities of virus-host interactions that impact both disease pathology and outcome. The knowledge acquired from the articles contained within this special issue may lead to the development of more specific peptide-based antiviral agents, monoclonal antibodies and novel vaccines that protect against infections in the future. Synthetic- and host-RNA-based immuno-modulatory compounds may act as potential treatment that reduce symptoms and disease. As these are host-targeted they may be suited for multiple viral-induced diseases. This area of research is absolutely essential and is urgently required in preparation of future pandemics.

We wish to convey our appreciation to all the authors who have participated in this Research Topic and the reviewers for their insightful comments.

# AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

# ACKNOWLEDGMENTS

We wish to convey our appreciation to all the authors who have participated in this Research Topic and the reviewers for their insightful comments.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Smed-Sörensen, Oh, Oshiumi and Hsu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Influenza A Virus Cell Entry, Replication, Virion Assembly and Movement

### *Dan Dou, Rebecca Revol, Henrik Östbye, Hao Wang and Robert Daniels\**

*Department of Biochemistry and Biophysics, Stockholm University, Stockholm, Sweden*

Influenza viruses replicate within the nucleus of the host cell. This uncommon RNA virus trait provides influenza with the advantage of access to the nuclear machinery during replication. However, it also increases the complexity of the intracellular trafficking that is required for the viral components to establish a productive infection. The segmentation of the influenza genome makes these additional trafficking requirements especially challenging, as each viral RNA (vRNA) gene segment must navigate the network of cellular membrane barriers during the processes of entry and assembly. To accomplish this goal, influenza A viruses (IAVs) utilize a combination of viral and cellular mechanisms to coordinate the transport of their proteins and the eight vRNA gene segments in and out of the cell. The aim of this review is to present the current mechanistic understanding for how IAVs facilitate cell entry, replication, virion assembly, and intercellular movement, in an effort to highlight some of the unanswered questions regarding the coordination of the IAV infection process.

Keywords: influenza A virus, viral ribonucleoprotein, hemagglutinin, viral entry mechanism, viral envelope proteins, HA and NA, viral replication, neuraminidase

# INFLUENZA VIRUSES

Influenza viruses belong to the *Orthomyxoviridae* family and are classified as either type A, B, C, or the recently identified type D (1, 2). Influenza A viruses (IAVs) and type B viruses (IBVs) contain 8, negative-sense, single-stranded viral RNA (vRNA) gene segments (**Figure 1A**) (3, 4), which encode transcripts for 10 essential viral proteins, as well as several strain-dependent accessory proteins (**Figure 1B**). In comparison, influenza type C and D viruses only possess seven vRNA gene segments, as the hemagglutinin–esterase fusion protein vRNA replaces the hemagglutinin (HA or H) and the neuraminidase (NA or N) vRNAs (1, 2). IAVs will be the main focus of this review since they are the primary agents responsible for influenza pandemics, and a major contributor to the annual influenza epidemics in the human population (5).

The natural reservoir for IAVs is wild aquatic birds, but they commonly infect other species, including humans, and have even been isolated from penguins in Antarctica (12–15). The ability to adapt to multiple species is a major reason why IAVs are more diverse than IBVs, which are essentially exclusive to humans. Despite the host-range differences, many similarities do exist between these two viruses. Both possess a host-derived lipid membrane, referred to as an envelope, which is decorated on the surface with the viral membrane proteins HA, NA, and to a lesser extent the matrix 2 (M2) protein (**Figure 1C**) (16–18). The envelope is supported underneath by the matrix 1 (M1) protein, and inside, the eight vRNAs are found as individual viral ribonucleoprotein (vRNP) complexes (**Figure 1C**, bottom). Each vRNP is comprised of a vRNA that is wrapped around numerous

#### *Edited by:*

*Alan Chen-Yu Hsu, University of Newcastle, Australia*

# *Reviewed by:*

*Bernard A. P. Lafont, National Institute of Allergy and Infectious Diseases (NIAID), United States Alan G. Goodman, Washington State University, United States Julie McAuley, University of Melbourne, Australia*

> *\*Correspondence: Robert Daniels robertd@dbb.su.se*

#### *Specialty section:*

*This article was submitted to Microbial Immunology, a section of the journal Frontiers in Immunology*

*Received: 17 April 2018 Accepted: 26 June 2018 Published: 20 July 2018*

#### *Citation:*

*Dou D, Revol R, Östbye H, Wang H and Daniels R (2018) Influenza A Virus Cell Entry, Replication, Virion Assembly and Movement. Front. Immunol. 9:1581. doi: 10.3389/fimmu.2018.01581*

**8**

cap, black lines denote the 10–13 nucleotide, host-derived primers that are obtained by the cap-snatching mechanism of the viral polymerase. A(n) corresponds to the 3′ poly-A tail produced by reiterative stuttering of the viral polymerase. The smaller mRNAs (empty boxes) represent transcripts that encode nonessential accessory proteins found in many strains, whereas those that are less prevalent (PB2-S1, M42, and NS3) are not illustrated (6–11). (C) Diagram of an influenza A or B virus. The viral membrane proteins HA, NA, and M2 are shown, along with the eight viral ribonucleoproteins (vRNPs), and the matrix protein M1 that supports the viral envelope. To highlight the vRNP components, the illustration beneath the virus is not to scale. A single vRNA gene segment is shown wrapped around multiple nucleoprotein (NP) copies with the conserved promoter regions in the 5′ and 3′ UTRs forming a helical hairpin, which is bound by a single heterotrimeric viral RNA-dependent RNA polymerase (PB1, PB2, and PA). (D) Top view of an influenza virus cross-section showing the vRNP "1 + 7" configuration. vRNPs are depicted with black circles as it is not known if the positioning of a particular vRNP is conserved or interchangeable.

copies of the viral nucleoprotein (NP) and bound by a single copy of the heterotrimeric viral polymerase, consisting of PB1, PB2, and PA (19–21). The polymerase binds the vRNAs at a helical hairpin that results from the base pairing between the conserved semi-complimentary 5′ and 3′ ends (21–23).

Morphologically, IAVs can either form spheres with a diameter of ~100 nm or filaments that can reach up to 20 µm in length [reviewed in Ref. (24)]. However, upon passaging in eggs, or MDCK cells, the filamentous form is generally lost (25, 26). Several studies have attributed the morphology change to M1, presumably through its function in supporting the envelope (27–29). Regardless of the virion shape, HA is the most abundant viral envelope protein, followed by NA, and M2 (30). Recent work has shown that the viral envelope also contains host membrane proteins (30, 31). These proteins are likely recruited based on the lipid composition at the plasma membrane budding site, which can differ between cell types (32, 33). Through possible interactions with each other and M1, the eight vRNPs typicallly form a 1 + 7 configuration inside the virus (**Figure 1D**) (34, 35). The 1 + 7 configuration may have a mechanistic function, as it is also conserved in type C and D viruses that only possess 7 vRNAs (36). Further supporting the mechanistic concept, it was recently shown that IAVs can package cellular ribosomal RNA (as a vRNP) when one of the vRNAs is made unavailable (37), possibly explaining how type C and D viruses acquire their "eighth" vRNA.

The classification of IAVs into subtypes is based on the genetic and antigenic properties of the surface antigens HA and NA, which mediate viral entry and release, respectively (17, 18). To date, 16 HA (H1-16) and 9 NA subtypes (N1-9) have been found in IAVs isolated from aquatic birds (13). Two additional subtypes for HA (H17 and H18) and NA (N10 and N11) have recently been identified in bats (38, 39), but in contrast to the HA and NA subtypes from the more traditional avian IAVs, these do not appear to recognize sialic acid (SA) (40–42). Despite the numerous possible subtype combinations, only three have consistently persisted in the human population, causing the following pandemics in the process: 1918 and 2009 (H1N1), 1957 (H2N2), and 1968 (H3N2) (43). Currently, only the H1N1 and H3N2 subtypes, as well as the two antigenically distinct IBV lineages (Victoria and Yamagata), are endemic in the human population (44), which is why many IAV vaccines include two representative IAV and IBV strains (5).

A significant challenge in battling IAVs is the constant evolution of the surface antigens (HA and NA) in response to pressure from the host immune system, which is referred to as antigenic drift and antigenic shift. Antigenic drift is most evident in circulating seasonal IAVs, where substitutions by the polymerase that cause mutations in the surface antigen epitopes have continuously been selected to enable reinfection of the same host (45). Antigenic shift is responsible for the development of the IAV pandemics, and it relies on the less

receptor-binding domain (located in the HA1 region) to associate with sialylated glycoconjugates on a host "receptor." Binding to the "receptor" triggers endocytosis. (ii) The virus then traffics to the endosome where the lower pH facilitates a conformational change in HA, exposing the fusion peptide (located in the HA2 region) for insertion into the endosomal membrane. (iii) The HA pre-hairpin conformation begins to collapse, forming a six-helix bundle that promotes hemifusion of the viral envelop with the endosomal membrane. At some point, the M2 channel opens to release the viral ribonucleoproteins (vRNPs) from M1 by acidifying the viral interior. (iv) HA further collapses into a trimer of hairpins to promote the formation of the fusion pore, which (v) releases the vRNPs into the cytosol. (vi) The exposed nuclear localization signals (NLS) on the vRNPs are recognized by the adaptor protein importin-α, leading to the recruitment of importin-β that (vii) facilitates the transport through the nuclear pore complex (NPC) and into the nucleus.

frequent process of reassortment, which involves the exchange of vRNAs between two IAVs during co-infection of a cell (43, 46, 47). While reassortment can happen between two related IAVs, antigenic shift occurs when the reassortment process yields a new IAV subtype.

IAVs are also under constant negative selection due to the functional requirements of the viral proteins, and the constraints of the limited genome. Several roles have been reported for most of the IAV proteins. These include the function of HA in receptor binding, as well as membrane fusion, and viral release by the sialidase activity of NA. To perform these functions, the proteins need to correctly fold, oligomerize, and as for the genome itself, they have to be properly trafficked and packaged into new virions. Thus mutations that benefit one property may hinder another. The goal of this review is to highlight these functional requirements by providing a summary of the mechanisms IAVs have evolved to facilitate cell entry, replication, virion assembly and movement, with particular attention to how IAVs coordinate the infection process.

# IAV CELL BINDING AND FUSION

IAVs initiate the infection process by using the HA molecules on the viral envelope. Upon reaching a potential host cell, the HA receptor-binding site attaches the virus to surface glycoconjugates that contain terminal SA residues (**Figure 2A**) (18, 48, 49). IAVs then scan the cell surface for the proper sialylated "receptor" by using the sialidase function of NA to remove local SAs and liberate nonproductive HA associations (50). Currently, the "receptor's" identity remains unknown, but it is generally thought that HAs from avian IAVs have higher specificity for receptors with α-2,3-linked SAs that have a "linear" presentation (51, 52), whereas HAs from human IAVs prefer an α-2,6 linkage, which results in a more "bent" presentation (**Figure 2A**) (53, 54). While these preferences correlate with SA linkages in the respective hosts (55), several studies have shown that matching HA receptor binding preferences with the SA linkages in a particular host is not essential for infection, but is more critical for transmission (56–59). This implies that the IAV "receptor" either displays significant cell tropism in the airways or that IAVs can potentially use more than one receptor.

Despite the unknown identity of the receptor, it is clear that HA-mediated binding to the receptor triggers endocytosis of the virion (**Figure 2B**, step i). The endocytosis can either occur in a clathrin-dependent manner, involving dynamin and the adaptor protein Epsin-1 (60–62), or by macropinocytosis (61, 63, 64). Once inside the cell, the virus is trafficked to the endosome, where the low pH activates the M2 ion channel (61, 65, 66), and causes a large conformational change in HA that exposes the fusion peptide (**Figure 2B**, step ii) (67–69). Opening of the M2 ion channel acidifies the inside of the viral particle, releasing the packaged vRNPs from M1 (**Figure 2B**, step iii), which enables the transfer of the vRNPs to the host cytoplasm following HA-mediated fusion (70, 71).

Fusion of the viral-endosomal membranes by HA occurs through multiple steps [reviewed in Refs. (72, 73), and requires cleavage of HA by host cell proteases into two subunits, HA1 and HA2 (55, 74, 75)]. The cleavage (see HA Proteolytic Activation at the *Golgi* or Plasma Membrane) is required to enable the exposure of the fusion peptide on the N-terminus of the HA2 upon the pH change in the endosome (69). Once exposed, the fusion peptide inserts into the endosomal membrane, while the C-terminal transmembrane domain (TMD) anchors HA2 in the viral membrane, creating a pre-hairpin conformation (see **Figure 2B**, step ii "box"). The HA2 trimers then fold back on themselves creating a hairpin that begins to position the two membranes in close proximity to each other (see **Figure 2B**, step iii "box"). The hairpin bundles then further collapse into a six-helix bundle, and in doing so, the two membranes come closer together enabling the formation of the lipid stalk, and the subsequent fusion of the two membranes (**Figure 2B**, step iv). To date, not all of these stages have been observed with HA and some have been inferred based on observations of related fusogens from other viruses.

# IAV GENOME TRAFFICKING TO THE HOST CELL NUCLEUS

In contrast to the early steps in IAV entry, vRNP trafficking to the nucleus following the fusion event is highly dependent on the host cell machinery and transport pathways [reviewed in Ref. (76)]. Supported by numerous studies, the current model is that the newly released cytoplasmic vRNPs use the importin-α– importin-β nuclear import pathway to gain entry to the host cell nucleoplasm (**Figure 2B**, steps vi and vii) (77–83). To initially engage this pathway, it is thought that the vRNPs use the surface exposed nuclear localization sequences from the numerous NP molecules to recruit the adapter protein importin-α (80–82). Upon binding to the vRNP, importin-α is recognized by the importin-β transport receptor, which directs the vRNP to the nuclear pore complex, where it is transported into the nucleoplasm.

Recent improvements in imaging and RNA labeling techniques have made it possible to monitor the entire entry process in single cells (61, 62, 83–85). The cumulative results from these studies show that IAVs can deliver their vRNPs from the cell surface to

Figure 3 | Transcription of the complimentary RNA (cRNA) and viral RNA (vRNA) by the heterotrimeric viral RNA-dependent RNA polymerase (PB2, PB1, and PA). (A) The viral polymerase initiates transcription of the positive-sense cRNA upon base-pairing of ATP and GTP with the complimentary nucleotides in the 3′ end of the vRNA. The subsequent formation of the A-G dinucleotide is followed by elongation of the cRNA transcript. Nucleoprotein (NP) molecules successively bind to the cRNA as it exits the polymerase, promoting cRNP assembly. cRNP formation is completed upon the termination of transcription and with the binding of a newly synthesized viral polymerase (yellow outline). (B) vRNA transcription proceeds in a similar manner as cRNA synthesis. Recent structures support a model where (i) ATP and GTP base pair to the nucleotides located 4 and 5 bases from the cRNA 3′ end, and there form a dinucleotide, which then disassociates and reanneals with the bases at positions 1 and 2. (ii) Alternatively, ATP and GTP could bind directly to the terminal nucleotides and form a dinucleotide. Both mechanisms would position the dinucleotide at the cRNA 3′ end, which is necessary to transcribe a full-length vRNA. Similar to cRNP formation, multiple NPs and a viral polymerase bind to the newly transcribed vRNA to produce a new viral ribonucleoprotein (vRNP).

the nucleus in approximately 1 h, with entry and fusion occurring rather quickly (~10 min), and nuclear import requiring the bulk of the time (85). A striking observation from these studies is the efficiency with which the eight vRNAs reach the nucleus, indicating how effectively vRNPs recruit the host nuclear import factors. Supporting this observation, it was shown that NP adaptation to the importin-α isoforms of a particular species is crucial for productive IAV infections (86). While the bulk of the vRNP trafficking work has been carried out using various immortalized cell lines, the potential species related differences, and the essential role of vRNP trafficking in reassortment, emphasize the need for further methodology development to examine the details of IAV entry in primary cells and tissue explants.

# REPLICATION OF THE vRNAs

Inside the nucleus, the heterotrimeric viral RNA-dependent RNA polymerase carries out the transcription and replication of the vRNAs [reviewed in Refs. (87, 88)]. The replication of the influenza genome involves two steps: transcription of complimentary RNA (cRNA), followed by transcription of new vRNA copies using the cRNAs as templates. The cRNAs are produced by an unprimed process that relies on the correct complementation of free rNTPs (generally GTP and ATP) with the 3′ end of the vRNA template (**Figure 3A**) (89, 90). The nucleotide complementation locks the vRNA template into the polymerase active site within the PB1 subunit and results in the formation of an A–G dinucleotide from which the cRNA is elongated (91). Upon exiting the polymerase, the cRNA associates with newly synthesized NP molecules and a single copy of the viral polymerase to assemble into a cRNP (90).

Currently, it is thought that the newly produced viral polymerases, which are incorporated into the cRNPs, generate multiple vRNA copies in a manner similar to cRNA transcription (**Figure 3B**). However, there is one distinction related to the difference in the positioning of the longer 3′ end of the positivesense cRNA. Due to the increased length, the cRNA is positioned in the polymerase such that the rNTP annealing and dinucleotide formation is likely to occur at the nucleotides located 4 and 5 bases from the cRNA 3′ end (**Figure 3B**, pathway i) (90, 92–94). The dinucleotide primer then has to dissociate and reanneal to the nucleotides at the 3′ end prior to elongation (**Figure 3B**). Alternatively, the cRNA 3′ end could reposition within the polymerase due to rNTP binding, resulting in the generation of full-length vRNA transcripts directly (**Figure 3B**, pathway ii). The transient nature of the rNTP annealing and dinucleotide formation makes it technically challenging to exclude either possibility. The remaining task of assembling a vRNP is analogous to cRNP formation.

Figure 4 | Transcription of IAV mRNAs by the viral polymerase. Viral mRNA transcription occurs when the viral ribonucleoproteins reach the host cell nucleus and is assisted by the association of the viral polymerase (PA subunit) with the cellular RNA polymerase II C-terminal domain (RNA pol II CTD). Transcription initiates by a "cap-snatching" mechanism where the PB2 subunit binds to the 5′ cap of a host mRNA (red). Cap binding positions the region of the mRNA 10–13 nucleotides downstream for cleavage by the endonuclease domain in the PA subunit. Following cleavage, a conformational shift repositions the acquired mRNA capped primer to the PB1 subunit where the 3′ end base-pairs with a complimentary sequence at the vRNA 3′ end. Following the priming event, the viral polymerase extends the mRNA transcript. The transcription is terminated by a "reiterative stuttering" process (depicted in the box), which occurs when the polymerase encounters the 5–7 consecutive uracil bases at the vRNA 5′ end. The "reiterative stuttering" function likely involves multiple cycles of dissociation and reannealing, and effectively polyadenylates [A(n)] the viral mRNA by continuously repositioning the elongating 3′ end on the uracil-rich region of the vRNA template.

# VIRAL mRNA TRANSCRIPTION

Viral mRNA transcription from the vRNA templates is primed, making it significantly more efficient than cRNA and vRNA transcription (95). The viral polymerase obtains the primers through a mechanism termed cap snatching (96), which is aided by the association with the cellular RNA polymerase II C-terminal domain (**Figure 4**) (97–99). For cap snatching, the viral polymerase uses the PB2 subunit to bind to 5′ caps of nascent host transcripts (100) and the PA subunit endonuclease domain to cleave 10–13 nucleotides downstream of the 5′ cap (101–103). The PB2 cap-binding domain then rotates to position the newly acquired capped primer into the PB1 catalytic center where it is extended using the vRNA as a template (95). Finally, each transcript is polyadenylated through a reiterative stuttering′ process, which occurs when the polymerase encounters the short poly-U sequence at the vRNA 5′ end (**Figure 4** "box")(104, 105). This process likely involves multiple cycles of dissociation, repositioning, and reannealing of the mRNA to this template region of the vRNA to achieve polyadenylation.

During the course of infection, mRNA synthesis occurs before cRNA and vRNA transcription, and mRNA transcription is much more abundant because the use of primers significantly increases the initiation efficiency (106). The initial mRNAs are transcribed by the vRNP-associated polymerases and exported from the nucleus for translation by cytoplasmic ribosomes (93). However, the M and NS transcripts also possess donor and acceptor splice sites that match well with those in human transcripts (107). These sites recruit the cell spliceosome, which produces the spliced transcripts that encode for the M2 and NS2 proteins, respectively (108–112). The NS transcript has been reported to maintain a similar ratio of non-spliced and spliced transcripts throughout infection (113), whereas the ratio of the spliced M transcripts (encoding M2) have been shown to increase during infection (114). These observations imply that NS1 and NS2 are always equally expressed, while M2 expression is more biased toward the later stages of infection. However, it is likely that the splicing efficiency of the NS and M transcripts differs between IAV strains (115, 116).

# ASSEMBLY AND TRAFFICKING OF vRNPs

IAV protein synthesis is entirely dependent on the translation machinery of the host cell. Following nuclear export [reviewed in Ref. (117)], the translation of the viral mRNAs is divided between cytosolic ribosomes (for PB1, PB2, PA, NP, NS1, NS2, and M1) and endoplasmic reticulum (ER)-associated ribosomes for the membrane proteins HA, NA, and M2 (**Figure 5**, steps i and ii).

Figure 5 | Coordination of viral ribonucleoprotein (vRNP) assembly and trafficking to the plasma membrane. Upon entry into the host cell nucleus, (i) the vRNP-associated viral polymerase transcribes the viral mRNAs. (ii) The mRNAs are either directly, or after alternative splicing, exported for translation by cytosolic ribosomes. (iii) Newly synthesized viral polymerase subunits (PA, PB1, and PB2) and nucleoprotein (NP) are imported back into the nucleus. (iv) Due to the inefficient dinucleotide priming, the vRNP-associated viral polymerase also infrequently transcribes complimentary RNA (cRNA) copies that assemble into cRNPs *via* (v) binding of a newly synthesized viral polymerase (PA, PB1, and PB2) and NP. (vi) The polymerase transcribes viral RNA (vRNA) copies from the positive strand in the cRNPs and these assemble into vRNPs by (vii) association with a new viral polymerase (PA, PB1, and PB2) and NP. Once assembled, the new vRNPs can (viii) transcribe additional viral mRNAs, (ix) transcribe new cRNA copies, or (x) associate with the newly synthesized viral proteins M1 and NS2 to facilitate the recruitment of CRM1, which (xi) mediates the nuclear export of the vRNP. (xiia) Once exported, the vRNPs then associate with Rab11 that assists in the trafficking of the vRNPs toward the cell surface. The vRNP trafficking either occurs by Rab11-containing vesicles associated with microtubules or (xiib) through Rab11 located in the modified endoplasmic reticulum (ER) membranes. How the vRNPs reach the budding site at the plasma membrane is currently not known.

Nuclear localization sequences on the newly synthesized NP proteins and polymerase subunits (PB1, PB2, and PA) target these proteins into the nucleus by recruiting the importin-α-importin-β pathway that is utilized for vRNP nuclear import (**Figure 5**, step iii). The NP and PB2 proteins are imported individually, whereas the PB1 and PA proteins are imported as a heterodimer (81, 118). In the nucleus, these newly synthesized proteins assist in viral mRNA transcription and vRNA replication. NP monomers bind to 12 nucleotide stretches with a partial G bias in vRNAs, and presumably cRNAs, to assemble vRNPs and cRNPs through a process that may be regulated by the NP phosphorylation (**Figure 5**, steps v and vii) (119–121). The heterotrimeric polymerase assembles and binds to the newly formed cRNPs to transcribe vRNAs (**Figure 5**, step vi) that upon formation into vRNPs can generate additional viral mRNA (**Figure 5**, step viii), or cRNA transcripts (**Figure 5**, step ix) (90, 93).

The viral RNA-binding protein NS1 is synthesized early and also imported into the nucleus, where it can act as an inhibitor of interferon signaling [reviewed in Ref. (122)]. In addition, NS1 may contribute to viral mRNA export from the nucleus by linking the viral transcripts to the cellular nuclear export components TAP/NXF1, p15, Rae1, E1B–AP5, and the nucleoporin NUP98 (123). NS2 (alternatively known as the nuclear export protein) and M1 are imported into the nucleus as well. Multiple studies have implicated these two proteins in the nuclear export of vRNPs (70, 71, 124–127). While the mechanism remains unclear, current data support a model where M1 acts as an adaptor protein linking NS2 to vRNPs (**Figure 5**, step x) (128, 129). Through established interactions with CRM1, NS2 is then able to target the vRNP complex to the CRM1 nuclear export pathway for transport to the cytoplasm (127), where M1 potentially prevents the re-import of vRNPs by blocking access to the NP nuclear localization sequences (**Figure 5**, step xi) (71).

Within the cytoplasm the vRNPs are trafficked toward the plasma membrane for viral assembly by Rab11. Rab11 facilitates the interaction by associating with the viral polymerase PB2 subunit (130), potentially providing a quality control mechanism that ensures new virions incorporate vRNPs carrying a polymerase. Earlier studies proposed that vRNPs specifically associate with Rab11 on recycling endosomes, which use microtubules for transport toward the cell surface (**Figure 5**, step xiia) (130–132). An alternative model has recently been proposed where infection causes tubulation of the ER membrane network and the vRNPs bind to Rab11 molecules that have localized to this network for trafficking toward the plasma membrane (**Figure 5**, step xiib) (133). Currently, it is not known how vRNPs are transferred to the plasma membrane in either model, or how IAVs incorporate all eight of the different vRNPs in a "1 + 7" configuration. While several studies have indicated that specific vRNP associations likely contribute to the packaging of the eight vRNPs (35, 134, 135), the underlying mechanisms remain to be established.

# ER TARGETING AND MATURATION OF THE IAV MEMBRANE PROTEINS

The IAV membrane proteins, which are ultimately destined for the viral envelope, are synthesized by ribosomes associated with the ER membrane. Similar to cellular secretory proteins, ribosome–nascent chain complexes containing NA, HA, or M2 are co-translationally directed to the ER by interactions of their hydrophobic targeting sequences with the signal recognition particle (SRP) (**Figure 6**, step ii) (136–139). The cleavable signal sequence on HA facilitates the interaction with SRP, whereas NA and M2 use their respective TMD as an ER targeting sequence. Once bound, SRP targets the ribosome–nascent chain complexes to the SRP receptor in the ER membrane (**Figure 6**, step iii), which transfers the ribosome to a Sec61 protein-conducting channel known as the translocon (140–142). Linked to the dependence on SRP, mutations that alter the targeting sequence hydrophobicity of cellular secretory proteins have been shown to decrease their ER targeting and subsequent synthesis (143, 144). Although this aspect has not been examined for the IAV membrane proteins, there is evidence that the hydrophobicity of their ER-targeting sequences change (138, 148), which suggests IAVs potentially use this mechanism to titrate NA and HA expression.

The translocon enables passage of the elongating NA, HA, and M2 polypeptides into the ER lumen and facilitates the membrane partitioning of their respective TMD segments through a lateral gate (145, 146). To activate the membrane integration, the TMD segments have to be of the appropriate length and hydrophobicity (146, 147). In human H1N1 and H3N2 viruses, these criteria are conserved in the TMDs of HA and M2, but not in the TMD of NA, as it has become progressively less hydrophobic in the H1N1 viruses (148). The uncharacteristic hydrophobicity loss was shown to be possible because of the NA TMD being positioned at the N-terminus (138). The positioning (~435 amino acids from the C-terminus), combined with the slow rate of ribosomal translation (~5 amino acids per second), likely provides these nontypical TMDs with significant time to properly orientate and facilitate membrane insertion during the co-translational translocation process.

During translocation, the N-terminus of HA and M2 is directly translocated into the ER lumen, whereas NA inverts, positioning the C-terminus in the ER lumen (137, 138). In addition, HA and NA receive multiple N-linked glycans. The glycans are transferred by the oligosaccharyltransferase to Asn–X–Ser/Thr sequences, and vary in number as well as positioning based on the strain, or subtype (149). One function of the glycans is to increase the folding efficiency of NA and HA by recruiting the lectin chaperones (calnexin and calreticulin) and the associated oxidoreductase ERp57, which aids in disulfide bond formation (136, 150–152). This is especially crucial for the HA and NA proteins that possess a significant number of intramolecular disulfide bonds (e.g., six in HAs, eight in N1, and nine in N2) (153–155). By contrast, M2 possesses two intermolecular disulfide bonds in its tetrameric conformation (156). Depending on the subtype, NA tetramers also possess 2 or more intermolecular disulfide bonds.

Oligomerization of HA involves the trimerization of independently folded monomers, whereas NA tetramerization has been proposed to result from the pairing of two co-translationally formed dimers, which assemble through a process involving the N-terminal TMD of NA (150, 157). In line with this model, it has been shown that the TMD is essential for proper NA folding, and that the decreasing hydrophobicity in the N1 TMDs functions to

mucus to facilitate movement of the virus to neighboring cells.

support the folding and oligomerization of the enzymatic head domain (158, 159). IAVs easily achieve the protein concentrationdependent requirement for oligomerization due to the abundance of HA and NA that is synthesized during an infection. However, these high synthesis levels at the ER can also be deleterious by activating the ER-stress response. Indeed, several studies have shown that IAV replication does activate the ER-stress induced unfolded protein response (160, 161), but this response is also mitigated by the inhibition of the eIF2α-kinase and stress granule formation through the functions of other viral proteins (162).

Despite everything that is known about the synthesis and assembly of the IAV membrane proteins, several aspects have yet to be addressed. These include obtaining atomic structures of fulllength HA and NA in a membrane, something that should become easier to address with the advances in cryo-electron microscopy structure determination. Identifying if the NA protein removes SA residues directly from substrates within the *Golgi*, as this could decrease the effectivity of nonmembrane permeable NA inhibitors. It is also unclear how IAVs regulate the timing and expression levels of the viral proteins as viral mRNA transcription shows little temporal variation (163, 164). While it is likely that M2 is regulated in part by splicing (112, 114), this does not apply to HA and NA. Recent work has linked NA and HA regulation to the nucleotide composition of the 5′coding regions for their ER-targeting sequences, which dramatically differ from the profile of corresponding regions in human secretory protein mRNAs (165, 166). An obvious candidate for post-transcriptional regulation is the viral RNA-binding protein NS1. Indeed, many studies have shown that NS1 can increase translation of particular mRNAs, possibly by enhancing the translation initiation rate through the recruitment of eIF-4G to the 5′region of viral mRNAs (165, 167–171). However, a clear mechanistic picture for influenza protein regulation is lacking.

# HA PROTEOLYTIC ACTIVATION AT THE *GOLGI* OR PLASMA MEMBRANE

HA traffics from the ER as a fusion incompetent precursor termed HA0. To gain its fusion function, HA must be cleaved into the subunits HA1 and HA2 (74, 172, 173). The cleavage occurs in either a monobasic, or a multibasic, cleavage site (55). Multibasic sites are commonly found in highly pathogenic avian IAVs and are cleaved by furin, a calcium-dependent serine endoprotease that is located within the *trans-Golgi* network (174). Furin is also ubiquitously expressed (175), which is one of the major reasons why avian IAVs with a multibasic cleavage site are generally more pathogenic.

By contrast, human (and low pathogenic avian) IAVs encode for HAs with a monobasic cleavage site, which have been shown to be processed by different proteases in human respiratory epithelial cells. These include the transmembrane protease serine S-1 member 2 (TMPRSS2), human airway trypsin-like protease (HAT), and possibly TMPRSS4 (176, 177). HAT localizes at the plasma membrane where it can either cleave newly synthesized HA or the HA found in cell-associated virions (178, 179). Similar to furin, TMPRSS2 resides in the *trans-Golgi* network, where it cleaves HA en route to the plasma membrane. The M2 ion channel is thought to prevent the premature activation of HA following cleavage by equilibrating the slightly acidic pH of the *Golgi* (180, 181). Distinct from furin, TMPRSS2 expression has been found to be more restricted to the upper and lower respiratory tract, whereas HAT was mainly shown to be expressed in the upper respiratory tract (182). These cell tropisms suggest that lower respiratory infections are likely mediated by TMPRSS2, and could be one of the primary reasons human IAVs are confined to the epithelial layer of the respiratory tract.

# IAV ASSEMBLY AND BUDDING

Compared with the bulk lipid profile of the plasma membrane, IAV envelopes are enriched in cholesterol and sphingolipids (32), indicating that they bud from distinct apical plasma membrane regions often referred to as "rafts" (183). However, infectious IAVs must possess mechanisms to target the eight vRNPs, M1, HA, NA, and M2 to these sites in the membrane (184, 185). HA is believed to localize to these distinct regions based on fatty acid modifications of the C-terminal cysteine that occur in the *Golgi* (186–189), whereas NA enrichment has previously been attributed to a property in the C-terminus of the TMD (190). In contrast, M2 has been shown to accumulate at the boundaries of these budding domains (191), and the cytosolic protein M1 has been proposed to localize to the budding region by associating with the short cytoplasmic tails of HA and NA (192). However, it is equally plausible that NA and HA create membrane domains with a unique lipid profile that have a high affinity for M1. Finally, the vRNPs, delivered to the cell periphery by Rab11, are thought to localize to the budding site by binding to M1 (193, 194).

In addition to orchestrating the assembly of the correct viral components at the apical budding site, IAVs also have to remodel the membrane to induce bud formation, and ultimately scission of the viral envelope from the plasma membrane. To promote bud formation, the virus must first induce significant curvature in the membrane and then constrict the two opposing membranes of the viral envelope to help to facilitate membrane scission. Curvature can be induced by (i) protein or "molecular" crowding on one leaflet of a bilayer, (ii) association of curved or "bending" proteins with the bilayer, (iii) biased accumulation of cone shaped lipids in one leaflet of the bilayer, or (iv) the cytoskeleton (195). Based on cumulative data regarding budding, IAVs appear to induce membrane curvature through a combination of these mechanisms. Indicative of using molecular crowding and bending proteins, several studies have demonstrated that HA and NA expression is sufficient to induce budding, and that the efficiency and shape uniformity benefit from the presence of M1 (196–199). These results indicate that the abundance of HA and NA on one side of the membrane can contribute to curvature. It also is intriguing to speculate that the asymmetric (154) shape of NA plays a role in this process as it is often seen clustering in the viral membrane (16, 199). By contrast, M1 appears to be analogous to a membrane-bending protein as it recruited to the cytosolic side of the membrane budding site, oligomerizes upon reaching the membrane, and these oligomers have been modeled to form curved structures (200–202). Based on these properties, it is plausible that M1 significantly influences the membrane curvature at the budding site, potentially explaining its role in discerning whether IAVs form spheres or filaments (27, 203).

The ion channel M2 localizes to the budding site boundary and has also been shown to contribute to IAV scission by functioning as a membrane-bending protein (191, 204). The membrane-bending property of M2 is localized in an amphiphilic α-helix that can incorporate the amino acid side chains from its hydrophobic face into a leaflet of the bilayer. With this domain positioned in the cytosol, the intercalation results in negative membrane curvature, which has been proposed to facilitate viral bud neck formation and scission, presumably by decreasing the distance between the two opposing membranes of the viral envelope (204). While much of the framework concerning IAV budding has been established, it has been difficult to identify the details of the budding process, in part due to the mobility and heterogeneity of the plasma membrane. The lack of strong phenotypes from domains proposed to contribute to budding could also imply that IAVs have built redundancy into the budding process (205–207). The possibility of redundancy is certainly plausible, as IAVs contain the necessary components to allow for a combination of lipid recruitment, molecular crowding, and a membrane-bending protein.

# IAV CELL RELEASE AND MOVEMENT

Once the newly assembled IAVs bud, their release is highly dependent on the sialidase activity of NA. NA is a homotetramer, and each subunit is comprised of a short N-terminal cytoplasmic tail (six amino acids), followed by a TMD, a length variable stalk, and a globular enzymatic head domain (208). The globular head domain forms a 6-bladed propeller structure, where each blade is comprised of four antiparallel β-sheets that are stabilized by disulfide bonds (155, 209, 210). The catalytic Tyr residue is found in a highly conserved active site that forms a deep pocket in the center of each monomer (211). All of the residues necessary for catalysis exist within each monomer (212), which has made it difficult to reconcile why NA evolved to function as a tetramer (208, 213, 214). Structures of the enzymatic head domain indicate that NA tetramers bind up to five calcium ions and calcium has been shown to contribute to NA activity (155, 208, 215). However, it remains unclear why influenza NA has evolved to position a calcium ion at the tetrameric interface.

NA facilitates viral release by catalyzing the hydrolysis of the glycosidic linkage that attaches SA to underlying sugar molecules (216–218). By removing local SA residues, NA prevents HA binding at the cell surface, which facilitates the release of the virus during budding (**Figure 6A** and step vi) (219, 220). NA has also been shown to promote the separation of IAVs by removing SA residues from the N-linked glycans located on the HA and NA molecules in the viral envelope (**Figure 6B** and step vii) (221). In contrast to HA, NAs from human IAVs show a general preference for α2,3-linked SA with variable abilities to cleave α2,6-linked SA residues (208, 222, 223). However, a thorough analysis of NA SA preference is lacking. More recent studies have found that some strains possess NAs that are inefficient enzymes, but still capable of SA binding, raising the question of whether a poor NA enzyme could contribute to, or replace, the HA receptor-binding function (224, 225).

The movement of IAVs from cell to cell in the respiratory epithelium is significantly different from that in immortalized cell lines grown in liquid culture due to the presence of different cell types and a mucus layer. The mucus layer provides a protective barrier for the epithelium and is rich in heavily glycosylated mucins that can interact with IAVs and limit cell binding (226, 227). Studies measuring viral movement through mucus and respiratory epithelial cells have shown that NA-mediated cleavage of SAs from mucins enhances IAV movement through the mucus layer and infectivity (**Figure 6C** and step viii) (226, 228, 229). Recent work showed that this function may also apply to transmission, as IAVs that possess low NA activity, and are inhibited by mucus, are deficient in aerosol and contact transmission (230).

# PERSPECTIVES

IAVs are constantly exposed to negative and positive selection pressure, which shapes how the virus evolves. The functional requirements of each IAV protein, such as enzyme catalysis, substrate binding, oligomerization, and domains that perform essential interactions with host proteins all combine to create substantial negative selection pressure that often manifests in the form of sequence conservation. Negative pressure can also come from functions within the vRNA sequences. These include promoters and "packaging signals," but are also likely to involve aspects such as the formation of structural elements, or possibly mediating vRNP interactions that generate the 1 + 7 assembly in viral particles. In addition, the exposure of IAVs to the immune response and constantly changing environments such as host, temperature, pH, cell type, and antivirals result in positive selection pressure. Experimentally, addressing each type of selection has its caveats, but clearly a holistic picture of both IAV and host functions are required to begin predictions of evolutionary constraints on the virus.

Most studies on the influenza evolutionary process focus primarily on antigenic drift and antigenic shift. However, all the viral transcribed RNAs are subject to replication errors by the viral polymerase, which are estimated at 1 per 2,000–10,000 nucleotides (231–233). Consequently, both the viruses and the viral proteins are likely to exist as large heterogeneous populations during an infection. As many IAV proteins are homo-oligomers this can potentially generate heterogeneity within individual protein complexes that could have functional advantages. By applying single particle and single cell analysis, these types of aspects are beginning to be investigated (234). Another interesting approach is deep mutational scanning, which has been used to examine the site-specific amino acid tolerance of IAV proteins in general, and in the context of different selection pressure (235–238).

Currently, the best characterized protein in IAVs is HA, which has two primary functions, (i) to initiate binding to the host cell and (ii) to deliver the vRNPs to the host cell cytosol by fusing the viral and endosomal membranes. These functions are efficiently divided between the two domains of HA (HA1 and HA2), created by proteolysis. The receptor-binding site responsible for entry is located in the considerably larger HA1 subunit that is known to be immunodominant, explaining the high sequence variability in this region (239). By contrast, the smaller HA2 subunit, containing the fusion peptide that is necessary to deliver the viral genome to the host cell, shows considerably higher sequence conservation. This organization is logical from the viral perspective as the large HA1 subunit likely blocks antibody recognition of HA2. The viral downside is the need to escape antibodies that inhibit the receptor-binding pocket without losing specificity and the binding function.

Based on this knowledge, several exciting new strategies are being developed to elicit the production of antibodies that target the more conserved region of HA (240–242). The hope is that these strategies will generate broadly neutralizing antibodies that recognize multiple HA subtypes from IAVs and the distinct lineages in IBVs, providing longer lasting immunity and alleviating the threat of potential pandemics. A similar approach using NA would likely provide additional benefits. However, our knowledge of NA lags behind HA. Currently, it is still not known why NA has evolved to function as a tetramer, which is relevant because this property presumably restricts the potential antigenic drift (mutations) it can accommodate and still function.

A relatively overlooked feature in the replication process is the contributions of host RNA-binding proteins (RBPs). Human cells are predicted to encode over 1,500 RBPs, 700 of which are predicted to interact with mRNAs (243). As a RNA virus, it is highly likely that IAVs have evolved to utilize this enormous network of RBPs, which is supported by observations that some RBPs inhibit IAV replication, whereas others contribute (244–246). It should also be considered that changes in RBPs have been associated with various cancers, which could possibly influence the susceptibility to influenza infections (247, 248). With the growing interest in RNA biology, this aspect of IAV infections is likely to receive considerable attention in the future.

In terms of IAV antivirals, the recent progress in determining the structures and mechanisms of the viral polymerase should significantly aid in the current development of drugs aimed at inhibiting different aspects of IAV transcription (249). Through

# REFERENCES


continued progress in defining the fundamental mechanisms that are necessary for IAV infections, replication and intercellular movement, it should become possible to minimize the annual burden caused by IAVs.

# AUTHOR CONTRIBUTIONS

RD wrote the review with input from DD, RR, HÖ, and HW. DD, RR, HÖ, and HW put together the figures and wrote the figure legends.

# FUNDING

RD is supported by grants from the Swedish Research Council (K2015-57-21980-04-4) and the Carl Trygger Foundation (CTS17:111).


promoters during viral RNA replication. *J Virol* (2006) 80:2337–48. doi:10.1128/JVI.80.5.2337-2348.2006


and fusion at pH 5.2. *Proc Natl Acad Sci U S A* (1981) 78:4133–7. doi:10.1073/ pnas.78.7.4133


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Dou, Revol, Östbye, Wang and Daniels. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# The Drivers of Pathology in Zoonotic Avian influenza: The interplay Between Host and Pathogen

*William S. J. Horman1,2, Thi H. O. Nguyen1 , Katherine Kedzierska1 , Andrew G. D. Bean2 and Daniel S. Layton2 \**

*1Department of Microbiology and Immunology, University of Melbourne at the Peter Doherty Institute for Infection and Immunity, Parkville, VIC, Australia, 2Australian Animal Health Laboratory, Health and Biosecurity, Commonwealth Scientific and Industrial Research Organisation (CSIRO), East Geelong, VIC, Australia*

The emergence of zoonotic strains of avian influenza (AI) that cause high rates of mortality in people has caused significant global concern, with a looming threat that one of these strains may develop sustained human-to-human transmission and cause a pandemic outbreak. Most notable of these viral strains are the H5N1 highly pathogenic AI and the H7N9 low pathogenicity AI viruses, both of which have mortality rates above 30%. Understanding of their mechanisms of infection and pathobiology is key to our preparation for these and future viral strains of high consequence. AI viruses typically circulate in wild bird populations, commonly infecting waterfowl and also regularly entering commercial poultry flocks. Live poultry markets provide an ideal environment for the spread AI and potentially the selection of mutants with a greater propensity for infecting humans because of the potential for spill over from birds to humans. Pathology from these AI virus infections is associated with a dysregulated immune response, which is characterized by systemic spread of the virus, lymphopenia, and hypercytokinemia. It has been well documented that host/pathogen interactions, particularly molecules of the immune system, play a significant role in both disease susceptibility as well as disease outcome. Here, we review the immune/virus interactions in both avian and mammalian species, and provide an overview or our understanding of how immune dysregulation is driven. Understanding these susceptibility factors is critical for the development of new vaccines and therapeutics to combat the next pandemic influenza.

Keywords: avian influenza virus, zoonosis, H7N9, H5N1, highly pathogenic avian influenza virus

# EMERGENCE OF AVIAN INFLUENZA (AI) VIRUS INFECTION IN HUMANS

Influenza A viruses have consistently posed a major threat to human health, both through seasonal infections and pandemic outbreaks (1). AI viruses have contributed significantly to this, and in recent years, highly pathogenic AI (HPAI) viruses have emerged as a major zoonotic threat. AI viruses naturally circulate in wild bird populations, including but not limited to, ducks and waterfowl, and can spill over to poultry birds such as chickens. Other than a few novel strains isolated in bats, all influenza A subtypes have been found in aquatic birds, which act as natural reservoirs for the viruses (2). These viruses typically replicate in the gastrointestinal and upper respiratory tract of both these natural hosts and chickens, and typically present as subclinical to mild disease (3, 4). Based on these clinical signs in chickens, these influenza viruses are classified as "low pathogenicity AI" (LPAI)

### *Edited by:*

*Ding Oh, Federation University, Australia*

#### *Reviewed by:*

*Alyson Ann Kelvin, Dalhousie University, Canada Elisa Vicenzi, San Raffaele Hospital (IRCCS), Italy*

> *\*Correspondence: Daniel S. Layton daniel.layton@csiro.au*

#### *Specialty section:*

*This article was submitted to Viral Immunology, a section of the journal Frontiers in Immunology*

*Received: 20 April 2018 Accepted: 23 July 2018 Published: 08 August 2018*

#### *Citation:*

*Horman WSJ, Nguyen THO, Kedzierska K, Bean AGD and Layton DS (2018) The Drivers of Pathology in Zoonotic Avian Influenza: The Interplay Between Host and Pathogen. Front. Immunol. 9:1812. doi: 10.3389/fimmu.2018.01812*

**25**

infections (5). Although many AI viruses circulate without causing serious disease, other viral subtypes can lead to more severe outbreaks within birds. Birds infected with these subtypes have more severe pathogenesis and rapid disease progression, often as the result of dissemination of the virus into tissues peripheral to the gastrointestinal and respiratory tracts. This type of infection in chickens defines the subtypes as HPAI (5). Highlighting the importance of HPAI viruses, recent outbreaks of HPAI H5N6 in China and the Philippines have caused approximately 37,000 bird deaths and 400,000 more culled at an economic cost of nearly \$USD40 million (6, 7), and novel strains of H7 viruses such as H7N4 continue to cause sporadic outbreaks (8). While such outbreaks represent a significant economic burden to the poultry industry, of greater concern is the potential for HPAI viruses to cross the species barrier into mammals, especially humans.

Despite there being a broad range of AI subtypes, fortunately, only a very select subset of these have been shown to infect humans with highly pathogenic consequences (9). The first known HPAI infections in humans were highlighted by the outbreak of H5N1 avian-derived influenza in Hong Kong in 1997, leading to 6 deaths from 18 confirmed cases (10). Since then, sporadic outbreaks of H5N1 have had highly pathogenic consequences in humans, resulting in over 450 deaths from approximately 900 cases (11–13). The emergence of the avianderived H7N9 strain infecting humans was first described in March 2013 in China's Yangtze River Delta, which has since caused 613 deaths out of 1,566 human cases throughout most of China as of January 2018 (14). This viral subtype is of a particular concern, as unlike H5N1, which is highly pathogenic in chickens and humans, H7N9 typically presents as an LPAI in chickens, but causes a high mortality rate in humans (40%), similar to that seen for H5N1 infections. H7N9 is one of several LPAI viruses in the H7 family capable of human infections, with viral transmission usually only acquired through close contact with host species (15–17). However, for reasons that are still unclear, H7N9 has greater transmissibility and more severe disease outcomes in humans than any other H7 viruses (18, 19). Thus, differences in clinical presentation across species, coupled with the potential of viruses such as H7N9 to cause a pandemic outbreak *via* evidence of human-to-human transmission (20), makes understanding the mechanisms by which these viruses cross the species barrier and become highly pathogenic in humans a critical area of investigation. Here, we discuss recent findings relating viral fitness to host susceptibility factors to understand how HPAI phenotypes are developed. We also outline how clinical manifestations following infection with LPAI or HPAI strains across different species provide further insights into the mechanisms underlying disease severity and susceptibility.

# CLINICAL MANIFESTATIONS OF DISEASE IN DIFFERENT SPECIES

Human cases of AI infection have become increasingly common since outbreaks of H5N1 in the late 1990s and accentuated by a dramatic increase in H7N9 infections during the recent "fifth wave" of epidemic infections in China (21–23). These viruses were commonly contracted by people in regular close contact with live poultry markets (24, 25), where outbreaks of AI viruses in chickens can be common yet go largely unnoticed, especially in the case of H7N9 infections. Despite the vast diversity of AI viruses, predominantly only viruses from three hemagglutinin (HA) subtypes have been recorded to naturally infect humans: H5Nx viruses, most notably H5N1; H7Nx viruses such as H7N9; and H9Nx viruses, commonly H9N2. H9Nx strains present as LPAI infections in birds and have less severe symptoms in human hosts compared to H5 and H7 strains (26–29). Despite this, as H9N2 viruses are regularly found co-circulating with H5N1 and H7N9 with assortment frequently occurring between these viruses in poultry there is a real possibility of a novel HPAI strain emerging from these subtypes (30, 31). While there have been isolated cases of human infections with other subtypes, such as H10 (32), and H6 viruses which have been discussed as a potential precursor to H5N1 with pandemic potential (33–38), H5/H7/H9-subtypes of AI remain of greatest concern for human infections and potential pandemics. Thus, understanding the clinical manifestations of these viruses in avian hosts is needed for our understanding of why these viruses are considered such a threat.

In the case of LPAI, infections tend to localize in the mucosal surfaces of the gastrointestinal tract of infected birds and although often asymptomatic, chickens may present with mild clinical signs following infection. These include excess mucus and congested tracheae, watery droppings, and mild respiratory inflammation, with rarely any other signs of respiratory disease associated with influenza infections (3, 26, 39, 40). Highest viral titers typically occur 2–3 days after infection with limited gross lesions evident, allowing the virus to replicate and be excreted into the environment with little to no effect on the host animal (41, 42). Interestingly, ducks can exhibit even more limited clinical signs following LPAI infection than chickens (43). Thus, it is interesting to note that even in avian hosts, there is a range of clinical severity to LPAI viruses, with chickens showing more moderate/severe signs compared to waterfowl infected with LPAI.

In contrast to LPAI viruses, HPAI viruses have the ability to induce severe disease and cause devastating outbreaks with high mortality rates in poultry (44, 45). In chickens, H5N1 causes acute illness with high levels of viral shedding and clinical signs such as dehydration, nasal discharge, and lesions in many tissue types (46–48). While LPAI infections tend to remain within the fecal–oral tract of infected birds, HPAI infections are often identified by virus spreading systemically to multiple tissues (49). Suzuki and colleagues (50) described the pathology of H5N1 infections in chickens, finding that clinical signs progressed from milder signs such as feather ruffling and depression behavior, to the more severe outcomes of hemorrhaging and edema in multiple tissues. These birds also showed severe respiratory distress not seen in LPAI infections (50). By contrast, ducks can present with a wider range of symptoms following infection with HPAI H5N1, with experimentally infected ducks having clinical signs as mild as depressive behavior without any other complications (51). Ducks may also show severe signs such as those seen in chickens, with common outcomes including neurological spread and hemorrhaging in the body extremities. In addition, Yamamoto and colleagues have also found that domestic ducks, unlike chickens, show corneal opacity following H5N1 infection and less severe hemorrhaging compared to chickens (52, 53). Likewise, wild ducks have been shown to exhibit less severe signs following H5N1 infection compared to other gallinaceous birds including domestic ducks, despite showing high levels of viral shedding consistent with an HPAI infection, which may suggest their role as a key reservoir species (48).

Avian influenza viruses also have the ability to infect pigs, which are housed in close proximity to human populations. Pigs often act as a "mixing vessel" for influenza viruses, which are able to reassort and thus infect humans (54, 55). While this is particularly the case for low pathogenicity viruses, there has been little evidence to suggest that pigs can contract highly pathogenic strains such as H5N1 and H7N9 to any great level, with H5N1 strains isolated from pigs in China found to be attenuated from the HPAI form (56). Although there have been no confirmed cases of H7N9 infection in pigs (57), H7N9 can replicate, cause pathology, and transmit among pigs during *in vivo* studies at low levels (58–60), as well as replicate in swine respiratory tissue *in vitro*, reinforcing the idea that pigs could still act as an important reservoir species for mammalian-adapted H7N9 (61). Of particular concern, reports of H7N2 infection in pigs (62) suggest that as the frequency of H7Nx cases increases, the likelihood of an H7N9 virus infecting pigs and potentially gaining stable mammalian transmissibility is a genuine possibility.

Studies in ferrets as a model for human infection have shown that mammalian-adapted AI viruses typically localize to the respiratory tract (63), however, these viruses have can have a limited ability to transmit *via* droplets (64–66). H5N1 viruses cause acute illness in the upper respiratory tract of ferrets with symptoms such as nasal discharge, high temperatures, and weight loss due to dehydration, and worsened pathogenesis, as highlighted by lung damage due to extensive infiltration of the lung tissue by inflammatory cells (18, 67, 68). H7N9 can also cause severe respiratory distress in this way, with lengthened time until viral clearance contributing to viral transmission and the substantial inflammation in the lungs of ferrets (69). Severe infections may cause complications such as viral pneumonia due to the breakdown of lung endothelial barriers, which contributes to this systemic spread and may lead to encephalitis or other neurological issues (18, 70). However, systemic spread to the central nervous system is more commonly associated with HPAI H5N1 than LPAI H7N9 (68).

Symptoms following human infection with AI are similar to those observed in the ferret model. Less severe cases present as more typical influenza infections, with symptoms such as fever and coughing among those commonly associated with influenzarelated illness (16, 71). However, these viruses can cause severe respiratory illness following infection in the lungs, which may manifest as atypical viral pneumonia and acute respiratory distress syndrome, and often patients who have contracted these infections die from respiratory failure (72, 73). Furthermore, much like in birds, dissemination of the virus away from the site of infection leads to other complications such as organ failure, encephalitis, and internal bleeding due to tissue destruction (10, 71, 74), all of which contribute to the lethal nature of these viruses. A summary of the varying degrees of clinical manifestations between the different species is depicted in **Figure 1**. These trends, including the ability for an LPAI AI viruses such as H7N9 to cause severe fatal disease in humans, is therefore of great importance to understand the pathology driving the variable signs and symptoms of these infections across species to best combat future infections.

# PATHOGENESIS OF HPAI INFECTIONS

Several factors appear to contribute to the worsened pathology observed in HPAI infections when compared with LPAI infections. One hallmark of HPAI pathogenesis is a rapid and robust cytokine response, often referred to as a "cytokine storm" or hypercytokinemia. This build-up of cytokines causes an inflammatory environment at the site of infection, leading to immune cell infiltration. In addition to the hypercytokinemia, there is a well-established loss of leukocytes, leading to the severe pathogenesis seen in these infections (75). Recently, the main factors associated with pathology were further described by Kuchipudi and colleagues (76). Following infection with H5N1 *in vitro*, chicken lung cells had increased expression of specific pro-inflammatory cytokines, particularly interleukins (IL)-6 and IL-8 when compared to cells infected with LPAI H2N3, suggesting that IL-6 and IL-8 may be key regulators leading to worsened pathology of HPAI compared to LPAI (76). IL-6 and IL-8 were also found to be significantly upregulated in the lungs of H5N1 infected ferrets, as well as in peripheral tissues, including the spleen, heart, and liver (77). Interestingly, this study also found that IL-6 and IL-8 were downregulated in the nasal turbinates following infection with pandemic H1N1 virus, which produced a less severe clinical infection compared to the H5N1 in ferrets (77). These findings were consistent with studies completed in rhesus macaques, which similarly showed upregulation of tumor necrosis factor α (TNFα), IL-6, and IL-8 in response to experimentally induced H5N1 infection, along with an increase in the antiviral interferons (IFNs), findings which correlate to the severe fever symptoms observed in the macaques at the peak of the fever response at day six postinfection (78). Moreover, infections with the recently emerged HPAI H5N6 in chickens, which caused human fatalities, was shown to have a very distinct immune response compared to other H5N6 strains by producing much higher levels of IL-6, IL-8, and other pro-inflammatory mediators such as TNFα compared to previously identified strains (79). While chickens and ferrets show similar pathogenesis, conversely, duck lung cells infected with the same viruses showed a decrease in IL-6 expression compared to the LPAI viruses, while IL-8 remained unchanged. It suggests that IL-6 may play a pivotal role in the regulation of AI pathology in birds and, as previously discussed, ducks typically show lessened disease severity following H5N1 infection compared to chickens (**Figure 1**). Moreover in humans, H5N1 elicits a similarly robust cytokine response, with the upregulation of IL-6, IL-10, and TNFα in response to H5N1 (80).

As the H7N9 virus is classified as an LPAI in chickens, it is interesting to note that a similar trend was observed when H7N9 of human origin were shown to induce increased production of pro-inflammatory IL-6 and IL-8 cytokines when compared to H7N9 of chicken origin (81). In the same study, it was demonstrated that IFNλ1 production was reduced for the human isolate, suggesting a possible modulation of the immune response. In addition, Wu and colleagues also found that H7N9 patients had higher levels of C-reactive protein expression in their plasma compared to H1N1 patients, and as C-reactive protein is associated with broader inflammatory responses, it suggests that this may be yet another factor alongside these regulatory cytokines contributing to disease severity (82). A similar pro-inflammatory response was observed when alveolar macrophages were infected with H7N9, however, when compared to H5N1, this cytokine response was demonstrated to be milder (83). Downregulation of these inflammatory responses to infection confers a level of immunity to these viruses in pigs, which hints at how pigs can act as mixing vessel species for AI without succumbing to severe disease. Human lung epithelial cells can express 100-fold higher levels of TNFα compared to pig lung epithelia, with suppressor of cytokine signaling 3 (SOCS3) identified as a key factor in reducing levels of TNFα in pig cells (84).

The drivers of cytokine production are resident and infiltrating immune cells, which release these cytokines in response to the infection to recruit other immune cells and hinder viral replication. This occurs following cellular activation and results in further activation of leukocytes recruited to the area in a positive feedback loop (85). As pro-inflammatory molecules are associated with apoptotic pathways in humans, increased cytokine production is likely to be a contributing factor in the loss of immune cells, or leukopenia, observed in severe cases of HPAI infection (55). These pro-inflammatory regulators lead to upregulation of the death signaling molecule Fas-ligand on the infected host cell to initiate the caspase-mediated Fas-associated pathway, in which Fas receptors on the immune cell (part of the TNF-receptor family) bind to Fas-ligand and subsequently recruit the Fas-associated death domain (FADD) molecule (86). FADD interacts with caspase-8, which initiates a signaling cascade within the immune cell, resulting in the destruction of cellular components and thus cell death, which may be causative of the pathology seen in influenza patients (87). Indeed, both H5N1 and H7N9 have been observed to cause leukopenia in hospitalized patients (88–90). According to Boonnak and colleagues, CD8+ T cells can be particularly affected by the Fasligand mediated pathway, where Fas-ligand was upregulated on plasmocytoid dendritic cells during lethal H5N1 infection in mice, which then lead to apoptosis of influenza-specific CD8<sup>+</sup> T cells in the lung draining lymph nodes (86). Furthermore, in hospitalized H7N9-infected patients during the emergence of H7N9 in 2013, the persistence of immune cell subsets within the blood contributed to disease severity and fatal outcomes, with the continuation of CD38<sup>+</sup>HLA-DR<sup>+</sup> CD8<sup>+</sup> T cell responses shown to be predictive of fatal outcomes, possibly due to longer-lasting inflammatory responses in the peripheral blood and lung (91). Loss of peripheral blood lymphocytes was also observed in human seasonal influenza A infections, with these cell subsets succumbing to apoptosis by the same apoptotic pathways described for AI viruses (92). In summary, LPAI versus HPAI viral strains differentially promote the induction of pro-inflammatory cytokines, which then alter disease outcomes in different species. The potential mechanisms by which certain HPAI AI viruses cause more severe disease will be further discussed in the following sections.

# VIRAL FITNESS ACROSS SPECIES

In order for AI viruses to be capable of infecting multiple host species they require viral adaptations allowing replication in different host cells, which ultimately increase their genetic fitness and create a sustainably replicating and transmitting virus (93). It is well established that for zoonotic transmission of AI viruses, the virus needs to present the appropriate HA binding specificity to allow viral fusion and entry. AI virus HA proteins typically have a binding preference for sialic acid residues with an α-2,3-Gal terminating sequence found on the surface of avian cells (**Figure 1**), resulting in restriction to avian cells (63, 94). Therefore, for AI viruses to gain entry into human cells displaying an α-2,6-Gal terminating sequence, modifications to the HA binding site may be required to allow this new interaction.

One of the commonly associated amino acid substitutions for LPAI strains converting from avian to mammalian receptorspecificity is a HA Q226L substitution (58, 95–97). For example, LPAI H9N2 isolates from birds can adopt a Q226L substitution, which increases its mammalian receptor binding affinity and potentially infect mammalian hosts (98, 99). However, with regards to H7N9 human isolates, Belser and colleagues described the Q226 bearing Shanghai/1 and the L226 bearing Anhui/1 as binding to largely avian α-2,3-Gal receptor analogs and mixed α-2,3/α-2,6-Gal receptors, respectively (64). Despite these differences, infections with each isolate produced effective replication in the lower respiratory tract of ferrets, suggesting additional factors are involved in mammalian adaptation. Similarly, this Q226L mutation has not been commonly observed in the HA of H5N1 HPAI viruses (100, 101). However, alternative substitutions, HA Q192H and HA I151T, can confer increased ability for replication in human hosts. Moreover, Herfst and colleagues demonstrated a closely related change to the HA of H5N1, HA Q222L, conferring more efficient replication and transmission in the ferret model (66). Additional to this HA mutation, an E627K change in the polymerase basic 2 (PB2) protein in H5N1 HPAI was also shown to be critical for transmission in ferrets (64, 102–105).

In addition to HA-sialic acid binding, key to viral fitness is the presence of a multi-basic cleavage site (MBCS), which has been shown extensively with H5N1, that the presence of an MBCS often dictates the use of the term HPAI (106, 107). The presence of an MBCS in the HA of influenza A viruses allows HA cleavage by additional enzymes such as furin-like proteases, whereas without an MBCS, the virus relies on only trypsin-like proteases (108). This flexible range of enzyme activity results in the virus being able to infect a greater range of cells and can lead to systemic infection (106, 107, 109). Though commonly associated with H5N1, this motif is seen in other avian-infecting HPAI viruses such as H7N3, however, these strains are less frequently transmitted to humans (19, 110). Interestingly, LPAI can also acquire MBCS motifs, changing the pathogenicity of the virus from low to high, such as in the case of an H7N8 outbreak in Turkey in 2016, where an LPAI virus caused a severe outbreak in the poultry due to the spontaneous addition of an MBCS, leading to over 800 bird deaths (111). However, the addition of an MBCS does not guarantee an LPAI virus to increase its pathogenicity, as recombinant H5 and H7 viruses do not exhibit HPAI pathology in chickens specifically due to the addition of an MBCS (112), which suggests that these motifs are one of many factors contributing to HPAI pathogenesis. H7N9 has also been shown to be able to obtain an MBCS to become highly pathogenic in chickens (113). Imai and colleagues showed increased disease severity of an MBCS-containing H7N9 virus in the ferret model compared to an LPAI H7N9 virus (18). However, H7N9 will still cause severe disease without an MBCS in most human cases, highlighting how unique this virus is in the AI landscape for its ability to show HPAI-like symptoms in mammals, while maintaining low pathogenicity in birds. Similarly for H5Nx viruses, several additional changes in the H5 HA protein, such as an N158D mutation, also allow greater replication of the virus in ferrets without the need for an MBCS, which combined with reassortment with human-adapted H1N1 gene segments shows the potential for these viruses to not only cross into humans but also cause severe, sustained human infection without systemic spread (114). This, coupled with H7N9's ability to cause severe disease without the requirement of an MBCS, suggests that while the MBCS still acts as a key virulence factor for AI viruses, it is not a sole-determining factor for HPAI in humans.

A key interaction between host and viral proteins is the interplay between Mx GTPases and viral nucleoprotein (NP), which may be pivotal in determining viral fitness. Human MxA and murine Mx1 protein have been shown to confer antiviral protection against influenza A viruses by interfering with the ability of NP to localize to the nucleus, inhibiting the viral replication cycle (115). While many influenza viruses are susceptible to Mx restriction, changes in the NP have been shown to confer resistance to this form of protection, particularly in the case of AI viruses such as H5N1 which shows greater susceptibility to MxA inhibition than pandemic H1N1 (116, 117). Moreover, LPAI H7N9 viruses were similarly shown to be affected by Mx1 in infected mice compared to H5N1 by Deeg and colleagues, who showed that these viruses required human adaptive motifs in their NPs to evade Mx restriction (118). Interestingly, avian species such as chickens have been found to have an Mx that does not display strong antiviral properties, suggestive of why these avian viruses do not commonly have NP capable of MxA evasion (119, 120). Therefore, while these AI viruses often appear to cause worsened disease progression due to their differences to human-infecting strains, in the case of Mx restriction it is a lack of human adaptation that may provide a level of protection to mammalian hosts.

Another key virulence factor elucidated in recent years is the ability of the non-structural protein 1 (NS1) to aid viral escape in the host immune system. In avian hosts, NS1 has been associated with worsened pathology through increases in iNOS and oxygen-reactive species for both LPAI H9N2 (121) and HPAI H5N1 (122). However, in contrast in mammals, the NS1 protein has been more closely associated with inhibition of host IFN responses. Jia and colleagues (123) showed that the H5N1 NS1 protein inhibits IFN production through interference with the JAK/STAT pathway. They found that expression of NS1 in HeLa cells prevented STAT phosphorylation and upregulated inhibitors of this pathway to prevent expression of IFNAR and SOCS3 proteins, which generally upregulate IFN expression (123). Furthermore, a naturally occurring deletion in the H5N1 NS1 effector domain can attenuate the virulence of the virus in both chickens and mice, suggesting that this protein is critical in the ability of H5N1 to suppress host immune antiviral responses across hosts (124). The pathogenicity of H5N1 in mice is also affected by the NS1 protein, as a single mutation (P42S) conferred greater pathogenicity to the virus by preventing nuclear factor-κB (NF-κB) and interferon-regulatory factor 3 (IRF3) signaling, and thus inhibiting IFN responses (125). Interestingly, in cats the NS1 protein can be associated with blocking NF-κB and IRF3 signaling in response to the emerging HPAI H5N6 virus, with inhibition of the IFN-β promoter blunting the feline IFN response (126), suggesting that NS1 may have different ways of interacting with influenza hosts across species to produce similar immune suppression. On the other hand, Thube and colleagues investigated the IFN responses of HPAI H5N1 compared to LPAI H11N1 and suggested that decreased IFN signaling occurred independently of NS1, suggesting other viral elements can also induce a reduced antiviral state in cells (127). It is worth noting that while the NS1 of LPAI H7N9 is inefficient at binding to CPSF30 (involved in pre-mRNA processing), a single I106M mutation restores CPSF30 binding to NS1 thereby blocking the expression of host antiviral genes. This renders the virus more virulent than other LPAI infections (128), which may explain why H7N9 causes more severe disease in humans compared to other LPAIs. These results also highlight inhibition of IFN-activation pathways as an important viral factor in preventing host immune responses to infection, allowing for more productive infection and potentially more severe clinical outcomes.

# HOST SUSCEPTIBILITY FACTORS

In addition to the ability of the virus to gain function through mutation, in recent years there has been an increasing focus on how host genetic factors can lead to changes in resistance or susceptibility to influenza A viruses. While many factors have been identified in preventing influenza A infection, a key which has come to light for AI host/pathogen interactions is the interferoninduced transmembrane (IFITM) protein family, which unlike other factors such as MxA seems to be predominantly the host, rather than the virus, that seems to control whether the virus is able to replicate. IFITMs are family of transmembrane antiviral proteins that are stimulated by the presence of elevated IFN levels, giving another reason why so many AI viruses attempt to quash the IFN response (129, 130). The IFITM proteins can interfere with viral entry to the cytosol *via* cell membranes (131), with Brass and colleagues showing that overexpression of human IFITMs 1, 2, and 3 effectively blocked infection with several influenza A pseudoviruses (retroviruses expressing influenza surface proteins), including those enveloped with H5 and H7 proteins (132). Moreover, IFITM3 specifically localizes to the endosomes due to phosphorylation of the Y20 tyrosine residue, enabling these proteins to intrinsically target pH-dependent viral pathways such as that seen with influenza A viruses (133).

The IFITM3 molecule can play a significant role in mice and human influenza infections. Mice inoculated with influenza antigen showing higher IFITM3 expression in the lungs developed a more robust lung tissue-resident memory CD8<sup>+</sup> T cell response as well as a longer duration of response even following reduction of IFN-α, suggestive of this molecule playing a role in not only innate immunity but also adaptive immunity as well (134). Of particular note, a single-nucleotide polymorphism (SNP) in the IFITM3 gene, *rs12252-C*, has been shown to strongly correlate to worsened disease progression, as this SNP leads to a truncated splice-variant that affects the protein's ability to localize to the membrane (135, 136). IFITM3 protein dysfunction can be associated with severe hypercytokinemia and worsened disease progression in H7N9-infected hospitalized patients. In support, Zhang and colleagues showed that the *rs12252-C* mutation correlated to severe seasonal H1N1 influenza cases in Chinese populations where the mutation appeared in higher frequencies (136, 137). However, studies have shown that while this SNP does affect IFITM3′s ability to localize, restriction of the virus may continue with the variant protein or with a Y20A mutation to affect the localization of the full protein (138). Furthermore, a recent study by Makvandi-Nejad and colleagues found that primary cell lines homozygous for the *rs12252-C* SNP expressed the non-truncated mRNA transcript and thus expressed the wild-type IFITM3 protein at levels greater than 99% when compared to the truncated versions (139), which may provide insights into why the *rs12252-C* mutation appears not to act as a significant risk factor in Caucasian populations, but perhaps in the Chinese patients. Moreover, an additional SNP has recently been identified, *rs34481144-A,* which affects the promoter for the *IFITM3* gene, resulting in lower IFITM3 expression compared to hosts without the mutation. Interestingly, both the *rs12252-C* and *rs34481144-A* mutation were found to be non-overlapping, as the risk allele for one was inherited with the protective allele of the other, suggesting a multifaceted IFITM response to influenza A viruses (140).

IFITM1 and IFITM3 distribution in mice has been investigated in the context of H9N2 infection, with the distribution of these proteins correlating with increased restriction of the virus' entry into host tissues, due to upregulation in the lungs and peripheral tissues of BALB/c mice following inoculation. Interestingly, when infected with a H9N2 strain with higher pathogenicity and ability for systemic spread due to a K627E mutation (rVK627E) in the viral PB2 protein, compared to the wild-type strain, IFITM3 was upregulated accordingly in the brain of rVK627E-infected mice to combat this virus' ability to cause viral encephalitis (141). IFITM3 responses to H5 and H7 proteins have also been assessed in pigs and bats, with both these HA subtypes showing restricted entry due to the action of these IFITM3 proteins (142), suggestive of the broad action of IFITM molecules across species known to contract and potentially disseminate AI viruses. That IFITMs restrict AI viruses in "mixing vessel" species (i.e., pigs and bats) suggests that these proteins may have a key role in preventing the spread of human-infecting AI viruses through these routes, as well as contributing to these species showing lessened pathology compared to other mammalian species.

The role of IFITMs against AI viruses in avian species has not been clearly defined, though a study by Smith and colleagues has shown that chicken IFITM3 (which is "human IFITM1 like") similarly restricts H5 and H7 expressing viruses (143). However, when chickens were infected with H5N1, expression of IFITM molecules was not highly upregulated compared to other human-infecting seasonal strains such as H1N1, with IFITMs showing the weakest inhibiting effect against H7N9 (144). Hence, IFITM expression in chickens appears to be limited and does not vary greatly whether the virus is of low or high pathogenicity. Conversely, the IFITM expression profile observed in ducks is far more robust and variable between viral strains, whereby infection with LPAI H5N2 virus was reported to consistently cause a 3-fold increase in IFITM1, 2, and 3 expression levels in the lungs and ileum on day one postinoculation, while infection with the HPAI H5N1 virus caused up to a 93-fold increase in IFITM3 expression in the lungs in a similar time frame (145). Therefore, understanding the differences in variable IFITM expression, and the reasons why chickens mount a lesser IFITM response to influenza viruses, may prove pivotal in understanding why some birds succumb to HPAI infection while others survive.

In addition to specific IFITM mutations, broad immunodeficiency can lead to worsened pathology and disease outcomes in humans and animals. For example, immunocompromised patients who contract LPAI viruses such as H9N2 suffer severe respiratory distress, and though many of these LPAI remain mild even in immunocompromised patients, H9N2 induces stronger cytokine responses than seasonal influenza viruses (146, 147). These trends are also observed in avian hosts, in which Nili and Asasi found that chickens co-infected with other pathogens such as *M. gallisepticum* showed worsened clinical signs such as severe necrotizing tracheitis, leading to a 19% mortality in the flock (148, 149). Therefore, it is apparent that both the host and pathogen can contribute to perturbed inflammation and severe disease outcomes. In addition, in humans, more severe AI subtypes have been associated with mortalities in very distinct age demographics when compared to seasonal influenza (**Figure 2**), and often manifest in age groups not commonly associated with immunodeficiency. Fatal cases in children (0–9 years) and younger adults (10–19 years) were predominantly caused by HPAI H5N1 infection, with 80.3% of H5N1 fatal cases seen in people aged 35 or under (12). Interestingly, a similar trend was observed with the pandemic [pdmH1N1(2009)] strain, which led to increased infection in younger age groups compared to

Figure 2 | Age-related mortality trends highlight impact of host–pathogen relationships. Frequency of age groups of patients who succumb to different strains of influenza is graphed as a proportion of total fatalities for a given strain. When we assessed the age of patients who succumb to different strains of influenza, as a proportion of the total mortalities for a given strain, trends emerge as to the host susceptibilities. For seasonal influenza, older patients (>60 years old) were the most susceptible, however, for a variation on seasonal influenza, pdmH1N1 2009, the age of patients who succumbed was reduced and included significant mortalities between 20 and 59 years old. Interestingly, the highly pathogenic AI H5N1 was predominantly fatal in those under 40 years old, whereas H7N9, a low pathogenicity AI strain, followed a similar trend to seasonal influenza.

other seasonal strains circulating at the same time (150). An interesting observation from the 2009 pandemic, however, was that disease outcomes were more mild in newly weaned ferrets infected with pdmH1N1(2009) as well as in younger children, suggesting the immune response in younger individuals may have a protective response to this strain (151, 152). Conversely, seasonal strains of influenza disproportionately impact older ferrets infected, which display a greater degree of morbidity and reduced HA and T cell responses (153). This is also true of human patients also whereby older patients (>60 years of age) are most susceptible to seasonal influenza strains (80% of mortality cases). With regards to H7N9 LPAI infections, the highest mortality was also skewed toward older people, which is likely due to the propensity for H7N9 cases to be found in live poultry farmers, typically older men (154, 155). This emphasizes the dynamic relationship between the pathogen and the human host, and how different strains of influenza viruses can lead to differential fatal outcomes across different ages, a concept further explored by Gostic and colleagues who suggest that pre-immunity to these viruses may confer differing levels of protection to AI based on whether they have been exposed to the viral HA class while still young (156).

While IFITM proteins aim to restrict viral entry, the interaction between influenza virus peptides and major histocompatibility complex (MHC) molecules is a key host–pathogen interaction affecting the outcome of disease following initial

which can or cannot present the antigen appropriately and efficiently to host T cells. Ultimately, these and other elements lead to changes in production of key

cytokines as well as cellular activation that drives inflammation, cell death, and clinical manifestation of disease.

infection. In humans, MHC molecules are encoded for by the human leukocyte antigen (HLA) system, with a vast array of alleles in the genes encoding for the molecules responsible for the recognition of antigenic peptides. As such, different HLA subtypes confer different levels of susceptibility to influenza A viruses, as not all HLA subtypes respond to influenza peptides in the same way. For example, human populations expressing HLA-A\*02:01 can elicit strong, cross-protective CD8<sup>+</sup> T cell responses following presentation of the internally conserved M158–66 epitope (157–160); this epitope is one of the most immunogenic influenza peptides observed in humans with HLA-A\*02:01 being the most common HLA alleles expressed worldwide (157). Moreover, individuals lacking common HLA alleles may be at greater risk of influenza A infections, such as those carrying the HLA-A\*24:02 allele, associated with increased mortality in individuals infected with pandemic H1N1 virus (161). Indigenous populations in particular are susceptible to influenza A due to a lowered prevalence of protective HLA variants toward these viruses (160). Wang and colleagues suggest that for AI viruses such as H7N9, MHC-interactions with internally conserved epitopes from other influenza A strains, such as pandemic H1N1, may explain why some populations show greater immunity through cross-reactive CD8<sup>+</sup> T cell responses than others (72). However, some H7N9 peptides may have cross-conservation with human host proteins that the immune system recognizes as "self," leading to an attenuated immune response to the virus and thus worsened disease progression due to T cell-mediated tolerance (162).

Interestingly, chickens show a restricted repertoire of MHC alleles compared to other species, and these haplotypes themselves are poorly characterized. As such, a few studies have been conducted into characterizing T cell epitopes in response to H5N1 infections, one such study predicting 25 potential T cell epitopes in the NP of H5N1 in four haplotypes (163). More recently, experiments into epitopes of a specific haplotype, BF2\*15, further characterized NP epitopes that may lead to protective immunity against H5N1 (164). This limited repertoire may explain why chickens show more severe clinical outcomes due to HPAI such as H5N1 compared to waterfowl, as ducks show extensive diversity in their MHC class I alleles which allows the immune system greater coverage for viral variation (165). A recent investigation into duck MHC class I molecules found that the duck *Anpl*-UAA\*01 complex showed similar peptide binding properties to HLA-A\*02:01 in humans and as such appears to cover a greater array of influenza A virus epitopes compared to similar chicken MHC molecules such as BF2\*2101 (166). Furthermore, migratory shorebirds which act as reservoirs for AI viruses (in particular LPAI H9N2) show increased diversity in their MHC alleles, likely as a mechanism for protecting against foreign pathogens that may be encountered during migration. A study in red knots found high MHC diversity with 36 alleles detected across eight birds, which when correlated to their low prevalence of shed AI virus and high antibody titers to AI viruses, they could mount effective immune responses toward these viruses, possibly *via* cytotoxic T lymphocyte responses recognizing novel peptide/ MHC complexes (167). Based on a culmination of human and animal evidence, the interactions between the various pathogen and host factors contributing to human influenza severity are summarized in **Figure 3**.

# CONCLUSION

The emergence of AI viruses is of major concern to the avian and human population. The lack of pre-existing antibody immunity and their ability to cause severe disease through multiple host and viral mechanisms makes these viruses difficult to counter. Currently, these viruses are yet to effectively replicate and transmit between humans, however, experiments in ferrets show that only a few mutations are needed for H5N1 and H7N9 viruses to quickly adapt and become a major pandemic threat (114, 168). Their ability to pass from birds to mammals commonly in contact with humans requires constant surveillance across all known bird reservoirs to limit the potential threat of an AI-derived pandemic. Characterization of the interactions between AI viruses and their hosts and how they illicit different degrees of clinical manifestations across species is of

# REFERENCES


utmost importance. Here, we extend our current knowledge of the "cytokine storm" model of AI pathogenesis and delve into more complex underlying viral and host genetic factors that may also contribute greatly to disease severity and susceptibility outcomes.

# AUTHOR CONTRIBUTIONS

WH and DL wrote and edited the manuscript. AB, TN, and KK edited the manuscript.

# FUNDING

KK is supported by a NHMRC SRF Level B Fellowship #APP1102792. WH is supported by CSIRO and The University of Melbourne 'One Health PhD Scholarship'.


(Vietnamese and Indonesian) in Pekin ducks (*Anas platyrhynchos*), with particular reference to clinical disease, tissue tropism and viral shedding. *Avian Pathol* (2009) 38(4):267–78. doi:10.1080/03079450903055371


dysfunction and predictive of fatal H7N9 infection. *Proc Natl Acad Sci U S A* (2014) 111(2):769–74. doi:10.1073/pnas.1321748111


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Horman, Nguyen, Kedzierska, Bean and Layton. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Influenza Virus: A Master Tactician in Innate Immune Evasion and Novel Therapeutic Interventions

*Alan Chen-Yu Hsu1,2\**

*1Viruses, Infections/Immunity, Vaccines & Asthma, Hunter Medical Research Institute, Newcastle, NSW, Australia, 2Priority Research Centre for Healthy Lungs, Faculty of Health and Medicine, The University of Newcastle, Newcastle, NSW, Australia*

Influenza is a contagion that has plagued mankind for many decades, and continues to pose concerns every year, with millions of infections globally. The frequent mutations and recombination of the influenza A virus (IAV) cast a looming threat that antigenically novel strains/ subtypes will rise with unpredictable pathogenicity and fear of it evolving into a pandemic strain. There have been four major influenza pandemics, since the beginning of twentieth century, with the great 1918 pandemic being the most severe, killing more than 50 million people worldwide. The mechanisms of IAV infection, host immune responses, and how viruses evade from such defensive responses at the molecular and structural levels have been greatly investigated in the past 30 years. While this has advanced our understanding of virus–host interactions and human immunology, and has led to the development of several antiviral drugs, they have minimal impact on the clinical outcomes of infection. The heavy use of these drugs has also imposed selective pressure on IAV to evolve and develop resistance. Vaccination remains the cornerstone of public health efforts to protect against influenza; however, rapid mass-production of sufficient vaccines is unlikely to occur immediately after the beginning of a pandemic. This, therefore, requires novel therapeutic strategies against this continually emerging infectious virus with higher specificity and cross-reactivity against multiple strains/subtypes of IAVs. This review discusses essential virulence factors of IAVs that determine sustainable human-to-human transmission, the mechanisms of viral hijacking of host cells and subversion of host innate immune responses, and novel therapeutic interventions that demonstrate promising antiviral properties against IAV. This hopefully will promote discussions and investigations on novel avenues of prevention and treatment strategies of influenza, that are effective and cross-protective against multiple strains/subtypes of IAV, in preparation for the advent of future IAVs and pandemics.

Keywords: influenza, influenza A virus, virulence factors, hemagglutinin, polymerase acidic 1-F2, NS1, therapeutics

# INTRODUCTION

Influenza A viruses (IAVs) and their continuous emergence/re-emergence are undoubtedly an important cause of global concern, morbidity and mortality, clinical, and socio-economical burden in humans. The emergence of highly pathogenic avian influenza virus H5N1 in 1997 and 2003, H7N7 in 2003, recent influenza H1N1 pandemic in 2009 (H1N1pdm09), and H7N9 in 2013 have caused tremendous mortality in the affected populations, and in our current inter-connected world, the concerns and impact of potential pandemics are truly global.

## *Edited by:*

*Amy Rasley, Lawrence Livermore National Laboratory (DOE), United States*

#### *Reviewed by:*

*Katie Louise Flanagan, RMIT University, Australia Irving Coy Allen, Virginia Tech, United States*

*\*Correspondence: Alan Chen-Yu Hsu alan.hsu@newcastle.edu.au*

#### *Specialty section:*

*This article was submitted to Microbial Immunology, a section of the journal Frontiers in Immunology*

*Received: 09 January 2018 Accepted: 26 March 2018 Published: 12 April 2018*

#### *Citation:*

*Hsu AC-Y (2018) Influenza Virus: A Master Tactician in Innate Immune Evasion and Novel Therapeutic Interventions. Front. Immunol. 9:743. doi: 10.3389/fimmu.2018.00743*

The ongoing circulation of IAVs in their natural hosts and the ever-mutating antigenicity frequently leads to the appearance of novel viruses with the ability to cross species barriers and infect humans with unpredictable pathogenicity. Furthermore, effective immuno-evasive and modulatory strategies employed by IAVs have made it difficult for both the innate and acquired immunity to combat the infections without the help of vaccination.

The early innate immune responses to IAV infections are essential in the immediate control of viral replication and spread. However, IAVs have also evolved to produce only a handful, but multi-functional proteins to combat the intricate layers of human innate immune signaling pathways. These virulence factors inhibit host antiviral immunity, while stimulating inflammatory responses in the airways (cytokine storm) that cause severe symptoms. It is, therefore, imperative to understand the molecular mechanisms of IAV-mediated modulation of innate immunity by these viral factors in order to identify potential therapeutic options.

While antiviral drugs are mostly used either prophylactically or in treatment, they have minimal impact on the course of the disease. Heavy reliance on antiviral drugs has also placed a strong selective pressure on IAVs to mutate and develop resistance (1, 2). The unpredictable nature of novel IAVs and limited prevention and treatment options, therefore, urges the research and development of novel approaches against IAVs. This review summarizes the current understanding of the mechanisms of IAV infection in humans, the ways by which IAV suppresses antiviral immunity and causes the inflammatory cytokine storm, and novel peptidebased anti-influenza drugs that may potentially be beneficial as preventative and treatment strategies in current and future IAV pandemics.

# IAV—SIMPLE AND ELEGANT

IAVs is a negative sense, single-stranded RNA virus (~80–120 nm in diameter) (3). The viral envelope features two surface glycoproteins, hemaglutinin (HA) and neuraminidase (NA), at a ratio of four HA to one NA (4). A small number of M2 ion channel is also embedded in the viral envelope at a ratio of one M2 channel to 101 –102 HA molecules (5) (**Figure 1**).

Within the envelope are the eight segments of influenza RNA, which are presented in a form of helical hairpin structure, each of which is coated with arginine-rich nucleoprotein (NP) (6–8). NP-RNA is complexed with heterotrimeric RNA-dependent RNA polymerase, which consists of three subunits, polymerase basic (PB)1, PB2, and polymerase acidic (PA) (3). Segment 1, 3, 4, 5, and 6 encodes for PB2, PA, HA, NP, and NA, respectively. Segment 2 encodes for PB1 and *via* a different reading frame an accessory protein PB1-F2 is expressed in most influenza A strains. Segment 7 encodes for matrix 1 (M1) protein and also M2 ion channel *via* alternative splicing. Segment 8 encodes for non-structural protein (NS)-1 and nuclear export protein. At

Figure 1 | Influenza A virus (IAV) structure. IAV contains an outer membrane envelope with hemaglutinin (HA), neuraminidase (NA), and matrix 2 (M2) ion channels, and an inner M1 protein layer that encloses an eight segmented RNA genome. The RNA segments are folded into a helical hairpin structure with nucleoprotein, which are complexed with heterotrimeric RNA-dependent RNA polymerase [polymerase basic (PB1), PB2, and polymerase acidic (PA)].

both 3′ and 5′ ends of each segment lies a non-coding region of varying length that acts as a promoter site for viral polymerase complex to initiate transcription. This region also contains an mRNA polyadenylation signal and a signal for virus assembly.

The primary site of infection by IAVs are epithelial cells of the respiratory mucosa. Airway epithelial cells are both susceptible and permissive to IAV infection. This occurs by the binding of HA to the sialyl sugar chain receptors on the host cell surface, allowing the virus to be internalized into endosomes in the host epithelial cells. The low pH environment of the endosome promotes HA to undergo conformational change that liberates and inserts a fusion peptide from the amino terminus of HA into the endosomal membrane. This spring-loaded mechanism fuses the viral envelope and the membrane together, thereby releasing viral RNP into the host cytoplasm (9, 10). The M2 ion channel also allows an influx of H<sup>+</sup> ions into the virion, and lowers the intra-virionic pH. This in turn disrupts the viral RNP-M1 protein interaction and subsequently releases viral RNP into host cellular cytoplasm (11–14). Released IAV RNAs and polymerases are then translocated into the nucleus where viral replication occurs. The newly synthesized viral structural proteins and viral segments then traffic to lipid rafts on the plasma membrane to be released (15, 16). Since the viral envelope is derived from the host membrane, which contains sialic acid glycoproteins, the newly formed virion remains intact on the host cell surface. The viral NA cleaves the host cell surface sialic acid residues, releasing the newly formed virions free from the host cell surface.

# HA—STRUCTURALLY PLASTIC AND ABSOLUTELY ESSENTIAL FOR INFECTIVITY, HUMAN TRANSMISSION, AND PANDEMIC

IAVs HA is responsible for the entry of the virus in the host cells by binding to host cell surface glycoproteins terminated with sialic acid residues at specific linkages. Human IAVs preferentially bind to glycoproteins containing the terminal SAα2,6Gal linkage, which are predominately found in human upper airway epithelium (17–19). In contrast, avian IAVs bind to that with terminal SAα2,3Gal linkage in the lower airways (17–21). This difference in binding specificity and distribution of sialic acid residues may in part explain why highly pathogenic avian influenza virus H5N1 is currently incapable of transmitting from human to human in a sustainable manner. In contrast, pig trachea contains both SAα2,6Gal and SAα2,3Gal linkages, indicating that pigs can act as an intermediate mixing host (22–24), allowing the reassortment of both avian and human viruses to occur. The prime example is the emergence of the 2009 H1N1 pandemic, which was the result of triple reassortment between avian, swine, and human IAVs in pigs (25, 26). The difference in their binding specificity can be explained by the amino acid residue at position 226 of HA glycoprotein. HA of human influenza viruses contains a Leu226 that results in the preferential binding to SAα2,6Gal linkage. In contrast, HA of avian influenza viruses have a Gln226, which binds to SAα2,3Gal-linked glycoproteins (27, 28).

Two recent important investigations further characterized the molecular changes that are required for H5N1 to become transmissible among humans. Imai et al. showed that Gln226Leu/Gly228Ser increased the binding of H5 HA to SAα2,6Gal while retaining the binding capacity to SAα2,3Gal. When Asn158Asp/Asn224Lys/Thr318Ile was introduced with Gln226Leu/Gly228Ser, this fully converted the virus with H5 HA in an H1N1pdm09 backbone to achieve sustainable aerosol transmission in a ferret model (29). In addition, Herfst et al. also demonstrated that mutant H5N1 containing His103Tyr, Thr156Ala, and Gln222Leu in the HA protein, and Glu627Lys in the viral polymerase protein PB2 was able to efficiently transmit between ferret models *via* aerosols (30). Interestingly, although not surprising, most of these residues (Asn158Asp; Gln222Leu; Asn224Lys; Gln226Leu; Gly228Ser) are at the sialic acid residue binding site of the globular domain.

When these amino acid residues in H5N1 HA were changed *in silico* to the ones in the airborne-capable HA, and structurally modelled using the wild-type H5N1 HA as a template in SWISS-MODEL (31), the binding cavity in the mutant, surrounded by the 130-, 190-, and 220-loop, may potentially be lengthened and result in a slightly larger binding cavity (**Figure 2**). Furthermore, Asn158Asp is also likely to make the receptor binding pocket slightly more acidic, further facilitating a more efficient airborne transmission. A mutant H1N1 with Tyr17His (pH 6.0) in HA

showed loss of airborne transmission, and was less efficient in contact transmission with reduced pathogenicity in mice, and His17Tyr (pH 5.3) revertant virus regained the airborne transmissibility and pathogenicity (32). This may also widen the binding capacity toward moieties with other linkages/modifications. While changes in these residues are important in airborne transmission in ferret models, it remains unclear whether similar or other changes in HA need to occur for a sustainable airborne transmission in humans, these studies provide valuable insight into cause of efficient human transmission, surveillance for potential human transmissible IAVs, and new approach into therapeutic designs.

Epidermal growth factor receptor, a glycoprotein with potential terminal SAα2,6Gal and SAα2,3Gal linkages (33, 34), has been shown to be important in IAV entry into host cells in a ligand (EGF)-independent, and phosphoinositide-3 kinase (PI3K)-activation-dependent manner (35–37). However, IAV has also been demonstrated to cause infection in the absence of its respective receptors. H3N2 and H1N1 was able to infect and replicate to the similar titer in the lung of the mice lacking receptors with SAα2,6Gal linkages compared to wild type mice (38). Consistent with this finding, while human bronchial epithelial cells showed higher levels of SAα2,6Gal compared with SAα2,3Gal linkages, human IAV H3N2, and a low pathogenic avian H11N9 have been shown to enter into bronchial epithelial cells at a similar rate (39). This indicates that sialic acid residues may not be absolutely essential in influenza virus entry, and other moieties, such as sulfonated or fucosylated glycoproteins may also be targeted by HA protein (40, 41).

## THE POLYMERASE CONSTELLATION

IAV RNA is always packaged in a heterotrimeric RNA-dependent RNA polymerase complex with viral polymerase basic (PB)1, PB2, polymerase acidic (PA), and nucleoprotein (NP), and upon successful entry into epithelial cells, this complex translocates to the nucleus where replication occurs. As IAV viral RNA does not contain a 5′ cap as they do on host RNAs, PB2 contains a capbinding domain that binds to the 5′ cap of nascent host RNAs, and PA then "steals" the cap off the host RNA by cleavage, which is then used as primers to initiate transcription of viral mRNAs. This process is called cap-snatching. This polymerase complex, therefore, is critical in the survival of IAVs, and has been implicated in the host range determination. Clements et al. demonstrated that the reassortment virus containing the PB2 gene of avian origin and all other genes from human IAV is able to replicate efficiently in the avian host, while showing restricted replication in the mammalian respiratory tract (42). This reassortment virus was then progressively passaged to generate mutant viruses that were able to replicate in mammalian tissues. Nucleotide sequence analysis revealed that this phenotype of the reassortment virus was due to the Glu627 of the PB2 gene (43). A single amino acid substitution to a lysine residue allowed this mutant reassortment virus to replicate efficiently with increased pathogenicity in the mammalian system (44–47). All the avian IAVs analyzed to date have a Glu627 in the PB2, whereas human IAVs have a Lys627, indicating this residue at position 627 of PB2 is an important host range determinant of IAVs (43). Furthermore, Glu627Lys substitution being a prerequisite for aerosol to airborne transmissibility switch is troubling, as the viruses isolated from the recent fatal cases of H7N9 infection in China carried Glu627Lys substitution, but the viruses isolated from poultry did not (48, 49).

The compatibility of other subunits of the polymerase complex is also involved in the host range specificity (50, 51). Reassortment viruses with human PA and avian PB1 and PB2, or with human PA and PB2 and avian PB1 showed a restricted viral replication in mammalian cells, indicating that a specific constellation of polymerase genes may be involved in the host range specificities (51).

# PB1-F2—A PYROGENIC DEATH DEALER

In the middle of searching for an IAV-derived out-of-frame polypeptide that can be recognized by CD8<sup>+</sup> T lymphocytes, an 11th gene, PB1-F2, was serendipitously discovered (52). PB1-F2 is an accessory protein that is translated from an alternative +1 open reading frame of PB1. It modulates innate immune responses at multiple signaling levels, leading to increased inflammatory and impaired antiviral responses. PB1-F2 contains a mitochondrial targeting sequence at the carboxyl-terminus and has a proapoptotic effect on immune cells (53). By binding to the inner and outer mitochondrial membrane transport protein adenine nucleotide translocator 3 and voltage-dependent anion channel 1 (VDAC1), PB1-F2 disrupts mitochondrial integrity, releasing cytochrome C, and leading to apoptosis. Another study also demonstrated that PB1-F2 can be translocated into the inner mitochondrial space in a TOM40-dependent manner, and impair mitochondrial function (54).

Host innate immune response is triggered by the pattern recognition receptors including retinoic acid inducible gene 1 (RIG-I). RIG-I binds to IAV RNAs and forms a complex with the adaptor protein tripartite motif-containing protein (TRIM)25 (55). This complex interacts with mitochondrial antiviral signaling (MAVS) protein to induce the activation of interferon (IFN)-regulatory factor (IRF) 3, which initiates the induction of important antiviral cytokines including type I and III IFNs (56, 57). These IFNs then stimulate the expression of over 300 IFN-stimulated genes (ISGs) such as protein kinase R (PKR) to limit viral replication (58). Toll-like receptor 3 also recognizes IAV RNAs and activates an important transcription factor nuclear-factor-kappa-B (NF-κB), which stimulates the expression of predominantly inflammatory cytokines, including interleukin (IL)-6 and IL-1β (59–61).

PB1-F2 has also been shown to inhibit RIG-I-TRIM25 mediated antiviral signaling pathway by direct interaction with MAVS (62, 63). Furthermore, this inhibition of type I and III IFNs was enhanced with Asn66Ser substitution. Intriguingly, PB1-F2 can also be expressed as a truncated (57 amino acids) protein devoid of the carboxyl terminus, which failed to translocate into the mitochondria (54). The full length PB1-F2 is mostly expressed by highly pathogenic viruses, such as H5N1, whereas low pathogenic subtypes, such as H1N1 (except PB1-F2 of 1918 H1N1) tend to express this truncated form. PB1-F2 also enhances the development of secondary bacterial pneumonia in mice with reduced survival rate. When PB1-F2 of A/PR/8 was engineered to become that of 1918 H1N1 by mutagenesis, the resulting mutant virus became more pathogenic with and without secondary bacterial infection (64). Surprisingly, this increased pathogenicity by PB1-F2 of 1918 H1N1 was also attributed to Asn66Ser (65). A mutant virus carrying H5N1 PB1-F2 gene with Asn66Ser in A/WSN/33 background led to increased pathogenicity with 100 fold increase in viral replication compared with wild-type virus in mice. Similarly, when the 1918 H1N1 virus was reconstructed to carry a pathogenicity-reducing substitution (Ser66Asn), the resulting virus led to attenuated viral replication and reduced mortality.

PB1-F2-mediated IAV pathogenicity could be the result of its interaction with NLRP3-inflammasome. PB1-F2 of H7N9 has been shown to form fibrillar higher molecular weight aggregates that were incorporated into the phagolysosome and induced ASC speck formation upon acidification (66). This resulted in the activation of NLRP3-mediated inflammasome, leading to increased production of pyrogenic cytokine IL-1β and enhance the pathogenesis of influenza viral pneumonia in mouse models (66, 67). PB1-F2 of avian H7N9 has also been shown to increase mitochondrial reactive oxygen species (ROS) and calcium (Ca2<sup>+</sup>) efflux, which contributed to the activation of NLRP3 inflammasome (68).

PB1-F2, therefore, appears to be a major driving force for excessive pro-inflammatory cytokine storm and pathogenesis (**Figure 3**), and Asn66 is important in the function of PB1-F2. Nevertheless, how IAV-mediated inflammatory cytokine storm holds an advantage in IAV survival remains elusive. The formation of this fibrillary PB1-F2 aggregate is consistent with the crystal structural data that indicated a strong propensity for PB1-F2 to adopt an oligomerization structure and form membrane pores in planar lipid bilayers (69, 70). However, it remains unknown whether it is a requirement for PB1-F2 to form this fibrillar structure to interact with all of its binding targets, and while the oligomerization domain is located in carboxyl-terminal α-helix, it is unknown if oligomerization is dependent on Asn66. It is also unclear if this structure is unique to highly pathogenic strains or is present in all subtypes/strains.

# NS1—THE INVERSE INTERFERON

When IFNs were first described by Isaacs and Lindenmann, it was found that cells treated with heat-inactivated, but not live IAV could produce these antiviral cytokines, an immuno-inhibitory phenomenon he described as "inverse interference" (71, 72). The identity of Lindenmann's inverse interferon was later revealed to be NS1 protein. NS1 protein is encoded by the NS gene together with nuclear export protein *via* alternative splicing. As NS1 is encoded by the shortest gene of all, NS1 protein is rapidly expressed to high levels in the infected epithelial cells (73, 74).

NS1 is a multi-functional protein with a RNA-binding domain at its amino-terminus (residue 1–73), and an effector domain (residue 74–230) at the carboxyl-terminus (73, 75, 76). The RNA-binding domain of NS1 recognizes dsRNA sequences and blocks the host RNA detection system. The effector domain of NS1 can stabilize the RNA-binding domain, but it predominantly interacts with host cellular proteins and interferes with host mRNA processing and host innate immune responses (**Figure 4**). NS1 inhibits host pre-mRNA endonucleolytic cleavage and polyadenylation by direct interaction and inhibition of cleavage and

terminal effector domain. RNA binding domain binds to viral RNAs and prevents detection by the host pattern recognition system. The effector domain binds to TRIM25, and inhibits RIG-I-MAVS interaction and downstream induction of type I and III IFNs. This domain also interacts with protein kinase R (PKR), therefore, inhibiting its pro-apoptotic activities. PI3K-p85β subunit is also targeted by the effector domain to inhibit apoptosis. NS1 binds with CPSF30 and poly A binding protein II (PABII), inhibiting host protein synthesis and facilitating viral protein synthesis. The ARSK tail is exclusively found in H3N2, and directly binds with hPAF1 complex and inhibits the transcription of antiviral genes. Protein visualization was performed in UCSF chimera molecular visualization application.

polyadenylation specificity factor −30 subunit (CPSF30) and poly A binding protein II (77–79). By shutting down the host cellular protein synthesis in the infected cells, this helps the virus to gain control of the host machineries required for efficient viral protein synthesis.

The major function of NS1 is to antagonize host innate immune responses during infection, and this occurs at multiple stages of the IFN signalling cascade. In the nucleus, NS1 of human IAV H3N2 was recently shown to contain carboxyl-terminal ARSK tail sequence (amino acids 226–229), which is analogous to the ARTK sequence on the lysine 4 of histone H3 (H3K4) (80). This ARSK tail could act as a H3K4 histone mimic that directly interacts with human polymerase II-associated factor 1 complex and impairs the transcription of antiviral genes. However, this ARSK tail appears to be unique to H3N2, and is absent in H5N1, H7N9, and H1N1pdm09. In the cytoplasmic space, NS1 inhibits RIG-Imediated signaling and subsequent induction of type I IFNs (81, 82). Specifically, the NS1 protein inhibits RIG-I ubiquitination mediated by TRIM25, which is crucial for maximal type I and III IFNs expression during viral infection (83). This NS1-TRIM25 binding event is dependent on the Arg38 and Lys41 in the RNA binding domain, and Glu96 and Glu97 in the effector domain of NS1 (84, 85). NS1 protein also interferes with functions of important intracellular antiviral ISGs, including PKR and oligoadenylate synthase (OAS). The RNA-binding domain of NS1 can bind to viral RNA and avoid detection by PKR (86). It also binds to PKR itself via the effector domain (residue 123–127) and inhibits PKR-mediated viral mRNA suppression and apoptosis (87–90). OAS, which detects and cleaves viral RNA by activating RNase L, can also be blocked by influenza NS1. The RNA-binding domain of NS1 can out-compete the RNA binding capacity of OAS, thereby inhibiting the antiviral response (91). Another important target of NS1 is the PI3K signaling pathway. NS1 activates PI3K pathway by direct interaction with p85β subunit, thereby increasing the rate of viral internalization and inhibition of apoptosis. NS1-p85β interaction is dependent on Tyr89/Met93 (92), Leu141/Glu142 (93), and Pro164/Pro167 (94) of the effector domain, all of which are located adjacent to each other within a cleft between the two NS1 monomers. The anti-apoptotic effect of NS1-PKR and NS1-p85β interaction, however, appears to conflict with the pro-apoptotic function of PB1-F2. The molecular equipoise of pro- and anti-apoptotic response by these two contradicting signals and the timing of apoptosis remain elusive.

The carboxyl-terminus of the effector domain is essential in the NS1-mediated inhibition of antiviral responses. A mutant H7N7 virus that lacked NS1 gene or with a large carboxylterminus deletion showed impaired IFN suppression activity and attenuated replication in mammalian and avian epithelial cells (95). Similar findings were also observed with influenza viruses of different species. Both swine and turkey influenza virus with NS1 carboxyl-terminal truncation showed attenuated replication and an increase in IFN response compared to the wild type infection in its respective epithelial cells (96, 97). When the NS1 gene of high pathogenic avian H5N1 was replaced with the one that had a natural deletion of residues 191–195 from low pathogenic swine H5N1, the resulting virus was attenuated in viral replication and its IFNs inhibition in chickens (98). The deleted residues were then engineered into this NS1 and the resulting virus gained virulence, demonstrating the importance of these residues at 191–195 in IFN inhibition. The overall suppression of host antiviral immunity by the NS1 protein appears to be more effective with human IAV H3N2 compared with a low pathogenic subtype (39). This difference in the levels of inhibition by the NS1 proteins was likely due to differences in the amino acid residues discussed previously, which might be driven by the selective pressure and the long evolution of the virus in human populations. The H5N1 NS1 was able to dramatically reduce the induction of antiviral cytokines, leading to much higher viral replication (99, 100). Influenza NS1 is no doubt a powerful and fast-deployable antagonist that rapidly controls critical host immune infrastructure (**Figures 3** and **4**). The dual wields of NS1 and PB1-F2 thus allow for maximal viral replication with minimal interference from the host immune system, while causing severe disease in the infected individuals.

# NOVEL THERAPEUTIC STRATEGIES

Influenza vaccination remains the cornerstone of efforts to protect against influenza, however, there are serious concerns regarding current vaccine prediction and manufacturing process. Mis-match or unforeseeable mutations between the circulating strain and the vaccine strain often results in compromised herd immunity and increased infections in that year (101). Vaccine manufacturing is a slow and problematic process, and this issue became evident during the 2009 H1N1 pandemic. The first vaccines only became available 5 months after the identification of the virus, and mutations can occur during the vaccine manufacturing period, leading to reduced effectiveness than anticipated. Antiviral drugs, such as M2 channel and neuraminidase inhibitors, are mostly used as treatment for influenza. However, these drugs were mostly given to patients who had symptoms after the virus had shed, therefore, had limited effect. Neuraminidase inhibitors have been shown to be more effective in reducing length of hospital stay than a reduction of mortality rate (102, 103). Heavy use of these drugs has also resulted in a massive development of resistance, including the emergence of oseltamivir-resistant H1N1pdm09 (A/Newcastle/1/2011) in Newcastle, NSW, Australia (2). There is, therefore, an urgent need to identify and develop novel therapeutic targets for influenza, in preparation for future emergence of influenza viruses and pandemics.

The peptide-based therapeutics is an ideal approach that offers specificity with relatively minor side-effects, and can be administered directly to the lung *via* the pulmonary delivery route. This delivery route also increases the adsorption area in the lung by the therapeutic peptides. For these reasons, a number of peptides have been discovered to inhibit IAV replication at different stages of infection.

IAV HA represents an initial and critical step to a successful infection, and would be an ideal target for drug design as a preventative strategy. By screening a HA fragment peptide library, Chen and Guo identified a HA-targeting peptide, HA-pep25, that inhibited influenza virus binding and infection, including human H1N1, human and avian H5N1, and avian H7N9 (104). HA-pep25 is an 18-mer mimicking peptide that corresponds to the sialic acid residue binding site (HA221–238), and specifically binds to the sialic acid residues, thereby shielding the cells from the viral attachment. Jones et al. also demonstrated that a 20-amino acid peptide derived from fibroblast growth factor 4 was able to inhibit viral binding to sialic acid residues by several IAVs, including H1N1, H3N2, and H5N1 (105). Prophylactic treatment in mice substantially reduced viral replication, and post-infection treatment was also as effective as rimantadine in protection against H5N1 in mice. Other peptide-based inhibitors, such as Flupep and Flufirvitide, have also been shown to be effective in inhibiting viral attachment to the cells. Flupep interacts with HA and inhibits viral binding (H1N1pdm09, H3N2, and H5N1) to the cell membrane (106). Flufirvitide is a mixture of hydrophobic α-helical peptides that also binds with HA and blocks viral internalization and infection, and is currently being tested in clinical trials (107).

A class of synthetic anti-lipopolysaccharide peptides (SALPs) has been reported to also inhibit IAV attachment to cell surface. SALPs were originally designed to inhibit bacteria-mediated lethal septic shock in mice (108), but also possess antiviral activities against IAVs. Specifically, SALP PEP19-2.5 peptide showed high binding affinity toward the sialic acid residues, and led to reduced viral attachment and internalization, including human H3N2, H1N1pdm09, and high pathogenic H7N7 (109). Given that primary IAV infection is frequently followed by secondary bacterial infection, this class of peptides would be an ideal therapeutic approach for IAVs as a preventative or post-infection treatment. IAV polymerase complex is also an ideal therapeutic target due to its importance in viral protein synthesis in the host cells. PB11–25 and PB1715–740 (110), and PB1731–757 (111) mimicking peptides have been separately shown to bind to a conserved PB1 binding site on PA subunit, and substantially reduced viral replication of a panel of viruses, including H1N1 and H5N1.

Advances in structural interactions between human broadly neutralizing antibodies and HA have directed several synthetic proteins and peptides that inhibit fusogenic conformational changes in the endosome. A panel of synthetic HA binder (HB) scaffold proteins have been shown to bind with the conserved region of the stems of HA protein (112, 113). HB36 and HB80 were shown to bind H1 and H5 HA stem with nanomolar affinity, and inhibited low-pH-induced conformational changes. HB80 also displayed similar levels of neutralization compared with the broadly neutralizing antibodies. Furthermore, Kadam et al. recently constructed a series of small cyclic peptides that resemble the epitopes of the stem-targeting broadly neutralizing antibodies (114). These peptidic fusion inhibitors bound to the conserved HA stem epitope at nanomolar affinity, and inhibited HA conformational change at low pH. Similar design strategies could also be applied to other conserved sites on HA, or to other virulence factors of IAV.

As PB1-F2 and NS1 are two major virulence factors that promote pro-inflammatory cytokine storm and inhibit important antiviral immunity, it is interesting to note that there has been no progress in identifying peptides that specifically target these two factors. Ideal therapeutic peptides need to either shield hosts from IAV internalization, or have anti-inflammatory properties without affecting or even increasing antiviral immune responses to infection. These peptides, therefore, are an attractive direction for therapeutics against IAVs, and may be combined with current antiviral drugs to increase the effectiveness of the anti-influenza treatment. The efficacy of peptide-based drugs as preventive agents, however, needs to be further explored, particularly in transmission models such as ferrets.

# CONCLUSION

The constant genetic mutations, recombination, and selection pressure ensure the survival of IAV and its continued threat to mankind. This rapid and threatening evolution of IAV enables the virus to infect multiple species, and with its superior immunomodulatory tactics, the emergence of novel IAVs is unpredictable and inevitable.

HA is the key for the virus to infect, spread, and to create a pandemic. Understanding the absolute molecular requirements for IAV to cross species barriers and achieve sustainable human transmission, and/or aerosol-airborne transmissibility switch will be vital in the development of next-generation virus-targeted therapeutics. This is crucial as the potential scenario of future IAV pandemics is likely to start with a lack of an IAV-specific vaccine, and current antiviral drugs would be available, but with questionable beneficial effect. The efficacy of peptide-based antiviral drugs as preventative agents will be an important first line of defense. However, when infection occurs, inhibiting viral polymerases with specific anti-PB1/2/PA peptides may also be feasible.

The dynamic duet of PB1-F2 and NS1 effectively controls the host cellular machineries for efficient viral replication, while instigating severe cytokine storm and disarming the host innate antiviral immunity in the infected individuals. Neutralization of their functions would be desirable to reduce infection and potentially fatal symptoms. By the time the infected individuals are admitted to hospital with symptoms, the virus would have shed. Future therapeutics will, therefore, also need to be host-targeted to treat symptoms such as the exaggerated airway or systemic inflammation that is left behind by the virus.

# REFERENCES


Peptide-based therapeutics may be the new generation of antiviral drugs, and would be protective against multiple strains/subtypes of IAV if targeted at the essential and conserved regions. This together with ever-improving nanoparticle delivery technologies could be the future of drug design and intracellular delivery system, and must be explored in the calm before the storm.

The great 1918 influenza pandemic happened before, and it will happen again.

# AUTHOR CONTRIBUTIONS

AH conceptualized and wrote the manuscript.


**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Hsu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Respiratory Mononuclear Phagocytes in Human influenza A virus infection: Their Role in immune Protection and As Targets of the virus

### *Sindhu Vangeti, Meng Yu and Anna Smed-Sörensen\**

*Division of Immunology and Allergy, Department of Medicine Solna, Karolinska Institutet, Karolinska University Hospital Solna, Stockholm, Sweden* 

#### *Edited by:*

*Francesca Chiodi, Karolinska Institutet (KI), Sweden*

#### *Reviewed by:*

*Elisa Vicenzi, San Raffaele Hospital (IRCCS), Italy Nicolas Ruffin, Institut Curie, France*

> *\*Correspondence: Anna Smed-Sörensen anna.smed.sorensen@ki.se*

#### *Specialty section:*

*This article was submitted to Viral Immunology, a section of the journal Frontiers in Immunology*

*Received: 30 March 2018 Accepted: 19 June 2018 Published: 03 July 2018*

#### *Citation:*

*Vangeti S, Yu M and Smed-Sörensen A (2018) Respiratory Mononuclear Phagocytes in Human Influenza A Virus Infection: Their Role in Immune Protection and As Targets of the Virus. Front. Immunol. 9:1521. doi: 10.3389/fimmu.2018.01521*

Emerging viruses have become increasingly important with recurrent epidemics. Influenza A virus (IAV), a respiratory virus displaying continuous re-emergence, contributes significantly to global morbidity and mortality, especially in young children, immunocompromised, and elderly people. IAV infection is typically confined to the airways and the virus replicates in respiratory epithelial cells but can also infect resident immune cells. Clearance of infection requires virus-specific adaptive immune responses that depend on early and efficient innate immune responses against IAV. Mononuclear phagocytes (MNPs), comprising monocytes, dendritic cells, and macrophages, have common but also unique features. In addition to being professional antigen-presenting cells, MNPs mediate leukocyte recruitment, sense and phagocytose pathogens, regulate inflammation, and shape immune responses. The immune protection mediated by MNPs can be compromised during IAV infection when the cells are also targeted by the virus, leading to impaired cytokine responses and altered interactions with other immune cells. Furthermore, it is becoming increasingly clear that immune cells differ depending on their anatomical location and that it is important to study them where they are expected to exert their function. Defining tissue-resident MNP distribution, phenotype, and function during acute and convalescent human IAV infection can offer valuable insights into understanding how MNPs maintain the fine balance required to protect against infections that the cells are themselves susceptible to. In this review, we delineate the role of MNPs in the human respiratory tract during IAV infection both in mediating immune protection and as targets of the virus.

#### Keywords: emerging, virus, influenza, respiratory, monocyte, dendritic cell, macrophage

**Abbreviations:** AMϕ, alveolar macrophage; ARDS, acute respiratory distress syndrome; BAL, bronchoalveolar lavage; CM, classical monocyte; DC, dendritic cell; IAV, influenza A virus; IFN, interferon; IM, intermediate monocyte; IMϕ, interstitial macrophage; ISG, interferon-stimulated gene; LRT, lower respiratory tract; MDC, myeloid dendritic cell; MNP, mononuclear phagocyte; Mo-DC, monocyte-derived dendritic cell; Mϕ, macrophage; NCM, non-classical monocyte; PDC, plasmacytoid dendritic cell; RE, respiratory epithelium; tipDC, TNF/iNOS-producing dendritic cell; URT, upper respiratory tract.

# INTRODUCTION

Emerging viruses including influenza viruses, contribute significantly to human morbidity and mortality. Influenza is one of the oldest diseases known to mankind, with historical reports of influenza outbreaks dating as far back as 1173 (1). Still, influenza viruses are considered emerging/re-emerging viruses due to their capacity to dramatically change and cause epidemics with high mortality rate (2–5).

There are two forms of influenza: seasonal and pandemic. Seasonal influenza epidemics are caused by influenza A and B viruses and seasonal strains undergo mutations referred to as antigenic drift. For influenza A viruses (IAVs) antigenic drift is typically more pronounced each season, while it is more gradual for influenza B (6–9). Seasonal influenza epidemics contribute heavily to global disease burden and to deaths associated with lower respiratory tract (LRT) infections. 3–5 million cases of severe illness and 290–650,000 deaths annually are estimated, especially in young children, immunocompromised, and elderly people (10–13). The clinical picture of IAV infection is broad, ranging from mild/no symptoms, to viral pneumonia, severe respiratory failure, or acute respiratory distress syndrome. IAV infection results in increased susceptibility to secondary bacterial infections, which also contribute to mortality (14–16). In addition, circulating IAV strains can, at unpredictable intervals, cause influenza pandemics when the virus undergoes more dramatic genetic changes known as antigenic shift. Four pandemics have occurred in the past century: the 1918 Spanish flu, the 1957 Asian flu, the 1968 Hong Kong flu, and the 2009 Swine flu. Influenza pandemics are usually characterized by higher mortality than seasonal epidemics, often in age groups that are not typically at risk for influenza infections (17–23).

The nature and severity of influenza disease are influenced by the properties of the virus, host genetics, pre-existing immunity, and the immune response generated to varying extents—their relative contributions remaining incompletely understood (24–30). Highly pathogenic strains, like the Spanish flu, induce massive immune responses, suggesting that too potent antiviral immune responses are pathogenic rather than protective and that immunopathology is central in influenza (19, 31–37). Still, robust immune responses against IAV are required to control and clear infection (38–40). Mononuclear phagocytes (MNPs)— monocytes, dendritic cells (DCs), and macrophages (Mϕ)—are important in IAV infection as they are capable of limiting virus release; sensing and phagocytosing pathogens; clearing virus and apoptotic cells; releasing cytokines to mediate inflammation; directing leukocyte traffic *via* chemokine release; processing and presenting viral antigens; and finally activating naïve T cells (41–49).

The distribution and function of immune cells, including MNPs, differ between anatomical compartments (50–53). However, the exact nature of MNP involvement in human IAV infection remains largely unclear. Sampling the human respiratory tract in patients during ongoing infection poses significant challenges of accessibility and risk of causing further injury to the mucosal barrier. Defining human respiratory MNP distribution, phenotype, and function during IAV infection can therefore offer valuable insights into understanding how the immune system maintains the fine balance required to protect against infections. In this review, we will summarize insights on the role of MNPs in the human respiratory tract during IAV infection both in mediating immune protection and as targets of the virus.

# HUMAN RESPIRATORY MNPs

The human respiratory tract encompasses a large mucosal surface with the densest vasculature of all organ systems, that is constantly exposed to the external environment with every inhalation (54, 55). MNPs are positioned along the respiratory tree, in anticipation of exposure to foreign material and respiratory pathogens. MNPs are dually tasked with both promoting inflammation and maintaining tolerance, without disrupting the mucosal barrier that separates the air-filled alveolar spaces from sterile blood in the capillaries (56). A detailed understanding of the distribution and function of respiratory MNPs from the nasal cavities to the alveoli is essential, yet currently incomplete, largely due to the challenges of accessing these tissues in humans. However, recent studies have generated important insight in this area and Mϕs, monocytes, monocyte-derived DCs (mo-DCs), and bona fide DC subsets have been identified from healthy human respiratory tissues. **Figure 1** summarizes the current understanding of the phenotype and distribution of human MNP subsets in respiratory tissues at steady state, as reported (41, 51, 52, 57–62).

Alveolar macrophages (AMϕs) are the most abundant phagocytes of the human lungs, responsible for internalizing inhaled pathogens and antigens, and comprising 95% of cells sampled *via* bronchoalveolar lavage (BAL) (51, 58, 60, 77). Interstitial macrophages, a functionally distinct population of Mϕs residing in lung parenchymal tissue, are less accessible and thus less well studied (63, 78). Similar to monocytes in blood, respiratory monocytes have been characterized as classical monocytes (CMs: CD14+ CD16−), intermediate monocytes (IM: CD14+ CD16+), and non-classical monocytes (NCMs: CD14− CD16+) (51, 52, 58, 64–66). IMs are more frequent in the airways, as opposed to blood, where CMs are in abundance; while NCMs seem to be the rarest monocyte subset (51, 58–60). CMs are the first cells to migrate out of blood to infiltrate sites of inflammation, release chemokines to attract other leukocytes; and can differentiate into mo-DCs and Mϕs (67, 68). IMs represent a population of differentiating monocytes that have been reported to expand during inflammation and/or infection (79–82). NCMs have been attributed with patrolling functions, debris removal, promoting wound healing (64, 81), and to some extent, TLR3 mediated type I interferon (IFN) production (69). Mo-DCs are an interesting subset that transiently arises in tissues from (primarily classical) monocytes recruited to the site of inflammation (46). In comparison to monocytes, DCs are rare in blood, and rarer still in the airways. Subsets of CD11c-expressing myeloid DCs (MDCs); CD1c+ MDCs, CD141+ MDCs, and more recently, langerin+ MDCs (with variable CD1a expression), as well as CD123+ plasmacytoid DCs (PDCs) have been described in the human respiratory tract (51, 57–60, 70–72). MDCs are excellent

Figure 1 | Mononuclear phagocyte (MNP) phenotype and distribution vary across human respiratory compartments. (A) Respiratory compartments and sampling sites. In the human upper respiratory tract, the initial site of influenza A virus infection, immune cells including macrophage (Mϕ), monocyte, and dendritic cell (DC) subsets from the nasal cavity and sinuses can be collected with nasal biopsies or nasal wash sampling. Along with pharyngeal palatine tonsils (and tubal and lingual tonsils), the adenoids form the Waldeyer's ring, an anatomical structure comprising a ring of lymphoid tissue guarding the pharynx. In the lower respiratory tract, bronchoscopy allows sampling of discrete regions of the airways and lungs. Bronchial washes can be used to sample the cells lining the bronchi and bronchioles. Endobronchial biopsies can also be obtained from the mucosal tissue of the bronchial walls. Bronchoalveolar lavages (BALs) sample the most distal airways and alveolar sacs. Finally, lung resection samples allow sampling of lung parenchyma and tissue-resident immune cells. (B) Distribution of human MNP subsets. Pie charts illustrate broadly pooled data from 21 published studies on human MNP subset distribution in blood, tonsils, BAL, and lung tissue to demonstrate the differential distribution of MNPs across anatomical compartments reported from many research groups (51, 52, 57–61, 63–76). As different studies utilize different strategies to specifically define MNPs, the pie charts show groups of cells typically including several subsets of cells: Mϕs (beige), monocytes (green), myeloid DCs (MDCs) (coral), and plasmacytoid DCs (PDCs) (teal). (C) Surface markers to identify MNP subsets across human tissues. The various MNP subsets across tissues can be identified using flow cytometry from HLA-DR+ leukocytes that do not express lineage (T cells, B cells, NK cells, and granulocytes) markers. Apart from CD123+ PDCs, the MNP subsets express different levels of the myeloid marker CD11c. Mϕs have been studied in detail in both BAL and lung tissue, where CD169 expression distinguished alveolar from interstitial Mϕs. Monocyte subsets can be identified from most tissues based on relative expression of CD14 and CD16, as first defined in blood. The major MDC subsets are defined by expression of CD1c or CD141. The extended MDC subsets are now distinguished by expression of CD207 (langerin), CD1a, or slan (51, 52, 57–61, 63–76).

antigen-presenting cells, CD141+ MDCs specialize in cross presentation *via* MHC I; and PDCs excel at type I IFN-mediated antiviral protection.

In the human respiratory system, the upper respiratory tract (URT) is comprised of the nasal cavity, sinuses, and the pharynx (**Figure 1A**). The LRT including the trachea, bronchi, bronchioles, and alveoli, is typically divided into the proximal conducting zone and the distal respiratory zone (**Figure 1A**) (83). The LRT accounts for a larger cumulative surface area and consequently higher likelihood of pathogen–immune cell interactions. However, it is the URT that is initially involved in prevention of pathogen entry (83). MNP distribution in the URT, especially at steady state, also remains poorly characterized. Recent studies have shown Mϕs, CMs, MDCs, and PDCs in the nasal cavities (84, 85); CMs in the sinuses; CMs, MDCs, and PDCs in the nasopharynx (43, 44); CD1c+ MDCs in nasal tissue (86); and Mϕs, CMs, and several DC subsets (PDC, CD1c+, CD141+, CD207+, slan+, Axl+, and CD4+) have been described in human tonsils (73–76). What is evident, however, is that the relative distribution of MNP subsets at steady state varies greatly across the different compartments of the respiratory tract (51, 87). For example, in blood, monocytes greatly outnumber all other MNP subsets, whereas in tonsils, PDCs are the most abundant MNP subset. In BAL, AMϕs make up almost 95% of all cells, but IMs are more frequent than DCs. In lung tissues, both alveolar and interstitial Mϕs can be found at different frequencies. Monocytes and MDCs are also present at greater frequencies than PDCs (**Figure 1B**). The immunological map of the human respiratory tree is becoming more detailed (**Figure 1C**), enabling a better understanding of how the respiratory immune system changes during disease including respiratory viral infections like IAV.

# MNPs: INNATE IMMUNE RESPONDERS IN IAV INFECTION

Respiratory MNPs function as mucosal sentinels and come into play rapidly after onset of IAV infection. Monocytes and DCs resident in the nasopharyngeal mucosa can rapidly sense the presence of IAV and elicit an early response featuring a predominance of monocyte-recruiting chemokines like CCL2, CCL17, CX3CL1, and MCP3 (45, 88, 89). Mϕs, that are abundant in the LRT, are less likely to be involved in uncomplicated human IAV infections, when the virus typically remains localized in the URT. However, when the virus spreads lower toward the lungs, not uncommon among pandemic IAV strains, Mϕs are likely central in the innate immune response.

The diverse functional capacity of monocytes translates into their involvement in several aspects of immunity to IAV, as depicted in **Figure 2**. Monocytes rapidly infiltrate the URT following IAV infection where increased nasal CM numbers and cytokine (MCP3, IFNα2, and CCL17) levels can predict disease severity (43–45). Similarly, in patients infected with the pandemic A/CA/07/09 (pH1N1) strain, high numbers of CD14+, TNF-producing monocytes were reported in blood, that positively correlated with disease severity in young, otherwise healthy adults (88, 90). In addition, exposure to IAV also drives differentiation of monocytes into mo-DCs *in vitro* (91). Studies on human IAV infections demonstrate causal association between CCR2-dependent lung monocyte and mo-DC recruitment and IAV-induced mortality in an NOS-2-dependent manner (91–95).

Mucosal tissue-resident DCs in peripheral tissues like the respiratory tract sense and take up antigens. They then migrate to draining lymph nodes to present processed antigen to T cells. Antigen-specific, clonally expanded T cells migrate back to the site of infection to control and clear infection (128–131) (**Figure 2**). This process is critical to restoration of homeostasis as well as for induction of potent adaptive immune responses. Murine models have elegantly demonstrated DC function during IAV infection (96–100, 132). What remains to be described is the exact role of human DC subsets. During pediatric IAV infection, MDCs and PDCs mobilize to the nasopharynx while DC numbers are reduced in blood (43, 44). The potential redistribution of DC subsets remains to be characterized in adults as well as over the course of infection. The different DC subsets each likely perform individualized tasks during IAV infection. CD1c+ MDCs are the most abundant MDCs in the airways (51, 58, 60), and are excellent at pathogen recognition (101, 133), inducing expansive T helper responses (107–109); and cytokine secretion (101). CD141+ MDCs possess superior MHC I cross-presenting abilities that can aid IAV clearance by CD8+ T cells (73, 107). TLR3 mediated cytokine production (TNF, IL-6, IL-12, and IFN-β) and importantly, type III IFN production by CD141+ MDCs, assist in enhanced innate MNP protection against IAV (57). PDCs mediate type I IFN-dependent antiviral protection that is beneficial during IAV infection. In addition to transcriptional activation of many IFN-stimulated genes (ISGs), PDCs also promote both T and B cell responses (75, 76, 102–104, 110–112).

Macrophages contribute during IAV infection by clearing cell debris, chemokine and cytokine production to modulate inflammation, recruitment of other MNPs, and to restore subsequent tissue homeostasis (**Figure 2A**) (105, 113). AMϕs are of particular importance when the infection reaches the LRT, where the AMϕs are in vast abundance. Severe influenza with LRT pathology is often accompanied by AMϕ involvement (114–118). Unhindered AMϕ-associated cytokinemia can result in devastating consequences for patients, ranging from delayed recovery to fatal lung pathology (116). Several factors control the extent of Mϕ involvement, two of the most likely contributors being IAV subtype/strain and Mϕ phenotype (90, 105, 113, 114, 118–120, 134) (**Figure 1C**). For example, Mϕ cytokine production differs across H5N1 and pandemic/seasonal H1N1 strains (119). The protective and pathologic roles of MNP subsets during IAV infection have also been summarized in **Figure 2B**.

Murine models of IAV infection have extensively characterized the role of MNPs in antiviral protection (92, 100, 132, 135–137). A potent immune response to human IAV infection is also likely dependent on synergy between the different MNP subsets and their functions (53, 138–141). However, MNP susceptibility to IAV infection can easily upset the balance, impacting both virus clearance and return to homeostasis.

# MNPs AND RESPIRATORY EPITHELIUM (RE): MUCOSAL BARRIERS AND TARGETS OF IAV INFECTION

During IAV infection, the virus is largely confined to the airways, where the RE is primarily targeted (13, 106, 142–147). The RE and MNPs represent an interesting functional dichotomy—both are targets of the virus and also capable of

Figure 2 | Human mononuclear phagocytes (MNPs) play a multitude of roles to mediate immune protection during influenza A virus (IAV) infection. (A) MNP subsets have many overlapping functions. Macrophages (Mϕs) clear up cell debris and release cytokines. Monocytes and dendritic cells (DCs) can also release cytokines and present antigens to initiate adaptive responses. (i) Following IAV infection of respiratory epithelium, Mϕs, monocytes, and DCs respond to the virus and cell debris, launching potent cytokine responses (TNFα, IL-6, IL-12p40, and IL-10), including interferon (IFN)α. Induction of interferon-stimulated genes (ISGs) promotes an antiviral state in bystander cells, protecting them from infection. (ii) The antigens taken up by monocytes/DCs are processed and presented *via* MHC I and II to CD8+ and CD4+ T cells, respectively. Antigen-specific CD8+ T cells perform effector functions *via* cytotoxic granule- and FasL-mediated caspasedependent apoptosis. (iii) CD4+ T cells mature into subsets with specific functions. Th1 cells primarily produce IFNγ, IL-2, and TNFβ; and aid CD8+ T cell proliferation. Th2 cells on the other hand, produce IL-4, IL-5, and IL-13 and assist B cells, especially during antibody class switching, promoting production of neutralizing antibodies. Induction of broadly neutralizing antibodies against all strains of influenza virus remains a challenge in the field of influenza immunology (45, 57, 75, 76, 96–106). (B) The table summarizes the individual functions of MNP subsets that can protect against IAV infection, but also contribute to pathology. Most MNP subsets are susceptible to IAV infection, as demonstrated by *in vitro* studies. As a consequence of IAV infection, MNP function can be directly affected, prompting them to respond in a protective or pathologic fashion (25, 37, 42–45, 73, 75, 76, 91, 102–105, 107–127).

immune functions to limit infection (25). The epithelial tight junctions constitute a mechanical barrier against the exterior and secrete antiviral molecules. The RE senses IAV *via* TLRs and RIG-I; with RIG-I signaling concentrated at the tight junctions, resulting in type I and type III interferon-mediated antiviral protection (106). Chemokines secreted from the RE aid neutrophil and MNP recruitment to the site of infection, enhancing innate protection. While potently responding to IAV, the RE is also highly susceptible to the cytopathic effects of IAV infection (**Figure 2A**). Loss of mucosal barrier integrity promotes bacterial adherence, contributing to secondary bacterial infections and lung pathology often associated with severe IAV infection (118, 147–149).

Mononuclear phagocytes are well located in the human respiratory mucosa to be targeted by the virus upon entry (135), and the endocytic and migratory properties of MNPs are likely favorable to viral infection and dissemination (120, 134). *In vitro* IAV infection of human Mϕs and DCs has been shown to result in productive infection with release of infectious particles (119–121) but has also been reported to result in abortive infection (42, 108, 121, 122), the contrast being discussed in great detail in Ref. (42). Which of these alternatives prevail in clinical cases, and what host factors determine their own fate, are questions that are yet to be answered. In addition, the negative implications of IAV infection, from an immunological perspective, they may be more pronounced for MNPs than for epithelial cells as MNPs are central in establishing a protective, specific immune response.

# CONSEQUENCES OF IAV INFECTION OF MNPs

Mononuclear phagocyte susceptibility to IAV infection can impair their many functions. For example, MDCs are crucial for T cell activation but they are also readily susceptible to IAV infection, impairing their ability to present antigens *via* both the direct presentation and cross presentation pathways (46, 121, 150). Most seasonal and low-pathogenic IAV strains infect respiratory human Mϕs and DCs but replication is typically abortive and therefore skews in favor of host defense (120). However, highly pathogenic strains of IAV can overcome this barrier and productively infect Mϕs and DCs, which in turn can impact viral amplification, dissemination, as well as pathogenicity and immunogenicity (123). Primary human monocytes exposed to H5N1 or highly pathogenic avian influenza strains *in vitro* exhibit a reduced antiviral response, as a consequence of impaired NF-κB signaling (91, 114, 115). In a murine model of IAV infection, CCR2+ inflammatory monocytes accumulate in lungs (92, 94). Impaired virus clearance by MNPs triggers IFN-mediated recruitment of CCR2+ monocytes inflammatory in a positive-feedback loop, resulting in severe lung pathology (92) (**Figure 2B**).

Impaired MNP responses have also been observed in IAV patients. Peripheral blood monocytes and to some extent PDCs, exhibit attenuated IFN responses indicating dysregulation at a systemic level, in particular in infants and the elderly, two of the largest risk groups for severe influenza disease (151–153). Human PDCs that potently produce large amounts of type I IFN, in response even to low doses of IAV, can rapidly undergo apoptosis when exposed to high doses of the virus (25, 124). Possibly related to that, it has been reported that pregnant women, a risk group for influenza, have fewer PDCs in circulation that are also less efficient at IFN production, which could contribute to more severe IAV disease during pregnancy (125) (**Figure 2B**).

As undesirable as depressed MNP function is, excessive activation of MNPs can also be equally dangerous, by contributing to IAV-induced immune pathology leading to fatal respiratory distress. Human monocyte-derived pro-inflammatory Mϕs exposed to IAV *in vitro* exhibit augmented phagocytic capability and strong cytokine responses (119). While this can encourage adaptive responses, it also contributes to the cytokine storm that is a hallmark of severe influenza disease (37, 126). Prolonged IFN signaling can also destroy alveolar epithelium and contribute to development of secondary bacterial infections, the most common complication associated with influenza infections (93). TNF/ iNOS-producing DCs, a subset of inflammatory DCs, accumulate in the LRT and promote CD8+ T cell responses in an IAV mouse model, but are also positively correlated with higher lethality (123). However, *in vitro*, human CD8+ T cells can rapidly induce monocyte differentiation into tip-DCs that in turn prime naïve CD4+ T cells and promote protective Th1 responses (154) (**Figure 2B**).

Not all respiratory MNP–IAV interactions have adverse implications. Virus-induced human *in vitro* mo-DCs express both CLEC9A and CD141, as do blood CD141+ MDCs. But uniquely, mo-DCs express CD141 on the cell surface and CLEC9A intracellularly (91). CD141+ DCs can efficiently prime and drive CD8+ T cell proliferation, while CLEC9A is linked to antigen uptake. CD141+ MDCs also subvert IAV infection by resisting virus entry in a RAB-15 dependent manner, instead relying on uptake of apoptotic virus-infected CD1c+ MDCs (and other cells) as a source of antigens (127) (**Figure 2B**). Virus-induced CD141+ DCs also exhibit type I IFN secretion and upregulate ISGs (tetherin, viperin, and IFITM3) and RIG-I/MDA5, suggesting an important protective role for them during infection; despite poor expression of co-stimulatory molecules (CD40, CD86, and HLA-DR), weaker pro-inflammatory cytokine expression, and impaired ability to activate naïve CD4+ T cells (46). Induction of CD141+ DCs could therefore be employed in vaccination/ therapeutic strategies. To summarize, while IAV infection of MNP compromises some aspects of innate protection, biological redundancy due to the overlapping functions of MNP subsets can likely prevent loss of essential immune responses.

# CONCLUDING REMARKS

Respiratory MNPs are important in the immune responses to IAV infection. At the same time, MNP susceptibility to IAV infection poses an interesting immunological challenge. Several key questions still remain to be further addressed to understand this dichotomy better. Does compromised MNP function result in altered innate immune responses? Do altered innate immune responses subsequently impair efficient induction of adaptive responses, ultimately contributing to increased host morbidity and mortality? If on the other hand, robust, unchecked innate responses lead to prolonged inflammation, causing irreparable damage to the host, is there a commonality in host responses across the various demographics affected by influenza? To answer these questions, and delineate the role of respiratory MNPs in human IAV infection, it will be critical to detail the function of the different MNP subsets—for example, functional assessment of sorted cells from the respiratory system and performing RNA sequencing or epigenetic analyses. Prospective studies of human IAV patients where detailed analyses of tissue samples can be correlated to clinical parameters are likely required to fully understand how MNPs contribute to disease severity.

## REFERENCES


# AUTHOR CONTRIBUTIONS

SV and MY performed the literature review. SV designed the figures. SV, MY, and AS-S organized and wrote the manuscript. SV and AS-S edited the manuscript.

# FUNDING

This work was supported by grants to AS-S from the following sources: the Swedish Research Council, the Swedish Foundation for Strategic Research, the Swedish Heart-Lung Foundation, and Karolinska Institutet.


the 1918 influenza virus. *Nature* (2007) 445(7125):319–23. doi:10.1038/ nature05495


from Langerhans cells, are consistently present in human tonsils. *Front Immunol* (2016) 7:197. doi:10.3389/fimmu.2016.00197


antiviral activity. *J Biol Chem* (2010) 285(37):28419–24. doi:10.1074/jbc. R110.145839


the response to R-848. *J Immunol* (2004) 173(7):4627–34. doi:10.4049/ jimmunol.173.7.4627


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling Editor declared a shared affiliation, though no other collaboration, with the authors.

*Copyright © 2018 Vangeti, Yu and Smed-Sörensen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Highly Pathogenic H5N1 Influenza A Virus Spreads Efficiently in Human Primary Monocyte-Derived Macrophages and Dendritic Cells

*Veera Westenius1 \*, Sanna M. Mäkelä <sup>1</sup> , Ilkka Julkunen2 and Pamela Österlund1*

*1Expert Microbiology Unit, Department of Health Security, National Institute for Health and Welfare, Helsinki, Finland, <sup>2</sup> Institute of Biomedicine, University of Turku, Turku, Finland*

Influenza A viruses cause recurrent epidemics and occasional global pandemics. Wild birds are the natural reservoir of influenza A virus from where the virus can be transmitted to poultry or to mammals including humans. Mortality among humans in the highly pathogenic avian influenza H5N1 virus infection is even 60%. Despite intense research, there are still open questions in the pathogenicity of the H5N1 virus in humans. To characterize the H5N1 virus infection in human monocyte-derived macrophages (Mɸs) and dendritic cells (DCs), we used human isolates of highly pathogenic H5N1/2004 and H5N1/1997 and low pathogenic H7N9/2013 avian influenza viruses in comparison with a seasonal H3N2/1989 virus. We noticed that the H5N1 viruses have an overwhelming ability to replicate and spread in primary human immune cell cultures, and even the addition of trypsin did not equalize the infectivity of H7N9 or H3N2 viruses to the level seen with H5N1 virus. H5N1 virus stocks contained more often propagation-competent viruses than the H7N9 or H3N2 viruses. The data also showed that human DCs and Mɸs maintain 1,000- and 10,000-fold increase in the production of infectious H5N1 virus, respectively. Both analyzed highly pathogenic H5N1 viruses showed multi-cycle infection in primary human DCs and Mɸs, whereas the H3N2 and H7N9 viruses were incapable of spreading in immune cells. Interestingly, H5N1 virus was able to spread extremely efficiently despite the strong induction of antiviral interferon gene expression, which may in part explain the high pathogenicity of H5N1 virus infection in humans.

Keywords: influenza A virus, avian influenza, macrophages, dendritic cells, viral replication, innate immunity

# INTRODUCTION

Influenza A viruses are one of the most important viral pathogens annually infecting even 5–10% of the human population and causing an estimated 250,000–500,000 deaths worldwide. Occasionally, influenza A viruses cause global pandemics of which the most devastating one was the pandemic caused by Spanish flu in 1918 leading up to 50 million deaths (1). The segmented genome of influenza A viruses confers evolutionary advantages and influenza A viruses have a great ability to evolve by two different mechanisms, antigenic drift and shift. All pandemics in the twentieth century have been of avian origin (2, 3). Avian influenza A viruses circulate continuously in birds and occasionally they have caused infections in humans. The highly pathogenic avian influenza (HPAI) H5N1 viruses

#### *Edited by:*

*Alan Chen-Yu Hsu, University of Newcastle, Australia*

#### *Reviewed by:*

*Maziar Divangahi, McGill University, Canada Anastasia N. Vlasova, The Ohio State University, United States*

> *\*Correspondence: Veera Westenius veera.westenius@thl.fi*

#### *Specialty section:*

*This article was submitted to Viral Immunology, a section of the journal Frontiers in Immunology*

*Received: 23 March 2018 Accepted: 04 July 2018 Published: 17 July 2018*

#### *Citation:*

*Westenius V, Mäkelä SM, Julkunen I and Österlund P (2018) Highly Pathogenic H5N1 Influenza A Virus Spreads Efficiently in Human Primary Monocyte-Derived Macrophages and Dendritic Cells. Front. Immunol. 9:1664. doi: 10.3389/fimmu.2018.01664*

have caused almost 900 human infections from 2003 until today and the mortality among humans has been over 50% (4). The H7N9 virus, which emerged in humans in 2013, has a surprising character as it causes only a mild infection in domestic poultry but an HPAI-like disease in humans. In the beginning of 2017, a human infection was detected with H7N9 virus possessing a multi-basic HA1-HA2 cleavage site motif in the hemagglutinin (HA) molecule characteristic of HPAI viruses (5). Thus far, there have been over 1,500 human infections caused by the H7N9 virus with a mortality rate reaching 40% (4, 6). Therefore, these avian influenza virus strains pose a serious risk for a pandemic. Despite active research, it is not completely clear which factors in H5N1 or H7N9 viruses or in host responses contribute to a severe disease in humans.

After infecting the epithelial cells of respiratory tract, influenza A viruses can spread to alveolar dendritic cells (DCs) and macrophages (Mɸs) which reside in the immediate proximity of the epithelium (7). DCs and Mɸs are the key cell types to orchestrate host immune responses (8, 9). Typically influenza A virus induces antiviral responses in immune cells by inducing type I and III interferons (IFNs) which are known to inhibit virus replication and propagation (10–12). Both alveolar and monocyte-derived Mɸs as well as DCs, including monocyte-derived DCs (moDCs), express receptors for both avian-adapted (α-2,3-linked sialic acids) and human-adapted (α-2,6-linked sialic acids) influenza A viruses (13–16). The viral HA binds to sialylated host cell surface receptor molecules and mediates virus entry. Influenza viruses enter the cells mainly *via* endocytosis and the fusion of viral and endosomal membranes. For the fusion to happen the precursor form of the HA, HA0 has to be cleaved into HA1 and HA2 subunits by host cells proteases. The membrane fusion mediated by the mature form of the HA occurs at low pH which enables the release of the segmented viral genome into the cytoplasm. The genome of the influenza virus is structured in eight separate viral ribonucleoprotein (vRNP) complexes which are transported into the nucleus for the transcription and replication of the virus. The viral proteins are translated in the cytoplasm but the viral proteins are assembled into vRNPs in the nucleus. Newly synthesized vRNPs are exported to the cytoplasm, virus particles are assembled at the cell membrane, and progeny virus particles bud out of the cell. All eight vRNAs have to be packed into a virion to produce infective progeny viruses and the infection to be productive. The mechanism behind the genome packaging is not fully understood but it is believed that influenza A virus packs its vRNAs in a specific manner by a selective packaging mechanism (17). Some studies suggest that most influenza A virus particles are noninfectious since they express incomplete set of viral gene segments and are incapable of inducing a secondary infection (18). However, three-dimensional analysis of the virions has shown that at least 80% of virions have all eight RNPs packaged (19). In addition, it is known that there are differences between various seasonal influenza virus strains in their ability to cause a productive infection (18) but the comparison between avian influenza and seasonal influenza virus strains in primary human cells have remained poorly characterized.

Previously, we have shown that human moDCs are susceptible to the avian influenza virus infection (12, 20). In this study, we show that the highly pathogenic H5N1 influenza A viruses can efficiently replicate and produce new infective particles in human primary moDCs and Mɸs and, despite the strong IFN-mediated antiviral responses induced by the infection, be able to spread throughout the whole immune cell culture. These results suggest that the excessive cytokine production ("cytokine storm") induced by H5N1 infection may in fact be due to extremely efficient spread of the virus infection in the infection site leading to greatly enhanced cytokine gene expression.

# MATERIALS AND METHODS

# Ethics Statement

The permission to import the human isolates of avian virus strains for research purposes was obtained from the Finnish Food Safety Authority (permission no 8634/0527/2012). Infective H5N1 and H7N9 viruses were handled strictly under Biosafety Level (BSL) 3 laboratory conditions at the National Institute for Health and Welfare (THL), Finland. Different virus subtypes were always handled in separate biosafety cabinets to avoid any possible creation of recombinant viruses. Adult human blood was obtained from anonymous healthy blood donors through the Finnish Red Cross Blood Transfusion Service (permission no 37/2016, renewed annually). Animal immunizations related to this study were approved by the Ethical Committee of the National Institute for Health and Welfare (permission no. KTL 2008-02).

# Cell Cultures

The buffy coats were obtained from healthy blood donors (Finnish Red Cross Blood Transfusion Service, Helsinki, Finland). Monocytes were purified from buffy coats as described previously (21). Human peripheral blood mononuclear cells were isolated by density gradient centrifugation over a Ficoll-Paque gradient (Amersham Biosciences).

To obtain monocytes for Mɸ differentiation, mononuclear cells were allowed to adhere onto plates or glass coverslips for 1 h at +37°C in RPMI 1640 (Sigma-Aldrich) supplemented with 0.6 µg/ml penicillin, 60 µg/ml streptomycin, 2 mM l-glutamine, and 20 mM HEPES. Nonadherent cells were removed by washing with phosphate-buffered saline (PBS), and the remaining monocytes were cultured in Mɸ serum-free medium (Life Technologies) supplemented with streptomycin and human recombinant granulocyte-macrophage colony-stimulating factor (rGM-CSF) (10 ng/ml; Nordic Biosite). The cells were differentiated into Mɸs for 7 days, with a change to fresh culture medium every 2 days.

To obtain moDCs, Percoll gradient (Amersham Biosciences) centrifugation was done after Ficoll-Paque gradient centrifugation. The fraction containing monocytes were collected and remaining T and B cells were depleted by using anti-CD3 and anti-CD19 magnetic beads (Dynal). Monocytes were allowed to adhere to plates (Sarstedt) for 1 h at +37°C in RPMI 1640 supplemented as described above. Non-adherent cells were removed by washing with PBS, and immature moDCs were generated by cultivating adherent monocytes in RPMI 1640 supplemented as described above and with 10% fetal calf serum (Integro), 10 ng/ml human rGM-CSF, and 20 ng/ml human recombinant interleukin-4 (R&D Systems). The cells were cultivated for 6 days, and fresh medium was added every 2 days.

In each experiment, cells from three to four different donors were cultured and used separately for infection experiments.

# Viruses

Human influenza A/Beijing/353/89 (H3N2) virus (originates from WHO Collaborating Centre for Reference and Research on Influenza, The Francis Crick Institute, UK) and human isolates of the avian influenza viruses A/Vietnam/1194/2004 (H5N1) and A/Hong Kong/156/1997 (H5N1) (Molecular Virology, Erasmus MC—Department of Viroscience, Rotterdam, Netherlands) and A/Anhui/1/2013 (H7N9) (WHO Collaborating Centre for Reference and Research on Influenza, The Francis Crick Institute, UK) were grown in the allantoic cavity of 10- to 11-day-old embryonated chicken eggs at +36°C for 2–3 days. Since the infectivity of influenza A viruses vary from one type of cell to another, the virus titers were determined by different assays. A hemagglutination titration was done with standard protocol using 0.5% turkey or guinea pig red blood cells. The infective virus titer in Madin–Darby Canine Kidney (MDCK) cells were determined by plaque assay. The viruses were serially diluted and inoculated to confluent MDCK cells on the 6-well plates for 1 h at 37°C. After incubation, cells were washed with PBS and covered with Avicel microcrystalline cellulose [Eagle-MEM containing 1.2% Avicel (#RC-591 NF, FMC BioPolymers)], 0.3% BSA, 2 μg/ml *N*-tosyl-l-phenylalanine chloromethyl ketone (TPCK) treated trypsin, 60 µg/ml streptomycin, 2 mM l-glutamine, and 20 mM HEPES. After incubation at 37°C for 1 day for supernatant samples and for 2 days for virus stocks Avicel was removed, cells washed with PBS, fixed with 4% paraformaldehyde (PFA), and stained with diluted crystal violet. The plaques were counted to obtain the concentration of infective viruses as plaque forming units (PFU)/ml.

The infectivity of the viruses was determined in Mɸs by immunofluorescence microscopy and in moDCs by flow cytometry. The number of infected cells with different virus dilutions was counted to obtain the concentration of infective virus particles as focus forming units per milliliter (FFU/ml). The multiplicity of infection (MOI) in Mɸs or moDCs is given according to the titers determined in Mɸs or moDCs, respectively (**Table 1**).

# Virus Infection Experiments

Macrophages and moDCs were infected with different MOI values of virus for different times, as indicated in the figures. Mɸs were differentiated on glass coverslips or cell culture plates, growth medium was removed, and virus dilution was added on cells. After 1 h at 37°C incubation, cells were washed with PBS, and fresh medium was added. For infection experiments in moDCs, the virus dilutions were added into cell cultures medium without changing the growth medium. For 6 h infectivity experiments in Mɸs and moDCs, cells were incubated with neuraminidase inhibitor, 20 nM oseltamivir carboxylate which is an active metabolite of oseltamivir phosphate (#RO0640802-002, Roche). All infection experiments with H5N1 and H7N9 viruses were performed in BSL-3 facility and only samples inactivated with validated methods were brought out from the BSL-3 laboratory.

# Immunofluorescence Microscopy

Macrophages were differentiated on glass coverslips, infected with different influenza A viruses at MOI values as indicated in the figures, incubated for 1 h at 37°C, washed with PBS, fresh Mɸ medium was added and incubated at 37°C with 5% CO2. At different times after infection, cells were washed, fixed with 4% PFA for 30 min at room temperature, washed, permeabilized with 0.1% Triton X-100 for 5 min, washed, and blocked with 0.5% BSA for 30 min. A/Beijing/353/89 (H3N2) virusinfected cells were stained with guinea pig antibody against influenza A virus H3N2 nucleoprotein (NP) (22). To visualize A/Vietnam/1194/2004 (H5N1) or A/Anhui/1/2013 (H7N9) virusinfected cells, guinea pig antibodies against H5N1 viral glycoproteins were used. Antibodies were prepared by immunizing guinea pigs and rabbits for four times every 2 weeks with A/ Indonesia/5/2005 vaccine antigen (2 + 6 in PR8) (4 µg of HA/ immunization) mixed with AS03 adjuvant (GlaxoSmithKline, Rixenart, Belgium) according to the instructions by the manufacturer. Secondary antibodies were fluorescein isothiocyanate (FITC)-labeled goat anti-guinea pig antibodies. In infectivity experiments, NucBlue Fixed Cell Ready Probes reagent (Life Technologies, DAPI) was added to secondary antibody staining solutions. Incubation time in every staining was 30 min in 0.5% BSA in PBS at 37°C and coverslips were washed three times with 0.5% Tween 20 (VWR) in PBS in every step. Finally, cells were washed with water and mounted in 25% Mowiol (Polysciences) in a solution containing 25 mM Tris–HCl (pH 7.5), 50% glycerol, and 2.5% 1,4-diazabicyclo(2,2,2)octane.


*The infectivity titers of viral stocks propagated in chicken eggs were determined in turkey and guinea pig red blood cells by hemagglutination assay (HA), in MDCK cells by the plaque forming assay (PFU), in human macrophages (Mɸs) and dendritic cells (moDC) by focus forming assay (FFU) with microscopy and flow cytometry, respectively.*

The cells were imaged with Leica TCS SPE confocal microscope with a 63 1.40-numerical-aperture (NA) oil objective maintaining the same image acquisition settings for all acquired images. In the infectivity experiments, the cells were imaged with a Zeiss Stallion fluorescence microscope with a Hamamatsu ORCA-Flash 4.0 LT sCMOS camera and a 20 0.4-NA objective by using Slidebook 6 software (Intelligent Imaging Innovations). In the productivity assay, cells were calculated with 20 0.6-NA air objective with Leica TCS SPE confocal microscope.

# Flow Cytometry

For determining the infectivity of viruses in moDCs, the cells were differentiated on 6-well plates. Cells were infected with different virus dilutions and times as indicated in the figures. After different times of infection, cells from four different blood donors were harvested and handled separately. Cells were fixed with 4% PFA for 30 min, permeabilized with 0.1% Triton X-100 for 5 min, and stored with 0.5% BSA in PBS. A/Beijing/353/89 (H3N2) virus-infected cells were stained with rabbit antibodies against H3N2 glycoproteins (23). To visualize A/Vietnam/1194/2004 (H5N1) or A/Anhui/1/2013 (H7N9) virus-infected cells, antibodies against A/Indonesia/5/2005 (H5N1) vaccine antigen were used (see above). The secondary antibody was FITC-labeled goat anti-rabbit IgG (H + L) (Invitrogen, #F2765). In all antibody stainings, the cells were stained at RT for 1 h and washed twice with 0.5% BSA in PBS. The samples were analyzed with a FACSCanto II (BD) device using FACSDiva software.

# Western Blotting

For protein analysis, Mɸs or moDCs from four blood donors were harvested and pooled. We pooled the donors to be able to analyze higher amount of donors to get a more global and general view of the protein expression changes. Cell pellets were lysed with passive lysis buffer (Promega) containing 1 mM Na3VO4. Total cellular proteins were boiled in a Laemmli sample buffer, proteins were separated in SDS-PAGE gels and transferred onto Hybond-P polyvinylidene difluoride membranes (Amersham Biosciences). The membranes were blocked with 5% milk protein (Valio Co., Helsinki) in PBS. Anti-influenza A virus NP and M1 and MxA protein-specific antibodies were prepared as described previously (12, 24). Monoclonal anti-influenza A virus H7 HA antibody was obtained through BEI Resources, NIAID, NIH and anti-H3N2 glycoprotein antibodies and anti H5N1 virus antibodies were produced as described above. Antibody stainings for phosphorylated form of IRF3 (P-IRF3), P-STAT2, STAT2, and GAPDH proteins (Cell Signaling Technology) were performed according to the manufacturer's instructions. For secondary antibodies, anti-rabbit or anti-mouse HRP-conjugated antibodies (Dako) were used. Protein bands were visualized on HyperMax films using a Pierce® ECL Western Blotting substrate (Thermo Fisher Scientific).

# qRT-PCR

Infected Mɸs or moDCs from four blood donors were pooled (25), and total cellular RNA was isolated using the RNEasy Mini kit (Qiagen) and DNase digestion was performed with RNasefree DNase kit (Qiagen). 0.5 µg of total cellular RNA was transcribed to cDNA using TaqMan Reverse Transcriptase kit (Applied Biosystems) with random hexamers as primers. cDNAs were amplified by PCR using TaqMan Universal PCR Mastermix and Gene Expression Assays for IFN-λ1 (Hs00601677\_g1), IFN-α1 (Hs00256882\_s1), IFN-β1 (Hs00277188\_s1), TNF-α (Hs01113624\_ g1), IL-1β (Hs00174097\_m1), CXCL10 (Hs00171042\_m1), CCL5 (Hs00174575\_m1) (Applied Biosystems) or with primers and probes for influenza A virus M1 (12). The data were normalized to 18S rRNA with TaqMan Endogenous Control kit (Applied Biosystems). The relative gene expression in relation to an unstimulated sample was calculated with the ΔΔCT method according to instructions provided by Applied Biosystems.

To quantify viral RNA from the supernatant samples, moDCs and Mɸs were infected as described above and the supernatant samples were collected at 1 and 24 h after the infection. RNA was isolated using RNEasy Mini kit (Qiagen) with Qiacube (Qiagen). cDNA synthesis was done with RevertAid H Minus Reverse Transcriptase (Thermo Fisher Scientific) kit according to the manufacturer's instructions with RiboLock RNase inhibitor (Thermo Fisher Scientific) and random hexamers (Roche) as primers. qRT-PCR was performed using QIAGEN® QuantiTect™ Multiplex PCR NoRox Kit (Qiagen) with the same influenza A virus M1 primer–probe pair as above. The relative viral gene expression in relation to a 1-h sample was calculated with the ΔCT method according to the instructions provided by Applied Biosystems.

# Quantitation of Propagation-Competent Virus Particles

Macrophages were grown on glass coverslips and infected with MOI of 0.01. After 1 h incubation, virus dilution was removed, and cells were washed with PBS. Cells were overlaid with 1.2% Avicel in E-MEM (containing 2 µg/ml TPCK-trypsin, 0.3% BSA, 60 µg/ml streptomycin, 2 mM l-glutamine, and 20 mM HEPES) and incubated for 15 h. After incubation, the cells were washed two times with PBS, fixed with 4% PFA for 30 min, washed with PBS, permeabilized with 0.1% triton X-100 for 5 min, and stained with antibodies of different specificity (above-mentioned) followed by immunofluorescence analysis [adapted from Ref. (18)]. The percentage of productive vs. total infection events was determined by eye using confocal microscope with 20× objective and cells were imaged with 40× objective (Figures 7A,B).

# Statistical Analysis

The data was analyzed with the Student's *t*-test to determine the statistical significance of differences between H5N1 virus and H3N2 or H7N9 viruses. Values of *p* < 0.05 were considered significant.

# RESULTS

# Highly Pathogenic H5N1 Virus Has an Ability to Efficiently Spread in Primary Human Immune Cell Cultures

To determine the infectivity of influenza A/Beijing/353/89 (H3N2), A/Vietnam/1194/2004 (H5N1), and A/Anhui/1/2013 (H7N9) viruses in human Mɸs and dendritic cells (moDCs), we infected the cells with different virus doses for 6 h. We determined the number of infective units [focus forming units per milliliter (FFU/ml), **Table 1**] in virus stocks for both cell models separately from the dilution where approximately 50% of the cells were infected (**Figures 1A,C,E**). We used the FFU/ml values to determine the actual MOI. In the single-cycle infection (6 h), the infectivity curves with different doses of H5N1, H7N9, or H3N2 viruses both in Mɸs (**Figure 1A**) and moDCs (**Figure 1C**) corresponded well to the determined FFU/ml values. However, during a multi-cycle infection (24 h), H5N1 virus was able to efficiently spread in the culture and the virus infected all cells even with the very low MOI value in the beginning of the infection (**Figures 1B,D**).

All virus stocks were propagated in low passages in chicken eggs and the stock virus infectivity titers in different cell types were assessed. To determine the titers of infective units in the virus stocks, we performed a plaque forming assay in MDCK cells. In these cells, the H5N1 strain from 2004 gave the lowest titers when compared with those of H7N9 and H3N2 viruses (**Table 1**). The reason for this could be that the H7N9 virus has a high affinity to both α-2,3- (avian type) and α-2,6-linked (human type) sialic acids whereas usually the avian type H5N1 viruses prefer the α-2,3-type receptor (26). However, in moDCs and Mɸs where both α-2,3- and α-2,6-linked sialic acids are expressed (14, 15), the infectivity titer of H5N1 stock was the highest when compared with H3N2 or H7N9 viruses. Moreover, it seems that human Mɸs are more permissive than moDCs to all of these viruses (**Table 1**).

Besides primary infectivity in Mɸs and moDCs, we investigated the replication of these viruses in Mɸs and moDCs by analyzing the expression of viral M1 RNA with qRT-PCR. During the first 6 h of infection, we did not observe any clear differences in M1 gene expression between H3N2, H5N1, or H7N9 viruses, which indicate that the rate of viral replication is similar between the viruses during the primary infection. In the secondary infection phase (24 h time point) in H5N1 infected cells, the expression of M1 gene reached the same submaximal levels regardless of the MOI values used (0.01–10) (**Figures 2A,B**). In H3N2 and H7N9 virus-infected cells, instead, there was a clear dose-dependent expression of viral M1 RNA in different time points (**Figures 2A,B**). In addition, there was a clear difference in the production of viral NP and M proteins between H5N1 and H7N9 or H3N2 virus-infected cells. In H5N1 virus-infected cells viral protein expression was high, independent of the MOI value used in the beginning of the infection, while in H3N2 or H7N9-infected cells viral protein levels depended on the virus dose at the beginning of the infection (**Figures 2C,D**).

# The HA of H5N1 Virus Is Efficiently Cleaved in Human Immune Cells

For the newly produced viruses to be infective, their HA precursor (HA0) has to be cleaved to HA1 and HA2 subunits by cellular proteases. Thus, we next analyzed how efficiently the HA molecule of H5N1, H7N9, and H3N2 viruses is cleaved in human immune cells. As visualized by immunoblotting in H3N2 virus-infected Mɸs, there was only some or none of the HA2 subunit detectable (**Figure 3A**). The A/Vietnam/1194/04 (H5N1) strain has multiple arginine and lysine residues at the HA1–HA2 cleavage site as a marker for a highly pathogenic form of an avian influenza virus. This multi-basic cleavage site can be cleaved by ubiquitous cellular proteases and, indeed, we noticed that in H5N1-infected human Mɸs a great amount of the HA0 protein produced at 24 h even with very low MOI was efficiently cleaved to HA1 and HA2 subunits (**Figure 3B**). However, like in H3N2 infection, in H7N9 virus-infected Mɸs, there was only some or none of the HA2 subunit detectable (**Figure 3C**). The HA of the different viruses was cleaved in similar fashion also in moDCs (**Figures 3D–F**).

# Addition of Trypsin Does Not Increase the Infectivity of H3N2 or H7N9 Viruses to the Levels Seen With the H5N1 Virus

Next we investigated whether the trypsin-induced cleavage of HA contributes to influenza virus spread in immune cells. Human Mɸs and moDCs were infected with H3N2, H5N1, and H7N9 viruses in the presence of TPCK-trypsin to induce the cleavage of HA0 into HA1 and HA2 subunits and the number of virus-infected cells was analyzed by immunofluorescence or flow cytometry, respectively. We noticed that trypsin increased the infectivity of the H3N2 and H7N9 viruses to some extent at the 24-h time point (compared **Figures 4A,B** with **Figures 1B,D**) but yet the infectivity of H3N2 and H7N9 remained at a much lower level when compared with that of the H5N1 virus. In Mɸs, even at a MOI of 0.004 nearly 100% of cells were infected by the H5N1 virus while in H3N2 or H7N9 infection less than 40% of the cells were infected (**Figure 4A**). In moDCs, the difference was equally clear and in H3N2 or H7N9 infection with the lowest MOI values only ca. 20% of the cells were infected while with similar H5N1 virus dose most cells were infected (**Figure 4B**). Unexpectedly, the infectivity of H3N2 or H7N9 virus, even with added trypsin, did not become as effective as was seen with H5N1 virus. This suggests that H5N1 can spread in human immune cells extremely efficiently and this spread is not solely dependent on the effective cleavage of HA.

# H5N1 Virus Is Able to Spread in Immune Cell Starting From Only a Few Infectious Virus Particles

Next, we analyzed how low amount of the H5N1 virus was sufficient to infect the whole immune cell culture. To do this, we made dilutions series of the stock virus and infected moDCs with MOI values ranging from 1 to 10<sup>−</sup>12 for 48 h. Interestingly, most moDCs were infected even at a MOI 10<sup>−</sup><sup>5</sup> of H5N1 virus and the infectivity decreased to background levels at a MOI of 10<sup>−</sup><sup>8</sup> , where there should not be any viral particles left (**Figure 5**). The findings suggest that the H5N1 virus is able to start a productive infection and spread eventually to the whole cell culture even from one infectious virus particle.

Figure 1 | Infectivity of the H3N2, H5N1, and H7N9 influenza A viruses in human immune cells. (A) Human monocyte-derived macrophages (Mɸs) were infected with different doses of a seasonal influenza virus A/Beijing/353/89 (H3N2) or human isolates of avian influenza viruses A/Vietnam/1194/2004 (H5N1) or A/ Anhui/1/2013 (H7N9) (A) for 6 h in the presence of oseltamivir carboxylate to prevent the spread of viruses, or (B) for 24 h in the absence of oseltamivir carboxylate. Infected Mɸs were detected with virus protein-specific antibodies by immunofluorescence microscopy. Results represent the mean values of three (A) or four (B) donors. From each donor at least 200 cells (average ca. 330) (A) or 300 cells (average ca. 840) (B) were counted. Human dendritic cells [monocyte-derived DCs (moDCs)] were infected with different virus doses (C) for 6 h in the presence of oseltamivir carboxylate, or (D) for 24 h in the absence of oseltamivir carboxylate. The proportion of infected moDCs was determined after staining the cells with antibodies against respective viral proteins for flow cytometry. From each donor, 10,000 events were analyzed. The mock sample (uninfected cells) was used to separate the uninfected and infected cells. Results are the mean values of four donors (C,D) analyzed separately. The infective particles per ml (FFU/ml) were calculated for Mɸs and moDCs separately, and multiplicity of infection (MOI) values was determined afterward. The MOI values in each experiment were calculated based on FFU/ml values determined in panels (A,C). (E) Representative images of Mɸs infected with H3N2, H5N1, or H7N9 viruses at MOI 10 or 0.5 stained with virus-specific antibodies and secondary fluorescein isothiocyanate-labeled antibodies.

# An Efficient Spread in Human Immune Cells and Efficient HA Cleavage Are General Features of HPAI H5N1 Viruses

Our next question was, whether the ability to spread in human immune cells is only associated with this specific HPAI H5N1 strain A/Vietnam/1194/04 or whether it is a more universal feature of all highly pathogenic avian-origin H5N1 viruses. Thus, we included another highly pathogenic H5N1 strain into the study. This A/Hong Kong/156/1997 H5N1 virus gave almost an equally high titer in MDCK cells than the H7N9 virus, whereas the other H5N1/2004 strain gave the lowest PFU titer (**Table 1**). It may be that the H5N1/1997 virus is already adapted to the mammalian MDCK cells, because H5N1/1997 virus has been cultivated at least four passages in MDCK cells before the propagation in chicken eggs. This likely explains the high PFU/ml titer of H5N1/1997 virus. The infectivity of A/Hong Kong/156/1997 (H5N1) virus was also determined in Mɸs and moDCs at 6 and 24 h by immunofluorescence microscopy and flow cytometry, respectively (**Figures 6A,B**), and calculated the FFU/ml values (**Table 1**) for virus stocks as was done for H3N2, H5N1/2004, and H7N9 viruses (**Figures 1A,C**). In multi-cyclic infection at 24 h, H5N1/1997 virus was spreading efficiently in Mɸs and moDCs (**Figures 6A,B**) in a similar fashion as the H5N1/2004 virus (**Figures 1B,D**). Infection with H5N1/1997 virus at MOI value of 10<sup>−</sup><sup>5</sup> reached over 60% infection rate in moDCs at 48 h. In addition, 10% of the cells got infected from the virus dose of MOI 10<sup>−</sup><sup>6</sup> where theoretically only one infectious virus particle was added into the culture (**Figure 6C**). Furthermore, the HA of the H5N1/1997 virus was efficiently cleaved from HA0 to HA1 and HA2 subunits as was also the case with the HA of the H5N1/2004 virus (**Figure 6D**). These data show that the H5N1/1997 virus has a similar ability to spread in human Mɸs and moDCs as the H5N1/2004 virus and the cleavage of HA is extremely efficient in these cell types suggesting that these features are general to HPAI H5N1 viruses.

# Primary Infective Units of H5N1 Virus Are More Often Propagation-Competent and Productive Than Those of H3N2 or H7N9 Viruses in Human Immune Cells

We assumed that one reason for the efficient spread of H5N1 virus may be that H5N1 virus particles are more often packed with a complete set of gene segments and therefore these virus particles are propagation-competent and infection events are productive. To investigate this possibility, we set up the propagationcompetent virus particle test (18). Human primary Mɸs were plated on coverslips followed by infection with H3N2, H5N1 from 2004 and H7N9 viruses at MOI of 0.01, overlaid with Avicel cellulose, incubated 15 h, fixed, and the infected cells stained for fluorescent microscopic analysis. Infection events were considered productive when three or more adjacent cells were infected. Our results show that significantly more infection events were productive with H5N1 virus, with almost 30% of the total infection events being productive, when compared with the 12 and 8% of events with H3N2 and H7N9 viruses, respectively (**Figure 7A**). We also noticed that the clusters of H5N1 virusinfected cells were larger and consisted of even 10 cells while those with H7N9 or H3N2 viruses were smaller and consisted of only a few cells (**Figure 7B**). Next, we analyzed whether the H3N2, H5N1, or H7N9 virus infections were productive by propagating new infectious virions in human Mɸs and moDC. For that, we performed plaque assays from the supernatant samples collected at 1 or 24 h after the infection. We noticed that the amount of infectious virus in the Mɸ cell culture supernatant was increased with all the studied viruses but H5N1 was superior, producing approximately 1,000-fold higher virus titers (**Figure 8A**). In moDCs, the virus titers increased even 10,000 fold in H5N1 virus infection after 24 h (**Figure 8B**). In contrast to the H5N1 virus infection, in infections with H3N2 or H7N9 viruses, the virus titers had a tendency to decrease (**Figure 8B**). In addition to analyzing the amount of secreted infectious virus, we investigated the viral RNA levels at 1 and 24 h in the Mɸs and moDCs supernatants of H3N2, H5N1, and H7N9 virus-infected cells. In H5N1 virus-infected Mɸs and moDC supernatants, total viral RNA levels increased over 100,000 and 100-fold, respectively, whereas in H3N2 or H7N9 virusinfected cells, viral RNA levels increased ca 10,000-fold in Mɸs and 10-fold in moDCs (**Figures 8C,D**). The data indicate that the H5N1 virus spreads more efficiently than H3N2 or H7N9 viruses in human immune cell cultures leading to efficient secretion of newly produced infectious H5N1 viruses.

# H5N1 Virus Infection Is Able to Spread in Human Immune Cell Cultures Irrespective of Induced IFN Responses

Influenza virus infection induces type I and III IFN production in response to the recognition of viral RNAs by host receptors.

GAPDH levels were analyzed to control the equal loading of the samples. A representative experiment out of two (Mɸs) or three (moDC) is shown.

the HA0, HA1, and HA2 of the H3N2 virus, (B,E) against the glycoproteins of H5N1 virus to detect the HA0, HA1 and HA2 of the H5N1 virus, or (C,F) against the HA of H7 virus to detect HA0 and HA2 of the H7N9 virus. In panels (A,B), HA1 and NA co-migrate and thus the heterologous signal is due to cross-reactivity of the NA protein. The levels of GAPDH were used as an internal loading control. A representative experiment out of two is shown.

We and others have previously shown that IFN-α can inhibit H3N2, H5N1, and H7N9 virus replication (12, 27). Here, we show that the H5N1/2004 avian influenza strain replicates and spreads efficiently in Mɸs and moDCs even at very low multiplicity infection (MOI 0.01) (**Figures 2A,B**), although the host cells respond to the infection with maximal cytokine gene expression (**Figures 9** and **10**). The H5N1/2004 virus induced strong IFNλ1, IFN-α1, and IFN-β gene expression in Mɸs (**Figures 9A–C**). This suggests that the H5N1 virus is able to spread irrespective of the activation of antiviral immune responses since otherwise IFNs should inhibit virus replication. Lower virus doses in H3N2 or H7N9-infected moDCs led to clearly weaker IFN responses when compared with H5N1-infected cells at the 24 h time point (**Figures 9A–C**) suggesting that there are likely less infected cells that produce IFNs. To get further view of H5N1/2004 virus induced "cytokine storm," we investigated the expression of chemokine CXCL10 and CCL5 and pro-inflammatory cytokine IL-1β and TNF-α genes in H3N2, H5N1/2004, or H7N9

Figure 4 | Infectivity of the H3N2, H5N1, and H7N9 viruses in the presence of trypsin. Monocyte-derived macrophages (Mɸs) (A) or monocyte-derived DCs (moDCs) (B) were infected with externally added TPCK-trypsin with A/Beijing/353/89 (H3N2), A/Vietnam/1194/2004 (H5N1), and A/Anhui/1/2013 (H7N9) influenza viruses with indicated multiplicity of infection (MOI) values and the samples were collected at 24 h after infection. (A) The proportion of infected Mɸs was analyzed with immunofluorescence microscopy by staining the cells with antibody against respective virus proteins. From each donor at least 300 cells (average ca. 550) were counted. (B) moDCs were stained with antibody against virus proteins and infected cells were analyzed with flow cytometry. From each donor 10,000 events were analyzed.

virus-infected Mɸs (**Figures 9D–G**). Like IFN responses, with low MOI value, also CXCL10, CCL5, and TNF-α responses were higher in H5N1-infected Mɸs with low MOI value than with low MOI values in H3N2 or H7N9 virus infection. Induction of IL-1β gene expression is very low with H3N2, H5N1, and H7N9 viruses (**Figure 9F**) as we have also previously shown in moDCs (12). IFN-λ1, IFN-α1, IFN-β, CXCL10, CCL5, IL-1β, and TNF-α gene expression was also investigated in H3N2, H5N1, and H7N9 virus-infected moDCs (**Figure 10**). Data from moDCs correlate with the data from virus-infected Mɸs though CXCL10 mRNA induction was clearly higher in moDCs than in Mɸs, whereas CCL5 gene was expressed in higher levels in Mɸs. Thus, we next analyzed the expression of antiviral proteins in H3N2, H5N1, and H7N9-infected Mɸs (**Figure 11A**) and moDCs (**Figure 11B**). With all analyzed viruses, the expression of IFN-induced antiviral MxA protein was the strongest at the 24-h time point (**Figures 10A,B**). We also analyzed the expression of phosphorylated transcription factors IRF3 (P-IRF3) and STAT2 (P-STAT2) and noticed that the protein expression of these proteins was weaker in H7N9-infected cells compared with those seen in H5N1 or H3N2 infected cells (**Figures 11A,B**) which correlates well with the gene expression of IFNs (**Figures 9** and **10**) as well as with our previous study (12). The data show that in a multi-cyclic infection the H5N1 virus is able to induce very strong IFN gene expressions and antiviral MxA protein expression.

# DISCUSSION

In 1997, a direct transmission of an avian influenza H5N1 virus from poultry to humans was documented for the first time (28). After that many subtypes of avian influenza A viruses have been reported to cause infections in humans, among them the H7N9 influenza type (4). High prevalence of influenza viruses in birds, the possibility of formation of new highly pathogenic reassortants and the ability of these viruses to spread to humans makes avian influenza A viruses of great global concern and emphasize the importance of virus–host cell interaction studies in human cell systems. In this study, we wanted to characterize different steps of H5N1 virus infection in human immune cells to reveal potential mechanisms explaining the severe clinical outcome of the H5N1 virus infection. In our previous study, we have shown that the avian H5N1 and H7N9 viruses can replicate in human moDCs (12). Here, we show that, unlike the H7N9 or seasonal influenza A viruses, the H5N1 virus is efficiently spreading in human immune cell cultures leading to a productive infection and robust IFN response. The remarkable ability of H5N1 virus to spread in human immune cells and likely other human cell types can be one explanation for the severe disease seen in humans with H5N1 virus infection. Here, we show that this ability is associated with at least two different HPAI H5N1 virus strains, and it may be a universal feature for all highly pathogenic avian-origin influenza A virus subtypes. There may be other viral or host factors regulating avian influenza pathogenicity in humans, since the LPAI H7N9 virus (LPAI in birds) has caused more human infections and deaths than the H5N1 subtype (4). In addition, there is now evidence that the H7N9 subtype has evolved to a HPAI H7N9 strain in birds, and this type of virus has already caused infections in humans (5).

Human infections with the HPAI H5N1 viruses are associated with a high viral load (29), and our data presented in this study confirm that the ability of the H5N1 virus to replicate and spread in immune cell culture is completely different from those of the LPAI H7N9 and seasonal H3N2 influenza A virus infections (**Figures 1**, **2** and **6A,B**). The efficacy of systemic spread of influenza viruses depends on the cleavage of HA0 into HA1

and HA2 subunits and the distribution of appropriate proteases in host tissues. We noticed that the HA0 of H5N1 virus is cleaved very efficiently in human immune cells while the cleavage of the HA0 of H3N2 or H7N9 viruses is inefficient (**Figures 3** and **6D**). However, externally added TPCK-trypsin, which cleaves the HA0 precursor to HA1 and HA2 subunits, increases the proportion of infected cells in H3N2 and H7N9 infection (**Figures 4A,B**). There is clear evidence that the multi-basic HA1–HA2 cleavage site constituting of several lysine or/and arginine residues in the HA molecule of the HPAI strains, in contrast to a mono-basic cleavage site in LPAI and seasonal influenza viruses, regulate the pathogenesis of influenza A viruses. The multi-basic cleavage site in the HA has been shown to be a critical determinant of systemic spread of HPAI H5N1 virus (30). The emerged HPAI form of the H7N9 virus expressing the multi-basic site in HA is showing mammalian adaptation with increased pathogenicity and

respiratory droplet transmission in ferrets (31). Yet, there must be other factors besides the multi-basic cleavage site that contribute to virus replication and spreading in mammals. Schrauwen and coworkers (32) have shown that the insertion of a multi-basic cleavage site in the HA of H3N2 virus enhances the cleavage of HA and replication in MDCK cells, but it did not increase the pathogenicity in ferrets. In addition, the study by Matthaei et al. (27) shows that the human H5N1 isolates replicate more efficiently in human epithelial cells than the avian H5N1 isolates, suggesting that there must be other differences in addition to multi-basic HA0 cleavage site between human and avian isolates that contribute to the pathogenicity of the virus. Zhao and coworkers have shown that a G158N mutation in HA (H3 numbering) of an avian HPAI H5N1 isolate enhanced viral production and induced stronger host immune responses in mammalian cells (33) which

indicates that multi-basic HA1–HA2 cleavage site is not the only factor affecting virulence. Also, our results are consistent with this study of Zhao et al. since both A/Hong Kong/156/1997 and A/Vietnam/1194/2004 H5N1 viruses have asparagine residue at position 158.

It is generally accepted that influenza A virus infection in epithelial cells is productive leading to the release of a great number of progeny viruses from these cells. There is, however, a controversy whether human DCs and Mɸs can efficiently produce infectious influenza A viruses. Bender and coworkers showed that a seasonal H1N1 influenza A virus infection in human DCs is abortive (34). Tate et al. (35) and Ioannidis et al. (36) obtained very similar results in mouse Mɸ and DC model systems with both seasonal H1N1 and H3N2 influenza A viruses. On the other hand, the study by Yu et al. (16) indicates that the

HPAI H5N1 virus can replicate productively in human alveolar Mɸs but the infection with seasonal H1N1 virus was abortive. Also, several other studies show that an infection in monocytederived Mɸs with seasonal H3N2 or HPAI H5N1 viruses seems to be productive (7, 14, 37, 38). In conclusion, it seems that a H1N1 type influenza A virus infection is likely abortive and, consistent with our results, HPAI H5N1 virus infection is productive (**Figures 8A,B**) in human immune cells. Nonetheless, there are inconsistent results on the productivity of seasonal H3N2 influenza A virus infection in human cells. In our analyses, a H3N2 influenza A virus as well as the LPAI human isolate of H7N9 virus appeared to cause abortive infection in human moDCs (**Figure 8B**). Instead, H3N2 and H7N9 virus infections seemed to be productive in Mɸs (**Figure 8A**). However, the productivity of these viruses was clearly lower when compared with that of the HPAI H5N1 virus. We used another way, namely analyzing the relative amount of virus-specific RNA from cell culture supernatants, to indirectly quantitate the amount of secreted influenza virus into cell culture supernatant. Based on this analysis, there was a good correlation between the amount of infectious virus

Figure 8 | The productivity of H3N2, H5N1, and H7N9 infections in human innate immune cells. Monocyte-derived macrophages (Mɸs) or monocyte-derived DCs (moDCs) from four different donors were infected with A/Beijing/353/89 (H3N2), A/Vietnam/1194/2004 (H5N1), or A/Anhui/1/2013 (H7N9) influenza viruses at multiplicity of infection (MOI) of 0.01 and after 1 and 24 h of infection the supernatant samples were collected. (A,B) The infective viral titers produced from Mɸs (A) or moDCs (B) were determined by plaque assay in Madin–Darby Canine Kidney cells. The relative virus titers were calculated over 1 h samples. The horizontal lines represent the geometric means of the results from four blood donors. (C,D) The RNA was isolated from the supernatant samples from Mɸs (C) and moDCs (D) and the viral RNA levels were detected by qRT-PCR. The relative viral RNA expressions were calculated over 1 h samples with ΔCT method. The horizontal lines represent the geometric means from different donors. The statistical significance of differences between H5N1 and H3N2 or H7N9 was determined by Student's *t*-test, \*<0.05.

and the viral RNA levels in the supernatants of H3N2, H5N1, and H7N9-infected Mɸs and moDCs (**Figures 8C,D**). A technical detail may, however, confound the results since in moDC infections the input virus was not washed away. It may be that, since in the plaque assay of the supernatant samples the background is relative high and H3N2 or H7N9 infections are only weakly productive, we are unable to clearly see an increase in the virus titers. The productivity of influenza virus infections in human DCs and Mɸs have remained poorly investigated and thus our result is a significant addition to the present knowledge.

There is limited amount of information of how many new influenza virus particles or infective units a single-infected cell can produce. Baccam et al. (39) estimated that a single H1N1 virus-infected cell of the upper respiratory tract can produce up to 22 new infectious virus particles leading to a productive infection. It is possible that a single H5N1 virus-infected cell produces even more infective progeny viruses, which may explain the fast and efficient spread of the H5N1 virus in our experimental systems. This study shows that in human immune cells the HPAI H5N1 virus can spread faster than the H3N2 or H7N9 viruses. This may contribute to the ability of H5N1 virus to cause a systemic infection associated with a fatal outcome in H5N1 virusinfected humans.

It is commonly thought that the IFN system is the key factor regulating the efficacy of host antiviral responses and the clearance of the virus. However, the high mortality in H5N1

A/Vietnam/1194/2004 (H5N1), or A/Anhui/1/2013 (H7N9) influenza viruses at multiplicity of infection (MOI) values of 1. Cellular protein lysates were collected at 3, 6, and 24 h after infection, samples from four donors were pooled and P-IRF3, MxA, P-STAT2, STAT2, and viral NP and M1 protein expression was analyzed by Western blotting using specific antisera. GAPDH protein expression was analyzed to control equal loading of the samples. A representative experiment out of two (Mɸs) or three (moDCs) is shown.

infection has been associated with an excessive inflammatory response in the lungs of virus-infected individuals (29). Our study shows that the H5N1 virus grows extremely well in human immune cells and it induces strong IFN responses (**Figures 1**, **2**, **9A–C** and **10A–C**), both of which may contribute to the clinical outcome of the H5N1 infection. Our data show that IFN and also pro-inflammatory cytokine and chemokine responses are higher in infection with H5N1 than with H3N2 or H7N9 viruses at low MOI at the 24 h time point after infection. At the same time, we see that after 24 h infection with the H5N1 virus all cells are infected whereas with the H3N2 or H7N9 virus only proportion of the cells is infected. This suggests that there is a positive correlation between the spread of infection and the cytokine responses. Thus, the strong IFN responses and virus replication are not exclusive of each other; for instance, Ngunjiri et al. (40) have shown that the highly pathogenic H5N1 virus is sensitive to the antiviral actions of IFNs. There is also evidence that H5N1 viruses with high replication competency can ultimately overcome the antiviral effects induced by IFN-α and IFN-β (41). Bordi et al. (42) have shown that IFN-λ and IFN-α have antagonistic antiviral activity against Crimean-Congo hemorrhagic fever virus. However, in our previous study (43) in influenza A virus infection, we did not see any antagonistic activity between IFN-α and IFN-λ, or IFN-λ and IFN-γ, or IFN-α and IFN-γ. We have shown that in H5N1 virus infection IFN-α is produced leading to strong MxA protein expression in the cells which proves that IFN signaling is intact in H5N1-infected cells (12). Still, in this study, we clearly show that HPAI H5N1 virus can replicate and spread extremely efficiently in human innate immune cells despite of strong activation of IFN gene expression and MxA protein expression (**Figure 11**). This indicates that due to efficient spread of H5N1 virus and high infectivity in the cell cultures there are more cells which can produce IFNs leading to a cytokine storm associated with H5N1-infected patients. This may in part explain the fatal outcome of a H5N1 virus infection in humans. Although the clinical outcome in humans is similar in H5N1 and H7N9 virus infections, surprisingly the antiviral IFN and pro-inflammatory cytokine gene expression is impaired in H7N9 virus-infected immune cells (**Figures 9** and **10**). Thus, it seems that the mechanism behind the pathogenicity is different in H5N1 and H7N9 virus infection which still demands further investigations.

In this study, we have shown that different types of influenza A viruses may behave very differently in immune cells, primary human monocyte-derived Mɸs and DCs functioning as our model cell systems. While HPAI H5N1 virus efficiently replicated and spread in immune cell cultures seasonal H3N2 and surprisingly LPAI H7N9 viruses showed reduced ability to establish a productive infection. More studies are clearly warranted to further reveal the mechanistic details regulating the ability of HPAI avian influenza viruses to replicate in human primary immune cells and to understand the relationship between the virus and host innate immune system.

# AUTHOR CONTRIBUTIONS

VW performed most of the experiments, analyzed the data, and wrote the manuscript. SM and PÖ helped with experiments. IJ and PÖ together with VW designed the experiments and wrote the manuscript.

# ACKNOWLEDGMENTS

We are grateful to Professor Ron A. M. Fouchier and Dr. John McCauley for providing the H5N1 and H7N9 viruses, respectively. The following reagent was obtained through BEI Resources, NIAID, NIH: Monoclonal Anti-Influenza A Virus H7 Hemagglutinin (HA) Protein, A/Netherlands/219/2003 (H7N7), Clone AT241.375.173 (produced *in vitro*), NR-48980. We thank our technical assistants Johanna Rintamäki, Teija Aalto, Nina Aho, Svetlana Kaijalainen, and especially Hanna Valtonen and Riitta Santanen for assistance in BSL-3 work, and we appreciate the help from the biosafety officer Susanna

# REFERENCES


Sissonen at THL. We also thank our trainees Riia Järvi and Olga Jänisniemi for their assistance in laboratory work.

# FUNDING

The study was funded by the Medical Research Council of the Academy of Finland (grant no. 297329), the Sigrid Jusélius Foundation, the Aatos Erkko Foundation, the Jenny and Antti Wihuri Foundation, the Finnish Cultural Foundation, Foundation for Research on Viral Diseases, and the Oskar Öflunds Stiftelse sr. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Westenius, Mäkelä, Julkunen and Österlund. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Potential Role of MicroRNAs in the Regulation of Antiviral Responses to influenza infection

*Thi Hiep Nguyen1,2, Xiaoming Liu1,2, Zhen Zhong Su3 , Alan Chen-Yu Hsu1,4, Paul S. Foster1,2 and Ming Yang1,2\**

*1Priority Research Centre for Healthy Lungs, The University of Newcastle, Callaghan, NSW, Australia, 2 Faculty of Health and Medicine, School of Biomedical Sciences and Pharmacy, The University of Newcastle, Callaghan, NSW, Australia, 3Department of Respiratory Medicine, The Second Hospital, Jilin University, ChangChun, China, 4 Faculty of Health and Medicine, School of Medicine and Public Health, The University of Newcastle, Callaghan, NSW, Australia*

Influenza is a major health burden worldwide and is caused by influenza viruses that are enveloped and negative stranded RNA viruses. Little progress has been achieved in targeted intervention, either at a population level or at an individual level (to treat the cause), due to the toxicity of drugs and ineffective vaccines against influenza viruses. MicroRNAs (miRNAs) are small non-coding RNAs that play critical roles in gene expression, cell differentiation, and tissue development and have been shown to silence viral replication in a sequence-specific manner. Investigation of these small endogenous nucleotides may lead to new therapeutics against influenza virus infection. Here, we describe our current understanding of the role of miRNAs in host defense response against influenza

#### *Edited by:*

*Shokrollah Elahi, University of Alberta, Canada*

#### *Reviewed by:*

*Laurence Amar, Centre national de la recherche scientifique (CNRS), France Ralph A. Tripp, University System of Georgia, United States*

*\*Correspondence:*

*Ming Yang ming.yang@newcastle.edu.au*

#### *Specialty section:*

*This article was submitted to Viral Immunology, a section of the journal Frontiers in Immunology*

*Received: 30 March 2018 Accepted: 21 June 2018 Published: 04 July 2018*

#### *Citation:*

*Nguyen TH, Liu X, Su ZZ, Hsu AC-Y, Foster PS and Yang M (2018) Potential Role of MicroRNAs in the Regulation of Antiviral Responses to Influenza Infection. Front. Immunol. 9:1541. doi: 10.3389/fimmu.2018.01541*

virus, as well as their potential and limitation as new therapeutic approaches.

Keywords: microRNA, immune responses, influenza virus, infection, inflammation

# INTRODUCTION

Influenza viruses belong to the *Orthomyxoviridae* family of single-stranded, negative sense RNA viruses with segmented RNA genomes (1, 2). There are three genera of influenza viruses including genus A, B, and C. Influenza A virus (IAV) causes significant respiratory infections in humans and global pandemics, while genus B virus can cause epidemics (but not pandemics) and genus C virus only leads to a mild disease (3).

Influenza viruses commonly cause acute respiratory infections and have posed serious threats to public health worldwide for many centuries due to their rapid and frequent mutation and recombination rate. Also, there is frequent and inevitable emergence of novel subtypes with unpredictable pathogenicity and transmissibility (4, 5). Each year, 3 to 5 million individuals experience severe influenza virus infections, with approximately 500,000 annual deaths worldwide (6–8). The clinical symptoms of acute respiratory infection by influenza viruses include high fever, body aches, headache, respiratory tract congestion, pharyngitis, and fatigue. In most cases, these symptoms are resolved in infected healthy subjects after 7–10 days. However, young children, elderly, and patients with chronic disorders such as asthma, chronic obstructive pulmonary disease (COPD), cardiovascular diseases, and diabetes are often at higher risk of developing complications such as severe bronchitis, pneumonia, or worsened medical conditions from influenza (9–14). Moreover, the severity of the disease differs between the virus subtypes (15). Although the pathogenesis and mechanisms of influenza virus infection are largely known, there is little progress in targeted intervention, either at a population level or at an individual level (to treat the cause). This is due to rapid and frequent mutations, and ineffective vaccines and antiviral drugs against influenza viruses. Furthermore, frequent antigenic changes (drift and shift) and constant emergence of drug resistant subtypes/ strains also undermine the effectiveness of current anti-influenza approaches (16–20). Therefore, it is urgently required to develop more efficient approaches for prevention and treatment of the disease.

# MICRORNAs (miRNAs)

MicroRNAs are a class of short non-coding single-stranded RNA sequences of about 20 bp first described two decades ago that negatively regulate gene expression in eukaryotes (21–24). These small RNAs are transcribed as long hairpin primary RNAs (pri-miRNAs) by RNA polymerase II. In nucleus, pri-miRNAs are cleaved by the microprocessor complex including Drosha ribonuclease III and the RNA-binding DGCR8 protein to form hairpin precursor miRNAs (pre-miRNAs, ~70 bp). Pre-miRNAs are exported to the cytoplasm by exportin-5 protein, belonging to the Ran-dependent nuclear transport receptor family, and are further cleaved by cytoplasmic endoribonuclease Dicer to form mature RNAs (**Figure 1**). Each miRNA gene has been recognized to generate two mature miRNAs that are designated as -3p miRNA and -5p miRNA (25–27). Both of them that can coexist are functional by associating with the RNA-Induced Silencing Complex. Mature miRNAs often are known to bind to 3-untranslated regions (UTRs) of target mRNAs to regulate gene expression. Most miRNA:mRNA interactions involve nucleotides 2–7 of miRNAs, a region called seed. Seed-based interactions lead to mRNA destabilization and/or translation inhibition.

Functional consequences of changed miRNA expression have been explored for diagnosis, prognosis, and severity of a wide range of diseases including infectious disease, autoimmune diseases, and cancer (28–31). Numerous investigations using anti-miRNA oligonucleotides, miRNA mimics/inhibitors, or mice deficient of a specific miRNA have demonstrated that a single miRNA can have extensive and crucial effects on the physiologic and pathological processes and that alterations in the function of miRNA can result in biological dysfunction and diseases (24, 31, 32). Despite the fact that the critical role of miRNAs in orchestrating cell differentiation, proliferation, and metabolism is well known (24, 33–35), we are only beginning to understand the contribution of these small RNAs to the innate immune responses to viral infections, to the regulation of gene expression programs, and immune cell activation (32, 34). In this review, we focus primarily on the role of miRNAs in regulating innate immune responses to IAV infections, and the potential use of miRNAs in the treatment of IAV infection.

# MICRORNAs AS DIAGNOSIS MARKERS

MicroRNAs are found in the intracellular niche and the extracellular fluids encapsulated in exosomes including blood plasma, serum, urine, saliva, and semen (36–38). These small molecules are also detected in body fluids independently of intracellular and exosome compartments (38). Interestingly, the differential expression of miRNAs in these compartments is linked to the development of IAV infection (29, 39–42), indicating that these small RNAs can be used as the diagnosis markers for the disease (**Table 1**). Evaluation of peripheral blood mononuclear cells (PBMCs) from critically ill patients with swine-origin pandemic H1N1 infection by qRT-PCR and receiver operating characteristic (ROC)/area under ROC (AUC) curve analyze has revealed that increased levels of miR-148 (>2-fold) and decreased levels of miR-31 and miR-29a (>2-fold) are valuable biomarkers for severe influenza virus infections with the AUC value ranging from 0.881 to 0.951 (39). The ROC tool is commonly used to evaluate the diagnostic accuracy of differentially expressed

cleaved to two strands by endoribonuclease Dicer, one strand becomes a mature miRNA and silence target mRNAs through mRNA degradation or translation repression and the other is degraded.



*a* ↓*, downregulation;* ↑*, upregulation.*

miRNAs for differentiating between IAV infection patients and healthy controls, and the accuracy is measured by AUC (39, 43). In another study, increased levels of miR-34c-3p (>4-fold) and decreased levels of miR-29a-3p, -30c-5p, and -181a-5p (>2-fold) have been demonstrated to be associated with the infection in the throat swab samples of H1N1-infected patients using qRT-PCR and ROC methods (43). The link between low levels of miR-29a-3p (threefold) and IAV infection has been further confirmed in the throat swabs of H1N1 infected patients using non-PCR MARS [microRNA-RNase-SPR (surface plasmon resonance)] assay (44). A similar study of peripheral blood samples from H3N2- or H1N1-infected patients using miRNA microarray and stem-loop PCR has identified that 14 miRNAs were linked to the pathogenesis of the disease (40). Among them, the levels of miR-229-5p, -335, -664, and -1260 were increased greater than twofold, while the levels of miR-18a, -26a, -30a, -34b, -185, -576-3p, -628-3p, -665, -765, and -1285 were decreased greater than fourfold. Expression of six (miR-26a, -335, -576-3p, -628- 3p, -664, and -1260) of these miRNAs was confirmed in H1N1 infected A549 cells and Madin Darby Canine Kidney (MDCK) cells (40). Moreover, H7N9 has been demonstrated to cause more severe infection in humans, and to-date, this H7N9 subtype has resulted in 1,533 human infections with 592 deaths in 2017 (45). MiRNA microarray and qRT-PCR analyses using serum samples from H7N9 infected individuals showed a slightly different miRNA signature, with miR-17, -20a, -106a, and -376c being upregulated (>1.5-fold) (42). The ROC curve analysis was used to discriminate H7N9 infected patients from healthy controls for each miRNA with AUC values ranging from 0.622 to 0.988, while its value for a combination of these four miRNAs is 0.96 (42). Furthermore, miR-150 levels assessed by qRT-PCR have been found to be significantly higher (>1.5-fold) in critically ill patients than those with milder disease and healthy controls in a human study of H1N1 infection, indicating the association of this miRNA with poor disease outcome (41). Collectively, these earlier studies indicate that infection not only by different subtypes of IAVs but also by the different strains of the same IAV subtype with varying pathogenicity elicit different miRNA expression patterns in similar samples, which can be valuable diagnostic and/or prognostic markers for influenza infection and severity of the disease. Identification of these miRNAs may greatly aid the design of analytical kits for rapid and precise diagnosis of IAV subtypes of infection, and potentially to develop customized therapeutic approaches to control infection.

# MICRORNAs DIRECTLY TARGET INFLUENZA VIRAL RNAs

Influenza A virus consists of eight gene segments that encode for 12 viral proteins including surface glycoprotein [hemagglutinin (HA) and neuraminidase (NA)], nucleoprotein (NP), two matrix proteins (M1 and M2), three polymerase complex proteins PB1, PB2, and PA, four non-structural proteins (NS1, NS2, PA-X, and PB1-F2) (2, 46, 47). HA and NA proteins predominantly regulate virus entry and exit from host cells, and their genes are the major genetic segments for influenza antigen drift and shift by genetic mutation and reassortment to create new strains/subtypes. In contrast, other IAV viral proteins are more conservative, which is essential for IAV replication. For example, viral polymerase complex proteins (PA, PB1, PB2) and NP form a viral ribonucleoprotein (vRNP), a minimal functional unit for influenza virus replication. M1 forms a coat inside the viral envelope and binds to viral RNA. Therefore, exploration of those miRNAs that directly target those conservative viral sequences could uncover novel therapeutics to control influenza replication and propagation.

Indeed, several lines of evidence have implied the feasibility of this concept. For example, miR -323, -491, and -654 destabilize PB1 mRNA by targeting the conservative region, as demonstrated in H1N1 infected cells that are treated with plasmids carrying those miRNA mimics or inhibitors, respectively (48). A similar investigation has shown that miR-485 directly binds to a conserved site of PB1 mRNA to regulate viral replication, in H5N1-infected HEK293T cells following miR-485 mimics or inhibitor treatment (49). Furthermore, multiple miRNAs may target the same seed sequence to regulate IAV replication. Khongnomnan and colleagues, through *in silico* analysis and a luciferase reporter assay have reported that the same conservative region of PB1 mRNAs of H1N1, H5N1, or H3N2 subtypes is targeted by miR-3145 (50). Neutralization of this miRNA by using plasmid encoded antimiRNA oligonucleotides restored the expression of PB1 mRNA and miR-3145 mimics treatment reduced PB1 expression in H5N1-, H1N1-, or H3N2-infected A549 cells (50). M1 is the most abundant protein in the IAV viral particle and regulates vRNP export, virus assembly and budding, and virus–host interactions (51, 52). Ma and colleagues have reported that let-7c precursor diminishes H1N1 replication by binding to the 3′-UTR of M1 mRNA and that let-7c inhibitor reinstates the expression of M1 protein and influenza infection in A549 cells (53). Certain micro-RNAs have also been shown to inhibit the expression of IAV viral proteins, not only in a direct manner but also through regulations of other host factors that affect viral replication. For example, miR-33a mimic suppressed the expression of NP and M1 proteins by directly binding to the 3′-UTR of Archain 1 (ARCN1) RNA in HEK293T, A549, and Hela cells infected with H1N1, H9N2, or H3N2, resulting in greatly decreased virus replication (54). ARCN1 is an important component of human coatomer protein complex, which regulates protein transport from the Golgi body to the endoplasmic reticulum and critically modulates influenza virus entry to host cells, viral membrane protein expression, and assembly (55, 56). Treatment with miR-33a inhibitor recovered the expression of ARCN1, NP, and M1 proteins, and thus increased H1N1, H9N2, or H3N2 replication (54). In the same study, miR-33a has also been shown to attenuate the replication of H1N1, H9N2, or H3N2 by reducing vRNP activity through an ARCN1-independent pathway in HEK293T cells (54), suggesting the multiple functions of this miRNA. A recent investigation has identified that miR-21 targets NP, PB1, PB2, PA, NA, and HA segments of H1N1, by using infected miR-21-deficient MDCK cells (28). It is promising that targeting NP segment or combination of both PA and NA segments of IAV simultaneously reduced IAV replication greater than twofold, as compared to other treatments (e.g., targeting sole segment and a combination of PA and HA) (28). Although the role of miRNAs in the pathogenesis of IAV infection should be further investigated, targeting these small viral RNAs may provide alternative approaches to reduce influenza infection by directly inhibiting expression of conserved viral proteins (e.g,. PB1, NP, or M1), regardless of the viral antigen drift and shift (**Figure 2** and **Table 2**).

# MICRORNAs CONTROL IAV-INDUCED INFLAMMATORY RESPONSES

Bronchial epithelial cells and pulmonary innate immune cells such as alveolar macrophages, dendritic cells, and natural killer cells provide the first line of defense against influenza infection (57, 58). Upon infection, molecular patterns of influenza viruses are recognized by host molecular pattern recognition receptors (PPRs) including retinoic acid-inducible gene I (RIG-I)-like receptors and toll-like receptors (TLRs) (59–61). These patternrecognition receptors are important sensors that recognize

Figure 2 | (A) Influenza structure: a lipid bilayer envelope containing glycoproteins M1 and M2 ion channel. Hemagglutinin and neuraminidase proteins on the outside of the envelope. Eight RNA genome segments inside the envelope encoding for three polymerase complex proteins (PB1, PB2, and PA), nucleoproteins (NPs), M1 and M2 matrix proteins, and non-structural proteins (NS1, NS2, PA-X and PB1-F2). (B) Cellular microRNAs (miRNAs) modulate influenza replication. Host cellular miRNAs inhibit influenza replication through targeting viral RNAs and proteins that are essential for influenza replication and translation such as PB1, M1, and nucleoprotein.



infectious pathogens and drive host defense responses. Among them, TLR3 and TLR7 induces the activation of interferon (IFN) regulatory transcription factor (IRF)-3 (IRF3)/IRF7 and TIR-domain containing adaptor inducing IFN-β (TRIF)/NF-κB signaling pathways after interacting with IAV RNAs, which leads to the production of type-I (IFN-α and IFN-β) and -III (IFNλ) IFNs and proinflammatory cytokines and chemokines (58, 62–65). RIG-I is a cytoplasmic RNA helicase that recognizes short double-stranded RNA produced during viral replication (66), and viral genomic single-stranded RNA (ssRNA) bearing 5′ phosphates (67). By binding to IAV ssRNAs, RIG-I facilitates the activation of IRF3 that directly regulate the production of type-I and -III IFNs (58, 62, 68). IAV infections are known to cause severe pro-inflammatory cytokine storm in the lung by the induction of overproduced aforementioned cytokines, which can also spread into systemic circulation, leading to severe symptoms such as leukopenia (69).

MicroRNAs have been shown to play key roles in the regulation of pro-inflammatory intracellular signaling pathways during influenza infection (32, 70–73). For example, increased expression of the IRF5 gene has been shown to be correlated with decreased levels of miR-302a (2.5-fold) in throat swab samples and PBMCs from patients with influenza infection, as compared to healthy controls (71). Treatment with miR-302a mimics decreased IRF5 expression by binding to its 3′-UTR and reduced IRF5-regulated production of IFN-β, TNFα, IL-6, IL-8, CCL2, and CCL5 in H1N1-infected PBMCs, leading to higher H1N1 viral production (71). Recently, miR-144 has been demonstrated in a mouse model to inhibit anti-IAV host responses by targeting the TNF receptor-associated factor 6 (TRAF6)/IRF7 signaling axis, which underpins type-I IFN responses against H1N1 infection (72, 74, 75). Another miRNA, miR-146a, has also been shown to directly downregulate TRAF6 in H3N2 infected human nasal epithelial cells (73). These findings suggest the importance of those miR-NAs in host immunity against IAV infection by targeting TRAF6.

NF-κB has been shown to play important roles not only in the production of pro-inflammatory genes in response to IAV infection but also in propagation of influenza viruses (76–79). Multiple miRNAs have been identified to regulate NF-κB activation, by targeting the 3′-UTRs of its components (80–82). NF-κB inhibitor β (NFKBIB, also known as IκBβ) is a regulatory protein for NF-κB that prevents nuclear translocation of NF-κB (RelA/ p65), and subsequent transcription of its target genes (83, 84). Treatment of H1N1-infected human primary bronchial epithelial cells (pBECs) with miR-4776 inhibitor led to increased expression of NFKBIB, resulting in lower viral replication (82). In contrast, miR-4776 mimic decreased the expression of NFKBIB and increased viral replication (82). Gui and colleagues have shown that H3N2 infection suppresses miR-302c expression in A549 cells (80). MiR-302c inhibitor treatment restored the expression of its targeting molecule, NF-κB-inducing kinase (NIK) that is a critical component of NF-κB pathway. MiR-302c mimic treatment prevented the nuclear translocation of NF-κB and downregulated IRF3/7 expression, leading to decreased expression of IFN-β (80). In a study of H1N1- or H3N2-infected A549 cells, miR-132, -146a, and -1275 simultaneously decreased the transcription and expression of interleukin-1 receptor-associated kinase 1, a key component of NF-κB pathway (81). These three miRNAs also target mitogen-activated kinases (MAPK) 3 that plays an important role in the activation of pro-inflammatory MAPK pathway. Those miRNAs associated with pro-inflammatory signaling pathways are delineated in **Figure 3** and **Table 3**. Interestingly, miRNAs may contribute to virus replication by suppressing host antiviral responses. For example, Dong and colleagues recently have demonstrated with H1N1- or H3N2-infected A549 cells that miR-9 promotes IAV replication through the suppression of monocyte chemoattractant protein 1-induced protein (MCP1P1) (85). MCP1P1 is a ribonuclease that plays an important role in antiviral immune responses (86) and inhibits IAV replication by decreasing the production of M and NP proteins (85). Treatment with a miR-9 mimic greatly augmented the production of NP and M1 proteins and IAV replication by decreasing production of MCP1P1. By contrast, an inhibitor of miR-9 blocked IAV replication (85). We have also recently shown that miR-125a/b directly targets A20 deubiquitinase, an enzyme that degrades receptor interacting protein 1 and inhibits NF-κB activation (87). Infection with H3N2 or H1N1 led to increased expression of miR-125a/b, which suppressed A20 deubiquitinase production and resulted in elevated NF-κB activity and production of pro-inflammatory cytokines in pBECs of COPD patients and in *in vivo* model (87). Similarly, the association between miR-125a/b and the expression of A20 deubiquitinase was observed in the mouse model of COPD (87). Collectively, miRNAs with altered expression play important roles in IAV replication, demonstrating their potential roles of antiviral host defense responses.

# MICRORNAs MEDIATE THE PRODUCTION OF ANTIVIRAL CYTOKINES

Three RIG-I like receptors are crucial in antiviral host defense and consists of RIG-I, melanoma differentiation-associated

by the damage-associated molecular patterns molecules released in influenza virus-infected cells and trigger TLR4-MyD88-signaling pathways. Induction of these TLRs can lead to activation of NF-κB, IRF 3, 5, and 7 and induce expression of type-I and -II IFN, IFN-stimulated genes, and inflammatory genes. Some miRNAs regulate these pathways through targeting critical components such as TRAF6, IRF3, IRF5, IFR7, interleukin-1 receptor-associated kinase 1, and IκBβ. Within the infected cells, RIG-I detects the 5-triphosphorylated RNA of replicating viral genomes in cytosol and associates with mitochondrial antiviral signaling protein (MAVS) to induce pro-inflammatory cytokines and type-I IFN. miRNAs can modulate this pathway by directly targeting RIG-I, MAVS, or NF-κB-inducing kinase.

protein 5 (MDA5), and laboratory of genetics and physiology 2 (LGP2) (88–90). RIG-I plays a central role in the induction of immune responses against IAV, by recognizing short RNA (67, 91). By contrast, MDA5 binds to long RNA and LGP2 acts as a positive regulator of RIG-I and MDA5 (92, 93). RIG-I recognizes viral RNA in the cytosol and interacts with the mitochondrial antiviral-signaling protein (MAVS), which is localized to the outer mitochondrial membrane (94). Aggregation of RIG-I, TRIM25, and MAVS then leads to the activation of IRF3/IRF7 by phosphorylation, which then induces the expression of type-I and -III IFNs (59, 94–96). These IFNs then bind to the respective IFN receptors on the same/neighboring cells and stimulate the expression of over 300 IFN-stimulated genes (ISGs), such as protein kinase R to limit viral replication. IAV-induced expression of miR-125a/b has also been shown to directly inhibit the expression of MAVs, leading to reduced production of type-I and III IFNs in pBECs of COPD patients and mouse model of COPD (87). Several miRNAs have been reported, either directly or indirectly, to regulate the RIG-I pathway for the antiviral response to IAV infection (97–99), this interaction is also shown in **Figure 3** and **Table 3**. For example, H1N5 or H1N1 infection increased the expression of miR-136 and miR-194 and both have been independently shown to suppress IFN-β expression by binding to the 3′-UTR of RIG-I transcript in IAV-infected A549 cells (97, 98). Furthermore, miR-483-3p, is highly expressed in the lung during infection of mice with H1N1, H5N1, or H7N9 (99). Transfection of H1N1-, H5N1-, or H7N9-infected MLE-12, a mouse cell line of lung epithelial cells, with miR-483-3p mimic led to decreased viral replication by targeting the transcript of RING-finger protein 5, which negatively regulates RIG-I signaling pathway (99). MiR-132 has also been shown to directly targets p300, an important component of IFN-β enhanceosome, which leads to reduced induction of IFN-β (100).



*NIK, NF-*κ*B-inducing kinase; IRF, interferon regulatory factor; IRAK1, interleukin 1 receptor-associated kinase 1; MAPK3, mitogen-activated kinase 3; HDAC, histone deacetylase; MCPIP1, monocyte chemoattractant protein 1-induced protein 1; pBECs, primary bronchial epithelial cell; RNF5, RING-finger protein 5; NFKBIB, NF-*κ*B inhibitor* β*; MLE-12, mouse cell line of lung epithelial cells; USP3, ubiquitin-specific protease 3; LIF, leukemia inhibitors factor; NEK7, NIMA-related kinase 7.*

New evidence has emerged on the potential role of miRNAs in IAV-induced host defense responses and in the modulation of the production of cytokines. MiR-26a significantly inhibits IAV replication by promoting the type-I IFN production and subsequent expression of ISGs in H1N1-infected A549 cells (101). IAV infection decreases the expression of histone deacetylase 1 (HDAC1) that plays an important role in the activation of type-I IFN response against IAV infection (102, 103). Furthermore, H1N1 or H3N2 infection in A549 cells induced increased levels of miR-449b (>7-fold) (104). A recent investigation has also linked the increased expression of miR-449b (>6-fold) to the decreased expression of HDAC1 and the increased expression of IFN-β in H1N1- or H3N2-induced in A549 cells (102). Interestingly, treatment of miR-449b mimics further suppressed the expression of HDAC1 and enhanced the expression of IFN-β in H1N1- or H3N2-infected A549 cells (102), suggesting the important role of this miRNA in host defense against influenza virus infections.

Interestingly, certain miRNAs have been shown to have opposing effects on regulating the antiviral response when host cells are infected at different doses of specific viruses. MiR-485 is induced by IAV infection and inhibits RIG-I pathways at low levels of IAV infection, which suppresses host antiviral responses and enhances virus replication (49). By contrast, this miRNA abates virus replication by degrading PB1 transcripts at higher levels of IAV infection (49). It is likely that varying multiplicity of infection triggers slightly different antiviral signaling pathways. Although the regulatory targets of miR-485 are currently unclear, this miRNA likely regulates those factors that are highly expressed in these conditions. Hence, the miRNA biological function should be carefully and fully examined before developing therapeutic approaches.

# MANIPULATION AND DELIVERY OF miRNA

Although miRNAs have been increasingly recognized in recent years as potential therapeutic targets for treating influenza infection, the successful miRNA manipulation can be difficult because of many factors such as short half-life of miRNAs, low cellular uptake and expression, pre-matured elimination by host immune cells, interruption of endogenous RNA processes and off-target effects. In particular, targeting a single miRNA may have limited success for treatment because an understanding of the molecular mechanisms underpinning many complex diseases (e.g., asthma, autoimmunity, and cancer) remains rudimentary. To elucidate the function of a miRNA family simultaneously, traditional antisense methods for a single miRNA are inadequate and laborious for targeting multiple miRNAs; to create genetically modified lab animals is even more arduous. As such, microRNA sponges have been designed to inhibit the function of a miRNA family by creating a single RNA sequence that consists of several tandem miRNA-binding sites for all the members (105). In fact, one of the unique features of miRNA function is that a family of miRNAs share an almost identical seed sequence, with often only a few nucleotides difference, although they may be expressed from different genomic loci (106–108). Therefore, targetting seed sequence may be particularly valuable using miRNA sponges to understand the pathogenesis of IAV infection and to treat the disease. A miRNA sponge could be designed to target those miRNAs that promote virus replication and IAV-induced inflammation.

How to effectively deliver miRNA sponges or mimics to manipulate host miRNAs is challenging for clinical application. For this reason, it is essential to carefully design and select candidate miRNA sequence. Non-specific responses induced by those molecules should also be examined with great caution. Any unexpected effect associated with miRNA's target should be quantified for the safety and success of miRNA manipulation approaches. Indeed, various types of strategies have been developed to optimize the delivery methods, including chemical modification of miRNA sponge molecules or to encapsulate them with macromolecules (e.g., polyamine, polyethylenimine, and basic complexes). Furthermore, the promoters used by miRNA sponge should be strongest and most suitable for the cells of interest to achieve highest efficacy. If the targeted miRNAs are expressed in many cell types in multiple tissues, the tissue- or cell-specific promoters could be employed to achieve precise expression, which could minimize potential side effects.

Many vectors that comprise plasmid, replication-deficient virus or transposons have been developed to accomplish successful gene interference *in vivo* (109–113, 124). Among these vectors, lentiviral vectors have been widely employed in the study of normal tissue physiology and processes of disease in animal models (114, 115). For example, lentiviral vectors carrying miR-30 mimic can inhibit viral replication in H1N1-infected MDCK cells by targeting viral NP and PB1 transcripts (116). IAV itself can also be modified to express exogenous miRNAs and modulate viral replication and for the treatment of the diseases (117), as incorporation of an artificial miRNA into IVA genome does not cause viral sequence instability or interfere with viral replication (117). Langlois and colleagues have shown that recombinant H1N1 expressing artificial miR-124 does not inhibit the function of other miRNAs only with limited repression of miR-124-star target by luciferase reporter system in transfected hamster kidney cells (118), suggesting the specificity of this delivery method. Indeed, live-attenuated IAV delivery has shown great potential in the development of more efficient IAV vaccine and in the treatment of infection respiratory diseases by carrying customized artificial miRNAs (119, 120). This notion is supported by a recent study, showing that exogenous miR-155 encoded by modified X31 IAV augments IAV-specific CD8<sup>+</sup> T cell response and neutralizing antibody production in a mouse model of IAV infection (120).

Adenoviral and adenovirus-associated viral (AAV) vectors are also valuable to silence candidate miRNAs *in vivo*. Indeed, recent progresses have been made in delivering a miRNA mimic to explore the therapeutic potential of these vectors in animal models (121, 122). However, adenoviruses induce off-target host defense response because they infect a wide range of cells (123). On the other hand, AAV vectors cause very few side-effects to host because they integrate at limited and defined location in the genome of transfected targets (121, 123). This unique feature of AAV vectors thus minimizes the chance of mutational insertion and induces effective gene silencing following either systemic or tissue-specific injection (121, 123). Furthermore, non-viral transfection approaches using nanoparticles and liposome have also attracted attention in the establishment of miRNA intervention. Thus, although delivery strategies are far from being optimal yet, significant progress has been achieved recently toward targeted therapy with limited off-target effects.

# LIMITATION

Current understanding of the roles of miRNAs in the pathogenesis of influenza is limited by a lack of sufficient characterization of the miRNA-associated molecular pathways in the context of influenza infection, IAV replication, and host immunity. Indeed, great challenge still exists in the field of the identification of miRNA targets within a living organism and in host defense processes that have many layers of molecular and cellular elements with specific spatiotemporal patterns. One of miRNA biological features is that they target multiple mRNAs; therefore, pathophysiological outcomes observed by modifying their function may correlate with subtle changes in the levels of diverse target mRNAs. Although *in vitro* assays using luciferase reporter system and miRNA mimics and inhibitors are usually employed to verify an interesting target transcript, caution should be taken to interpret the important biological function in dynamic living systems by exclusively relying on these methods. Furthermore, as miRNAs potentially act as master regulators of disease and inflammation, ill-designed manipulation of a miRNA may generate unexpected side-effects. This is particularly important when considering the observations that different species infected with different IAV strains/subtypes generate different miRNA signatures. In addition, delivery methods that limit miRNA mimics and inhibitors precisely to the infected and inflamed tissues are also needed to be further developed as it is essential for not causing any disruption to normal function of surrounding tissues.

Although many candidate miRNAs largely play suppressive roles in IAV infection, certain host miRNAs may assist viral replication. One example is that the miR-664 was highly upregulated approximately fourfold in H7N9-infected A549 cells, and treatment with miR-664 inhibitor reduced H7N9 replication (125). MiR-664 is predicted by *in silico* pathway analysis to target the 3′-UTRs of leukemia inhibitory factor (LIF) and NIMA-related kinase 7 (NEK7) whose activation leads to lower H7N9 replication (125). MiR-144 increased IAV infection by suppressing the activity of TRAF6-IRF6 axis posttranscriptionally as demonstrated in H1N1 infected miR-144 deficient mice and mouse lung epithelial cells (72). Although a specific miRNA may play either pro- or anti-IAV role, numerous investigation suggest that miRNAs have great potential as diagnostic biomarker and treatment of human diseases when considering the profound biological function regulated by these small RNAs and their extensive links to IAV infection.

# CONCLUSION

Based on the unique features of miRNAs, a new generation of IAV vaccine may be developed by incorporating miRNA response elements (MRE, miRNA recognition sequences) into viral genomic segments such as NP, NS, or PB1. An attempt to generate novel attenuated IAV vaccine has yielded promising results by inserting a let-7b MRE into H1N1 PB1 gene, which significantly reduced viral replication in bronchial epithelial cells (126). Although preliminary, this method to generate attenuated IAV vaccine has been proven effective in a mouse model (122).

Manipulation of miRNAs needs to be approached with caution for the reason that intervention of miRNA function may predispose to impaired immunity, cancer, or other unforeseen biological abnormalities. However, miRNA has become more and more attractive as diagnostic biomarkers and potential clinical intervention targets as effective prevention and treatments for IAV infection is poorly available. Furthermore, direct targeting of key miRNAs that underpin IAV infection may lead to new and more specific therapeutic interventions as these small RNAs are

# REFERENCES


implied in regulating specific gene clusters triggered by infection (e.g., cytokine driven inflammation). It is particularly important to explore those miRNAs that can both degrade IAV RNAs and alleviate virus-induced inflammation, as they may concurrently control both virus replication and over-reactive immune responses. Understanding the role of miRNA in fundamental processes associated with IAV infection is necessary to fully characterize their potential in disease diagnosis and prognosis and ultimately for the treatment of disease.

# AUTHOR CONTRIBUTIONS

TN and MY wrote and edited the paper*.* XL, ZS, AH, and PF edited the paper.

# FUNDING

This work was supported by project grant from the National Health and Medical Research Council (NHMRC) of Australia.


mammalian cells, stem cells and transgenic mice by RNA interference. *Nat Genet* (2003) 33:401–6. doi:10.1038/ng1117


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Nguyen, Liu, Su, Hsu, Foster and Yang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# The Hurdles From Bench to Bedside in the Realization and Implementation of a Universal Influenza Vaccine

*Sophie A. Valkenburg1,2\*, Nancy H. L. Leung2 , Maireid B. Bull 1,2, Li-meng Yan2 , Athena P. Y. Li1,2, Leo L. M. Poon2 and Benjamin J. Cowling2*

*1HKU Pasteur Research Pole, The University of Hong Kong, Pokfulam, Hong Kong, 2WHO Collaborating Centre for Infectious Disease Epidemiology and Control, School of Public Health, The University of Hong Kong, Pokfulam, Hong Kong*

#### *Edited by:*

*Alan Chen-Yu Hsu, University of Newcastle, Australia*

#### *Reviewed by:*

*Stephen Kent, University of Melbourne, Australia Alice Sijts, Utrecht University, Netherlands*

*\*Correspondence: Sophie A. Valkenburg sophiev@hku.hk*

#### *Specialty section:*

*This article was submitted to Viral Immunology, a section of the journal Frontiers in Immunology*

*Received: 27 April 2018 Accepted: 14 June 2018 Published: 02 July 2018*

#### *Citation:*

*Valkenburg SA, Leung NHL, Bull MB, Yan LM, Li APY, Poon LLM and Cowling BJ (2018) The Hurdles From Bench to Bedside in the Realization and Implementation of a Universal Influenza Vaccine. Front. Immunol. 9:1479. doi: 10.3389/fimmu.2018.01479*

Influenza viruses circulate worldwide causing annual epidemics that have a substantial impact on public health. This is despite vaccines being in use for over 70 years and currently being administered to around 500 million people each year. Improvements in vaccine design are needed to increase the strength, breadth, and duration of immunity against diverse strains that circulate during regular epidemics, occasional pandemics, and from animal reservoirs. Universal vaccine strategies that target more conserved regions of the virus, such as the hemagglutinin (HA)-stalk, or recruit other cellular responses, such as T cells and NK cells, have the potential to provide broader immunity. Many pre-pandemic vaccines in clinical development do not utilize new vaccine platforms but use "tried and true" recombinant HA protein or inactivated virus strategies despite substantial leaps in fundamental research on universal vaccines. Significant hurdles exist for universal vaccine development from bench to bedside, so that promising preclinical data is not yet translating to human clinical trials. Few studies have assessed immune correlates derived from asymptomatic influenza virus infections, due to the scale of a study required to identity these cases. The realization and implementation of a universal influenza vaccine requires identification and standardization of set points of protective immune correlates, and consideration of dosage schedule to maximize vaccine uptake.

#### Keywords: influenza virus, universal vaccine, T cell, hemagglutinin-stalk, clinical trials

# INTRODUCTION

Influenza A viruses have over 18 different hemagglutinin (HA) subtypes, and continual antigenic drift of seasonal H3N2 and H1N1 viruses generates new variants. In addition, there are distinct lineages of influenza B viruses that also exhibit antigenic drift, meaning there is a plethora of influenza viruses that pose a threat to public health (1). Reports of global influenza infection rates estimate that up to 18% of the population can be infected during annual influenza epidemics (2), causing excess morbidity and mortality resulting in projected economic losses of nearly US\$87 billion (3). Influenza vaccines are the most widely used vaccines in the world due to annual updates on circulating strains and health authority recommendations for at risk groups (4). The groups most commonly targeted for influenza vaccination programs are children and elderly, pregnant women, immunocompromised, and healthcare workers (HCWs). Inactivated influenza vaccines (IIV) administered intramuscularly have been available since the 1940s and progressive developments have been made to increase breadth of immunity provided by the vaccine, from monovalent to bivalent and then trivalent, to most recently quadrivalent formulations (5). The use of split and subunit vaccines has provided a more purified formulation, and the use of improved adjuvants with reduced side effects in recent years, such as MF59 and AS03 has enabled antigen sparing and increased immunogenicity of vaccine antigens (5, 6). One important advance was the release of live-attenuated influenza vaccines (LAIV) by MedImmune to American markets in 2003, delivered as a nasal spray (7). The quadrivalent cell-grown recombinant HA protein vaccines, FluCelVax (Seqirus) available from 2012 (8) and FluBlok (Sanofi Pasteur) available from 2013 (9), provide an expedient pipeline for pandemic vaccine responsiveness and avoid egg adaptations generated during vaccine production.

A combination of issues exists for the current influenza vaccines (10), including egg adaptations (11), lag between strain selection and vaccine availability (10), and breadth and duration of immunity (12). Annual vaccine effectiveness (VE) is variable and contingent upon antigenic distance between vaccine and circulating strains and the individual's immune history (13). Shortcomings in VE for IIV and LAIV are repeatedly reported with a recent average VE found to be 78.4% and 30.7% against H1N1pdm09 infections, respectively, in 2- to 17-year olds in 2015/2016 in the UK, US, Canada, and Finland (14), while IIV VE reported by the CDC ranges from 10 to 60% from 2004 to 2016 (15). Therefore, current vaccines are not effective enough, with negative or low VE reported, and LAIV does not appear to improve upon VE over IIV consistently (16–18), hence, the need for universal vaccines. Furthermore, current IIV VE decreased over time by one-third from 3 to 6 months post vaccination (19), and reduced VE estimates over time were seen for LAIV (20, 21). Targeting the elderly for vaccination is a logical step as they are the demographic that have the highest morbidity and mortality risk from an influenza virus infection; however, current vaccines are even less effective at conferring protection within this susceptible age group (22).

# CRITERIA FOR DESIGN OF NEXT-GENERATION UNIVERSAL INFLUENZA VACCINES

WHO published in 2017 the Preferred Product Characteristics for Next-Generation Influenza Vaccines which lays out the targets for influenza vaccine development over the next 5 and 10 years (23). In the first 5 years, the WHO encourages the evaluation of currently available vaccine and vaccine technologies to achieve greater protection against vaccine-matched or drifted influenza strains and protection against severe influenza for at least 1 year. In 10 years, by 2027, the WHO encourages research and development in next-generation vaccines to provide universal protection against severe influenza A illness for at least 5 years. In addition, a strategic work plan for the design of a universal vaccine has been outlined by the NIH NIAID (10, 24, 25). To achieve the goal of a universal influenza vaccine capable of providing protection beyond 1 year and with broader immunity against antigenically diverse strains, the work plan identified areas for expanded research efforts to address this goal, with an emphasis on research in the areas of (1) influenza transmission, natural history, and pathogenesis; (2) development of influenza immunity and correlates of protection; and (3) rational design of universal influenza vaccines.

Ultimately, an ideal universal influenza vaccine would provide protection (1) against seasonal influenza epidemics by drift variants between seasons or pan-influenza A and B viruses, (2) against influenza pandemics with limited prior population immunity, and (3) against zoonotic (e.g., avian) influenza infections with severe disease outcomes. However, this "ideal" vaccine is still stuck at the laboratory bench (26), with fundamental questions about immune correlates of protection required for universal protection against influenza viruses yet to be answered.

# STRATEGIES TO INCREASE THE STRENGTH, DURATION, AND BREADTH OF VACCINE-INDUCED IMMUNE RESPONSES

# Timing of Priming for T Cell Immunity

Existing inactivated virus-based vaccine technologies could be improved to increase the strength and duration of the vaccineinduced responses to overcome seasonal influenza epidemics of drifted variants. While natural infection may generate protective immunity for 2–10 years (2, 27), it has been reported that IIV sero-protection fell below 60% one year after vaccination (28). Therefore, bridging the gap between protection afforded by infection and vaccination, by defining immune correlates of protection (**Table 1**) associated with better outcomes of natural infection, severity of infection, and the protection afforded by current and other next-generation vaccines in clinical development (**Table 2**) is critical to future vaccine design.

Live-attenuated influenza vaccines have been shown to induce T cell responses in children but not adults (58), and in particular children under 10 years of age have greater T cell boosting (59). In mice, early-priming preserves optimal influenza-specific CD8<sup>+</sup> T cell function and diversity and protects against age-related immune decline (60), with similar age-associated effects of VE observed for LAIV (20). Furthermore, the thymus involutes during puberty greatly reducing naïve T cell output, while "inflammaging" impacts T cell priming (61). In another study, memory CD8<sup>+</sup> T cells have been observed and stable longitudinally for over more than a 10-year period, most likely due to multiple reinfections since childhood (27). It is likely that repeated boosting of T cell immunity to conserved, immunodominant epitopes of NP and M1 antigens (62) results in long-term maintenance of T cell responses associated with protection from symptomatic infection (2). In older adults, individuals who received 3–4 years of annual repeated IIV vaccination, rather than single vaccination, had higher response magnitude, long-term durability, and multifunctional quality cross-reactive memory CD4<sup>+</sup> T cells (63). Indeed, the T cell-based modified vaccinia virus Ankara (MVA) vector expressing NP + M1 influenza vaccine (MVA-NP + M1) could boost antigen-specific T cell memory responses in adults Table 1 | Broadly reactive correlates of protection from symptomatic influenza virus infection from human studies.


(*Continued*)

#### TABLE 1 | Continued


Table 2 | Clinical trial phase, size, scale, and influenza vaccines in development.


*a Included studies on 'Influenza' or 'Influenza vaccine' listed in ClinicalTrials.gov (56). bExtracted from WHO Tables on clinical evaluation of influenza vaccines (57).*

over 65 years of age (64). The promising MVA-NP + M1, which is currently in phase II clinical trials (**Table 2**), has been proposed to be used in conjunction with current IIV (65) and shown to broaden both humoral and cellular immunity. Therefore, a window for priming optimal T cell immunity with longevity exists and could be considered for vaccine design to maintain effective T cell immunity.

In humans, the lungs are enriched with CD8<sup>+</sup> T resident memory (TRM) that rapidly generate effector cytokines upon influenza infection (66). Furthermore, prime-pull strategies have been tested in mouse studies to seed local TRM responses (67), whereby a vaccine is given first parentally, i.e., intramuscularly like traditional vaccination route, to prime the T cell responses; and then inflammatory or secondary vaccine is given locally, i.e., intranasally, to pull responses to the lung, but this has had limited effect in the lung where cognate antigen presentation is required to maintain TRM (68).

However, the protective efficacy of localized TRM responses vs. circulating responses can only be tested in animal models. A 7-month limitation of protection has been identified by lung-resident TRM in mouse models (69), but may not reflect the decay of human peripheral T cell memory in labeling and tracking studies (70). Whether lung TRM have a similar critical role in modulating disease outcome in humans is unknown, but it is essential to be understood for optimal vaccine design.

# Stalling of the HA-Stalk

Development of next-generation vaccines that provide broader immune responses will be needed to protect against influenza pandemics and zoonotic influenza infections. A consensus on immune arms which are capable of providing broader immunity are split over a dichotomy, which are focused on either the anti-HA-stalk antibodies or T cell immunity. Theoretically, a HA-stalk vaccine has an exciting and promising potential, with subtype specific, multigroup, and even pan-influenza A and B antibodies being identified (71). Impressive *in vitro* and animal studies have shown the breadth of HA-stalk antibodies, yet passive transfer in human clinical trials have shown high concentrations are needed but with little therapeutic effect (72). While HA-stalk antibodies have been found to be enriched in some individuals infected with the 2009 H1N1 pandemic virus, these antibodies are notoriously low in frequency with universal antibodies such as F10 representing only 0.001% of circulating antibodies (73). There is a paucity of data on the protective role of HA-stalk antibodies in human infection studies (46) (**Table 1**). A human challenge study found that higher baseline level hemagglutination inhibition (HAI) antibodies were accompanied by increased HA-stalk-specific antibodies and reduced viral shedding but not symptom severity, while anti-neuraminidase (NAI) antibodies were the strongest correlate of protection (CoP) for symptomatic infection (46). Therefore, the independent role of HA-stalk antibodies remains to be defined separately from HAI and NAI antibodies. Ultimately, harnessing HA-stalk-specific B cells capable of universal immunity may also require repeated boosting to overcome immune waning, which limits the duration of current IIV.

# Alternative Strategies in Development

There are a large number of universal vaccine strategies in development in animal models, and only 61 in phase I and 189 in phase II clinical trials (**Table 2**) which are designed for pandemic potential viruses (56), and in total only 12.8% of these are designed to be effectively T cell stimulating (74). Apart from HA, additional viral proteins including the NP, NA, M1, and M2, are proposed to be possible targets for universal vaccines (75). Cross-reactive antibodies against these viral proteins from different subtypes have been identified and they are shown to have heterosubtypic protective effects in animals and humans. Various strategies using recombinant proteins/peptides, recombinant DNA, recombinant RNA, virus-like particles, viral vectors, and synthetic viruses for inducing heterosubtypic protective effects have been reported. Some of these approaches do not only aim at inducing broadly reactive antibodies but also cross-reactive T cell immunity against influenza infections. Clinical trials of experimental vaccines, such as proteasomal adjuvanted IIV by nasal delivery and MVA-NP + M1 have been assessed by experimental challenge and immune correlates of protection evaluated (**Table 1**). Previous reports also show many experimental vaccines are not undergoing clinical trials or approved for human use, suggesting a bottleneck to preclinical development (76), which could be attributed to limitations of some animal models to show vaccine efficacy or support needed from industry funding for increasing scale of clinical studies (**Table 2**).

# HURDLES IN EXTENDING EXPERIMENTAL FINDINGS TO COMMUNITY: IDENTIFICATION OF IMMUNE CORRELATES OF PROTECTION

# Correlating Immune Responses to Infection and Illness Severity

Hemagglutination inhibition and single radial hemolysis assays are the only accepted serological methods used both in the US and Europe for accelerated licensure of seasonal IIV and only recognized immune CoPs for influenza currently (77–79). Other candidates of CoPs (**Table 1**) have been evaluated against experimental or natural human influenza virus infections and vaccine efficacy studies. The route of vaccination (intramuscular or intranasal) determines systemic vs. local immunity, and variability in sampling techniques at the mucosa may hinder the precise evaluation of mucosal antibody responses (53). Furthermore, different CoPs may be identified depending on the outcome measure that is used across the spectrum of severity, for example, from asymptomatic infection to severe illness leading to hospitalization. Therefore, the context under which each CoP was determined should be considered, and a comprehensive analytical approach is needed for clinical studies (80).

Cellular immunity is important for protection from clinical disease. For example, the Flu Watch study highlighted NP-specific CD4<sup>+</sup> and CD8<sup>+</sup> T cells correlated with lower nasal viral shedding (2). Other studies identified dysregulation of cytokines (namely, IL-10, MCP3, and IL-6) (35) and reduced cellular responses (including T, NK, and MAIT cells) (39, 40, 42) are associated with severe disease. The baseline presence and increasing titer of secreted IgA (33, 47) and NAI (30, 32, 45, 46) have also been identified across studies and appear as more effective correlates of protection from symptomatic infection than HAI. In addition, reports are emerging that HA-stalk (46) and antibody dependent cellular cytotoxicity (ADCC)-activating antibodies (51) have been associated with reduced viral titers upon infection. de Vries et al. showed HA-stalk-specific ADCC responses were boosted in children post-infection (81), while H7-cross-reactive ADCC antibodies were cumulative and detectable from 2 years of age but plateauing by 17 years of age (82). Early exposures to influenza boost ADCC antibodies, while older adults have limited rises in ADCC antibodies post-infection (83). A titer of HA-specific ADCC antibodies >320 correlated with reduced risk of infection, symptom scores, and viral shedding in a human challenge study (51).

While a HAI titer of 40 is believed to provide 50% protection from symptomatic infection (84), new thresholds are being defined for T cell immunity. IIV does not effectively boost T cell immunity, hence the need for new universal vaccines. The longevity derived from memory generated by natural infection is also limited, with repeated infection during our lifetime, estimated every 2–10 years (2, 27). Therefore, universal vaccines will need to do better than nature to provide longer duration immunity from symptomatic infection. From community cohort studies with baseline samples prior to symptomatic infection, Hayward et al. (2) defined the protective threshold for symptomatic infection as >20 SFU/106 PBMCs (by stimulation with overlapping peptides for NP/M1); and from a LAIV children cohort study Forrest et al. (20) defined protective T cell threshold for symptomatic influenza infection as evaluated by ELISPOT was >100 SFU/106 PBMCs. Inactivated vaccination from the study by Koutsakos et al. resulted in a modest boost of influenza-specific CD4<sup>+</sup> T cells, while CD8<sup>+</sup> T cells were not boosted (85). While CD4+ T follicular helper cells (Tfh), correlate with greater antibody production and HAI titers (86), and are therefore important for current IIV efficacy. Future universal vaccines need to overcome limited immunogenicity of inactivated and LAIV vaccines by more immunogenic vaccine vectors (74). While universal vaccines, such as MVA-NP + M1, which uses systemic vaccination of a one-step replication vector encoding conserved NP and M1 proteins, boosted influenza-specific CD8<sup>+</sup> T cells in adults over >65 years of age (64), a notoriously difficult population for increased cellular immunity. Therefore, universal vaccines in development already show improved ability to establish T cell memory.

The immune correlates of protection from influenza are mostly derived from the comparison of infected subjects on a spectrum of severity (**Table 1**). However, there is a difference between correlates of protection against all infection vs. correlates of protection against symptomatic infection. Furthermore, studies of current IIV for boosting of T cell responses as correlates of protection are not ideal as IIV is not designed to stimulate cellular immunity and can impinge the cellular immunity that is being developed during natural infection (87). Rather, studies of uninfected but exposed and asymptomatic cases (low or no viral shedding) from naturally acquired infection could define immune correlates on a larger scale than possible with human challenge studies (**Table 1**) (2, 29, 30).

# Limitations by Prior Immunity

Prior immunity may impact vaccine efficacy, original antigenic sin, and similarly "HA-imprinting" may skew antibody and CD4+ T cell helper profiles by the viral subtype in the first exposure (88, 89). Furthermore, the level of neutralizing antibodies in a population will affect influenza transmission, and Bolton et al. proposed that T cell activating vaccines will have different efficacy depending on the population's prior immunity (90). For example, due to prior immunity to seasonal H3N2 viruses but not to avian H7N9 viruses, a T cell-activating vaccine would be more efficacious for H7N9 viruses. Vaccinating an immune population with biased prior immunity may reduce vaccine efficacy, and universal vaccine strategies may differ by age group due to HA imprinting and immunosenescence. Therefore, use of a universal vaccine in younger demographics could exploit immunological imprinting to their advantage. Previously, highantibody titers generated from childhood influenza infections which were maintained have been seen to be cross-reactive to antigenically drifted strains (91, 92).

On the other hand, seasonal influenza vaccination history may not always play a positive role in heterologous protection against subsequent influenza infection. Bodewes et al. (87) have compared the influenza A virus-specific cellular and humoral responses between 14 annually immunized children with cystic fibrosis and 27 unvaccinated healthy control children during winter season 2009–2010. A similar level of influenza-specific CD4<sup>+</sup> T cell responses and neutralizing antibody titers were found between vaccinated and unvaccinated groups of children, but an age-dependent increase in the frequency of virus-specific CD8<sup>+</sup> T cells were only observed in unvaccinated children. These findings indicated repeated annual influenza vaccinations might hamper the development of influenza A virus-specific CD8<sup>+</sup> T cell immunity. One report in mice recently from Rowell et al. (93) also addressed such issue, which presented varied heterologous protection from a candidate universal influenza vaccine (A/NP + M2-rAd) following a history of conventional IIV vaccination. Interestingly, they found that humoral and cellular responses induced by universal vaccine could be enhanced, inhibited, or unaffected by selected prior vaccinations, and such variations may be affected by many factors including vaccine preparation and specific vaccine components.

# Standardization of Assays and Findings Across Studies

Community cohort studies to identify natural influenza virus infections and measure immunity are established in the UK (2), US (94), Vietnam (95), Hong Kong (96, 97), China (98), and Nicaragua (99) with recruitment and experiments ongoing, making this area of research an exciting area to watch. The seasonality of influenza and year-to-year variation in infectivity of viruses requires these studies to span multiple years to generate robust data, for example, the Flu Watch study spanned 2006–2011 to capture 205 infections from baseline responses (2). Peripheral blood sampling will continue to be a proxy for cellular immune correlates for influenza virus infection, and simplified and standardized assays for immune signatures or biomarkers may aid future vaccine trials (**Table 2**).

One of the challenges in conducting studies to identify new correlates of protection is the sample size required. Typical community-based studies can follow up more than a thousand people over multiple years (2, 30, 97), measuring immune status before the season as baseline immunity and then identifying infections after influenza activity. Dunning et al. commented that data from 1,000 to 2,000 persons may be needed for a reasonably precise estimate of an influenza CoP (49). However, such sample size is logistically challenging, and the size scale of existing studies ranged from 16 to 226 infected individuals to stratify cases by severity to derive immune correlates (**Table 1**), and community cohort studies such as those by Sridhar et al. (25 cases from 342 participants) (29), Hayward et al. (205 cases from 1,414 participants) (2), and Couch et al. (226 cases from 1,509 participants) (30). The scale of vaccine efficacy trials precludes many vaccine studies, especially considering the need to show an improved standard of care from current IIV, which can be reasonable when well-matched viruses are in circulation but are limited for novel viruses.

The French Interior Milieu project (100) has provided baseline immune responses of 1,000 individuals over 2 time-points, sampled the individual's genetic background, skin biopsy, nasal swab, urine, and fecal samples, and uses 10 unique panels by flow cytometry, and 40 stimuli for characterizing adaptive and innate cellular responses. The panels measure in parallel innate cells and adaptive cells, including innate lymphoid cells, NK cells, mucosal associated invariant T cells, dendritic cells, neutrophils, B cells, and T cells (1, 2, 17, reg). Stimulation determines the individual's ability to respond to viral, microbial, agonists, and ligands, such as influenza and Sendai viruses, *Helicobacter pylori*, Poly I:C, Flagellin, TNFα, and CD3 + CD28 (101). The implementation of standardized assays, such as the stimulation of PBMCs with influenza viruses directly at blood collection by TruCulture tubes, was essential for multicenter experimental success (102). However, due to the use of an archetypal and outdated laboratory strain, A/Puerto Rico 8/1934, the results in regard to determining relevant baseline influenza virus-specific immunity were obsolete. The study protocols from the Interior Milieu project are now being extrapolated to other ethnicities and countries to provide a spectrum of a "healthy" baseline immune system and may provide a model for assays on a larger scale beyond HAI needed for universal vaccine design. The feasibility and scale of larger community cohort studies is beyond the capacity of a single research group for processing, storage, and experimental measures (80, 103) and needs commercial partners. Therefore, consensus and synergy with other established cohorts and network design to share expertise is essential to get past the bench to define quantifiable thresholds of immune correlates of protection.

# MAXIMIZING THE USE AND EFFECTIVENESS OF INFLUENZA VACCINES IN THE COMMUNITY

Once a vaccine has been licensed to be truly effective within a population a certain coverage threshold must be reached. In 2009, the European Centre for Disease Prevention and Control set out to achieve 75% influenza vaccination coverage in the elderly and those suffering from chronic medical conditions by the winter season 2014/15. However, this target was only reached by one EU Member State in the 2013/14 season and during the 2014/15 influenza season no Member States were able to reach this coverage rate (104). Whereas the vaccination coverage rate for adults aged 18-64 years is even lower, reaching only 36.7% during 2013–2014 in the US (105). This demonstrates that current approach to vaccination, in the case of influenza, is insufficient and even with the development of novel vaccines, strategies for their implementation needs to be carefully considered.

# Considerations and Strategies to Increase Accessibility and Uptake

The live-attenuated influenza vaccine only represents 8% of the vaccine market share (106), and production in the US has been threatened by low VE in recent years. Other enhanced influenza vaccines, such as QIV Fluzone by intradermal vaccination (107), have also been threatened by a dwindling market share. Combination and heterologous approaches may complicate adherence to vaccine schedules. Various methods have been developed to stimulate HA-stalk antibodies, such as HA-headless or chimeric HA, and combination strategies of prime boost for four doses (108). However, anti-HA-stalk antibody-stimulating vaccine regimes by heterologous prime boost used in mouse studies to elicit HA-stalk antibody responses may not be feasible in practice in the community, with each regime requiring separate licensure and multiple doses reducing vaccine adherence. The human papilloma virus (HPV) and hepatitis B virus (HBV) vaccines both require a homologous 3-dose regime within 2 years for optimal sero-protection, and HBV also requires a 10-year dose booster. Adherence to HPV vaccine 3-dose schedule is only 28% (109), and similarly 29% for HBV vaccination (110). A comprehensive vaccination record system will be instrumental for orderly vaccination schedules.

An increasingly difficult barrier to successful vaccination strategies is "vaccine hesitancy." The WHO Strategic Advisory Group of Experts (SAGE) on Immunization has defined vaccine hesitancy as a "delay in acceptance or refusal of vaccination despite availability of vaccination services" (111). Vaccine hesitancy can develop into refusal and the encouragement of others to refuse vaccination, leading to unvaccinated clusters within a community and severe public health consequences. One of the more concerning effects of vaccine hesitancy is the effect on vaccination coverage in HCWs. Vaccination for HCWs is recommended in most countries but mandatory vaccination programs vary. A survey of HCWs in China found a coverage rate of only 9.5% in the seasons of 2009/2010 to 2011/2012 (112). However, vaccination in HCWs in the US has increased since the 2010/2011 season, reaching 64.8% in 2014/2015 (113), demonstrating this issue varies greatly by country due to policy decisions and cultural factors. If HCWs themselves are hesitant about current vaccines, novel vaccines that utilize "non-traditional" approaches for universal immunity may require extensive explanation and promotion to HCWs to encourage self-vaccination and increase vaccine recommendations to patients.

# Indirect Protection in the Community With Vaccine Uptake

Many studies have shown that increasing vaccination uptake in children and younger adults reduces influenza burden in older adults (114–116, 117). Older children and adolescents have been shown to be the key age groups affecting the initial spread of influenza infections within a community (118). Elderly individuals often come into contact with children and young adults in household and urban settings, public areas and transportation. One of the clearest examples of this was seen in Japan, when influenza vaccination of school children ceased in 1994, leading to an increase in elderly mortality rates (119). A study analyzing US vaccine data also found that in areas where there was ≥31% vaccine coverage in younger adults, the elderly had a 20.6% lower chance of being diagnosed with influenza than in areas with a ≤15% coverage rate (120). Vaccination of healthy

children could also form the basis of establishing early T cell memory and broader HA imprinting from an immunological perspective.

# CONCLUSION

Universal influenza vaccine research is a growing trend (**Figure 1**), with first reports in the 1970s of heterologous immunity in the absence of antibodies for recombinant vaccines being developed following the antigenic switch from H1N1 to H3N2 viruses (121). A large increase in the universal vaccine research field has been seen since 2003, coinciding with zoonotic infections from avian and equine sources and pandemic viruses becoming a real threat to public health. Therefore, the drive for increased breadth of coverage for influenza vaccine has been a long-term objective, and the recent NIAID push has been a "call to arms" to address this issue. An increasing number of immune biomarkers that are associated with protection against influenza virus infection and disease severity *in vitro* and *in vivo* have been identified, leading to vaccines designed to elicit these immune markers at different stages of clinical trials. Such a strategy assumes that these markers are correlates of protection in humans, but whether such assumptions hold is yet to be confirmed in large epidemiological studies. Ultimately, immune correlates should be compared in parallel and defined

# REFERENCES


within a weighted hierarchy to drive vaccine design which can stimulate multiple immune arms effectively. Alternatively, despite measurable influenza-specific T and B cell immunity, all healthy adults experience repeat infections in their lifetime. Additionally the WHO goals to promote longevity of responses may also require a vaccine that elicits a "better than nature" response. With increased attention and funding for this area, particularly from the National Institutes of Health, there is real hope for the successful development of universal influenza vaccines.

# AUTHOR CONTRIBUTIONS

SV, NL, MB, YLM, AL, LP and BC wrote and prepared the review. SV and NL prepared the figures and tables.

# FUNDING

This work was supported in part by the NIAID Centers of Excellence for Influenza Research and Surveillance (CEIRS Contract No. HHSN272201400006C), the Theme-based Research Scheme (TRS) from the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. T11-705/14N), and Health and Medical Research Fund (Project No. 14130672).

influenza vaccination coverage and the 2009 influenza pandemic have had very little impact on improving influenza control and pandemic preparedness. *Vaccine* (2017) 35(36):4681–6. doi:10.1016/j.vaccine.2017.07.053


A systematic review and meta-analysis. *Vaccine* (2017) 35(2):212–21. doi:10.1016/j.vaccine.2016.11.013


polyfunctional and maintain diverse TCR profiles. *J Clin Invest* (2018) 128(2): 721–33. doi:10.1172/JCI96957


responses to influenza vaccination. *Sci Transl Med* (2013) 5(176):176ra132. doi:10.1126/scitranslmed.3005191


**Conflict of Interest Statement:** BC has received research funding from Sanofi Pasteur, and honoraria from Sanofi Pasteur and Roche. The authors report no other potential conflicts of interest.

*Copyright © 2018 Valkenburg, Leung, Bull, Yan, Li, Poon and Cowling. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Treating Influenza Infection, From Now and Into the Future

### Sophia Davidson\*

*Inflammation Division, The Walter and Eliza Hall Institute of Medical Research, Parkville, VIC, Australia*

Influenza viruses (IVs) are a continual threat to global health. The high mutation rate of the IV genome makes this virus incredibly successful, genetic drift allows for annual epidemics which result in thousands of deaths and millions of hospitalizations. Moreover, the emergence of new strains through genetic shift (e.g., swine-origin influenza A) can cause devastating global outbreaks of infection. Neuraminidase inhibitors (NAIs) are currently used to treat IV infection and act directly on viral proteins to halt IV spread. However, effectivity is limited late in infection and drug resistance can develop. New therapies which target highly conserved features of IV such as antibodies to the stem region of hemagglutinin or the IV RNA polymerase inhibitor: Favipiravir are currently in clinical trials. Compared to NAIs, these treatments have a higher tolerance for resistance and a longer therapeutic window and therefore, may prove more effective. However, clinical and experimental evidence has demonstrated that it is not just viral spread, but also the host inflammatory response and damage to the lung epithelium which dictate the outcome of IV infection. Therapeutic regimens for IV infection should therefore also regulate the host inflammatory response and protect epithelial cells from unnecessary cell death. Anti-inflammatory drugs such as etanercept, statins or cyclooxygenase enzyme 2 inhibitors may temper IV induced inflammation, demonstrating the possibility of repurposing these drugs as single or adjunct therapies for IV infection. IV binds to sialic acid receptors on the host cell surface to initiate infection and productive IV replication is primarily restricted to airway epithelial cells. Accordingly, targeting therapies to the epithelium will directly inhibit IV spread while minimizing off target consequences, such as over activation of immune cells. The neuraminidase mimic Fludase cleaves sialic acid receptors from the epithelium to inhibit IV entry to cells. While type III interferons activate an antiviral gene program in epithelial cells with minimal perturbation to the IV specific immune response. This review discusses the above-mentioned candidate anti-IV therapeutics and others at the preclinical and clinical trial stage.

Keywords: influenza, therapeutics, treatment, antiviral, immunomodulation

# INTRODUCTION

Influenza viruses (IVs) are a continual and re-emerging threat to human health. Annual epidemics infect approximately 1 billion individuals, leading to three to five million cases of severe illness and up to half a million fatalities worldwide (1, 2). Influenza A Virus (IAV), Influenza B Virus (IBV) and Influenza C Virus (ICV) are all members of the Orthomyxoviridae family. IV genomes are segmented, which allows for reassortment within, but not between, family groups. Although IBV

#### Edited by:

*Alan Chen-Yu Hsu, University of Newcastle, Australia*

#### Reviewed by:

*Bi-Hung Peng, The University of Texas Medical Branch at Galveston, United States*

> \*Correspondence: *Sophia Davidson Davidson.s@wehi.edu.au*

#### Specialty section:

*This article was submitted to Viral Immunology, a section of the journal Frontiers in Immunology*

Received: *08 June 2018* Accepted: *07 August 2018* Published: *10 September 2018*

#### Citation:

*Davidson S (2018) Treating Influenza Infection, From Now and Into the Future. Front. Immunol. 9:1946. doi: 10.3389/fimmu.2018.01946*

**102**

and ICV do cause disease in humans (IBV being responsible for approximately 25% of seasonal influenza infections) IAV strains are responsible for the majority of human infections and are most likely to cause severe disease. IAV are further classified into subtypes based on the antigenic properties of two viral surface glycoproteins, hemagglutinin (HA) and neuraminidase (NA), to date 18 HA (H1–H18) and 10 NA (N1–N10) antigenic subtypes been identified (3, 4). Unlike IBV and ICV, IAV infects a broad range of species including humans, pigs, horses, wild mammals, and birds (5). Due to different preferences for sialic acid moieties direct zoonosis of IAV between birds and humans is rare, however when it does occur, the mortality rate is staggeringly high, approximately 60% for H5N1 and 30% for H7N9 (6). In worrying contrast, transmission of IAV strains from swine to humans is common (7).

In healthy humans, IV infection induces a robust immune memory response, in spite of this the average adult will experience two IV infections per decade throughout their lifetime (8). IVs are able to evade IV-specific host immunity through two mechanisms: antigenic drift and shift. Antigenic drift occurs as IV genomes do not have RNA proofreading enzymes and consequently, point mutations accumulate in the genome through successive replication. This leads to alterations in the appearance of viral antigens and eventual emergence of new IV strains which are unrecognizable to pre-existing host immunity (9). Significantly more dramatic and, within the Orthomyxoviridae family, believed to be specific to IAV is antigenic shift. Infection of a single host cell with two or more strains of IAV results in the reassortment of genomic segments. IAV genome segments are packaged into viral particles by the host cell without respect to the original strains, leading to progeny virions which possess new HA and/or HA and NA proteins, such as those of avian or swine origin, but may retain the ability to effectively infect humans. Antigenic shift gives IAV pandemic potential, indeed it is thought that the majority of pandemics of the Twenty-First century have been caused by reassortment events that resulted in avian or swine IAV being able to stably infect humans (10).

The severity of IV induced disease is a function of the interplay between viral virulence and the host immune response. In a mild infection the inflammatory response is controlled and cleared rapidly. However, in highly pathogenic IV infections the host immune response can become excessive. Termed the cytokine storm, severe IV infection in humans is characterized by aberrant cytokine and chemokine responses that associate with infiltration of inflammatory cells, particularly monocytes and neutrophils. This inflammation coincides with destruction of the epithelial layer and consequently, respiratory dysfunction or acute respiratory distress syndrome (ARDS) (11). Ex vivo analysis of clinical samples, experimental infection models and clinical trials all indicate that the cytokine storm positively correlates with tissue injury and severe IV induced disease (12–17).

To add to the multifarious nature of IV infection, it can be complicated by secondary bacterial infection. Bacteria which normally colonize the upper respiratory tract such as Streptococcus pneumoniae or Staphylococcus aureus can cause pneumonia and septicaemia in IV infection (18). It is thought that opportunistic bacteria take advantage of changes in the pulmonary environment wrought by IV infection. Many mechanisms have been proposed to explain this phenomenon, for example IV infection induces a robust type I interferon (IFNαβ) response, which blocks the recruitment of neutrophils, a cell type particularly important for clearance of bacterial infection (19). Furthermore, monocytes and monocyte-derived cells recruited to the lung during IV infection induce the apoptosis of airway epithelial cells via TNF-related apoptosisinducing ligand (TRAIL), this facilitates bacterial colonization and systemic dissemination by compromising epithelial layer ntegrity (20).

Undeniably, there is a real and present need for effective broad spectrum anti-IV therapies. Given the high mutagenicity of the IV genome vaccine development is fraught with difficulty, current IV vaccines are strain specific and therefore a new immunization is required for each new season (21). Moreover, the rapid emergence of the 2009 H1N1 pandemic strain demonstrated how under prepared we are for a serious IAV pandemic. This review reports current treatments for IV and discusses new therapies at clinical or pre-clinical stage. As IAV has pandemic potential and is most likely to cause severe disease in humans many of the treatments discussed are primarily directed at this virus, however they may be effective against other Orthomyxoviridae family members. For clarity, therapies are categorized based on point of action in IV infection, specifically, (1) IV: proteins and genomes, (2) Host immune response: cytokines/chemokines and other inflammatory modulators, and (3) Target cells for IV replication: respiratory epithelium.

# DIRECT TARGETING OF IV

# Current Treatment

IV surface proteins HA and NA are responsible for virion attachment to and detachment from sialic acid moieties on the host cell surface. HA attaches to cell surface sialic acid receptors to initiate viral entry and promote fusion of viral and cellular membranes, while NA acts as a sialidase, cleaving the α-ketosidic bond linking a terminal neuraminic acid residue to the adjacent oligosaccharide moiety. This enzymatic action of NA releases IV particles from infected cells and thereby allows the spread of IV to naive cells (22). NA sialidase activity also facilitates the movement of IV through the sialic acid-rich mucous of the human respiratory tract (23). NA is essential for productive IV infection and the catalytic sites of NA are conserved across IAV and IBV strains, making this glycoprotein an attractive target for antiviral therapy (24). Accordingly, in the 1990s Neuraminidase inhibitors (NAIs) were developed. NAIs are sialic acid analogs which competitively bind to the active site on NA molecules to inhibit the release of IV progeny from the cell surface (25).

NAIs are the only antivirals currently recommended to treat IV infection, oseltamivir and zanamivir are used worldwide, laninamivir is approved in Japan and peramivir is approved in China, Japan, South Korea, and the United States (26). Oseltamivir (**Table 1**) is most commonly used and has been shown in vitro to have activity against human and avian IAV subtypes and IBV strains (27). NAIs have been employed

#### TABLE 1 | Summary of key treatments discussed.


*Potential therapeutics for human IV infection are summarized. Treatments are separated based on which aspect of IV infection is targeted. Viability of each therapeutic is rated based on data discussed in this review.*

successfully for over decade, however between 2007 and 2009 resistance to oseltamivir in seasonal IAV strains surged from less than 1% to over 90% (28–31). IV strains resistant to NAIs typically contain mutations in the NA which reduce the inhibitor binding ability by altering the shape of the NA catalytic site. Although several resistance conferring mutations have been reported, the most common for IAV is H274Y. In order for oseltamivir to bind correctly, NA must undergo rearrangements to form a binding pocket. Key to these rearrangements, is the amino acid E276 rotating and binding to R224 (32, 33). In vitro modeling and X ray crystallography revealed that H274Y inhibits this rotation of the E276 residue thereby preventing pocket formation (32, 34). Such a dramatic uptake of the H274Y mutation at the population level is unlikely to be driven by individual patient use, instead H274Y-mutant IAV strains may have acquired advantageous epidemiologic fitness, allowing for rapid global transmission (35, 36). Fortunately, the 2009 H1N1 IAV pandemic strain did not carry this mutation when it emerged, and as this is the current dominant seasonal strain, the frequency of NAI resistance in circulating IAV strains is now low. However, localized clusters of oseltamivir-resistant IAV have been detected (37), and mutations which confer decreased sensitivity to oseltamivir in IBV strains have also been reported (38). The rapid emergence of oseltamivir-resistance observed between 2008 and 2009 demonstrates that NAI-resistance can develop at no cost to viral fitness and these mutations can easily spread throughout the population.

Aside from concerns regarding resistance, the effectiveness of NAIs is limited when delivered over 48 h after symptom onset. Indeed, multiple systematic reviews have concluded that oseltamivir does not reduce IV related hospitalizations, and that there is little evidence of reduction in complications of IV infection (39–42). Although, another meta-analysis did demonstrate that oseltamivir was effective for prevention of influenza at the individual and household levels (43). Use of oseltamivir and other NAIs has demonstrated the need for development of anti-IV drugs that improve treatment effectiveness, particularly when delivered late in the progression of disease, and have a low propensity for driving the emergence of viral resistance.

# Potential IV Targeted Therapies

The IV surface protein HA binds to host cell receptors to initiate infection. This glycoprotein consists of a globular head and a stem region that are folded within six disulfide bonds, plus several N-glycans that produce a homotrimeric complex structure (44). The majority of IV neutralizing antibodies elicited by vaccination or infection bind to the globular head of HA and recognize homologous strains within a given subtype (45). Antibodies to the HA head neutralize virus infectivity by blocking sialic acid receptor binding either directly, by interacting with the receptor binding site at the tip of the molecule, or indirectly, by projecting over the binding site and rendering it inaccessible (46–48). However, N-linked glycosylation sites on the HA globular head are highly variable across different IV subtypes and some IAV strains can further avoid host antibody responses by acquiring additional N-glycan modifications in the HA head region (49, 50). In contrast, N-linked glycosylation sites in the HA stem region are relatively well conserved among IAV strains. Antibodies to IAV HA stem motifs occur naturally and have activity against a broad range of IAV subtypes, however they are immune-subdominant and are only induced in very low titres during natural infection. Mechanistically, anti-stem antibodies control IAV by inducing antibody-dependent cellular cytotoxicity of infected cells (51–53). Given their potential, several monoclonal antibodies targeting the highly conserved stem region of the HA molecule are being evaluated in clinical trials. In particular, MHAA4549A and MEDI8852 have demonstrated high-affinity binding to 16 IAV HA subtypes and VIS410 has confirmed binding to 7 (54–56) (**Table 1**). MHAA4549A, MEDI8852, and VIS410 were all shown to be effective in protecting IV infected hosts by inhibiting pulmonary viral load in preclinical animal models (55–59). VIS410 was found to be safe and well tolerated in a phase 1 study and is now under phase 2 investigation (60). MHAA4549A and MEDI8852 were both reported to control viral shedding in humans in phase 2a clinical trials (58, 59). Furthermore, MHAA4549A was reported to lower patient influenza symptom scores and significantly, levels of inflammatory cytokines in serum and nasopharyngeal samples compared to placebo controls (58).

In a clinical trial setting MHAA4549A and MEDI8852 both performed comparably to oseltamivir, yet neither antibody improved oseltamivir effectiveness when used in combination (clinical trials: NCT02293863 and NCT02603952), indicating that these antibodies do not offer better protection than NAIs. However, compared to oseltamivir, which must be given twice daily (61), HA stem antibodies have superior pharmacokinetics, the half-life of MHAA4549A is approximately 3 weeks in humans (58) and MHAA4549A, MEDI8852 and VIS410 all have demonstrated protection against IV induced disease with only one to two doses (55–59). Furthermore, both MHAA4549A and MEDI8852 have been shown to confer protection beyond 48 h post infection, a point at which oseltamivir has lost effectivity in small animal models (55, 56, 58, 59). Excellent pharmacokinetics and a longer therapeutic window make HA stem antibodies strong candidates for treatment of IV infection.

The IV RNA-dependent RNA-polymerase (RdRp), is responsible for transcription and replication of IV's genome and is highly conserved across different strains. It is a heterotrimeric protein containing three virally encoded subunits: PB1, PB2, and PA. PB1 has polymerase activity, PB2 is involved in cap-binding of host cell pre-mRNAs and PA cleaves capped host pre-mRNAs and initiates transcription (62). Cap-snatching by PB2 essential for RNA transcription, PB2 first binds to the 5′ -methyl cap of host pre-mRNA which is then cleaved by PA's endonuclease site to produce a capped primer for IV transcription initiation (62). JNJ63623872 (formerly known as VX-787) (**Table 1**) is a compound that binds to key residues in the PB2 cap binding domain preventing the docking of the natural ligand: 7-methyl GTP. Preclinical in vivo and in vitro studies have demonstrated that JNJ63623872 has varying degrees of activity against a range of IAV strains, however due to the differences in IAV and IBV PB2 protein JNJ63623872 is ineffective against IBV (63). When directly compared in a mouse model of IAV infection JNJ63623872 was more effective than oseltamivir in controlling IV induced disease severity (64). A placebo-controlled phase IIa study showed JNJ63623872 to be well tolerated and resulted in a 94% reduction in viral shedding and quicker resolution of flulike symptoms compared to controls (65). However, the dosing regime of JNJ63623872 is similar to oseltamivir and variant strains with reduced susceptibility to JNJ63623872 have been isolated from in vitro culture (66), indicating that this therapy in its current form may not supersede NAIs. JNJ63623872 is now in phase II trials alone (NCT02342249) and in combination with oseltamivir (NCT02532283). Interestingly, a phase I trial has been initiated to evaluate the safety and pharmacokinetic interaction of JNJ63623872 with AL-974, a PA inhibitor that is in early-stage development (NCT02888327).

Favipiravir (also known as T705) (**Table 1**) is a ribonucleotide analog (6-fluoro-3-hydroxy-2-pyrazinecarboxamide) that inhibits viral RdRps. However, the mechanism by which this inhibition occurs is not understood, indeed, even the viral proteins targeted by Favipiravir are not yet defined. In vitro studies have revealed that serial passage with increasing concentrations of Favipiravir drives guanosine to adenine nucleotide mutations in IV, essentially resulting in the production of non-viable IV particles (67). Several studies in mice have demonstrated Favipiravir administration up to 72 h post infection with seasonal IAV strains such as H1N1 and avian strains: H5N1 and H7N9 result in a dose-dependent reduction lung viral titres and host mortality (68–71). Favipiravir has been shown to be to have potent inhibitory activity against several RNA viruses in vitro and appears especially effective for IVs (72). This acute susceptibility of IV may be due to IV's lack of RNA proofreading enzymes. Furthermore, Favipiravir appears to have an exceptionally high barrier for drug resistance, currently only one mutation (V43I in PB1; obtained in virus-infected cell cultures under selection) was found to confer a slight increase in resistance to Favipiravir (73). Favipiravir is highly promising as a broad acting anti-IV therapy and as such, has been approved for select use in Japan and has completed phase III trials in the USA and Europe.

Along with proteins for replication, assembly and infection, IV genomes also code for a protein which can inhibit the host immune response: non-structural protein (NS1). NS1 is a highly conserved multifunctional protein which inhibits host antiviral responses, particularly, induction of types I and III IFNs (IFNαβ and IFNλ). NS1 antagonism of host immunity varies between IV strains; NS1 can prevent IV-mediated activation of key inflammatory transcription factors such as IFN Regulatory Factor 3 (IRF-3) and nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) (74–76). NS1 limits host recognition of IV through the pattern recognition receptor (PRR): retinoic acid inducible gene-I (RIG-I) by sequestering dsRNA (which is a RIG-I agonist) and inhibiting RIG-I ubiquitination and therefore activation (77–83). NS1 is key to viral fitness, strains deficient for NS1 induce markedly higher secretion of antiviral IFNs from cells in vitro and are non-pathogenic in mouse models of IV infection (84–87). Thus, the NS1 protein is a suitable target for anti-IV therapeutics. JJ3297 (**Table 1**) is a second-generation chemical inhibitor of NS1 function that has been shown in an in vitro assay to restore levels of IFNαβ-mRNA to those seen when cells were infected with a NS1 deleted mutant (88). While the exact mechanism of action is not understood, JJ3297 mediated inhibition of NS1 absolutely requires the function of cellular RNase L, indicating that an intact interferon system is essential for function of the compound (88). Further development of JJ3297 has resulted in the generation of another compound: A22 and NS1 inhibitors are now being investigated in in vivo models of infection (89). Additionally, SP600125, a C-Jun-N-terminal kinase inhibitor reduces the replication of IV in vitro and in vivo by indirect inhibition of NS1-mediated functions in the early stages of infection (90) and small molecules such as polyphenol and quinoxaline derivatives have also been proposed to inhibit NS1 (91). More study is required to determine if NS1 inhibitors are suitable for clinical use. However, given the direct correlation between host inflammatory response and IV-induced disease severity, use of NS1 inhibitors, particularly late in infection, should be cautiously evaluated.

# STEPPING INTO THE STORM

Limiting IV replication curbs disease severity not only by decreasing number of virions able to propagate the infection, but also by limiting immune stimulation. All cell types will secrete cytokines and chemokines to varying degrees upon recognition of IV pathogen associated molecular patterns. Cytokines and chemokines drive the recruitment and activation of both innate and adaptive immune cells which, while vital for resolution of infection, can also exacerbate disease through tissue damage. Therefore, at later time points in infection when viral load is already limited, it is more important to control the inflammatory response. Use of interventions which target the host response is an excellent strategy to combat severe IV infection. Host directed therapeutics are unlikely to drive the emergence of resistant strains and their effectivity is not strain specific. However, which immune drives are the most appropriate to target remains an open question. Severe IV infection induces many cytokines; IFNαβ, TNFα, IFNγ, C-X-C motif chemokine (CXCL) 10 (CXCL10), CXCL9, C-C motif ligand (CCL) 2 (CCL2), CCL4, CCL5 and interleukin (IL)−6 (IL-6), IL-2, IL-8, and IL-10 have all be observed to be upregulated during severe IV infection in humans (14, 15, 17, 92, 93). Yet studies in animal models demonstrate that there is yet to be a setting where complete absence of a specific cytokine or its cognate receptor entirely ablates IV induced cytokine storm. As TNFα and IFNαβ correlate well with disease severity in both clinical and experimental IV infection and are potent immunomodulators, known to be upstream of proinflammatory cytokine and chemokine secretion from many cell types, multiple studies have proposed treatment with these cytokines to promote viral clearance, or blockade of these cytokines to minimize host mediated tissue damage (12, 15, 94– 101).

TNFα drives the activation of multiple intracellular signaling pathways through the activation of NF-κB (102). In response to IV infection TNFα promotes the secretion of the antiviral cytokine families: type I, II, and III IFNs through upregulating RIG-I and toll-like receptor 3, Myeloid differentiation primary response 88 (MyD88), TIR-domain-containing adapter-inducing interferon-β (TRIF), and IRF7 genes. TNFα drives IV clearance via induction of apoptosis, stimulation of reactive oxygen species and activation of Nicotinamide Adenine Dinucleotide Phosphate Hydrogen (NADPH) oxidases in neutrophils and macrophages, such as NADPH oxidase 2 (NOX2), resulting in the generation of superoxide (103). Yet, TNFα is dispensable for control and clearance of IV, TNF deficient mice exhibited comparable mortality to controls upon H5N1 infection (104). Anti-TNF therapy in a murine H1N1 infection model reduced pulmonary recruitment of inflammatory cells, cytokine production by T cells and the severity of IV induced disease without preventing virus clearance (96). Similarly, treatment of mice lethally infected with H1N1 IAV with etanercept (**Table 1**), a soluble TNF receptor decoy, significantly reduced inflammatory cell infiltration, production of inflammatory cytokines and downregulated NFκB signaling, yet enhanced host control of virus replication, resulting in a 30% increase in host survival (105).

Interestingly, etanercept is used to treat a range of inflammatory conditions such as Rheumatoid Arthritis (RA). While patients with RA do exhibit an increased risk of IAV infection, treatment with etanercept does not contribute to this. In a retrospective cohort study Blumentals et al. found that etanercept or use of other biologics did not significantly affect the rate of influenza infection or its complications in RA patients (106). Yet whether or not etanercept lowered IV induced disease burden in treated patients compared to controls could not be assessed, as this data was not consistently recorded. Conversely, there is also evidence that TNFα is required for controlling the extent of IV induced immunopathology and tissue injury. In a mouse model of H1N1 infection Damjanovic et al. found that TNF-/- mice exhibited prolonged expression of inflammatory chemokines such as CCL2 leading to an exaggerated immune response and consequent damage to pulmonary epithelial cells (107). Further investigation by DeBerge et al. revealed that it is soluble, and not membrane bound, TNFα that is required to limit the IV induced immune response and tissue damage (108). Therefore, it is unclear if TNFα blockade is a suitable treatment for severe IV induced disease, however given the multiple components of the TNFα signaling system, TNFR1 vs. TNFR2 and the differing activities of membrane bound and soluble TNFα, there is the possibility to specifically inhibit certain aspects of TNFα signaling while not interfering with others.

IFNαβ are the canonical antiviral cytokine family in fact, they were discovered in the context of IV. IFNαβ induces the expression of hundreds of genes, such as MX dynamin like GTPase 1 (Mx1) and interferon induced transmembrane protein 3 (IFITM3) which have direct anti-IV activity. As such, IFNαβ has been periodically suggested as a therapy for IV (94, 97, 100, 101). Prophylactic or very early on treatment with IFNαβ in rhesus macaques, ferrets, guinea pigs and mice experimentally infected with IAV controls virus replication and spread thereby protecting against severe IV induced disease (101, 109–114). However, it appears the therapeutic window is short, later treatment with IFNαβ during infection still controls viral load but exacerbates disease by driving the cytokine storm and TRAIL mediated airway epithelial cell death (101, 109, 115). While there have been no studies directly assessing the effectiveness of IFNαβ blockade during IV infection in humans, IFNαβR deficient mice exhibit a range of susceptibility to IV induced disease depending on the virulence of the infecting IV strain and the genetic background of the mice (86, 115–118), demonstrating that the activity of IFNαβ on host immune response to IV is too complex to extract the immunopathogenic from the protective effects on the host.

Due to the pleiotropic actions of TNFα and IFNαβ direct targeting of these cytokines may not be the most suitable approach. Instead, a general dampening on the immune response may be more effective. Recently, chemical agonism of the sphingosine-1-phosphate (S1P) receptor (S1PR) pathway has been shown to blunt IV induced inflammation. The sphingosine analog: AAL-R (**Table 1**) agonizes S1P receptors 1, 3, 4, and 5. Treatment of IV infected mice with AAL-R during infection resulted in reduced release of proinflammatory cytokines and chemokines including IFNαβ and inhibited inflammatory cell infiltration and thereby decreased damage to pulmonary tissue. AAL-R treatment did not affect antibody responses and pulmonary viral load was comparable between treatment and control groups, however AAL-R did suppress dendritic cell maturation and inhibited IV specific T-cell responses (119, 120). Although the IV T cell response is dispensable for clearance of IV, it provides the host with herterosubtypic immunity, thus AAL-R is too immunosuppressive to be applied as an anti-IV therapy. But based on the promise of AAL-R, two agonists specific S1P1R: CYM-5442 and RP-002 (**Table 1**) were tested. Like AAL-R, CYM-5442 and RP-002 significantly reduced cytokine and chemokine responses associated with IV induced lung injury without effecting viral load. Yet, unlike AAL-R, neither CYM-5442 and RP-002 effected dendritic cell and Tcell responses (120, 121). Teijaro et al. proposed that agonism of S1PRs on endothelial cells was responsible for the blunted proinflammatory cytokine levels in the lung (121, 122). However, in follow up studies this group also found that S1P1R agonists act directly on plasmacytoid dendritic cells to block their secretion of IFNα (123, 124). Furthermore, these results defined signaling downstream of MyD88 in multiple cell types to be a key amplifier of IAV induced cytokine storm which could be inhibited by S1P1R agonism. Further characterization of S1PR agonists as IV-therapeutics is ongoing in mouse and ferret models (123).

Cyclooxygenase enzymes (COX) catalyze the conversion of arachidonic acid to prostaglandins, which can modulate the inflammatory response (125). Interestingly, there are two isoforms of COX: the constitutively expressed COX-1 and the inducible COX-2 which have divergent roles in influenza infection. Carey et al. demonstrated that in H3N2 IAV infection COX-2 deficient mice, compared to wild type controls, had lower levels of proinflammatory cytokines (IL-6, TNFα, IL-1β, and IFNγ) and inflammatory cells recruited to the lung during infection, and this correlated to a moderate increase in survival. While in contrast, COX-1 deficient mice in the same study exhibited a higher pulmonary inflammatory burden compared to wild type controls. The cost of this blunted inflammation in COX-2-/- mice was a higher viral burden early in infection, however by day six all three mouse strains had comparably low pulmonary titres of H3N2 IAV (126). In another study, COX-2 deficiency correlated to higher levels of the prostaglandin: PGE<sup>2</sup> which has an inhibitory effect on proinflammatory cytokine expression, the adaptive immune response and macrophage apoptosis in mice infected with H1N1 (127). Furthermore, COX-2 expression is elevated in autopsy tissue samples from patients infected by H5N1 IAV and induction of proinflammatory cytokines such as IL-6, TNFα, IFNα, and IFNβ by H5N1 in monocyte derived macrophages could be blocked by a COX-2 inhibitor (nimesulide) (128). Thus, there is strong evidence that COX-2 is an upstream driver of IV induced inflammation, however, the specific mechanism of action remains to be determined. In a follow up study, Carey et al. found that treatment of wild type mice with COX-1 inhibitor (SC-560) or a COX-2 inhibitor [celecoxib (**Table 1**)] prior to and during IAV infection resulted in the same pattern of susceptibility (COX-2 inhibition being protective and COX-1 inhibition being detrimental) yet, neither treatment drastically altered pulmonary cytokine profiles, viral load or inflammatory cell recruitment (129). Furthermore, another in vivo study found that celecoxib alone did not protect H5N1 infected mice from mortality, although the authors did observe a protective effect of celecoxib administration when used in combination with zanamvir and mesalazine (a PPARγ agonist, see below) in mice challenged with H5N1 IAV. Significantly, combination treatment was administered post IAV infection. This protection did correlate to a moderate decrease in proinflammatory cytokine concentrations and a modest elevation PGE<sup>2</sup> in the lung late in infection however, it also correlated to decreased viral loads at this time point which may explain the change in pulmonary cytokine profile (130).

Currently, a phase III clinical trial is running to assess efficacy and safety of celecoxib used in combination with oseltamivir in patients with severe IAV infection (NCT02108366). While this is an exciting development for the use of immunomodulating drugs in the treatment of IV, in high concentrations celecoxib can also inhibit COX-1 (131), which may prove problematic. As demonstrated by Carey et al. COX-1 plays an anti-inflammatory and protective role in IV infection (126). Moreover, treatment with nonselective COX inhibitors such as aspirin and diclofenac confer an increased risk of mortality in animal models of infection and it has been proposed that an increase in aspirin use during the 1918 pandemic contributed to the October death spike (132, 133).

In 2006 Fedson proposed the use of statins to modulate IV induced cytokine storm (134). Statins (**Table 1**) block cholesterol synthesis by competitively inhibiting the enzyme 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase (135). Commonly employed to reduce the risk of cardiovascular disease by lowering cholesterol levels, statins are inexpensive and widely available, therefore making them an attractive candidate for IV treatment. Statins can inhibit IV induced disease through multiple mechanisms, in vitro studies have shown that statins can interfere with viral replication (136, 137), block the induction of proinflammatory cytokines and chemokine such as IL-6 and TNFα and inhibit the activation of key signaling molecules including Signal transducer and activator of transcription 3 (STAT3) (138, 139). Animal studies have shown promise, Haidari et al. demonstrated statin treatment lowered pulmonary viral load and host mortality in murine H3N2 and H1N1 IAV infection models and An et al. demonstrated combination treatment with a statin, a NAI and a fibrate, protected mice from H5N1 mediated mortality (136, 140). In an intriguing study Liu et al. combined statins with another readily available drug: caffeine, and found that combination therapy lowered pulmonary viral load and ameliorated lung damage in H5N1-, H3N2-, and H1N1-infected mice (141). However, other studies conducted in mice have reported little to no effect of statins on IV clearance or cytokine profile (142, 143).

As statins are so widely used in the human population, there is a substantial amount of data on their use in the context of IV infection. Five retrospective studies conducted in four separate countries (Netherlands, United Kingdom, USA and Mexico) reported that to varying degrees, statin treatment associated with reduced IV-related pneumonia and a lower IV induced mortality rate (144–147). In contrast, Fleming et al. and Kwong et al. conducted retrospective studies over a 6 and 10 year periods (respectively) and found no association between statin treatment and decrease IV induced disease burden (148, 149). There are many caveats to these studies, including what other treatments patients were on during the study period and a lack of defined IV specific outcomes. Furthermore, the use of different statins and strains of infecting IVs likely contributes to the varied results. Overall, there is evidence that statins can ameliorate severe IV induced disease, and the availability of this class of drugs certainly makes it an attractive therapeutic option. Further study is required to delineate the specific actions of statins which block viral replication and inhibit over activation of the innate immune response, thereby allowing us to capitalize on these properties. Excitingly, a phase II trial has begun to test the effectivity of atorvastatin in minimizing IV induced disease severity in patients infected with seasonal IV (NCT02056340).

Peroxisome proliferator-activated receptors (PPARs) are nuclear receptors and ligand-activated transcription factors that control a number of target genes upon assembly of a transcriptional complex. PPARs regulate energy balance, including glucose homeostasis, fatty acid oxidation, and lipid metabolism (150). PPAR agonists are commonly used to treat patients with cardiovascular diseases and diabetes mellitus. Drugs which specifically antagonize PPARγ appear to be the most promising as therapeutics for IV. Treatment of mice, prior to and during IAV infection, with PPARγ agonist: pioglitazone (**Table 1**) was shown to temper recruitment of Ly6Chigh myeloid cells termed: TNF-α/inducible nitric oxide synthase (iNOS) producing DCs (tipDCs), although likely these are comparable to what other studies have reported as inflammatory monocytes or exudate macrophages (20, 115, 151). Pioglitazone lowered pulmonary concentrations of chemokines known to attract tipDCs to the lung (CCL2 and CCL7) and this associated with a decrease in IAV induced morbidity and mortality. Importantly, pioglitazone treatment did not alter the rate of IAV clearance from the lung, as was observed when tipDC recruitment was entirely ablated through the genetic deletion of CCR2 (152). In a follow up study, this group also demonstrated that rosiglitazone (another PPARγ agonist) mediated better protection than pioglitazone (or vehicle control) in mice infected with H1N1 IAV (153). Finally, treatment of mice with 15 deoxy-112,14-prostaglandin J2 (15d-PGJ2), 1 day post infection blunted IV induced proinflammatory cytokine secretion in the lung and increased host survival in a PPARγ dependent manner (154). As with statins, PPARγ agonists could be easily employed an adjunct therapy for IV induced disease, however human studies must be performed. Indeed, there is a somewhat surprisingly little amount of data about immunomodulating agents and IV infections. Although imperfect, retrospective studies on patients treated with immunomodulating agents such as IFNαβ for multiple sclerosis or hepatitis C, or any number of anti-inflammatory agents for heart disease may provide informative preliminary data in terms of effectivity and safety.

# TARGETING THE EPITHELIUM

In general, productive IV replication is restricted to airway epithelial cells, as these cells exclusively express proteases required for HA maturation (155). Damage to the respiratory tract in the form of virally induced necrosis, immune mediated apoptosis or other forms of cell death leads to ARDS. Finding a way to directly target the cells which support IV replication is highly desirable in anti IV treatment design. As such, many of the treatments discussed in this review are delivered via inhalation. However, by focusing on features relatively specific to the epithelial cells therapies can directly protect the epithelium during infection or promote healing post viral clearance. For example, Fludase (**Table 1**), is a recombinant fusion protein consisting of a sialidase catalytic domain derived from Actinomyces viscosus fused with the epithelial anchoring domain of human amphiregulin. Fludase is effectively a neuraminidase mimic, it tethers to, and cleaves both α(2,6)-linked and α(2,3) linked sialic acid receptors, thereby removing IV's entry point into epithelial cells (156). This drug is administered as an inhaled dry powder with microparticles of 5–10µm in size, enabling the drug to access the upper and central, but not the lower respiratory tract. In vitro studies on human airway epithelial cells have shown that Fludase removed approximately 90% of sialic acid receptors within 15 min of treatment and desialylation lasted at least 2 days (157). Serial passage of IAV and IBV under increasing selective pressure of Fludase selected for several mutations in HA (G137R, S136T, S186I) and NA (W438L, L38P) which resulted in IVs with increased receptor binding, coupled with significantly reduced NA on the cell surface. These mutations lead to an attenuated phenotype in vitro and no change in virulence in a mouse model of IV infection. Furthermore, the resistance phenotype was unstable and was reversed after withdrawal of Fludase (158).

As it targets the common entry point of IVs Fludase has been shown to be effective at inhibiting a broad range of IAV and IBV strains in vitro (159–161). Prophylactic treatment of mice with Fludase inhibited establishment of infection by IAV strains H1N1, H5N1, H7N9 and therefore protected against host mortality. Furthermore, these studies reported that Fludase inhibited IV replication and therefore host mortality when given up to 3 days post infection, albeit with less effectivity than prophylactic treatment (156, 162, 163). Malakhov et al. also demonstrated effectivity of Fludase in a ferret model of H1N1 infection (156). Fludase has begun clinical trials and was generally well tolerated in phase I trial (164). A phase II trial performed over three influenza seasons (2009–2011) in otherwise healthy IV-infected participants demonstrated that Fludase was well tolerated and patients under a multi-dose treatment regime exhibited a significant decrease viral load and viral shedding (165).

While Fludase is a promising anti-IV therapy there are potential pitfalls to broad use. Sialic acid is catabolized by S. pneumonia, IV-mediated release of this metabolite is thought to facilitate bacterial colonization and consequent pneumonia (166). In a preclinical study Hedlund et al. demonstrated that Fludase treatment did not alter S. pneumonia colonization in an in vitro model of a human lung cell line (A549) or in healthy mice. This study also reported that Fludase treatment 24 h post infection with H1N1 or H3N2 strains of IAV protected mice from S. pneumonia colonization and therefore morbidity and mortality (167). However, it is important to note that Hedlund et al. administered the secondary bacterial infection 2 days after a single dose of Fludase in IAV infected mice, which, given that airway epithelial cells begin to recover sialylation by 2 days post treatment (157) may be too late to see direct effects of Fludase treatment on bacterial colonization in the context of IV infection. Furthermore, the authors employed a lethal dose of IV, with all vehicle control mice exhibiting highly similar morbidity and mortality regardless of secondary S. pneumonia infection. It is therefore unclear whether or not the inoculum of S. pneumonia used in this study actually increases disease burden (167). Further studies are required to understand if Fludase alters host susceptibility to secondary bacterial infection.

IFNαβ signal to all cell types in the body and, as discussed, are therefore too inflammatory to be used as anti-influenza therapeutics. However, type III IFNs (IFNλ) (**Table 1**) are an intriguing alternative. Discovered in 2003, IFNλ are induced during IV infection via the same pathways as IFNαβ and utilize an almost identical signaling cascade to activate transcription of ISGs (168–170). However, IFNλ engages a separate receptor complex with a limited tissue distribution, compared to the ubiquitously expressed IFNαβR. IFNλ receptor expression is predominantly restricted to mucosal surfaces, such as that of the lung, and only select immune cells, primarily neutrophils (86, 169, 171, 172). There is some evidence to suggest IFNλ may be more critical for protection against IV infection than IFNαβ. In vitro and in vivo analysis has revealed that IFNλ is produced more rapidly and in higher concentrations than IFNαβ by epithelial cells in response to IV infection (101, 170, 172), however this could be attributed to the sensitivity of the assays employed to detect various IFNs. More convincingly, Klinkhammer et al. have recently demonstrated in mice that prophylactic treatment with IFNλ, but not IFNα, confers sustained antiviral protection in the upper airways and blocks IV transmission to uninfected animals (173). In terms of employing IFNλ as in anti-IV therapy, IFNλ treatment consistently administered from 48 to 120 h post infection did not enhance proinflammatory cytokine signaling in the lung but did inhibit IV replication, lowered airway epithelial cell death and consequently promoted host survival (101). Kim et al. reported similar findings and Galani et al. further demonstrated IFNλ signaling to neutrophils also promotes IV clearance (172, 174). Pegylated recombinant IFNλ (PEG-IFNλ) was originally developed to treat Hepatitis C infection, however it was superseded by more specialized treatment options for the disease. Yet during development, PEG-IFNλ passed Phase I and II clinical trials, demonstrating desirable pharmacological properties and a safer drug profile than IFNαβ (175). PEG-IFNλ therefore constitutes a highly promising new broad-spectrum candidate for the treatment IV.

Apoptosis is an important process for resolution of IV infection, not only for elimination of infected cells but also for removing inflammatory cells such as CD8+ T cells, from the pulmonary environment once IV has been cleared. Death-inducing members of the TNF superfamily, including TRAIL and first apoptosis signal (Fas) ligand (FasL) have been shown to induce apoptosis of cells during IV infection (176– 180). DNA microarray analysis performed by Kash et al. found that FasL/Fas signaling related genes in the lung are associated with IAV induced mortality in mice (181). Additionally, ex vivo assessment of human macrophages has shown that TRAIL expression and secretion is enhanced in severe IV induced disease and human peripheral blood mononuclear cells upregulate TRAIL upon IV infection. Furthermore, IAV infection of a human lung epithelial cell line increases cell susceptibility to TRAIL mediated apoptosis (182, 183). Blocking extrinsic apoptosis by inhibition of Fas/FasL interaction though treatment with a recombinant decoy receptor for FasL or interruption of TRAIL signaling, either by genomic deletion or monoclonal antibody (mAb) blockade (**Table 1**) during IAV infection can increase the survival rate of mice after IV infection (115, 151, 179, 182–184). Furthermore, mAb blockade of TRAIL signaling protects against secondary bacterial infection (20). Protecting airway epithelial cells from death during IV infection associates with better prognosis. However, it is a fine balance, as mentioned FasL and TRAIL are also used to control inflammatory cells in the lung. Indeed, in severe IAV infection TRAIL deficient mice are more susceptible to IAV induced disease due to accumulation of cytotoxic CD8+ T cells in the lung (180). As yet, blockade of apoptosis in human IV infection has not been assessed.

An alternative approach to entirely blocking apoptosis is to try to target it specifically to infected cells. B-cell lymphoma 2 (Bcl-2) family members such as Bcl-xL, are key regulators of apoptosis and as such Bcl-2 inhibitors have been developed to treat cancer. It was recently proposed that Bcl-2 inhibitors could also be repurposed for antiviral drug development (185). A series of compounds (ABT-737, ABT-263, ABT-199, WEHI-539, A-1331852) have been show to induce premature death of IAV-infected cells at concentrations that were not toxic for non-infected cells in vitro (186). Furthermore, Bulanova et al. showed that A-1155463 (**Table 1**) limited viral spread (186). The authors hypothesize that recognition of IV infection by the cell causes the release of proapoptotic proteins from Bcl-xL to initiate mitochondrial membrane permeabilization, ATP degradation, and caspase-3 activation. Subsequent addition of Bcl-2 inhibitors in low concentrations acts synergistically, further driving apoptosis of IV infected cells. It appears this phenomenon is not specific to IV, as transfection with plasmid DNA elicited similar effects (186, 187). As ABT-199 (as known as Venetoclax) is approved for use in humans for treatment of chronic lymphocytic leukemia, this class of drugs may have potential to be used as anti-IV therapeutics. However, Kakkola et al. did report that ABT-263 treatment of IV-infected mice resulted in an altered pro-inflammatory cytokine profile in the lung and a slightly higher viral load, which associated with decreased host survival, indicating that these treatments may need to be supplemented with other therapeutics which modulate the inflammatory response or promote viral clearance (187).

# CONCLUDING REMARKS

Globalisation and the continual growth of the world population means that we are living closer together and traveling further distances with greater ease and speed. Emerging strains of IV can transverse the globe in a matter of days. Furthermore, increased demand of fowl and swine products has enlarged the interface between humans and animal reservoirs of IAV, elevating the likelihood of zoonotic transmission. Under these circumstances it is not a case of "if " another IV pandemic emerges but "when." To combat future IV pandemics we need therapeutics to supplement or replace oseltamivir and other NAIs. Of trials registered on clinicaltrials.gov assessing combination therapies to treat IV (25 results, July 2018), all involve a NAI (primarily oseltamivir) and another therapeutic targeted to IV, with the exception of a single celecoxib/oseltamivir trial (NCT02108366). Combinations of antivirals which inhibit different aspects of IV's replication cycle such as inhibitors for PB and PA (NCT02888327) may have synergistic effects and reduce the likelihood of resistant strains developing. However, trials combining anti-HA stem antibodies and NAIs (NCT02293863 and NCT02603952) have reported no decrease in symptom severity and duration compared to monotherapies. As discussed, severity of IV induced disease is a function of the host immune response, therefore combining antivirals with immunomodulatory drugs will likely prove more effective in treating IV infection. Host directed therapies are less likely to drive drug resistance, are more apt for protecting the delicate epithelium from immune mediated cell death and consequently, may be superior at decreasing disease burden. Repurposing of clinically approved immunomodulators is a simple solution. More trials are needed to assess the feasibility of other immunomodulatory drugs to be used as adjuncts to oseltamivir or other antivirals. Selection of appropriate candidates should be based on in vivo models and retrospective studies. Furthermore, taking advantage of inhalers to deliver drugs directly to the site of infection and tailoring therapeutics to epithelial cells, where IV replication occurs will also improve effectivity of treatment while minimizing harmful side effects.

# AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and approved it for publication.

# FUNDING

SD acknowledges funding from NHMRC ECF: GNT1143412.

# ACKNOWLEDGMENTS

SD thanks members of the Wack (Francis Crick Insitute for Medical research) Masters (Walter and Eliza Hall Institute) and Kile (Monash University) laboratories for advice and discussion on this review.

# REFERENCES


healthy adults and children. Cochrane Database Syst Rev. (2014) CD008965. doi: 10.1002/14651858.CD008965.pub3


acute, uncomplicated influenza A. Open Forum Infect Dis. (2017) 4:S519. doi: 10.1093/ofid/ofx163.1352


bearing 5'-phosphates. Science (2006) 314:997–1001. doi: 10.1126/science. 1132998


front-line antiviral protection against influenza virus infection without compromising host fitness. Immunity (2017). 46:875–90.e6. doi: 10.1016/j.immuni.2017.04.025


**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Davidson. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# A Review of DNA Vaccines Against Influenza

### *Leo Yi Yang Lee, Leonard Izzard and Aeron C. Hurt\**

*World Health Organisation Collaborating Centre for Reference and Research on Influenza at the Peter Doherty Institute, Melbourne, VIC, Australia*

The challenges of effective vaccination against influenza are gaining more mainstream attention, as recent influenza seasons have reported low efficacy in annual vaccination programs worldwide. Combined with the potential emergence of novel influenza viruses resulting in a pandemic, the need for effective alternatives to egg-produced conventional vaccines has been made increasingly clear. DNA vaccines against influenza have been in development since the 1990s, but the initial excitement over success in murine model trials has been tempered by comparatively poor performance in larger animal models. In the intervening years, much progress has been made to refine the DNA vaccine platform—the rational design of antigens and expression vectors, the development of novel vaccine adjuvants, and the employment of innovative gene delivery methods. This review discusses how these advances have been applied in recent efforts to develop an effective influenza DNA vaccine.

#### *Edited by:*

*Alan Chen-Yu Hsu, University of Newcastle, Australia*

#### *Reviewed by:*

*Jan Willem Van Der Laan, Medicines Evaluation Board, Netherlands Masaaki Miyazawa, Kindai University, Japan*

#### *\*Correspondence:*

*Aeron C. Hurt aeron.hurt@influenzacentre.org*

#### *Specialty section:*

*This article was submitted to Microbial Immunology, a section of the journal Frontiers in Immunology*

*Received: 18 April 2018 Accepted: 25 June 2018 Published: 09 July 2018*

#### *Citation:*

*Lee LYY, Izzard L and Hurt AC (2018) A Review of DNA Vaccines Against Influenza. Front. Immunol. 9:1568. doi: 10.3389/fimmu.2018.01568*

Keywords: DNA vaccine, influenza, adjuvant, hemagglutinin, immunization

# INTRODUCTION

Seasonal influenza epidemics continue to challenge public health systems worldwide, causing 3–5 million cases of severe respiratory disease and 290–650 thousand deaths annually (1). Despite annual updates to the seasonal vaccine, in 2017 overall vaccine effectiveness for Australia was estimated to be only 33% (2), and interim estimates from the United States were similarly low for the 2017–2018 influenza seasons (36%) (3). In addition, current seasonal vaccines provide little or no protection against novel pandemic viruses of animal origin (4). Consequently, research efforts have increased to improve seasonal vaccines and develop new vaccine platforms to achieve better protection against both seasonal and potentially pandemic influenza A viruses.

DNA vaccines possess numerous properties ideal for influenza control and have been trialled for a range of diseases, including viral and bacterial infections, and some cancers (5–7). Whilst inactivated influenza vaccines (IIVs) largely rely on antibody production to achieve effective protection (8), DNA vaccines can efficiently engage both humoral and cell-mediated immune responses (9). Their production does not require the growth of live virus and can be rapidly upscaled in response to emerging pandemic influenza (10, 11). Despite these advantages, promising immunogenic responses achieved in small animal models, predominantly mice, are rarely replicated in larger animals (12, 13). Murine model data are based on immune responses in highly inbred animals to mouse-adapted influenza viruses—an unreliable comparison to vaccination in the outbred human population against circulating influenza viruses (14, 15). Larger animal models susceptible to human influenza virus provide more relevant data—ferrets exhibit clinical signs, lung pathology, and transmission similar to humans (16, 17), whist human-like immune responses to influenza in cynomolgus macaques are good predictors of vaccine efficacy in humans (15, 18). As such, achieving sufficient immunogenicity in larger animals has required the development of potent delivery systems and adjuvants (19, 20). This review summarises innovations in the design, formulation, and delivery of DNA vaccines against influenza, and the major obstacles impeding their implementation (**Figure 1**).

# INFLUENZA VACCINES: PRODUCTION AND MECHANISMS OF PROTECTION

Inactivated influenza vaccines and live attenuated influenza vaccines (LAIV) are the most widely used forms of influenza vaccine, and are generated by harvesting viruses grown in embryonated hen's eggs (21). The delivery of viral antigens derived from this process induces the production of antigen-specific antibodies, particularly against the haemagglutinin (HA) surface glycoprotein, to protect against future infections (22). However, egg-based vaccine production is time-consuming and resource-intensive, and manufacturing delays have previously caused severe vaccine shortages (23, 24). The overall vaccine effectiveness against seasonal influenza ranges from 40 to 60% during typical seasons, but is significantly reduced when antigenic mismatch occurs (25, 26). Furthermore, antigenic mismatch can be exacerbated by mutations which allow vaccine viruses to grow in eggs, which may also alter antigenic sites (27).

DNA vaccines are able to avoid many issues associated with egg-based vaccine production by generating viral proteins within host cells. To create a DNA vaccine, an antigen-encoding gene is cloned into a non-replicative expression plasmid, which is delivered to the host by traditional vaccination routes (28). Host cells which take up the plasmid express the vaccine antigen which can be presented to immune cells *via* the major histocompatibility complex (MHC) pathways. CD4+ T helper cell activation following MHC class II presentation of secreted DNA vaccine protein is critical for the production of antigen-specific antibodies (29),

whist CD8+ T cell immunity, important for viral clearance, is predominantly activated by endogenously expressed antigens presented on MHC class I molecules (30).

# DESIGNING ANTIGENS FOR INFLUENZA DNA VACCINES

The protection conferred by conventional IIV is based on the induction of HA-specific serum antibodies, which interfere with virus attachment to inhibit cell entry and limit infection (8, 31). Early mouse studies using DNA vaccines encoding H1 HA genes reported that protection against lethal homologous challenge correlated with increasing titres of HA-specific serum antibody (32, 33). As observed in LAIV, other correlates of immunity are less well defined, but the induction of local inflammation and cytotoxic T cell responses have been implicated as key mechanisms to enhance vaccine cross-reactivity and reduce the severity of infections (34, 35). As such, the nature of the host response to influenza DNA vaccination can be manipulated by the encoded antigen.

"Universal" influenza vaccines are being developed to induce broadly protective responses against drifted variant viruses and animal-origin strains that may result in a pandemic. Protection induced by evolutionarily stable influenza antigens is associated with viral clearance mediated by broadly reactive cytotoxic CD8+ T cells, reducing the severity of clinical disease (36). Candidate universal influenza vaccine targets include the nucleoprotein (NP), matrix proteins (M1 and M2), and the RNA-directed RNA polymerase catalytic subunit (PB1). Individual plasmids encoding NP (37) and M2 (38) have each been reported to decrease viral load and enhance survival against lethal heterologous challenge viruses in BALB/c mice. Combined immunisation with matrix protein, NP, and PB1 plasmids has been reported to induce protection against heterologous challenge in mice (39), pigs (40), ferrets (41), and macaques (42). Chimeric protein antigens designed to increase the breadth of host responses can be delivered by DNA vaccines. Plasmid-encoded fusion proteins of H1N1 HA and the conserved M2-ectodomain improved the cross-reactivity of antibody responses to drifted H1N1 viruses compared to a plasmid encoding HA alone in mice (43).

Attempts to create HA-based universal influenza vaccines have targeted the conserved stem region of the HA protein (31). Mice vaccinated with plasmids encoding a PR8 "headless HA" antigen developed serum antibody responses to a greater range of influenza viruses than wild-type HA DNA-vaccinated animals (44). The expression of consensus HA sequences has also increased the cross-reactivity of antibody responses (45–47). Chen et al. (48) constructed a plasmid encoding a consensus H5 HA generated from 467 HA sequences, which induced protection against a wide spectrum of lethal H5N1 reassortant challenge viruses in mice. Broadly reactive responses have also been induced using polyvalent formulations similar to currently available trivalent and quadrivalent IIVs (26). Huber et al. (49) generated crossreactive antibodies against multiple H3 drift variants in mice by vaccinating with three different H3-expressing plasmids. Rao et al. (50) achieved similar success against several variant H5N1 viruses in chickens using vaccines containing up to 10 different H5 HA plasmids.

Efficient antigen expression *in situ* is a key factor for DNA vaccine effectiveness which can be modulated by altering the antigen coding sequence. Encoding antigens using codons optimised for expressing within the host species is a commonly used strategy to enhance influenza DNA vaccine expression (51–53). Jiang et al. (54) used a lethal H5N1 challenge model in chickens to compare the protective efficacy of DNA vaccines encoding either the wildtype HA or HA codon-optimised for chickens. Chickens receiving the codon-optimised HA plasmid demonstrated up to fourfold increases in antibody titre compared to animals inoculated with wild-type HA plasmids, resulting in greater survival rates during viral challenge.

DNA vaccine antigen design can direct the post-translational trafficking of expressed proteins to influence the development of host immunity. The human tissue plasminogen activator leader sequence promotes high levels of protein secretion and has improved antibody responses to an H5 HA DNA vaccine in rabbits (55). Grodeland et al. (29) encoded DNA vaccine antigens consisting of H1 HA linked to MHC class II-targeting units which enhanced its delivery to antigen-presenting cells (APCs). Ferrets and pigs vaccinated with plasmids expressing the targeted H1 fusion protein generated significant antibody titres, whereas H1 DNA alone failed to cause seroconversion. A similar DNA vaccine strategy expressing APC-targeted H7 fusion proteins was found to improve anti-HA serum antibody and cytotoxic T cell responses to highly pathogenic avian influenza (56).

# DNA VACCINE DELIVERY PLATFORMS

For influenza DNA vaccines, the route of administration is critical to vaccine effectiveness as it dictates the cell types that will be transfected. DNA vaccines were initially tested in the murine model using intramuscular injection of naked plasmids to produce antigens in passively transfected myocytes (muscle cells) (57). This method relies on the influx of leucocytes following local inflammation to expose the immune system to DNA vaccine antigens (58). Outside of the murine model, effective intramuscular administration of plasmids depends on adjuvants and delivery systems to achieve sufficient immunogenicity (59). More recently, cutaneous delivery has become a highly desirable route for DNA vaccines, as the epidermis is abundant in Langerhans cells, which can efficiently transport and present DNA vaccine-encoded antigens in the lymph node (58).

Alternative delivery devices have been developed to improve upon traditional needle and syringe inoculation for parenteral administration. In small animal models, the gene gun induces immune responses successfully with low doses of DNA by delivering gas-propelled plasmid-coated gold microparticles directly into epidermal cells (60, 61). Human clinical trials of influenza DNA vaccines have successfully employed the Biojector system (iHealthNet, GA, USA), which uses pressurised CO2 to transport a liquid inoculum to the intradermal or intramuscular layer (62–64). Recently, the development of patches composed of micron-length needles has enabled the dermal delivery of lyophilised DNA vaccine (65–67). HA DNA vaccination in mice using dry-coated microneedle patches were reported to induce antibody titres and T cell responses up to five times higher than an equivalent intramuscular dose (68).

Parenteral gene delivery has been further enhanced by electroporation, which temporarily increases the permeability of local cell membranes with electrical pulses (69). Its early use alongside intramuscular delivery required highly invasive electrodes associated with excessive inflammation and the production of lesions (70). Updated devices such as the CELLECTRA system (Inovio, PA, USA) can target the dermal and subcutaneous layers and are optimised to be minimally invasive for clinical use (71). Electroporation has been reported to enable influenza DNA vaccines to generate robust antibody titres and T cell responses in guinea pigs (72), swine (73), and macaques (47, 74).

The mucosa is an appealing site of inoculation for influenza vaccines as it is easily accessible and is the clinical site of entry for influenza viruses (75). Existing mucosal influenza vaccines such as FluMist (MedImmune, MD, USA) mediate protection through local mucosal inflammation and the production of secretory IgA (34, 76). The enrichment of dendritic cells and M cells at mucosal surfaces is ideal for the immune presentation of DNA vaccine antigens (77, 78). However, successful mucosal delivery of plasmids in large animal models requires specialised adjuvants or highly optimised delivery systems (79, 80). Torrieri-Dramard et al. (81) reported that an intranasal HA DNA vaccine failed to elicit detectable IgA titres unless the plasmid was complexed with a polyethylenimine nanocarrier. To induce detectable seroconversion in sheep, Rajapaksa et al. (82) used a novel acoustic nebuliser to produce aerosols of an optimal size to deliver HA plasmids to deep lung tissue.

# ADJUVANTS

The co-administration of adjuvants with influenza DNA vaccines is a common strategy to elicit adequate levels of protection *in vivo*. The mechanisms of action for licensed conventional adjuvants include the formation of antigen depot at the inoculation site, the activation of inflammatory pathways, and the recruitment of APCs (83).

The goal of adjuvant design is to increase the immune response to vaccine antigens, a critical hurdle in the DNA vaccine field. Mineral salts such as alum are widely used in human vaccines and have resulted in up to fivefold increase in HA DNA vaccine-induced antibody titres (84). Cytokine expression vectors exploit host signalling pathways to heighten immune stimulation (85). IL-6 is an important inflammatory mediator involved in B cell stimulation and the recruitment of leucocytes (86, 87). Co-administration of an IL-6-expressing plasmid in HA DNAvaccinated mice has been reported to reduce the duration of influenza illness (88). Lee et al. (60) reported that only 50% of HA and NP DNA-vaccinated mice survived a lethal homologous challenge, whereas mice receiving the additional IL-6-expressing plasmid were fully protected. The cytokine activity of high mobility group box 1 protein has been shown to increase the survival of mice vaccinated with NP DNA in a homologous viral challenge, and has been found to enhance antibody production induced by HA DNA vaccines by twofold (89). Cytokine adjuvants have also demonstrated effectiveness in large animal models such as macaques, where the use of adjuvant plasmids encoding GM-CSF, a potent immune cell proliferation and differentiation factor, resulted in up to fivefold increases in the serum antibody titre compared to a HA DNA vaccine delivered alone (90).

Adjuvant compounds developed as delivery reagents aim to improve the transfection efficiency of DNA vaccine plasmids. The efficiency of the cellular uptake of DNA vaccines is determined by cell membrane permeability and the susceptibility of foreign DNA to host enzymes. It is estimated that only 1% of a naked plasmid inoculation is able to reach the nuclei of target cells for protein expression—most plasmids remain in the extracellular space to be cleared by host processes (91). Synthetic nanocarriers form structures that protect DNA from host enzymes and facilitate its entry through the cell membrane lipid bilayer (85). Cationic lipids form vesicles known as liposomes, which interact electrostatically with negatively charged DNA to form lipoplexes that efficiently enter host cells through endocytosis (92). Vaxfectin (Vical, San Diego) is a cationic lipid-based system that has boosted influenza DNA vaccine immunogenicity in numerous large animal models (93–95). Other nanoparticle-forming polymers have been reported to enhance influenza DNA vaccine formulation including poly(lactic-co-glycolic) acid (96, 97), chitosan (98), and polyethylenimine (81).

# PRIME-BOOST STRATEGIES

The administration of novel vaccine types including adenovirus vectors (99, 100), subviral particles (101), and recombinant protein antigens (102) in combination with conventional influenza vaccines has been reported to enhance seroconversion and antibody cross-reactivity. Wang et al. (103) demonstrated that a primary HA DNA vaccine followed by a seasonal trivalent inactivated vaccine (TIV) boost induced significantly higher antibody titres compared to two doses of either DNA vaccine or TIV in rabbits. Similar results have reported using DNA vaccines to prime LAIV in ferrets (104) and recombinant HA-protein vaccine in chickens (105). However, human trials applying this strategy against circulating seasonal influenza failed to significantly improve seroconversion compared to TIV alone (106, 107).

Despite this, studies have indicated that DNA vaccines may have a clinical application in pandemic settings. Chang et al. (108) demonstrated that mice which had been pre-exposed to H1N1 were significantly protected from lethal H5N1 challenge after DNA vaccination with H5N1 NP- and M1-expressing plasmids. Given the commonality of H1N1 exposure amongst the public, this suggests DNA vaccines could be rapidly deployed to protect a large susceptible population against H5N1 outbreaks. During the 2009 pandemic, a Phase 1 human clinical trial was conducted using an A(H1N1)pdm09 DNA vaccine produced 2 months before the licensed monovalent inactivated vaccine (MIV) (64). Seroconversion was observed in 30% of recipients after three doses of DNA vaccine delivered by Biojector, and the response rate rose to 72% after a booster dose of MIV. Similar human trials of DNA prime-MIV boost vaccines against H5N1 (109) and H7N9 (110) have reported significant improvements in antibody responses compared to MIV alone, indicating that DNA vaccines can effectively prime the immune system against viruses where there is low pre-existing immunity in the population. These recent developments indicate the potential for further research into combined DNA vaccine/IIV strategies as viable control measures against novel influenza outbreaks.

# FUTURE PROSPECTS

After two decades of research, DNA vaccine technology is gaining maturity—several veterinary DNA vaccines are currently licensed for West Nile virus and melanoma (111), and significantly, the first commercial DNA vaccine against H5N1 in chickens has recently been conditionally approved by the USDA (112). In addition, ongoing large animal trials of DNA vaccines against other diseases such as against HIV (6, 113, 114), hepatitis (115, 116), and Zika virus (117, 118) offer valuable insights that can be applied to influenza DNA vaccine design. Promising approaches have arisen from the numerous studies evaluating different DNA vaccine formulations and delivery systems, but a strategy that consistently elicits protection against influenza in large animal models has not yet emerged. Successful plasmid

# REFERENCES


delivery and the use of appropriate adjuvants remain key challenges that need to be addressed before influenza DNA vaccines become effective for human use.

# AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication.

# ACKNOWLEDGMENTS

We would like to thank Paulina Koszalka for editing and critiquing early drafts of this review.

# FUNDING

The Melbourne WHO Collaborating Centre for Reference and Research on Influenza is supported by the Australian Government Department of Health. LI and LL were supported by NHMRC development grant #1112870.


dissolving microprojection arrays. *Small* (2010) 6:1785–93. doi:10.1002/ smll.201000326


effective than using DNA or inactivated vaccine alone in eliciting antibody responses against H1 or H3 serotype influenza viruses. *Vaccine* (2008) 26:3626–33. doi:10.1016/j.vaccine.2008.04.073


**Conflict of Interest Statement:** The authors declare that they have no commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Lee, Izzard and Hurt. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Advancements in Host-Based interventions for influenza Treatment

*Tsz-Fung Yip1 , Aisha Sami Mohammed Selim1 , Ida Lian2 and Suki Man-Yan Lee1 \**

*1HKU-Pasteur Research Pole, School of Public Health, The University of Hong Kong, Hong Kong, Hong Kong, 2School of Life Sciences and Chemical Technology, Ngee Ann Polytechnic, Singapore, Singapore*

Influenza is a major acute respiratory infection that causes mortality and morbidity worldwide. Two classes of conventional antivirals, M2 ion channel blockers and neuraminidase inhibitors, are mainstays in managing influenza disease to lessen symptoms while minimizing hospitalization and death in patients with severe influenza. However, the development of viral resistance to both drug classes has become a major public health concern. Vaccines are prophylaxis mainstays but are limited in efficacy due to the difficulty in matching predicted dominant viral strains to circulating strains. As such, other potential interventions are being explored. Since viruses rely on host cellular functions to replicate, recent therapeutic developments focus on targeting host factors involved in virus replication. Besides controlling virus replication, potential targets for drug development include controlling virus-induced host immune responses such as the recently suggested involvement of innate lymphoid cells and NADPH oxidases in influenza virus pathogenesis and immune cell metabolism. In this review, we will discuss the advancements in novel host-based interventions for treating influenza disease.

#### *Edited by:*

*Alan Chen-Yu Hsu, University of Newcastle, Australia*

#### *Reviewed by:*

*Stavros Selemidis, Monash University, Australia Anastasia N. Vlasova, The Ohio State University, United States*

#### *\*Correspondence:*

*Suki Man-Yan Lee suki@hku.hk, myleesuki01@gmail.com*

#### *Specialty section:*

*This article was submitted to Microbial Immunology, a section of the journal Frontiers in Immunology*

*Received: 13 April 2018 Accepted: 22 June 2018 Published: 10 July 2018*

#### *Citation:*

*Yip TF, Selim ASM, Lian I and Lee SMY (2018) Advancements in Host-Based Interventions for Influenza Treatment. Front. Immunol. 9:1547. doi: 10.3389/fimmu.2018.01547*

Keywords: host factors, influenza, cytokines, metabolism, immunomodulation

# INTRODUCTION

Influenza remains a source of public health concern. Influenza A virus (IAV) has been the cause of historical noxious pandemics, such as the Spanish flu 1918 H1N1, Asian flu H2N2 1957, Hong Kong H3N2 flu 1968, and more recently the pandemic of H1N1 2009 (Swine flu). Influenza also causes seasonal epidemics and outbreaks with high morbidity and mortality rates such as the 2015 H1N1 outbreak in India (1, 2). The error-prone nature of the viral RNA polymerase (RdRP) and virus' capacity for genetic re-assortment (antigenic drift and shift) result in the viral components' susceptibility to mutations, allowing the viruses to evade the immune system and increases their resistance to control strategies.

Currently, influenza vaccination and two classes of antiviral drugs—M2 ion channel blockers (amantadine and rimantadine) and neuraminidase (NA) inhibitor (oseltamivir, zanamivir, and peramivir)—and the novel treatment option using polymerase inhibitor (favipiravir) are considered as mainstays in influenza infection treatment and control. The use of influenza vaccinations remains challenging due to antigenic drifts and shifts, with seasonal variation of new circulating species. Production of vaccine is time consuming with efficacy concerns, especially in the case of pandemic. Variations in vaccine efficacy caused by age should be aware, with studies suggesting that vaccineconferred protection may not be optimal in certain age groups (3).

The disadvantages of using the conventional antiviral drugs have also been a concern. Significant levels of resistance to both classes of drugs have been repeatedly reported (4, 5). High level of resistance (up to 91%) to M2 blockers has been reported in H3N2 virus strain in American isolates (6). Resistance has also been reported in H5N1 virus (7). IAV resistance to NA inhibitors has also become an increasingly prevalent concern, with the recent highly fatal outbreak of influenza A(H1N1)pdm09 in India 2015 associated with oseltamivir drug resistance (8, 9). In addition, a large cluster of influenza A(H1N1)pdm09 viruses in Japan was found to have increased oseltamivir and peramivir drug resistance (5). There is an urgent need to search for alternative targets to treat influenza virus infections, including non-viral targets such as host cellular factors; which are promising as viruses rely on the host machinery for replication. While host immune response is intended to confer a degree of protection against the infection, an impaired or exaggerated host immune response could be detrimental—IAV H5N1 and H7N9 virus infection was reported to exaggerate aberrant cytokine release, resulting in a cytokine storm that caused accelerated host death (10–12).

Many recent studies have focused on the investigation of targeting host factors to control virus replication as well as modulate immune response, which we have previously evaluated (13). In this review, we will discuss the latest studies (in the past 5 years) on the investigation of novel host-based approaches with potential for influenza treatment.

# STRATEGIES TARGETING HOST CELL MACHINERY

The replication cycle of IAV can be grossly divided into four different stages: (1) entry, (2) genome nuclear import, (3) replication and protein synthesis, and (4) genome nuclear export, apical transport, assembly, and budding. As an obligate intracellular pathogen, IAVs are heavily dependent on host machinery for replication and propagation. To this extent, studies employing genome-wide RNA interference (RNAi) to screen for host factors involved in IAV replication cycle have been performed (14, 15) and an increasing number of approaches targeting these host factors to control IAV replication have been investigated.

# Entry of IAV

Entry of IAV into the host cell is divided into several steps (16, 17). First, hemagglutinin (HA) on the surface of IAV binds to the terminal α-sialic acid on the host cell receptor. This induces the internalization of the viral particle by clathrin-dependent, caveolin-, and clathrin-independent endocytosis (18). Macropinocytosis was revealed as an alternative entry pathway for IAV (19), which subsequently enters the canonical endocytic pathway (20, 21). The vesicle-containing viral particle forms an early endosome (also known as sorting endosome), which matures into a late endosome as the endocytic pathway progresses. A gradual decrease in intraluminal pH from pH 6.5 to 5.0, mediated by V-ATPase proton pump (22), takes place as the endosome matures (23, 24). This pH drop in the endosomal lumen induces a conformational change in HA, which is activated by proteolytic cleavage to generate HA1 and HA2 from precursor molecule HA0 (25, 26). This conformational change triggers the fusion of the viral envelope with the endosomal membrane, releasing the viral genome into the cytoplasm.

Acidification of the endosome causes the subsequent acidification of viral lumen *via* the IAV M2 proton channel (27), which in turn promotes the dissociation of M1 layer from both the viral envelope (24) and the viral ribonucleoprotein (vRNP) complex (28). Interestingly, a sharp decrease in pH from neutral to an acidic pH of 5.0 as utilized by acid bypass has been observed to be sub-optimal for viral replication. It is hence proposed that a gradual decrease in endosomal pH is necessary for sequential reduction in viral stiffness, dissociation of M1 from the NP in the vRNP complex, destabilization of M1 layer from the viral envelope, and the eventual conformational change of the HA for the release of viral genome and proteins to the cytoplasm from late endosome (24).

### Inhibition of Proteolytic Cleavage of HA

Proteolytic cleavage of HA0 to HA1/HA2 is an important step in IAV replication. This cleavage relocates HA2, converting previously uncleaved HA0 to a metastable conformation that induces membrane fusion at acidic pH (29). Inefficient cleavage and activation of HA leads to low infectivity (30). As identified proteins encoded by the viral genome do not possess proteolytic properties, the virus is dependent on host protease for the cleavage of HA. This provides a potential target to control IAV infection. HA is commonly cleaved by trypsin-like proteases at the single arginine residue at position 329. Human airway epithelium serine proteases HAT and TMPRSS2 were identified as the host factors for cleavage at this residue (31).

Aprotinin, purified from bovine lung (32), is a protease inhibitor with a long history of clinical use as an antifibrinolytic agent in cardiac surgery (33). Its potential as an anti-IAV drug has been recognized for over a decade (34) and has been shown to reduce the infectivity of a broad spectrum of IAV strains (34, 35) both *in vitro* (26) and *in vivo* (36). Once withdrawn from the Western drug market due to its association with mortality (33), aprotinin has been approved as a locally administered, small-particle aerosol drug for the treatment of IAV infection in Russia (36). However, side-effects associated with the systemic administration of aprotinin raises the need for an alternative protease inhibitor for use in treatment of IAV infections.

Camostat, a serine protease inhibitor, was reported to demonstrate anti-IAV potential in mice dating back to 1996 (37), but little to no research has been conducted to develop it into an anti-IAV treatment. It was revisited and proven to be one of the most efficient serine protease inhibitors for the inhibition of IAV replication in primary human tracheal epithelial cells *in vitro* when tested compounds were used at similar molarities (35). At present, camostat is widely administered for the treatment of liver fibrosis, chronic pancreatitis, and cancer (38, 39), making it a highly promising candidate for drug repurposing. Despite the lack of association between camostat and increased mortality (as with aprotinin), reports of camostat potentially inducing acute eosinophilic pneumonia (38) warrants the need for careful consideration and further research into the repositioning of drugs from the same class.

Highly pathogenic IAV, such as the H5 and H7 subtypes, are reported to have HA cleavage sites rich in basic residues (30). The polybasic nature of the cleavage sites provides multiple targets for a broad spectrum of proteases, including the more ubiquitously expressed intracellular proteases such as furin (40). This increased protease spectrum could be utilized by these viruses for the activation of HA prior to viral budding, allowing for evasion of potential inhibition by exogenously administered serine protease inhibitors. Furthermore, an *in vivo* study utilizing mice treated with a single protease inhibitor prior to infection with H7 virus bearing a polybasic cleavage site showed poor efficacy despite good results were obtained for infection with H1N1 virus bearing single cleavage site (41), suggesting strain specificity in using serine protease inhibitors to treat IAV infections.

## Inhibition of Endosomal Acidification

Endosomal acidification is required for the release of IAV genome (in the form of a vRNP complex) into the cytoplasm (24). Research has shown that an increase in endosomal pH during the early phases of infection could inhibit IAV infection *in vitro* (42), bringing to light the possibility of controlling IAV infection through the prevention of endosomal acidification.

The V-ATPase inhibitor bafilomycin A1, when used at high concentrations (10–100 nM) has been proven to inhibit IAV replication through the efficient suppression of V-ATPase (43, 44). However, prominent cytotoxicity to host cells was also observed at such concentrations (44). Interestingly, lower concentrations (0.1 nM) of bafilomycin A1 lack inhibitory effects on V-ATPase attenuated IAV replication due to disruption of endosomal trafficking. Thus, bafilomycin A1 is suggested to exert its antiviral function *via* distinct mechanisms at differing concentrations.

Diphyllin, isolated from the plant *Cleistanthus collinus*, is a natural compound able to induce a V-ATPase inhibitory effect (45). In contrast to bafilomycin A1, diphyllin is well-tolerated *in vitro* without inducing obvious cytotoxic effects (46). Most notably, diphyllin is found to effectively inhibit replication of viral strains resistant to amantadine and/or oseltamivir (46). Since drug resistance to these widely administered antivirals is of major public health concern (13), diphyllin is regarded as a promising antiviral against drug-resistant IAV strains.

## Controlling Cholesterol Homeostasis

The release of IAV genomic material during replication requires the fusion of the endosomal membrane with the viral envelope. Since cholesterol plays a major role in controlling the fluidity of the lipid bilayer in cells, it is hence suspected to have a role in the infection cycle of IAV.

Interferon-induced transmembrane proteins (IFITMs) are proteins expressed in many vertebrates (including humans) and are found on the plasma membrane, the membranes of early and late endosomes, as well as on lysosomes (47, 48). While humans express IFITM1, IFITM2, IFITM3, IFITM5, and IFITM10, only IFITM 1, 2, and 3 are both immune-related as well as interferon (IFN)-inducible (48), and have been observed to restrict the replication of different viruses, including IAV (49). Studies suggest that IFITMs limit viral infection by reducing membrane fluidity and hence restrict the hemifusion (the mixing of lipid bilayer without the release of viral content) of viral and endosomal membranes (50), probably *via* the disruption of cholesterol homeostasis of late endosomes, where viral fusion and genome release conventionally take place (51). A recent study using RNAi also demonstrated that cholesterol homeostasis can be regulated *via* acid phosphatase 2 (ACP2)-mediated Niemann–Pick C2 activity and impaired the membrane fusion of IAV and influenza B virus (IBV) (52), further suggesting the importance of controlling cholesterol homeostasis in the release of viral genome to cytoplasm.

On the contrary, later studies suggest that IFITM3 exerts its antiviral activity in a cholesterol-independent manner, showing that an increase in cholesterol composition of late endosomal membranes fail to inhibit viral membrane fusion (53). In addition, studies suggested the accumulation of cholesterol level in the late endosome does not inhibit the IAV genome release into cytoplasm (54, 55).

With the modulation of cholesterol levels in host endosomal membrane as a mean to inhibit IAV host cell entry is still under debate, further studies are required before clear conclusions can be drawn.

## Other Possible Targets for IAV Entry Inhibition

By comparing the miRNA profiles of the IAV-permissive HEK 293T cells and (less permissive) HeLa cells, miRNA-33a has been identified as a negative regulator for IAV infection *via* the inhibition of archain 1 (ARCN1, also known as δ-COPI) (56). ARCN1 is a subunit of the COPI complex that is required for intracellular trafficking and endosome function (57), depletion of which has been reported to inhibit IAV infection (14). Despite impaired IAV internalization caused by ARCN1 depletion *via* siRNA (56, 58), it was not able to recapitulate through acute inhibition of COPI complex by pharmaceutical means (58). It is hypothesized that the long-term (lasting days) perturbation on ARCN1 by RNAi affected the general endosomal trafficking network, a phenomena which cannot be recapitulated by acute pharmaceutical inhibition to block IAV infection (58). The potential of targeting ARCN1 for IAV treatment deserves further investigation, despite the favorable results from RNAi studies.

# Blocking the Nuclear Import of vRNP Complex

Nuclear import of vRNP complexes from the cytoplasm following fusion of the viral and the endosomal membrane is required for replication to take place (59). An early study suggested that vRNP complexes could be transported to the periphery of the nucleus (60), while recent studies report that vRNP complexes utilize the importin-α-importin-β1 (IMPα-IMPβ1) system for nuclear import (59, 61) and lacking of importin-α7, in an importin-α7 knockout mouse model were found to be resistant to IAV infection (62).

Ivermectin has long been clinically administered for the treatment of parasitosis (63), but has recently come to attention as a potential inhibitor of IMPα/β (64). Ivermectin inhibition of IMPα/β has shown to inhibit the replication of RNA viruses such as dengue virus and HIV-1 (64). Ivermectin was recently tested for the inhibition of IAV *in vitro*, with nuclear import of vRNP complex (of both wild-type and antiviral MxA escape mutant) efficiently inhibited (65). Given ivermectin's longstanding record of clinical applications and FDA-approved status, repurposing of this drug for the treatment of IAV should be considered, especially while under threat of pandemic IAV outbreak.

# Genomic Replication and Protein Synthesis

Following the import of the vRNP complex into the nucleus of the host cell, RdRP uses the vRNA as a template to synthesize mRNA or cRNA. Synthesized cRNA remains in the nucleus for new vRNA generation, while mRNA is exported out of the nucleus for translation. Viral protein products are either transported to the cell surface *via* Golgi (in case of HA and NA) or imported back into the nucleus to bind with vRNA, forming new vRNP complex (59). Numerous host factors are involved in this process and hence could be possible targets for therapeutic intervention.

## Regulation of the Splicing of Pre-mRNA

Out of the eight genome segments of IAV, the M and NS segments are well known for undergoing splicing to generate at least two different mRNAs per individual segment (66, 67). Cdc2-like kinase 1 (CLK1) is a kinase which regulates alternative splicing of pre-mRNA (68). Inhibition of CLK1 by the chemical TG003 or knockdown of CLK1 is shown to cause a decrease in M2 mRNA generation and disrupt downstream M2 protein expression, prominently reduced IAV propagation (15).

Clypearin and corilagin were both found to be potent anti-IAV compounds, with a higher therapeutic index than TG003 *in vitro* (69). Clypearin is isolated from herbs used by Chinese medicine practitioners for treating respiratory tract diseases (69), while corilagin is isolated from medicinal plants and herbs. The identification of effective compounds and the systemic investigation of the use of traditional Chinese medicine (TCM) in the treatment of IAV infection open new frontiers in research and therapeutics.

## Inhibition of mRNA Export

During replication, viral mRNA is exported from the nucleus to cytoplasm, where protein synthesis takes place. Human RNA polymerase II activity is found to be correlated with IAV replication through the inhibition of nuclear export of certain viral mRNAs, such as M1 mRNA (70).

Cyclosporine A (CsA) is a FDA-approved drug with immunomodulatory functions (71) that has been found to have an anti-IAV effect in both cyclophilin A (CypA)-dependent and -independent manners (72). The CypA-dependent effect was found to correlate with nuclear export of vRNP complex (see Targeting Nuclear Export Complex). The CypA-independent effect caused inhibition of host RNA polymerase II. CsA is a prospective drug candidate for treatment of IAV infections with a relatively high barrier for development of intrinsic drug resistance, as opposed to commonly used antivirals (73).

Nuclear RNA export factor 1 (NXF1) is a host factor that has been identified to be involved in the nuclear export of IAV mRNA. The knockdown of NXF1 in HEK 293T cells revealed prominent viral mRNA nuclear retention in host cell nucleus (74). Protectin D1 (PD1), an endogenously produced lipid in the respiratory tract, has been identified to have potent anti-inflammatory and antiviral effects (75). PD1 production was notably found to be reduced in the lungs of IAV-infected mice. Therapeutic administration of PD1 was shown to significantly reduce IAV mRNA expression, lower lung viral titer, as well as improve survival of IAV-infected mice. Mechanistic studies revealed attenuated cytoplasmic translocation of viral mRNA with such treatment. A decrease in recruitment of viral transcripts to NXF1 was observed while nuclear export of host RNA remained largely unaffected, suggesting a role of PD1 in regulating NXF1 in nuclear export of viral RNA. Natural PD1 expression in the human airway makes this an ideal candidate for novel therapeutics in the treatment of IAV infection.

## Inhibition of mRNA Translation

The eukaryotic initiation factor-4A (eIF4A) family plays an important role in protein translation (76, 77). eIF4A impairment has been proven to be related to antiviral activity in a broad spectrum of RNA viruses *in vitro* (78), with inhibition of IAV mRNA translation (79). The eIF4A inhibitors, silvestrol and pateamine A were demonstrated to arrest viral protein synthesis, thus blocking viral genome replication *in vitro* (80). Although both silvestrol and pateamine A caused high cytotoxicity at the concentration required effective for IAV inhibition, drugs targeting mRNA translation for various diseases have been approved by FDA or are under active development (81). As such, inhibition of IAV infections by disrupting mRNA translation may well be a therapeutic approach in the future.

### Inhibition of HA Maturation

Post-translational modifications during protein maturation ensure proper function of proteins, with proteins of IAV no exception. Nitazoxanide, a FDA-licensed drug used to treat enteritis, was found to be effective in controlling IAV infection by interfering with HA N-glycosylation as well as intracellular trafficking in host cell and eventually led to a reduction in viral budding (82). Despite the mechanism of nitazoxanide being presently unknown, its ability to inhibit replication of numerous viruses [IAV, respiratory syncytial virus, coronavirus, hepatitis B virus, and many others (83)] suggests that it may act on host machinery. The drug has also been proven *in vitro* to inhibit the propagation of many circulating strains of human IAV, including those resistant to oseltamivir or zanamivir (84). Nitazoxanide has a high barrier of resistance to IAV (83) and other viral strains resistance to neuraminidase inhibitors (85), making it a very promising therapeutic target for IAV treatment. The drug is currently under phase III clinical trials (83).

# Nuclear Export, Assembly, Apical Transport, and Viral Budding

In the later stage of viral replication, viral RNAs of IAV packed with RdRP and NP (known as vRNP complexes) are exported from the nucleus (59), assembled (86), and transported to the plasma membrane [apical in polarized cells (87)] for budding.

# Inhibition of Nuclear Export

### *Targeting Nuclear Export Complex*

Exportin 1 (XPO1, also known as CRM1) is well known for its function in the nuclear export of protein (88) and RNA, including viral RNA (89). Similar to HIV (89, 90), IAV viral RNA does not directly bind to XPO1 but is instead held together by several viral proteins. The viral nuclear export protein (NEP, or previously known as NS2) and the vRNP complex have been proposed as the nuclear export complex (91). Cellular XPO1 has been proven to be crucial in the nuclear export of the vRNP complex, with early studies using leptomycin B (LMB), a potent XPO1 inhibitor, revealing that *in vitro* inhibition of XPO1 led to nuclear retention of vRNP complex (92, 93). However, LMB was deemed unsuitable for development as a potential drug in the phase I clinical trial due to observed cytotoxic effects (94).

Verdinexor (also known as KPT-355) is a new bioavailable selective inhibitor of XPO1. It has been shown to be effective against different strains of IAVs both *in vitro* and *in vivo* as prophylactic and therapeutic treatments (95, 96). It is worth mentioning that delayed administration of verdinexor at day 4 post-infection was still deemed beneficial, with reduced viral load *in vivo* (96). This suggests a prolonged therapeutic time window when compared to the mainstay antiviral drugs such as oseltamivir, where recommended administration is at the early stage of infection (within 48 h of symptom onset) (97). Currently, verdinexor has passed the phase I clinical study trials, suggesting that it does not pose severe cytotoxic effects as LMB does.

In addition, a recent report demonstrated that a new drug, DP2392-E10, which binds and inhibits the function of XPO1, can suppress IAV replication *in vitro* (98) further strengthens the concept of IAV intervention by targeting XPO1.

Viral M1 protein is crucial in assisting the nuclear export of vRNP complex. It was commonly suggested that M1 protein links vRNP complex to viral nuclear export protein NEP which interacts with XPO1 for nuclear export (59). Thus, viral M1 protein may serve as a target to inhibit nuclear export of vRNP. As previously mentioned (see Inhibition of mRNA Export), CsA inhibits IAV replication *via* both CypA-dependent and -independent mechanisms. A recent study using a transgenic mice overexpressing CypA showed greater resistance to IAV challenge (99). In the CypA-dependent mechanism, CsA enhances the binding of CypA to M1 protein (72), increases the self-association of M1, and hinders M1 nuclear import (100). CsA also promotes the CypA-dependent degradation of viral M1 protein (72, 101). CsA seems to be a promising drug to inhibit the nuclear export of vRNP complex by inhibiting viral M1 protein stability and function.

Recently, CD151, a tetraspanin (defined by four transmembrane domains with conserved residues) that is expressed abundantly in lungs and interacts with integrins has been implicated in the regulation of IAV replication *in vitro* and *in vivo* (102). Knockdown of CD151 in primary human nasal epithelial cells resulted in the nuclear retention of host XPO1, viral NP, NEP, and M1 proteins, with an increased survival rate observed in IAVinfected CD151 knockout mice. Co-immunoprecipitation assays suggest that CD151 interacts with viral NP, M1, and NEP proteins (102); however, the exact domains involved in interaction and the mechanism of CD151 function in nuclear export remain unclear. Given that a small molecule inhibitor for CD151 is now under development (103), more data revealing the role of CD151 in IAV infection and subsequent use in targeting CD151 as anti-IAV therapy is anticipated.

# *Targeting the Raf/MEK/ERK Pathway*

During IAV infection, Raf/MEK/ERK signaling cascade is activated, while the inhibition of MEK by U0126, probably mediated *via* myosin (light chain) (104), a known motor protein, impairs the nuclear export of vRNP complexes (105). Suppressing IAV replication by inhibition of Raf/MEK/ERK signaling cascade has been illustrated both *in vivo* (106) and *in vitro* (105). The replication of IBV (107) as well as Borna disease virus (108) was shown to be inhibited by U0126, suggesting the versatility of this approach in controlling infection by different viruses. Despite being effective when administered locally to lungs *via* aerosol, U0126 has little effect when administered orally (106).

Another MEK inhibitor, CI-1040 (also known as PD184352) was shown to have high potency against IAV *in vitro* (106). CI-1040 has completed phase II clinical trials as an anti-tumor drug, with the application of CI-1040 as a potential anti-IAV drug candidate recently revisited. Unlike U0126, CI-1040 is orally bioavailable and oral administration of CI-1040 at 48 h post-infection protected 60% of the IAV-infected mice, while the oseltamivir-treated group experienced a 100% death rate (109). Oseltamivir is known to be effective only when administered in the early stages of IAV infection. This suggests the potential use of CI-1040 as an agent used in IAV treatment due to its potentially longer therapeutic time window than mainstay antivirals.

Formyl peptide receptor 2 (FPR2) located at the host cell surface was identified as an ERK stimulator (110). Antagonizing FPR2 promoted the survival of IAV-infected mice (110). Furthermore, FPR2 antagonists have been described to possess antiviral activity against not only IAV but also IBV infection (111), promoting the idea that antagonizing FPR2 to suppress Raf/MEK/ERK signaling cascade could potentially be a novel approach for the treatment of a broad spectrum of influenza viruses.

## Apical Transport of Viral Components

After the nuclear export of the vRNP complexes, host cell's intracellular transport mechanism is required to deliver vRNP complexes to the host plasma membrane for the assembly of viral RNAs and proteins at the final stage of viral replication. Among the various vesicular compartments found in a cell, the Rab11A+ endosomes are known to recycle endocytosed membrane proteins and lipids to the plasma membrane for membrane homeostasis (112), a property utilized by many RNA viruses, including IAV (87, 113–115). IAV progeny virus production was found to be significantly reduced in Rab11A<sup>+</sup> knockdown human cell lines (116). Furthermore, vRNP complex plasma membrane transport perturbation was observed in *Rab11A* knockdown cells (114, 115); in cells expressing deletion mutant of Rab11 family interacting proteins (87); as well as cells treated with chemicals to interfere microtubule (114). Direct interaction of vRNP complex with Rab11A has also been verified (114, 115), demonstrating the dependence of vRNA complex transport on Rab11A<sup>+</sup> vesicles and the microtubule network during viral replication. Since Rab11A proteins do not confer any mobile properties to the vesicle, molecular motors such as kinesins are required for the active transportation of vesicles through cytoskeletons.

KIF13A, a kinesin-3 family member, was recently identified as a molecular motor for plasma membrane transportation of vRNP-loaded Rab11A<sup>+</sup> vesicles (117). KIF13A knockdown was found to reduce progeny virus production. Overexpression of a mutant form of KIF13A lacking in motor capacity resulted in disruption of the plasma membrane distribution of vRNP complex during later stages of infection. This data suggest that the apical transport of viral components *via* Rab11A or KIF13A could potentially serve as therapeutic targets against IAV infection. Further examination is merited.

Tubulin acetylation and deacetylation affects microtubule stability (118). Histone deacetylase 6 (HDAC6) was found to deacetylate α-tubulin, one of the subunits of microtubule (119). A study has demonstrated that HDAC6 is involved in IAV replication (120). Inhibition of HDAC6 by tubacin or knockdown of HDAC6 gene resulted in an increase of progeny virus production with vRNP complex redistributed toward the periphery of infected cells. In addition, transportation of HA to the plasma membrane for viral budding was also found to be inhibited by HDAC6. This data suggests that activation of HDAC6 by its stimulant could be a potential approach to anti-IAV therapy, despite HDAC6 stimulants still being under development.

### Interference of Viral Budding

While several studies have suggested IAV transmission between cells through apical membranes (121) and intercellular connections (122), virus budding from cell membranes remains the major route for transmission of viruses to uninfected cells. NA is responsible for the cleavage of sialic acid to prevent the interaction between HA and the host cell during viral budding. Besides, viral NA, viral HA, M1 as well as M2, are also suggested to play an important role in the initiation of the budding process (123, 124).

In Section "Controlling Cholesterol Homeostasis," we discussed the involvement of host cholesterol in viral membrane fusion and viral genome release to cytoplasm. Recent studies have demonstrated that host cholesterol may also play an important role in viral budding. It was demonstrated that overexpression of annexin A6 (AnxA6), a phospholipid binding protein, could lead to a decrease in cholesterol levels within the Golgi apparatus and plasma membrane (55), ultimately causing a reduction in egression of progeny virion from infected cells (54). This reduction could be reversed by the addition of exogenous cholesterol (55). Similar to AnxA6 overexpression, addition of a hydrophobic polyamine, U18666A, could reduce cholesterol level in plasma membrane, also inhibited viral replication (55). Since IAV is assumed to bud from lipid rafts (cholesterol-rich plasma membrane domains) (123), it was demonstrated that AnxA6 overexpression or U18666A treatment could hinder progeny virus production by lowering the cholesterol content in the plasma membrane. This hypothesis was strengthened through recent studies resolving the cholesterol-binding site of viral M2 protein, suggesting that IAV M2 clustering (which provides membrane curvature for scission) is mediated by cholesterol (125). A recent report utilizing two different FDA-approved cholesterol-lowering drugs, gemfibrozil and lovastatin, stated that there was reduction in stability and infectivity of progeny virus compared to that replicating within cholesterol-sufficient host cells (126). Taken together, this data suggests that controlling cellular cholesterol content would be an effective alternative with drugs available for repurposing IAV treatment. Further *in vivo* works are needed to confirm this hypothesis.

The Gi-type G-protein coupled receptor α2-adrenergic receptors (α2-ARs) have been recently identified as a key host factor involved in IAV replication (127). Apical transport of the viral protein HA is inhibited by low intracellular cAMP level after stimulating the α2-AR-mediated signaling. *In vitro* stimulation of α2- AR by its agonist clonidine inhibits IAV replication. Therapeutic administration of clonidine reduced pulmonary edema and improved survival rate of IAV-infected mice. Development of a new antiviral targeting the α2-AR-mediated signaling seems promising and deserves further investigation.

# Interrupting the Virus Replication Cycle by Combinatory Use Targeting Both Virus and Host Factors

Although targeting host factors for viral interventions generally provides a better resistance barrier, emergence of resistance may still arise (61). Therefore, combined use of interventions targeting both virus and host factors have been recommended to reduce opportunities for viral development of resistance. One such example would be the combined administration of NA inhibitor (oseltamivir) alongside an anti-host factor [such as V-ATPase inhibitor diphyllin (46), HA maturation inhibitor nitazoxanide (85), FPR2 antagonists (111), and XPO1 inhibitor verdinexor (96)]. While further direct assessment for the ease of emergence of escape mutants between single and combinatory use of drugs is required, the synergistic effects of a combined, multi-drug approach observed thus far highly suggest an increased effectiveness over a single-drug approach.

**Table 1** summarizes novel host targets regulating IAV replication. Compared to RNAi, small molecular chemicals remain the best choice as drug candidates due to their fast acting and easy-todeliver properties. Although small molecular chemicals targeting certain host factors aforementioned have yet to be developed, their RNAi-identified involvement in the IAV replication cycle provide leads for the development of new IAV interventions.

# REGULATION OF ABERRANT IMMUNE RESPONSES IN IAV INFECTION

The immune system aims to protect the host from infection and clear the pathogen once an infection occurs. In addition, the complex networks formed between the host physiology and the immune system co-operatively shape the disease outcome; modulations on the networks could alleviate disease severity in IAV infections.

The immunological responses elicited by IAV infection has been reviewed in detail (128–130). At the initial stage of IAV Table 1 | Advancements on targeting host factors for antivirals.


infection, the respiratory epithelial cells are the primary target for infection. Once the infection is initiated, the recognition of infection is accomplished *via* the detection of pathogen-associated molecular patterns (PAMPs) by pattern recognition receptors (PRRs) (see Toll-Like Receptors), and lead to the expression and secretion of different cytokines and chemokines, such as IL-6, IL-8, tumor necrosis factor (TNF)-α, and CCL2 as well as type I and III IFNs. As sentinel cells, alveolar macrophages could also be infected, inducing cytokines and is the main source of type I IFNs (128, 129). Type I IFNs are known inducer for the upregulation of death receptor 5, which is the receptor for TNFrelated apoptosis-inducing ligand (TRAIL), in lung pneumocytes (128). IL-8 and CCL2 produced by both epithelial cells and macrophages act as chemoattractants for neutrophils and monocytes, respectively. Neutrophils are one of the earliest immune cells being recruited to the site of infection (131) with transmigration of neutrophils carry out by adhesion molecules, such as CD11a, CD11b, and CD18 (132). In addition to the antiviral activity of neutrophil-released reactive oxygen species (ROS), defensin and pentraxin (132), uptaking IAV by neutrophils could also help in controlling viral propagation as these cells do not support replication of IAV (133). Besides controlling viral replication, neutrophils also play an important role in guiding the migration of IAV-specific CD8<sup>+</sup> T-cells in the infection site by secreting and leaving a trail of CXCL12 (134). Infiltrated monocytes will, however, differentiate into macrophages or dendritic cells (DCs). The monocytes-derived macrophages are reported to be a permissive host for IAV production (135), sustaining inflammation by producing cytokines in a magnitude larger than that of the resident alveolar macrophages. The monocyte-derived DC as well as the resident airway CD11clowB220<sup>+</sup> plasmacytoid DC (pDC) and two types of conventional DCs (CD103<sup>+</sup>CD11blow and CD103<sup>−</sup>CD11bhi) acquire the antigen of the invading pathogen through either direct infection or up-taking infected dead cells (129). In the presence of type I IFNs, DCs mature when encountering PAMPs from invading pathogen (129). Depending on the sub-cellular localization of the antigen, cytosolic and endosomal antigen will be loaded onto major histocompatibility complex (MHC) class I and II molecules respectively (130). Once mature, DCs migrate from the infection site to the draining lymph nodes *via* the interaction of CCR7 and CCL19/CCL21 (130, 136) for antigen presentation *via* MHC class I and II to naïve CD8<sup>+</sup> and CD4+ T-cells, respectively (137–140). Interestingly, monocytesderived DCs that engulfed the infected dead cells are poor antigen presenters for CD8<sup>+</sup> T-cells and require the transfer of intact MHC class I/peptide complex to lymph node-resident CD8α+ DCs which are the most efficient antigen-presenting cells to CD8<sup>+</sup> T-cells (137). In addition to antigen presentation, pDC are well known for their high ability in type I IFNs production to limit viral propagation (141).

Within the lymph node, naïve CD8<sup>+</sup> T-cells are activated by the DCs, differentiate and clonal expand into cytotoxic T-lymphocytes (CTLs) with the aid of various cytokines, including IFN-γ, IL-12, type I IFNs, and IL2 (142, 143), and the help from activated CD4<sup>+</sup> T helper cells (144). Differentiated CTLs downregulate their lymph node homing receptor CCR7 and upregulate CCR4 and CXCR3 for the migration to the site of infection. Within the site of infection, CTLs control viral replication by targeting and inducing apoptosis of virus-infected cells *via* the secretion of perforin and granzymes as well as the ligation of death receptors on the infected cells by TNF, Fas ligand, and TRAIL. On the other hand, CD4<sup>+</sup> T-cells are activated by the presentation of MHC class II/ antigen complex by DCs, with co-stimulatory receptors such as CD28 expressed on the T-cells and the ligand for CD28 (CD80 and CD86) expressed on DCs playing an important role (144). Activation of CD4<sup>+</sup> T-cells lead to differentiation into different effector cells subsets, including the classical Th1 and Th2, and the more recently identified regulatory T cells, follicular T helper cells, Th9, and Th17 subsets (144). Th1 cells regulate to the differentiation of CTLs as mentioned whereas Th2 cells contributes to the activation of B-cells through CD40L. Within the pregerminal center of the lymph node, the follicular T helper cells interact with antigen-primed B-cells and promote their proliferation. Antigen-primed B-cells differentiates into plasmablast and undergo antibody class-switching in the germinal center (145). Detailed functions of regulatory T cells, follicular T cells, Th9, and Th17 cells are discussed elsewhere (144, 145). Plasmablasts enter the blood-stream, are recruited to the inflamed tissue, and terminally differentiate into plasma B cells which specialize in the production of antibody for pathogen neutralization, opsonization, and antibody-dependent cell-mediated cytotoxicity, etc. Memory T- and B-cells are also developed during the maturation process, and has been discussed and reviewed elsewhere (146–149). A schematic diagram showing a summary of the immune response after IAV infection has been illustrated in **Figure 1**.

The Yin and Yang theory is always used to describe the importance in balancing the host immune response. In the light of this theory, the treatment strategy aims to suppress the overwhelming activation of the host immune response and in reverse to compensate any unfavorable suppression.

Although adaptive immune responses are important in viral clearance, the immediate innate immunity play an important role in the early control of an infection, and conversely, is a major factor for disease severity due to immunopathology. Dysregulated immune responses caused by viral infections have been implicated in severe disease development (150, 151), such as acute lung injury (ALI). ALI in its most severe form, known as acute respiratory distress syndrome (ARDS), is reported to be the most prevalent cause of mortality in IAV-infected patients (152).

Studies suggested that IAV strains could be associated with either over-activating (human infection by avian H5N1 and H7N9) (153, 154) or suppressing (H1N1, H3N2) (155) immune response.

# Regulation of Neutrophil Infiltration and Neutrophil Extracellular Trap

Recent history has seen the outbreak of IAV pandemics of varying severity takes place at the cost of millions of lives. One such example would be the deadly Spanish flu of 1918, which claimed the lives of 20–50 million of the 500 million people infected worldwide.

The pathological examination of lung sections from mice infected with reconstituted 1918 IAV virus revealed necrotizing bronchiolitis and severe alveolitis in tissue, with neutrophils observed as the predominant inflammatory cell type present (156), suggesting neutrophil involvement in the pathogenesis of IAV infection.

The majority of immune cells in blood circulation are neutrophils; of which they are among the first innate immune cells recruited to the site of infection (131). Neutrophils characteristically control microbial infections by generating bactericidal (157) neutrophil extracellular traps (NETs), consisting of granule proteins, histones, and decondensed chromatin (131). Both protective and destructive role of neutrophils in IAV infections have been described. The contrasting role of neutrophils could be explained by factors such as viral strain and viral dose used in different experimental setup, etc.

The protective role of neutrophils was observed when mice infected with a low, non-lethal dose of IAV H3N2 strain HKx31 displayed neutrophil-mediated viral clearance *via* phagocytosis (132, 158). Depletion of neutrophils has found to enhance viral load in the IAV-infected animals (158).

On the contrary, this protective nature is disputed due to the association of neutrophil-generated NETs. Extensive NET formation was observed in mice infected with PR8, an IAV strain highly pathogenic to mice (159). Histones and myeloperoxidase within the NET induce cell death of lung epithelium and endothelium (157), leading to the loss of integrity of the alveolarcapillary barrier, a characteristic of ALI. Yet, while histones have

(CTLs) in the later stage of infection. Infiltrated monocytes will further differentiate into monocyte-derived macrophages and monocyte-derived dendritic cells (MoDC). Constant surveying of the airway and uptake of virus-infected dead cells by DCs lead to their maturation. Upregulation of CCR7 results in a CCL19/ CCL21-dependent lymph node homing of DCs. Within the lymph node (middle), MoDCs cross-dress CD8α+ DC. CD4+ and CD8+ T-cells are activated by DCs in the presence of cytokines and undergo clonal expansion. Antigen-primed B-cells mature with the aid from follicular T helper cells and further differentiate into plasmablasts in the germinal center. Differential expression of receptors upon maturation of T- and B-cells prompt them to leave the lymph node, enter the blood-stream and recruited to the lung. In the inflamed lung (right), neutrophils leave a trail of CXCL12 to guide CTLs migration. Measures utilized by CTLs for killing infected cells are depicted in inset. Plasmablasts further terminally differentiate into plasma cells and increase antibody production for IAV neutralization.

been shown to suppress IAV replication *in vitro* (160), *in vivo* study demonstrated that there was increase in lung inflammation and damage in IAV-infected mice treated with histones (161). Interestingly, co-treatment of lethally infected mice with anti-histone antibody and oseltamivir resulted in an increase in animal survival when compared to infected mice groups treated solely with oseltamivir (161).

In agreement with the *in vitro* and *in vivo* data, it has been reported that NET produced by cultured neutrophils from patient with H7N9 and severe H1N1 infection increased alveolar epithelial cell permeability (162) leading to ALI. More importantly, plasma NET level positively correlated with the disease severity index (including higher acute physiology and chronic health evaluation II score) and multiple organ dysfunction syndrome (162), further demonstrating the detrimental role of NET in the pathogenesis of severe IAV infections.

Studies have demonstrated the involvement of superoxide dismutase and myeloperoxidase in NETosis, the formation of NET (159). The presence of anti-myeloperoxidase antibody as well as the superoxide dismutase inhibitor (DETC) significantly reduced NETosis. Finally, tetrahydroisoquinolines (163) and a panpeptidylarginine deiminase (PAD) inhibitor, named Cl-amidine (164) have been suggested to inhibit NETosis. Despite it has been reported that during IAV H1N1 infection, PAD4 knockout mice displayed only slight improvement in weight loss and a slight prolonged but no end-point survival advantage was observed compared to WT mice (165), based on the extensive findings presented above, targeting NET to prevent ALI in the severe case of IAV infection, including the highly pathogenic avian IAV, remain promising and may warrant further investigation.

# Innate Lymphoid Cells (ILCs)

Innate lymphoid cells are cells of lymphoid lineages that do not express antigen-specific B- or T-cell receptors (166). Similar to T-helper cells, they are classified into subsets by their ability to produce type 1 (Th1), type 2 (Th2), and type 3 (Th17 and Th22) cytokines.

Previous studies confirmed the involvement of ILCs of group 2 linage (ILC2) in IAV infection and airway inflammation (166, 167). On the positive side, during the recovery phase of IAV infection, ILC2 expresses amphiregulin which promote airway epithelium repair (166, 168), thus facilitating the recovery of the infected lung.

On the other hand, in response to IL-33 produced by macrophages, DCs, and NKT cells, ILC2 secretes IL-5 and IL-13 and induce airway hyper-responsiveness. Recruitment of eosinophils by IL-5 to the lung also mediates airway inflammation (166). Since eosinophilia is a characteristic of allergic asthma and influenza is a major cause for morbidity and mortality in asthma patients (166), it will be of particular interest to investigate the role of ILC2 in IAV infection, particularly in asthma patients.

ILC1s have been initially described as immature NK cells residing in the liver and share many phenotypic similarities with NK cells (169). It was recently appreciated that tissue-resident ILC1s other than the previously recognized NK cells are the major early source of the antiviral IFN-γ at the primary site of various viral infection, including IAV (170). Interestingly, IFN-γ was found to suppress ILC2 activity and reduce IL5 production which exacerbates disease severity during influenza A(H1N1) pdm09 infection (171). This data may highlight a link between ILC1 and ILC2 and suggesting ILC1 can suppress ILC2 activity *via* IFN-γ production during IAV infection.

With ILCs finally identified, functions of these cells and their role in immune response to tumors and pathogen infections have been massively investigated in recent years. Type I IFNs, prostaglandin I2, corticosteroids, and testosterone have been reported to suppress ILC2 activity (172, 173). In addition to IL-33, the epithelial cytokines IL-25, thymic stromal lymphopoietin, as well as the lipid mediator prostaglandin D2 were found to activate ILC2 (173). The therapeutic potential of these ILC2 activators and suppressors is yet to be deduced. With more and more studies demonstrating the involvement of ILC in IAV infection, the interplay between different ILC subtypes in IAV infection would, therefore, be an interesting area to explore and modulate the ILC activity may be a future approach to combat IAV infection.

# Reactive Oxygen Species

Reactive oxygen species, generated by specialized enzymes such as NADPH oxidases, are released during IAV infection (174). The NADPH oxidase family consists of enzymes containing different catalytic subunit named Nox1–5 and dual oxidase (Duox) 1 and 2. ROS have been reported to display both beneficial (limiting viral replication) and detrimental (promoting ALI) effects in the course of IAV infection. Interestingly, the protective or destructive effect of ROS is dependent on the enzyme of which the ROS is generated (174).

Dual oxidase1 and 2 are found to be host-protective (174, 175). *In vitro*, ROS generated by nuclear Duox indirectly regulates the splicing of IAV mRNAs *via* the nuclear speckle-associated splicing complex (175). In addition to altering viral mRNA splicing, ROS generated by Doux2 has been attributed to the production of IFN-λ, an important anti-IAV IFN. In response to IAV infection, increased viral mRNA replication was observed when *Duox2* was silenced *in vitro* (176). Increased viral replication was also observed in mice with *Doux* silenced (175), further depicting the protective role of Doux in IAV infection.

Unlike Doux, Nox2 activation could be harmful to host. IAV infection was reported to induce Nox2-dependent endosomal ROS production (177). ROS could target the conserved Cys98 on Toll-like receptor (TLR) 7, and inhibit TLR7-mediated type I IFN expression during a mild IAV H3N2 infection *in vivo* (177). IAV-infected mice treated with specific Nox2 inhibitor, cholestanol-conjugated gp91ds-TAT, were found to have reduction in endosomal ROS production, restored TLR7 activity, and displayed a decreased viral load (177). In addition to Nox2, Nox4 dependent ROS production has also been reported to activate MAPK/ERK signaling (178), enhancing the export of vRNP complex, thus increasing viral replication (see Targeting the Raf/ MEK/ERK Pathway). *Nox4* knockdown resulted in a reduction of viral replication *in vitro* (178).

Targeting the different NADPH oxidase isoforms, instead of scavenging ROS should be considered as the therapeutic approach for IAV infection, as Doux-mediated ROS production is beneficial (175, 176), while Nox2 and Nox4 are harmful during IAV infections (177, 178). Finally, NS1 (not to be confused with IAV NS1 protein) has been demonstrated to be a Nox inhibitor, which could inhibit the activity of Nox1, Nox2, and Nox4. A study demonstrated that NS1 suppresses IAV-induced Nox2 and significantly inhibits IAV virus replication (179). Besides cholestanolconjugated gp91ds-TAT and NS1 aforementioned, apocynin, a phagocytic Nox2 inhibitor as well as ROS scavenger (180–182), has been demonstrated to ameliorate hyper upregulation of cytokines induced by IAV infection through SOCS1 and SOCS3 *in vitro* (154) and reduce peri-bronchial inflammation and viral titer *in vivo* (183). Interestingly, ebselen, another Nox2 inhibitor and glutathione peroxidase mimetic, could reduce inflammatory status measured in bronchoalveolar lavage fluid (BALF) of mice pre-exposed to cigarette smoke and subsequently infected with IAV (184). Taken together, these reports highlight the potential use of NADPH oxidases inhibitors and ROS scavengers to treat IAV infections.

# Soluble Mediators and Receptor-Based Immunomodulation

Dysregulated cytokine production has been associated with the elevated mortality rate observed in severe IAV infections (185, 186). As such, the immunomodulation of cytokines are regarded as promising therapeutic tactics. Recent advancements developed with this approach will be highlighted in the following section.

# Tumor Necrosis Factor

Tumor necrosis factor has two main functions during viral infection—it activates NF-κB, inducing the expression of cytokines responsible for the host immune response; and induces apoptosis through activation of a signaling cascade involving TRADD, FADD, and caspase 3, 7, 8, and 10 (187–189). TNF is known to be highly upregulated in IAV-infected hosts, especially in hosts infected with highly pathogenic IAV (153, 190). However, it is both protective and counter-protective functions associated with TNF that makes it a target in the treatment of IAV.

The protective role of TNF is observed during infection by low pathogenic IAV, where extrinsically derived TNF is responsible for attenuating tissue-damaging CD8<sup>+</sup> T-cell response (191). In addition to recruiting monocytic cells to the infection site, CD8<sup>+</sup> T-cells response was observed to deteriorate lung pathology (192) and damage healthy, non-infected lung epithelial cells (193) upon IAV infection. Furthermore, TNF deficiency has been associated with an increased detection of IL-15 and IL-6 in BALF (192), which promote the survival of and proliferation of CD8<sup>+</sup> T-cells (194, 195) and subsequent tissue damage. Exacerbated lung pathology caused by the upregulation of the monocyte chemoattractant protein-1 was observed in *TNF*<sup>−</sup>/<sup>−</sup> mice infected with sub-lethal dose of IAV (196). In addition, decreased CD8<sup>+</sup> T-cell contraction due to enhanced expression of the anti-apoptotic protein Bcl-2 was observed in sub-lethally IAV-infected TNFdeficient mice when compared to WT mice (192). As a whole, there is substantial evidence supporting the protective role of TNF in IAV infection.

On the other hand, the correlation of TNF with pulmonary edema has been well-documented (197). TNF has been observed to stimulate the expression of CXCL2 in alveolar epithelial cells in a transgenic mice model resembling extensive IAV infection in lung tissue, causing alveolar damage, lung edema, and hemorrhage (198). In addition to lung edema, TNF has also been reported to correlate with IAV-associated encephalopathy (199, 200). However, it is notable that despite IAV-associated encephalopathy, direct invasion of the central nervous system is rare (201), suggesting that IAV-associated encephalopathy could instead be a result of peripheral infection. Furthermore, TNF has been shown to increase the permeability of the blood–brain barrier (BBB) (202, 203), contributing to neural damage (204). These studies further support an anti-TNF approach as a potential therapy for severe IAV infection.

At present, etanercept, an anti-TNF drug administered in the treatment of rheumatoid arthritis, is the only TNF inhibitor (or even TNF directed treatment) tested for IAV treatment. Etanercept has been shown to protect against the *in vivo* lethal infection of mice with a highly virulent, mouse-adapted IAV strain (205), with observations made of an increased survival rate with decreased morbidity, expression of the proinflammatory cytokine IL-6, lung injury, and edema (205).

## IL-6 and IL-27

The protective role of IL-6 was demonstrated in mice challenged with sub-lethal IAV infection. IL-6-deficient mice displayed exacerbated pulmonary damage (206, 207) and lung injury due to an observed decline in the survival of alveolar type II cells and alveolar epithelial cells (207). IAV suppresses the anti-apoptotic Mcl-1 and Bcl-XL expression, causing cell death of neutrophils which are critical in viral clearance (206). Addition of IL-6 restored the expression of Mcl-1 and Bcl-XL *in vitro* and is considered as the underlying mechanism for the observed survival advantage of WT mice over *IL-6* knockout mice during mild IAV infection.

IL-6 has also been shown to induce the proliferation of lung IL-10<sup>+</sup> regulatory T cells and IL-27, which act to limit excessive proliferation of CD8<sup>+</sup> T-cells and subsequent CD8<sup>+</sup>-inflicted damage. This would hence prevent the tissue damage observed in lung immunopathology (208).

Despite the apparent protective role of IL-6, high levels of IL-6 in serum or cerebrospinal fluid have been reported in severe neurologically complicated IAV cases, with IL-6 used as a marker for prognosis (199–201, 209, 210). The role of IL-6 in regulation of BBB permeability was reported (211), with potentially detrimental neurological complications. As such, the suppression of hyper-induced IL-6 as a form of therapy in severe IAV infection should be considered. One such option is the anti-IL6 antibodybased drug tocilizumab, which is currently administered clinically for the treatment of rheumatoid arthritis. However, study on the usage of this drug to treat hyper upregulation of IL-6 due to severe IAV infection has yet to be conducted. On the other hand, in a case of H1N1 virus-induced ARDS, the use of an extracorporeal cytokine hemoadsorption device to remove cytokines including TNF and IL-6 from the bloodstream (212) has showed beneficial to the patient (213). More research is required to confirm whether the removal or neutralization of IL-6 could be a potential therapy for severe IAV infections.

The activation of CD8<sup>+</sup> T-cell is crucial for viral clearance. It should, however, be tightly regulated to limit CD8<sup>+</sup> T-cell inflicted host cell damage. IL-6 mediates IL-27 induction (208). IL-27 acts to suppress CD8+ T-cells and reduce morbidity through IL-10 and regulatory T-cells (208). Much like other immunomodulatory approaches, the timing for applying IL-27 should be carefully assessed. Compared to placebo-treated IAVinfected group, early administration of IL-27 to IAV-infected mice in fact led to poorer viral clearance, increased morbidity, and deteriorated lung histopathology, while IL-27 administration during the recovery phase (5–10 days post-infection) accelerated recovery and improve lung immunopathology (214). Notably, IL-27 could also suppress Th17 responses and increases susceptibility to secondary *S. aureus* infection (215). Therefore, co-administration of antibiotics should be considered when utilizing IL-27 as potential IAV treatment.

## Type I and III Interferons

Both type I and III IFNs have antiviral properties, with viruses counteract IFNs to gain an advantage for their propagation. The IAV viral protein NS1 inhibits the production of IFNs by antagonizing IRF-3, a key transcriptional factor for IFNs. This prevents the processing of cellular pre-mRNAs (including those for IFNs) and directly interacts with retinoic acid-inducible gene (RIG)-I receptors, which are critical in innate sensing, to suppress IFN production during infection (216, 217). In addition to inhibiting IFN expression, the induction of SOCS3 inhibits IFNs signaling by suppressing cytokine signaling has been documented (155).

The recognition of 5′ triphosphate on viral RNA by RIG-I receptor is shown to induce the expression of SOCS3, which in turn represses type I IFNs expression (155). Due to IFNs being a key contributor to antiviral immune response, an impairment of type I or III IFN production may cause the escalation of otherwise mildly pathogenic IAV infection into a life-threatening one (218).

While type I IFN has been demonstrated to inhibit IAV replication *in vitro* (219); the *in vivo* administration of type I IFN in animal models only displayed effectiveness in a prophylactic capacity. A lowered viral titer was detected in the nasal wash of test animals. However, host susceptibility to IAV infection remained unchanged (219). Notably, this protective effect is only conferred by an optimal dose of type I IFN of low to moderate amounts (10–100 units per mice daily); with higher dosages (1,000–10,000 units per mice daily) shown to increase morbidity (220). In addition, clinical trials demonstrated that prophylactic administration of type I IFN reduced disease severity and lowered susceptibility to IAV in males and participants aged 50 or above (221).

Despite relatively successful results seen in the prophylactic use of IFNs, its therapeutic use is of greater clinical relevance. Mice treated with type I IFN post-IAV infection showed a successful reduction in lung IAV titer but displayed increased morbidity and mortality in comparison to vehicle-treated mice (222). A possible explanation for this phenomenon is the induction of excessive inflammatory response and TRAIL-DR5-mediated epithelial cell death by type I IFN (223), which accounts for the observed lung pathology in IAV-infected animals treated with type I IFN (224). In addition, downregulation of γδ T-cells by type I IFN has been correlated with increased susceptibility to secondary *S. pneumoniae* infection (225), further arguing against the potential use of type I IFNs for the treatment of IAV infection.

In comparison to type I IFNs, the administration of type III IFNs may provide advantages in the control of IAV replication (176, 222, 224) without the risk of previously reported type I IFNs-mediated immunopathologic side-effects (222, 224, 226). However, a recent study aiming to stimulate IFNs signaling through the systematic administration of RIG-I ligand post-IAV infection demonstrated that type I, but not type III IFNs signaling is important in conferring protection during fatal IAV infection *in vivo* (227). Though, this study did not measure the production of type I and III IFNs as well as any changes in viral load with respect to *Ifnar* or *Ifnlr* knockout. In addition, while human immune cells are not primary targets in IAV infection, they could be susceptible to IAV and become efficient host cells for virus replication. They are reported to possess a subpar response to type III IFNs (222); leading to the preliminary conclusion that solely using type III IFN as treatment may not be feasible. As such, reports suggesting the use of type III IFNs over type I IFNs as a front-line therapeutic agent to counter IAV infections may require further investigation.

## Prostaglandin E2

The inhibition of COX-2 by selective inhibitors, nimesulide and celecoxib, was previously demonstrated to suppress the hyper upregulation of pro-inflammatory cytokines induced by highly pathogenic avian IAV (228–230). In addition, the use of zanamivir in tandem with a specific COX-2 inhibitor was shown to increase the survival rate of mice lethally infected with avian H7N9 IAV, when compared to mice treated solely with zanamivir (229).

Activated COX-2 regulates downstream prostaglandin production. One such example is PGE2, a major type of prostaglandin recently demonstrated to play an important role during IAV infection. PGE2 was significantly upregulated in response to IAV infection, leading to the inhibition of antiviral type I IFN production in macrophages and the subsequent increase in virus replication (231). The use of chemicals AH6809 and GW627368X to antagonize PGE2 downstream signaling molecules EP2 and EP4 respectively, was shown to induce antiviral type I IFN production. The *in vivo* treatment of mice lethally challenged IAV with both EP2 and EP4 antagonists significantly improved the survival rate.

A recent study demonstrated the ability of a modified TCM decoction to reduce PEG2 production and subsequent morbidity in mice lethally challenged with IAV. Improved lung pathology was observed (232). The long history of clinical TCM use supports the clinical feasibility of PEG2 inhibition as an option to treat severe IAV infections.

### Toll-Like Receptors (TLRs)

Pattern recognition receptors on host cells sense specific PAMPs present on the viral surface or generated during replication. PRRs can be broadly divided into two classes by their function or location. When defined by location, PRRs are classified into 3 groups—membrane-bound (TLRs and C-type lectin receptors), cytosolic (RIG-I-like and NOD-like receptors), and secreted (collectins and pentraxins) (233).

Significant research has been conducted on PRRs with regards to IAV infection. TLRs and RIG-I receptors have been extensively studied for their major roles in eliciting host immune responses (cytokine and IFN expression) during IAV infection (234–236). RIG-I receptors have been investigated for their functional relevance to IAV infection and targeting these receptors as a form of IAV treatment has been extensively reviewed (237–239). This section will cover recent research on TLRs and the targeting of different TLRs to treat IAV infection.

Humans have been identified to express TLR1–10, while mice have been identified to express functional TLR1–9 as well as TLR11–13 (240). Most TLRs—with the exception of TLR3 utilize MyD88 as an adaptor protein during signal transduction. TLR3 utilizes TRIF as an adaptor. TLR4 is known for its ability to utilize either MyD88 or TRIF, with the choice of adaptor dependent on its sub-cellular location (241). Different TLRs, such as TLR3, 7, and 8 (240) as well as TLR2, TLR4, and most recently TLR10 (235), have been revealed to play a role in the orchestration of host immune responses contributing to IAV pathogenesis.

With TLR10 being an exception (242–244), TLR activation largely causes the release of pro-inflammatory cytokines, with hypercytokinemia leading to ALI as a major cause of mortality in severe IAV infections. In addition to dysregulated cytokine release, excessive production of ROS has been associated with ALI development. In fact, lung injury during severe pulmonary infections, such as IAV and SARS, could be caused by oxidative stress (245). IAV infection activates NADPH oxidase that subsequently produces oxidized PAPC, an endogenous phospholipid. The oxidized PAPC serves as an agonist for TLR4, activating a TLR4-TRIF-TRAF6-NF-κB signaling cascade to eventually trigger the release of IL-6, ultimately inducing the onset of ALI. In addition to oxidized PAPC, the induction of endogenous protein S100A9 upon intracellular PRR DDX21 recognition of IAV subsequently induces the activation of TLR4, further contributing to IAV-induced mortality (246). Since TLR4 has been proven to be important in ALI induction (and hence IAV-related mortality), manipulating the stimulation and antagonism of TLR4 could potentially reduce the severity of IAV infections.

Eritoran (E5564) is a specific TLR4 antagonist initially purposed for the treatment of sepsis, but a failed a phase III clinical trial due to improved patient care in the placebo group prevented its eventual use in sepsis treatment (247). *In vivo* administration of eritoran in mice lethally infected with IAV resulted in improved clinical score, lung pathology results, and reduced viral titer. Delayed administration of eritoran, at day 6 after infection beyond the recommended therapeutic time window (within 48 h after the first display of clinical symptom) for use of oseltamivir (248), also demonstrated a significant benefit to infected mice compared to non-treated group, suggesting a prolonged therapeutic time window for IAV treatment when compared to mainstay antiviral drug treatment. A newer and structurally simpler specific TLR4 antagonist, FP7 (249), alongside a newly developed decoy peptide 2R9 that has been shown to disrupt TLR2, 4, 7, and 9 signaling *via* TIRAP, has been shown to protect mice from lethal IAV infection (250). These results support the potential use of TLR4 antagonism as a means to treat severe IAV infection.

The suppression of other TLR signaling pathways—such as blocking TLR2-mediated signaling through the use of an anti-TLR2 antibody, significantly protected against lethality when administered on day 2 and 4 post-IAV infection (251).

A study also demonstrated that H5N1-infected TLR3 knockout mice had better survival than H5N1-infected wild-type mice, which is evident through the significantly faster regaining of body weight post-infection, lower viral titer in the lung, and fewer pathological changes in the lung (252).

An increasing number of TLR antagonists are now under development (253, 254), alongside several other agents also shown to have effects on TLRs. Polysaccharides isolated from *R. isatidis*, a traditional Chinese medicinal herb used to treat IAV infection, have recently been shown to inhibit pro-inflammatory cytokines such as IL-6 and CCL-5 *in vitro* by down-regulating upstream TLR3 expression (255). MENK, an endogenous protein expressed in the adrenal medulla, was shown to both prophylactically and therapeutically increase the survival rate while reducing viral-caused lung pathology and viral titer in mice lethally challenged with IAV (256). This was determined to be caused by the downregulation of TLR7. These results suggest the potential of down-regulating TLR expression in the treatment of IAV infection.

The above-mentioned data suggest modulation of TLR signaling or expression as a promising approach in treating severe influenza disease and deserves immediate investigation. **Table 2** summarizes new immunomodulatory approaches to combat IAV infections.

# MODULATION OF METABOLISM

It is well documented that patients with diabetes mellitus have a greater tendency to develop severe IAV infection than healthy patients (257). Hyperglycemia increases susceptibility of the


host to IAV infection *via* viral uptake, through the promotion of V-ATPase assembly (258) and immunosuppression (257). In addition, viruses rely on host metabolism to perform essential functions during replication (259–262). These processes exert a large energy demand on the host within a very short period of time (263); energy of which is supplied by and is dependent on host metabolism. IAV viruses have been reported to modify the metabolic state of the host. For example, increased c-Myc-dependent glycolysis and glutaminolysis has been demonstrated in infected cells (264). The changes in glucose and glutamine metabolism were reversed upon the addition of BEZ235, which inhibited the IAV-mediated c-Myc induction. Administration of BEZ235 2 days prior to infection and up to 4 days post-infection was shown to decrease lung viral titer and improve the survival rate in IAV-infected mice. Small molecules such as clotrimazole and α-mangostin that target lipid metabolism have also been demonstrated to suppress IAV replication *in vitro* (264).

In addition to being important for generating energy and biosynthesis, recent research demonstrates that cellular metabolism affects immune cell function. Dysregulated immune responses observed in many diseases are associated with specific metabolic configurations. Viruses, influenza inclusive (265), were found to induce drastic alterations in metabolic levels and programs (263). Macrophages in infected hosts were observed to have marked differences in the Krebs cycle, a key metabolic pathway. This is of significance due to the role of macrophages, which are immune cells critical in the pathogenesis of many inflammatory diseases (263, 265, 266).

In activated macrophages, succinate, a Krebs cycle intermediate, was found to possess inflammatory signal. Accumulation of succinate generates ROS, leading to subsequent activation of hypoxia-inducible factor 1α and the induction of cytokines such as IL-1β (267). A recent study identified the ability of itaconate, another Krebs cycle-derived metabolite, to block the production of inflammatory factors. This prevented inflammation, protecting mice from lethal levels of inflammation that can occur during infection (268). This data suggest the critical roles of Krebs cycle intermediates in regulating cytokine profiles and inflammation. Metabolites generated by innate immune cells in distinct configurations could have different roles beyond that of bioenergetics, with functions in signaling regulation, transcription, and orchestrating innate immune responses.

Despite the lack of research conducted thus far on the application of immunometabolic approaches to influenza treatment, the prospect of manipulating immune responses by modulating immune cell metabolic state is promising. Further research should focus on the identification of metabolites for modulation of immune cell function with substantial improvement of therapeutic strategies to treat IAV disease.

Latest advancements in high-throughput technologies, e.g., metabolomics is a useful approach to systematically investigate the changes of metabolic mechanisms during IAV infections. Identification of important metabolites involved during IAV infection should be a new approach by modulating the host metabolism for interventions.

# CONCLUDING REMARKS

Multiple host-based intervention strategies against influenza have been developed or are under development. While approaches targeting host machinery required for virus replication seem to be promising thus far, additional research is needed to determine the effect of modulating host immune response on influenza treatment. This is increasingly important, since targeted host factors may play distinct roles in response to infection by different influenza viral strains (252), making the management of influenza through solely targeting a single specific host factor is difficult.

Host-based interventions offer obvious advantages over conventional antivirals, such as a higher barrier to drug resistance (73, 83, 107) due to greater genetic stability of host factors than the mutation-prone nature of viral components. In addition, administration feasibility is a key factor to consider the usage of drugs. The mainstays of antivirals for IAV infections, the NA inhibitors, and M2 blockers, are recommended to be administered within 48 h of symptom onset for optimal antiviral activity. This short treatment window may not be fully fulfilled in a clinical setting. Novel host-based interventions were reported to have therapeutic time windows longer than this conventional timeframe (96, 109, 214, 251), even up to 6 days post-infection (248), providing a clear clinical advantage over NA inhibitors and M2 blockers. In addition, hypercytokinemia and ARDS could contribute to disease severity and mortality in instances of severe influenza infection, with virustargeting antivirals providing little to no alleviation of such complications.

Since host immune response is indispensable in host defense against invading pathogens, the use of immune-modulators to suppress detrimental effects while retaining beneficial protection of the host remains challenging. The timing and dosage of medication administration would be critical in determining the drug effectiveness in influenza treatment.

Targeting virus-induced metabolic changes to restore host normal metabolism may be a new direction to combat influenza disease. Further research in the immunometabolism field, alongside studies on modulating immune response to infectious disease by altering host metabolic processes; would create a new direction for future research and is expected to yield significant discoveries that may provide new therapeutic options in the treatment of IAV infections.

# AUTHOR CONTRIBUTIONS

SMYL conceptualized the work. IL and ASMS drafted some review sections and TFY and SMYL wrote the manuscript.

# FUNDING

Funding was provided by Research Grants Council of Hong Kong, General Research Fund #17111714, and Theme-based Research Scheme #T11-705/14-N; Health and Medical Research Fund #12111822 and 14130662; and Procore France/Hong Kong Joint Research Scheme #F-HKU708/16T.

# REFERENCES


metabolite of nitazoxanide. *Antiviral Res* (2017) 147:142–8. doi:10.1016/j. antiviral.2017.10.002


metastatic triple negative breast cancer cell lines. *Proc Biochem* (2018) 66:254–62. doi:10.1016/j.procbio.2017.12.004


junction-associated occludin in acute liver failure. *Liver Int* (2010) 30(8): 1198–210. doi:10.1111/j.1478-3231.2010.02211.x


year. *Influenza Other Respi Viruses* (2013) 7(5):854–62. doi:10.1111/irv. 12094


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Yip, Selim, Lian and Lee. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# RiG-i-Like Receptors as novel Targets for Pan-Antivirals and vaccine Adjuvants Against emerging and Re-emerging viral infections

*Hui Yee Yong1,2,3\* and Dahai Luo 1,2\**

*<sup>1</sup> Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore, Singapore, 2NTU Institute of Structural Biology, Nanyang Technological University, Singapore, Singapore, 3School of Biological Sciences, Nanyang Technological University, Singapore, Singapore*

Emerging and re-emerging viruses pose a significant public health challenge around the world, among which RNA viruses are the cause of many major outbreaks of infectious diseases. As one of the early lines of defense in the human immune system, RIG-I-like receptors (RLRs) play an important role as sentinels to thwart the progression of virus infection. The activation of RLRs leads to an antiviral state in the host cells, which triggers the adaptive arm of immunity and ultimately the clearance of viral infections. Hence, RLRs are promising targets for the development of pan-antivirals and vaccine adjuvants. Here, we discuss the opportunities and challenges of developing RLR agonists into antiviral therapeutic agents and vaccine adjuvants against a broad range of viruses.

### *Edited by:*

*Hiroyuki Oshiumi, Kumamoto University, Japan*

### *Reviewed by:*

*Junji Xing, Houston Methodist Research Institute, United States Carlos Alberto Guzmán, Helmholtz Centre for Infection Research,Germany*

### *\*Correspondence:*

*Hui Yee Yong hyong005@e.ntu.edu.sg; Dahai Luo luodahai@ntu.edu.sg*

#### *Specialty section:*

*This article was submitted to Viral Immunology, a section of the journal Frontiers in Immunology*

*Received: 28 March 2018 Accepted: 04 June 2018 Published: 20 June 2018*

#### *Citation:*

*Yong HY and Luo D (2018) RIG-I-Like Receptors as Novel Targets for Pan-Antivirals and Vaccine Adjuvants Against Emerging and Re-Emerging Viral Infections. Front. Immunol. 9:1379. doi: 10.3389/fimmu.2018.01379*

Keywords: RIG-I-like receptor, pan-antivirals, vaccine adjuvants, interferon, interferon-stimulated genes, RNA therapeutics

# INTRODUCTION

RNA viruses account for a third of all emerging and re-emerging infections (1). Due to the changes of abiotic and biotic landscape encountered by RNA viruses and the error-prone nature of viral replication, RNA viruses evolve quickly and contribute to the outbreak of infectious diseases (2). Many recent outbreaks of emerging and re-emerging viruses involve RNA viruses, and thus, there is an urgent need to develop antivirals against these viruses.

The innate immune system confronts viral infection *via* a specialized group of receptors known as pattern recognition receptors (PRRs) (3). Some PRRs recognize RNA viral infections including toll-like receptors 3 and 8 (TLRs), NOD-like receptors NLRP6 and 9 (NLRs), certain DDX/DHX helicases, and RLRs (retinoic acid inducible gene 1 (RIG-I) and melanoma differentiation associated protein 5 (MDA5)) (4–9). These PRRs usually activate interferon production and the secretion of pro-inflammatory cytokines (10). Interferon activates the Janus kinase/signal transducers and activators of transcription (JAK-STAT) signaling pathway in surrounding cells and the expression of interferon-stimulated genes (ISGs). ISGs inhibit virus replication and spread to surrounding cells by degrading viral nucleic acids and inhibiting viral gene expression (11, 12). Here, we focus on RLRs, the major sensors for pathogenic RNA species which trigger antiviral responses and discuss how modulation of RLRs may lead to broad-spectrum antivirals and new vaccine adjuvants.

# RIG-I-LIKE RECEPTORS

RIG-I-like receptors are a class of DExD/H box RNA helicases which recognizes double-stranded RNA (dsRNA) (13–17). RLRs consist of RIG-I, MDA5, and laboratory of genetics and physiology 2 (LGP2) (18). RIG-I and MDA5 have similar structural domains with N-terminal caspase activation and recruitment domains (CARDs), central helicase domain, and C-terminal domain, which recognizes viral RNA ligands (19–21). RIG-I recognizes short dsRNA and binds to blunt-ended RNA with 5′ triphosphate moiety (22–27). In contrast, MDA5 binds to the stem region of longer dsRNA in a cooperative manner (28–30). LGP2, on the other hand, only have the helicase and C-terminal domain and are involved in the regulatory function of RIG-I and MDA5 (31, 32).

The CARD domains of RIG-I and MDA5 are involved in the activation of downstream signaling event *via* a protein known as mitochondria antiviral signaling protein (MAVS) (33–36). RIG-I binds to unanchored lysine-63 polyubiquitin chains and promotes efficient interaction with the CARD domain on MAVS (37, 38). MAVS protein polymerizes and forms fibrils when activated and will be polyubiquitinated and phosphorylated (38–42). The MAVS oligomer act as a platform to promote downstream antiviral signaling by recruiting several different proteins, such as tumor necrosis factor receptor type-1-associated death domain (TRADD), receptor interacting serine/threonine-protein kinase 1 (RIP1), Fas-associated protein with death domain (FADD), tumor necrosis factor receptor-associated factors (TRAF6, TRAF2, and TRAF3), as well as caspase 8 and caspase 10 (43, 44). TRAF3 activates TANK binding kinase 1/IκB kinase ε/IκB kinase γ/ TANK (TBK1/IKKε/IKKγ/TANK) complex which phosphorylates and dimerizes interferon regulatory factors 3 and 7 (IRF3 and IRF7). The activated IRF3 and IRF7 translocate into the nucleus and activate IFN production (45, 46). TRAF 2 and 6 activate the IKKα/β/γ (also known as NEMO) by ubiquitination and resulting in activation of NFκB and the expression of pro-inflammatory cytokines (**Figure 1**) (41, 47).

# PAN-ANTIVIRALS TARGETING RIG-I

Since RLRs are the key component for the antiviral immune response, these sensors are targets for antiviral therapeutics development. Current antiviral interventions focus on the use of direct-acting antivirals (DAAs), which target the essential components in the life cycle of a virus and thus are virus-specific (48). Although DAAs are highly effective, the low fidelity replication of the RNA virus genome could ultimately lead to the emergence of DAA therapies escape mutant (49). To circumvent this problem, broadly targeting antiviral therapeutics need to be used synergistically with DAAs. To this end, RIG-I agonists or RIG-I pathway activators represent a novel group of promising antiviral candidates. Lists of the antiviral candidates are discussed below as three categories based on their chemical nature (**Table 1**).

# NUCLEOTIDE-BASED ANTIVIRALS

A dinucleotide-derived small molecule compound, SB9200, has been shown to induce IFN *via* RIG-I and nucleotide-binding oligomerization domain containing protein 2 (NOD2). SB9200 is believed to interact with RIG-I and NOD2 that are associated with pre-genomic RNA thus blocking the HBV viral polymerase from replicating the genomic RNA (60). SB9200 was shown to confer dose-dependent and long-lasting induction of IFNα, IFNβ, and ISGs in liver tissue (50). Treatment of this compound in woodchucks infected with Woodchuck Hepatitis Virus (WHV) showed no sign of toxicity with reduced hepatic WHV antigen and nucleic acid. The sequential treatment of WHV-infected woodchuck with SB9200 followed by entecavir (ETV), a currently used antiviral to treat Hepatitis B (HBV), showed a reduction in viremia and delayed recrudescence of viral replication. The viral reduction from the treatment of SB9200 was comparable with current antivirals, such as Emtricitabine, Tenofovir, and Adefovir, when administered for up to 12 weeks (51). Drugs such as Emtricitabine, Tenofovir, and Adefovir are commonly used for the treatment of HBV. These drugs, however, may cause side effects such as lactic acidosis and possible liver, and kidney failure. SB9200 is effective for HCV patients with relapse after DAA and interferon treatment and could serve as a promising treatment option for patients who are not responding to the current regimen of DAA therapy (52). The phase 1 clinical trial on naïve adult with chronic hepatitis C showed an association between the decline in viral RNA and the peak of SB9200 detection in plasma (Clinical trial no NCT01803308). Currently, SB9200 is being tested in phase 2 clinical trials for treating subjects chronically infected with the HBV.

# RNA-BASED ANTIVIRAL CANDIDATES

5′ triphosphorylated and diphosphorylated short dsRNAs are RIG-I specific ligands (22, 26, 61, 62). Goulet et al. showed that 5′pppRNA could activate a broad spectrum of antiviral and inflammatory genes such as IRF3, IRF7, NFkB, and the downstream ISGs. Treatment of lung epithelial cells A549 with 5′pppRNA confers protection against vesicular stomatitis virus (VSV), vaccinia virus, and dengue virus (DENV). The antiviral effect of 5′pppRNA was also detected against HIV in CD4+ T cells and HCV in Huh 7.5 cells (53). Besides that, 5′pppRNA was also an effective antiviral against influenza virus infection *in vitro* and *in vivo*. Treating mice with 5′pppRNA prior to influenza virus challenge also reduces pneumonia related to influenza virus infection (53). In another study, 5′pppRNA was shown to stimulate host antiviral response and reduce the infectivity of DENV and chikungunya virus (CHIKV) in human myeloid, fibroblast, and epithelial cells *via* RIG-I specific activation (54).

Several studies were also carried out to determine factors such as sequence, length, and structure of 5′pppRNA to enhance the antiviral activities of RIG-I (55, 56). Chiang et al. showed that 5′pppRNA with uridine-rich sequence with 99 nucleotides hairpin (M8) triggered higher interferon response when compared to other RIG-I aptamer and poly(I:C). M8 specifically activates RIG-I without triggering MDA5 or TLR3 activation. Prophylactic and therapeutic treatment using M8 protect cells from dengue and influenza viral infections. Furthermore, administration of M8 followed by influenza virus challenge improves the survival rate of mice with low lung virus titer detected at day 3 post-infection (55). In another study carried out by Lee et al., different RNA fold was shown to elicit different antiviral properties *via* RIG-I. Short hairpin RNA with a bent in the stem structure with phosphorothioate backbone was used as antiviral and was more potent than oseltamivir against influenza A H1N1 virus *in vitro* (56). Linehan et al. recently showed that short RNA with stable tetraloop at one

Figure 1 | Viral RNA is recognized by RIG-I-like receptors (RLRs), RIG-I, or melanoma differentiation-associated protein 5 (MDA5). Activated RLRs interacts with mitochondria antiviral signaling protein (MAVS) adapter protein *via* CARD–CARD interactions. Activated MAVS then interacts with tumor necrosis factor receptorassociated factors 3 (TRAF3), tumor necrosis factor receptor-associated factors 6 (TRAF6), tumor necrosis factor receptor type-1-associated death domain (TRADD), receptor interacting serine/threonine-protein kinase 1 (RIP1), Fas-associated protein with death domain (FADD), and other signaling molecules. TRAF3 activates TANK binding kinase 1 (TBK1) and IκB kinase ε (IKKε), which phosphorylates interferon regulatory factors 3 and 7 (IRF3 and IRF7). The phosphorylated IRF3 and IRF7 dimerize and translocate into the nucleus to induce type 1 interferon response. On the other hand, MAVS interaction with receptor interacting serine/ threonine-protein kinase 1, FADD, TRAF6, and TRADD. TRAF 6 ubiquitinate NF-kappa-B essential modulator (NEMO) which then activates IκB kinase and activates NF-κB. NF-κB transcription factor drives the expression of type 1 interferon and proinflammatory cytokines.

end of duplex RNA triggers a robust IFN 1 response *in vivo*. These short stem-loop RNA (SLR) induces a subset of genes involved in antiviral and effector responses as well as represses gene involved in T cell maturation and could potentially be developed into a highly effective antiviral or vaccine adjuvant (57).

# SMALL MOLECULAR COMPOUNDS

High throughput screening (HTS) of small molecule compounds identified a group of novel agonists of the innate immune pathway. The isoflavone-like compound confers protection against HCV and Influenza A virus *in vitro*. These compounds were also shown to activate a narrower subset of genes and thus have potential to be useful antiviral without causing cytokine toxicity (58).

Another class of small molecule compounds, hydroxyquinolines, identified *via* HTS of compound library in cell culture induces the expression of innate immune antiviral genes such as RIG-I, IFIT1, IFIT2, IFITM1, OAS3, and MX1. Remarkably, although these compounds were able to induce high expression of antiviral genes, the expression of type I and III interferon remains low, suggesting the activation of distinct antiviral pathway than that of RIG-I agonists. The specific target(s) of these hydroxyquinoline


(*Continued*)

#### TABLE 1 | Continued

*RLRs, RIG-I-like receptors.*

compounds are not known. These hydroxyquinoline compounds were effective antivirals against a broad range of RNA viruses from the families Flaviviridae, Filoviridae, Paramyxoviridae, Arenaviridae, and Orthomyxoviridae. Interestingly, these compounds exhibit both prophylactic and therapeutic activity against infection and could be used in combination with other antivirals (59).

# INNATE IMMUNE POTENTIATOR AS VACCINE ADJUVANTS

Adjuvants act as an immune enhancer to a vaccine. Several different classes of adjuvant had been approved for use in human vaccines such as alum and oil in emulsion MF59, AS03 (oil in water emulsion), virosomes, and AS04 (aluminum with monophosphoryl lipid A) (63). Alum and MF59 act by increasing antigen uptake at the injection site and activates pro-inflammatory responses (64–66). Alum mainly acts *via* Th2 cellular immune response, which does not confer the best protection for viral infections such as HCV and HIV. Moreover, there are safety concerns with the use of alum as an adjuvant with reported cases of hypersensitivity and erythema (67, 68). Well characterized agonists of innate immunity may serve as a better candidate of targeted vaccine adjuvants (**Figure 2**).

A small molecule compound named KIN 1148, discovered *via* HTS, was shown to activate IRF3 nuclear translocation. When this compound was tested with influenza split virus vaccine H1N1 A/ California/07/2009, it confers protection from lethal challenge of influenza virus strain A/California/04/2009. KIN1148 together with the vaccine confers protection *via* IL-10 and Th-2 response to T cells in lung and lung-draining lymph nodes. Immunization with vaccine and KIN 1148 showed a significant increase in IgG antibodies with serum from mice receiving prime-boost immunization conferring protection to naïve mice from influenza challenge. KIN1148 was shown to be able to work alongside the vaccine to boost protective immunity and protect against influenza strain A/California/04/2009 (**Table 2**) (69).

5′triphosphorylated duplex RNA was tested as an adjuvant by Beljanski et al. M8 a potent triphosphorylated RNA was used in conjunction with virus-like particle (VLP) expressing H5N1 influenza hemagglutinin and neuraminidase. The combination of VLP and RNA increases the survival rate of mice infected with H5N1 influenza virus and induces higher antibody titer against influenza virus as compared to other adjuvants such as alum, addavax and poly(I:IC). Furthermore, vaccination with VLP and RNA stimulates TH1-biased CD4 T cells response in mouse sera (70). Another 5′triphosphorylated RNA derived from Sendai virus defective interfering RNA (SeV DI RNA) was also tested as adjuvant together with the H1N1 2009 pandemic vaccine and was shown to enhance production of influenza-specific IgG antibodies and influenzaspecific IgA antibodies indicating that this 5′triphosphorylated RNA could potentially be used as influenza vaccine adjuvant (71).

# DELIVERY OF RNA-BASED AGENTS REMAINS CHALLENGING

To be effective as therapeutics, functional RNA species must be internalized into targeted cells. The delivery methods commonly used for RNA-based or nucleotide specific antivirals includes the lipid-oligo complexes, nanoparticle-based delivery, and viral-based delivery. The hydrophilic, negatively charge nature of RNA hinders the direct uptake of naked oligos into cells. The administration of RNA *via* inhalation was poorly efficacious in

Figure 2 | The use of innate immune potentiator as adjuvant triggers the stimulation of adaptive immune responses. Innate immune potentiator stimulates RIG-I-like receptors (RIG-I) and interacts with MAVS adapter. This results in the activation of downstream signaling pathways and release of type I interferon. Type I interferon couple with the presence of antigen trigger DC maturation by enhancing surface marker expression and antigen presentation. The activated DCs interact with CD4+ T cells and thus stimulate Type 1T helper (TH1) cells. TH1 cells in turn interact with B cells to produce antibodies and trigger clonal expansion of B cells and T helper cells.

Table 2 | Innate immune potentiator as virus vaccine adjuvants.


lung macrophages and only intratracheal administration leads to efficient delivery of RNA into the targeted site (72). Besides suffering from poor uptake, naked RNA is often prone to degradation in plasma (73).

One favored method of delivery for RNA is lipid-oligo complexes. This delivery method is more efficient due to the tendency of lipid to interact with the cell membrane and improve the uptake of RNA (74). A biocompatible lipid-based carrier can further reduce undesired immunogenic activation. However, the cationic nature of lipid is reported to interact with proteins in serum, and these aggregates are cleared by organs such as the spleen, lung and liver (75).

Nanoparticle-based delivery is a versatile method for RNA delivery with many organic or inorganic materials available as nanocarriers. Nanoparticle requires less RNA material with a large surface area for interaction with cells (76). The material used as a carrier could also be tailored for application such as dose, circulation time, as well as passive or active release of RNA (77). To overcome the issue of toxicity or uncontrolled immune activation, multistage delivery of RNA from nanoparticles could be carried put (78). The downside of this strategy is the need to test various materials for RNA delivery and this would drive up the cost of therapeutics or vaccine.

# THE DANGER OF UNCONTROLLED IMMUNE ACTIVATION ALWAYS EXISTS

The therapeutic and prophylactic use of pan-antivirals was previously demonstrated in viral infection of influenza and dengue (55, 69, 71). Hotz et al. demonstrated that the pre-exposure of murine APC to synthetic poly(I:C) inhibits RLR activation while augmenting the sensitivity of TLRs. This would also imply a narrow therapeutic window for the use of pan-antiviral RNA targeting RIG-I (79). For clinical usage, the dosage of therapeutics is important to minimize side effects such as exacerbated cytokine storms and toxicity. Prater et al. showed that the injection of pregnant C57BL/6 mice with a high dose of CpG-ODN resulted in high fetal resorptions and craniofacial/limb defects (80). RIG-I agonists face similar concerns.

# RIG-I SNPs MAY LEAD TO POOR OR HYPER-RESPONSIVENESS

There are 324 RIG-I SNPs identified from NCBI SNPs database with 8 resulting in amino acid changes or truncation. The S183I mutation on RIG-I weakened the antiviral signaling and produces a low level of IFNβ and NFκB upon IAV and SeV challenges in a cell-based assay. Another SNP of RIG-I at P229 resulted in frameshift mutation at the CARDs domain and triggers constitutive expression of IFNβ suggesting that individual with this mutation could be linked to hyper-responsiveness in the immune system or autoimmune diseases (81). Another commonly found SNP (rs10813831) of RIG-I resulted in the substitution of R7C which could alter RIG-I interaction with MAVS (82). Individuals with these SNPs have a lower rubella-specific IgG titer when immunized with live measles-mumps-rubella (MMR-II) vaccine (83). This SNPs mutation also increased

# REFERENCES


the IFNβ level and RIG-I transcription in human dendritic cells when infected with Newcastle disease virus (NDV) (82). Individuals with these SNPs were also shown to have complications of brainstem encephalitis when infected with enterovirus 71 (84). Several intronic SNPs also alter the cellular and humoral response to the measles vaccine. Genotype of individual carrying the SNP in minor allele of RIG-I (for rs12555727, rs12006123, and rs17289116) also showed less virus-specific IFN-γ secretion against measles. These findings imply that genetic variants are also involved in initial antiviral responses to vaccination (85) The haplotype of RIG-I rs3739674 which is located in the 5′UTR is associated with higher EV71 HFMD risk possibly by altering the expression level of the gene (86). In order to target RIG-I as pan-antiviral or vaccine adjuvant, the different haplotypes affecting the disease outcome should be considered. Dosage concern should be taken into account to enhance the effectiveness of RIG-I as a broadly targeting antiviral or vaccine adjuvant.

# CONCLUSION

Emerging and re-emerging viruses present a significant public health concern, and there is an urgent need for novel vaccination and treatment strategies. RIG-I agonists as new adjuvant candidates may work alone or couple to vaccine agents such as VLPs or recombinant proteins to improve the safety and efficacy of conventional vaccines. Antivirals targeting the innate arm of immunity (host-directed therapy) would be useful to confer protection against emerging and re-emerging viruses (87). However, the development of such vaccines and antivirals is still in its infancy and many challenges related to the production and safety evaluation of vaccines and antivirals. Several key issues still need to be addressed including production platform, formulation, delivery, safety, and the ability of such class of the vaccine adjuvant or antivirals to be used in immunocompromised and elderly.

# AUTHOR CONTRIBUTIONS

DL and HY discussed, wrote, and revised the manuscript.

# ACKNOWLEDGMENTS

This work was supported by (1) a start-up grant from Lee Kong Chian School of Medicine, Nanyang Technological University, (2) MOE2016-T2-2-097 grant, and (3) National Research Foundation grant NRF2016NRF-CRP001-063.


type I interferon-independent stimulation of the innate antiviral response. *J Virol* (2014) 88:4180–94. doi:10.1128/JVI.03114-13


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Yong and Luo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Pathogenicity and Viral shedding of Mers-coV in immunocompromised rhesus Macaques

*Joseph Prescott <sup>1</sup> , Darryl Falzarano1 , Emmie de Wit1 , Kath Hardcastle2 , Friederike Feldmann2 , Elaine Haddock1 , Dana Scott <sup>2</sup> , Heinz Feldmann1 and Vincent Jacobus Munster1 \**

*<sup>1</sup> Laboratory of Virology, Division of Intramural Research, Rocky Mountain Laboratories, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Hamilton, MT, United States, 2Rocky Mountain Veterinary Branch, Division of Intramural Research, Rocky Mountain Laboratories, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Hamilton, MT, United States*

#### *Edited by:*

*Alan Chen-Yu Hsu, University of Newcastle, Australia*

### *Reviewed by:*

*Byron Emanuel Eusebio Martina, Erasmus University Rotterdam, Netherlands Dirk Dittmer, University of North Carolina at Chapel Hill, United States*

> *\*Correspondence: Vincent Jacobus Munster munstervj@niaid.nih.gov*

#### *Specialty section:*

*This article was submitted to Microbial Immunology, a section of the journal Frontiers in Immunology*

*Received: 13 November 2017 Accepted: 24 January 2018 Published: 12 February 2018*

#### *Citation:*

*Prescott J, Falzarano D, de Wit E, Hardcastle K, Feldmann F, Haddock E, Scott D, Feldmann H and Munster VJ (2018) Pathogenicity and Viral Shedding of MERS-CoV in Immunocompromised Rhesus Macaques. Front. Immunol. 9:205. doi: 10.3389/fimmu.2018.00205*

Middle East respiratory syndrome coronavirus (MERS-CoV) has recently emerged in the Middle East. Since 2012, there have been approximately 2,100 confirmed cases, with a 35% case fatality rate. Disease severity has been linked to patient health status, as people with chronic diseases or an immunocompromised status fare worse, although the mechanisms of disease have yet to be elucidated. We used the rhesus macaque model of mild MERS to investigate whether the immune response plays a role in the pathogenicity in relation to MERS-CoV shedding. Immunosuppressed macaques were inoculated with MERS-CoV and sampled daily for 6 days to assess their immune statues and to measure viral shedding and replication. Immunosuppressed macaques supported significantly higher levels of MERS-CoV replication in respiratory tissues and shed more virus, and virus disseminated to tissues outside of the respiratory tract, whereas viral RNA was confined to respiratory tissues in non-immunosuppressed animals. Despite increased viral replication, pathology in the lungs was significantly lower in immunosuppressed animals. The observation that the virus was less pathogenic in these animals suggests that disease has an immunopathogenic component and shows that inflammatory responses elicited by the virus contribute to disease.

Keywords: Middle East respiratory syndrome coronavirus, immunosuppression, pathology, shedding, macaque monkey

# INTRODUCTION

A novel coronavirus (CoV) emerged in Saudi Arabia in June of 2012 that is the causative agent of a severe respiratory disease called Middle East respiratory syndrome (MERS) (1). Thus far, there have been almost 2,100 diagnosed cases (2). Despite increased surveillance and the identification of many new cases, the case fatality rate has remained high and is currently approximately 35%.

Although there have been a high number of cases, little is known about the mechanisms of pathogenesis and the disease progression in humans is poorly described. Clinical features range from asymptomatic infection, to an acute respiratory distress syndrome, and multi-organ failure (3). The majority of patients that have succumbed to MERS-CoV have had comorbidities (4, 5) and disease is thought to be more severe in immunocompromised patients. However, the actual mechanisms of disease remain to be elucidated. The virus has been shown to replicate in human primary epithelial and *ex vivo* human lung cultures, especially in non-ciliated bronchial epithelial cells and alveolar type II pneumocytes (6–8) and the receptor has been identified as dipeptidyl peptidase 4, which is expressed on these cell types (9). MERS-CoV shedding is higher in patients with more severe disease manifestations compared to milder cases (10).

Our laboratory has recently developed two non-human primate models of MERS, utilizing the rhesus macaque and the common marmoset (11–13). Rhesus macaques develop a mild pneumonia upon intratracheal inoculation with MERS-CoV (12). In this model, virus replicates within the respiratory tract to modest levels, and is detectible in oral and nasal swabs. However, clinical disease is most prominent within the first few days after inoculation and animals show signs of disease resolution soon after. Disease in rhesus likely models the mild form of the human disease, where the infection is self-limiting and clinical signs and symptoms are mild (10, 14, 15). In an effort to examine whether the immune status of an individual influences the disease severity and pathogenicity and replication kinetics of the virus, we downregulated the immune system of rhesus macaques using immunosuppressive drugs. We found that MERS-CoV replicated to significantly higher titers and disseminated outside of the respiratory tract in immunosuppressed animals, yet pathology was markedly reduced in these animals, showing that disease has an immunopathogenic component.

# MATERIALS AND METHODS

# Ethics Statement

The use of study animals was approved by the Institutional Animal Care and Use Committee of the Rocky Mountain Laboratories and experiments were performed following the guidelines of the Association for Assessment and Accreditation of the Laboratory Animal Care by certified staff in an approved facility. The guidelines and basic principles in the United States Public Health Service Policy on Humane Care and Use of Laboratory Animals and the Guide for the Care and Use of Laboratory Animals were followed. All procedures were carried out under anesthesia using Ketamine by trained personnel under veterinarian supervision and efforts were made to provide for the welfare of animals and to minimize suffering. All animals were humanely euthanized at the endpoint of the study (6 days post-inoculation) by exsanguination under deep anesthesia. All standard operating procedures for MERS-CoV were approved by the Institutional Biosafety committee of the Rock Mountain Laboratories, and sample inactivation was carried out according to approved standard operating procedures prior to removal from high containment.

# Virus Propagation

Middle East respiratory syndrome coronavirus (isolate EMC/2012) was propagated in Vero E6 cells in DMEM (Sigma) supplemented with 2% FBS (Logan), 1 mM l-glutamine (Lonza), 50 U/mL penicillin, and 50 µg/mL streptomycin (both from Gibco).

# Rhesus Macaque Immunosuppression and Inoculation

Five Rhesus macaques (female, weighing 7–11 kg, 11 years of age) were enrolled in this study. Immunosuppression (animals ISCoV1-3) was achieved by administration of cyclophosphamide (CyP) (Roxane Laboratories) (10 mg/kg dissolved in 30 mL of a meal supplement (Boost) and delivered *via* an orogastric tube under anesthesia every other day starting 16 days prior to virus inoculation and ending 2 days after inoculation), and dexamethasone (Dex, 2 mg/kg daily by subcutaneous injection beginning 16 days prior to virus inoculation and ending 5 days after inoculation). Mock immunosuppression (CoV1-2) was performed following the same schedule, but orogastric feeding did not contain CyP and injections consisted of sterile PBS. Immunosuppression was confirmed by monitoring white blood cell (WBC) populations using a HemaVet (Drew Scientific). For inoculation, 7 × 106 TCID50 of MERS-CoV was diluted in 7 mL of DMEM and delivered *via* intratracheal (4 mL) oral (1 mL) nasal (1 mL), and ocular (1 mL) routes as previously described (12). Clinical exams were performed on days −18, −16, −10, −4, −2, 0, and +1 to +6 relative to virus inoculation. Blood was obtained at these times as well as nasal and oral swabbing and chest radiographs starting on day 0. Six days after inoculation, all five animals were euthanized and necropsies performed to obtain samples of the following tissues: lungs (all six lobes), bronchi, oro/ nasopharynx, trachea, tonsils, heart, liver, spleen, kidney, adrenal gland, pancreas, inguinal, axillary, mesenteric, and mediastinal lymph nodes.

# Virus Quantitation

We used a one-step real-time quantitative RT-PCR to measure viral RNA in the samples. RNA was extracted from swabs using the QiaAmp Viral RNA extraction kit and tissues using the RNeasy kit (both from Qiagen). RNA was then used along with a MERS-CoV-specific primer/probe set using the Rotor-Gene Probe kit (Qiagen). Tissue culture infectious dose 50% (TCID50) equivalents were calculated by comparing cycle threshold values to a standard curve generated from virus stocks of known titer. Primers and probe sequences were described previously (16).

# Histopathology

Tissues were fixed in 10% neutral buffered formalin with two changes, for a minimum of 7 days and processed with a Sakura VIP-5 Tissue Tek, on a 12 h automated schedule, using a graded series of ethanol, xylene, and ParaPlast Extra. Embedded tissues were then sectioned at 5 µm and dried overnight at 42°C prior to staining.

For immunohistochemistry (IHC), tissues were processed using the Discovery XT automated processor with a DAPMap kit (both from Ventana). Specific primary antibodies used were: anti-HCoV-EMC polyclonal rabbit antibody (17) against CoV at a 1:1,000 dilution, anti-CD3 (2GV6) rabbit monoclonal primary antibody applied neat, and anti-CD20 (Thermo Scientific) at a 1:100 dilution. For CD3 and CD20, IHC stained sections were scanned with an Aperio ScanScope XT (Aperio Technologies, Inc., Vista, CA, USA) and analyzed using the ImageScope Positive Pixel Count algorithm (version 9.1). Approximately 25 mm squared were evaluated at 2× magnification. The default parameters of the Positive Pixel Count (hue of 0.1 and width of 0.5) were used.

# RESULTS

# Rhesus Macaque Immunosuppression

To assess the contribution of the immune response to protection from MERS disease, we immunosuppressed three rhesus macaques using CyP and Dex for 16 days prior to inoculation with 7 × 106 TCID50 of MERS-CoV. Two animals were used for mock immunosuppression controls and received the identical inoculum. Throughout immunosuppression, we monitored WBC populations in the blood to determine the efficacy of the immunosuppression regimen. Total WBC counts were decreased by approximately twofold compared to the control animals in response to CyP and Dex administration at the time of MERS-CoV inoculation. This reduction was due to decreases in all measured cell types (lymphocytes, neutrophils, monocytes, eosinophils, and basophils) (**Figure 1**). Following MERS-CoV inoculation, the absolute counts of these cell populations remained low in the immunosuppressed animals, suggesting their inability to respond to infection. Conversely, the two mock immunosuppressed animals had increased numbers of monocytes and eosinophils in response to infection. To quantify immunosuppression at the tissue level, the spleens and mediastinal lymph nodes of all animals were stained immunohistochemically with T cell and B cell markers (CD3 and CD20, respectively) post-mortem. The amount of staining was quantitatively assessed using imaging software. The quantity of CD3 staining was approximately 2-fold lower in the spleen and 2.7-fold in a mediastinal lymph node in the immunosuppressed animals. Likewise, CD20 was 2.4-fold lower in the spleen and 3-fold lower in the lymph nodes, showing a general reduction in lymphocytes in these tissues and confirming that the suppressive drug therapy was effective in reducing immune cell populations (**Figures 2A,B**). Immunosuppression also disrupted the normal architecture of these tissues.

# Virus Shedding and Replication

Oral and nasal swab samples were obtained daily throughout the course of MERS-CoV infection to monitor the shedding of viral RNA. While shedding was detected from all animals, detection of viral RNA occurred earlier, persisted longer, and was several logs higher, in the immunosuppressed animals compared to the control animals (**Figures 3A,B**). Viral RNA was detectable at 6 days post-inoculation in nasal swabs from all three immunosuppressed animals and oral swabs from two animals, whereas the control animals were negative by this time point.

To assess virus replication in the tissues, we enlisted three rhesus macaques from a previous study to serve as historic controls, along with the two controls in this study (11). All animals were given the same inoculum (from the same stock of virus) *via* the same route and all were euthanized 6 days post-inoculation. When comparing the viral abundance in the lungs (all six lobes) between the controls from this study and the historic controls,

there was no significant difference in the geometric means of viral RNA abundance between these two groups, although the historic controls had slightly more measurable viral RNA (**Figure 3C**). However, the immunosuppressed animals had

Values shown are the percent of positive (red) staining within a tissue and are the average for all animals within each group.

significantly increased MERS-CoV replication (measured by RNA abundance) in the lungs (Kolmogorov–Smirnov test, *p* ≤ 0.001). Similarly, there was significantly more virus detected in several respiratory, or respiratory tract-associated

confidence interval from each group. qRT-PCR was performed on an additional three animals from a previous macaque study (CoV historic) to compare to the immunosuppressed animals. Tissues from the respiratory tracts were also analyzed for RNA levels (D) and the samples from the historic control macaques were grouped with the two control animals from this study (black circles). Data are represented as the geometric means of the groups. To compare the data sets within (C,D), a Kolmogorov–Smirnov test was performed to determine statistical differences between the control and immunosuppressed groups. Asterisks (\*) indicate significant differences, with \**p* ≤ 0.05 and \*\*\**p* ≤ 0.001.

tissues, including the bronchi, trachea, tonsils, and mediastinal lymph nodes of immunosuppressed animals, compared to the controls (**Figure 3D**) (Kolmogorov–Smirnov test, *p* ≤ 0.05). When assessing viral dissemination in tissues outside of the respiratory tract, immunosuppressed animals were positive for low levels of viral RNA in several tissues, including the liver and spleen, as well as several lymph nodes, whereas virus was undetectable outside of the respiratory tract in the control animals, with the exception of one inguinal lymph node sample (**Table 1**).

# Lung Histopathology

Samples from all animals were evaluated for the presence of histopathologic changes. Each of the animals, with the exception of IS-CoV3, developed some degree of pulmonary pathology upon examination of tissue following necropsy 6 days after inoculation (**Figure 4A**). Lesions were characterized as multifocal, mild-tomarked, interstitial pneumonia and were frequently centered on terminal bronchioles. The pneumonia was characterized by thickening of alveolar septae by congestion, edema and fibrin,



*(–) indicates that no viral RNA was detected, (*+*) indicates that viral RNA was detected by qRT-PCR. LN, lymph node.*

and small to moderate numbers of macrophages and neutrophils. Alveoli contained moderate numbers of pulmonary macrophages and neutrophils. In lungs with marked changes, there was abundant alveolar edema and fibrin with the formation of hyaline membranes. Multifocal type II pneumocyte hyperplasia

was noted and there were also perivascular infiltrates of inflammatory cells within, and adjacent to, affected areas of the lung. Samples from each lung lobe for each animal were individually scored for the presence and extent of pathologic changes, with scores ranging from 0 (no pathology) to 4 (multiple coalescing inflammatory foci with fibrin and edema). Animals CoV1 and CoV2 had average histology scores of 1.3 and 1.2, with individual lung lobe scores ranging from 0 to 4 and from 0 to 3, respectively (**Table 2**). Immunosuppressed animals displayed much milder pathology with average scores for IS-CoV1, IS-CoV2, and IS-CoV3 of 0.2, 0.6, and 0, respectively, and no lobe showing a score greater than 1.

The animals that did not undergo immunosuppression developed the most severe pulmonary pathology, but demonstrated little or no viral antigen in the lung tissue examined by IHC (**Figure 4B**), reflecting the qRT-PCR results, where much less viral RNA was detected and viral RNA was undetectable in many of the individual lung lobes. Conversely, macaques that had undergone immunosuppression had very mild lung lesions, but demonstrated MERS-CoV viral antigen multifocally throughout the lung; predominantly within type I pneumocytes. This suggests that pulmonary pathology associated with MERS-CoV in these animals may be tightly associated with the immune response.



*(0)* = *no pathology, (1)* = *few inflammatory foci scattered between multiple lung lobes; alveolar interstitium minimally thickened by congestion and small numbers of neutrophils and macrophages; few neutrophils and macrophages within alveoli, (2)* = *multiple inflammatory foci in multiple lung lobes; alveolar interstitium is mildly thickened, edema and small numbers of neutrophils and macrophages; small numbers of neutrophils and macrophages within alveoli, (3)* = *multiple inflammatory foci scattered within single lung lobes; alveolar interstitium is moderately thickened, edema and moderate numbers of neutrophils and macrophages; many neutrophils and macrophages within alveoli; small amounts of fibrin and edema in alveoli, (4)* = *multiple to coalescing inflammatory foci within a single lung lobe; alveolar interstitium is markedly thickened by congestion, edema, fibrin, and large numbers of neutrophils and macrophages; large numbers of neutrophils, macrophages, cellular debris, fibrin, and edema within alveoli.*

# DISCUSSION

Little is known regarding how several emerging zoonotic viruses infecting the respiratory tract cause disease, and what risk factors contribute to poor outcome. Some viruses are thought to cause disease by dysregulating the immune response, whereby destruction of infected cells or secretion of pro-inflammatory mediators leads to immunopathology, as in the case of hantaviruses (18). Conversely, pathogenesis caused by some viruses correlates with deficiencies, or inefficiencies of the immune response, such as in the case of pathogenic viruses affecting the very young and elderly, or immune compromised individuals, as in the case of influenza virus (19). The mechanisms by which the recently emerged MERS-CoV causes disease in humans, and what host factors are associated with either resistance or a poor outcome are not known (15). These questions are important for the development of countermeasures that either directly target the virus to inhibit its replication or modulate the immune response to limit immunopathogenesis.

Early epidemiologic studies of MERS suggested that patients with comorbidities fared worse than healthy patients upon infection, and the number of comorbidities correlated with a worse outcome (20). In addition, a few patients with lethal infections were reported to be immunosuppressed (21–23). These reports were from cases where diagnoses were primarily in patients already in hospitals, including a relatively large number of nosocomial infections affecting 23 patients in a hospital outbreak in Al-Hasa (24). Since these initial case reports, more than 2,000 additional cases of MERS-CoV infection have been confirmed (2). Although many of these new cases are reported to be health care related, either stemming from patients or health care workers, it is unclear how many of these cases involve immunocompromised individuals.

Risk factors that have been associated with disease (or infection) include weakened immune systems and chronic diseases, such as diabetes, cancer, and chronic lung disease, as well as co-infections (5, 20, 25). Although these comorbidities clearly affect the status of the immune response, acute immunosuppression using drugs, as we have done here, provides a more controlled deficit in immune responses, whereas immune dysregulation brought on by chronic disease, infection, and aging is a complex phenomenon that involves deficiencies in the immune response, chronic inflammatory responses, and other known and yet to be described complex changes. For instance, the majority of patients had diabetes as a comorbidity and several others were immunosuppressed with HIV/AIDS. Both of these conditions alter the immune response in a way that both inhibits normal T cell functions, as well as inducing an inflammatory response by altering Th17 responses and secretion of inflammatory cytokines (26–30). The immunosuppression in our study mimics some aspects of the human condition in these patients, such as inhibition of CD4<sup>+</sup> T cell responses by HIV-infected patients. However, the chemical immunosuppression used herein is unlikely to mimic the chronic inflammatory state in many of these patients. This higher basal level of immune activation associated with these conditions may be important contributions to the manifestation of the clinically overt serious disease following MERS-CoV infection, and would imply that the immune system plays a role in the pathogenesis of MERS-CoV. This agrees with our observation that upon simple immunosuppression, MERS-CoV replicated to higher levels and showed greater dissemination and shedding, while the pathology was actually reduced in these animals. Pathology was likely lessened due to the absence of inflammatory cells and mediators, as observed histologically in the lung tissues. This suggests that the virus itself might cause little damage to the cells that it infects and this would lead toward a mechanism in which the absence of an efficient immune response allows the virus to replicate to high levels, whereas pathology can be attributed to the overactive inflammatory response, which patients with comorbidities are prone to possess. This is supported by data in the resus and marmoset animal models, which show that increased viral replication and the local immune response to this plays an important role in the pulmonary severity of disease (31). Although not performed in this study, a control group treated with immunosuppressive drugs, and not challenged, would be necessary for a comprehensive picture of the immune status of these animals at the time of necropsy.

Recent experiments using human-derived blood cells have shown that infection with MERS-CoV results in a dramatic increase in the production of cytokines and immune cellrecruiting chemokines and the authors hypothesize that these inflammatory responses could lead to severe inflammation and tissue damage (32, 33). This is supported by the observation that bronchoalveolar lavage fluid of humans infected with MERS-CoV contain high numbers of neutrophils and macrophages (1, 22). Furthermore, lymphopenia has been associated with disease and is potentially caused by infiltration of lymphocytes in the lung tissue and egress from the blood (34). Taking these findings into account, we can envision a model in which infection of the lung tissue and resident immune cells, such as alveolar macrophages, leads to the hyper-production of inflammatory cytokines and immune cell recruitment chemokines, which together limit virus replication, but result in an immunopathologic state. Upon immunosuppression of our macaques, the virus was still able to infect and replicate in the lung tissue, and likely induced local cytokine and chemokine expression; however, the depletion of immune cell populations upon chemical immunosuppression inhibited recruitment of inflammatory cells to the lungs (or infected tissues) and, thus, limited pathology. This is the first direct experimental evidence showing that MERS-CoV has an immunopathogenic component. This is in line with the observation in one patient in South Korea, which was taking prolonged high-dose corticosteroid therapy to control lymphoma activity and hemolytic anemia and displayed persistant viral shedding without clinical progression of the disease (23).

The shedding of MERS-CoV was more extensive in the immunosuppressed animals, both in duration as in peak shedding. This suggests that the immune status has direct influence on virus shedding and subsequent potential of transmission. The epidemiological analyses of the 2015 MERS-CoV outbreak in South Korea clearly showed that only the level of MERS-CoV shedding was directly associated with transmission potential. Where spreaders had statistically lower Ct values compared to non-spreaders (25). The persistent MERS-CoV shedding in immunocompromised patients (23) could, therefore, contribute to enhanced nosocomial transmission.

As of yet, no specific treatment options have been identified for MERS-CoV infection. The results presented herein show that inflammatory responses contribute to the pathogenic process. This would suggest that treatment for patients with symptomatic infections would benefit from additional therapy that lessens the inflammatory response, especially in the lung, and not be based solely on therapies that are aimed at controlling virus replication.

# ETHICS STATEMENT

The use of study animals was approved by the Institutional Animal Care and Use Committee of the Rocky Mountain Laboratories and experiments were performed following the

# REFERENCES


guidelines of the Association for Assessment and Accreditation of the Laboratory Animal Care by certified staff in an approved facility. The guidelines and basic principles in the United States Public Health Service Policy on Humane Care and Use of Laboratory Animals and the Guide for the Care and Use of Laboratory Animals were followed. All procedures were carried out under anesthesia using Ketamine by trained personnel under veterinarian supervision and efforts were made to provide for the welfare of animals and to minimize suffering. All animals were humanely euthanized at the endpoint of the study (6 days post-inoculation) by exsanguination under deep anesthesia. All standard operating procedures for MERS-CoV were approved by the Institutional Biosafety committee of the Rock Mountain Laboratories, and sample inactivation was carried out according to approved standard operating procedures prior to removal from high containment.

# AUTHOR CONTRIBUTIONS

JP, DF, EW, HF, and VM designed the study; JP, DS, HF, and VM analyzed the data; JP, HF, and VM wrote the manuscript; and JP, KH, FF, EH, VM, and DS performed the experiments and assays.

# ACKNOWLEDGMENTS

We would like to thank Doug Brining for veterinarian services; Dan Long, Tina Thomas, and Rebecca Rosenke for histology support; and Anita Mora for graphics. We would also like to thank Drs. Bart Haagmans and Ron Fouchier (Erasmus MC, Netherlands) for providing the MERS-CoV isolate.

# FUNDING

This work was supported by the Intramural Research Program of the National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH).


marmoset. *PLoS Pathog* (2014) 10:e1004250. doi:10.1371/journal.ppat. 1004250


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Prescott, Falzarano, de Wit, Hardcastle, Feldmann, Haddock, Scott, Feldmann and Munster. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Flavivirus Receptors: Diversity, Identity, and Cell Entry

Mathilde Laureti 1,2†, Divya Narayanan1,2†, Julio Rodriguez-Andres 1,2†, John K. Fazakerley 1,2 and Lukasz Kedzierski 1,2 \*

*<sup>1</sup> Department of Microbiology and Immunology, Peter Doherty Institute for Infection and Immunity, University of Melbourne, Melbourne, VIC, Australia, <sup>2</sup> Faculty of Veterinary and Agricultural Sciences, University of Melbourne, Melbourne, VIC, Australia*

Flaviviruses are emerging and re-emerging arthropod-borne pathogens responsible for significant mortality and morbidity worldwide. The genus comprises more than seventy small, positive-sense, single-stranded RNA viruses, which are responsible for a spectrum of human and animal diseases ranging in symptoms from mild, influenza-like infection to fatal encephalitis and haemorrhagic fever. Despite genomic and structural similarities across the genus, infections by different flaviviruses result in disparate clinical presentations. This review focusses on two haemorrhagic flaviviruses, dengue virus and yellow fever virus, and two neurotropic flaviviruses, Japanese encephalitis virus and Zika virus. We review current knowledge on host-pathogen interactions, virus entry strategies and tropism.

#### Edited by:

*Alan Chen-Yu Hsu, University of Newcastle, Australia*

#### Reviewed by: *Alec Jay Hirsch,*

*Oregon Health & Science University, United States Namal P. M. Liyanage, The Ohio State University, United States*

\*Correspondence:

*Lukasz Kedzierski lukaszk@unimelb.edu.au*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Viral Immunology, a section of the journal Frontiers in Immunology*

Received: *12 May 2018* Accepted: *04 September 2018* Published: *26 September 2018*

#### Citation:

*Laureti M, Narayanan D, Rodriguez-Andres J, Fazakerley JK and Kedzierski L (2018) Flavivirus Receptors: Diversity, Identity, and Cell Entry. Front. Immunol. 9:2180. doi: 10.3389/fimmu.2018.02180* Keywords: flaviviruses, Japanese encephalitis virus, Zika virus (ZIKV), dengue virus, yellow fever virus, entry receptor

# INTRODUCTION

The Flavivirus genus consists of more than 70 small, positive-sense, single-stranded RNA viruses transmitted by arthropods, in particular mosquitos and ticks. These include globally important human pathogens such as West Nile virus (WNV), Japanese encephalitis virus (JEV), dengue virus (DENV), Murray Valley encephalitis virus (MVE), tick-borne encephalitis virus (TBEV), Yellow Fever virus (YFV), and Zika virus (ZIKV). These viruses are responsible for some of the most severe arbovirus infections affecting humans, pose a serious threat to global health and have the potential to cause severe outbreaks. These are exemplified by the global distribution of DENV (1), the recent ZIKV outbreak in South America (2), YFV outbreaks in Africa (3), and Brazil (4) and the spread of WNV across North America (5). Flavivirus infections range from asymptomatic, through mild fever and arthralgia to life threatening haemorrhagic or encephalitic diseases (6). Flaviviruses are also able to persist in patients and can be responsible for long-term morbidities (7). There are no antiviral treatments for flavivirus infection currently in clinical use, and despite licensed vaccines against several of the viruses including YFV, JEV, TBEV, or DENV, outbreaks still occur highlighting challenges in implementing effective vaccination programs (8).

The flaviviral genome of ∼11 kb contains a single open reading frame flanked by untranslated regions, and encodes 3 structural proteins (C, M, and E) and 7 non-structural proteins (NS). The mature virion features a surface densely covered with E glycoproteins and M proteins and a core consisting of capsid (C) protein and the RNA genome (9, 10). The entry into the target cell is dependent on E protein contact with its cognate receptor. E protein initially binds to attachment factors such as glycosaminoglycans. This effectively increases viral density on the cell surface, leading to high affinity receptor binding (11). The E protein ectodomain consists of three domains (E-DI, E-DII, E-DIII) of which E-DIII

**163**

is thought to interact with attachment factors and receptors (12). E-DIII domain's importance is highlighted by the fact that a vast majority of potent, neutralizing antibodies has been mapped to this region. Nevertheless, anti-DI and DII antibodies, although less potent, show broader cross-reactivity and form a major pool of anti-E protein specific immonoglobulins (13). Receptor binding is followed by clathrin-mediated endocytosis (14), which is considered to be a major mechanism of flavivirus cell entry, although there are exceptions described below. This leads to formation of endosomes and low pH dependent changes in the E glycoprotein with subsequent membrane fusion and release of nucleocapsid into the cytosol (15). In vitro, flaviviruses are able to infect a plethora of cell lines originating from rodents, non-human primates, humans and mosquitos. However in vivo, fewer cell types seem able to support flavivirus replication (16). A wide range of cell surface receptors has been implicated in flavivirus entry into different cells types (11). Amongst the entry receptors postulated to be involved in flavivirus entry, the best characterized to date include αvβ<sup>3</sup> integrins (17, 18), C-type lectin receptors (CLR) (19–23), phosphatidylserine receptors TIM (Tcell immunoglobulin and mucin domain) and TYRO3, AXL and MER (TAM) (24). Recent studies indicate that flaviviruses can produce a range of structurally different virions. This structural heterogeneity may expand tissue tropism and ability to infect different cell types both in invertebrate and vertebrate hosts (25).

Flaviviruses are deposited into the skin epidermis by a mosquito bite where they encounter cells permissive to infection such as keratinocytes and skin dendritic cells (Langerhans cells) (26). Dendritic cells in particular appear to be a common initial target for flaviviruses. When infected, dendritic cells migrate to lymphoid organs where viral replication takes place allowing for flavivirus dissemination into circulation and internal organs (12). Viruses such DENV (27), JEV (28), ZIKV (29) have been shown to infect skin dendritic cells, and although there are no reports on YFV infecting Langerhans cells, it can nevertheless infect other types of dendritic cells (30). This interaction is mediated by DC-SIGN for JEV (28) and ZIKV (29), but appears to be DC-SIGN independent in case of DENV (31) and YFV (30).

Many flaviviruses are neuroinvasive and neurovirulent and cause central nervous system (CNS) damage (32). Neuroinvasive infections are observed with JEV, TBEV, and WNV (33, 34), and occasionally with haemorrhagic viruses including DENV (35). There is a paucity of knowledge regarding factors involved in CNS cell entry. While CLRs and TIMs are expressed by cells of the CNS (36–38), they are not expressed by neurons (39–41). However, members of the TAM family of receptors are expressed by different neuronal subtypes (42), though they are dispensable for ZIKV infection as ZIKV was able to infect and replicate in TAM receptor knockout mice (43).

As natural vectors, mosquitos and ticks are highly permissive to flavivirus infection. The virus can replicate in a range of arthropod tissues and cells (44, 45). Given that flaviviruses have only one glycoprotein, it seems likely that the mechanism of entry into vertebrate and invertebrate cells is evolutionarily conserved. A number of the cellular receptors implicated in flavivirus entry into mosquito cell lines overlaps with those identified for mammalian cells (46). Some flaviviruses are more selective regarding their arthropod host than others. For example, DENV is spread mainly by Aedes spp. mosquitos (6), WNV by Culex spp. (47), whereas JEV is transmitted by Aedes, Anopheles and Culex spp. (48). There appears to be a more restricted receptor repertoire used by flaviviruses for insect cell entry compared to mammalian cell entry. The range of clinical manifestations of flaviviral infection in the mammalian host suggests that these viruses may use a wide range of receptors. Mammalian tissues in general offer much greater range of receptors compared to invertebrates.

Identification of flavivirus entry receptors, particularly those involved in CNS infection, could lead to identification of novel therapeutic targets. For this review we will focus on four major flaviviruses of humans—DENV, JEV, ZIKV, and YFV, and discuss the differences and similarities in their mechanisms of entry into arthropod and mammalian cells.

# DENGUE VIRUS

DENV is one of the most common mosquito-borne viruses, mainly transmitted by Aedes aegypti mosquitoes, and occasionally by Ae. albopictus. Symptoms of DENV infection range from fever and muscle and joint pain (Dengue fever) to potentially life threatening haemorrhagic fever or shock syndrome. While DENV was endemic in <10 countries in the 1970's, it is presently a threat in over 128 countries and is responsible for almost 400 million human infections every year. About 24% of infections manifest in severe clinical symptoms. There is currently no treatment for DENV serotypes. There are four virus serotypes, and recovery from one serotype provides lifelong homologous immunity (49).

DENV is an icosahedral particle of 50 nm with a positive, single-stranded RNA genome of 10–11 kb (50). As in other flaviviruses, E protein is involved in receptor binding and fusion (51) and has the ability to bind to a wide range of cellular receptors to initiate DENV entry. The E-DIII domain has a role in cellular recognition (52) and has been suggested as a target for the development of a DENV vaccine (53).

Over the years, several cell membrane receptors involved in DENV entry have been identified. These include carbohydrate molecules (54–56), lectins (57, 58), and claudin-1 cell receptors (59). Carbohydrate molecules such as glycosaminoglycans (GAGs), sulphated polysaccharides, and glycosphingolipids (GSL) are widely expressed cell surface co-receptors for DENV entry and are believed to enhance viral entry efficiency. The highly sulphated form of GAGs, the heparan sulfates (HS) and heparan sulfates proteoglycans (HSPG), are essential for cellular adhesion to extracellular matrix and binding of polypeptide growth factors involved in intracellular signaling (56). It has been suggested that DENV first contacts HSPG, and that this weak interaction facilitates binding of virus to other receptors, which then results in virus internalization (55). Several studies have shown that pre-treatment with heparin can reduce or block DENV-2 infection (60, 61). However, the efficiency of inhibition of viral entry was dependent on numerous factors, such as the virus strains and the target cell (61). GSLs, a member of the same family of carbohydrate molecules as HS, are ubiquitous cellular components of eukaryotic plasma membranes that can also facilitate entry and binding of virus (54). However, GSLs are not required for DENV entry as the virus was able to enter GSL-deficient cells (62).

Cellular C-type lectin receptors (CLRs) are part of the host immune response to fungal, bacterial and viral infections (57). CLRs in mammalian cells include DC-SIGN/L-SIGN, mannose receptors (MR), and CLEC5A. DC-SIGN receptors are widely known because of their association with HIV. These receptors are also involved in DENV binding and internalization into dendritic cells (63). MR has been found to be the primary DENV cell receptor in macrophages. The CLEC5A receptor cooperates with DC-SIGN or MR to increase DENV binding and stability (58).

Other studies have suggested claudin-1 as a putative cell receptor for DENV entry through a direct interaction with the viral prM protein. Claudins are vital components of tight junction complexes and are essential for normal permeability of the epithelia (59, 64). DENV-2 entry was significantly reduced in claudin-1 deficient cells (59). Also, it has been demonstrated that caudin-1 is upregulated early in infection in order to facilitate entry and downregulated in late stage of infection (64).

Protein binding assays and mass spectrometry analysis have identified several additional potential flavivirus cellular receptors (65–67). Among them, the tubulin and tubulin-like proteins in C6/36 Ae. albopictus cell line (65). Heat shock proteins (HSPs) of ∼70 kDa and 80 kDa were also identified as cellular receptors for all four DENV serotypes in C6/36 cell line (66, 68). HSPs are chaperone proteins involved in the regulation of folding and unfolding of cellular, and upon infection, viral proteins (69). The 70 kDa protein, also known as heat shock cognate protein (HSC70) or HSPA8, acts as a chaperone protein during DENV entry (70, 71). Modulation of HSC70 expression was observed during DENV-2 infection, with an increase on the cell membrane during infection, suggesting that DENV-2 utilizes HSC70 for entry into mosquito cells (67). In addition to its role in viral entry, HSP70 is involved in virion biogenesis and RNA replication (71). It appears that all four DENV serotypes are dependent on this chaperone protein family, which makes HSP70 proteins an interesting target for the design of a tetravalent DENV therapy or vaccine (71).

HSP90, another heat shock protein, can also act as chaperone. This protein interacts with six DENV proteins (69). While the involvement of HSP70 and HSP90 in DENV binding to host cells has been reported (70–72), these proteins are not involved in internalization of virus into the host cell (73).

As mentioned before, TIM/TAM family receptors have been implicated in flavivirus entry. DENV express on its surface phosphatidylserine (PS) and phosphatidylethanolamine (PE) molecules. Both PS and PE are known to directly interact with TIM/TAM receptors and DENV is able to enhance its entry by exploiting these interactions (74).

After binding to cellular receptors, internalization of viral particles occurs. For DENV, internalization occurs via pHdependent endocytosis. Several endocytosis pathways are currently known, but clathrin-mediated endocytosis is the main pathway for DENV (75). The DENV use of clathrin-mediated endocytosis was demonstrated in C6/36 mosquito cells by biochemical inhibition of cell receptors (76), and in several human cell lines through siRNA silencing of genes associated with clathrin-mediated receptors (75, 77, 78). While this inhibition and specific gene silencing resulted in a decrease in viral load, a complete inhibition was not achieved, suggesting the existence of alternative entry pathways in mosquito and mammalian cells.

In addition to the exploitation of clathrin-mediated endocytosis, the host immune system can also promote viral entry (79). This phenomenon, known as antibody-dependent enhancement (ADE), was first described in 1964 by Hawkes (80) for WNV and JEV, and observed for DENV more than a decade later (81). Antibody-virus complexes are internalized by phagocytosis via Fc gamma receptors (FcγR) into macrophages, monocytes and dendritic cells (82) (**Figure 1**), thus facilitating virus entry (83). It has recently been shown that ADE increases membrane fusion activity, promoting DENV entry (79). Moreover, prM antibodies have the capacity to convert noninfectious, immature DENV particles into infectious particles and enhance their infectivity to levels comparable to wild-type

virus (50). ADE has been linked to the observation that one flavivirus infection can enhance another (84). However, a recent study showed that ADE is dependent on the level of neutralizing antibodies, particularly IgG and IgM (85); only patients with a low level of neutralizing antibodies showed enhancement of DENV infection (85). Antibodies, even at low concentration, against the EDIII domain were able to block viral entry of the four DENV serotypes without inducing antibody-dependent enhancement (86). However, high IgG titres were observed in patients with ADE following DENV infection, in particular IgG1 levels were the highest in patients with dengue fever or shock syndrome (87). ADE has been recently identified as consequence of sensitisation with Dengvaxia quadrivalent vaccine, leading to severe vaccine-enhanced disease resulting in hospitalization (88).

# JAPANESE ENCEPHALITIS VIRUS

Globally, Japanese encephalitis is the most clinically important arboviral encephalitis, with an estimated annual prevalence of up to 50,000 cases (89). As is the case with most arboviral encephalitic infections, humans are dead-end hosts unable to develop a sufficiently high viremia to transmit to feeding mosquitos. The majority of JEV infections are asymptomatic. Approximately a third of clinical cases are fatal and half of survivors have neurological or neuropsychiatric sequelae with symptoms resembling Parkinsonian movement disorders, poliomyelitis-like paralysis or impaired cognition (90). Disease is most common in children up to 14 years of age. JEV has been expanding its endemic areas in Asia (91) and poses an unpredictable and emerging global threat.

In humans, JEV has been found in different anatomical compartments and a variety of cell types is able to support its replication. These include endothelial cells, granulocytes, dendritic cells, macrophages and cells of the CNS including astrocytes, neurons and microglia (92). The virus spreads from dermal tissues (93) to lymphoid organs (94) and during the acute stage of infection can be found in blood (95). Although highly neuroinvasive, the mechanism of JEV entry into the CNS is unclear. Transport along the olfactory nerve and across the blood brain barrier have been implicated in JEV invasion of the CNS (96, 97). Studies in rodent models indicate that the blood brain barrier is disrupted following neuroinvasion (98), and might be a consequence of invasion rather than an entry route. Once in the brain, JEV can infect pericytes (99), astrocytes (100) and microglia (101), and has a predilection for developing neurons and neuronal progenitors (102, 103). As described below, a number of receptors mediate entry into different cells types. The distinctive neuronal tropism suggests the existence of JEV-specific receptors in the CNS, but their nature remains elusive (104).

In vitro studies on mouse neuroblastoma cells indicate heat shock protein (HSP) 70 as a putative entry receptor present on neuronal cells (105). This has not been corroborated by in vivo experiments, but in human hepatoma Huh7 cells, HSP70 is required for entry (106). Recently, a member of the HSP70 family, glucose-regulated protein (GRP) 78, has been implicated in JEV entry into Neuro2a and BHK-21 cells (107, 108). In addition to HSP70 and GRP78, HSP90β also interacts with E protein and may be used by JEV to enter mammalian cells (109). Another member of the HSP70 family is heat shock cognate (HSC) protein 70. HSC70 has been suggested to be a receptor for entry into mosquito cells (110). HSC70 isoform D is essential for clathrin-dependent endocytosis of JEV into C6/36 cells (111). Clathrin dependence seems to be critical for JEV entry into mammalian cells with the exception of neuronal cells (112), where JEV internalization into rodent neuroblastoma cell lines has been shown to be clathrin-independent (113, 114) and independent of HSP70 family proteins. JEV can enter human neuronal cells by caveolin-mediated endocytosis (115), a process that is receptor-independent (116). Interestingly, JEV has been shown to utilize the dopaminergic signal transduction pathway to increase neuronal susceptibility to infection (117). Infection of human dopaminergic neuroblastoma cells in vitro leads to increased levels of secreted dopamine and activation of the phospholipase C cascade. The latter enhances formation of structures known as focal adhesions on the cell surface and increases JEV binding and entry. One of the main components of focal adhesions is αvβ3 integrin that recruits vimentin to the cell surface (118), and is involved in JEV binding and infection of BHK-21 cells (18). Vimentin is a putative JEV receptor (119, 120). Thus, by signaling through dopamine D2 receptors and activating the phospholipase C cascade, JEV induces recruitment of surface molecules that enhance and propagate infection in adjacent cells. Enhanced infection of dopaminergic neurons also explains why JEV is predominantly found in brain areas rich in these cells including the thalamus and the midbrain (121, 122). Whereas JEV infection of neurons may be most directly relevant for disease, other cells types are also likely to have an important role in the disease process. Microglial cells may be a viral reservoir due to long-term, high level of virus production in these cells (123). CD4 has been identified as a major receptor for JEV entry into microglia (124). Presumably, CD4 can be used by JEV to enter other CD4 positive cells such as T cells, macrophages or dendritic cells. Published data are scarce, however, JEV productive infection of splenic macrophages and T cells has been reported in a mouse model of infection (125). T lymphocytes have also been reported as a reservoir of latent JEV in asymptomatic children following recovery from acute infection (126). The involvement of CD4 in microglial cell entry has not been reported for any other flavivirus, however CD4 is the main receptor for retroviral entry and is primarily localized in lipid rafts (127). As mentioned above, HSP70 in lipid rafts is involved in JEV entry into human cells and in general lipid rafts play a critical role in JEV entry (128, 129). Moreover, lipid rafts, as well as clathrin-coated pits and caveolae, contain sphingolipids such as sphingomyelin (SM) that is involved in JEV attachment and entry (130). Studies in SM synthase 1 deficient mice infected with JEV showed a reduction in disease, indicating a role for SM in JEV infection models (130).

Despite advances in identification of new receptors associated with JEV entry and its clear tropism for neuronal cells in the CNS, the identity of a specific neuronal receptor remains elusive. Notably, JEV has the ability to infect cells in the absence of above mentioned putative receptors although at a reduced rate (104). This suggests that the entry process involves multi protein interactions with high degree of redundancy and a single, specific entry receptor might not exist. Alternatively, the inability to identify such a receptor highlights the limitations of in vitro systems commonly used to investigate virus-cell interactions.

# ZIKA VIRUS

ZIKV is a mosquito-borne emerging pathogen that poses significant public health concerns due to recent rapidly expanding outbreaks. ZIKV was relatively unknown until 2007, when an outbreak occurred in Yap Island (Micronesia) (131). The virus was first isolated in the Zika Forest in Uganda from a rhesus monkey in 1947. In 1948 a second isolate from Ae. africanus mosquitoes was obtained from the same forest (132). Prior to the recent serious outbreak in French Polynesia, New Caledonia, the Cook Islands and Easter Island in 2013 and 2014 (133), Zika has not been reported to cause significant disease. Data from French Polynesia during the ZIKV epidemic documented the occurrence of Guillain-Barre syndrome and other neurological complications (134). The pathogenesis of ZIKV infection is poorly understood and involves a multifaceted interaction between viral and host factors. ZIKV has shown a significant tropism to the CNS and causes neurodegeneration, particularly of neural progenitor cells (135–137). ZIKV is also the only flavivirus known to have teratogenic effects in humans, including microcephaly, intracranial calcification and fetal death (138). As a result, the World Health Organization announced in 2016 that the ZIKV outbreak was a health emergency of international concern (139). Like other flaviviruses, ZIKV likely enters host cells through endocytosis instigated by an interaction of E glycoprotein with cell surface receptors. Identification of the entry receptor(s) for ZIKV is essential to understanding viral tropism and pathogenesis, and could lead to the development of novel therapeutics to treat the infection.

The first barrier for the virus to enter the host cell is the skin epidermis. ZIKV is transmitted by Aedes spp. mosquitoes, which deposit virus in the epidermis and dermis during the blood meal. Both dermal fibroblast and epidermal keratinocytes are permissive to ZIKV infection as are skin dendritic cells. Several entry receptors including the innate immune receptor DC-SIGN, transmembrane protein TIM-1 and TAM receptors (TYRO3, AXL, MER), have been shown to facilitate entry and enhance ZIKV infection (29). RNA silencing of TIM-1 and AXL in subsets of human skin cells showed a significant reduction in ZIKV titre in AXL knockdown, and in double AXL and TIM-1 knockdown, indicating that AXL is a major receptor for ZIKV entry at least in human skin cells. However, a recent study (43) investigating different infection routes of ZIKV, including subcutaneous, transplacental, vaginal, and intracranial infections in wild-type and TAM receptor null mice, showed no difference in viral titres. TAM receptors, at least in mice, are therefore not essential for ZIKV infection. Interestingly, WNV infection of neurons can be enhanced in mice lacking AXL and MER. This increase in infectivity was associated with changes in blood brain barrier permeability (140), suggesting that AXL and MER do not serve exclusively as receptors and might have other roles in WNV infection of the brain.

To reach the fetal brain, ZIKV must first be transported to the fetal circulation, and cross the placental barrier. The placental barrier is composed of placental barrier cells, trophoblasts and fetal endothelial cells, which separate the fetal blood in capillaries from maternal blood. ZIKV has been reported in the amniotic fluid of fetuses in Brazil (141). This observation strengthened the association of ZIKV with microcephaly in neonates. Moreover, it has been shown that microcephaly caused by maternal viral infection in mice could result from direct viral infection of the fetus via the trans-placental route as well as from a placental inflammatory response that affects fetal development (142). ZIKV can efficiently infect fetal endothelial cells, whereas WNV and DENV do not, highlighting ZIKV unique tropism among flaviviruses (143). These differences between flaviviruses are due to ZIKV ability to efficiently use AXL receptor to enter fetal endothelial cells (143).

TIM-1 was also observed to have an important role in placental entry of ZIKV (144). ZIKV was able to infect different human primary placental cell types and explants from chorionic villi. AXL, TYRO3, and TIM-1 were present in the primary placental cells and are found at the uterine-placental interface. Particularly high expression of TIM-1 has been observed in cells where maternal blood perfused placenta including basal decidua and neighboring chorionic villi. Expression of AXL and TYRO3 varied with explant donor, gestational age and cell type. Specific pharmacological inhibition of TIM-1 by duramycin (145) could inhibit ZIKV infection at the uterine-placental interface, indicating that TIM-1 is a putative receptor for ZIKV placental cell entry. However, the role of AXL, variation in the expression of AXL and TYRO3 in pregnant women and whether TIM-1 is the sole receptor for ZIKV infection of the placenta need further study.

Numerous studies on ZIKV infection in mice having defective interferon signaling, including IFNα/β knockout mice (146, 147), double knockout of IFNα/β and IFNγ (148), and triple knockout of IRF-3,-5,-7 (136, 149) showed viremia, microcephaly and death in young mice and viremia with recovery in adult mice. However, cell death and reduced proliferation was observed for adult neural stem cells (136) suggesting possible long term effects in adult brain followin ZIKV infection. ZIKV also infects other cell types, especially in the eye. ZIKV-inoculated mice develop ocular defects including conjunctivitis, pan uveitis, and infection of the optic nerve, cornea, iris, and ganglion and bipolar cells in the retina (150). AXL is expressed at high levels in retinal progenitor cells (151) suggesting a possible role in ZIKV infection of ocular cells. However, the ocular abnormalities were shown to be independent of AXL or MER, given that AXL−/−, MER−/−, and AXL−/<sup>−</sup> MER−/<sup>−</sup> double knockout mice sustained levels of infection similar to those of control animals. Nevertheless, AXL might have a role in ZIKV infection of glial cells via Gas6 mediated activation of AXL kinase (152).

In vitro and in vivo systems to study ZIKV infection of neural cells have been developed. ZIKV neuro-infection models using cultured neural precursor cells (NPCs), cortical organoids, mouse brains, and human fetal brain materials have been studied (153). Microcephaly in these models is associated with inflammation, reduced proliferation of NPCs and neuronal cell death. ZIKV-BR can infect mouse fetuses and infection of pregnant mice also causes disease in embryos with intrauterine growth restriction, including signs of microcephaly (154). This study also demonstrated that ZIKV-BR infects human cortical progenitor cells, increasing the rate of cell death. ZIKV was found to directly infect human NPCs with high efficiency providing a plausible explanation for the observed developmental phenotypes and associated teratogenicity in the neonatal brains (137). Based on previous studies (29, 151, 152), AXL is a strong candidate receptor for the entry of ZIKV into cells of the developing brain. The potential role of AXL to facilitate ZIKV infection of the neonatal brain was explored by determining, at a single cell level, RNA expression profiles in the developing human cerebral cortex (151). The study revealed a higher expression of AXL in the radial cells and neural stem cells of the developing brain throughout neurogenesis and in capillaries and astrocytes. However, loss of AXL expression following CRISPR/Cas9 gene editing, did not affect ZIKV infectivity into hNPCs or cerebral organoids (155).

As is the case with JEV, the identity of the receptors involved in ZIKV receptor mediated endocytosis remains to be elucidated. There might be tissue specificity in receptor mediated viral entry, with variation in receptor repertoire in the skin, placenta, neurons, and other cell types. Alternatively, given ZIKV unique ability to cross the placenta and infect developing neurons in the fetal brain, there might be some as yet unidentified receptors facilitating this process.

# YELLOW FEVER VIRUS

YFV is the prototype and namesake virus of the Flavivirus genus; flavi means yellow in Latin. When infecting humans YFV replicates in liver, heart, kidneys, and lungs causing a broad spectrum of clinical symptoms. These vary from asymptomatic infection to renal and hepatic failures with severe haemorrhagic disease (156). A live attenuated YFV vaccine 17D was created over 70 years ago and has been used safely in over 500 million people. The parent strain of 17D is the virulent Asibi strain (157) isolated in Africa in 1927. 17D was passaged more than 230 times in mouse and chicken embryonic tissue. The adaptation of 17D to grow in tissue culture resulted in loss of viscerotropism, neurotropism, and mosquito tropism (156), making it an ideal candidate for a vaccine. The genome of both strains has been sequenced (158). The extensive passage history gave rise to 68 nucleotide mutations and 32 amino acid substitutions. Most of the genetic differences occur in the envelope (E) protein gene (157). Interestingly, the molecular determinants and mechanisms of this attenuation remain largely unknown. It has been suggested that the differences in the E protein and its involvement in cell entry are determinants of the difference in pathogenicity between the 17D and Asibi strains (159, 160). Mutations in the E gene have been suggested to allow the 17D strain to bind and enter hosts cells more efficiently.

YFV shares genome organization and entry by clarithinmediated endocytosis (CME) with most of the other flaviviruses. During YFV infection, the E protein binds to an unknown entry receptor that traffics the virion to endosomes. Similarly to other flaviviruses, increase in acidification of the endosome results in conformational changes in the E protein, membrane fusion and nucleocapsid release into the cytoplasm (156). The vaccine strain uses a clathrin- and caveolin-independent, but dynamin-2-dependent, pathway for infection (160). Dynamin-2 is a GTPase involved in cleaving off endocytic vesicles from the plasma membrane (161). The entry pathway of the 17D strain was further characterized as Rac1, Pak1, and cortactin independent (160). Clarithin-independent entry has been reported to mediate the internalization of a variety of viruses, such as rotavirus, human rhinovirus, influenza, and interestingly, JEV vaccine strain in neuronal cells (114, 162–164). Cells infected with 17D have been found to produce more viral RNA and INF-β, IL-29, ISG56, CCL5, and CXCL10 mRNA than those infected with the parental Asibi strain. In addition, 17D infected cells secrete INF-β, whereas cells infected with the Asibi strain do not. Virus entry through a clathrin-independent pathway allows for more efficient virion delivery into endosomes or protection from degradation, relative to entry via the classical clathrin-mediated route. This former entry route has been suggested to allow for a higher amount of viral RNA released into the cytoplasm (160). Viral RNA in the cytoplasm is detected by RIG-I, MDA5 and TLR7 (165), triggering strong innate immune responses. The Asibi strain on other hand, replicates at lower levels and inhibits the innate immune system. This difference in entry mechanism has been suggested to account for the differences in cytokine response between the two YFV strains, though further mutations in other proteins, such as NS2A could also be involved.

# CONCLUSIONS

Glycoprotein E is responsible for receptor-mediated attachment of flaviviruses to the host cell and membrane fusion. Although E protein of different flaviviruses share approximately 40% sequence identity (e.g., DENV and TBEV), their overall structural features are almost identical and this is assumed to apply to all flaviviruses (166). Cell entry is facilitated by a conserved peptide of 16 amino acids, located in E-DII region of the envelope glycoprotein (167). This conservation, coupled with highly organized conformational changes upon exposure to low pH (168), suggests evolutionary constraints allowing flaviviruses to enter both mammalian and arthropod cells. Yet, flavivirus receptors show diversity and significant cell type specificity. It is not unusual that a single molecule can bind to variety of targets as exemplified by immunoglobulins. However, their diversity and specificity are governed by V(D)J recombination, while the flaviviral glycoprotein E is conserved. The flavivirus infection is a consequence of multiple complex interactions between the virus and the target cell. It is clear that the flavivirus can exploit different endocytic routes that can be either clathrin or caveolae dependent or independent. The neurotropism of specific flaviviruses raises the question, is there a single specific neuronal receptor? What is the identity of this receptor and is the same receptor being used by all encephalitic flaviviruses? Another unresolved question is whether all flaviviruses share the same features of infection in the developing brain, or whether viruses such as microcephalycausing ZIKV, exhibit a different infection pattern. It is also relevant to note that the expression of entry receptors (e.g., CLRs or TAM) does not account for flavivirus tropism and cellular models lacking those receptors are still permissive to

# REFERENCES


infection. Identifying the relevant entry receptors is essential to deciphering the mechanisms of pathogenesis, tropism and viral biology. A better understanding of those processes will uncover new strategies for designing therapeutics and vaccines against flaviviruses.

# AUTHOR CONTRIBUTIONS

ML, DN, JR-A, and LK wrote sections of the manuscript, JF and LK edited and critically evaluated the manuscript.

for cellular attachment and infection. J Virol. (2006) 80:1290–301. doi: 10.1128/JVI.80.3.1290-1301.2006


regulation of actin reorganization. Future Microbiol. (2016) 1:1227–48. doi: 10.2217/fmb-2016-0002


alters proliferation. Cell Stem Cell (2016) 9:593–8. doi: 10.1016/j.stem.2016. 08.005


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Laureti, Narayanan, Rodriguez-Andres, Fazakerley and Kedzierski. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Cell-Mediated Immune Responses and Immunopathogenesis of Human Tick-Borne Encephalitis Virus-Infection

Kim Blom<sup>1</sup> \*, Angelica Cuapio<sup>1</sup> , J. Tyler Sandberg<sup>1</sup> , Renata Varnaite<sup>1</sup> , Jakob Michaëlsson<sup>1</sup> , Niklas K. Björkström<sup>1</sup> , Johan K. Sandberg<sup>1</sup> , Jonas Klingström<sup>1</sup> , Lars Lindquist 2,3, Sara Gredmark Russ 1,2 and Hans-Gustaf Ljunggren<sup>1</sup> \*

#### Edited by:

Alan Chen-Yu Hsu, University of Newcastle, Australia

#### Reviewed by:

Alessandro Marcello, International Centre for Genetic Engineering and Biotechnology, Italy Manuela Zlamy, Innsbruck Medical University, Austria

#### \*Correspondence:

Kim Blom kim.blom@ki.se Hans-Gustaf Ljunggren hans-gustaf.ljunggren@ki.se

#### Specialty section:

This article was submitted to Microbial Immunology, a section of the journal Frontiers in Immunology

Received: 27 March 2018 Accepted: 03 September 2018 Published: 26 September 2018

#### Citation:

Blom K, Cuapio A, Sandberg JT, Varnaite R, Michaëlsson J, Björkström NK, Sandberg JK, Klingström J, Lindquist L, Gredmark Russ S and Ljunggren H-G (2018) Cell-Mediated Immune Responses and Immunopathogenesis of Human Tick-Borne Encephalitis Virus-Infection. Front. Immunol. 9:2174. doi: 10.3389/fimmu.2018.02174 <sup>1</sup> Department of Medicine Huddinge, Center for Infectious Medicine, Karolinska Institutet, Karolinska University Hospital, Stockholm, Sweden, <sup>2</sup> Department of Infectious Diseases, Karolinska University Hospital, Stockholm, Sweden, <sup>3</sup> Unit of Infectious Diseases, Department of Medicine Huddinge, Karolinska Institutet, Karolinska University Hospital, Stockholm, Sweden

Tick-borne encephalitis virus (TBEV) is a flavivirus that belongs to the Flaviviridae family. TBEV is transmitted to humans primarily from infected ticks. The virus causes tickborne encephalitis (TBE), an acute viral disease that affects the central nervous system (CNS). Infection can lead to acute neurological symptoms of significant severity due to meningitis or meningo(myelo)encephalitis. TBE can cause long-term suffering and has been recognized as an increasing public health problem. TBEV-affected areas currently include large parts of central and northern Europe as well as northern Asia. Infection with TBEV triggers a humoral as well as a cell-mediated immune response. In contrast to the well-characterized humoral antibody-mediated response, the cell-mediated immune responses elicited to natural TBEV-infection have been poorly characterized until recently. Here, we review recent progress in our understanding of the cell-mediated immune response to human TBEV-infection. A particular emphasis is devoted to studies of the response mediated by natural killer (NK) cells and CD8 T cells. The studies described include results revealing the temporal dynamics of the T cell- as well as NK cell-responses in relation to disease state and functional characterization of these cells. Additionally, we discuss specific immunopathological aspects of TBEV-infection in the CNS.

Keywords: cell-mediated immunity, flavivirus, NK cells, T cells, tick-borne encephalitis, tick-borne encephalitis virus, viral immunopathogenesis

# INTRODUCTION

Tick-borne encephalitis virus (TBEV) is a flavivirus that belongs to the Flaviviridae family. Flaviviruses comprise many human pathogens including the commonly known Dengue virus (DENV), Japanese encephalitis virus (JEV), West Nile virus (WNV), Yellow fever virus (YFV), and Zika virus (ZIKV) (1). With respect to TBEV, three subtypes of the virus exist: European (TBEV-Eu), Siberian (TBEV-Sib), and Far Eastern (TBEV-FE) (2).

**174**

TBEV is transmitted to humans primarily from infected ticks, mainly from the Ixodes family. The virus can also be transmitted from unpasteurized dairy products from infected livestock (3–5). Infection with TBEV causes tick-borne encephalitis (TBE), an acute viral infection that affects the central nervous system (CNS) with often severe long-term neurological consequences (3, 4, 6, 7). The first TBE-like disease was described as early as in the eighteenth century in Scandinavian church records (8). Traditionally, the disease is described as a syndrome with a biphasic course beginning with an influenza-like illness followed by a second neuroinvasive phase with neurological symptoms of variable severity, ranging from meningitis to severe meningoencephalitis with or without myelitis (3, 4, 6) (**Figure 1**). It shall be noted, however, that also monophasic patterns of disease development have been described (9). Upon infection, virus is detected in serum in the first phase of the disease but rarely in the second phase (10).

Due to increased geographic distribution of TBEV as well as a marked increase in morbidity in many areas, TBEV-infection has more recently caught attention as a public health problem. TBE is now observed in large parts of Europe as well as in northern Asia (3, 4). The main risk areas for TBE in Europe are primarily parts of central and eastern Europe as well as the Baltic and Nordic countries. With respect to central Europe, risk areas extend from Switzerland in the west into northern Italy and the Balkan countries (11). The incidence of TBEVinfection in endemic countries varies from year to year (12– 14), however, an overall upsurge has been reported in certain parts of Europe, including the borders between Austria, Slovenia, and Italy (15, 16). These changes have been related to climatic, ecological, environmental, and socioeconomic factors that all can lead to an increased risk of human exposure to infected ticks (17–20).

The total number of annual cases has been estimated to be up to 13,000, and as such the infection constitutes the most important tick-borne viral disease (4). More than 30% of patients with clinical symptoms from TBEV-infection develop prolonged sequelae, some of which may become life-long including neuropsychiatric symptoms, severe headaches, and a general decrease in quality of life (3, 4, 6, 7). The mortality rates differ between the strains. Infection with the Far Eastern strain (TBEV-FE) has a mortality rate of 5–35%, whereas the other two strains (TBEV-Eu and TBEV-Sib) have mortality rates of 1–3% (3, 4). There is no specific treatment (e.g., antivirals) for TBE; rather, symptomatic treatment is the only available option (3, 4, 9).

Of importance, TBE may be prevented by vaccination. There are in total four licensed vaccines to TBE. Two vaccines based on TBE-Eu subtype are licensed in Europe and two are licensed in Russia. Additionally, a TBEV-vaccine based on the Far Eastern subtype is produced and marketed in China. All vaccines are based on formalin-inactivated strains of TBEV (3, 4, 21, 22). In areas where the disease is highly endemic, WHO recommends that vaccination should be offered to all groups above 1 year of age (4, 23). Primary vaccination against TBE includes three doses of the vaccine within the first year, followed by revaccinations every third to fifth year to maintain immunity. Vaccination is generally considered effective and TBE incidence has decreased substantially in TBEV-endemic regions with successful vaccination-programs (24). Randomized controlled trials in large populations have shown high immunogenicity with often-strong antibody production and acceptable rates of adverse events following vaccination (25–28). Breakthrough TBE after vaccination is generally considered rare (4). However, over the last years, vaccine failures have been reported, in particular in middle-aged and elderly individuals, who have completed the primary vaccination (29–31).

Infection with TBEV triggers humoral and cell-mediated immune responses. A confirmed diagnosis of TBE is established by the detection of specific IgM and IgG in serum. IgM antibodies have been observed in sera very early in symptomatic TBE disease, whereas IgG antibodies peak in the convalescent phase of disease (32). IgG antibodies can persist over lifetime and prevent TBE (4, 33). Early after clinical disease onset, TBEV-specific antibodies can also be found in the cerebrospinal fluid (CSF) (32, 33). In contrast to the humoral immune response, the cellmediated immune responses elicited to natural infection have been rather poorly studied until recently. The latter responses may contribute both to host resistance against infection as well as to pathological reactions affecting the target organ of the virus, i.e., primarily the CNS.

Here, we review recent progress in studies of the cell-mediated immune response to human TBEV infection. A particular emphasis is devoted to natural killer (NK) cell- and T cellmediated responses. Responses to TBEV are discussed in context of cell-mediated immune responses toward other flavivirus infections. We also discuss some immunopathological aspects of TBE with a particular emphasis on cell-mediated immune reactions in the CNS. Cell-mediated immune reactions in the CNS may contribute to neural damage with severe consequences of brain function, and could in the worst cases lead to fatal outcome. First, however, some aspects of the TBEV itself are covered.

# TBEV AND OTHER FLAVIVIRUSES

All flaviviruses are enveloped and have a positive-sense single stranded RNA genome, which per se acts as messenger RNA upon entrance in the host cell. The RNA encodes for a polyprotein, which is co- and post-translationally cleaved by viral and cellular proteases into three structural proteins; capsid (C), precursor membrane (prM) and envelope glycoproteins (E), and seven non-structural proteins including NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5 (34). Flaviviruses enter the cell through clathrindependent endocytosis upon attachment of the E protein to a receptor. Heparan sulfate has been identified as such receptor for TBEV (35); however, there are most likely also other yet not identified receptors for the virus. Following cell entry, the flavivirus is delivered to endosomes (36), in which the low pH triggers the E protein to fuse with the endosomal membrane and the nucleocapsid is released into the cytosol. The assembly of immature flavivirus virions (36–38), including TBEV (39, 40), occurs in the ER, and the viral particles are transported to the

Golgi apparatus. The virion particles are immature until the envelope protein is rearranged and prM is cleaved by the host enzyme furin in the acidic environment in the Golgi apparatus. Immature particles are non-infectious and proteolytic cleavage of prM is a prerequisite for viral infectivity. However, studies have shown that complete cleavage of prM is not necessary for viral infectivity (41–43).

In general, species of flaviviruses have many similarities, but their preferred host cells differ. TBEV is shown to replicate 10,000-fold higher in human neuronal cells as compared with epithelial cells (44). A similar infection pattern has recently been shown for ZIKV (45).

# NK CELLS

NK cells are innate lymphocytes, though recent studies have revealed "adaptive" features of these cells (46, 47). They are perhaps best known for their ability to kill virus-infected and tumor cells. NK cell cytotoxicity is regulated by the expression of numerous activating and inhibitory receptors that sense ligands on neighboring healthy and altered cells. Several activating receptors recognize molecules that are up-regulated on cells during conditions of cellular stress, such as viral infection and transformation [reviewed in (48, 49)] whereas many inhibitory receptors, e.g., human killer cell Ig-like receptors (KIR) bind to HLA class I molecules. Additionally, NK cells have an important role in producing cytokines and chemokines, as well as by other means interacting with other immune and non-immune cells.

Human NK cells are classically defined as CD3<sup>−</sup> (T cell receptor negative), CD56<sup>+</sup> cells and represent about 15% of peripheral blood lymphocytes. These cells have for long been divided into two main subsets; CD56bright and CD56dim cells (50). The CD56bright NK cells are thought to be less mature and are commonly known as primarily cytokine-producing cells with low cytotoxic ability, whereas CD56dim NK cells are best known for their potent cytotoxic activity upon target cell recognition (51). However, the latter are also ample cytokine producers upon interaction with target cells (51). Both "natural" and antibody-mediated NK cell cytotoxicity is mediated by exocytosis of cytoplasmic granules containing perforin and granzymes (52). Cytotoxic responses may also to various degrees involve TRAIL- and Fas-ligand-mediated induction of apoptosis (53, 54). CD56dim NK cells frequently express CD16 (FcγRIII), KIRs, and CD57, which regulate their function and define distinct stages of NK cell maturation (55), whereas CD56bright NK cells largely lack expression of these molecules.

# THE ROLE OF NK CELLS IN HUMAN TBEV INFECTION

Direct evidence for a protective role of NK cells has been found in experimental models of several viral infections, including cytomegalovirus and influenza, and a number of studies have indicated that they play a role also in protection against viral infections in humans. For example, NK cell-deficiencies in humans result in severe herpes virus infections in childhood and adolescence (56). NK cells may also have a protective role in human TBEV-infection. At the same time, responses mediated by these cells may be associated with development of symptoms in the course of TBEV-infection. Although there is only little known about NK cells in TBE, NK cells have been detected in CSF of patients with TBE (57), an observation that indicates transmigration through the blood brain barrier (BBB).

To gain a better understanding of the NK cell response to human TBEV-infection, we recently performed a longitudinal study providing an in-depth analysis of the human NK cell response to acute TBEV-infection in a well-defined cohort of TBE patients. The study had an emphasis on NK cell responses during the second stage of disease from which clinical samples were available. NK cell activation, as measured by expression of the proliferation marker Ki67, was apparent at the time of hospitalization (58) (illustrated in **Figure 2**). Concomitant with the increase in NK cell activation in the acute stage of disease, augmented levels of IL-12, IL-15, IL-18, IFN-γ, and TNF were detected in patient plasma. In parallel with high levels of activation, the activated NK cells expressed less perforin, granzyme B, and Bcl-2. By 3 weeks after hospitalization, the NK cell activation decreased to levels seen in healthy controls. This TBEV-induced NK cell activation was restricted predominantly to more differentiated CD57+CD56dim NK cells. Functionally, CD56dim NK cells responded poorly to target cells at the time of hospitalization, but they recovered functional capacity to healthy control levels during the convalescent phase. The poor functionality of NK cell responses was exclusive for target cell recognition, since NK cell responses induced by IL-18 and IL-12 remained unchanged throughout the disease (58).

# NK CELL RESPONSES TOWARD OTHER ACUTE FLAVIVIRUS INFECTIONS IN HUMANS

To be able to interpret the above-mentioned NK cell responses to acute TBEV infection, it is important to understand NK cell responses to other acute virus infections, including acute flavivirus infections. In this respect, NK cells have to various extent been studied ex vivo in other acute flavivirus infections, including DENV (60, 61) and WNV (62, 63) as well as hepatitis C virus (HCV), a distant relative within the Flaviviridae family (64, 65). NK cells have also been studied after vaccination with the live attenuated YFV 17D vaccine (66–68). They have been suggested to influence disease severity and outcome, and to contribute to viral control in these infections, even though underlying mechanisms are not well studied.

In this context, it was observed that the absolute number of NK cells in patients with a mild form of infection with DENV was higher as compared to patients with the more severe form of the infection, dengue hemorrhagic fever (DHF) (69). Reduced numbers of NK cells in the circulation may be indicative of migration toward peripheral target organs. Furthermore, a higher frequency of NK cells expressing CD69 early on during the infection in children developing severe DHF has been reported (70). In recent studies of DENV-infection, we found NK cells to be robustly activated during the first week after symptom debut. Here, the response seemed to be confined largely to the CD56bright subset of NK cells and less mature CD56dim NK cells (our own unpublished studies). Noteworthy in the context of acute TBEV infection, activation of NK cells may also occur very early, even before the onset of symptoms. This possibility is supported by the observation that the highest levels of NK cell activation in most TBEV infected patients were observed already at the time of hospitalization (58). This notion is corroborated in studies of YFV vaccinated individuals. An early response by NK cells was observed in study subjects vaccinated with YFV-17D, where expression of both Ki67 and CD69 was increased on NK cells as early as 1 week after vaccination (66). Accumulation of adaptive-like NK cells expressing the activating receptor CD94/NKG2C has been reported in some human viral infections (71–73); however, no expansion of NKG2C<sup>+</sup> NK cells in blood has been observed in TBEV-infection or any other flavivirus infection (58). It can, however, not be excluded that this type of expansion could occur locally, e.g., at the site of infection.

In addition to the observed activation of NK cells in vivo in different flavivirus infections, a protective role of NK cells is also supported by in vitro data. For example, primary activated human NK cells have been shown to inhibit WNVinfection of Vero cells (63) and IFNα-activated NK cells can kill HCV-infected hepatoma cells in vitro (65). In addition, flavivirus-infected target cells have been reported to display virus-mediated up-regulation of MHC class I (74), and could thereby theoretically evade lysis from NK cells by engaging inhibitory receptors. The dampened NK cell responses to target cells in acute TBEV-infection further support this notion (58). On the other hand, increased MHC class I expression could result in enhanced T cell responses. In such a scenario, one may speculate that flaviviruses may have been driven more toward escape from innate immunity rather than from adaptive T cell immunity (75).

# T CELLS

In contrast to NK cells, CD8 and CD4 T cells recognize specific foreign peptide sequences presented by HLA class I and II molecules, respectively (76). Like NK cells, major functions of CD8 T cells are to kill infected cells through the release of perforin and granzymes, and to secrete cytokines such as IFN-γ, TNF, and IL-2. The cytotoxic T cell response to acute infection can typically be divided into three phases; priming and expansion, resolution and contraction, and memory formation. During the first phase, naïve CD8 T cells divide and differentiate into effector cells acquiring high cytotoxic ability (77). Following viral clearance, the effector T cell population contracts and the majority of the pathogen-specific T cells enter apoptosis. A small pool of pathogen specific T cells (5–10%) survives as memory cells in the third stage (78). Memory T cells are a principal component of immunity against intracellular pathogens such as viruses. They are distinguished by their capacity to survive long-term, and undergo rapid and robust proliferation and acquisition of effector function upon antigen re-exposure (78). Memory T cells can vary in their phenotype, localization, and function allowing them to protect the host against a broad array of potential insults.

Distinct stages of CD8 T cell differentiation are defined by the expression of specific surface markers such as the isoforms of CD45 and expression of the homing receptor CCR7. These stages of differentiation are useful in the characterization of responses to, e.g., anti-viral responses. The set stages define CD45RA+CCR7<sup>+</sup> as naive (TN), CD45RA−CCR7<sup>+</sup> as central memory (TCM), CD45RA−CCR7<sup>−</sup> as effector memory (TEM), and CD45RA+CCR7<sup>−</sup> as effector memory RA (TEMRA)

CD8 T cells (79, 80). TCM cells primarily reside in secondary lymphoid organs, possess the greatest proliferative potential among the memory T cell subsets and can rapidly expand and differentiate following re-challenge. TCM cells have higher sensitivity to antigenic stimulation, are less dependent on co-stimulation and provide better feedback to DCs and B cells compared to T<sup>N</sup> cells. TEM cells can migrate between tissues and secondary lymphoid organs and provide immune surveillance.

# THE ROLE OF T CELLS IN HUMAN TBEV INFECTION

Due to difficulties in identifying the acute phase of viral infection in humans, T cell responses to viral infections have to a large extent been addressed in studies of pathogens causing chronic infections such as HIV-1, EBV, HCV, and CMV (81–85). Such responses can be very robust, as exemplified by the massive clonal expansion of antigen-specific CD8 T cells seen in many infections (83, 85). Based on these studies, it has also become clear that the resulting populations of human CD8 T cells display striking phenotypic differences, as determined by the expression profiles of surface markers (79–81). In contrast to many other infections, including some flavivirus infections (86–89), there are only few studies of T cell responses to TBEV-infection in humans. This hold true for acute as well as cases with prolonged TBE disease. Noteworthy, however, one report has shown that TBEV-specific CD4 T cells from naturally infected patients show a higher level of polyfunctionality in response to antigen in the convalescent phase of disease, as compared to TBE-vaccine specific T cells (90).

The general lack of studies more systematically characterizing the human T cell response to TBEV-infection prompted us to study the primary T cell-mediated immune response in patients diagnosed with TBE with a particular emphasis of CD8 T cells (59). Similar to our study on NK cells (58), the T cell study focused on responses during the second stage of disease from which clinical samples were available. During this phase, CD8 T cells were strongly activated, as detected by increased expression of Ki67, within 1 week of hospitalization (illustrated in **Figure 2**). A large part of these CD8 T cells expressed high levels of perforin and granzyme B, and low levels of the anti-apoptotic protein Bcl-2. In contrast to CD8 T cells, CD4 T cells showed only low or at most moderate levels of activation. The TBEV-antigen specific CD8 T cells had a TEM PD-1<sup>+</sup> phenotype throughout the course of disease. TBEVspecific CD8 T cells were predominantly Eomes+Ki67+T-bet<sup>+</sup> in the acute stage of disease. This pattern was replaced by an Eomes−Ki67−T-bet<sup>+</sup> profile in the convalescent phase of disease. TBEV-specific CD8 T cells were mainly monofunctional in the acute stage of disease, and tended to become more polyfunctional in the convalescent phase when clinical symptoms retracted (59).

# T CELL RESPONSES TOWARD OTHER ACUTE FLAVIVIRUS INFECTIONS IN HUMANS

To be able to better interpret the above-mentioned T cell responses to acute TBEV infection, we compared the present results with T cell responses to other acute virus infections, including infections by other flaviviruses. The live attenuated YFV-vaccine strain can replicate after vaccination leading to a detectable viral load similar to a mild infection. Thus, this vaccine can be utilized as a controlled model to study mild acute viral infection in humans. CD8 T cells become activated within 1–2 weeks after vaccination with the YFV vaccine (86, 91, 92). YFV antigen-specific CD8 T cells predominantly display a TEM PD-1 <sup>+</sup> phenotype, which transition into a TEMRA PD-1<sup>−</sup> memory phenotype (86). With respect to DENV-infection, a high level of functionality of DENV-specific T cells is associated with a better disease outcome (93). Similarly, patients hospitalized with (severe) TBE show a low level of T cell functionality in the acute stage of disease (59), indicating the importance of high function among virus-specific T cells for beneficial disease outcome. CD8 T cells have been shown primarily to respond with IFN-γ to JEV in asymptomatic JEV-exposed donors (87).

Activation of CD4 T cells with an optimal magnitude, specificity and kinetics may be a requirement for viral clearance and protective immunity. In the immune response induced by the YFV vaccine, activation of CD4 T cells (peak at 10 days after vaccination) precedes that of CD8 T cells, and this may be of importance to elicit strong immunological memory (86). Furthermore, CD4 T cell release of IFN-γ may have an impact on disease outcome since CD4 T cells, and not CD8 T cells, were shown to dominate the IFN-γ response in recovered Japanese encephalitis (JE) patients. In addition, a high quality polyfunctional CD4 T cell response can be associated with better disease outcome in JE patients (87). In murine models, a perforin-dependent mechanism by the CD8 T cells has been shown to clear WNV from infected neurons, thereby suggesting an immunopathological role of T cells in mice (88). In this context, it is of interest to note that TBEV-specific T cells have a high content of both perforin and granzyme B (59), but whether the same perforin-dependent mechanism is causing immunopathogenesis in acute infection with TBEV remains to be investigated.

# CROSS-REACTIVITY WITHIN THE FAMILY OF FLAVIVIRUSES

Immunological cross-reactivity between TBEV and other species within the flavivirus family may also contribute to disease (94, 95). Antibody-dependent enhancement (ADE) is a described phenomenon that can occur when non-neutralizing antibodies facilitate virus entry into host cells, leading to increased infectivity in the cells. ADE is commonly observed in vitro in cell culture-based models (96), but it is questioned as to which degree this phenomenon occurs in vivo. A recent study from non-human primates in vivo, did not observe increased ZIKV titers after prior infection with heterologous flaviviruses (97). Protective cross-reactivity of flaviviruses has been reported as well, opposing increased pathogenesis upon pre-exposure to other species of flaviviruses (98). TBEV has been suggested to cause both pathogenic and protective cross-reactivity. Polyclonal sera against members of the TBE serocomplex (including TBEV, Kyasanur Forest disease virus, Omsk hemorrhagic fever virus, and Langat virus) enhance viral replication of TBEV in vitro (96). However, it has recently been shown that antibodies generated from TBEV infection or from the TBEV vaccine can mediate cross neutralization against other, if not most, of the members of the TBE virus complex (99). Furthermore, sera from some individuals vaccinated against TBEV and JEV neutralized WNV, and the neutralization was enhanced by YFV vaccination in some recipients (95), altogether indicating that previous flavivirus exposure may sometimes provide a degree of protection to new flaviviruses.

Cell-mediated immunity and cross-reactivity caused by TBEV and other flaviviruses has been less well studied. Recently, a study demonstrated that JEV- and JE vaccine-specific T cells cross-react with DENV (87, 100). In line with this, it was also recently shown that vaccination with YFV vaccine could induce ZIKA-specific T cells, thereby suggesting cross-protection of flavivirus-specific T cells (101). The latter phenomena opened up a discussion as to the possibility of utilizing the YFV vaccine to protect against Zika virus infection (101). In the present context, it remains to be investigated whether YFV vaccination elicits protective cross-reactive immunity also toward TBEV.

# IMMUNOPATHOLOGICAL ASPECTS OF TBEV INFECTION IN THE CNS

In contrast to the significant interest in emerging infections such as the recent Zika pandemic [reviewed in (45, 102, 103)], studies of the immune response toward TBEV-infection as such, and TBEV-induced immunopathology in particular, have been rather limited.

Immune and none-immune mechanisms have been proposed contribute to the crossing of TBEV over the BBB and invasion of the CNS [reviewed in (104)]. Cytokines may facilitate this process. Cytokines such as TNF-a and IL-6 have an impact on endothelial cell permeability that may induce a BBB disruption (105, 106), leading to crossover of the virus into the CNS. A distinct mechanism by which the TBEV could possibly cross the BBB is the Trojan Horse mechanism (107), by which TBEV-infected immune cells such as dendritic cells, neutrophils, monocytes, macrophages, and T cells would migrate into the parenchymal compartment causing infection of neurons or other cells in the brain and the spinal cord. Yet, an alternative route is invasion via the olfactory epithelium (108, 109).

After the landmark discovery of the lymphatic system present in the meninges that connects the CNS to the peripheral blood (110, 111), the classical concept of the CNS as an immune privileged site has been replaced by a view of an immune regulated site. Hence, under normal conditions a continuous transmigration of lymphocytes, monocytes, DCs and macrophages occurs. They may serve to detect any kind of infection or injury in the brain [reviewed in (112)]. Similar to other CNS infections, increased frequencies of T cells have been reported in CSF of TBE patients (57). Hence, activated T cells are crossing the BBB in TBE; however, the role of T cells at this site not well understood (113, 114). On one hand they could contribute toward clearing viral infection but on the other hand they may mediate immunopathology within the CNS. Corroborating the latter speculations are findings in which granzyme B<sup>+</sup> CD8 T cell infiltrates have been linked to cell-death in infected human neuronal tissue (113) and, in parallel, mice with CD8 T cell deficiency have been shown to have prolonged survival upon infection with TBEV compared to mice with adoptively transferred CD8 T cells to immuno-competent mice (114). Furthermore, studies of post-mortem tissue of TBE patients have shown a predominance of macrophages/microglia and CD3<sup>+</sup> T cells (both CD4<sup>+</sup> and CD8+) in brain parenchyma (113). As seen in other flavivirus infections, macrophages and microglia also play a role in tissue destruction in human TBE.

In relation to NK cells and their possible role in causing immunopathogenesis, it is of interest to note that also these cells have been detected in the cerebrospinal fluid (CSF), albeit in low numbers, in patients with severe TBE meningitis or encephalitis (57). Activated NK cells may be protective, but they may also, like T cells, take part in immunopathological reactions as they are known to participate in direct killing of infected cells, indirect killing through cytokines or chemokines, or by the recruitment of inflammatory cells into the tissues (115, 116). Although recent results support a role for NK cells in clinical TBEV-infection, more studies are needed to provide a better understanding of the role NK cells play in pathogenic processes of human TBE infection.

Knowledge and experience gained in the field of the immunopathogenesis of other diseases affecting the CNS and its immunological compartments could be helpful in understanding TBE-specific diseases patterns. For example, in multiple sclerosis (MS), an inflammatory disease with pathology affecting the CNS (117), the concentration of the Sphingosine-1-Phosphate (S1P) in CSF is elevated and S1P-signaling is altered. In MS, binding of S1P to S1P1-receptors expressed on lymphocytes leads to invasion of autoreactive T cells

## REFERENCES


into the CNS, the latter contributing to the hallmarks of the disease including demyelination and neurodegeneration (118). Interestingly, during the phase of acute infection in TBEV-infected patients, the levels of S1P in blood and CSF are highly elevated (119). This increase might promote a proinflammatory response. An increased production of extracellular S1P can be regulated by modulators of the S1P pathway, such as fingolimod, which is an immunomodulatory drug used in the treatment of MS (118). Therefore, therapeutic options used in other CNS diseases that share common immunopathogenic mechanisms with TBE could be used as models to aid in the development of new strategies for TBE treatment.

# CONCLUDING REMARKS

In the present review, we have focused our attention to recent insights into the cell-mediated immune response to human TBEV infection, with an emphasis on studies of NK cell and CD8 T cell mediated responses. Until recently, the latter have been poorly studied. As yet, however, much more needs to be learnt with respect to these responses and research in this area should be encouraged. We have also addressed some aspects of TBEV CNS pathogenesis, a process still far from understood in detail. Clearly, however, cell-mediated immune responses likely play an important role in this process. As TBE continues to be an increasing global health problem and challenge, much more research is needed into this emerging disease. Several areas of research of the TBEV itself, and the clinical disease TBE merit further studies. Not the least, the specific organ pathogenesis caused by TBEV and the immune response, including infiltrating immune cells, needs more investigation. Furthermore, the possibility of antiviral treatment and other possible treatment modalities needs much more thorough investigation to prevent disease development and the often severe sequeale following infection of humans with TBEV.

# AUTHOR CONTRIBUTIONS

KB, AC, and H-GL wrote the manuscript. All other authors provided valuable contributions and insights into the manuscript.

# FUNDING

The authors were founded by the Swedish Research Council, the Swedish Foundation for Strategic Research, Karolinska Institutet and the Stockholm County Council.


focus of infection. Eur J Epidemiol. (1996) 12:373–5. doi: 10.1007/BF00 145300


encephalitis vaccine FSME-IMMUN: results of two large phase 3 clinical studies. Vaccine (2006) 24:5256–3. doi: 10.1016/j.vaccine.2006.03.061


virus 17D as it matures from an effector- to a memory-type response. J Immunol. (2013) 190:2150–8. doi: 10.4049/jimmunol.1202234


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Blom, Cuapio, Sandberg, Varnaite, Michaëlsson, Björkström, Sandberg, Klingström, Lindquist, Gredmark Russ and Ljunggren. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Advances in Designing and Developing vaccines, Drugs, and Therapies to Counter ebola virus

*Kuldeep Dhama1 \*, Kumaragurubaran Karthik2 , Rekha Khandia3 , Sandip Chakraborty <sup>4</sup> , Ashok Munjal3 , Shyma K. Latheef <sup>5</sup> , Deepak Kumar <sup>6</sup> , Muthannan Andavar Ramakrishnan7 , Yashpal Singh Malik8 \*, Rajendra Singh1 , Satya Veer Singh Malik9 , Raj Kumar Singh10 and Wanpen Chaicumpa11*

*1Division of Pathology, ICAR-Indian Veterinary Research Institute, Bareilly, India, 2Central University Laboratory, Tamil Nadu Veterinary and Animal Sciences University, Chennai, India, 3Department of Biochemistry and Genetics, Barkatullah University, Bhopal, India, 4Department of Veterinary Microbiology, College of Veterinary Sciences and Animal Husbandry, Agartala, India, 5 Immunology Section, ICAR-Indian Veterinary Research Institute, Bareilly, India, 6Division of Veterinary Biotechnology, ICAR-Indian Veterinary Research Institute, Bareilly, India, 7Division of Virology, ICAR-Indian Veterinary Research Institute, Nainital, Uttarakhand, India, 8Division of Biological Standardization, ICAR-Indian Veterinary Research Institute, Bareilly, India, 9Division of Veterinary Public Health, ICAR-Indian Veterinary Research Institute, Bareilly, India, 10 ICAR-Indian Veterinary Research Institute, Bareilly, Uttar Pradesh, India, 11Center of Research Excellence on Therapeutic Proteins and Antibody Engineering, Department of Parasitology, Faculty of Medicine SIriraj Hospital, Mahidol University, Bangkok, Thailand*

#### *Edited by:*

*Hiroyuki Oshiumi, Kumamoto University, Japan*

#### *Reviewed by:*

*Oscar Negrete, Sandia National Laboratories (SNL), United States Pei Xu, Sun Yat-sen University, China Yohei Kurosaki, Nagasaki University, Japan*

#### *\*Correspondence:*

*Kuldeep Dhama kdhama@rediffmail.com; Yashpal Singh Malik malikyps@gmail.com*

#### *Specialty section:*

*This article was submitted to Microbial Immunology, a section of the journal Frontiers in Immunology*

*Received: 01 April 2018 Accepted: 23 July 2018 Published: 10 August 2018*

#### *Citation:*

*Dhama K, Karthik K, Khandia R, Chakraborty S, Munjal A, Latheef SK, Kumar D, Ramakrishnan MA, Malik YS, Singh R, Malik SVS, Singh RK and Chaicumpa W (2018) Advances in Designing and Developing Vaccines, Drugs, and Therapies to Counter Ebola Virus. Front. Immunol. 9:1803. doi: 10.3389/fimmu.2018.01803*

Ebola virus (EBOV), a member of the family *Filoviridae*, is responsible for causing Ebola virus disease (EVD) (formerly named Ebola hemorrhagic fever). This is a severe, often fatal illness with mortality rates varying from 50 to 90% in humans. Although the virus and associated disease has been recognized since 1976, it was only when the recent outbreak of EBOV in 2014–2016 highlighted the danger and global impact of this virus, necessitating the need for coming up with the effective vaccines and drugs to counter its pandemic threat. Albeit no commercial vaccine is available so far against EBOV, a few vaccine candidates are under evaluation and clinical trials to assess their prophylactic efficacy. These include recombinant viral vector (recombinant vesicular stomatitis virus vector, chimpanzee adenovirus type 3-vector, and modified vaccinia Ankara virus), Ebola virus-like particles, virus-like replicon particles, DNA, and plant-based vaccines. Due to improvement in the field of genomics and proteomics, epitope-targeted vaccines have gained top priority. Correspondingly, several therapies have also been developed, including immunoglobulins against specific viral structures small cell-penetrating antibody fragments that target intracellular EBOV proteins. Small interfering RNAs and oligomer-mediated inhibition have also been verified for EVD treatment. Other treatment options include viral entry inhibitors, transfusion of convalescent blood/serum, neutralizing antibodies, and gene expression inhibitors. Repurposed drugs, which have proven safety profiles, can be adapted after high-throughput screening for efficacy and potency for EVD treatment. Herbal and other natural products are also being explored for EVD treatment. Further studies to better understand the pathogenesis and antigenic structures of the virus can help in developing an effective vaccine and identifying appropriate antiviral targets. This review presents the recent advances in designing and developing vaccines, drugs, and therapies to counter the EBOV threat.

Keywords: Ebola virus, Ebola virus disease, vaccines, prophylactics, drugs, therapeutics, treatment

# INTRODUCTION

Ebola virus (EBOV; *Zaire ebolavirus*) is the causative agent of a severe hemorrhagic fever disease, Ebola virus disease (EVD; formerly called Ebola hemorrhagic fever). It was first recognized in 1976 in northern Democratic Republic of Congo, at that time Zaire (1–3). Since then, EVD is endemic in Africa. Fruit bats are the best-known reservoirs of EBOV (4). EVD is a well-established zoonotic disease; the initial cases of the EVD outbreaks occur after contact with reservoir or materials contaminated with the virus and followed by human-to-human transmission (5). EBOV is not only a serious public health issue but now also designated as category A pathogen and considered as a potential bioterrorism agent (6, 7). EBOV causes high mortality rates of up to 88% in the infected humans (8); therefore, it is classified as a risk group 4 agent and handled under biosafety level-4 containment. The risk of mortality is relatively greater in the elderly and/or patients with high viral load and poor immune response at the initial stage of the infection (9).

The EBOV belongs to the *Filoviridae* family and has a unique thin filamentous structure that is 80-nm wide and up to 14-µm long. Its envelope is decorated with spikes of trimeric glycoprotein (GP1,2) which are responsible for mediating viral entry into target cells (function of GP1) (10) and release of viral ribonucleoprotein from endosome to cytoplasm for replication (function of GP2) (11, 12). EBOV infects primarily humans, simians, and bats; but other species such as mice, shrew, and duikers may also contact infection (3, 13). Of the five identified EBOV species, four species, *viz*., EBOV, Sudan virus (SUDV; *Sudan ebolavirus*), Tai Forest virus (TAFV; *Tai Forest ebolavirus*, formerly *Côte d'Ivoire ebolavirus*), and Bundibugyo virus (BDBV; *Bundibugyo ebolavirus*), are known to infect humans and cause disease, whereas Reston virus (RESTV; *Reston ebolavirus)* is non-human primate (NHP) pathogen.

After an initial incubation period of 3–21 days, the disease progresses quickly to fever, intense fatigue, diarrhea, anorexia, abdominal pain, hiccups, myalgia, vomiting, confusion, and conjunctivitis (14) which may lead to the loss of vision (15). EBOV can spread from males to females through semen (16) and from mother to fetus and infant during gestation and lactation, respectively (17). Of the note, in an EBOV-infected patient, higher concentration of Ebola viral RNA in semen was noticed during the recovery period than the viral concentration in the blood during peak time of infection, suggesting male genital organ as virus predilection site for replication (18). Usually the human immune system mounts a response against infectious pathogens by sensing the pathogen-associated molecular patterns *via* a variety of pathogen-recognition receptors. Nevertheless, in the case of EBOV, innate immunity is impaired by the immunosuppressive viral proteins including VP35 and VP24, and lymphocytes are depleted as a result of apoptosis caused by inappropriate dendritic cell (DC)–T-cell interaction (7, 19). A thorough understanding on the pathogenesis of this deadly virus is essential because of its severe health impacts (20).

The increased incidences and fast spread of EBOV paving into a pandemic flight has compelled more focus of research to develop strategies and remedial measures for mitigating the impact and consequential severity of the viral infection. Even before delineating the less studied Ebola viral genome fully, researchers throughout the globe and health industry were pressured to focus on the development of effective and safe Ebola vaccines and therapeutics (21, 22). As of now, no licensed vaccines and direct-acting anti-EBOV agents are available to protect against the lethal viral infection or to treat the disease. To minimize the suffering, EBOV-infected patients are only provided with symptomatic treatment and supportive care. Because of its high pathogenicity and mortality rate, preventive measures, prophylactics, and therapeutics are essential, and researchers worldwide are working to develop effective vaccines, drug, and therapeutics, including passive immunization and antibody-based treatments for EVD (23–26). Prior to the 2014–2016 EBOV outbreak in West Africa, which has been the deadliest EBOV outbreak to date, convalescent blood products from survivors of EVD represented the only recommended treatment option for newly infected persons. Administration of monoclonal antibody (mAb) cocktails (ZMapp, ZMAb, and MB-003) as post-exposure prophylactics have been found to reverse the advanced EVD in NHPs and/or effectively prevented morbidity and mortality in NHPs (27–30).

There is the need for an effective vaccine against EBOV, especially in high-risk areas, to prevent infections in physicians, nurses, and other health-care workers who come into contact with diseased patients (31). Regular monitoring and surveillance of EBOV is essential to control this disease. In the EBOV outbreak, novel surveillance approaches include contact tracing with coordination at the national level and "lockdown" periods, during which household door-to-door reviews are conducted to limit the spread of the virus. Swift identification and confirmation of the Ebola cases and immediate follow-up of appropriate prevention and control measures, including safe burial of dead persons, are crucial practices to counter EBOV (32).

After the onset of EVD, treatment is required, whereas, when EBOV is circulating in population dense areas before infection, prophylactic measures like vaccination are necessary. One of the main challenges in containing EBOV is its presence in remote areas that lack technology and equipment to limit the virus spread. Because of its lethality, EBOV can only be handled in laboratories with biosecurity level-4 containment; thus, only few laboratories in the world can conduct EBOV research and testing of the counter measures against the authentic virus. Recent efforts by several organizations have focused on identifying effective therapies and developing appropriate vaccination strategies (33). Several drugs and vaccines have been developed against EBOV, and the production of low-cost drugs and vaccines against EBOV is essential for everyone, including those in the high-risk areas of the world, to be protected (26, 34). As of the acquisition of better knowledge against the pathogen due to improvement in the field of genomics and proteomics, there has been expansion in the field of vaccine synthesis where epitope-based vaccines are gaining top priority (35–37).

The present review aims to discuss advances in designing and development of EBOV vaccines, drugs, antibody-based treatments, and therapeutics, and their clinical efficacy in limiting EVD, thereby providing protection against the disease and alleviating high public health concerns associated with EBOV.

# ADVANCES IN DEVELOPING VACCINES AGAINST EBOV

There is a clear need for an effective vaccine to prevent the rapid spread of EVD. An inactivated EBOV vaccine was first produced in 1980. This vaccine was tested for efficacy in guinea pigs (7). Since that time, several vaccines against EBOV have been developed, but no vaccine is licensed and available in the market (7). After the massive 2014–2016 outbreak of EBOV, several researchers have begun working to develop an effective vaccine (38). For an EBOV vaccine candidate, a long-lasting immune response is essential; as EBOV remains in the seminal fluid of EVD survivors as long as 401 days post-infection (39, 40). Keeping this window of virus persistence, a vaccine conferring immunity at least for 2 years is recommended by the Wellcome Trust-CIDRAP Ebola Vaccine Team B initiative (41). Vaccines like the chimpanzee adenovirus type 3 (ChAd3)-based non-replicating ChAd3-EBO vaccine, prime-boost recombinant adenovirus type 26 vector (Ad26.ZEBOV) followed by the modified vaccinia Ankara vector (MVA-BN-Filoa) vaccine, adenovirus 5-vectored EBOV vaccine, EBOV DNA vaccine, and recombinant vesicular stomatitis virus (rVSV) vector-based vaccine are undergoing clinical trials to evaluate their efficacy against EVD (38). The RNA-dependent RNA polymerase (L) epitope-based vaccine was designed using immunoinformatics. Various software have been used to analyze immunological parameters, and this epitope vaccine was found to be a good candidate for use against EVD (42). Two conserved peptides of EBOV, 79VPSATKRWGFRSGVPP94 from GP1 and 515LHYWTTQDEGAAIGLA530 from GP2, were identified as targets for the development of an epitope-based vaccine (43). Collection of the sequences of EBOV glycoproteins and examination for determining the proteins with greatest immunogenicity have been performed using *in silico* methods. The best corresponding B and T cell epitopes included peptide regions encompassing residues 186–220 and 154HKEGAFFLY162, respectively. Such predicted epitopes can confer the long-lasting immunity against EBOV with better ability of protection (36).

Ebola virus-GP fused with the Fc fragment of a human IgG1 subunit vaccine administrated with alum, QS-21, or polyinosinicpolycytidylic acid-poly-l-lysine carboxymethylcellulose adjuvant induced strong humoral immunity in guinea pigs (44). Effectiveness of a ring vaccine using rVSVΔG/EBOVGP in cases of simulated EBOV disease was studied and even this approach can be employed during an outbreak situation (45). Notably, the neutralizing antibodies play a major role in conferring protection against EBOV infections. Thus, an EBOV vaccine capable of effectively inducing a long-lasting neutralizing antibody response is desirable for developing appropriate prevention strategies in combating the infection. In this line, the mucin-like domain of EBOV envelope glycoprotein GP1 has been identified to be critical in induction of protective humoral immune response (46, 47). Filorab 1 vaccine revealed desirable immunogenicity without the side effects. The main advantage of this vaccine is its higher immune response induction in chimpanzees (captive) when given orally and also with a single dose [instead of multiple doses as is required by virus-like particle (VLP) vaccine] (48). Modified mRNA-based vaccine constructs, formulated with lipid nanoparticles (LNPs) to facilitate delivery, are being tested against EBOV challenge in guinea pigs. These mRNAs induced robust immune responses and conferred up to 100% protection from the infection (49).

It is important to note that compilation of data in relation to immune responses (both induced by vaccines and natural infection) and the records of community members showing IgG seropositivity should be kept systematically. Assimilation of such information will help to handle next outbreaks with more rigidity, thereby helping to check EVD-associated disasters at an early stage (50). *Vis-à-vis* public health workers should also be vaccinated and mass vaccination programs should be undertaken through standardized and coordinated efforts (51).

The following section describes the various types of vaccines and vaccine platforms which are being explored for the development of a successful EBOV vaccine.

# Inactivated Vaccines

Even though inactivated vaccines suffer with the problem of reversion to virulence due to inadequate viral inactivation, various strategies have been constantly explored in developing safe and potent non-replicating vaccine candidates for combating the EBOV infection (52). Both heat- and formalin-inactivated EBOV have been found protective against EBOV infection in a guinea pig model. Inclusion of inactivated vaccine with EBOV E-178 along with interferon (IFN) and immune plasma saved the life of a scientist working on EBOV (53). The protective efficacy of liposome-encapsulated irradiated EBOV, tested in a mouse model, was 100%. However, these viral particles failed to protect NHPs (54). This suggests that murine model is excellent for evaluating vaccine efficacy, but the level of protection might be different in different species and, hence, it is essential to test vaccines in NHPs before proceeding to clinical trials in humans. Heat-, formalin-, or gamma irradiation-killed EBOV vaccines have been found ineffective against EBOV disease; thus, the novel effective vaccine is essentially required (55).

# DNA Vaccines

In DNA vaccines, plasmids are used to express immunogenic antigens. This is an attractive vaccine approach because of the ease of production and simplicity. In addition, DNA vaccine induces both humoral and cellular immune responses. A threeplasmid DNA vaccine comprising the transmembrane-deleted GP sequences from EBOV species Zaire and SUDV-Gulu as well as nucleoprotein (NP) sequence from EBOV was tested in healthy adults. The vaccine was well tolerated, and both CD4<sup>+</sup> and CD8<sup>+</sup> T cell responses were elicited (56). An EBOV GP DNA vaccine designed on a consensus alignment of GPs (from strains obtained during 1976–2014), delivered intramuscularly and then electroporated, elicited a strong T cell response, and protected 100% of experimental mice from lethal challenge with EBOV (57). The DNA from three strains of EBOV was used to prime human volunteers and boosted with attenuated adenovirus, which acted as delivery vehicle for EBOV DNA into antigen-presenting cells, induced significant humoral- and cell-mediated immune (CMI) responses (58). Intramuscular inoculation of the DNA vaccine through electroporation with DNA plasmid containing codon-optimized GP genes of EBOV elicited high levels of IgG and a strong CMI response (measured by IFN-γ ELISpot assay) in cynomolgus macaques (59).

Though the preliminary trials using DNA constructs have provided the acceptable safety profiles, the development of low immune titer for a shorter window necessitates repeated vaccinations to overcome this problem. Thus, the use of a potent vaccination regimen based on DNA vaccine platforms does not appear logical for a large population (60, 61).

# Virus-Like Particles

Ebola VLPs (EBOV-VLPs or eVLPs) are generated from the expression of viral transmembrane glycoprotein (GP) and structural matrix protein (VP40) in mammalian cells, which undergo self-assembling and budding from host cells and display morphological similarity to infectious EBOV particles (47). Baculovirusderived eVLPs comprising GP, VP40, and NP of EBOV have been found to induce human myeloid DC maturation, suggesting their immunogenicity. Baculovirus-generated VLPs were able to elicit similar levels of protection as 293T cell-derived VLPs and showed protection against virus challenge in a dose-dependent manner (62). Nano-VLPs, produced by sonication of VLPs and filtering to have a mean diameter of approximately 230 nm, increased their thermostability. Unlike native VLPs where GP protein is denatured in a solution by heating, the nano-VLP maintained the conformational integrity of the GP protein at temperature up to 70°C and could confer protection in a mouse model (63). VLP containing only VP40 was sufficient to protect mice from EBOV infection. VLP injection leads to an enhanced number of natural killer (NK) cells, which play a crucial role in innate immune protection against lethal EBOV. NK cell protection is dependent on perforin, but not recombinant viral vector vaccines on IFN-gamma secretion (64).

Ebola virus VP40 and GP have been demonstrated to interact with the host protein, BST2, and are associated with viral infections by trapping the newly assembled enveloped virions at the plasma membrane in the infected cells, ultimately induce NF-κB activity. The effects of EBOV GP1,2, VP40, and BST2 converge on an intracellular signaling pathway leads to neddylation, resulting in the additive response with respect to the induction of NF-κB activity. Exploring the dynamics of this interaction could provide targets for vaccine developments and therapies that can modulate the inflammatory response during EVD (65).

Quantitation of EBOV antigenic particles using proteomic assays like liquid chromatography high resolution mass spectrometry method can be employed for determining the batch quality of vaccine constructs as well as in optimizing the dosages by assessing the amount of GP1 needed to confer effective protection (47).

It is to be noted that though anti-EBOV antibody can mediate effective protection, VLP-vaccinated murine models were shown to survive the EBOV challenge in the absence of detectable serum anti-EBOV antibodies (66). It could also be revealed that adjuvant signaling may circumvent the necessity for B-cell immunity in conferring protection against EBOV. These studies can be valuable for the future characterization, development, and optimization of effective EBOV vaccine candidates (66).

# Virus-Like Replicon Particles (VRPs)

The VRPs are the alternative to live-attenuated vaccines. The use of VRPs eliminates the risk of reversion to the original pathogenic form of live vaccine strains. To generate VRPs, generally filoviruses or alphaviruses are required. Here, while keeping the genes essential for replication, viral structural genes are deleted from full-length genomic cDNA clones. Viral structural genes are replaced with alternative gene(s) coding for an immunogen. Such replicons are able to replicate and transcribe upon transfection in competent cells. The resulting VRPs are able to infect cells only for one cycle. Because of the lack of structural genes, viral progeny are not formed. Viruses such as Venezuelan equine encephalitis virus (VEEV) can be used for production of EBOV antigen instead of structural proteins for the replicon vector. Thus, such vaccines are also quite safe (67). The gene inserted is typically GP, the main target of neutralizing antibodies. VRPs expressing EBOV VP24, VP30, VP35, and VP40 have been evaluated for their protective efficacy in a mouse model, but these were found not to be as protective as EBOV GP and NP antigens. VEEV replicons containing GPs from both EBOV and SUDV showed promising results in cynomolgus macaques after administration of a single dose. Here, two VRPs were constructed that contained the GP of EBOV or SUDV. The animals intramuscularly injected with both of the VRPs, survived viral challenge without exhibiting any clinical signs. The final results indicated that VRP-EBOV GP was able to confer cross-protection against SUDV, whereas VRP-SUDV GP was unable to provide complete protection against EBOV-Zaire challenge (68).

Recently, Ren et al. (69) constructed an alphavirus Semliki forest virus based recombinant replicon vector DREP for efficient and unchecked *ex vivo* co-expression of EBOV GP and VP40. Active immunization with recombinant DREP vectors possessing GP and VP40 induced cellular and humoral immune responses in murine model against EBOV antigens. This path breaking approach may provide key insights and strategies for designing further effective vaccines to contain EBOV permanently.

# Reverse Genetics System for EBOV Vaccine

A full-length recombinant EBOV infectious clone was constructed using cDNA. By employing reverse genetics method, viable but replication incompetent virus lacking entire VP30 ORF was constructed. The resultant EbolaΔVP30 is biologically contained and replication deficient, until VP30 is provided extraneously. Virus replication in cell culture was allowed by growing the virus in Vero cell line that stably expresses VP30, designated VeroVP30 (70). The safety of EbolaΔVP30 has been evaluated in mice and guinea pig model and was able to protect from lethal infection (71). The EbolaΔVP30 virus inactivated by using hydrogen peroxide protected NHPs after a single immunization. To avoid any incidence of potential recombination events that might result in regaining the replicative efficiency, the vaccine candidate was inactivated by hydrogen peroxide, that creates nicks and breakages in single- or double-stranded DNA or RNA and the virus is completely inactivated while retaining antigenic determinants unaffected (72).

# Recombinant Viral Vector Vaccines

Engineered viruses are gaining popularity because of their ability to efficiently induce CMI responses (a major part of adaptive immunity along with humoral response), as the antigen is expressed and processed in the cytoplasm. Replicationcompetent rVSV and chimpanzee adenovirus 3 (ChAd-3/cAd3) are the most efficient platforms for designing new vaccines (73). A recombinant vesiculovirus vector containing EBOV GP region (rVSVΔG/EBOVGP) was found to be highly effective after a single injection in NHPs (74, 75). The vaccine evaluated in pigs showed no disease development and no viral shedding. This indicated that the vaccine could be utilized for herd immunization and it also suggested the safety of the live-attenuated rVSVΔG/ EBOVGP vaccine (76). Recently, this rVSVΔG/EBOVGP vaccine was evaluated in a randomized double blinded placebo phase III trial in 1,197 humans. There were no adverse effects or death following vaccination, supporting its use as a vaccine (77). The vaccine protected immunocompromised rhesus macaques that had a high number of CD4<sup>+</sup> T cells (78). The rVSVΔG/EBOVGP vaccine was also studied for its efficacy as a therapy in rhesus monkeys after exposure to EBOV-Makona. This vaccine showed minimal prophylactic efficacy after exposure (79). Efficacy trials initiated to test the rVSV-vectored EBOV vaccine showed greater efficiency at the time of EVD outbreak, if deployed following the strategy of ring vaccination (80).

Another recombinant vaccine (VSV based), i.e., rVSV-Zaire EBOV has been shown to provide substantial protection. From 10th day of vaccination with this vaccine, no report of any disease was documented, which proved efficacy and effectiveness of rVSV-vectored vaccine in preventing EVD (81). It is interesting to note that seroconversion has been noticed in recipients of recombinant VSV-EBOV (rVSV-EBOV) vaccine by the end of fourth week (i.e., by 28 days) against the Kikwit strain glycoprotein (82). Another recombinant vaccine *viz*., rVSV-EBOV vaccine was tested as a candidate vaccine. This particular vaccine is under trial in human (phase II/III). It provides protection against only EBOV and is clinically efficient in the clinical set up of ring vaccination format (38, 83, 84). EBOV and SUDV glycoproteins have been assimilated into a cAdVax vector (adenovirus-based vaccine). In mice, this vaccine has provided full protection (85, 86). During recent outbreak in Democratic Republic of the Congo (DRC), rVSVΔG-EBOV-GP is being used for ring vaccination in the affected area. Though the vaccine is yet not approved and still under investigation.

In Russia, clinical trial of a vaccine, GamEvac-Combi, has been performed and has been approved to enter in phase III clinical trial (87). The vaccine GamEvac-Combi contained two heterologous expression systems. One is live-attenuated rVSV and the second is a recombinant replication-defective adenovirus type-5 (Ad5). Both the vectors are expressing the same glycoprotein. The rationale to use a combination of two vectors expressing glycoprotein of EBOV is that widely present preexisting immunity to Ad5 limits the use of Ad5 and also a negative correlation between EBOV glycoprotein-specific immune response and preexisting antibodies to Ad5 has been reported (88). Hence, prime immunization with VSV vectored vaccine and then boosting with AD5 vectored vaccine might contribute in compensating negative impacts of preexisting immune response to Ad5. This heterologous vaccine evoked glycoprotein-specific immune response in 100% volunteers on day 28th. Also, the vaccine is well tolerated and did not significantly altered the body physiological parameters and vital organs. In Liberia, Sierra Leone, and Guinea; the VSV and ChAd3 vectored vaccine are in focus (89).

Another study in mice models has reported that the adoption of a heterologous prime-boost vaccine strategy can result in a durable EBOV-neutralizing antibody response. The chimpanzee serotype 7 adenovirus vectors expressing EBOV GP (AdC7-GP) was used for priming and a truncated version of EBOV GP1 protein (GP1t) was used for boosting. Vaccination response studies showed that AdC7-GP prime/GP1t boost strategy was more potent in generating a sustained and strong immune response as compared to using an individual vaccine construct (90).

Replication-defective recombinant chimpanzee adenovirus type 3-vectored EBOV vaccine (cAd3-EBO) elicited both cellmediated and humoral immunity in NHPs. A vaccine dose of 2 × 1011 particle units was found sufficient to induce protective immunity in the NHPs and to eliminate the effect of prior immunity to cAd3 (91). Recombinant VSV vaccine expressing EBOV GP and A/Hanoi/30408/2005 H5N1 hemagglutinin (VSVΔG-HA-ZGP) protected mice against challenge with both viruses and also cross-protected against H5N1 viruses (92).

The utility of adenovirus-vectored EBOV vaccines is limited with preexisting anti-adenoviral antibodies, which significantly lower the GP-specific humoral and T cell responses (88). Six mutations in the genome of MVA virus restrict its host specificity and make it unable to replicate in mammalian cells. A randomized study of a multivalent MVA vaccine encoding GPs from EBOV, SUDV, Marburg virus (MARV), and TAFV NP (MVA-BN-Filo) conducted in 87 participants resulted in no fever. The quadrivalent vaccine formulation has demonstrated the boosting up of both cellular and humoral immune responses against EBOV to several folds (93). Twenty-eight days after immunization, GP-specific IgG was detected with EBOV-specific T cell responses (94). EBOV GP and TAFV NP expressed in an MVA platform assembles into VLPs. Heterologous NPs enhanced VLP formation and offered GP-specific IgG1/IgG2a ratios comparable to those of MVA-BN-Filo (95).

Recombinant cytomegalovirus expressing EBOV GP was found to evoke protective immunity in rhesus monkeys challenged with EBOV (79). Baculovirus-expressed EBOV-Makona strain GP administered with Matrix-M (saponin adjuvant) showed better immunogenicity. Administration of Matrix M-adjuvanted vaccine resulted in increased IgG production and CD4<sup>+</sup> and CD8<sup>+</sup> T cell production (96). A human parainfluenza virus type 3-vectored vaccine expressing the GP of EBOV (HPIV3/EboGP) was developed as an aerosolized vaccine, and studies in Rhesus macaques showed 100% protection against challenge with EBOV (97).

Adenovirus 26 vectored glycoprotein/MVA-BN vaccine has recently passed the phase I trial (94). In the European countries including United Kingdom and United States, for the purpose of clinical trial, administration of ChAd-3 vectored vaccine has been adopted. This vaccine expresses the EBOV GP and is available in monovalent and divalent forms (91, 98).

Ebola vaccine potency trials employing replication defective adenoviral vectors (rAd) encoding EBOV GP have come up with promising results in NHP models. Based on such studies, multiple mutant glycoproteins were developed (such as glycoprotein with deleted transmembrane domain) which offers reduced *in vitro* cytopathogenicity but possessed reduced vaccine-mediated protection. In contrast to this, a point mutated glycoprotein has been reported to offer minimal cytopathogenicity and appropriate immune protection even with a two logs lower vaccine dose (99).

# Plant-Based Vaccines and Antibodies

Viral antigens, including GP, VP40, and NP, elicit protective immune responses. ZMapp, the cocktail of antibodies being used to treat EBOV, is a biopharmaceutical drug. To note, the component antibodies in ZMapp are manufactured in *Nicotiana benthamiana* using a rapid antibody manufacturing platform. Gene transfer is mediated by a viral vector, and the expression is transient. *N. benthamiana*-derived antibodies produced stronger antibody-dependent cellular cytotoxicity than the analogous anti-EBOV mAbs produced in a mammalian Chinese hamster kidney cell line (100). Phoolcharoen (101) expressed a GP1 chimera with the heavy chain of 6D8 mAb, forming an immune complex that was co-expressed with the light chain of the same mAb in leaves of tobacco plant. The ammonium sulfate-precipitated purified antibodies, along with poly(I:C) adjuvant, a synthetic analog of double-stranded RNA capable of interacting with toll-like receptor (TLR)-3, was found to elicit strong neutralizing anti-EBOV IgG. In addition, the immune complex along with poly(I:C) adjuvant was capable of stimulating a Th1/Th2 response. The experiment suggested the potential application of plant-produced Ebola immune complexes as vaccine candidates. EBOV VP40 was expressed in tobacco plants, and a mouse immunization study showed results that suggested this approach can be used to produce an EBOV vaccine (102).

The utility of plants as bioreactors for the bulk production of ZMapp could be considered to meet the required demand. The glycosylation pattern of mAbs may alter their efficiency and bioactivity, including their binding with the antigenic epitope. Several glycoforms of EBOV mAb13F6 have been prepared using a magnICON expression system. These glycoforms have humanlike biantennary N-glycans with terminal N-acetylglucosamine, resulting in a structure similar to that of human mAbs. Hence, these are beneficial for humans (103).

Both RNA and DNA viruses have been modified to serve as plant-based vectors for the expression of heterologous proteins. Bean yellow dwarf virus, a single stranded-DNA virus, can replicate inside the nucleus of plant cells using their cellular machinery. A vector containing deletions in the coat-encoding genes and gene for the desired antigen may be inserted to form an expression cassette. The delivery of vectors to plants is *Agrobacterium*mediated (23). mAbs against EBOV are produced by the process of agroinfiltration. In this context, it is noteworthy that lettuce acts as a very good host for the process of agroinfiltration. In lettuce cells, *Agrobacterium tumefaciens* has been used for delivering viral vectors (104). Neutralizing and protective mAb6D8 against EBOV has been expressed at a concentration of 0.5 mg/g of leaf mass. This quantity is similar to that generated in magnICON expression system (105). The plant-derived approach to vaccine development is attractive because of the large amount of transient proteins that can be expressed, with the potential for use during high demand for therapeutics and prophylactics (106). Advances in the field of vector expression like plant transient expression system and associated host cell engineering and manufacturing processes paved way for developing biopharmaceutical proteins and therapeutics in commercial basis. The great potentials of such novel approaches have been exploited for evolving therapeutics to counter emerging pandemics of EBOV and influenza that is evidenced from the production of experimental ZMapp antibodies (107).

An overview of various types of vaccines for countering EVD is presented in **Table 1** and depicted in **Figure 1**.

# ADVANCES IN DEVELOPING DRUGS AND THERAPIES AGAINST EBOV

Momentous leap has been witnessed toward the designing of efficacious EBOV drugs and therapeutics during the short span of only few years, even though the efficacy of several biologicals and vaccines were evaluated during the recent West African outbreak, it remained elusive to ratify a licensed EBOV disease treatment regimen (123).

Management of suspected or confirmed EVD patients includes quarantine, symptomatic, and supportive treatments, including fluid replacement, electrolyte imbalance correction, treating complicated infections, and preventing shock (124). For mitigation of the huge fluid loss and resultant hypovolemia, oral rehydration solutions should be provided adequately and if required anti-diarrheal and anti-emetic drugs need to be administered (125). Brincidofovir, a drug used to treat dsDNA viruses such as adenovirus, herpesviruses, orthopoxviruses, papillomavirus, and polyomaviruses, was approved for emergency treatment of two patients with EBOV infection; however, the clinical efficacy of the drug is unknown (126). Many drugs are being tested to identify specific antiviral drugs to treat EBOV, and new drug candidates are being developed by researchers worldwide (26, 127, 128).

Favipiravir (T-705), an antiviral drug found useful in treating influenza, has been studied and found effective against EBOV (129–131). Insertional mutagenesis, a high-throughput method to identify genes responsible for virus replication, can be used to develop drug candidates (132). Molecular docking experiments with EBOV GPs can be used for drug designing and the development of therapeutics (133, 134). Novel flexible nucleosides called fleximers were found to be effective against recombinant EBOV in Huh7 cells (135). Ribavirin antiviral can be recommended for the treatment of EBOV, since in mouse and monkey models, treatment with ribavirin delayed the death and increased survival rate (136). However, adverse effects associated with its use may limit ribavirin use (137). Lamivudine, an anti-retroviral drug, has been tested by Liberian doctor on 15 EBOV patients with survival of 13 patients (138). However, study by Cong et al. (139) found no survival benefits in Guinea pig model. Similar results were obtained by Hensley et al. (140), with no significant antiviral Table 1 | Vaccines for treating Ebola virus disease.


(*Continued*)

EBOV Vaccines, Drugs, and Therapies

#### TABLE 1 | Continued


EBOV Vaccines, Drugs, and Therapies

activity of lamivudine against EBOV in Vero E6 cells. Hence, the use of lamivudine may not be advocated.

The docking of VP40, VP35, VP30, and VP24 has been achieved using small molecules belonging to the class of flavonoids and derivatives. Gossypetin and Taxifolin (the two flavonoids of top ranks) showed higher docking scores for every EBOV receptor (141). A virtual analysis of more phytochemicals could help to identify plant-derived products with comparatively higher efficacy and lower toxicity. Adenosine nucleoside analogs such as BCX4430 have been found effective against EBOV in a mouse model. GS-5734, also a nucleoside analog, was found effective against EBOV in a NHP model (142). Using molecular dynamics simulations, graphene sheets are found to associate strongly with VP40 (matrix) protein of EBOV and disrupt VP40 hexamer–hexamer association, crucial to form virus matrix, thereby graphene and similar nanopolymers may be used as therapy or at least disinfectant to reduce the risk of transmission at time of epidemic (143).

The potential of retro-type drugs (molecules that block the retrograde trafficking of bacterial and plant toxins within mammalian cells) must be explored for designing novel therapy against filovirus (144). Retro-2 along with its other two derivatives, Retro-2.1 and compound 25 could effectively block EBOV and MARV progression *in vitro*. The derivatives were shown to be more potent inhibitors of filoviral penetration, replication, and progression when compared with their parent compound, as evidenced by pseudo-typed virus assays (144).

# EBOV Entry and Inhibitors

The cell entry of EBOV involves virus binding to the cell surface receptors followed by internalization through macropinocytosis, processing by endosomal proteases, and transport to Niemann– Pick C1 (NPC1; an internal receptor for EBOV) containing endolysosomes. Phosphatidylinositol-3-phosphate 5-kinase is essential for maturation of endosome, a critical step to the EBOV infection (145). *In vitro* studies using apilimod, an antagonist of phosphatidylinositol-3-phosphate 5-kinase, showed inhibition of EBOV by blocking the viral particle trafficking to NPC1 containing endolysosomes (146). IFNs are natural antivirals, and type I (IFN-α/β), particularly, is being widely used for the treatment of viral diseases. Type I IFN-α2b has been evaluated for the treatment of EBOV; however, IFN-α2b was not successful, as only delayed death but could not prevent mortality in EBOV-exposed monkeys. IFN-γ reduced the mortality rate in mice when administered either before or after EBOV infection (147), suggesting its promise as a prophylactic and/or therapeutic drug for use in EBOV infections. As IFN-γ has already been used to treat certain chronic medical conditions and has been approved by the FDA, it can be readily adapted for use against EBOV infections.

Silvestrol extracted from *Aglaia foveolata* was found to inhibit replication of EBOV and has been suggested as a therapeutic drug to treat acute EBOV infection (148). Supportive treatments like oral rehydration therapy are recommended for children under 5 years of age (149). Web-based identification of therapeutic agents indicated that a single siRNA can inhibit mRNA transcription of three species of EBOV, whereas 75 siRNAs can inhibit at least two species of EBOV. The web server Ebola VCR has been developed, with details available for the development of suitable therapeutic agents (150). Numerous treatment options for EVD are discussed below.

Ebola virus possesses only one surface protein and is responsible for both the receptor binding and fusion of virus-to-host cell endosomal membrane. EBOV glycoprotein binds with lectin receptor DC-SIGN (151). The infection initiated with the binding of EBOV glycoprotein to lectin receptors and internalization of virus majorly through macropinocytosis and as alternative mechanism through clathrin-dependent endocytosis (152). In the low pH of endosome, cysteine proteases including cathepsin B and L proteolytically cleave GP. Possibly this proteolytic cleavage exposes the putative receptor-binding region that interacts with NPC1, a receptor facilitating the filovirus entry (153). TIM-1 receptors directly interact with phosphatidylserine on the viral envelope, suggestive of GP independent virus attachment onto the cells. In poorly permissive cells, EBOV infection enhanced by exogenously expressed TIM-1 by 10- to 30-folds (154). Other phosphatidylserine interacting proteins like TIM-4 and Axl (a receptor tyrosine kinase) also have been demonstrated to enhance the infection of several enveloped virus. A benzylpiperazine adamantane diamide-derived compound obtained after screening of a library of small molecules, targets endosomal NPC1, and inhibit infection by VSV particles (VSV) pseudotyped with EBOV GP (155).

Tetrandrin is a potent drug that inhibits the EBOV entry into the cells (156). Two estrogen receptor drugs, clomiphene and toremifene, have been reported to hinder EBOV infection in mice by blocking cell entry and fusion with host cells (157). Amiodarone, an ion channel blocker, has been found to inhibit EBOV entry into cells (158, 159). Dendrimers and fullerene C60 have unique symmetrical properties and were recently found effective in inhibiting EBOV entry *in vitro* (160). Clarithromycin, an antibiotic, inhibits the release of calcium (stimulated by nicotinic acid adenine dinucleotide phosphate) from lysosome and exhibit anti-EBOV activity. Alike clarithromycin, posaconazole, an anti-fungal agent also shows similar anti-EBOV activity. In addition, it also inhibits the functions of NPC1 protein and acid sphingomyelinase activity. Both drugs, i.e., clarithromycin and posaconazole, ultimately inhibit the entry of EBOV into the host cell (161). The drug 5-(*N*-ethyl-*N*-isopropyl) amiloride inhibits the process of macropinocytosis (a process required for uptake of large filamentous virions, like EBOV) and thus interferes with the viral entry into the cell. Compounds like MLS000394177 and MLS000733230 also inhibit the viral entry into cells (162).

*Prunella vulgaris,* a Chinese herb, was found to inhibit EBOV entry into cells, using an EBOV-GP-pseudotyped-human immunodeficiency virus (HIV)-1-based vector system (163). Pseudovirions containing EBOV-GP were used for screening of the Prestwick chemical library, which contains 1,200 FDA approved drugs. The assay was based on cell entry of HIV-1-based surrogate in 384-well format. Twenty chemicals were found to inhibit more than 80% entry and 16 out of them were identified as G protein-coupled receptor (GPCR) antagonists, which target a range of GPCRs including adrenergic receptors, 5-HT (serotonin) receptors, histamine receptors, muscarinic and acetylcholine receptors. The time-of-addition studies suggested that EBOV entry is stopped at level of initial attachment prior to fusion of virus and cell membrane (164). Quercetin 3-β-*O*-d-glucoside (Q3G), a flavonoid derivate, was found to protect mice against EBOV challenge by targeting viral entry (165).

During replication of EBOV, surface GPs undergo proteolytic cleavage in the endosome by several proteases, including cathepsin B (CatB) (166). Thus, proteases may be a good target for the inhibition of EBOV replication. One study showed that, using a synthetic serine protease inhibitor, nafamostat mesilate (NM), caused a reduction in CatB release in rat pancreases. NM was also found to have anti-coagulant properties, which would also be useful in EBOV infections, as EBOV causes disseminated intravascular coagulation. Thus, this drug should be examined in clinical trials to be approved for the treatment of EVD (167). Chemically modified human serum albumin with 3-hydroxyphthalic anhydride (HP-HSA) has been demonstrated with the potential of a therapeutic candidate in resisting the EBOV infection (168).

## Transfusion of Convalescent Blood/Serum

Convalescent serum by definition contains immunoglobulins IgM and IgG but is devoid of red blood cells and clotting factors. Transfusing convalescent whole blood and convalescent plasma from disease survivors has been found to neutralize EBOV and reduce its load; thereafter, the immune response of the patient can provide protection against EBOV (169, 170). The use of whole blood and convalescent serum was approved by the World Health Organization (WHO) during critical EBOV conditions (171, 172). Screening of plasma is needed to rule out the presence of residual EBOV RNA and other blood-transmitted pathogens such as HIV, hepatitis B virus, and hepatitis C virus. Protection is conferred in NHPs through antibody therapy (post-exposure). In humans, this has ultimately paved the way for filovirus therapy by the use of polyclonal/mAb (approved by Food and Drug Administration) (173).

Valuable emergency therapeutics for the treatment of EBOVinfected persons include passive immunization with neutralizing antibodies by the transfer of sera from individuals recovering from EVD (174, 175), although it is not considered to render 100% protection especially after exposure (3 days post-exposure) to EBOV (e.g., Zaire Ebolavirus Makona) (176). Precise immunoglobulins retrieved from equine serum against EBOV were found safe and effective as prophylactic therapy in non-allergic patients (177). Recently developed mAb-based treatments for EVD include mAb114 and MB-003, ZMAb, ZMapp, and MIL-77E cocktails (25). ZMAb, consisting of three murine mAbs (1H3, 2G4, and 4G7), administered at a dose of 25 mg/kg three times, completely protected cynomolgus macaques against EVD. Administration of ZMAb with adenovirus-vectored IFN-α resulted in 75 and 100% survival of cynomolgus and rhesus macaques, respectively (29). mAbs that bind to the base of GP (4G7 and 2G4) are neutralizing antibodies, whereas mAbs that bind to the glycan cap (mAb114, 1H3, and 13C6) are nonneutralizing antibodies. The chimeric human mAbs 13C6, 6D8, and 13F6 possess the variable region from mice and Fc region of human; mAbs 13C6 and 6D8 neutralize EBOV in the presence of complement proteins (24). By repeated immunization of mice with glycoproteins of filovirus, generation of pan-EBOV-specific (as well as pan-filovirus) mAbs have been obtained. These pan-EBOV mAbs have shown reaction with RESTV, SUDV, and other viruses (178).

The components of ZMapp are mAbs (chimeric), *viz*., c13C6 from MB-003 (already known cocktail of antibody) and c2G4 as well as c4G7 from ZMab (different cocktail of antibody). This drug reversed clinical signs in 100% of rhesus macaques, even when administered as late as 5 days after EBOV exposure (30). Use of ZMapp, humanized-mouse antibodies, as a therapeutic agent has shown promise in NHPs (30, 84, 179) and it is a WHOapproved treatment regimen for EVD. Recently, a baby born to an EBOV-infected mother was found positive for EBOV on the first day of life. The baby was treated with ZMapp and the broad spectrum antiviral GS-5734, and, on day 20, the baby was found negative for EBOV (180). MB-003 is another mAb cocktail which is found to be effective in NHPs against variants of EBOV that are resistant to ZMapp (181).

The mechanism of action of mAbs is that they identify the inter-protomer epitope of the GP fusion loop, which is essential for viral membrane fusion, and also neutralize the entry of virus (182). Although several mAbs are available that can neutralize EBOV, there are few mAbs that can neutralize GPs from different EBOV species. In a study by Duehr et al. (183), a panel of eight murine mAbs derived from animals immunized with *Zaire ebolavirus* was evaluated. The mAbs were tested for binding breadth using a set of recombinant surface GPs from RESTV, TAFV, BDBV, EBOV, SUDV, and MARV. Of the eight, two mAbs (KL-2E5 and KL-2H7) showed binding ability. These two mAbs did not neutralize EBOV; however, they protected mice from infection with a VSV expressing the *Zaire ebolavirus* GP. Duehr et al. (183) also suggested that Fc-FcR interactions are responsible for the protection of mice in the absence of neutralization. Although ZMapp was found to be effective against Zaire EBOV, it has not shown cross-protection against other species of EBOV. FVM04 (a mAb) has shown cross neutralizing activity against SUDV. So, it can be used to replace one of the components of ZMapp, thereby increasing the range of protection against SUDV, ultimately leading to generation of cross-protective mAbs cocktail (184).

One Fab, KZ52, obtained by panning of phage display library, was derived from the bone marrow of an EVD survivor. Fab KZ52 exhibited 50% neutralization at a concentration of 8 nM (185). The mAb KZ52 protected guinea pigs from lethal *Zaire ebolavirus* challenge; however, when an experiment was carried out in rhesus macaques, the antibody failed to protect animals prophylactically and did not inhibit viremia (186). EBOV GP is processed by cathepsins, and the cleaved GP fuses with host cells to form a fusion pore, a passage for the EBOV genome to enter the cytosol for replication. Human mAb KZ52 and monkey mAb JP3K11 bind to conformation-dependent epitopes of GP. KZ52 is directed to bind a conformational non-glycosylated epitope at base of GP and a total 23 residues of GP residues remain in contact with antibody. Out of 23, 15 are contacted through van der Waals interactions and remaining 8 through direct hydrogen bonds (187). At 0.4 µg/ml dose, KZ52 lead to 50% neutralization. KZ52 protective efficacy is due to inhibition of cathepsin mediated cleavage of GP (23).

Exploring the synergistic effect of different pairs of neutralizing and non-neutralizing anti-EBOV mAbs could provide 100% protection in mice, revealing the scope of this approach in designing and developing immunotherapeutics and vaccines (188).

Bispecific Trojan-horse antibodies neutralizing other filoviruses have been found to provide protection in mice from multiple EBOV infection (189). Cell-penetrable human scFvs (HuscFvs) (transbodies) that bind to EBOV VP40, a matrix protein pivotal for viral assembly and budding, produced by phage display technology, revealed inhibition of the EBOV-like particles (VLPs) egress from hepatic cells (190). These transbodies were effective in blocking viral assembly and budding within the cells as they bind to several cationic patches in the VP40 C-terminal domain. The transbodies inhibit the function of VP40 by additional mechanisms also; such as binding to N-terminal domain and L-domain peptide WW binding motifs, suggesting the potential of these transbodies as direct acting anti-EBOV agents in future (190). Cell-penetrable HuscFvs specific to a highly conserved interferon-inhibitory domain (IID) of VP35 of EBOV inhibited the VP35 biofunctions in the EBOV replication cycle including polymerase cofactor activity and host IFN–antagonism by forming interface contact with residues of the first basic patch, the central basic patch, end-cap, and residues important for IID multimeric formation for dsRNA binding (191). The cell-penetrable small antibody fragments (HuscFvs) or superantibodies [the term coined by Kohler and Paul (192)] can cross the membrane of all cells but get accumulated intracellularly only where the target antigen is present. Thus, disappearance of the superantibodies from the blood circulation does not imply that they are eliminated from the body. The transbodies to the highly conserved EBOV VP40 and VP35 should be evaluated further using authentic EBOV in animal models of EVD and clinical trials before they can be considered a broadly effective and promising alternative to existing treatment approaches for EVD.

Recently, three mAbs produced in tobacco plants that target the EBOV GP were tested and showed good results in humans (193). Human mAbs against BDBV GP were isolated from patients who survived during the 2007 Uganda outbreak. These mAbs were found to have a neutralizing effect against multiple EBOV species, suggesting the possibility of the use of single mAbs as cross-protecting antibodies (194). Another investigation showed that EBOV GPs were conserved across different EBOV species. ELISA revealed that four mAbs namely S3, S12, S17, and S33 were found to show cross-reaction with GPs of five different species of EBOV (195). The discovery of cross-protective antibodies can aid in the development of therapeutic strategies for treatment of EBOV disease (196). In another study, 349 EBOV GP mAbs were isolated from survivors of EVD in an outbreak in Zaire, and 77% of the mAbs were found to neutralize EBOV (197). Three mAbs of EBOV-GP (Q206, Q314, and Q411) were isolated during the West African EVD outbreak in 2014. Recognition of the novel epitopes has been performed for Q206 and Q411, wherein these mAbs were found protecting mice against EBOV (198). A therapeutic vaccine based on mAbs has been proposed to sufficiently resolve replication of invasive EBOV, even if administered as a single dose 4 days post-infection (199). Non-neutralizing mAb 5D2 or 7C9 expressing adeno-associated virus (AAV), consistently released mAb in body and was found 100% protective against mice adapted EBOV strain. Neutralizing mAb 2G4 conferred 83% protection and a cocktail of these two mAbs provided 100% protection when given 7 days prior to infection and sustained protection when immunized animals were challenged 5 months post AAV-mAb immunization (200).

Potential limitations of mAb-based therapies include the requirement for high doses and mAb mixtures that are outbreakspecific owing to constant viral evolution. Furthermore, epitope mutations could reduce efficacy of the therapeutic mAbs used. Hence, these limiting factors need to be taken care of accordingly with mAbs usages.

## EBOV Gene Expression Inhibitors

Viral gene expression is dependent on host cell machinery and is critical for virus replication. A conserved guanine-rich sequence in the EBOV L gene has been reported to assemble into quadruplex RNA, targeted by cationic porphyrin TmPyP4 that directs inhibition of the expression of L gene at the RNA level (86, 201). BCX4430 (a nucleoside analog) is a viral RNA polymerase inhibitor, and it has been found effective in protecting mice against lethal challenge of EBOV (202–204). In addition, double-stranded RNA binding protein 76 has been reported to inhibit EBOV polymerase activity (205).

Small molecular inhibitors needed for the synthesis of polyamine have been found to block the expression of EBOV gene. The eukaryotic initiation factor 5A (eIF5A) hypusination and spermidine (a polyamine) are essentially required for the replication of EBOV. However, if eIF5A hypusination is blocked, the gene expression of EBOV is inhibited which subsequently blocks the replication of the virus. Therefore, in-depth understanding of this mechanism at molecular level is essential for developing anti-EBOV drugs (205).

# Repurposed Drugs

It is time taking task to develop a new therapeutic against an infectious agent and till that time new therapy divulged; drug repurposing, *i.e.*, already existing drugs may be screened for their efficacy against pathogen. Owing to the lack of approved EBOV therapies, the screening of potentially efficacious drugs revealed that few of the drugs could be repurposed for EBOV treatment (206) (**Table 2**). Amiodarone, dronedarone, and verapamil, which are used for tachycardia, arrhythmias, and high blood pressure or angina, respectively, have been screened for their ability to inhibit the entry of filoviruses into cells and found efficacious in *in vitro* models (158). The use of statins, angiotensin-converting enzyme inhibitors, and angiotensin receptor blockers has also been suggested to attenuate EBOV infection (126). Phosphoinositide 3-kinases inhibitor LY294002 and calcium/calmodulin kinase (CAMK2) inhibitor KN-93 have been reported to reduce EBOV infection in Vero E6 cells. The p38 mitogen-activated protein kinase inhibitor SB202190 was shown to check virus-mediated cytokine storm, as studied in monocyte-derived DC of humans (207). Estrogen re-uptake modulators, *viz.*, toremiphene and clomiphene, although cause *in vitro* inhibition of the virus entry, but are not free from unwanted side effects like ocular adverse reaction (in case of clomiphene) and serious derangements of electrolytes (in case of toremiphene) at the higher doses. To overcome this, combination therapy is suggested while using such drugs (206). Brincidofovir, a cidofovir analog conjugated with a lipid, can prevent EBOV replication; however, its exact efficacy in an *in vivo* model needs to be determined (208). Because cyclophilin A (CypA) is not essential for EBOV replication, alisporivir, which inhibits the host protein CypA, has shown limited antiviral effects against EBOV strains (Makona, Mayinga) (209). Emetine, an anti-protozoal agent, and its desmethyl analog cephaeline have potently inhibited EBOV replication and cephaeline is well tolerated in patients than emetine (210).

Rosuvastatin, atorvastatin, and pravastatin have been reported to alleviate inflammation, reduce C-reactive protein and TNFα levels, and impede cholesterol-supported EBOV membrane biosynthesis (232). In EBOV infection, overexpression of the procoagulant tissue factor in monocytes and macrophages and participation of endothelial cells leads to an imbalance in coagulation. The use of recombinant nematode anticoagulant protein c2, an inhibitor of tissue factor-mediated blood coagulation, was found to improve survival of macaques from Ebola hemorrhagic fever, and hence suggested to act as a good treatment module targeting the disease development (228). Anti-malarial drugs such as chloroquine and its structural analogs (hydroxychloroquine, pamaquine, primaquine, and plasmaquine) also act as lysosomotropic agents, preventing endosomal/lysosomal acidification, and thus limiting certain viral infections (233). There are conflicting reports on the therapeutic effects of chloroquine in mouse, hamster, and guinea pig models of EVD. Chloroquine was found to inhibit virus replication in *in vitro* studies but failed to protect against EBOV infection and disease development in mice, hamsters, and guinea pigs (234, 235). Esomeprazole and omeprazole were also found to inhibit viral entry during *in vitro* studies but higher concentrations of these drugs may be required when to be used *in vivo* (236). During the EBOV outbreak in Liberia in 2014, a reduction in fever cases was observed following mass administration of malaria chemoprevention drugs (237).

Because of its competitive anti-heparin potential and interference with viral replication and entry into the cell, the antitrypanosomal agent, Suramin (Germanin or Bayer-205) has been proposed to treat EVD (229). A pyrazine carboxamide derivative namely Favipiravir (an anti-flu medicine), which was used earlier as an inhibitor of influenza virus replication, has been found useful in both therapy and prophylaxis during EBOV epidemic in West Africa (238–240). Favipiravir and the pyrazine carboxamide derivative T-705 showed positive results in treating patients with medium to high viremias, although these drugs were not found to be effective with very high viremias, but revealed acceptable results during EBOV infection in mouse (219, 221, 241).

The microtubule inhibitor drugs (vinblastine, vinorelbine/ navelbine, and vincristine), commonly used as anticancer agents,


Table 2 | Repurposed drugs used in Ebola virus disease therapy.

(*Continued*)

### TABLE 2 | Continued


have been found effective in inhibiting EBOV VLP entry into HeLa cells even at low concentrations (48–140 nM). Colchicine, a microtubule modulator primarily used for gout, has also been found to show anti-EBOV activity (225, 242).

Screening of 1766 FDA approved drugs and 259 experimental drugs revealed that Indinavir, an HIV protease inhibitor, may be effective in reducing the severity of EVD (230). The antiviral drugs including Maraviroc, Abacavir, Telbivudine, and Cidofovir could target the MTase domain of EBOV and inhibit the viral RNAdirected RNA polymerase (230). The Computational Analysis of Novel Drug Opportunities platform was recently developed to screen drugs approved by the FDA. Drugs like enfuvirtide, vancomycin, bleomycin, octreotide, lanreotide, somatostatin, and ubidecarenone (CoQ10) have shown higher activity against EBOV (231). Recently, virtual screening of several thousands of repurposing drugs from Drug Bank has been performed and ibuprofen was selected by realizing its possible inhibitory effect on EBOV infection. The drug has been found to show detectable antiviral effect in cell culture and can thus be used as a very useful molecular template for anti-Ebola viral drug development (243).

# Nucleotide Analog Prodrug

GS-5734 developed by Gilead Sciences falls under this category. Interestingly, clinical trials have been conducted and it has been found that the drug is effective in clearing virus from semen (181, 244). Administration of GS-5734 in rhesus monkey through intravenous route resulted in suppression of replication of EBOV. It is also important to note that in NHPs, this compound can provide protection post-exposure (245).

# Interferons

Interferons act as potent inhibitors of EBOV as has been proved through *in vitro* studies conducted involving various types of cells. IFN-β-1a treatment protected mice against a lethal challenge of EBOV (206). The virus clearance from the blood stream is enhanced by IFN β-1a leading to resolution of the symptoms of the disease at an early stage (246). Even though there is increased therapeutic usage of IFNs, but certain side effects are also associated with such treatment, *viz*., fever and myalgia, which must be kept in mind while opting for their use. Moreover, the occurrence of malaria additionally should be ruled out before initiating IFN therapy (206). Tilorone hydrochloride induces IFN response in mice and has been found effective against EBOV due to its action mainly mediated through pathway of innate immunity (IFN related) (247).

# Oligomer-Mediated Inhibition

RNAi and advanced antisense therapies have been reported to provide post-exposure protection against lethal filovirus infections (248). Small interfering (si)RNA targeting RNA polymerase L protein has shown inhibition of EBOV replication and promising results for its use as a post-exposure therapeutic option (249). siRNAs and phosphorodiamidate morpholino oligomers (PMOs) targeting the EBOV RNA polymerase (L) protein protected NHPs against EVD (248, 249). PMOs target co-polymerase protein VP35 and membrane-associated protein VP24 for this protection (250). Antisense PMO-based drugs like AVI 6002, AVI-6003, and LNPs/siRNA (TKM-Ebola) are in clinical trials. TKM-Ebola is a mixture of three siRNAs that target the L, VP24, and VP35 proteins of Zaire EBOV. In NHPs, it efficiently provided postexposure protection (249, 251). Although the results were promising, clinicians did not use much TKM-Ebola as it could lead to lethal overproduction of cytokines (a dangerous EBOV-induced inflammatory response) (232). A siRNA LNP product named TKM-130803 has been developed for EVD therapy. Although, the infusion of TKM-130803 at a dosage of 0.3 mg/kg/day through intravenous route to adult patients with severe clinical signs of EVD was comparable, it did not show an improved protection to the existing and classical controls (252). LNP-encapsulated short siRNAs protected 100% of rhesus monkeys exhibiting viremia and clinical illness (253). LNP encapsulation is also an effective drug delivery system (127, 253).

The inhibitors of hsa-miR-1246, hsa-miR-320a, and hsa-miR-196b-5p have been found to decrease EBOV GP cytotoxicity during *in vitro* studies; hence, miRNA technology can be used to develop useful therapeutics (254). Antiviral drug AVI-7537, targeting the VP24 gene of EBOV, has been shown to be efficacious in mice and monkeys (255). The EBOV-GP (cleaved) molecule acts as a ligand for NPC1, a transmembrane transfer protein. For entry of the virus into the target cell, an interaction between EBOV-GP and NPC1 domain C is necessary. Two small molecules, a sulfonamide (MBX2254) and a triazole thioether (MBX2270), have been identified as novel EBOV inhibitors suppressing EBOV infection in an *in vitro* model by blocking the entry of virus into target cell *via* inhibiting GP–NPC1 protein interaction (256). This strategy of targeting viral entry could pave way for the development of an anti-EBOV therapeutic agent.

A brief summary of investigated drugs/biomolecules and therapeutics to treat EBOV infection is presented in **Table 3** and depicted in **Figure 2**.

Of the note, a luciferase reporter system (EBOV-like particle/ EBOVLP) has been developed. It helps in evaluating the *in vivo* anti-EBOV agents, *viz*., vaccines and drugs without the necessity of biosafety level-4 facilities. The system appears suitable in studying the process of viral entry also (259). The molecular tweezer CLR01 has been recently reported to inhibit EBOV and Zika virus infection. CLR01 interacts with the lipids in the viral envelop but not with the cellular membrane, thereby it is having very less effect on viability of cells (270). This small molecule has earlier been shown to possess antiviral activity against HIV-1 and herpes viruses. Such broad-spectrum antiviral agents need to be further explored to develop an effective drug against EBOV.

Currently, priority is being given toward investigating various proteins in the host system and viral targets (druggable) (271). Further research works need to be strengthened to identify potent viral or host targets that can be exploited to treat EVD or inhibit EBOV. With advances in bioinformatics tools, it is now possible to identify the active sites of the viral targets which can be utilized as a critical step toward designing and discovering anti-EBOV drugs (272). The involvement of computational tools has widened our approach toward designing drugs (target based) widely. Computational approaches can also countervail the endemic burdens in development of drugs traditionally (271, 273). Large libraries can now be effectively screened, ultimately stimulating research activities toward identifying potent anti-EBOV drugs. Therapeutic applications of cytokines, recombinant proteins, RNAi technology/RNA interference, TLRs, avian egg yolk antibodies, plant-based pharmaceuticals, nanomedicines, immunomodulatory agents, probiotics, herbs/plant extracts, and others may be explored appropriately to combat EBOV, as these have been found promising against other viral pathogens (2, 249, 274–282).

# CONCLUSION AND FUTURE PERSPECTIVES

The 2014 EBOV outbreak has been marked as the most widespread lethal viral hemorrhagic attack and prompted a hasty leap in the researches for developing effective vaccines and therapies to counter it. In the case of Ebola, deviations in the touchstone drug/vaccine research approaches may be permitted by authorities to an appropriate extent, considering the devastating and alarming pandemic threat from the disease. In recent years, several therapies have emerged to tackle lethal EBOV infections. A plant-derived formulation of humanized mAbs: "ZMapp" has been used to treat some patients. However, the shortage of ZMapp supply warrants the evaluation and development of new mAbs.

#### Table 3 | Investigated drugs/biomolecules to treat Ebola virus (EBOV) infection.


EBOV Vaccines, Drugs, and Therapies

#### TABLE 3 | Continued


EBOV Vaccines, Drugs, and Therapies

Dhama et al.

#### TABLE 3 | Continued


*aDose not mentioned.*

EBOV Vaccines, Drugs, and Therapies

Various drugs have been repurposed to treat potentially lethal disease like EVD. There is a long list of repurposed compounds that have been evaluated as inhibitors of EBOV, including microtubule inhibitors, estrogen receptor and reuptake modulators, kinase inhibitors, histamine antagonists, and ion channel blockers. In-depth studies are still required to understand the pathogenesis and the role of different EBOV peptides, proteins, and antigens and host–virus interactions in EVD. There is also a need to develop economic and effective antivirals and vaccines against EBOV having approach/utility to any part of the world including resource poor countries.

Although the development of vaccines against EBOV began in 1980, there is still no effective vaccine available to prevent this deadly disease. Hence, the hunt for an effective vaccine is still on. Ebola VLPs play an imperative role in high-throughput screening of anti-EBOV compounds. Because five EBOV species have been reported, a polyvalent vaccine having immunogenic determinants such as GP from each of species would provide broader immunity; indeed, in nonhuman primate experimental studies with a DNA vaccine, this is commonly true. The best first-generation vaccine candidates for EBOV are rVSV and ChAd3, as reflected by their application in providing long duration protection during sporadic outbreaks. Various combinations of antigens from different species of EBOV may be explored to achieve higher protective immune response. The rVSV-based vaccine is being used in Democratic Republic of the Congo. Due to absence of preexisting immunity to VSV, it eliminates several drawbacks and safety concerns associated Ad5-based vaccine. Also, it has show long-term protection in several NHP models, it is an ideal vaccine platform to be used at time of outbreak. Together, the GamEvac-Combi vaccine also seems to be equally promising as it generated immune response in 100% volunteers.

In addition, mAbs with broad cross-reactivity that will neutralize all five species of EBOV are required to be developed and evaluated for prophylactic and therapeutic uses. Furthermore, effective antibodies may be engineered for homogeneity with human antibodies. Many nucleic acid-based modalities like siRNA, miRNA, and PMOs have been tested against EBOV and found functional. In the era of genomics, a computational approach may also be employed to screen large numbers of inhibitory molecules to safeguard human health. Available treatments within the disaster settings; mostly combination of appropriate supportive care and boosting of patient's immune responses, need to be optimized to ensure minimum research/ medical ethics being followed in such settings.

There is always scope for future investigations on the basis of clinical studies that are designed well and statistically supported. Maximum use of supportive therapy (MUST) should be introduced for studying the effects of new therapeutics. The side effects of newer drugs can also be revealed very efficiently by MUST and for this more resources are needed for the Ebola clinics. Though several drugs have been evaluated and vaccines are in development; however, more research is required to develop potent therapeutic and prophylactic agents against EBOV. Apart from these advances, adaptation of appropriate preventive measures and strict biosecurity principles are essential to stop the EBOV outbreaks, limit the spread of virus, and address its public health significance.

# REFERENCES


# AUTHOR CONTRIBUTIONS

All the authors substantially contributed to the conception, design, analysis and interpretation of data, checking and approving final version of the manuscript, and agreed to be accountable for its contents. KD, RK, AM, and KK initiated this review compilation. SC, SL, and RK updated various sections. RKS, YM, DK, and MR reviewed virology and biotechnology aspects. RKS, SM, RS, and WC reviewed recent vaccines and therapies. RK designed tables. KK designed the figures. WC, AM, and KD overviewed and edited final.

# ACKNOWLEDGMENTS

All the authors acknowledge and thank their respective Institutes and Universities.

# FUNDING

This compilation is a review article written, analyzed and designed by its authors, and required no substantial funding to be stated.

terrestrial small mammals of the Central African Republic. *Microbes Infect* (1999) 1:1193–201. doi:10.1016/S1286-4579(99)00242-7


virus immunoprotectant. *Proc Natl Acad Sci U S A* (2011) 108(51):20690–4. doi:10.1073/pnas.1108360108


disease virus-vectored vaccine candidate against Ebola virus elicits a neutralizing antibody response. *Vaccine* (2010) 29(1):17–25. doi:10.1016/j. vaccine.2010.10.024


(GP)-mediated virus entry and infection. *Antiviral Res* (2016) 127:20–31. doi:10.1016/j.antiviral.2016.01.001


glycoproteins show protection in a murine challenge model. *J Virol* (2017) 91(16). doi:10.1128/JVI.00652-17


interferon-ß and combination therapies with nucleoside analogs as effective inhibitors of Ebola virus. *PLoS Negl Trop Dis* (2016) 10(1):e0004364. doi:10.1371/journal.pntd.0004364


**Conflict of Interest Statement:** All authors declare that there exist no commercial or financial relationships that could in any way lead to a potential conflict of interest.

*Copyright © 2018 Dhama, Karthik, Khandia, Chakraborty, Munjal, Latheef, Kumar, Ramakrishnan, Malik, Singh, Malik, Singh and Chaicumpa. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Neutrophil Activation in Acute Hemorrhagic Fever With Renal Syndrome Is Mediated by Hantavirus-Infected Microvascular Endothelial Cells

Tomas Strandin<sup>1</sup> \*, Satu Mäkelä<sup>2</sup> , Jukka Mustonen<sup>2</sup> and Antti Vaheri <sup>1</sup>

<sup>1</sup> Department of Virology, Medicum, Faculty of Medicine, University of Helsinki, Helsinki, Finland, <sup>2</sup> Department of Internal Medicine, Faculty of Medicine and Life Sciences, Tampere University Hospital, University of Tampere, Tampere, Finland

### Edited by:

Alan Chen-Yu Hsu, University of Newcastle, Australia

#### Reviewed by:

Erdong Cheng, University of Pittsburgh Cancer Institute, United States Albert Rizvanov, Kazan Federal University, Russia Tione Buranda, University of New Mexico, United States

> \*Correspondence: Tomas Strandin tomas.strandin@helsinki.fi

#### Specialty section:

This article was submitted to Viral Immunology, a section of the journal Frontiers in Immunology

Received: 31 March 2018 Accepted: 24 August 2018 Published: 18 September 2018

#### Citation:

Strandin T, Mäkelä S, Mustonen J and Vaheri A (2018) Neutrophil Activation in Acute Hemorrhagic Fever With Renal Syndrome Is Mediated by Hantavirus-Infected Microvascular Endothelial Cells. Front. Immunol. 9:2098. doi: 10.3389/fimmu.2018.02098 Hantaviruses cause hemorrhagic fever with renal syndrome (HFRS) and hantavirus cardiopulmonary syndrome (HCPS) in humans. Both diseases are considered to be immunologically mediated but the exact pathological mechanisms are still poorly understood. Neutrophils are considered the first line of defense against invading microbes but little is still known of their role in virus infections. We wanted to study the role of neutrophils in HFRS using blood and tissue samples obtained from Puumala hantavirus (PUUV)-infected patients. We found that neutrophil activation products myeloperoxidase and neutrophil elastase, together with interleukin-8 (the major neutrophil chemotactic factor in humans), are strongly elevated in blood of acute PUUV-HFRS and positively correlate with kidney dysfunction, the hallmark clinical finding of HFRS. These markers localized mainly in the tubulointerstitial space in the kidneys of PUUV-HFRS patients suggesting neutrophil activation to be a likely component of the general immune response toward hantaviruses. We also observed increased levels of circulating extracellular histones at the acute stage of the disease supporting previous findings of neutrophil extracellular trap formation in PUUV-HFRS. Mechanistically, we did not find evidence for direct PUUV-mediated activation of neutrophils but instead primary blood microvascular endothelial cells acquired a pro-inflammatory phenotype and promoted neutrophil degranulation in response to PUUV infection in vitro. These results suggest that neutrophils are activated by hantavirus-infected endothelial cells and may contribute to the kidney pathology which determines the severity of HFRS.

Keywords: hantavirus, HFRS, neutrophils, IL-8, endothelial cells, degranulation, NETs, NETosis

# INTRODUCTION

Hantaviruses are the causative agents of two human diseases: hemorrhagic fever with renal syndrome (HFRS) in Eurasia and hantavirus cardiopulmonary syndrome (HCPS) in the Americas. Each hantavirus species is carried by its specific rodent host with no or minimal signs of disease but cause occasional human spillover infections with immune-mediated pathology (1). A hallmark of both hantavirus diseases is increased vascular permeability which mediates kidney and lung failure associated with HFRS and HCPS, respectively (1). Endothelial cells lining the vasculature are also the prime target of viral replication in patients (2, 3). Puumala hantavirus (PUUV) circulates in Northern Europe and Russia causing a relatively mild form of HFRS as compared to Hantaan (HTNV) or Dobrava hantavirus (DOBV)-caused HFRS and especially Andes (ANDV)- or Sin Nombre (SNV)-caused HCPS in which fatality rates can reach 40% (4).

Typical laboratory findings in acute PUUV-caused HFRS are leukocytosis, thrombocytopenia, increased Creactive protein (CRP) level, and as signs of acute kidney injury (AKI), proteinuria, haematuria and elevated serum creatinine concentration (5). In addition to thrombocytopenia, hematological abnormalities include increased coagulation and fibrinolysis, complement activation, and elevated levels of proinflammatory cytokines which all have the potential to contribute to vascular permeability. Vascular permeability, in turn, could be the underlying cause of proteinuria which typically precedes AKI (6). The pathophysiology of PUUV-HFRS associated AKI is usually described as tubulointerstitial nephritis and infiltration of several immune cell types such as lymphocytes, monocytes, and polymorphonuclear leukocytes into kidneys have been observed (7, 8).

Neutrophils are the most abundant circulating leukocytes in humans and play a fundamental role in the innate immune response (9). Circulating neutrophils are the first type of immune cells recruited to sites of inflammation or infection. In humans, one of the most important chemotactic factors for neutrophils is interleukin (IL)-8, released at the site inflammation/infection (10). After receiving chemotactic signals neutrophils interact with endothelial cells lining the vasculature in order to traverse the endothelium and into the inflamed tissues (11). In vivo, the interactions between neutrophils and endothelial cells include initial rolling followed by firm adhesion and finally transendothelial migration. Firm adhesion is facilitated, among other factors, by endothelial intercellular adhesion molecule (ICAM)-1 expressed on the surface of endothelial cells and CD11b/CD18 integrin complex (also known as Mac-1, CR3 or αMβ2) on neutrophils (12). Once neutrophils reach the site of infection their primary role is to kill invading microbes by production of reactive oxygen species (ROS) and the release of antimicrobial proteins such as myeloperoxidase (MPO) and human neutrophil elastase (HNE) in a process of degranulation (13, 14). In addition to ROS formation and degranulation, neutrophils are also able to release neutrophil extracellular traps (NETs), consisting of extracellular chromatin decorated with histones and granular proteins such as MPO and HNE, and with the potential to entrap and kill pathogens (15, 16).

The activation of neutrophils in several bacterial infections is well-described but their role in virus-mediated diseases has been neglected to a large extent (17). The involvement of neutrophils in the pathogenesis of hantavirus diseases is suggested by acutely increased serum levels of cytokines and chemokines with known functions in neutrophil chemotaxis, differentiation and mobilization (18–20). Increased numbers of circulating histones, cell-fee DNA and histone-DNA complexes have been described in the acute stage of PUUV infection (21–23), suggesting that neutrophils release NETs during HFRS. NETosis could potentially explain capillary leakage in HFRS (24). Interestingly, neutrophils are crucial for increased vascular permeability observed in response to HTNV infection in SCID mice (25). Furthermore, HTNV has also been shown to cause NETosis in vitro (21).

In this study we wanted to elucidate the role of neutrophil activation in HFRS by determining markers of neutrophil activation (MPO, HNE, histones, and IL-8) in blood and tissues of patients suffering from acute PUUV-caused HFRS. In addition, to directly determine the role that hantavirus plays in mediating neutrophil responses in PUUV-HFRS, we investigated the potential of purified PUUV or PUUV-infected endothelial cells to activate neutrophils in vitro. Taken together, we found that the levels of circulating and tissue-localized MPO, HNE and IL-8 are elevated in acute PUUV-HFRS and correlate with kidney dysfunction, thereby corroborating the role of neutrophils in hantavirus pathogenesis. Mechanistically, our results do not support direct virus-mediated neutrophil activation but rather an indirect mechanism through infected endothelial cells. Finally, the antiviral function of neutrophils was pinpointed strongly to the release of proteases from neutrophils.

# MATERIALS AND METHODS

# Patient Samples

The study was approved by the Ethics Committee of Tampere University Hospital (Nos. 99256 and R04180). All subjects gave written informed consent in accordance with the Declaration of Helsinki. The study consisted of plasma samples from patients treated for serologically confirmed acute PUUV infection at the Tampere University Hospital, Finland, during September 2000– March 2009. We assessed the extracellular circulating levels of MPO, HNE, histone H3 and IL-8 in plasma samples obtained from patients with acute PUUV-caused HFRS at 1st day of hospitalization (acute stage; median days after onset of fever 4 ± 2), early recovery phase (15–30 days after hospitalization) and healthy controls.

The study included Boiun-fixed, paraffin-embedded kidney biopsies obtained at Tampere University Hospital during 1985– 1987. The biopsies among patients with PUUV-HFRS were performed for clinical reasons at the time when there was no reliable serological test for PUUV infection available. All these biopsies were performed during the acute phase of the disease. The highest measured serum creatinine level of the patients ranged from 220 to 1,050 µmol/L. The biopsy findings were acute interstitial in two cases and acute tubulointerstitial nephritis in three, both findings being typical for PUUV-induced AKI (5).

PUUV-negative cases served as controls. Indications for renal biopsies were AKI in one case, microscopic hematuria and/or proteinuria in four cases. The biopsy findings were normal morphology in two cases, acute tubulointerstitial nephritis, mesangial proliferative glomerulonephritis and IgMglomerulonephritis in one case each. In both groups of biopsy patients a positive (PUUV-HFRS cases) or negative (controls) PUUV-serology was determined using stored serum samples obtained at the time of renal biopsy.

# Primary Antibodies

Primary antibodies used in this study were rabbit polyclonal antibodies to MPO (Thermo scientific, #RB-373-A), Histone H3 (Abcam, #ab18521) or HNE (Abcam, #ab68672) and mouse monoclonal antibodies to IL-8 (RnD Systems; #Mab-208), CD18 (Millipore, #TS1-18) and ICAM-1 (RnD systems, #BBIG-I1) in addition to IgG1 isotype control (Immunotools). Rabbit polyclonal antibodies for PUUV nucleocapsid protein and glycoproteins have been described before (26).

# Histone Quantification

Histones were measured by a dot-blot assay where 2 µl of patient or healthy control EDTA-plasma were pipetted on nitrocellulose membrane, air-dried and probed with Histone H3 antibody in blocking buffer [1.5% milk in Tris–EDTA–NaCl– Tween (TENT)]. After washing with TENT, the primary antibody was detected with IR800-conjugated anti-rabbit antibody (Li-cor) in blocking buffer. After additional washing, signal intensity was determined by Odyssey instrumentation (Li-cor). Recombinant Histone H3 (New England Biolabs) was used for standard preparation.

# Enzyme-Linked Immunosorbent Assays (ELISAs)

The levels of MPO, HNE, and IL-8 were measured from patient or control plasma using ELISA kits provided by Abcam (for MPO, HNE, and IL-8) or Immunotools (for IL-8).

# Immunohistochemistry (IHC)

IHC was performed on kidney sections after heat-mediated antigen retrieval using the protocol provided in Vectastain ABC Elite HRP kit from Vector labs. Avidin/biotin blocking kit, biotinylated goat anti-mouse IgG or anti-rabbit secondary antibodies and DAB substrate were used as suggested by the vendor (Vector labs). The mean percentage of DAB positive area was counted from four images taken (1.5 × 1.5 mm) from each individual section using Fiji Image J software.

# Cultures of Primary Blood Microvascular Endothelial Cells (BECs)

BECs were obtained from Lonza and maintained in endothelial basal medium (EBM-2) supplemented with SingleQuotsTM Kit containing 5% fetal bovine serum (FBS), human endothelial growth factor, hydrocortisone, vascular endothelial growth factor, human fibroblast growth factor-basic, ascorbic acid, R<sup>3</sup> insulin like growth factor-1, gentamicin and amphotericin-B (Lonza). For experiments the cells were used at passages 7–10.

# Virus Propagation and Titration

PUUV Kazan strain was propagated in Vero E6 cells (green monkey kidney epithelial cell line; ATCC no. CRL-1586) grown in Minimum Essential Medium containing 10% FCS, penicillin and streptomycin (cMEM). For experiments, PUUV was purified from Vero E6 cell supernatant through a 30% sucrose cushion by ultracentrifugation. Virus titers were measured by incubating diluted virus stocks on Vero E6 cells for 24 h at 37◦C and subsequently staining acetone-fixed cells with a polyclonal antibody specific for PUUV nucleocapsid protein and AlexaFluor488-conjugated secondary antibody. Focus-forming units (FFU)/ml were counted under an UV-microscope.

# Infection of BECs

Confluent BECs were infected for 1 h at 37◦C using purified PUUV diluted in BEC growth medium at multiplicity of infection (MOI) of 10. For inactivation of PUUV, virus stocks were kept under UV light for 30 min.

# Immunofluorescence

Immunofluorescence of either mock-, UV-PUUV or live PUUV-infected BECs was performed on cells grown on black, glass-bottomed cell culture plates. After fixation with 4% formaldehyde, cells were permeabilized (3% BSA, 0.3% TritonX-100) for 10 min and incubated with primary antibodies to MPO, HNE, ICAM-1 or viral nucleocapsid protein followed by appropriate Alexafluor488- or Alexafluor594 conjugated secondary antibodies (Thermo Scientific). To stain the nuclei, cells were incubated with Hoechst 33258. After washing, fluorescence intensity was counted using Hidex sense microplate reader (Hidex) and images taken using Leica TCS SP8 X confocal microscope (Biomedicum Imaging Unit, Biomedicum, University of Helsinki, Helsinki, Finland). Neutrophil activation was quantitated by the percentage of decondensed polymorphonuclear cells with simultaneous MPO relocalization in 4 immunofluorescence images (500 × 500µm) taken randomly from each well by confocal microscope.

# PMN Cultures

Fresh blood from healthy volunteers was drawn into EDTAtubes and PMNs immediately isolated using Polymorphoprep separation medium (Axis-Shield) according to manufacturer's protocol. PMN purity and viability were assessed by phenotypic polymorphonuclear characterization using Hoechst 33258 fluorescence microscopy and trypan blue exclusion test, respectively (both were routinely found to be over 90%). Isolated PMNs were diluted into BEC growth medium (10<sup>6</sup> cells/ml) and incubated with mock-, UV-PUUV or live PUUV-infected BECs for 1 or 3 h at 37◦C (10 times excess PMNs over BECs). Cells were washed and subjected directly to immunofluorescence staining as described above for determination of PMN binding. When PMN-BEC co-cultures were incubated in the presence of 10µg/ml neutralizing antibodies to IL-8, CD18 or viral glycoproteins, PMNs were pre-treated with FcR blocking reagent (Immunostep) for 10 min.

# Neutrophil Activation Assays

Purified virus or virus-containing Vero E6 cell culture supernatants were incubated with freshly isolated PMNs (MOI 1) in cMEM for 3 h at 37◦C. The incubation time was chosen based on the optimal time needed to detect adequate levels of PMN activation marker expression with low concurrent spontaneous activation due to culturing. After fixation with 2% formaldehyde, extracellular DNA was quantitated with the fluorescent, cell-impermeable Pico-Green dsDNA binding reagent (Thermo Scientific) using Hidex Sense microplate reader. Peroxidase activity was assessed from supernatants of pelleted PMNs (400 × g 5 min) by the chromogenic peroxidase substrate 3,3′ ,5,5′ -tetramethylbenzidine (TMB).

# PUUV Viability Assay

Purified PUUV was incubated for 3 h with freshly isolated PMNs (in a 1:1 ratio) in assay buffer (10 mM Hepes pH 7,4; 150 mM NaCl). Neutrophil activator phorbol myristate acetate (PMA; 1µg/ml) and one of the following inhibitors NaN<sup>3</sup> (0.01 or 0.002%), PMSF (0.2 or 1 mM), EDTA (2 mM) or DNAse I (10 U/ml) were added where indicated. After incubation, PMNs were pelleted by centrifugation (400 × g 5 min) and one third of the supernatant used for PUUV titer measurement as described above. The rest of the sample was used for immunoblotting by standard procedures for PUUV structural proteins (27) using polyclonal antibodies specific for Gn, Gc or N followed by IRD800-conjugated secondary antibody (Li-Cor) and detection by Odyssey.

# Statistics

Significant differences between groups of normally distributed data was assessed by student's T-test and non-normally distributed data by Mann-Whitney U-test or Kruskal-Wallis H-test for multiple populations. The normality of data was estimated by Shapiro-Wilk test. Correlations between parameters were assessed using Spearman's rank correlation test. All analyses were done with SPSS software version 24 (SPSS Inc., Chicago, IL, USA) or GraphPad prism (La Jolla, CA, United States).

# RESULTS

# Circulating Levels of MPO, HNE, Histones, and IL-8 Are Elevated in Acute PUUV-Caused HFRS

Circulating levels of MPO, HNE, histones, and IL-8 were all significantly higher in the acute stage of the disease as compared to the recovery stage or healthy controls (**Figure 1**). The levels of all the measured markers remained elevated also in the recovery stage as compared to controls but this difference was not statistically significant. Strongly elevated circulating levels of MPO and HNE suggest that neutrophils are activated in acute PUUV-HFRS. Furthermore, increased numbers of extracellular histones imply that neutrophil activation could, at least in part, be due to NETosis (21). Interestingly, in the present study, neutrophil activation was accompanied by elevated levels of IL-8, a chemotactic and priming factor for neutrophils.

To get further insight to the role of neutrophil activation in the pathogenesis of PUUV-HFRS by we correlated the acute plasma levels of MPO, HNE, histone H3, and IL-8 to variables reflecting severity of AKI (maximum plasma creatinine level measured during the hospital stay), and to hematological variables (minimum blood platelet count, maximum blood leukocyte count, and plasma tissue plasminogen activator (tPA) level) (**Table 1**). We found that HNE and histone H3 positively correlated with the severity of AKI. In addition, MPO and HNE correlated with all the tested hematological variables. Not surprisingly, given the likely neutrophil origin of MPO and HNE, they also correlated strongly with each other. The acute IL-8 levels correlated significantly with low platelets and exceptionally strongly with MPO, HNE and histones (p < 0.005 for all) suggesting that IL-8 could play a role in neutrophil activation.

# Localization of MPO, HNE, and IL-8 in the Kidneys of Acute PUUV-HFRS

In order to investigate whether markers of neutrophil activation could also be detected in tissues of patients suffering from acute PUUV-HFRS, we made use of archival biopsies to detect the expression of MPO, HNE and IL-8 in patient kidneys by immunohistochemistry. We did not include histones in this analysis since we expected that differentiating NET-associated histones from viable cells would be challenging. Elevated expression of MPO, HNE and IL-8 could be detected more readily in acute PUUV-HFRS patients than PUUV-negative control patient samples (**Figure 2A**). By counting the tissue area staining positive for MPO, HNE, or IL-8 we could observe a statistically significant difference in MPO expression between PUUV positive and negative cases (n = 5; p = 0.04; **Figure 2B**). In the case of HNE and IL-8 higher expression was found in 2 and 3 PUUV cases, respectively, as compared to controls but this was not enough to reach statistical significance (n = 5; p = 0.08 and p = 0.15, respectively). The most prominent localization of MPO, HNE and IL-8 in HFRS patients was the tubulointerstitial space, in line with the diagnosis of tubulointerstitial nephritis. Therein we could observe both cell-associated (arrowheads) and extracellular expression (filled arrows) of MPO and HNE suggesting both neutrophil infiltration and activation, respectively. Interestingly, HNE and IL-8 localized also to in the tubular epithelial cells in two HFRS patients (empty arrows). IL-8 could also occasionally be observed in the tubular cells of PUUV-negative patients albeit with a lower intensity and frequency as compared to HFRS patients (**Figures 2A,B**). Taken together, these results show that MPO is a component of the inflammatory response toward PUUV in the kidneys of acute HFRS, possibly accompanied by HNE and IL-8, and suggest that neutrophils (and their activation products) infiltrate kidneys through the capillary endothelium. In addition, the current findings together with our previous observations of elevated IL-8 levels in urine of acute PUUV-HFRS (28) suggest that IL-8 is produced locally by kidney epithelial cells in PUUV-HFRS and likely acts as a chemotactic factor inviting neutrophil recruitment and extravasation in the kidney.

# Live PUUV Does Not Activate Neutrophils in vitro

It has been shown that HTNV can directly bind CD18 integrin on neutrophils and induce NETosis (21). We wanted to determine whether PUUV can also induce NETosis which could explain our findings of neutrophil activation in acute PUUV-HFRS. We did this by incubating mock- or PUUV-infected Vero E6 cell culture supernatants or purified PUUV (with the same infectious titer) with freshly isolated PMNs and subsequently quantified the release of extracellular DNA from PMNs. NETosis was

FIGURE 1 | Circulating levels of MPO, HNE, histone H3 and IL-8 in PUUV-HFRS. Plots of the acute (median days after onset of fever 4 ± 2) and recovery stage (15–30 days post onset) levels of MPO (A), HNE (B), histones (C) or IL-8 (D) in PUUV-infected patient plasma samples (n = 32, 30, 53 and 36, respectively) as compared to healthy controls (n = 5, 5, 5, and 8, respectively) distributions across groups were compared by Kruskal-Wallis H test and statistically significant differences indicated as either \*p < 0.05 or \*\*p < 0.01.

clearly induced above mock control by PUUV-containing Vero E6 supernatant but not with purified PUUV (**Figure 3A**). To elaborate on the finding that live PUUV does not cause NETosis, UV-inactivation of PUUV-infected Vero E6 supernatant did not affect its ability to induce NETosis (**Figure 3B**). These findings indicate that live PUUV alone is not capable of inducing NETosis but instead involves other factor/s released from hantavirusinfected Vero E6 cells supernatants which might act alone or synergistically with the inactivated virus. In order to analyze whether purified hantaviruses could induce PMN degranulation we incubated fresh PMNs with purified PUUV, TULV, HTNV, or PMA (as a positive control) and analyzed peroxidase activity in cell supernatants (**Figure 3B**). MPO activity was induced by PMA but not with any of the hantaviruses tested, suggesting that hantavirus do not induce significant PMN degranulation. These results imply that neutrophil activation in HFRS is not caused by direct contact with hantavirus but instead requires additional factors. Thus, we wanted to analyze whether PUUV-infected endothelial cells could play a role in neutrophil activation during HFRS.

# PUUV Induces a Pro-Inflammatory Phenotype in BECs

Endothelial cells are the prime target of hantavirus infection in vivo (2, 3). Thus, we hypothesized that PUUV could mediate neutrophil activation through infected endothelial cells. To test this, we either mock-infected or infected primary blood endothelial cells (BECs) with purified live PUUV or UVinactivated PUUV (UV-PUUV). By visualizing the expression of viral nucleocapsid protein N in the cells by immunofluorescence we observed that BECs were close to 100% infected at 3 days post infection (dpi), followed by significant drop in infectivity levels at 6 dpi (**Figure 4A**). The downregulation of PUUV infection in BECs is type I interferon-mediated, as reported previously (29), and is probably due to MxA protein sequestering the viral N protein (30) We analyzed the supernatants of infected BECs for the presence of IL-8 by ELISA and observed that BECs infected with live PUUV upregulate the secretion of IL-8 at 3 dpi as compared to mock- or UV-PUUV infections (**Figure 4B**). However, longer culturing of mock- or UV-PUUV infected BECs also upregulated IL-8 in the cell supernatants which finally led to


The clinical variables creatinine, platelet and leukocyte counts are maximum or minimum values from the course of hospital stay whereas tPA and markers of neutrophil activation are from 1st day of hospitalization. r, Spearman's rho correlation coefficient; Sig., Significance of correlation (p); n, number of patients; MPO, Myeloperoxidase; HNE, Neutrophil elastase; IL-8, Interleukin-8; tPA, Tissue plasminogen activator; \*p < 0.05, \*\*p < 0.01.

comparable, high levels of IL-8 irrespective of PUUV infection at 6 dpi. Since IL-8 could act as a general marker of inflammation (10), we wanted to see whether PUUV-infected BECs would upregulate also other inflammatory factors potentially important for neutrophil recruitment. Thus we assessed the expression of ICAM-1, a ligand for neutrophil-expressed CD11b/CD18 integrin complex, on the plasma membrane of BECs by immunofluorescence-based imaging and quantification assays. We found elevated levels ICAM-1 on the surface of PUUVinfected BECs as compared to mock- or UV-PUUV infected BECs at 3 dpi (**Figure 4C**). The elevated ICAM-1 levels remained only slightly lower as compared to TNF-α treated BECs, used as a positive control. At 6 dpi, ICAM-1 expression returned to baseline levels in PUUV-infected BECs concomitantly with the reduction of viral replication. As expected, TNF-α was found to be a robust inducer of ICAM-1 in BECs and thus we hypothesized that TNF-α could induce ICAM-1 also in PUUVinfected BECs. However, we did not find any evidence for TNF-α in PUUV-infected BEC supernatants by ELISA suggesting that the induction of ICAM-1 in PUUV-infected BECs is independent of this cytokine (data not shown).

# PMNs Adhere to PUUV-Infected BECs

Next, we wanted to determine whether freshly isolated PMNs (containing mainly neutrophils) could adhere to PUUV-infected BECs which express elevated levels of IL-8 and ICAM-1. Fresh PMNs were incubated with mock-, UV-PUUV or live PUUVinfected BECs for 1 h, non-bound PMNs removed by washing and remaining numbers of BEC-bound PMNs determined by immunofluorescence-based imaging and quantitation of MPO (used as a marker of PMNs). In addition to MPO, bound PMNs were differentiated from BECs based on their segmented nuclear morphology by DNA staining. We found that PMN binding to PUUV-infected BECs was significantly elevated as compared to mock- or UV-PUUV infected BECs at 3 dpi, but not at 6 dpi (**Figures 5A,B**), consistent with the elevated expression of ICAM-1 and higher virus replication in PUUVinfected BECs at 3 dpi. As positive and negative controls, we used TNF-α treated BECs and PUUV infected-BECs without PMN incubation, respectively. Based on these results we consistently allowed PMNs to bind the 3-day BEC cultures for 1 h in the following experiments assessing PMN-BEC interaction as shown in **Figure 5**.

We wanted to analyze further whether the binding of PMNs to PUUV-infected BECs is mediated by chemotactic signals driven by IL-8 and/or is dependent of the neutrophil-expressed CD11b/CD18 binding to ICAM-1 on BECs. We did this by incubating PMNs with either BECs infected with either live PUUV or UV-PUUV as the negative control for infection in the presence of neutralizing antibodies to IL-8 or CD18. We could determine that binding to PUUV-infected BECs was dependent on CD11b/CD18 on neutrophils but not on the presence of IL-8 (**Figures 5A,C**). Not surprisingly, CD11b/CD18 seemed to mediate the adhesion of PMNs also to TNF-α treated BECs, although less dramatically (**Figure 5D**). The fact that CD18 neutralizing antibody was not as efficient in blocking PMN adhesion to TNF-α treated BECs could be due to the excess level of ICAM-1 upregulated by TNF-α as compared to PUUV infection (**Figure 4C**). Given that HTNV has been shown to bind CD18 integrin (21) we also tested the effect of viral glycoprotein (Gn and Gc) neutralizing antibodies (31) on PMN-BEC interaction. However, we could not observe any significant effects (**Figure 5E**) by Gn or Gc neutralizing antibodies suggesting that PMN adhesion to infected BECs is mediated by host-derived inflammatory factors but not viral proteins. Unfortunately, we could not reliably determine the role of ICAM-1 in PMN-BEC interaction since the presence of ICAM-1 specific antibodies resulted in further elevated binding of PMNs to BECs regardless of BECs being infected with PUUV or not (data not shown). We hypothesize that this phenomenon could be due to antibody-mediated cross-linking of neutrophiland BEC-associated ICAM-1 proteins. In addition, ICAM-1 antibodies are known to activate cross-linking and activation

IL-8 was evaluated as percentages in all sections and mean values ± standard deviation reported for acute PUUV-HFRS and PUUV-negative sections. Significant differences were assessed by student's T-test and statistically significant differences indicated as \*p < 0.05.

of ECs on cell surfaces in some conditions (32) which could potentially explain this finding.

# PUUV-Infected BECs Induce PMN Degranulation

Next, we wanted to determine whether more extensive coculturing of PMNs and PUUV-infected BECs could result in PMN activation (degranulation or NETosis) which could explain our findings of increased levels of extracellular neutrophil markers in PUUV-HFRS patients (**Figures 1**, **2**). We observed a time-dependent morphological change in PMNs bound to PUUV-infected BECs but not in mock- or UV-PUUV infected BECs which was visible after 3 but not 1 h of co-culture (**Figures 6A,B**). This was evident by increased nuclear swelling of PMNs together with altered localization of MPO from diffuse to plasma membrane-associated staining. Furthermore, MPO could be observed also extracellularly, which strongly suggests that PMN degranulation was taking place. The fluorescent phenotype of PMNs bound to PUUV-infected BECs was strikingly similar although less pronounced as observed for PMNs bound to TNF-α treated cells after a 3 h of co-culture. We did not find any evidence for NETosis in PMNs bound either on PUUV-infected or TNF-α treated cells or by PUUV-infected BEC supernatants (data not shown) in these experimental conditions.

# Antiviral Effect of PMNs Is Mediated by Protease Release in vitro

Finally, we wanted to determine whether neutrophils possess antiviral function and if so, by which mechanism. We incubated purified PUUV with freshly isolated PMNs, which were either non-activated or activated with PMA. Furthermore, incubations

FIGURE 3 | Live PUUV does not induce neutrophil activation. (A) Mock- or PUUV-containing Vero E6 supernatants (Sup) or purified PUUV were incubated with freshly isolated PMNs for 3 h (MOI 1) (n = 2). The release of extracellular DNA was assessed by an impermeable DNA binding fluorescent dye. (B) Mock-, live PUUV- or UV-inactivated PUUV-containing Vero E6 supernatants were incubated with PMNs and assessed for DNA release as in (A) (n = 3). (C) Purified PUUV, TULV or HTNV were incubated with fresh PMNs for 3 h (MOI 1) and peroxidase activity assessed from cell culture supernatants by TMB (n = 3). PMA-treated PMNs were used as a positive controls. Difference between groups were assessed by one-way ANOVA + Dunnett's multiple comparisons test and statistically significant differences indicated as either \*p < 0.05 or \*\*p < 0.01.

and assessed for viral nucleocapsid protein expression at 3 and 6 days post infection by immunofluorescence (red). The nuclei of BECs were visualized with Hoechst 33420 (blue). (B) IL-8 was measured from the respective supernatants of mock, UV-PUUV or PUUV-infected BECs by ELISA. (C) ICAM-1 expression in TNF-α treated or mock-, UV-PUUV or PUUV-infected BECs was visualized by immunofluorescence (green) and shown as an overlay with Hoechst 33258 staining (blue). Fluorescence intensity of ICAM-1 expression on BECs was quantified and reported as fold change to mock-infected cells. Differences between groups were assessed by one-way ANOVA with Dunnett's multiple comparisons test and statistically significant differences indicated as \*\*p < 0.01. n = 2 in all panels. Results shown are representatives of three independent experiments.

and indicated as \*\*p < 0.01. n = 2 in all panels. Results shown are representatives of three independent experiments.

of virus with PMA-activated PMNs were performed in the presence of one of the following inhibitors: NaN<sup>3</sup> (an inhibitor of MPO activity), phenylmethylsulfonylfluoride (PMSF; inhibitor of serine proteases), EDTA (inhibitor of metalloproteinases) and DNAse (degradation of NETs). After incubation, samples were subjected to immunoblotting in order to detect degradation of

viral structural proteins Gn, Gc and N protein (**Figure 7A**) and virus titer determinations in Vero E6 cells (**Figure 7B**). The PUUV Gc and N proteins migrated as expected based on their molecular size of ∼54 and 50 kDa, respectively, whereas Gn is known to form SDS-stabile tetramers of ∼240 kDa in SDS-PAGE [**Figure 7A**, (27)]. PMA-treated PMNs complete destroyed PUUV infectivity and the surface glycoproteins Gn and Gc while it has almost no effect on N protein, which is retained inside the virus particle, whereas quiescent PMNs decreased PUUV viability only marginally (50%) with the concomitant low-level cleavage of Gc. PMSF totally blocked the cleavage of Gn and Gc while retaining almost 10% of PUUV infectivity indicating strong involvement of serine proteases (such as HNE) in the antiviral effect of activated PMNs. Also, EDTA treatment retained significant levels of infectivity (5%) although not being able to block the degradation of Gn and Gc significantly. However, NaN<sup>3</sup> or DNAse treatments had no effect on the ability of PMA-treated PMNs to kill virus or degrade its glycoproteins, indicating that MPO or the formation of NETs do not play a direct role in the antiviral effects of PMNs in this experimental setup. The fact that the infectivity of PUUV was not significantly reduced after treatment with quiescent PMNs is further proof (see **Figure 3**) that hantaviruses are unable to significantly activate PMNs.

PMSF), metalloproteinases (1 mM EDTA) or NETs (10 U/ml DNAse) and (A) subjected to immunoblotting for viral proteins Gn, Gc and N protein and infectious titer measurement in Vero E6 cells (B). C, Control without the addition of PMNs.

# DISCUSSION

We observed elevated levels of circulating extracellular MPO, HNE and histones in the acute stage of PUUV-caused HFRS as compared to the recovery stage and healthy controls. MPO and HNE are expressed in neutrophil granules and released to the extracellular space through neutrophil degranulation or alternatively as components of NETs. Histones, on the other hand, are essential components of NETs but could be also released from other types of dying cells. The circulating levels of histone H3 correlated with MPO and HNE suggesting that all these factors are released simultaneously from neutrophils during NETosis, which presence in acute PUUV-HFRS has been reported previously (21). Interestingly, we found strong positive correlations between the levels of HNE or histone H3 and variable reflecting the severity of AKI (maximum creatinine); indicating that neutrophils could play a role in mediating kidney damage, the hallmark of HFRS. We could observe elevated expression of MPO in kidneys of acute PUUV-HFRS where it mainly localized to the interstitial space surrounding tubules, supporting earlier findings that infiltration of neutrophils into kidneys is a part of the inflammatory response toward PUUV (7).

In addition, we found that the major chemotactic factor for neutrophils, IL-8, is upregulated in the acute stage of PUUV-HFRS in blood, as previously reported (33), and correlated strongly with neutrophil activation markers (MPO, HNE and histone H3). This was also accompanied by a tendency of higher IL-8 expression also in the kidneys of acute PUUV-HFRS patients. The kidney IL-8 was found to be localized to the tubulointerstitial space, similarly to MPO and HNE, but also to tubular epithelial cells. This suggests IL-8 could be expressed directly by the kidney in PUUV-HFRS and plays a major role in recruiting neutrophils into this tissue. Furthermore, local expression of IL-8 in the kidney probably explains the high levels of IL-8 found in the urine of PUUV-HFRS patients (28).

AKI is the hallmark symptom of HFRS but the disease manifests as systemic vascular dysfunction as indicated by capillary leakage in several organs (1). It is of interest to note that neutrophil activation markers not only correlate with kidney injury, but also with different hematological parameters indicating the extent of thrombocytopenia, leukocytosis and fibrinolysis (low platelets, high leukocytes and high tPA, respectively). This suggests that neutrophil activation either contributes to or is in fact caused by the same underlying factors which lead to capillary leakage in PUUV-HFRS and is not an isolated phenomenon which effect would be restricted to the kidney. This calls for future studies aimed at determining the extent of neutrophil activation also in other forms of hantavirusmediated diseases such as HCPS which severely affect lungs instead of kidneys. In fact, neutrophils have been shown as the main culprits of pulmonary vascular permeability in mouse model of HTNV infection (25).

A previous report indicates that HTNV binds CD11b/CD18 integrin complex on the surface of neutrophils and, as shown using HTNV-containing Vero E6 cell culture supernatants, activates NETosis in freshly isolated PMNs (21). Given the high degree of similarity between different hantaviruses (1), we hypothesized that also PUUV is be able to activate in NET formation in PMNs. However, we observed that, while PUUVcontaining Vero E6 cell culture supernatant was able to activate NETosis, purified PUUV was not (**Figure 3A**), suggesting that NETosis is not mediated by live, intact virus but rather other factor/s present in the cell culture supernatant of infected Vero E6 cells. We therefore sought for an alternative mechanism of neutrophil activation in PUUV-HFRS and turned our attention toward microvascular endothelial cells, the prime target of hantavirus infection in vivo. To begin with, we found that PUUVinfected blood microvascular cells (BECs) secreted elevated levels of IL-8 and expressed more ICAM-1 on their surface than non-infected BECs; suggesting that PUUV infection induces a pro-inflammatory phenotype in BECs. Importantly, elevated expression of ICAM-1 has been also detected in kidneys of acute PUUV-HFRS (8). We observed that freshly isolated PMNs adhered to PUUV-infected, pro-inflammatory BECs which was dependent on CD11b/CD18 integrin complex on the surface of neutrophils but not IL-8 or viral glycoproteins. We then asked whether the BEC-adhered PMNs would become activated to explain the observed elevated levels of MPO, HNE and histones in patients. We found that extensive co-culturing of fresh PMNs together with PUUV-infected BECs resulted in morphological changes in PMNs, which suggested degranulation of PMNs. However, we did not find any evidence of increased NETosis in these co-cultures.

These results imply that PUUV infection causes proinflammatory changes in endothelial cells to attract and activate neutrophils on the surface of the endothelium by an IL-8 dependent mechanism. It is previously noted that proinflammatory endothelial cells can induce NETosis partially through EC-derived IL-8 causing endothelial cell death (34). The interactions of neutrophils with the endothelium in inflammatory conditions is well-documented (9, 12, 35, 36) and typically these events are not associated with clinically significant vascular permeability. In typical inflammatory reactions, however, the chemotactic signals that induce inflammation originate from tissues from where neutrophil diapedesis across the endothelium is locally dictated. This is in contrast to hantavirus infections where the viruses replicate in endothelial cells to cause systemic inflammation of the vasculature. In addition to endothelial cells, hantavirus glycoproteins have been detected in renal tubular cells in acute HTNV-caused HFRS (37), where we observed increased levels of HNE and IL-8 in this study, suggesting that hantavirus-infected tubular cells could behave similarly to infected endothelial cells and recruit and activate neutrophils by an IL-8 dependent mechanism. The fact that depletion of neutrophils suppress lung pathology in a mouse model for HTNV certainly supports the idea of neutrophils as the driving force of HFRS pathogenesis (25). Similar mechanisms could play a role also in other hemorrhagic fevers (38). It is known that for instance that flavi- and filoviruses induce pro-inflammatory changes in endothelial cells (39–41). Of note, ICAM-1 polymorphism is associated with dengue disease severity (42).

Neutrophils are well-known for their antibacterial effects but their ability to counteract viral infections are only beginning to be unraveled (17, 35, 43). Early research indicates that defensins, which are peptides released from neutrophils upon activation, show antiviral activity against enveloped viruses (44). MPO can also cause oxidative damage to viruses upon neutrophil oxidative burst and degranulation (45). Lately, the antiviral effects of NETs have been implicated (46). In the present study we observed that serine proteases (such as HNE) play a major role in the

# REFERENCES


virucidal function of neutrophils against PUUV. However, we did not obtain evidence for a direct role of NETs or MPO-mediated oxidative effects in killing PUUV. Furthermore, the fact that PUUV was not killed by PMNs which were non-activated at the start of the experiment, show that PUUV doesn't directly cause significant protease release by PMNs. The inability of PMNs to produce NETs through direct contact with hantaviruses is not surprising since typically NETs are produced in response to larger microbes (47) but given the known expression of several Toll-like receptors by neutrophils (48), their inability to release proteases (degranulate) in response to PUUV is peculiar. However, some viruses are known to block the activation of neutrophils (49, 50) and whether this is also the case for hantaviruses remain a topic for future studies.

To conclude, we have found that patients suffering from acute PUUV-caused HFRS show elevated levels of circulating MPO, HNE, histone H3 and IL-8 indicating neutrophil activation through degranulation and/or NETosis. The expression levels of these markers positively correlate with parameters reflecting disease severity suggesting that neutrophil activation could play a major role in the pathogenesis of HFRS. Mechanistically, our data indicates that neutrophil activation is more likely to occur indirectly via virus-infected microvascular endothelial cells rather than directly through virus contact with neutrophils. Future studies are still needed to elucidate whether either neutrophil degranulation or NETosis has the stronger impact on hantavirus pathogenesis.

# AUTHOR CONTRIBUTIONS

TS and AV conceived and designed the research. TS performed the experiments and analyzed the data. SM and JM contributed to sample collection. TS wrote the paper. All authors reviewed the manuscript.

# FUNDING

This work was supported by the Academy of Finland (Grant 1275597 to TS), Sigrid Jusélius Foundation (to JM and AV), Magnus Ehrnrooth Foundation (to AV), the Competitive State Research Financing of the Expert Responsibility Area of Tampere University Hospital (grant 9P0312 to JM) and Tampere Tuberculosis Foundation (to JM).

# ACKNOWLEDGMENTS

We thank Ms. Sanna Mäki, Ms. Katriina Ylinikkilä, and Ms. Marja-Leena Koskinen for excellent technical assistance.


Dis (Lond). (2017) 49:321–32. doi: 10.1080/23744235.2016.12 74421


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Strandin, Mäkelä, Mustonen and Vaheri. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.