# RESPIRATORY VIRUS INFECTION: RECENT ADVANCES

EDITED BY : Kelvin To, Ville Peltola and Shin-Ru Shih PUBLISHED IN : Frontiers in Microbiology, Frontiers in Medicine and Frontiers in Public Health

### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88963-862-8 DOI 10.3389/978-2-88963-862-8

### 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

# RESPIRATORY VIRUS INFECTION: RECENT ADVANCES

Topic Editors: Kelvin To, The University of Hong Kong, Hong Kong Ville Peltola, Turku University Hospital, Finland Shin-Ru Shih, Chang Gung University, Taiwan

Citation: To, K., Peltola, V., Shih, S.-R., eds. (2020). Respiratory Virus Infection: Recent Advances. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-862-8

# Table of Contents


Ji Eun Hong, Yoon-Chul Kye, Sung-Moo Park, In Su Cheon, Hyuk Chu, Byung-Chul Park, Yeong-Min Park, Jun Chang, Jae-Ho Cho, Man Ki Song, Seung Hyun Han and Cheol-Heui Yun

*39 Recent Aspects on the Pathogenesis Mechanism, Animal Models and Novel Therapeutic Interventions for Middle East Respiratory Syndrome Coronavirus Infections*

Sinosh Skariyachan, Sneha Basavaraj Challapilli, Swathi Packirisamy, Supreetha Toplar Kumargowda and Vaishnavi Sneha Sridhar


Olivier Vandenberg


# Editorial: Respiratory Virus Infection: Recent Advances

### Ville Peltola<sup>1</sup> \*, Shin-Ru Shih<sup>2</sup> and Kelvin Kai-Wang To<sup>3</sup>

*<sup>1</sup> Department of Paediatrics and Adolescent Medicine, Turku University Hospital and University of Turku, Turku, Finland, <sup>2</sup> Research Center for Emerging Viral Infections, Chang Gung University, Taoyuan, Taiwan, <sup>3</sup> Department of Microbiology, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Pokfulam, Hong Kong*

Keywords: respiratory virus, influenza, epidemiology, pathogenesis, therapeutics, vaccines

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

### **Respiratory Virus Infection: Recent Advances**

Respiratory virus infections cause an enormous disease burden both in children and in adults in all parts of the world. Influenza A and B viruses and respiratory syncytial virus (RSV) are responsible for the highest numbers of deaths and hospitalizations but other respiratory viruses, including rhinovirus, parainfluenza viruses, adenovirus, human bocavirus and coronaviruses, are also currently known to cause severe diseases in addition to mild upper respiratory tract infections. Furthermore, respiratory viruses of animal origin keep on emerging as novel human pathogens with epidemic or pandemic potential. Such viruses include avian influenza viruses and the Middle East Respiratory Syndrome (MERS) and Severe Acute Respiratory Syndrome (SARS) coronaviruses.

### Edited and reviewed by:

*Marc Jean Struelens, European Centre for Disease Prevention and Control (ECDC), Sweden*

> \*Correspondence: *Ville Peltola ville.peltola@utu.fi*

### Specialty section:

*This article was submitted to Infectious Diseases - Surveillance, Prevention and Treatment, a section of the journal Frontiers in Medicine*

> Received: *13 January 2020* Accepted: *12 May 2020* Published: *09 June 2020*

### Citation:

*Peltola V, Shih S-R and To KK-W (2020) Editorial: Respiratory Virus Infection: Recent Advances. Front. Med. 7:257. doi: 10.3389/fmed.2020.00257*

Despite the high clinical importance of respiratory viruses, the mechanisms of pathogenicity, transmission and evolution are only partially known, and most respiratory virus infections lack effective treatment or prevention modalities. With this Research Topic we aimed to focus on these knowledge gaps. We sought to collect original research articles and focused reviews from leading experts all over the world on respiratory virus epidemiology, pathogenesis, diagnostics, therapeutics and vaccines. We acknowledge all authors who submitted high-quality manuscripts. We accepted nine papers including six original research articles and three reviews. Influenza viruses are well-presented in the Topic by five papers. Two articles focus on RSV and one each on rhinovirus and MERS coronavirus.

Seasonal influenza A and B virus epidemics globally cause up to 600,000 deaths per year. Influenza A viruses have an animal reservoir in avian species, which forms the basis for a permanent threat of pandemics. Vaccines and specific antiviral drugs against influenza are available but they have limitations including variable vaccine efficacies. McAuley et al. review the current knowledge on the structure of the influenza virus neuraminidase and link the structural features of this surface glycoprotein with the multiple functions of neuraminidase in virus infection and fitness. Guo et al. studied the genetic changes that occur in avian influenza A H5 viruses when they infect humans. Identification of the important steps in adaptation of these viruses to humans is needed for evaluating the pandemic risk related to avian influenza viruses.

National reporting of clinical microbiology results is an important part of surveillance for respiratory virus epidemics. Clinical microbiology services have recently been consolidated in many countries to a limited number of central laboratories. Van den Wijngaert et al. studied the effects of such laboratory merging on influenza surveillance in Belgium. They report benefits of the integration of whole genome sequencing of influenza viruses performed in centralized clinical microbiology laboratories into the public health surveillance system.

**4**

Neuraminidase inhibitors are widely used for treatment (and sometimes prevention) of influenza infection, and they have demonstrated efficacy. However, new drugs are needed to reach better treatment results in vulnerable patient groups particularly in the scenario of developing antiviral resistance. Several new antiviral drugs against influenza have been developed or are under development, including polymerase inhibitors and monoclonal antibodies. These new drugs are reviewed in the paper by Principi et al..

Vaccines are the cornerstone in protection against influenza. However, currently used influenza vaccines have limitations and the process of production in eggs is vulnerable. Yamada et al. report preclinical results on the efficacy of a recombinant hemagglutinin protein vaccine candidate produced in human Expi293F cell culture. Interestingly, recombinant hemagglutinin protein of H1N1pdm09 virus protected mice not only against H1N1pdm09 virus but also against avian H5N1 influenza virus.

RSV occurs in yearly epidemics that cause high number of severe lower respiratory tract infections in young infants and also in the elderly. There is variation in the timing of RSV epidemics between geographic areas. Recognition of the onset of the epidemic is important in order to accurately start adequate control measures, particularly passive immunization of high-risk infants by monoclonal antibodies such as palivizumab. Yamagami et al. developed a mathematical model that was able to detect the onset of RSV season in different parts of Japan with various climate conditions. Another interesting study on RSV by Hong et al. focused on enhancing antiviral activity of alveolar macrophages by administration of Bacillus subtilis spores in a mouse model. Their findings may help in development of novel RSV therapeutics in the future.

Rhinovirus is frequently detected in children with pneumonia, but its clinical significance remains unclear. Hartiala et al. compared rhinovirus-positive and rhinovirus-negative children hospitalized with community-acquired pneumonia. They found that the clinical picture was rather similar in both groups. Premature birth associated with rhinovirus-positive pneumonia. More research is needed on the pathogenic role of rhinovirus in pneumonia.

Skariyachan et al. reviewed the pathogenesis, animal models and therapeutics for MERS coronavirus infections. This review is timely not only for the continuing identifications of human infections by MERS coronavirus in the Middle East but also because of the potential of other novel coronaviruses to jump from animals to humans. Most recently, a novel beta coronavirus has led to an outbreak of respiratory tract infection in Wuhan, China.

We believe that the articles published in this Research Topic provide important new knowledge on respiratory virus infections. They identify areas of lacking knowledge that need to be addressed by future research. We hope that this collection of articles is beneficial for the readers and advances the field.

### AUTHOR CONTRIBUTIONS

VP drafted the editorial and all authors revised it.

**Conflict of Interest:** 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 © 2020 Peltola, Shih and To. 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 Neuraminidase Structure and Functions

*Julie L. McAuley1 \*, Brad P. Gilbertson1 , Sanja Trifkovic1,2 , Lorena E. Brown1 and Jennifer L. McKimm-Breschkin1*

*1Department of Microbiology and Immunology, The University of Melbourne at the Peter Doherty Institute for Infection and Immunity, Melbourne, VIC, Australia, 2Department of Infectious Diseases, St Jude Children's Research Hospital, Memphis, TN, United States*

With the constant threat of emergence of a novel influenza virus pandemic, there must be continued evaluation of the molecular mechanisms that contribute to virulence. Although the influenza A virus surface glycoprotein neuraminidase (NA) has been studied mainly in the context of its role in viral release from cells, accumulating evidence suggests it plays an important, multifunctional role in virus infection and fitness. This review investigates the various structural features of NA, linking these with functional outcomes in viral replication. The contribution of evolving NA activity to viral attachment, entry and release of virions from infected cells, and maintenance of functional balance with the viral hemagglutinin are also discussed. Greater insight into the role of this important antiviral drug target is warranted.

### *Edited by:*

*Ville Peltola, Turku University Hospital, Finland*

### *Reviewed by:*

*Kirsty Renfree Short, The University of Queensland, Australia Takashi Irie, Hiroshima University, Japan*

*\*Correspondence:* 

*Julie L. McAuley jmcauley@unimelb.edu.au*

### *Specialty section:*

*This article was submitted to Virology, a section of the journal Frontiers in Microbiology*

*Received: 25 October 2018 Accepted: 10 January 2019 Published: 29 January 2019*

### *Citation:*

*McAuley JL, Gilbertson BP, Trifkovic S, Brown LE and McKimm-Breschkin JL (2019) Influenza Virus Neuraminidase Structure and Functions. Front. Microbiol. 10:39. doi: 10.3389/fmicb.2019.00039*

Keywords: influenza, neuraminidase, hemagglutinin, sialic acid, infection, HA:NA balance, NA

# INTRODUCTION

Influenza viruses are members of the family *Orthomyxoviridae*, a group of enveloped viruses containing a segmented negative-sense single-stranded RNA genome. Three main types of influenza viruses (A, B, and C) infect humans, with influenza A and B viruses causing significant morbidity and mortality annually. While yearly outbreaks of influenza in the human population induce the development of neutralizing antibody against common circulating strains, new strains arise constantly in a process referred to as antigenic drift. This occurs due to errors in copying of the genome by the viral polymerase and selection of those mutants with changes in the antigenic sites that allow escape from neutralization [reviewed in (Wilson and Cox, 1990)].

The influenza virus major surface glycoproteins, hemagglutinin (HA), and neuraminidase (NA) dominate the virion surface and form the main targets for these neutralizing antibodies. In addition to the mutations that arise due to antigenic drift, the HA and NA of influenza A viruses (IAVs) can exist in different forms. Based on HA and NA antigenicity using serologic tests with hyperimmune sera, there have been a total of 16 HA (H1-16) and 9 NA (N1-9) subtypes identified in birds. These are expressed in numerous combinations of viruses isolated from aquatic avian species, and an additional two combinations, H17N10 and H18N11, have been identified in bats (Tong et al., 2012, 2013). IAVs of subtypes H1N1 and H3N2 are endemic in humans, circulating constantly within the population and giving rise to seasonal outbreaks. Zoonotic transmission from birds and swine of viruses with certain other HA and NA combinations (e.g. H5N1, H7N9, and H9N2) sporadically occurs; however, these viruses need to accrue additional mutations in order to gain the ability to transmit readily between humans (Cox and Subbarao, 2000; Harris et al., 2017). If viruses bearing a novel HA subtype,

**6**

which is often accompanied by a novel NA, gain the ability to transmit between humans, the potential exists to cause another pandemic, as no one will have relevant pre-existing neutralizing antibody immunity to this novel virus.

The functions of both HA and NA involve interaction with sialic acid, a terminal structure bound to underlying sugar residues expressed by glycoproteins or glycolipids at the cell surface (Gottschalk, 1958; Palese et al., 1974). The binding of HA to sialic acids presented by cellular receptors triggers cell entry by clathrin-mediated endocytosis, although other endocytic routes, including micropinocytosis, may also be used [reviewed in (Lakadamyali et al., 2004)]. A major function of NA occurs in the final stage of infection. Viral NA removes sialic acids from both cellular receptors and from newly synthesized HA and NA on nascent virions, which have been sialylated as part of the glycosylation processes within the host cell (Palese et al., 1974; Basak et al., 1985). NA cleavage of sialic acids prevents virion aggregation and stops virus binding back to the dying host cell *via* the HA, enabling efficient release of virion progeny and spread to new cell targets (Palese et al., 1974).

The role of the viral HA in attachment and infection has been well explored, yet examination of the role of NA in the IAV infection cycle has been largely limited to its role in aiding exit of virion progeny from infected cells. The majority of reviews on NA have focused on viral inhibitors that target NA and block this function. While some studies have suggested NA function does not influence the early stage of IAV infection (Liu et al., 1995), arguably, the sialidase activity of NA aids the virus to gain access to cells by catalyzing the cleavage of sialic acids presented by decoy receptors, such as mucins (Kesimer et al., 2009; McAuley et al., 2017), potentially providing NA with an important role in viral entry. In addition, experiments showing a decrease in infection of cells in the presence of NA-blocking drugs provide evidence for a role of NA in a virus entry step (Matrosovich et al., 2004; Ohuchi et al., 2006; Su et al., 2009; Gulati et al., 2013). As such there is a need for better understanding of the complex role of NA in the influenza infection and replication cycle, particularly with consideration to how the disparate roles of HA and NA glycoproteins need to achieve a functional balance in order to maintain viral fitness. Therefore, we sought to review the existing literature to evaluate the NA structure and function in relation to its role in the IAV infection cycle.

# NA STRUCTURE

The NA assembles as a tetramer of four identical polypeptides and, when embedded in the envelope of the virus, accounts for approximately 10–20% of the total glycoproteins on the virion surface, with about 40–50 NA spikes and 300–400 HA spikes on an average sized virion of 120 nm (Varghese et al., 1983; Ward et al., 1983; Moules et al., 2010). The four monomers, each of approximately 470 amino acids, fold into four distinct structural domains: the cytoplasmic tail, the transmembrane region, the stalk, and the catalytic head (**Figure 1**). Cryoelectron tomography studies have indicated that the NA tetramer exists in local clusters on the virion surface or as isolated spikes surrounded by HA (Harris et al., 2006). Depending on the length of the stalk region, the NA may protrude slightly more (Harris et al., 2006) or less (Matsuoka et al., 2009) above the viral envelope than the HA, which may influence the overall enzymatic activity of the virus.

# Cytoplasmic Tail

Suggesting that the NA cytoplasmic tail is involved in critical viral functions, the N-terminal domain sequence is nearly 100% conserved across all IAV subtypes and consists of the sequence MNPNQK (Blok and Air, 1982). Reverse engineered viruses containing site-specific mutations in this domain exhibit altered virion morphology and reduced replicative yields (Mitnaul et al., 1996; Jin et al., 1997; Barman et al., 2004). IAV engineered to encode an NA lacking a cytoplasmic tail could still be rescued albeit with a markedly attenuated phenotype (Garcia-Sastre and Palese, 1995). The altered morphology and attenuated infectivity of viruses expressing NA lacking the cytoplasmic tail domain are thought to be due to a lack of interaction with the membraneassociated matrix M1 viral protein (Enami and Enami, 1996), which ultimately alters the efficiency of budding from the infected host cell (Jin et al., 1997; Ali et al., 2000; Barman et al., 2001; Mintaev et al., 2014). Determinants in both the cytoplasmic tail domain and the transmembrane domain contribute to the transport of the glycoprotein to the apical plasma membrane (Kundu and Nayak, 1994; Kundu et al., 1996). However, the role of the tail domain in packaging the surface NA into virions remains unclear. A complete loss of the tail domain (Garcia-Sastre and Palese, 1995) resulted in a 50% reduction in the amount of NA in infected cells. This corresponded to a reduction in the amount of NA incorporated into virions, suggesting efficient packaging of existing NA. However, the absence of all tail amino acids except for the initiating methionine gave rise to virus that also showed markedly less incorporation of NA into virions, but in this case, NA was present at the plasma membrane at similar levels to wild-type virus (Mitnaul et al., 1996).

# Transmembrane Domain

The N-terminal hydrophobic transmembrane domain, which attaches the NA to the viral envelope (Bos et al., 1984), contains a variable sequence of amino acids spanning residue numbers 7–29 and is predicted to form an alpha helix (Blok and Air, 1982; Air, 2012) with interspersed polar residues driving subunitsubunit interactions (Nordholm et al., 2013). The transmembrane domain provides signals for translocation from the endoplasmic reticulum to the apical surface, as well as association with lipid rafts (Barman and Nayak, 2000). The N-terminal amino acids (positions 1–74), which include both the cytoplasmic tail, the transmembrane domain, and some of the stalk region have been reported to be sufficient to target the cell membrane and for the formation of the NA tetramer complex (Kundu et al., 1991, 1996; da Silva et al., 2013; Nordholm et al., 2013). Implicating the function of the transmembrane domain directly in the translocation of NA to the apical membrane, membrane trafficking can also occur in the absence of the stalk and head domain (Ernst et al., 2013). Mutation of specific amino acids

within the transmembrane domain can induce stable architectural differences in the anchoring signal region that result in diminished transport to the plasma membrane (Ernst et al., 2013).

### Stalk

The stalk domains of NAs of different IAV subtypes share some structural features, but the number and sequence of amino acid residues can vary considerably (Blok and Air, 1982). Regardless of this variability, all NA stalk domains share some structural features, including at least one cysteine residue and a potential glycosylation site (Blok and Air, 1982; Air, 2012). The cysteine residue(s) may assist with tetramer stabilization by enabling disulfide bonds to form between each monomeric NA unit (Blok and Air, 1982; Ward et al., 1983). While the cysteine residues may occur at variable positions in the NA stalks of different subtypes, the tetrameric structure of NA allows them to align, so that the disulfide bonds can form between the pairs of cysteine residues on neighboring monomers (Blok and Air, 1982). The presence of carbohydrate side chains within the stalk is thought to contribute further to the stability of the tetramer (Blok and Air, 1982).

The length of the stalk region of different subtypes can have significant impact on particular virus characteristics. Using reverse engineering techniques, mutant viruses unable to produce the NA stalk were able to be rescued in tissue culture cells and replicate to the equivalent titer as the unmodified parent virus, but could not replicate in eggs or mice (Castrucci and Kawaoka, 1993). Using a series of NA mutants differing only in stalk length, studies showed that while there was no correlation between the stalk length and the ability to cleave fetuin or a small substrate *in vitro*, enhanced virus replication in eggs correlated closely with increasing stalk length (Els et al., 1985; Castrucci and Kawaoka, 1993). Viruses presenting NAs with shortened stalk domains have also been reported to elute less efficiently from chicken erythrocytes (Els et al., 1985; Castrucci and Kawaoka, 1993). Reduced stalk length has been commonly thought to impact NA activity of virions because the diminished height may hinder access to cellular sialic acid expressing receptors and that the towering HA blocks the shorter NA catalytic domain from gaining access to the sialic acids (Baigent and McCauley, 2001) (**Figure 2**).

More recently, it has been postulated that the limited access to substrate theory may not fully explain the reduced sialidase activity of stalk-deletion mutants, with the argument that the clustering of NAs on the virion surface would diminish any blocking effects of nearby HA. This view was backed up by molecular dynamics simulations providing evidence that a reduced stalk size also alters the geometry and dynamics of the enzymatic pocket itself, modifying its affinity for sialic acids (Durrant et al., 2016). Further evidence for the impact of the stalk structure on the activity of the NA catalytic domain comes from experiments where the same NA head domain was expressed with different tetramerizing domains as artificial stalks (Schmidt et al., 2011). A tetrabrachion stalk conferred similar properties as the native NA, whereas the yeast stalk (GCN4-pLI) reduced substrate affinity and inhibitor binding. Others have also demonstrated that a single mutation in the stalk can significantly affect enzyme activity, presumably affecting the stability of the tetramer (Zanin et al., 2017).

There is also a mounting evidence for a relationship between NA stalk length and virulence in mammalian models of infection. NA stalk truncation mutants of the 2009 pandemic virus

A(H1N1)pdm09 showed greater lethality in mice and virulence in ferrets than the untruncated counterpart (Park et al., 2017). Loss of the glycosylation site in the stalk likewise enhanced virulence in mice (Park et al., 2017). A deletion of 20 amino acids is frequently observed upon transmission of IAV from waterfowl to domestic poultry (Matsuoka et al., 2009; Hoffmann et al., 2012; Blumenkrantz et al., 2013; Sun et al., 2013; Bi et al., 2015). As such, the length of the stalk domain may be a species-specific determinant of viral adaptation and virulence that we are yet to fully understand (Li et al., 2014; Stech et al., 2015; Park et al., 2017).

### Head Domain

Peptide maps from crystallized NA catalytic heads were first detailed in 1978 (Laver, 1978). However, the actual structure of this domain was first described in 1983 for IAVs and in 1992 for IBVs (Varghese et al., 1983; Burmeister et al., 1992). The catalytic head of all NAs consists of a box-shaped structure comprising four monomers (**Figure 3**). Each monomer is in the form of a six-bladed propeller structure, with each blade having four anti-parallel β-sheets that are stabilized by disulfide bonds and connected by loops of variable length (Varghese et al., 1983). A functional catalytic site is present on the surface of each monomer and is directed sideward rather than upward, a property consistent with the ability to cleave sialic acids from nearby membrane glycoproteins to prevent virus trapping (Colman et al., 1983; Burmeister et al., 1992). These catalytic sites are characterized by a large cavity with an unusually large number of charged residues in the pocket and around its rim (Colman et al., 1983; Varghese et al., 1992). The tetrameric form of NA is considered optimal for enzyme activity, and mutations that lead to instability of the tetramer lead to decreased enzyme activity (McKimm-Breschkin et al., 1996b; Fujisaki et al., 2012; McKimm-Breschkin et al., 2013). While it has been reported that monomers alone have no enzyme activity (Air, 2012) and usually expression of recombinant soluble NA heads requires a synthetic tetramerization domain for active NA (Schmidt et al., 2011), there are reports of expression of soluble recombinant monomeric influenza NA heads in both yeast and mammalian cells that have comparable properties to the native enzyme (Yongkiettrakul et al., 2009; Nivitchanyong et al., 2011). When the head domain of NA is proteolytically cleaved from the remaining NA tetrameric stalk embedded in the virion, the enzymatic properties remain active and the heads retain the tetrameric state of purification (Laver, 1978; McKimm-Breschkin et al., 1991).

The NA active site consists of an inner shell of eight highly conserved residues that interact directly with sialic acids (Arg118, Asp151, Arg152, Arg224, Glu276, Arg292, Arg371, and Tyr406) (**Figure 4**). In addition, there is an outer shell of 10 residues, which do not contact sialic acid, but which have an important structural role and are defined as framework residues. These comprise Glu119, Arg156, Trp178, Ser179, Asp198, Ile222, Glu227, Glu277, Asn294, and Glu425 (Colman et al., 1983, 1993; Burmeister et al., 1992). Three arginine residues (Arg118, 292, 371) interact with the carboxylate of the sialic acid substrate. Arg152 binds to the acetamido group on the sugar ring, while Glu276 interacts with the 8- and 9-hydroxyl groups on the glycerol side chain. The enzyme active site is said to be highly conserved in both spatial orientation and sequence properties, making it an ideal target for drug inhibition.

NA active site variants that either occur naturally or are created *via* genetic modification have shown that both framework

stabilized by disulfide bonds and connected by loops of variable length. Sialic acid (yellow structure) is observable on the head of each monomer. The boxed region is magnified in the inset and shows the receptor-binding pocket to which sialic acid (yellow compound) binds. The structure was generated in Pymol using structural information from Protein Data Bank code 4GZX.

second site binds sialic acid by making contacts with Ser367, Ser370, Ser372, Asn400, Trp403 and Lys432 (highlighted in yellow), but bound sialic acid at this site is not released by the NA activity. The structure was generated in Pymol using structural information from Protein Data Bank code 1MWE for the A/tern/Australia/G70C/75 NA/Neu5Ac complexed at 4 °C.

and catalytic residues can alter the viral replicative ability, transmissibility, and susceptibility to antiviral inhibitors to varying degrees (Lentz et al., 1987; McKimm-Breschkin, 2000, 2013; Abed et al., 2006; Meijer et al., 2009; Richard et al., 2011; Chao et al., 2012; Govorkova, 2013). An H1N1 virus that emerged in 2007–2008 had a single H275Y amino acid change conferring oseltamivir resistance (Hurt et al., 2011; Takashita et al., 2015). This virus showed no decrease in fitness, spreading globally over the next year. Therefore, it remains important to monitor antiviral susceptibility of viruses circulating in the community.

# STRUCTURAL RELATIONSHIPS BETWEEN NA SUBTYPES

Phylogenic mapping, which included comparisons of genetic and structural relationships between NAs from different viruses (not including the recently discovered bat viruses) revealed that IAV NAs fall into two distinct groups, regardless of their serotype identification (i.e. N1–N9) (Russell et al., 2006). Group 1 contains N1, N4, N5, and N8 serotypes, while group 2 contains N2, N3, N6, N7, and N9 serotypes. Crystal structures of the group 1 and group 2 NAs reveal a particularly striking feature in the catalytic domain referred to as the '150 loop.' This loop comprises residues 147–152, which forms one corner of the enzyme active site, and is able to exist in at least two conformations (Russell et al., 2006). Group 1 NA structures have an additional cavity next to the active site, which is created by the movement of the 150 loop during conformational changes brought about by substrate binding within the active site. Structurally, the group 2 NAs do not appear to have the cavity. By X-ray crystallography, the A(H1N1)pdm09 NA also appears to lack the 150-cavity and has more structural similarities to the group 2 NAs (Li et al., 2010; Amaro et al., 2011). However, molecular dynamics studies suggest that the A(H1N1)pdm09 NA and the group 2 NAs do have the 150-loop flexibility, but there may be fewer of the population of the open 150-loop conformation, hence not seen in the static structures in crystals (Amaro et al., 2011). As NA inhibitors have been developed based on crystal structures of group 2 NAs, there is now a great interest in the development of novel inhibitors that target the cavity of group 1 NAs (Russell et al., 2006; Li et al., 2010; Amaro et al., 2011). Sialic acid derivatives that selectively lock the 150 loop in the open cavity conformation inhibit the group 1 sialidases, including the A(H1N1)pdm09 NA, further supporting the fact that this pandemic NA does have an accessible 150 cavity that is exposed to some extent upon the binding of an appropriate inhibitor (Rudrawar et al., 2010).

### NA Hemadsorption Site

X-ray crystal structures of N6 and N9 that have revealed sialic acids can also bind at a discrete second site on the NA head (Varghese et al., 1997; Air, 2012; Streltsov et al., 2015), resulting in the binding of red blood cells (hemadsorption) to the NA. The second site, a shallow pocket located next to the catalytic site, is made up of three surface loops and binds sialic acid by making contacts with Ser367, Ser370, Ser372, Asn400, Trp403, and Lys432 (**Figure 4**), but bound sialic acids at this site are not released by the NA activity (Varghese et al., 1997). The second site appears to be a common feature of avian NAs of different subtypes, while swine influenza strains have lost several of these conserved residues, so theoretically they do not have the second site (Kobasa et al., 1997; Varghese et al., 1997). However, the Brownian dynamics simulation of human N2 from the 1957 pdmIAV and the A(H1N1)pdm09 NA suggests that some of the key structural features are retained (Sung et al., 2010), and NMR studies subsequently added support to the presence of the second site in these viruses (Lai et al., 2012). Substitutions at different positions in the three loops of the second sialic acid-binding site on the NA of human H1N1 IAV have been shown to have a profound effect on binding and cleavage of multivalent, but not monovalent, receptors and significantly affect virus replication (Du et al., 2018). Linking the second sialic acid-binding site with host tropism, it has been reported that for avian viruses that have succeeded in zoonotic transmission to man, such as the pandemic 1918 and 1968 IAVs and the emerging H7N9 IAV, this site contains point mutations (Uhlendorff et al., 2009; Dai et al., 2017). It has been hypothesized that substitutions in the second sialic acidbinding site enable the enzymatic activity of the NA of newly emerging viruses to be compatible with an HA that is evolving toward human receptor-binding specificity (Du et al., 2018).

Understanding the biological function of the NA second sialic acid-binding site has been challenging. One study has shown that the ability of NA to bind red blood cells correlated with the cleavage efficiencies of multivalent substrates like fetuin (Uhlendorff et al., 2009). The observation that substrate binding *via* the second sialic acid-binding site in H7N9 viruses enhances NA catalytic efficiency against the same substrate (Dai et al., 2017) is possibly achieved by recruiting and keeping multivalent sialosides close to the active site (Uhlendorff et al., 2009; Dai et al., 2017). The second site has also been shown to be a highly conserved target for a novel influenza inhibitor found in the saliva of mice that impact the ability of the infection to progress to the lungs of infected mice when the inoculum is initially confined to the upper respiratory tract (Ivinson et al., 2017). The amino acids at positions 368–370 (N2 numbering) on the rim of the second site dictated the *in vitro* susceptibility of different viral strains to the salivary inhibitor and their ability to progress to the lungs of mice (Gilbertson et al., 2017).

### NA FUNCTIONAL ROLES IN REPLICATION

### Virus Entry

NA activity and cleavage of sialic acids have long been thought to enable movement of the virion through mucus (Burnet, 1948). Airway mucus is comprised of a large network of sialylated glycoproteins known as gel-forming mucins (MUC5AC and MUC5B in particular), which interconnect and serve as a scaffold to present secreted antiproteases, antioxidants, antimicrobial proteins, secretory immunoglobulins (IgA), cytokines, and other innate defensive molecules (Lillehoj and Kim, 2002). Hypersecretion of mucus during the course of respiratory tract infection can occlude the airways and increase the lung pathology (Rose and Voynow, 2006). In murine models of influenza infection, overexpression of the major respiratory tract gel-forming mucin, Muc5ac (non-human form of MUC5AC), revealed that this glycoprotein presents SAα2-3Gal, which can bind virus and limit infection of the underlying epithelia (Ehre et al., 2012). This supports the proposal that the physical properties of the gel-forming mucins serve as a trap by presenting decoy receptors to which the inhaled pathogen binds and is then cleared by way of the mucociliary escalator (Button et al., 2012). IAV has been shown to interact with secreted mucus on frozen human trachea and bronchus tissue sections, and bead-bound mucins inhibited the NA cleavage of substrate (Cohen et al., 2013). To add credence to the hypothesis that NA functions to aid viral movement through the respiratory mucus layer, NA inhibitors have been shown to block IAV entry into differentiated human tracheobronchial and nasal epithelial cells, as well as porcine cells that secrete mucus (Matrosovich et al., 2004; Yang et al., 2014). Exogenous NA also enhanced passage through the mucus layer (Yang et al., 2014). This suggests that NA is needed to remove decoy sialic acids presented on mucins, cilia, and cellular glycocalyx in order for virus to efficiently access functional receptors on the surface of target cells.

### Receptor Binding

In addition to the non-catalytic sialic-binding site that is structurally distinct from the NA active site in N6 and N9 NAs (Varghese et al., 1997; Air, 2012; Streltsov et al., 2015), more recent studies have shown that the NAs of human H3N2 viruses isolated since 1994 can also demonstrate agglutination of red blood cells after passage in MDCK cells, but not in eggs (Lin et al., 2010; Hooper and Bloom, 2013; Mohr et al., 2015). This property of the more recent N2 NAs was first noted because many H3N2 isolates showed weak HA-mediated binding to chicken red blood cells allowing NA-dependent hemagglutination to be detected. NA agglutination is inhibited by NA inhibitor drugs (NAIs) but only poorly inhibited by post-infection ferret antisera, thus distinguishing it from HA binding. In contrast to the N6 and N9, both sialidase and receptor-binding functions reside in the N2 active site, yet the catalytic and receptor-binding sites do not appear to be identical since relative sensitivity to inhibition of the two functions varies with oseltamivir, zanamivir, and peramivir (Mohr et al., 2015). Substitution of aspartate at position 151 near the active site to glycine, alanine or asparagine, or threonine 148 to isoleucine in H3N2 NAs (Lin et al., 2010; Mohr et al., 2015) or glycine 147 to arginine in N1 NAs (Hooper and Bloom, 2013) correlates with the acquisition of receptor binding. An H150R substitution has also recently been shown to correlate with NA receptor binding and has been found in both clinical samples and passaged viruses (Mogling et al., 2017). Interestingly, the affinity of the NA receptor-binding site to sialyl lactose is much stronger than the corresponding affinities of HA with its sialylated receptors (Zhu et al., 2012). The fact that entry and infection of MDCK cells with viruses having NA D151G can be blocked by NAIs (Gulati et al., 2013) suggests this NA active site-associated receptor binding function may play an important biological role for these H3N2 isolates.

### Virus Internalization

NAIs were found to reduce infection efficiency of cell lines without inhibiting virus binding or fusion activity, supporting a role for the NA during the viral entry process (Ohuchi et al., 2006). It was proposed that the NA facilitated movement of the virus across the cell surface by repeated binding and release steps from an endocytosis inactive site on the cell, to an active site, thereby increasing the efficiency of viral uptake. With the recent development of a biolayer interferometry assay, there is now clear evidence that viral NA plays a major role in driving virus particles over sialylated receptor surfaces. Using this method, Guo et al. (2018) showed that NA contributed to the initial rate of virus binding to sialoglycoproteins after which multiple low-affinity HA-sialic acid interactions take place. The rapid association and dissociation of these allow the NA to remove sialic acids and create a receptor density gradient that enables the rolling of virus particles across the surface.

NA has also been shown to enhance HA-dependent influenza virus fusion and infectivity using a cell-cell fusion assay and an HIV-based pseudotype infectivity assay (Su et al., 2009). When the NA gene from H9N2, H5N1, or A(H1N1)pdm09 virus was expressed on a PR8 background, the replication kinetics were similar *in vitro* (MDCK cells) and *in vivo* (mice) (Chen et al., 2013), yet the initial infection kinetics and virusinduced fusion and elution from erythrocytes were affected, implicating a role for NA during the early stage of infection (Chen et al., 2013).

### Catalytic Activity

By far, the most characterized function of NA is its action as a sialidase enzyme, enabling release of new virion progeny by enzymatically cleaving sialic acids from cell surface receptors and from carbohydrate side chains on nascent virions (Gottschalk, 1958; Palese et al., 1974). When NA activity is inhibited by the use of antivirals that target the enzymatic site, or through alteration of key amino acid residues, such as those identified to be integral in the catalytic process, the budding virions aggregate on the cell surface instead of being released (Lentz et al., 1987; Tarbet et al., 2014; Yang et al., 2016). This clumping of virions is due to HA on newly released virus binding to the sialic acids expressed on receptors in the vicinity of the budding site and to carbohydrate side chains on the HA and NA of progeny viruses that still contain terminal sialic acids in the absence of NA activity.

The catalytic mechanism of NA has not yet been completely resolved but is expected to begin with the binding of substrate to the active site *via* interactions with the catalytic residues and involves salt-bridge formation between the carboxylate of the sialic acid and the three arginine cluster at one end of the active site. Functional and structural evidence for the formation of a covalent intermediate between the C-2 on the sugar ring and the Tyr406 was obtained using a 2,3-difluoro sialic acid derivative (DFSA), which exhibits slow turnover, permitting accumulation of the covalent intermediate (Kim et al., 2013; Vavricka et al., 2013). This confirmed that the Tyr406 functions as the catalytic nucleophile. This leads to a change in the chair conformation of sialic acid with the carboxylate in the axial position, to a boat conformation with the carboxylate rotated into the pseudo equatorial position, then eventual cleavage of the sialic acid molecule from the preceding galactose residue. This cleavage first results in the release of sialic acid in an α-anomer conformation, then conversion into a β-anomer state shortly thereafter (Air, 2012).

The optimal activity of NA occurs at a pH range of 5.5–6.5 (Mountford et al., 1982; Lentz et al., 1987; McKimm-Breschkin et al., 2013); however, some viruses have been reported to have a stable NA activity at a lower pH range of 4–5, which has been shown to enhance replication kinetics (Takahashi and Suzuki, 2015). The presence of Ca2+ is thought to be essential during the reaction for both thermostability and enzyme activity of the NA. Using common fluorometric activity assays, increasing calcium ion concentration was shown to increase NA activity (Dimmock, 1971; Potier et al., 1979; Chong et al., 1991; Johansson and Brett, 2003). In crystal structures of NA bound to sialic acid, up to 5 Ca2+ ions per subunit that forms each tetramer are observed (Varghese et al., 1983; Russell et al., 2006; Xu et al., 2008; Lawrenz et al., 2010). An X-ray crystal structure of a whale N9 NA revealed that there were structural alterations near the substratebinding site in the absence of calcium (Smith et al., 2006).

### NA Substrate Specificity

Avian IAVs express HAs that typically bind to galactose in an α2-3 linkage (SAα2-3Gal). For avian influenza viruses to undergo human to human transmission, the HA must acquire the capacity to bind SAα2-6Gal through mutations within the receptorbinding pocket (Gambaryan and Matrosovich, 2015). In some

*<sup>(</sup>Continued)*

cell-surface mucins) by NA enables HA access to sialic acids expressed by entry receptors and efficient endocytosis. After endosome escape of the viral genome and its migration to the nucleus, replication of the genome, synthesis of viral mRNAs, and translation of viral proteins take place. New virions assemble at the cell surface and are released from the cell by budding. As the viral components bud from the cell, NA cleaves sialic acids from receptors near the budding site to prevent virions binding back to the dying cell. NA cleavage of sialic acids from the carbohydrate side chains of nascent HA and NA also prevents newly budded virus from clumping together. Both these functions enable efficient release of the nascent virions from the cell. (B) If the HA and NA are mismatched and have mutations in important binding or catalytic sites that alter function, the relative activity of the two proteins may be imbalanced. If the sialidase function of NA is suboptimal, virus may remain bound by decoy receptors, which may shed and block virus entry into the cell. As the virus buds from the cell, an imbalance of HA and NA function may result in the lack of release of the virions due to the binding of HA to the sialic acids expressed at the cell surface that have not been removed by the NA. (C) Alternatively, if the sialidase activity of NA is too strong when compared to the HA-binding activity, sialic acids may be removed from receptors at the expense of the HA being able to bind and trigger endocytosis. FIGURE 5 |

instances, the mutations are such that the HA will retain SAα2- 3Gal binding and have dual specificity. Through further evolution in humans, the HA can become solely specific for SAα2-6Gal (Couceiro et al., 1993; Matrosovich et al., 1997, 2007). Similar to HA, the specificity of the active site of the viral NA evolves with time in the human host toward SAα2-6Gal (Kobasa et al., 1999; Gambaryan and Matrosovich, 2015). However, unlike the HA, it always maintains the ability to cleave SAα2-3Gal, even in viruses whose sole HA specificity is for SAα2-6Gal (Baum and Paulson, 1991; Kobasa et al., 1999). This evolution of specificity for sialic acid bound to galactose in different conformations is most likely due to the presentation of these glycoproteins on the surface of the target epithelium. In the upper airways, human tracheal epithelium expresses sialylated glycoproteins that are bound in an SAα2-6Gal linkage (Couceiro et al., 1993). In contrast, human bronchial mucus secretions contain large glycoproteins that express SAα2-3Gal, and as such, maintenance of NA activity for this linkage may be necessary for virion movement through the mucus barrier.

Functional evolution of NA has been shown to occur by amino acid substitutions that subtly alter the conformation of the NA catalytic domain to enable a different form of sialic acid to bind to the active site (Kobasa et al., 1999). A single change of isoleucine 275 to valine in N2 NA enables the shift in NA specificity toward increased activity for SAα2-6Gal, while other mutations are thought to subtly alter the conformation of the active site to accommodate this linkage of sialic acid (Kobasa et al., 1999).

### HA:NA BALANCE

With respect to the ability for IAV to circumnavigate the mucosal environment and successfully infect underlying epithelial cells, the HA and NA need to have complementary receptor and ligand-binding specificity. It is also imperative that the relative activity of the two proteins is balanced to maintain the ability to infect and to release from cells efficiently (**Figure 5**). The importance of this functional balance was initially demonstrated when the first NA inhibitor resistant mutants were analyzed. Unexpectedly, rather than mutations in the NA, these drug resistant viruses had mutations in the HA (McKimm-Breschkin et al., 1996a,b, 1998). The HA mutations were found to reduce the affinity for receptors, so that less NA activity was required for virion release. However, while they had a fitness advantage in the presence of NA inhibitors, in the absence of NA inhibitors, the receptor binding was so poor for some of these mutants that the NA was able to cleave off the receptors before HA binding could take place. Such viruses are thus drug dependent (McKimm-Breschkin et al., 1996a; Blick et al., 1998). Others subsequently confirmed the need for balanced HA and NA activities (Wagner et al., 2002). The relevance to *in vivo* adaptation of influenza virus was shown by the isolation of several H3N2 viruses from patients that were reported to have little or no NA activity (Ferraris et al., 2006). A weak-binding HA was found to compensate for the absence of NA activity (Richard et al., 2012). Obviously, evolution of an HA or NA that negatively impacts viral attachment, replication, and transmission results in a less fit virus. In order to survive, compensatory mutations are needed to restore fitness (Lin et al., 2010; Mohr et al., 2015). Thus, for a human virus to gain efficient access to the cell surface *in vivo*, it needs to have a combination of HA and NA activities that enable escape from inhibition in the mucus layer by having either an HA with low avidity for mucin-bound SAα2-3Gal or an NA with greater activity for SAα2-3Gal, or a combination of the two. To attach and enter a cell, HA avidity for SAα2- 6Gal must be strong enough to enable binding before the NA can cleave receptors. However, HA binding cannot be too strong, since release of progeny virions and prevention of aggregation at the cell surface need access of the NA to cleave the SAα2-6Gal.

Traditionally, incubation of IAV with red blood cells at different temperatures enables the functional balance of the HA and NA to be explored. As NA is inactive at 4°C but active at 37 °C, the ability of HA to agglutinate red cells at 4°C and of NA to elute virus at 37°C can be measured. However, this is no longer so clear cut. The isolation of the NA inhibitorresistant HA mutants revealed that elution can occur rapidly with little NA activity, due to a weak HA (McKimm-Breschkin et al., 2012, 2013). More insight into the relative roles can be obtained by further modifications of the HA elution assay. When the eluted virus/red blood cells are mixed and re-incubated at 4 °C, if elution is due to a weak HA, virus will rebind. If the NA has cleaved the receptors, virus cannot rebind as in the traditional reaction. Alternatively, the addition of NA inhibitors will prevent elution if it is by the NA, but not by the HA (McKimm-Breschkin et al., 2012, 2013). Furthermore, NA inhibitors will also prevent agglutination if it is through the NA active site, as seen for the H3N2 viruses (Lin et al., 2010; Mohr et al., 2015). However, even with these variations on the assay, binding and elution vary depending on the species from which the red blood cells are derived from and are due to different sialic acid linkages presented at their surface (Ferraris et al., 2006; Richard et al., 2012).

Thus, the optimal functional balance of the HA and NA glycoproteins is difficult to measure as a number of physical characteristics of the virus can influence the properties of the HA- and NA-mediated receptor interactions. HA and NA affinity for and kinetics of interaction with sialic acids are the foremost determinants for the ability of the virion to infect a cell. On an average virion, there are 300–400 HA spikes and 40–50 NA spikes (Harris et al., 2006). The excess of HA over NA is perhaps reflective of the weak interaction of HA for sialic acids (Sauter et al., 1989) and the need to form several connections to form a stable interaction. The position and amount of NA present on the virion then plays an important role in gaining access to the cell surface sialic acids; its catalytic activity can directly influence efficiency of viral release, movement through mucus, and potential infection of new cells. Virion morphology can also contribute to altering the protein distribution and amount of NA and HA on the virion surface, potentially altering viral fitness and replication kinetics (Wasilewski et al., 2012). Finally, the ability for NA to access substrate either by the distribution of the NA on the virion or by variation in the length of protrusion of the NA spike can significantly influence both HA binding to a receptor and cleavage by NA from the receptor (Els et al., 1985; Castrucci and Kawaoka, 1993; Baigent and McCauley, 2001; Durrant et al., 2016). The recent study of Guo et al. (2018) further implicates the critical importance of HA:NA functional balance on virion movement through the mucus layer and over epithelial surfaces, a dynamic rolling process that may also be involved in cell-to-cell spread across the respiratory epithelium.

### CONCLUSION

Rather than just a sialidase that facilitates virus release from infected cells, the NA is a complicated multifunctional protein with an important role at many stages of the infectious process. While the NA is the main target for current antiviral therapies (Ison, 2015), recent approaches to new influenza therapy include targeting the HA with monoclonal antibodies (Nachbagauer and Krammer, 2017). However, given the NA also has the capacity to bind receptors, there needs to be caution in this approach, as it is possible that compensating mutations in the NA may allow escape from inhibition of the HA. As antibody levels against NA in children, adults, and the elderly correlate well with functional neuraminidase inhibition titers (Rajendran et al., 2017), altering vaccine strategies to enable efficient boosting of broadly cross-reactive antibodies against neuraminidase (Sandbulte et al., 2007; Marcelin et al., 2011; Liu et al., 2015; Chen et al., 2018) may be an important consideration in the campaign against the incredibly adaptable influenza virus.

### AUTHOR CONTRIBUTIONS

JM and JM-B contributed to the conception and design of the review. JM wrote the first draft of the manuscript and designed figures. BG, ST, LB, and JM-B contributed to various sections of the manuscript, including figures. All authors contributed to manuscript revision, read, and approved the submitted version.

### FUNDING

This work was supported by the Australian Government National Health and Medical Research Council grant 1079924 and program grant 1071916.

### REFERENCES


### ACKNOWLEDGMENTS

The authors would like to acknowledge Dr. Victor Streltsov (Florey Institute of Neuroscience and Mental Health, Victoria, Australia) for his technical assistance in developing the structural images and Dr. Lynne Waddington, CSIRO Melbourne, Australia for provision of the 2D-electron micrograph image.


A virus N1 neuraminidase contributes to enzymatic activity. *J. Virol.* 92, e01243–e01218. doi: 10.1128/JVI.01243-18


neuraminidase and its alteration in pandemic viruses. *Arch. Virol.* 154, 945–957. doi: 10.1007/s00705-009-0393-x


**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 © 2019 McAuley, Gilbertson, Trifkovic, Brown and McKimm-Breschkin. 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.*

# Detection of the Onset of the Epidemic Period of Respiratory Syncytial Virus Infection in Japan

Hidetomi Yamagami <sup>1</sup> \*, Hirokazu Kimura<sup>2</sup> , Takafumi Hashimoto<sup>3</sup> , Isao Kusakawa<sup>4</sup> and Satoshi Kusuda<sup>5</sup>

<sup>1</sup> Medical Affairs, AbbVie GK, Tokyo, Japan, <sup>2</sup> Department of Health Science, Gunma Paz University Graduate School of Health Science, Gunma, Japan, <sup>3</sup> Data Science and Stat, AbbVie GK, Tokyo, Japan, <sup>4</sup> Department of Pediatrics, St. Luke's International Hospital, Tokyo, Japan, <sup>5</sup> Department of Pediatrics, Faculty of Medicine, Kyorin University, Tokyo, Japan

### Edited by:

Sunil Dhiman, Defence Research and Development Establishment (DRDE), India

### Reviewed by:

Karen Bohmwald, Pontificia Universidad Católica de Chile, Chile Bernhard Resch, Medical University of Graz, Austria

\*Correspondence: Hidetomi Yamagami hidetomi.yamagami@abbvie.com

### Specialty section:

This article was submitted to Infectious Diseases – Surveillance, Prevention and Treatment, a section of the journal Frontiers in Public Health

Received: 20 December 2018 Accepted: 14 February 2019 Published: 07 March 2019

### Citation:

Yamagami H, Kimura H, Hashimoto T, Kusakawa I and Kusuda S (2019) Detection of the Onset of the Epidemic Period of Respiratory Syncytial Virus Infection in Japan. Front. Public Health 7:39. doi: 10.3389/fpubh.2019.00039 Respiratory syncytial virus (RSV) is a leading cause of lower respiratory tract infection in young children worldwide. An annual epidemic of RSV infection generally begins around autumn, reaching a peak at the end of year in Japan, but in 2017 it started in early July and peaked in September. As the onset timing of RSV season varies, it is important to detect the beginning of an epidemic, to enable the implementation of preventive measures. However, there are currently no specified criteria or methods to determine the onset of RSV season in a timely manner. Therefore, we developed a model to detect the epidemic onset, based on data from the Infectious Diseases Weekly Report from 2012 to 2017. The 47 prefectures of Japan span 11 climate zones, which affect the timing of epidemic onset. Therefore, the onset of RSV season was assessed separately in each prefecture. Non-linear regression analysis was performed to generate a mathematical model of the annual epidemic cycle for each prefecture. A search index was used to determine the onset of RSV season, which was estimated using the number of RSV reports per week within an epidemic period (RSV-reports/w) and the number of reported cases included within an epidemic period relative to the total number of RSV reports (capture rate). A number of RSV-reports/w, which was used as a threshold (a number at onset line) to determine the condition of the onset of RSV season, was then estimated based on the search index. The mean number at the onset of RSV season for 47 prefectures was 29.7 reports/week (median 21.0, range 6.0–121.0 reports/ week). The model also showed that the onset of RSV season in 2017 was more than 1 month earlier than the previous year. In conclusion, the model detected epidemic cycles and their onset conditions in all prefectures, despite the 11 climate zones of Japan. The results are expected to contribute to infant medical care by allowing medical personnel to take preventive measures promptly at the beginning of the epidemic RSV season.

Keywords: respiratory syncytial virus infection, palivizumab, IDWR, epidemic period, the onset of RSV season

### Yamagami et al. Criteria for RSV Epidemic Onset

# INTRODUCTION

Respiratory syncytial virus (RSV) is a major cause of lower respiratory tract infection in infants, and more than 50% of infants are infected for the first time by the age of 1 year, and almost 100% of infants by the age of 2 years (1, 2). Lifelong immunity is not acquired after the primary infection; therefore, reinfection with RSV is universally observed in nearly 50% of infants by the age of 2 years (1). In infants, about 50% of cases of pneumonia, and around 50% to 90% of cases of bronchiolitis are attributed to RSV infection (3). In particular, in premature infants and infants with chronic respiratory diseases such as bronchopulmonary dysplasia, RSV infection tends to be severe (4–7). Also, RSV infection tends to be severe even in infants with congenital heart diseases (8, 9) 1 and infants with immunodeficiency or Down's syndrome (10). It is important to take preventive measures for these high-risk infants, as RSV infection may take a fatal course in some cases (11).

Because the medicine for RSV infection has not been developed, patients receive supportive treatment mainly by transfusion and respiratory management. In Japan, public health insurance coverage was approved in 2002 for palivizumab (product name: Synagis <sup>R</sup> ), an anti-RSV humanized monoclonal antibody, which is used as a prophylactic medicine to prevent severe lower respiratory tract disease associated with RSV infection in high-risk infants. As palivizumab needs to be administered once a month throughout the epidemic period of RSV infection, for its appropriate use, it is important to accurately detect the onset of the epidemic period.

RSV infection is monitored via pediatric sentinel sites under the Act on Prevention of Infectious Diseases and Medical Care for Patients Suffering from Infectious Diseases (11). In this program, physicians at designated notification facilities (∼3,000 pediatric medical institutions at fixed points all around Japan to cover 10% of pediatric institutions in Japan) are required to report patients diagnosed with RSV infection based on their symptoms and findings of RSV antigen or by PCR, and the information is published in the Infectious Diseases Weekly Report (IDWR) as national surveillance data by the National Institute of Infectious Diseases (NIID)<sup>1</sup> . In Japan, the epidemic of RSV infection had generally begun around autumn, and the number of reported cases had increased rapidly in autumn, reached a peak at the end of the year, and continued until spring. No significant fluctuations were observed in the epidemic tendency nationwide until 2015; however, the number of reported cases increased from early August in 2016. In 2017, the number of reported cases increased rapidly from early July and reached a peak in September. Tendencies of fluctuations in the onset of the epidemic period are evident, as the number of RSV reports started to increase earlier than in the past several seasons (**Figure 1**). To resolve the issue, the guideline for the use of palivizumab in Japan was immediately revised in accordance with the observed epidemical change of RSV. As the onset of RSV season changes every year and there are no specified decision criteria for the onset of the epidemic RSV season, the onset of the RSV season should be determined on the basis of the IDWR and other included reports<sup>2</sup> .

In response to this, we investigated the onset of the epidemic period of RSV infection based on the IDWR surveillance data using a newly developed algorithm. The main objective of this study was to detect the onset weeks of the annual RSV epidemics in Japan to enable timely access to preventive measures for severe RSV disease.

# METHODS

### Data Source

Six epidemic periods with a 1-year cycle of RSV infection for 46 prefectures and five epidemic periods with a 1-year cycle of RSV infection for Okinawa were analyzed using the IDWR surveillance data from the first week of 2012 to the 24th week of 2018<sup>3</sup> . The numbers of sentinel medical institutions (SMI) in each prefecture were used on the basis of the average numbers of sentinel medical institutions in each prefecture in 2016. Weeks ending logs were referenced from the NIID report<sup>4</sup> .

### Modeling the RSV Season as a Periodic Oscillation

RSV incidence is reported weekly in each prefecture in Japan. Since epidemic cycles vary among regions in Japan, we generated a mathematical model of the annual epidemic cycle for each prefecture by applying non-linear regression analysis to data reported from the first week of 2012 to the 24th week of 2018. The following sine function was used to model the cyclical nature of the annual epidemics, where "Y" represents the number of RSV reports each week, "w" represents the week of the year (1–52) in which the cases were reported, and a0, a1, a2, and a<sup>3</sup> represent fixed parameters.

$$\mathbf{Y} = \mathbf{a}\_0 + (\mathbf{a}\_1 \times \sin(\mathbf{a}\_2 + \mathbf{a}\_3 \times (\mathbf{w} \times \pi/180))).\tag{1}$$

Because we focused on the rhythm of the epidemics, year-toyear variations in RSV reporting numbers (the amplitude of the function) were not incorporated into the model, and an upper limit was applied to variable Y. The model parameters were calculated using the "NLIN" non-linear regression procedure in SAS software (version 9.4; SAS Institute, Cary, NC, USA), with Gauss-Newton iteration. We confirmed that the convergence criterion was met for each prefecture. Also, visual confirmation of graphs of each data set confirmed that the incidence of RSV cases in each prefecture of Japan had sinusoidal periodicity.

When there was more than one peak between the 1st week and the 52nd week in a plot, the peak corresponding to the highest reported number of cases was selected as the main peak for calculation purposes. When there was more than one trough between the 1st week and the 52nd week in a plot, the trough falling just before the main peak was selected as the central trough for calculation purposes. The range of a 1-year cycle based on the sinusoidal curve was identified between one trough and next

<sup>2</sup>https://www.jpeds.or.jp/modules/news/index.php?content\_id=345 <sup>3</sup>https://www.niid.go.jp/niid/en/survaillance-data-table-english.html <sup>4</sup>https://www.niid.go.jp/niid/en/calendar-e.html

<sup>1</sup>https://www.niid.go.jp/niid/ja/rs-virus-m/rs-virus-idwrc/7509-idwrc-1734.html

trough. The range of data used to identify epidemic starting points was a 6-year cycle for 46 prefectures, and a 5-year cycle for 1 prefecture (Okinawa). An epidemic period was defined as the period between the onset and the next trough.

nationwide surveillance data in Tokyo and in Japan.

The validity of the model was confirmed by determining the proportion of the number of RSV reports falling below the onset line (horizontal line of RSV reporting number at onset) of the RSV season and within the bounds of the sine function.

### Determining the Onset of RSV Season

The main objective of this study was to determine the number of RSV reports at the onset of RSV season in Japan to enable timely access to preventive treatment. Ideally, the determination of period for the epidemic should include as many patients as possible. However, the higher number of RSV reports included, the longer the specified epidemic period becomes. From a health economics perspective, more information about a short epidemic period with more RSV reports is desirable. Accordingly, both the number of RSV reports included within epidemic periods relative to the total number of RSV reports (the capture rate) and the number of RSV reports per week within an epidemic period (RSV-reports/w) must be taken into account when determining epidemic starting points. The search index for both the capture rate and RSV-reports/w was then calculated to determine the onset of the epidemic cycle (see the formula below).

The search index was expressed by the following equation:

$$A(\mathbf{x}) = \frac{\Sigma\_{i=1}^{s} \Sigma\_{j=\text{onset}\_i}^{\text{Through}\_i} k\_j}{\Sigma\_{i=1}^{s} (Trough\_i - onset(\mathbf{x})\_i + 1)}$$

$$B(\mathbf{x}) = \frac{\Sigma\_{i=1}^{s} \Sigma\_{j=\text{onset}\_i}^{Trough\_i} k\_j}{\Sigma\_{j=Trough(\mathbf{x})}^{Trough(\mathbf{x})} k\_j}$$

$$INDEX(\mathbf{x}) = \frac{A(\mathbf{x})}{\max\left\{\mathbf{x} \, \middle| A(\mathbf{x})\right\}} + \frac{B(\mathbf{x})}{\max\{\mathbf{x} \, \middle| B(\mathbf{x})\}}$$

A(x) denotes RSV-reports/w (the number of RSV reported per week)

B(x) denotes capture rate (the number of RSV reports included within onset—trough periods relative to the total number of RSV reports in the dataset).

INDEX(x) denotes the search index. The capture rate (A(x)) and RSV- reports/w (B(x)) were converted into the same scale and summed.

The other variables were as follows:


<sup>∗</sup>The procedure to set the onset was as follows: The number of RSV reports for each prefecture was transposed into percentile rank, and the minimum percentile rank that was greater than the percentile value was selected. For the 101 percentile values, 101 percentile ranks were selected for each prefecture. The reporting week corresponding to the intersection of these percentile ranks and the increasing slope of the sinusoidal curve was taken as the onset.

Percentile rank of the highest index was set as the onset line, which provided a number of RSV reports/w and was used as a threshold horizontal line to detect the onset of the RSV season. The onset of each epidemic was identified from the intersection of the onset line of the epidemic and the increasing slope of the sinusoidal curve.

The validity of the determined onset was confirmed by calculating a capture rate.

The analyses were first performed on the data reported in Tokyo, the model was then validated using the data for 47 prefectures.

### RESULTS

# Determination of the Onset of RSV Season in Tokyo

The number of RSV reports/w at the onset of RSV season in Tokyo was 81 reports/week (**Figure 2**). When 81 reports/week was set as the onset line of the epidemic period in Tokyo, the onset of epidemic periods was detected through all the years from 2012 to 2017.

The onset of the epidemic period was the 32nd week (August 6–12) in 2012, the 31st week (July 29–August 4) in 2013, the 36th week (September 1–7) in 2014, the 35th week (August 24–30) in 2015, and the 31st week (August 1–7) in 2016. The onset was the 27th week (July 3–9) in 2017, which was earlier as compared to those previously observed (**Figure 2**, **Table 1**).

When converting the value to the number of cases reported per SMI with the view to generalize the findings to other prefectures, 81 reports/week was converted to 0.31 reports/week per SMI for 259 pediatric medical institutions at fixed locations in Tokyo.

### Application of the Approach to All 47 Prefectures in Japan

As in the case of Tokyo, the model identified the RSV epidemic cycles as periodic oscillations in other prefectures as well, despite the 11 climate zones of Japan (**Table 1**, **Supplemental Figures**). The minimum proportion of RSV reports falling below the onset line of RSV season and within the bounds of the sine function was 83.1% (mean 91.3%, median 91.4%, and max 97.1%). The

### TABLE 1 | The onset of RSV season in 47 prefectures.


(Continued)


RSV, respiratory syncytial virus.

A number at an onset line of RSV season was used as a threshold to determine an onset of RSV season.

mean number of RSV reports/w at onset of the RSV season for all prefectures was 29.7 reports/week (median 21.0, min 6.0, and max 121.0 reports/week), and the mean number of weekly reports/SMI to define the onset of the RSV season was 0.52 (median 0.53, min 0.22, and max 1.0, **Table 1**). In general, the onset of the epidemic RSV season in 2017 was more than 1 month earlier than in the previous year, as demonstrated by the data.

### DISCUSSION

Japan is under 11 significantly different climate zones due to its geographical location between the Sea of Japan and the Pacific Ocean, and shape, which spans from north to south. There are 47 prefectures in Japan; the epidemic of RSV infection begins from the south/western area and then tends to spread to the eastern area<sup>1</sup> . The epidemic curve based on RSV reports reaches a peak during summer in Okinawa, the southernmost prefecture and a subtropical region, unlike other prefectures in Japan (12). For this reason, it may be difficult to define the integrated conditions of the onset of the RSV season on a national basis in Japan. Therefore, the conditions of the onset of the RSV season were investigated separately for each prefecture in this study.

First, periodicities of the RSV season in each year were explored for each prefecture using a sinusoidal curve. The

TABLE 1 | Continued

fluctuation of the recent epidemic period was found to be periodic every year in all prefectures (**Table 1**). We then explored the onset of the RSV season, assuming it would come after the trough of the sinusoidal curve for each prefecture.

Previous studies have shown similar modeling approaches for the epidemics of RSV (13–16), in which non-linear regression analyses were applied. In the present study, in addition to the non-linear regression analyses, the model was developed taking account of the following factors and their balance for the epidemic period: (1) the number of RSV reports per week within epidemic periods (a high rate of RSV-reports/w), and (2) the majority of the reports included within the RSV season (a high capture rate).

Generally, if the epidemic period is narrowed down to the vicinity of the peak, although the number of RSV-reports/w will be high, the capture rate will be low, because many RSV infections have been reported before it. Thus, it may be too late to define the epidemic start based on this period definition. Conversely, when the epidemic period is set too early, the capture rate will be sufficiently high, but the number of RSVreports/w will be too low, and therefore it may be too early to define the onset of the RSV season. Consequently, we considered that a balance between RSV-reports/w and the capture rate of reports is important to detect the onset of the RSV epidemic. To accomplish this task, the RSV-reports/w and capture rate in the epidemic period were quantified to the same scale from 0.0 to 1.0, so that the sum of both could be used as an index. The timing of the maximum value was taken as the epidemic start.

It is worth mentioning that it is mandatory for about 10% of medical institutions (SMIs) to report the RSVpositive cases, and the data are disseminated by the IDWR in Japan, indicating that the weekly reports well represent the real epidemic RSV status in Japan, whereas other case data (hospitalization and RSV-positive rates) used in the previous studies for the modeling in the US were obtained in certain geographical regions (13, 15, 16). Such difference in the data coverage used for the modeling development, as well as that in the modeling methods, make a direct comparison difficult between our study and previous studies.

Criteria for the onset of RSV season for each prefecture were set in the present study, and the results showed a remarkable change in the onset of RSV season by years. The model detected that the onset was 1 or 2 months earlier in 2017 than in 2016. The 2017 RSV epidemic began in July in 34 (72%) prefectures and in June in 4 (8.5%) prefectures (**Table 1**). The National Institute of Infectious Disease and Tokyo Metropolitan Government announced that the RSV epidemic in 2017 started in July, indicating that our results support the findings of the NIID and the Tokyo Metropolitan Government.

The model also revealed that the changes were small in the southern prefectures. Furthermore, Miyazaki and Kagoshima prefectures, which are located between the temperate zone and the subtropical zone in the south, showed two peaks where RSV prevails, in winter and summer. In Okinawa prefecture, which is located in the subtropical zone at the south end, the epidemic period was in summer. Because of the prefecture-specific tendency in the patterns of RSV cycles, the application of our model for each prefecture will provide more precise and timely information regarding the pattern of RSV seasons.

A limitation of the study is that the model does not predict the onset of the RSV season in advance, and therefore it may not provide timely information regarding its start. However, the model enables us to detect the onset at an early stage so that we can prepare, e.g., to administer palivizumab to infants to prevent the severe RSV infection. Another limitation is that the model may not identify potential onset timing if the number of reported cases fluctuates substantially or if high numbers of cases are observed throughout a year.

The result would be used to prepare for the RSV season by estimating the onset of the next RSV season from the past onsets.

Converting to the reported number per SMI, the differences in the pediatric population by prefecture are corrected, so it would be able to refer to the situation in the neighboring prefecture. For example, many people in the Ibaraki prefecture are commuting to Tokyo.

The result would contribute timely detection of the onset of RSV season even if the season starts earlier than the presumption. It would be expected to help reduce the severe RSV infection in infants by signaling the need for the prompt start of treatment with palivizumab and hence contribute to pediatric and neonatal medical practice.

# CONCLUSION

In this study, our model detected the onset week of RSV season in 47 prefectures based on the national surveillance data in Japan for 6 years (2012–2017). The model also verified that the onset of RSV season was in early July in 2017. A timely detection of the onset of RSV season is expected to help reduce severe RSV infection in infants by signaling the need for the prompt start of treatment with palivizumab, and thereby contribute to pediatric and neonatal medical practice.

# DATA AVAILABILITY

All datasets generated for this study are included in the manuscript and/or the supplementary files.

# AUTHOR CONTRIBUTIONS

All authors prepared the manuscripts and approved the manuscript for publication. HY and TH designed the study. TH conducted data analysis. HK, IK, and SK provided efficient advice, from the data analysis planning to obtainment of results.

# FUNDING

The design, study conduct, and financial support for the study were provided by AbbVie. AbbVie participated in the interpretation of data, review, and approval of the publication.

### ACKNOWLEDGMENTS

Mayumi Saotome, Ryo Yano, and Ayumi Kodama of DOT WORLD Co., Ltd., Tokyo, Japan, provided medical writing and editing services in the development of this manuscript. AbbVie provided funding to DOT WORLD Co., Ltd. for this work.

### REFERENCES


# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpubh. 2019.00039/full#supplementary-material

Data Sheet 1 | Detection of the start of the epidemic period of respiratory syncytial virus (RSV) infection using the Infectious Diseases Weekly Report surveillance data (2012-2017 seasons) of 46 prefectures. Black line, number of RSV report; light blue line, RSV epidemic cycle; dotted red line, onset line of RSV season; vertical red line, onset week; vertical blue line, trough of epidemic cycle.


**Conflict of Interest Statement:** AbbVie GK planned the study design and conducted the study. HK, IK, and SK are advisors to AbbVie GK. HY and TH are employees of AbbVie GK.

Copyright © 2019 Yamagami, Kimura, Hashimoto, Kusakawa and Kusuda. 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.

# Alveolar Macrophages Treated With Bacillus subtilis Spore Protect Mice Infected With Respiratory Syncytial Virus A2

Ji Eun Hong<sup>1</sup>† , Yoon-Chul Kye<sup>1</sup>† , Sung-Moo Park1,2, In Su Cheon1,3, Hyuk Chu<sup>4</sup> , Byung-Chul Park<sup>5</sup> , Yeong-Min Park<sup>6</sup> , Jun Chang<sup>7</sup> , Jae-Ho Cho<sup>8</sup> , Man Ki Song<sup>3</sup> , Seung Hyun Han<sup>9</sup> and Cheol-Heui Yun1,2,5 \*

<sup>1</sup> Department of Agricultural Biotechnology, Research Institute of Agriculture and Life Sciences, Seoul National University, Seoul, South Korea, <sup>2</sup> Center for Food and Bioconvergence, Seoul National University, Seoul, South Korea, <sup>3</sup> Laboratory Science Division, Department of Molecular Vaccinology, International Vaccine Institute, Seoul, South Korea, <sup>4</sup> Division of Zoonoses, Center for Immunology and Pathology, Korea Center for Disease Control and Prevention, National Institute of Health, Cheongju, South Korea, <sup>5</sup> Institute of Green Bio Science and Technology, Seoul National University, Seoul, South Korea, <sup>6</sup> Department of Immunology, School of Medicine, Konkuk University, Chungju, South Korea, <sup>7</sup> Graduate School of Pharmaceutical Sciences, Ewha Womans University, Seoul, South Korea, <sup>8</sup> Academy of Immunology and Microbiology, Institute for Basic Science, Pohang, South Korea, <sup>9</sup> Department of Oral Microbiology and Immunology, DRI, and BK21 Plus Program, School of Dentistry, Seoul National University, Seoul, South Korea

Respiratory syncytial virus (RSV) is a major pathogen that infects lower respiratory tract and causes a common respiratory disease. Despite serious pathological consequences with this virus, effective treatments for controlling RSV infection remain unsolved, along with poor innate immune responses induced at the initial stage of RSV infection. Such a poor innate defense mechanism against RSV leads us to study the role of alveolar macrophage (AM) that is one of the primary innate immune cell types in the respiratory tract and may contribute to protective responses against RSV infection. As an effective strategy for enhancing anti-viral function of AM, this study suggests the intranasal administration of Bacillus subtilis spore which induces expansion of AM in the lung with activation and enhanced production of inflammatory cytokines along with several genes associated with M1 macrophage differentiation. Such effect by spore on AM was largely dependent on TLR-MyD88 signaling and, most importantly, resulted in a profound reduction of viral titers and pathological lung injury upon RSV infection. Taken together, our results suggest a protective role of AM in RSV infection and its functional modulation by B. subtilis spore, which may be a useful and potential therapeutic approach against RSV.

Keywords: respiratory syncytial virus, Bacillus subtilis, spore, alveolar macrophage activation, MyD88

# INTRODUCTION

Respiratory syncytial virus (RSV) causes serious disease symptoms of bronchiolitis and acute lower respiratory tract infection resulting in up to 200,000 annual deaths worldwide, especially for the infants and elderly (Nair et al., 2010). Recent study has also suggested that children younger than 6 months old suffering from RSV amounted to 1.4 million hospital admissions and

### Edited by:

Ville Peltola, Turku University Hospital, Finland

### Reviewed by:

Diana Boraschi, Istituto di Biochimica delle Proteine (IBP), Italy Vahid Salimi, Tehran University of Medical Sciences, Iran

### \*Correspondence:

Cheol-Heui Yun cyun@snu.ac.kr; cyun@snu.kr †These authors have contributed equally to this work

### Specialty section:

This article was submitted to Virology, a section of the journal Frontiers in Microbiology

Received: 16 September 2018 Accepted: 20 February 2019 Published: 12 March 2019

### Citation:

Hong JE, Kye Y-C, Park S-M, Cheon IS, Chu H, Park B-C, Park Y-M, Chang J, Cho J-H, Song MK, Han SH and Yun C-H (2019) Alveolar Macrophages Treated With Bacillus subtilis Spore Protect Mice Infected With Respiratory Syncytial Virus A2. Front. Microbiol. 10:447. doi: 10.3389/fmicb.2019.00447

**27**

27,300 for in-hospital death (Shi et al., 2017). Despite the numerous trials, effective preventive strategies against RSV infection have been limited and turned out to be unsuccessful. This is because of the unique ability of RSV to disarm host of both innate and adaptive immunity (Meng et al., 2014) causing overall failure of host immune defense mechanisms, which is mainly due to non-structural 1 and 2 proteins. Indeed, a poor innate response has previously been well documented particularly at the initial stage of RSV infection (Marr et al., 2013) and several attempts have been made to improve this stage in RSV infection. In this regard, it has been suggested that successful protection against RSV can be mediated by activation of innate immune cells via tolllike receptor (TLR) 2 and 4 stimulation (Kurt-Jones et al., 2000; Murawski et al., 2009), but not by caspase-1 pathway (Shim and Lee, 2015).

One of the primary innate immune cell types in the lung is alveolar macrophages (AM), which play a pivotal role to maintain homeostasis and to induce effective defense mechanism. AMs have unique properties compared to other macrophages in different tissues because they have direct contact with external environment, which could allow rapid and direct recognition of antigens derived from invading pathogens or allergens followed by the initiation of immune responses (Hashimoto et al., 2011; Herold et al., 2011). Recent studies have shown that failure of early AM-mediated defense response caused insufficient protection against various viral pathogens infecting respiratory tracts and led to diminished recruitment of immune cells followed by disrupted lung homeostasis (Maus et al., 2002; Snelgrove et al., 2008). Despite the importance of AMs in various respiratory viral diseases, the precise role and protective mechanisms of AM in RSV infection are yet to be investigated.

Previous attempts using probiotics have gained an insight for the priming innate immune system where the administration of such probiotics induced a rapid activation of innate immunity and consequently enhanced protective efficacy against respiratory viral infection (Garcia-Crespo et al., 2013; Park et al., 2013; Lee et al., 2016). Furthermore, spore derived from probiotics, as we used in this study, is known to be one of the alternatives not only for inducing enhanced innate immune responses but also for safety and stability (Hong et al., 2008). Bacillus subtilis has been known to change its status into spore form under the harsh circumstances such as starvation. After sporulation, it can survive with resistance to heat, cold, or other enzymatic assaults (Tan and Ramamurthi, 2014). Because of their stability, several researchers have tried to use spores as probiotics or adjuvants to enhance the health and to protect against infections caused by microbes such as influenza virus (Song et al., 2012) or Clostridium difficile (Colenutt and Cutting, 2014) in mice and human trials (Lefevre et al., 2015, 2017).

In this study, we tried to use B. subtilis-derived spore to overcome RSV infection and investigated whether and how B. subtilis-derived spore would influence on the AMs to induce effective innate immunity and subsequently to protect host from RSV infection.

# RESULTS

# Intranasal Delivery of B. subtilis Spore Enhanced Anti-viral Immunity to RSV Infection

It has been reported that in vivo administration of prebiotics or probiotics such as Lactobacillus can facilitate protective immunity against influenza virus infection (Kiso et al., 2013; Lee et al., 2013). These previous reports prompted us to test the potential of bacterial particular substance, B. subtilis-derived spore to protect against RSV infection. To address this, a group of mice was administered with or without spore intranasally (i.n.) prior to RSV infection and then infected them with RSV. The effect of spore was clear in body weight loss; untreated control mice lost 10–20% total body weight with a peak reduction of day 2 post-infection (p.i.), which is well-known symptom of RSV infection in mouse model, whereas mice treated with spore did not lose their body weight (**Figure 1A**). It should be noted, however, that the weight loss after RSV infection appeared to be transient, which is a limitation of RSV mouse infection model, so the control mice began to gradually gain weight back to near normal level by day 4 post-infection. To support the body weight data, there was about 4–5-fold decrease in viral loads in the lung of the spore-treated mice compared to those of control mice (**Figure 1B**). These results suggest that delivery of B. subtilis spore not only prevent the pathology, but also help clearance of virus against RSV infection.

# I.n. Administration of B. subtilis Spore Increases the Number of Alveolar Macrophages

To identify the cellular mechanism by which B. subtilis spore influences on anti-viral immunity, we analyzed various innate immune cells in the lung with or without spore treatment. We found that there was the most notable increase in AM both in the lung and bronchoalveolar lavage (BAL) along with inflammatory monocytes or neutrophils (**Figures 1C,D**), implicating AMs might play an important role in mediating innate immunity to RSV. In line with increased numbers of AM, spore-treated mice showed increased gene expression of GM-CSF and well-known M1 macrophage-related cytokines (Murray et al., 2014) including TNF-α, IFN-γ, IL-12p40, and IL-6 (**Supplementary Figure S1**). The cytokine producing ability of spore was also apparent in protein level of effector cytokines such as IFN-β and IL-12p40 (**Figure 1E**), which are also categorized in M1 macrophagerelated cytokines. These results suggest that administration of B. subtilis spore induced significantly enhanced number of AMs, and augmented expression of effector cytokines which are related to M1 macrophage differentiation.

# Depletion of Alveolar Macrophages Aggravates Disease Severity in Mice Infected With RSV

In viral infection condition, it is known that AM have a critical role during the early phase of infection (Schneider et al., 2014a)

FIGURE 1 | Pre-treatment with spore through intranasal route reduces the disease severity following RSV infection and induces the population change of alveolar macrophages (AM) and enhances antiviral effector molecules. Mice were administered with 1 × 10<sup>9</sup> CFU of spore via i.n. route at 5 days prior to RSV infection with 2 × 10<sup>6</sup> PFU per mouse (n = 3). (A) Body weight was monitored daily after the infection and (B) viral load in the lungs was analyzed by plaque assay at 4 days post-infection (DPI). Change of various innate immune cells in the (C) post-lavaged lung and (D) BAL fluid was analyzed by flow cytometry at 0 to 4 DPI. Empty and filled circles indicate PBS and spore pre-treated mice, respectively. (E) IFN-β and IL-12p40 in BAL fluid were measured by ELISA. Data are expressed as mean ± S.E.M. for the group. Significant differences from results with the PBS control are <sup>∗</sup>P < 0.05; ∗∗P < 0.01; and ∗∗∗P < 0.001, respectively.

because of the alarming failure and lack of scavenger activity against infected cells (Kumagai et al., 2007; Pribul et al., 2008). Moreover, AM-depleted mice are unable to deliver the antigen to dendritic cells for the antigen presentation (Ugonna et al., 2014) in order to strengthen the following adaptive immune response. To examine the particular role of AM in RSV infection, we conducted the depletion of AM by injection of clodronateencapsulated liposome (Clod) through the intratracheal (i.t.) route. As a result, absolute number of AM in the lungs and BAL from Clod-treated mice declined by approximately 85% compared to that in control (**Supplementary Figure S2**) without significant histological changes (**Figure 2C**). At day 1 postinfection of RSV, AM-depleted mice showed a sharp decline in body weight (**Figure 2A**) and at day 4, they have significantly higher viral loads in lung than control (**Figure 2B**). Pathological results also showed the severity of infection with displayed thickened alveolar epithelium, destruction of epithelial walls, overall alveolar swelling, and the accumulation of immune cells in the lung (**Figure 2C**). Furthermore, the accumulation of inflammatory cells in the interstitial space, bronchi, and vessel

was apparent in lungs from AM-depleted mice (**Figures 2D–G**). Taken together, depletion of AM at the initial phase of RSV infection leads to exacerbated disease severity coincident with increased pulmonary inflammation.

### Spore Directly Enhances the Antiviral Function of Alveolar Macrophages

To explore direct role of AM in cellular level and the effect of B. subtilis spore, AM cell line, MH-S, was treated with spore for 24 h and then infected with RSV. The number of plaques of RSV was significantly reduced when the cells were incubated with spore in a dose-dependent manner (**Figure 3A**). Inflammatory cytokines, IL-12p40 and IL-6, in the supernatant from the cells treated with spore followed by RSV infection were measured as an initial assessment for their antiviral activity (Ugonna et al., 2014). The results showed that the spore treatment led to substantial increase of IL-6 and, to a lesser extent, IL-12p40 (**Figure 3B**). These results demonstrated that B. subtilis spore directly promotes the anti-viral activity of AM, especially at the early time point after the infection, in a dosedependent manner.

FIGURE 3 | Alveolar macrophages pre-treated with spore extend their antiviral effects. MH-S cells were stimulated with spore for 24 h, with a ratio of MH-S:Spore at = 1:0, 1:1, 1:10, and 1:50. Then, the cells were washed and infected with 1 MOI RSV for 12 h. (A) The viral load was analyzed by using plaque assay and (B) the production of IL-12p40 and IL-6 was measured by ELISA. Data are expressed as mean ± S.E.M. of three replicates. <sup>∗</sup> and ∗∗ indicate significant differences at P < 0.05 and P < 0.01, respectively.

# Alveolar Macrophages Activated by Spore Have a Pivotal Protective Role in Mice Infected With RSV

To validate direct involvement of spore-activated AM for the protection from RSV infection in vivo, we depleted the AMs in spore pre-administered mice before RSV infection. As mentioned above, mice treated with spore did not have any significant weight loss (**Figure 4A**), in line with the ability of spore to clear the virus (**Figure 4B**). However, when we depleted AM by clodronate in spore-treated mice, they showed a moderate decrease of body weight (**Figure 4A**) and led to viral counts be increased in the lung (**Figure 4B**), suggesting that AM play an important role in the protection against RSV infection by B. subtilis spore treatment. The spore treatment can efficiently protect the epithelium and prevent infiltration of immune cells in alveolus (**Figure 4C**) and bronchi (**Supplementary Figure S3**), which is seen in the pathological score (**Figures 4D–G**) in support of the effect of AM. These data suggested that spore treatment prevents RSV infection efficiently, the effect is partially dependent on the AM. Taken together, these results support that the B. subtilis spore enhances host protection against RSV infection with support of AM.

# The Effects of Spore on Protection Against RSV Infection Are Dependent on MyD88 Signaling

The next question was which signaling pathways are involved in the capacity of spore to modulate anti-viral functions of AM. Since previous reports suggest TLR2 or 4 is in charge of viral infection as mentioned above, we presumed that spore might trigger MyD88 signaling upon direct engagement with TLRs on AM. In support of this idea, we used MyD88−/<sup>−</sup> mice for RSV infection model with B. subtilis spore administration. MyD88−/<sup>−</sup> mice were unable to induce protective response to RSV infection with reduced body weight (**Figure 5A**) and significantly high viral load (**Figure 5B**). More importantly, the administration of spore did not work on the recovery of MyD88−/<sup>−</sup> mice at all. Consistent with these data, infiltration of AM in the lung (**Figure 5C**) and BAL (**Figure 5D**) was far less than wild-type mice and the inflammation with severe destruction of epithelium was also shown in alveolus (**Figures 5E–I**) and bronchi (**Supplementary Figure S3**) of MyD88−/<sup>−</sup> mice even with B. subtilis spore treatment.

To confirm the direct engagement of MyD88 signaling upon B. subtilis treatment in cellular level, we used bone marrowderived macrophages (BMMs) from wild type or MyD88−/<sup>−</sup> mouse and treated spore in vitro with RSV infection and conduct plaque assay with culture supernatant. In the same context with in vivo data, significantly higher viral loads were observed in MyD88-deficient BMMs compared to wildtype BMMs (**Figure 6A**). As expected, the capability of spore to reduce viral load was prominent in wild-type BMMs in a dose-dependent manner; unexpectedly, however, the effect of spore was, albeit less, also apparent even in MyD88 deficient BMMs. Consistent with data from viral loads, IL-12p40 production was markedly augmented by spore treatment and substantially higher in wild-type BMMs than MyD88-deficient BMMs (**Figure 6B**).

Taken together, these results suggest that B. subtilis spore can enhance the protective capacity against RSV infection affecting directly on the AMs, and these activation and M1-related effector cytokine production is dependent on MyD88 signal.

# DISCUSSION

Importance of AM for host defense mechanism during early viral infection in respiratory tract is well established, however, the precise role of AM, especially in mice infected with RSV is less clear. In the present study, we focused on the specific role of AM treated with B. subtilis spore in RSV infected mice for their protection and brought the following four notable findings: (1) AM have a pivotal role for protection during the initial stage of RSV infection; (2) Administration of B. subtilis spore induces the activation of AM with up-regulation of GM-CSF and M1 macrophage-related cytokines; (3) The i.n. delivery of B. subtilis spore induces the increase of antiviral effector molecules against RSV infection; and (4) MyD88 signals in the AM are critical for the protection against RSV.

5 days prior to RSV infection and injected with control or clodronate-encapsulated liposome through intratracheal on days 3 and 1 before the infection and sacrifice on 4 DPI. (A) Body weight was monitored daily after the infection and (B) viral load in the lung was examined at 4 DPI, respectively (n = 3). (C) At 4 DPI, perfused lungs were stained with H&E for histological examination by microscopy at 200 × magnifications and (D–G) scored for histopathology. Arrows indicated are as follows; orange, epithelium thickness and destruction; green, pulmonary edema; red, inflammatory cells; and black, cell death. Significant differences from results with the PBS control are <sup>∗</sup>P < 0.05; ∗∗P < 0.01; and ∗∗∗P < 0.001, respectively.

Induction of innate immune response in the early phase of viral infection is necessary for host immunity (McGill et al., 2009; Copenhaver et al., 2014), and thus over the years various attempts have been tried to improve the innate immunity in order to achieve the better protection against viral infection. Delivery of probiotics and their related materials is one of the strategies (Kiso et al., 2013; Yaqoob, 2014) and in the present study we sought a potential ability of spore to enhance innate immunity against RSV in line with other studies (Hong et al., 2008; McKenney et al., 2013).

It been reported that AM play an important function in diverse respiratory infections such as a primary recognition of antigens and scavenger to infected cells (Tate et al., 2010) and that the absence of AM-mediated defense mechanism led to the failure of protective immune responses, the exacerbation of pathology, and the accumulation of debris following pulmonary infection

magnification and (F–I) scored for histopathology at 4 DPI. Arrows indicated are as follow; orange, epithelium thickness and destruction; green, pulmonary edema; red, inflammatory cells; and black, cell death. Significant differences from results with the PBS control are <sup>∗</sup>P < 0.05; ∗∗P < 0.01; and ∗∗∗P < 0.001, respectively.

(Kolli et al., 2014) but the precise role and protective mechanism of AM in RSV infected mice are yet to be illuminated. Direct evidence showing important role of AM in the present study came from in vivo experiment with depletion of AM. The results showed that the AM-depleted mice suffered from high viral load and intense pathological episode as shown by the accumulation of dead cells and debris in bronchi and blood vessels, indicative of early pulmonary inflammation. All these findings clearly support the critical role of AM in controlling disease severity of mice infected with RSV. AM treated with B. subtilis spore rendered

less susceptible to RSV infection and more importantly, anti-viral function of spore was markedly reduced upon depletion of AM, implying AM as a major target for B. subtilis spore as well as a key player for the protection.

How does then the spore influence on anti-viral function, differentiation, and/or migration of AM that would potentially benefit protective immunity to RSV? Our results showed that mice pre-treated spore through i.n. route rapidly up-regulated GM-CSF and M1 macrophage-related cytokines such as TNF-α, IFN-γ, IL-6, and IL-12p40. GM-CSF is a well-known essential factor for the differentiation (Guilliams et al., 2013), survival (Shibata et al., 2001), replenishment (Steinwede et al., 2011), and the ability of host defense (Ghoneim et al., 2013) of macrophages. Also, IFN-γ could lead monocytes differentiate into classically activated macrophages (M1-macrophages). This polarization is important in limiting tissue damage (Barros et al., 2013) and induces the increase of cytokines that enhance the ability of AM for the killing of intracellular pathogens (Martinez and Gordon, 2014). Thus, as a working mechanism for B. subtilis spore, it is conceivable that pre-treatment with B. subtilis spore might enhance the viability of macrophages, which promotes their antiviral functions. Alternatively, B. subtilis spore may facilitate AM proliferation and/or differentiation, skewing M1 vs. M2 balance toward favoring of M1 macrophage, which leads to an environment in which immune-stimulatory cytokines dominate for host protection against RSV. Supporting this idea, pre-treatment with spore indeed failed to induce TGFβ and PPAR-γ, which are known to induce differentiation of immune-suppressive M2 macrophages (Ginhoux, 2014; Schneider et al., 2014b). However, it is not always better to have M1 responses than M2 responses depending on how strong the immune response is and when the response was induced. It is important to induce appropriate and balanced immune responses because it has been reported that overwhelming pro-inflammatory responses with iNOS exacerbate the lung damage (Mgbemena et al., 2013). In addition, Th2 cytokines such as IL-4 and IL-13 induced after RSV infection could alleviate the damage of tissue (Shirey et al., 2010). Whether the preferences of M1/M2-type cytokines by spore indeed reflect AM proliferation/differentiation or selective enhancement of migration will need further investigation.

Besides the preference of M1 cytokines, mere increase of AM numbers we found after spore treatment would also have an advantage for protection against RSV infection, given that AM is known to play a role as reservoirs for inhaled antigens (Copenhaver et al., 2014) and thus the rapid increase of AM may effectively boost not only innate immunity but also prime adaptive immunity. Consistent with these, mice pre-treated with B. subtilis spore before RSV infection displayed a certain protective immunity as shown by significantly low level of viral titer and maintenance of body weight. Furthermore, B. subtilis spore treatment induced inflammatory monocytes with Ly6C high phenotype, known to regulate type I IFN signaling during acute viral pneumonia in mice (Seo et al., 2011) with an ability to differentiate into macrophages (Geissmann et al., 2010). Indeed, mice pre-treated with B. subtilis spore showed remarkably high IFN-β. Recent study showed that AM are the major source of type I IFNs during RSV infection and an underappreciated facet of type I IFNs-dependent resistance leads to a cell-extrinsic responses through rapid recruitment of antiviral inflammatory monocytes to the site of infection (Goritzka et al., 2015).

It would be important to postulate what are the mechanisms behind the modulatory ability of spore for AM. The most plausible explanation is that spore itself and/or its degraded components acts as a TLR ligand. TLR2 and TLR4 signaling pathways, in which MyD88 serves as a downstream molecule, are essential for the protection in RSV infection (Shirey et al., 2010). Consistent with those reports, in the present study, MyD88−/<sup>−</sup>

mice showed persistence of high viral counts in parallel with impaired body weight gain after the infection suggesting that MyD88 is indispensable for the protection against RSV infection. Furthermore, it was evident that MyD88−/<sup>−</sup> mice failed to recruit AM into the lung region, irrespective of B. subtilis spore administration, suggesting that B. subtilis spore-induced increase of AM in the lung and BAL was dependent on MyD88.

MyD88−/<sup>−</sup> mice also displayed severe pathological signs in the lung compared to wild type mice. These results imply that MyD88-dependent signaling might be the key mechanism by which AM treated with B. subtilis spore contribute to the protection of mice against RSV infection. This requirement of MyD88 signaling may account for the ability of spore to promote AM capable of producing GM-CSF and M1-related cytokines, which enhance their viability and M1 signature cytokines including IFN-γ and TNF-α, as well as producing IFNβ and IL-12p40, which may serve as an alarm signal and facilitate IFN-γ production. Whether all these MyD88-dependent effects by spore are indeed mediated through direct binding on TLR and whether there are MyD88-independent TLR signaling such as NOD2 being involved will need to be addressed. Collectively, our results provided a key modulatory role of B. subtilis spore in the anti-viral function of AM and have useful implication for effective treatment against RSV infection.

Even after depletion of AM, to a lesser extent, spore treatment can reduce the viral load of RSV. There are other possibilities than macrophages which may contribute to the protective function of spore against RSV infection. For instance, components derived from commensal microbes have been reported to induce antimicrobial peptides from epithelial cells that act as a first line defense against viral infection (Gomez and Prince, 2008). Furthermore, TLR2 signal can up-regulate the tight junction proteins in epithelial cells (Ragupathy et al., 2014) which can enhance the tolerance to viral infection.

Several studies have been done with B. subtilis spore to protect against infections in mice and human (Colenutt and Cutting, 2014; Lefevre et al., 2015, 2017). Even though the spores have been reported to be safe in human trials, safety should be more elucidated further, for example, on administration of spores in the respiratory tracts or as subcutaneous injection. In addition, the status of spore as a live or heat-killed form should be studied more in terms of safety. Collectively, we proved a key function of B. subtilis spore in antiviral activity and protection of lung environment against RSV-induced damage in mice infected with RSV via MyD88.

# MATERIALS AND METHODS

### Mice and Ethics Statement

Female BALB/c mice, 6–8 week-old, were purchased from Orient Bio Inc., Korea. MyD88−/<sup>−</sup> mice were purchased from Jackson laboratory (Bar harbor, ME, United States). Animals were maintained and procedures were performed with approval of the IACUCs of Seoul National University (Approval No.: SNU-130527-5) and International Vaccine Institute (Approval No.: 2012-022) in accordance to Laboratory Animal Act of Korean Ministry of Food and Drug Safety for enhancing the ethics and reliability on animal testing through appropriate administration of laboratory animals and animal testing.

# Preparation and Isolation of B. subtilis Spore

Bacillus subtilis was spread in an agar plate containing 3% Trypticase soy broth (TSB), 0.5% Yeast extract (YE), and 1.5% Bacto Agar (all from BD Biosciences, San Diego, CA, United States) and incubated at 37◦C for 9 h. One colony was picked and inoculated in 25 ml of 3% TSB and 0.5% YE liquid media. Then, it was incubated for 5 h in the shaking incubator at 150 rpm at 37◦C until the OD value reached between 0.45 and 0.6. For sporulation, culture was transferred to 500 ml of the autoclaved media containing 5 ml of 10% KCl, 5 ml of 1.2% MgSO4·7H2O (pH 7.6), 0.5 ml of 1 M Ca(NO2)3, 0.01 M MnCl2, and 1 mM FeSO4. The culture was incubated at 37◦C for 48 h with shaking at 150 rpm. The cells were collected by centrifugation at 5516 × g for 10 min, re-suspended in distilled water, and incubated at 4◦C for 48 h on the rocker. Then, the cells were sonicated at 35% amplitude (1 watt) for 90 s with 0.5 s pulse. Spore loaded on the layers of 35, 25, and 15% OptiPrep Density gradient (Sigma-Aldrich, Wt. Louis, MO, United States) was centrifuged at 10,000 × g for 40 min at 25◦C without break for the purification. The spore was washed 3 times with distilled water and re-suspended in 1 ml of distilled water.

### Preparation and Isolation of Respiratory Syncytial Virus A2

Respiratory syncytial virus A2 strain was amplified as follows: HEp-2 cells (ATCC, Manassas, VA, United States) were grown in MEM containing 10% of FBS and 1% of antibiotics (Jang et al., 2011). When the cells reached at approximately 80% confluence, the cells were washed and inoculated with 0.01–0.05 MOI of virus in MEM containing 1% of antibiotics and 25 mM HEPES (Gibco, Grand Island, NY, United States). The cells were incubated for 2 h at 37◦C, added MEM containing 6% of FBS only, and then incubated for additional 72–96 h. The cells were scraped and combined into the conical tubes on ice and centrifuged at 1,400 rpm for 4 min at 4◦C. The cell pellets were collected into conical tubes, re-suspended with cold conditioned media with 60% sorbitol, sonicated for 15 min in slurry ice and centrifuged at 4,000–5,000 rpm for 10 min at 4◦C, and then the sonication and centrifugation step was repeated under the same condition. The supernatants were transferred into new tubes, centrifuged at 23,000 rpm for 1 h at 4◦C, and then discarded. The resulting whitish virus pellet was re-suspended with 500 µl of cold MEM and the titer was determined by standard RSV plaque assay.

### Isolation and Culture of Bone Marrow-Derived Macrophages and Macrophage Cell Line

Bone marrow was flushed from femurs and tibias of the mice using Dulbecco's Modified Eagle Medium (DMEM, Gibco) supplemented with 2% FBS (Gibco). Red blood cells were lysed with ACK lysing buffer (Gibco) and whitish marrow cells were

seeded in 90 × 15 mm Petri dish with complete media containing 10% FBS, 1% antibiotics with 20% L-929 conditioned media (Zhang et al., 2008) and cultured at 37◦C in humidified incubator with 5% CO<sup>2</sup> for 7 days. On day 3, another 5 ml of fresh complete media was added to each dish. On day 7, only adherent cells were collected using non-enzymatic cell dissociation solution (Gibco). MH-S cells (ATCC, Manassas, VA, United States), mouse AM cell line, were grown in RPMI-1640 GlutaMax medium containing 10% FBS, 1.5% antibiotics (all from Gibco) in a humidified incubator with 5% CO<sup>2</sup> at 37◦C.

### RNA Isolation and Quantitative Real Time-PCR

For quantitative analysis for the expression of GM-CSF, TNF-α, IFN-γ, IL-12p40, IL-6, PPAR-γ, and TGF-β1 at mRNA level, quantitative real time polymerase chain reaction (qRT-PCR) was conducted. RNA was extracted from perfused lungs using TRIZOL (Invitrogen, Carlsbad, CA, United States). Total RNA was isolated by adding chloroform followed by centrifugation at 4◦C, 12,000 × g for 15 min and addition of isopropanol for 10 min at room temperature for RNA precipitation. RNA pellet obtained by washing with 75% ethanol was air dried for 10–15 min. Then it was re-suspended with DEPC distilled water (Sigma-Aldrich, Wt. Louis, MO, United States) and quantified with NanoDrop (Amersham Bioscience, United States) at A260. One microgram of RNA was reverse transcribed into cDNA and amplified with murine primers specific for GM-CSF (forward primer: 5<sup>0</sup> -CTGCCTTAAAGGGACCAAGAGA-3 0 , reverse primer: 5<sup>0</sup> -TTCCGCTGTCCAAGCTGAGT-3<sup>0</sup> ), TNF-α (forward primer: 5<sup>0</sup> -GCCAACGGCATGGATCTC-3<sup>0</sup> , reverse primer: 5-GTGGGTGAGGAGCACGTAGTC-3<sup>0</sup> ), IFN-γ (forward primer: 5<sup>0</sup> -GCCATCGGCTGACCTAGAGA-3<sup>0</sup> , reverse primer: 5<sup>0</sup> -GCAGTGTGTAGCGTTCATTGTCT-3<sup>0</sup> ), IL-12p40 (forward primer: 5<sup>0</sup> -GAAAGGTGCGTTCCTCGTAGA-3<sup>0</sup> , reverse primer: 5<sup>0</sup> -GGAACACATGCCCACTTGCT-3<sup>0</sup> ), IL-6 (forward primer: 5<sup>0</sup> -CACAGAGGATACCACTCCCAACA-3<sup>0</sup> , reverse primer: 5<sup>0</sup> -TCAGAATTGCCATTGCACAAC-3<sup>0</sup> ), PPARγ (forward primer: 5<sup>0</sup> -CAGGAGCCTGTGAGACCAACA-3<sup>0</sup> , reverse primer: 5<sup>0</sup> -ATCAGTGGTTCACCGCTTCTTT-3<sup>0</sup> ), TGFβ1 (forward primer: 5<sup>0</sup> -TCGTCTGCATTGCACTTATGC-3<sup>0</sup> , reverse primer: 5<sup>0</sup> -GTGGTGCCCTCTGAAATGAAA-3<sup>0</sup> ), and GAPDH (forward primer: 5<sup>0</sup> -CTCCACTCACGGCAAATTCA-3 0 , reverse primer: 5<sup>0</sup> -GCCTCACCCCATTTGATGTT-3<sup>0</sup> ). Real-time PCR was performed using Power SYBR Green PCR master mix (Applied Biosystem, Waltham, MA, United States) and analysis of the data was performed by One-step RT PCR (Applied Biosystem). Target gene expression was normalized to GAPDH expression.

### Measurement of Cytokine Production

Broncho-alveolar lavage (BAL) samples were collected via tracheotomy using 600 µl of PBS and the cells were separated from the BAL fluid by centrifugation at 1400 rpm for 5 min at 4 ◦C. Forin vitro experiments, supernatants from BMMs or MH-S after virus infection were carefully collected. Production of IFNβ (BioLegend, San Diego, CA, United States), IL-12p40, GM-CSF, and TNF-α (R&D System, Minneapolis, MN, United States) was examined using ELISA kit.

### Phenotypic Characterization of the Cells

To examine the absolute number of AM and other innate immune cells, BAL cells were collected as above and perfused lungs were isolated and minced through 70 µm cell strainer using MEM. The cells were stained with FITC-conjugated anti-CD11c (HL3), PE-conjugated anti-Siglec-F (E50-2440), PerCP-conjugated anti-Ly6C (AL-21), PE-Cy7-conjugated anti-Ly6G (1A8), APC-conjugated anti-CD11b (M1-70) or APCconjugated anti-F4/80 (BM8), and APC-Cy7-conjugated anti-CD45 (30-F11) (all from BD Biosciences except anti-F4/80 from BioLegend). The cells were acquired using FACS LSR II and flow cytometric data were analyzed by using FlowJo software (Tree Star, San Carlos, CA, United States). We defined CD45+Ly6C−Ly6G−CD11c+Sigleg-F+F4/80<sup>+</sup> as AMs, CD45+CD11b+Ly6C+Ly6G<sup>−</sup> for inflammatory monocytes, and CD45+CD11b+Ly6G+Ly6C<sup>−</sup> for neutrophils.

# Selective Depletion of Alveolar Macrophages

To selectively deplete AM, 350 mg per mouse of clodronateencapsulated liposome (FormuMax Scientific Inc., CA, United States) was given i.t. in a volume of 50 µl at 1 and 3 days before the RSV challenge with or without spore treatment. To verify the depletion of AM, naïve mice were given control liposome via i.t.

# Virus Titration in the Lung

To determine the viral titers, lungs from RSV-infected mice were isolated at day 4 post-infection. The lungs were minced through 70-µm cell strainer using cold MEM. Cell lysates were collected and RSV titers were determined by plaque assay using HEp-2 cells. The virus titers in the whole lung were normalized to weight of the lung tissue and indicated as PFU/g.

# Lung Histology and Pathology Scoring

For histology studies, mice were administered with spore and/or clodronate encapsulated liposome prior to RSV infection. For control, mice were administered with PBS or control liposome. At 4 days post infection, BAL fluid was removed from lung and blood was perfused. Then, lungs were fixed with 4% paraformaldehyde and embedded in paraffin. Lung sections were produced and stained with Hematoxylin and Eosin to examine the abnormalities. Four inflammatory parameters were scored independently from 0 to 5 for each section: alveolitis (inflammatory cells within alveolar spaces), interstitial pneumonitis (increased thickness of alveolar walls associated with inflammatory cells), peribronchiolitis (inflammatory cells surrounding a bronchiole), and perivasculitis (inflammatory cells surrounding a blood vessel). Slides were randomized, read blindly, and scored for each parameter.

### In vivo or in vitro Spore Administration and RSV Infection

In vivo mouse model, each mouse was administered with either 1 × 10<sup>9</sup> CFU of spore or, PBS as a control, i.n. in a volume of 20 µl at 5 days before the infection. Then, the mice were infected i.t. with 2 × 10<sup>6</sup> PFU of live RSV A2. BMMs or MH-S cell line were cultured in 12-well plate (1 × 10<sup>6</sup> cells/ml) with various ratio of spore for 24 h. Cells were washed with PBS and then infected with 1 MOI of live RSV A2 for additional 12 h.

### Statistical Analysis

fmicb-10-00447 March 8, 2019 Time: 17:38 # 11

Results are presented as means ± SEM. Statistical differences between two means were evaluated using unpaired Student t test. For comparisons of multiple groups, one-way ANOVA was used. Statistical significance was set at a P value of < 0.05.

### AUTHOR CONTRIBUTIONS

C-HY conceived the study. MS, SH, and C-HY designed the study. JH, Y-CK, S-MP, and IC performed all the experiments. JH and C-HY wrote the manuscript. HC, B-CP, Y-MP, JC, and J-HC analyzed and discussed the data.

### REFERENCES


### FUNDING

This work was carried out with the support of "Cooperative Research Program for Agriculture Science and Technology Development (Project No. PJ01336401)" Rural Development Administration, South Korea. This work was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2018R1A2B2006793 and NRF-2015R1D1A1A02061577). This work was also supported by the BK21 Plus Program of the Department of Agricultural Biotechnology, Seoul National University, Seoul, South Korea.

### ACKNOWLEDGMENTS

We thank Dr. Cutting SM (Royal Holloway University of London) for kindly providing Bacillus subtilis PY79 strain spore.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2019.00447/full#supplementary-material



GM-CSF is critical for the differentiation of fetal monocytes into alveolar macrophages. Nat. Immunol. 15, 1026–1037. doi: 10.1038/ni.3005


**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 © 2019 Hong, Kye, Park, Cheon, Chu, Park, Park, Chang, Cho, Song, Han and Yun. 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.

# Recent Aspects on the Pathogenesis Mechanism, Animal Models and Novel Therapeutic Interventions for Middle East Respiratory Syndrome Coronavirus Infections

Sinosh Skariyachan\*, Sneha Basavaraj Challapilli, Swathi Packirisamy, Supreetha Toplar Kumargowda and Vaishnavi Sneha Sridhar

R&D Centre, Department of Biotechnology, Dayananda Sagar College of Engineering, Bengaluru, India

### Edited by:

Kelvin To, University of Hong Kong, Hong Kong

### Reviewed by:

Man Lung Yeung, University of Hong Kong, Hong Kong Mahmoud Mohamed Shehata, National Research Centre, Egypt

### \*Correspondence:

Sinosh Skariyachan sinoshskariya@gmail.com; sinosh-bt@dayanandasagar.edu

### Specialty section:

This article was submitted to Virology, a section of the journal Frontiers in Microbiology

Received: 28 October 2018 Accepted: 05 March 2019 Published: 26 March 2019

### Citation:

Skariyachan S, Challapilli SB, Packirisamy S, Kumargowda ST and Sridhar VS (2019) Recent Aspects on the Pathogenesis Mechanism, Animal Models and Novel Therapeutic Interventions for Middle East Respiratory Syndrome Coronavirus Infections. Front. Microbiol. 10:569. doi: 10.3389/fmicb.2019.00569 Middle East Respiratory Syndrome Coronavirus (MERS-CoV) is an emerging zoonotic virus considered as one of the major public threat with a total number of 2 298 laboratory-confirmed cases and 811 associated deaths reported by World Health Organization as of January 2019. The transmission of the virus was expected to be from the camels found in Middle Eastern countries via the animal and human interaction. The genome structure provided information about the pathogenicity and associated virulent factors present in the virus. Recent studies suggested that there were limited insight available on the development of novel therapeutic strategies to induce immunity against the virus. The severities of MERS-CoV infection highlight the necessity of effective approaches for the development of various therapeutic remedies. Thus, the present review comprehensively and critically illustrates the recent aspects on the epidemiology of the virus, the structural and functional features of the viral genome, viral entry and transmission, major mechanisms of pathogenesis and associated virulent factors, current animal models, detection methods and novel strategies for the development of vaccines against MERS-CoV. The review further illustrates the molecular and computational virtual screening platforms which provide insights for the identification of putative drug targets and novel lead molecules toward the development of therapeutic remedies.

Keywords: MERS-CoV, emerging zoonotic virus, mechanisms of pathogenesis, animal models, probable drug targets, vaccine development

# INTRODUCTION

The coronaviruses such as HCoV-229E, HCoV-NL63 (α-coronavirus) and HCoV-OC43 and HKU1 (β-coronavirus) circulated in the human population and caused mild respiratory infections (Hamre and Procknow, 1966; Fouchier et al., 2004; van der Hoek et al., 2004; Woo et al., 2005). Over the last two decades, there have been two zoonotic emergence of coronaviruses into the human population, both linked with Severe Acute Respiratory Syndrome Coronavirus (SARS-CoV)

**39**

(Drosten et al., 2003; Kuiken et al., 2003) and Middle East Respiratory Syndrome Coronavirus (MERS-CoV) (Zaki et al., 2012; Chafekar and Fielding, 2018).

Middle East Respiratory Syndrome Coronavirus belonged to β-coronavirus, clade-c (van Boheemen et al., 2012; de Groot et al., 2013) and was initially known as Human Coronavirus Erasmus Medical Center/2012. or HCoV-EMC/2012 (Chan et al., 2012). MERS-CoV infections resulted high mortality rate in human, and till date, very limited therapeutic intervention or vaccine are available for the treatment of infections caused by the virus (Mustafa et al., 2017; Hui et al., 2018).

Middle East Respiratory Syndrome Coronavirus found in various natural hosts such as the dromedary calves (Camelus dromedarius), bats (Vespertilio superans and Neoromicia capensis), and European hedgehog (Erinaceus europaeus) (Zhang et al., 2016; Lau et al., 2017). The replication of the virus has observed in various other animals such as rabbit, goat, civet, pig, camelid, and horse (Muller et al., 2012; Chan et al., 2013a; Al-Tawfiq et al., 2014; Eckerle et al., 2014; Meyer et al., 2015). The prevalence MERS-CoV was recently reported in non-camelid domestic mammals such as sheep, goat, donkey, and cow that were in contact with camels (Kandeil et al., 2019). The origin of human MERS-CoV is considered to be from the bats. Several coronaviruses were isolated and found that which were genetically similar to human MERS-CoV. Phylogenetic analysis of MERS-CoV showed that a close relationship to Pipistrellus bat coronavirus HKU5 and Tylonycteris bat coronavirus HKU4 (Selinger et al., 2014). Studies have suggested that there were few evidences where the virus showed camel origin (Zhang et al., 2016). Studies revealed that the sero-prevalence of the virus was greater among the people those who were exhibited direct and indirect contact with dromedary camels when compared to those who were exhibited with the general population (Watson et al., 2014; Zhou et al., 2015; Conzade et al., 2018). Recent study revealed that the prevalence of MERS-CoV infection in camel workers was found to be very high in Saudi Arabia (Alshukairi et al., 2018). Thus, this review provide the recent perspectives on the epidemiology and genome composition of MERS-CoV with a special emphasis on the viral entry, clinical manifestations, mechanism of pathogenesis, animal models, detection, and development of vaccines with an insight on various computational biology perspectives.

# EPIDEMIOLOGY

The first case of MERS-CoV infection was reported in Saudi Arabia in 2012 (Zaki et al., 2012). From the period of its emergence, the virus has infected and propagated across more than 20 countries, contributing to a wide number of laboratory-confirmed cases. In 2015, the MERS-CoV outbreak in South Korea involved 186 cases which included 38 fatalities. A total of 83% cases of transmission were due to five super spreaders, and 44% of 186 cases were in the patients those who were experienced nosocomial transmission at 16 hospitals. The epidemic lasted for 2 months and the government authorities quarantined 16,993 persons for 14 days to manage the outbreak. This outbreak provides an exclusive opportunity to seal the gap in our awareness of MERS-CoV infection (Oh et al., 2018). Recent statistics suggested that as of 31 December 2018, the total of 2,279 laboratory-confirmed cases of MERS-CoV with 806 associated deaths (case fatality rate: 35.4%) were reported in 27 countries [Harcourt et al., 2018; World Health Organization (WHO), 2018a]. Among various cases, 1901 confirmed cases with 732 related deaths (fatality rate 38.5%) were reported from Saudi Arabia [Harcourt et al., 2018; World Health Organization (WHO), 2018a].

In a comprehensive report by WHO, 1841 laboratory cases were confirmed between 2012 and 2016 in which 80% of them were reported from Kingdom of Saudi Arabia. The remaining cases were reported from 27 countries in Middle East, Asia, North Africa, United States of America, and Europe [World Health Organization (WHO), 2016a]. A total of 13 cases were reported between October to November 2017, amongst which 12 cases were reported in male and only one case was reported in female. Most of the infected individuals were hospitalized due to the contact with camels and consumption of camel milk with the virus [World Health Organization (WHO), 2017a]. Additionally, two laboratory cases were also reported in which the patients belonged to the countries such as Malaysia and United Arab Emirates [World Health Organization (WHO), 2018a]. The laboratory confirmed cases and deaths reported in several countries from the year 2012 to 2018 are shown in **Table 1**.

# THE GENOME OF MERS-CoV

In the early years of 2000, coronaviruses were not acquired great importance in research and development. With the emergence of SARS in the year 2002 and MERS in 2012, the interest of many researchers was stimulated toward coronavirus (Perlman, 2015).

The morphology of coronavirus includes spherical or pleomorphic shapes with a diameter of 80–120 nm (Masters, 2006). The genome of the coronavirus consists of 6 and 7 open reading frames (ORFs). The ORF 1a and 1b encompass twothird of the viral genome which encodes the non-structural poly-proteins and the other four ORFs on the downstream side encode for the structural proteins such as envelope protein (E), Spike protein (S), nucleocapsid protein (N), and membrane protein (M). In some coronaviruses, hemagglutininesterase (HE) gene is present in the region between ORF 1b and S (Marra et al., 2003; Rota et al., 2003; Snijder et al., 2003). These structural proteins are folded and entered in to the endoplasmic reticulum (ER) and transported to the ER-Golgi transitional slot (Masters, 2006). During the replication of coronavirus, substantial amounts of structural proteins are synthesized in order to assemble the progeny virions (Fung and Liu, 2014). They occupy the RNA genome which encodes structural proteins such as S protein, M protein, and N protein (Snijder et al., 2003).

The genome size of the first imported MERS-CoV strain was identified to be 30,114 nucleotide (nt) long, including the 3<sup>0</sup> and 5<sup>0</sup> UTRs. This strain showed the genome structural ornaisation of typical betacoronavirus such as a 5<sup>0</sup> -untranslated


TABLE 1 | Laboratory confirmed cases of MERS-CoV infections and deaths reported in several countries from the year 2012 to 2018.

region (UTR) (nt 1 to 272), replicase complex ORF1ab (nt 273 to 21508), S gene (nt 21450 to 25511), ORF3 (nt 25526 to 25837), ORF4a (nt 25846 to 26175), ORF4b (nt 26087 to 26827), ORF5 (nt26834 to 27508), E gene (nt 27584 to 27832), M gene (nt 27847 to 28506), N gene (nt 28560 to 29801), ORF8b gene (nt 28756 to 29094), and 30UTR (nt 29094 to 30114). The two main poly-proteins such as pp1ab and pp1a are cleaved into 15/16 non-structural proteins (nsp) by 3C-like protease (3CLpro) and papain-like protease (PLpro). These proteases are cleaved from pp1ab along with other ORFs encoding nsp which are essential for activating RNA dependent RNA polymerase, helicase, exo-ribonuclease, endo-ribonuclease, and methyltransferase. These nsp were identified to be nsp12, nsp13, nsp14, nsp15, and nsp16. The nsp14 is necessary for proofreading the ubiquitous mutation of RNA virus (Ziebuhr et al., 2000; Snijder et al., 2003; Gorbalenya et al., 2006; Smith et al., 2013; Raj et al., 2014; Durai et al., 2015). Studies revealed that the accessory ORF proteins play an important role in MERS-CoV infection and pathogenesis (Menachery et al., 2017). This study showed that the absence of major accessory ORFs such as deletion of ORF3, -4a, -4b, and -5 (dORF3-5) played major role in the viral replication and pathogenesis. Further, the attenuation of the mutant dORF3-5 was found to be responsible for the dysregulated host responses such as augmented interferon (IFN) pathway activation, disrupted cell processes, and robust inflammation. Thus, the disruption of accessory ORFs probably provide platform to the attenuation of future emergent strains of MERS-CoV accessory ORF mutants. The accessory ORF functions can be targeted for both therapeutic and vaccine treatments in response to MERS-CoV and related group 2C corona viruses (Menachery et al., 2017). The graphical representation of the genome organization of MERS-CoV is illustrated in **Figure 1**. Coronaviruses have demonstrated the establishment of double membrane vesicles (DMVs) in the infected cells. Based on immune-histochemistry and electron microscopic studies, the coronavirus replicase proteins are colocalized with the DMVs and are assumed to be the site of replication/transcription complex (Fung and Liu, 2014).

An interesting fact of coronaviruses is that they display recombination, high rates of mutation, and propensity to cross species. One such event was reported during the 2002–2003 epidemic, where the transmission of SARS-CoV has occurred from Chinese horseshoe bats to human populations (Perlman and Netland, 2009). MERS-CoV has the capability to adjust new environments and acquire various virulence factors and enhance their ability to transfer these factors from humanto-human due to continues outbreak (Zumla et al., 2015). The phylogenetic analysis revealed that all human and camel MERS-CoVs were identified to be homologous to each other. Additionally, the hedgehog and bat MERS-CoVs demonstrated para-phyletic group to human and camel MERS-CoV clade (Zhang et al., 2016).

Kim (2016) reported that the occurrence of a mutant strain of MERS-CoV toward the human CD26 receptor during the South Korean outbreak. Furthermore, they isolated 13 new viral genomes in which, 12 possessed point mutation in the receptorbinding domain (RBD) of S protein (Kim, 2016). The S protein is suggested to be essential for the host tropism through its interaction with the host CD26 receptor (van Doremalen et al., 2014; Wang et al., 2014).

# PATHOGENESIS

# Viral Entry

The entry of MERS-CoV to the host is facilitated by its type-I transmembrane glycoprotein known as S protein (Xia et al., 2014). The RBD of MERS-CoV consists of S1 subunit which ranged from 367 to 606 amino acids and can be grouped into an external sub domain and core domain as shown in **Figures 1C,D**. The receptor-binding motif, V484 – L567 of RBD is situated at the external sub domain and the core sub domain consists of anti-parallel β-sheet (five-stranded) and six helices in the center. The two small β-strands make a globular fold. The core domain structure is balanced by three disulfide linkages. The MERS-CoV RBD external sub domain contains a β-sheet along with

one small and three large strands which are organized in antiparallel manner. The RBD core is attached to this region by special intervening loops and it binds to the core sub-domain by a clamp at the lower and upper regions. Most of the connecting loops and two small 3<sup>10</sup> helices are located in the inner region of the sheet. The fourth disulfide linkage is stabilized among the residues starting from C503 to C526 and β6-strand linked with η3-helix. Three helices (HR1) at the center and three chains (HR2) near to the core promote the release of the viral particle into the cytoplasm and promote the progression of the infection (Durai et al., 2015).

Middle East Respiratory Syndrome Coronavirus can also enter the cell through an auxiliary pathway on the cell surface via transmembrane proteases. The binding of MERS-CoV into host cell and replication of the virus is shown in **Figure 2**. The host protease cleaves coronavirus S protein into two functional domains distinctive to each other at the N-terminal region (denoted as the S1 subunit) which comprised of RBD, and C-terminal portion (S2 subunit) comprised of a fusion peptide, two heptad repeats (HR2 and HR1) and the transmembrane (TM) domain (Weiss and Navas-Martin, 2005). The membrane fusion is conciliated by main conformational rearrangement which exposed the fusion peptide resulting in the development of six-helix bundle (6HB). The core of 6HB consists of a triplestranded coiled like structure constructed by HR1s of the three spike subunits forming trimers; the HR2 elements are found to be packed within the grooves of the coiled coil in an anti-parallel direction. Owing to their central role in membrane fusion, it was reported that various antiviral peptides interfere the formation 6HB (Forni et al., 2015).

# Transmission and Clinical Presentation

The clinical presentation of MERS-CoV compasses from severe respiratory diseases to subclinical infections (Drosten et al., 2013; Hui et al., 2018). Infected patients often indicate the presence of hemoptysis, sore throat, fever, cough, shortness of breath, and other gastrointestinal symptoms such as diarrhea and vomiting (Albarrak et al., 2012; Assiri et al., 2013a,b; Guery et al., 2013; Health Protection Agency, 2013; Omrani et al., 2013). A low range of lower pulmonary infiltrate associated with viral pneumonia was observed in the radiograph of patients infected with MERS-CoV (Memish et al., 2013a; Wang, 2014). It was observed that greater than 60% of the initially reported cases of MERS-CoV infection, the patients experienced severe disease which demanded intensive care treatments like extracorporeal membrane oxygenation and mechanical ventilation. Hematological aberrations delineated for the clinical cases of neutrophilia (8%), lymphopenia (34%), thrombocytopenia (36%), and lymphocytosis (11%). While the conditions of kidney failure necessitate renal replacement therapy for a considerable number of MERS-CoV cases, several studies detected MERS-CoV antigens and particles in the renal tissues in vivo, which is the direct indication of the virus replication in renal tissues and long term acute infection (Yeung et al., 2016; Alsaad et al., 2018).

complete double membrane vesicle and lastly through exocytosis and MERS-CoV is released out of the host cell (Du et al., 2009).

The first evidence of human-to-human transmission of MERS-CoV was reported in few cases in United Kingdom, when an adult male traveled to Saudi Arabia and transferred the infectious virus to two of his family members (Perlman and Netland, 2009). It was suggested that MERS-CoV infections were due to the introduction of the virus in humans and the close contact with camels was one of risk factors for the transmission of MERS as reported by World Health Organization [World Health Organization (WHO), 2016b].

The transmission of MERS-CoV is determined as spasmodic, often healthcare associated, intra-familial, and required prolonged and close contact (Memish et al., 2013a,b, 2014a,b; Puzelli et al., 2013; Drosten et al., 2014). In a domestic study conducted with MERS-CoV infected patients, 14 out of 280 contacts of 26 positively indexed patients (antibody positive or RNA) suggested that the frequency of the transmission even during outbreaks was approximately 3% (Drosten et al., 2014). Several studies suggested that the virus was not self-sustained, as during local epidemic of MERS-CoV, which has not been readily transmitted to more than one human and has been reported in majority of human related cases (Bauch and Oraby, 2013; Breban et al., 2013; Cauchemez et al., 2014; Chowell and Nishiura, 2014). The mode of transmission of the virus is illustrated in **Figure 3**.

The serological analysis and extensive investigation of those potentially exposed to the patients suggested that there were no secondary infections [World Health Organization, 2013b]. As a minimum of 18 cases of asymptomatic MERS-CoV infections were reported in health care personals, the function of these subclinical cases in the transfer of infection was found to be uncertain (de Sousa et al., 2014).

# Molecular Mechanism of the Pathogenesis

While there were intensive studies on coronavirus research over the past 5 years, limited reports are available on the pathogenesis of MERS-CoV. Recent post-mortem histopathological studies revealed that the localization of viral particles in the pulmonary and extra pulmonary tissue of a 33-year-old male patient of T cell lymphoma, who acquired MERS-CoV infection (Ng et al., 2016; Alsaad et al., 2018). The histopathological analysis of the biopsies were collected from lung, brain, liver, heart, kidney, and skeletal muscle demonstrated that there were pulmonary diffuse alveolar damage, necrotizing pneumonia, acute kidney injury, portal and lobular hepatitis and myositis with muscle atrophic changes. Further, the viral particles were found to be localized in the pulmonary macrophages, pneumocytes, renal proximal

tubular epithelial cells and macrophages infiltrating the skeletal muscles (Alsaad et al., 2018).

Studies suggested that the pathogenesis of MERS-CoV in human and animals are mainly due to three mechanisms such as DPP4 (dipeptidyl peptidase-4) mediated mechanism, papain like protease PLpro mediated mechanism and accessory proteins like p4a and membrane M protein mediated mechanism. Hence, these proteins are probably considered as one of the potential therapeutic targets for MERS-CoV infections (Durai et al., 2015). The molecular mechanisms exhibited by various modes of pathogenesis are illustrated in **Figure 4**.

### DPP4 MEDIATED MECHANISM

DPP4 is a type-II transmembrane glycoprotein consists of approximately 766 amino acids and function as major receptor for MERS-CoV. The three dimensional structural analysis revealed that this receptor has α/β-hydrolase domain and a β-propeller domain with eight blades that helped in the binding of receptor binding domain in MERS-CoV (Raj et al., 2014). The crystal structure of DPP4 is illustrated in **Figure 1C**. A DPP4 binding protein, adenosine deaminase, is considered as one of the major inhibitors of the S protein of MERS-CoV (Drosten, 2013).

A number of cell lines such as human-derived HFL, Calu-3, Caco-2, Huh-7, HEK, His-1 cell lines, CD8+, and CD4+ that are susceptible to MERS-CoV were reported (Chan et al., 2013b; Shirato et al., 2013; Chu et al., 2016). Two S protein ecto-domains can be divided into fusion-catalyzing domains (FD) and RBD. The S RBDs bind to DPP4 after which the FDs are exposed through unfolding. Further, the cellular and viral membranes were joined together by refolding (Raj et al., 2013). The unfolding FD required the cleavage of S protein by host proteases. Out of various proteolytic cleavage sites between the FD and RBD, the cleavage at S1/S2 site is essential for the infection of Calu3 cells by MERS-CoV (Park et al., 2016).

It has been revealed that upon infection, monocyte chemo attractant protein-1 and IFN-γ-inducible protein-10 were induced and suppressed the proliferation of human myeloid progenitor cells (Zhou et al., 2014). MERS-CoV can infect T-cells from human lymphoid organs and causes the peripheral blood inducing apoptosis by intrinsic and extrinsic pathways

(Chu et al., 2016). A study conducted by Al-Qahtani et al. (2017) suggested that the inhibition of DPP4 mitigated the induction of PPARγ (the transcriptional repressor) and IRAK-M (negative regulator of TLR signaling) which indicated that the immunosuppressive action of S glycoprotein regulated by DPP4.

### PAPAIN LIKE PROTEASE PLpro MEDIATED MECHANISM

Upon entry into the cells, two ORFs at the 5<sup>0</sup> -end of the viral genome are translated into two poly-proteins. They are processed by two viral proteases, PLpro and 3C-like proteinase (3CLpro) and form 16 nonstructural proteins which are necessary for the membrane-associated replication complex. PLpro is located in nonstructural protein 3. PLpro was found to be multifunctional enzymes with deISGylating (removal of ISG15 conjugates from host cell factors) and deubiquitinating (cleavage of ubiquitin from host cell factors) properties in addition to protease activity in many coronaviruses (Clasman et al., 2017). This resulted in the host antiviral immune response being antagonized and promoting the viral replication. PLpro can inhibit the mitochondrial anti-viral signaling protein induced IFN-β reporter activity and reduce TNF-α induced NF-κB reporter activity. Hence, MERS-CoV PLpro is considered to be important antiviral target (Mielech et al., 2014; Harrigan et al., 2018).

The coronavirus genome is encoded by two ubiquitin-like domains (Ubl1 and Ubl2), out of which Ubl2 is considered to be an important component which influences the PLpro activity. Clasman et al. (2017) reported that PLpro has not showed the deISGylating, de-ubiquitinating or protease activities in the absence of Ubl2 domain. There were no variations in the process of inhibition, substrate specificity, catalytic efficiency and thermal stability (Clasman et al., 2017).

### ACCESSORY PROTEINS MEDIATED MECHANISM

The viral genome is recognized by melanoma differentiationassociated protein-5 (MDA5), retinoic acid inducible gene-1 (RIG-1) and endosomal toll-like receptor 3 (TLR3) as pathogenassociated molecular patterns. This recognition resulted in the formation of type-1 interferon (IFN1). As an evasion mechanism, virus synthesize proteins that hinder the production IFN1 in the pathway (Zinzula and Tramontano, 2013).

In the absence of MDA5 and RIG-1, MERS-CoV used the protein such as 4a (p4a) which contain RNA-binding motif for binding (Rabouw et al., 2016). By a direct binding, this motif masks the dsRNA of MERS-CoV. The p4a possesses αβββα fold with the β1-β2 loop and α1 helix which bind to the minor groove of dsRNA. The mutational studies revealed that p4a-dsRNA complex stabilized by the amino acids such as K27, W45, K63, and K67 (Batool et al., 2017). In a study carried out by Canton et al. (2018) suggested that, NF-κB remained in the cytoplasm of MERS-CoV infected cells while 4b was found to be attached with the nucleus. In the absence of 4b, expression of pro-inflammatory cytokines was observed as NF-κB translocated to the nucleus. This was also observed in the case of 4b mutants that devoid of nuclear localization signal, indicating that nuclear localization signal mediated process of 4b is essential for NF-κB expression (Canton et al., 2018).

As a component of the viral envelope, M protein is involved in various functions such as viral assembly along with other proteins such as E, S, N. The M, E, and S proteins interact with N protein in the ER-Golgi complex. This interaction hinders the fusion of viral and cellular membranes (Mustafa et al., 2018). It was observed that the expression of M protein in MERS-CoV might suppressed the type-I IFN expression in response to poly (I: C) induction or Sendai virus infection. This reaction was detected to be specific for the activation of IFN regulatory factor 3 (IRF3), however, not activated the nuclear factor-κB. The interaction of MERS-CoV M protein with TRAF3 and the association of disrupted TRAF3–TBK1 lead to reduced activation of IRF3 (Lui et al., 2016).

# HOST IMMUNE RESPONSE

Middle East Respiratory Syndrome Coronavirus can readily infect human respiratory epithelial cells, macrophages, T-cells, dendritic cells and can influence the production and induction of pro-inflammatory cytokines and chemokines (Zhou et al., 2015). Studies on ex vivo human lung tissue and respiratory epithelial cell lines showed that the induction of antiviral interferons followed by MERS-CoV infection by stimulated epithelial cells was found to be limited. A study by Menachery et al. (2014) described that the cells infected with MERS-CoV demonstrated distinctly altered chromatin structures such as repressive histone markers, that could limit the transcription factors from binding to the promoter regions of interferon-stimulated gene. The changes in DNA methylation found to be one of the reasons for down regulation of IFN-γ–associated antigen-presenting gene after the infection (Menachery et al., 2018). MERS-CoV proteins such as 4a, 4b, M, and PLpro were found to be suppressed the induction of interferons (Yang et al., 2013; Mielech et al., 2014; Canton et al., 2018). Further, the induction of pro-inflammatory cytokine response in these cells was found to be limited. A delayed induction of pro-inflammatory cytokines/chemokines such as IL-6, IL-8, and IL-1β were observed upon the viral infection (Lau et al., 2013; Canton et al., 2018).

Middle East Respiratory Syndrome Coronavirus could infect and replicate in human monocyte-derived macrophages and immature monocyte-derived dendritic cells, while it failed to perform similar functions in mature monocyte-derived dendritic cells. This indicate that the maturation restrict the viral replication. Immature dendritic cells were unable to stimulate T-cell even after the antigen uptake, delaying the activation of T-cells and permitting further viral replication (Cong et al., 2018). The infection of human macrophages induced the expression of pro-inflammatory cytokines/chemokines such as MCP-1/CCL-2, IFN-α2, IFN-γ, MIP-1α/CCL-3, IP-10/CXCL-10, RANTES/CCL-5, IL-8, TNF-α, IL-12p40, and IL-6

(Zhou et al., 2015; Cong et al., 2018). The infected dendritic cells induced the expression of IL-12, RANTES/CCL-5, IP-10/CXCL-10 and IFN-γ, however, showed low expression of IFN-α and no expression of IFN-β (Zhou et al., 2015).

T-cells can be directed to the infection site by type-I IFN stimulated secretion of IL-10 and CXCL10. Due to their uncontrolled and strong expression, the expression of IFNγ and IL-12 were inhibited and restricted the activation of T-helper cells. With the added effect of down-regulated antigenpresentation pathways, the activation of T cells is inhibited. Hence, the sequestered T cells failed to target the virus (Ying et al., 2016). Previous studies revealed that when the serum samples and bronchoalveolar lavage were analyzed from two patients, it was observed that the patients with low level secretion of IFN-α were failed to survive and the patients displayed high level secretion of type-I IFN were survived (Faure et al., 2014).

### ANIMAL MODELS

The animal models are one of the vital components for translation of the findings from drug discovery process from bench to bedside. These animal models are indispensable for the deeper understanding of any diseases and safe and protective vaccine development process (Lorenzen et al., 2015). The main criteria for the validation of animal models in preclinical studies are face validity, predictive validity and target validity. The face validity means that the proper similarity between the animal model and the human disease in terms of the biology and symptoms of the disease. However, assessing face validity is a tedious task due to the lack of understanding on the core biology of the disease symptoms. Predictive validity means the display of clinical interventions which exhibited similar effect in the animal models. This is also very difficult to achieve due to the incomplete correlation between animal models, the mechanism of human diseases and the incapability of standard drugs to active in many animal models. Target validity is another essential component for the validation of animal models in which the target under study must have an analogous function in the disease model as in the clinical condition (Denayer et al., 2014).

An ideal animal model is the one where an immunocompetent animal, on receiving the challenge virus at a realistic dose through a suitable inoculation route, is able to replicate the characteristics of a human disease as closely as possible (Doremalen and Munster, 2015). Ferrets, hamsters, mice and other small laboratory animals are not ideal animal models for MERS-CoV owing to the variations in DPP4 receptor. The rhesus macaque and the common marmoset are the first non-human primate species (NPS) to be used as the animal models for MERS-CoV. Glycosylation of mouse DPP4 (mDPP4) prevents its binding to S protein of MERS-CoV. Hence, transgenic hDPP4 (human DPP4) mice were developed as animal model (Peck et al., 2015). The major advantages and limitations of MERS-CoV animal models are listed in **Table 2**.

### Mice

Studies suggested that the glycosylation of certain amino acids prevented the binding of mouse DPP4 (mDPP4) to S protein of MERS-CoV (Doremalen and Munster, 2015). The first transgenic MERS-CoV mouse model was established by Agrawal et al. (2015). The mice expressed hDPP4 and developed severe respiratory MERS-CoV infections. The transgenic mice infected with an intra-nasal dose of 10<sup>6</sup> TCID50 of MERS-CoV developed pneumonia, weight loss, acute pulmonary viral infection, ruffled fur, squinting, and death within few days (Agrawal et al., 2015). A transgenic mouse model developed by Zhao et al. (2015) showed that limited utility for pathogenic studies due to the systemic dissemination and development of severe neurological diseases. The other lung infection models were lack mortality and weight loss, however, developed lung pathology in response to 50% TCID<sup>50</sup> of MERS-CoV (Tao et al., 2015).

Recent studies suggested that the transgenic mice models were developed where the full-length mDPP4 gene replaced by hDPP4 (Coleman et al., 2017). The infection and replication of MERS-CoV was observed in the lungs, however, disseminated slow

TABLE 2 | Advantages and limitations of MERS-CoV experimental animal models.


impact to the other organs during the post injection with dose range of 102–105 PFU. The pathology of infected mice showed pneumonia-like symptoms (Coleman et al., 2017).

The CRISPR-Cas9 gene editing tools were used to produce mice susceptible to MERS-CoV infection by modifying their genome sequence (positions 288 and 330) homologous to the sequence of hDPP4. In certain mice model the characteristic symptoms such as pulmonary hemorrhage, low survival, decreased pulmonary function and weight loss were observed. The 288-330+/+MERS-CoV mouse model demonstrated severe respiratory infection as observed in humans (Cockrell et al., 2016).

### Rhesus Macaques and Marmosets

Upon infection with MERS-CoV, rhesus macaques and marmosets were developed moderate respiratory disease which are comparable with mild human MERS-CoV infection. As observed in human cases, they also showed complete blood count abnormalities. Rhesus macaque and marmoset models were found to be suitable for pathogenic studies of mild infections. Both the models can be used for the efficacy testing of prophylactic and therapeutic countermeasures (Baseler et al., 2016).

Yu et al. (2017) reported that when 2-3 old rhesus macaques and common marmosets were intra-tracheally injected with hCoV-EMC, hematoxylin and eosin stained tissues of the infected model demonstrated lesions primarily in the lungs with various degrees of inflammation, hemorrhaging, pulmonary oedema, interstitial pneumonia, eosinophil infiltration and necrosis in bronchial epithelial cells and pneumocytes. The lungs of the infected marmosets demonstrated widespread pulmonary oedema and hemorrhaging. The oedematous alveolar cavities showed fibrinous exudates and neutrophil infiltration. In both the cases, DPP4 was widely expressed in alveolar macrophages and type-I and II pneumocytes were also actively involved. The pathological changes as a result of viral infection were not observed in any other organs of the infected animal (Yu et al., 2017).

### Rabbits

In most cases, MERS-CoV causes severe infection in the lower respiratory tract in humans. Nasal swabs collected from the infected rabbits demonstrated the presence of MERS-CoV. They generally develop respiratory diseases such as asymptomatic and mild human infections as sometimes observed in immunecompetent patients (Haagmans et al., 2015).

Recently, Houser et al. (2017) revealed that the infected New Zealand rabbits were developed asymptomatic pulmonary infections with high levels of viral antigens and RNA. Multiple lung lobes showed peri-vascular inflammation. The antibodies developed by the rabbits lacked neutralizing activity, and as a result the animals were susceptible to re-infection. In fact, enhanced pulmonary inflammation was observed after re-infection. However, re-infection elicited the neutralizing antibodies (Nabs). Hence, it was suggested that upon re-infection, there might be greater risk of severe lung disease in those who do not develop neutralizing antibody response, or those whose neutralizing antibody titers have decreased after the initial infection (Houser et al., 2017). The clinical signs and lesions along with the occurrence of MERS-CoV RNA and antigens observed in MERS-CoV infected animal models are shown in **Table 3**.

# DIAGNOSTIC APPROACHES

The primary diagnosis of MERS-CoV infection was performed by molecular techniques such as real-time reverse transcriptase PCR (RT-PCR) (Corman et al., 2012a), reverses transcription– loop-mediated isothermal amplification (RT-RTPA) (Shirato et al., 2014) and reverses transcription-recombinase polymerase amplification (RT-LAMP) (Abd et al., 2013). Numerous serological assays were used to detect MERS-CoV or closely related viruses in seropositive camels; these tests were protein microarrays like indirect enzyme-linked immunosorbent assay (ELISA), recombinant spike immunofluorescent assay, spike pseudoparticle neutralization and microneutralization assay (Buchholz et al., 2013; Perera et al., 2013). However, none of the serological assays showed evidence of the precise occurrence of MERS-CoV in camels (Song et al., 2015). Recent study reported that a rapid and specific assay for the detection of MERS-CoV such as nucleic acid visualization technique which combine the reverse transcription loop-mediated isothermal amplification technique and a vertical flow visualization strip (RT-LAMP-VF) to detect the N gene of MERS-CoV (Huang et al., 2018). The study suggested that in comparison with the

TABLE 3 | Clinical signs and lesions along with the occurrence of MERS-CoV RNA and antigens as seen in MERS-CoV infected animal models.


real-time RT-PCR (rRT-PCR) method recommended by the World Health Organization, the RT-LAMP-VF assay is easy to perform and the rapid detection of MERS-CoV within 35 min was achieve by this approach (Huang et al., 2018).

An assay was developed based on the detection of N protein of MERS-CoV (Song et al., 2015). The NP protein preferred as better target than the S protein. In this type of assay, firstly, the antigen binds to gold labeled monoclonal antibodies which lead to the formation of antigen-antibody complex. The complex further passed through nitrocellulose membrane and bind to monoclonal antibody at the test region and formed band which could be conveniently analyzed (Tang, 2014).

Currently, multiple purified monoclonal antibodies are available in clinical and preclinical studies for the development of antimicrobials (Rockx et al., 2010). Studies suggested that two human antibodies (REGN3048 and REGN3051) are known to bind with MERS-CoV RBD and prevent the interaction of cellular receptor DPP4 with S protein and effectively neutralize MERS-CoV infection (Pascal et al., 2015). The antibodies such as REGN3048 and REGN3051 showed no binding with S protein, thus, employed as effective combination for inhibiting MERS-CoV (Aderem et al., 2011). REGN3048 and REGN3051 cointeract with MERS-CoV RBD, which suggested that these antibodies bind to discrete epitopes as they bind to regions of MERS-CoV S protein that are conserved during the natural evolution of the virus. These two antibodies were blocked entry of the virus into susceptible cell lines and neutralize the infection (Pascal et al., 2015).

The diagnostically relevant variations in the neutralization activity have not been detected in several isolates of MERS-CoV. In order to detect the serological response to a specific type of single MERS-CoV serotype, specific protein based sero-assays are essential to be performed (Muth et al., 2015). The development of potential serological assays demand well delineated human or animal sera and antibodies specific to MERS-CoV (Meyer et al., 2014).

# VACCINES AND THERAPEUTICS

Previous studies revealed that both cellular and humoral responses are required for the complete protection against MERS-CoV infections (Faure et al., 2014; Park et al., 2015; Corman et al., 2016). Nabs and T-cell responses are the main agents in providing protective immunity against coronaviruses. In a study by Zhao et al. (2014) reported that MERS-CoV cleared by mice which lack B-cells while those deficient in T-cells failed to perform similar functions. Further, since virus-specific humoral responses were decreased over time, developing a vaccine which could induce long-lived memory T-cell response considered to be beneficial. However, any vaccine compounds which are able to induce both T and B-cell responses are required for immune protection (Okba et al., 2017).

### Current Vaccine and Antiviral Candidates

MERS-CoV vaccine candidates are still under development with DNA based, protein based, live attenuated and vectored-virus based vaccines (Arabi et al., 2017). Out of the four structural proteins and sixteen non structural proteins encoded by the viral genome, S protein showed the highest immunogenicity which induced T-cell responses. In most of the MERS-CoV, Nabs target the RBD of the S protein. The variations in amino acid sequences of the S protein detected amongst MERS-CoV is limited, thus, designing a vaccine against one strain most likely to be protected against all the other circulating strains (Okba et al., 2017). The S1 protein located outside the RBD probably considered as potential drug target to develop therapeutic strategies against MERS-CoV. RBD-specific mAbs are considered to be displayed greater neutralizing activity than those were targeted the S1 or S2 region outside RBD (Du et al., 2013).

DPP4-targeting therapeutic agents are capable of blocking the interaction between MERS-CoV RBD and DPP4, and therefore inhibit MERS-CoV infection. The anti-DPP4 (CD26) antibodies such as 1F7, 2F9, and YS110 prevent the entry of MERS-CoV into susceptible cells, obstructing the virus infection (Ohnuma et al., 2013). Additionally, DPP4-binding protein adenosine deaminase (ADA) competes the binding with MERS-CoV RBD to DPP4, specifically at crucial residues such as Q286 and L294, defining their role as a naturally occurring antagonist (Raj et al., 2014).

Recent studies revealed that the DNA vaccines are the safest vaccines for immune-compromised patients. They possessed comparatively greater amounts of DNA (e.g., 0.5–2 mg) thus requires multiple inoculations to induce potent immune responses in rhesus macaques and mice (Wirblich et al., 2017). Chi et al. (2017) developed a DNA vaccine which encoded the first 725 amino acids of S protein in MERS-CoV which induced the cell mediated and antigen-specific humoral immune responses in mice. Post three immunizations, it was observed that high titers of Nabs (up to 1: 104) were produced without any adjuvant. A prophylactic DNA-plasmid vaccine known as GLS-5300, which encodes for S protein, is the first MERS-CoV vaccine to be considered for the human trials developed by GeneOne Life Inc., which has been enlisted in the on-going list of vaccines by WHO [Rabaan et al., 2017; World Health Organization (WHO), 2018b].

The vaccines that imitate the natural infection, such as recombinant viral vectors or live-attenuated virus present the antigens in natural ways which further stimulate the activation of humoral and cell-mediated immunity to the native protein conformation. As the live attenuated viruses have been the most immunogenic platforms available, they exhibited the capability to present numerous antigens across the viral life cycle in their native conformations (Henderson and Moss, 1999). Agrawal et al. (2016) vaccinated the mice with an inactivated MERS-CoV vaccine and which demonstrated hypersensitivity risk factors in MERS-CoV infection. Although there were extensive research in the last few years, limited anti-MERS-CoV therapeutics have been approved for human use. Originally, conventional IFAs were employed in the human sero-surveys and these relied on the MERS-CoV infected cell cultures. These cell cultures were the primary antigen sources for detecting the presence of human anti-MERS-CoV Nabs, IgM, and IgG in human samples (Corman et al., 2012b). Further, it was reported that human monoclonal

antibody m336 demonstrated significant neutralizing properties against MERS-CoV in vitro conditions (Houser et al., 2016).

Recent studies revealed that there were few antiviral agents in the developmental stage. It was reported that the antiviral drug chlorpromazine demonstrated strong anti-MERS-CoV potential; however, the high cytotoxicity of the drug limiting the potential window for drug utilization. Similarly, another drug Toremifene showed moderate activity when tested in antigen presenting cells, however, the high cytotoxicity of this drug also narrowing its application as a therapeutic remedy against MERS-CoV (Cong et al., 2018). Further, studies suggested that a nucleoside analog GS-5734 known as Remdesivir, which showed potential antiviral activity against human and zoonotic CoVs in vitro and the drug inhibited MERS-CoV at 50% effective concentration values (EC50) (Agostini et al., 2018).

### NOVEL THERAPEUTIC APPROACHES

The conserved nature of the N protein was used to develop vaccines that can induce an adaptive immune response against these proteins. In a study by Zhao et al. (2016) reported that a vaccine candidate induced airway memory CD4+ T cell response against N-specific epitopes of MERS-CoV. In another study suggested that vaccination with recombinant N-terminal domain (rNTD) triggered greater T-cell response than the mice vaccinated with rRBD. The neutralizing activity was also observed in the sera (Jiaming et al., 2017). Recent study revealed that the combination of fusion inhibitory peptide which targets the protein HR1 domain of MERS-CoV S2 and a neutralizing antibody specific for the S1 protein RBD showed synergistic therapeutic activities against MERS-CoV (Wang et al., 2019). Further, the researchers have developed novel neutralizing nanobodies (Nb) which specifically bind to the RBD of MERS-CoV S protein. The Nb was found to be interacted with the conserved domain of MERS-CoV RBD with high affinity and blocked binding of RBD to the receptor of MERS-CoV (Zhao et al., 2018).

Some immunodominant epitopes within the protein structure may contribute negatively to the neutralization activity of the vaccine. For example, a short region of RBD was able to induce higher titer of IgG when compared to other longer regions (Ma et al., 2014). The negative effects of these epitopes can be surmounted by immunofocusing; mapping of the most neutralizing RBD fragment and eliminating unnecessary fragments. This can help in focusing the immune response to important epitopes (Ma et al., 2014). Pallesen et al. (2017) suggested that the structure-based design of S proteins improved their immunogenicity where the high titers of Nabs were observed against MERS-CoV. The major categories of vaccines developed and tested against various animal models are shown in **Table 4**.

### Therapeutic Screening by Computational Biology

The technological progression is reflected upon a series of genome wide molecular screening platforms and computational biology approaches that provided novel insights to prompt response against emerging viral diseases. The prototype of these approaches included the in vivo animal model, tissue culture model, human challenge model, and vaccine studies. These model systems are anxious by challenges, preferably yielding a spectra of the severity of the diseases like lethal versus sub-lethal in order to increase divergence for downstream data mining and modeling (McDermott et al., 2011; Drosten et al., 2013; Law et al., 2013). A diversity of computational methodologies and network approach were used to detect the regulatory networks of various viral systems by de novo studies (Hou and Zhao, 2013; Mitchell et al., 2013).

Recently, studies have shown that peptides can be utilized as highly potent signal transduction agents for various viral infections. It has been suggested that antimicrobial peptides (AMPs) can be used as potential therapeutic option against MERS-CoV infections (Mustafa et al., 2018). Computational biology approach in combination with virology can be employed to design novel peptide based therapeutics against various strains of MERS-CoV with high efficacy. Recent studies have suggested that repurposing of existing and clinically approved anti-viral peptides can be used as promising lead molecules for the development of novel anti-MERS-CoV agents (Mustafa et al., 2018).

The RBD and S protein based vector vaccines are considered to be one of the efficient targets for designing MERS-CoV vaccine, as it comprised of multiple neutralizing epitopes which offer higher potential to prompt Nabs against the virus (Shi et al., 2015). The vast developments in computational biology have led to the new line of immunoinformatics databases and tools, which are other operative ways for vaccine development (Terry et al., 2015).


TABLE 4 | Categories of vaccine developed against MERS-CoV, their potential candidate, and their target long with their immunological response and efficacy.

The screening of dominant immunogens based on the availability of genome sequence data is crucial for designing novel vaccines, which is assisted by immunoinformatics (Firbas et al., 2006). The epitope directed lead design was one of the critical steps involved in the lead identification by immunoinformatics approaches (Shi et al., 2015). MHC-I restrained CD8+ cytotoxic T-lymphocytes (CTLs) identification provides profound insight in designing novel antiviral agents (Terry et al., 2015).

In conclusion, MERS-CoV infection continued to be fatal disease and has emerged as one of the epidemic concern. Irrespective of the various studies conducted to determine the distribution, the exact intermediates remain unidentified. There are various animal models available which provided profound insight in understanding the transmission and pathogenecity of the virus in human. The common pathogenesis mechanisms of the virus such as DPP4, PLpro, accessory proteins like p4a and membrane M protein provided significant insight for the screening of novel drug targets for vaccine development. The detailed understanding of the binding mechanism of various

### REFERENCES


inhibitors toward the structural, non-structural, and accessory proteins of the virus probably provide profound insight for lead development. The integration of genome analysis, proteomics studies, immunoinformatics, and systems biology approaches on various animal models have made the recognition of new targets and lead molecules much easier in than the approaches available during earlier time. By considering MERS-CoV infections as one of the great public threats, there is high demand for undertaking the coronavirus research at the deeper molecular level to understand the mechanism of viral infection, development of advanced and rapid detection methods and futuristic therapeutic strategies to combat MERS-CoV infections.

### AUTHOR CONTRIBUTIONS

SS, SC, SP, SK, and VS collected data and prepared the complete manuscript. Further, SS reviewed, revised, and edited the manuscript.



Doremalen, N. V., and Munster, V. J. (2015). Animal models of middle east respiratory syndrome coronavirus infection. Antiviral Res. 22, 28–38. doi: 10.1016/j.antiviral.2015.07.005


Drosten, C. (2013). Is MERS another SARS? Lancet Infect. Dis. 13, 727–728.


induces protective immunity against Middle East respiratory syndrome coronavirus in nonhuman primates. Sci. Transl. Med. 7:301ra132. doi: 10.1126/ scitranslmed.aac7462



assessment 5 December 2016 WHO/MERS/RA/16.1. Available at: https://www. who.int/emergencies/mers-cov/en/


implications for pathogenesis. J. Infect. Dis. 209, 1331–1342. doi: 10.1093/infdis/ jit504


**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 © 2019 Skariyachan, Challapilli, Packirisamy, Kumargowda and Sridhar. 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.

# Drugs for Influenza Treatment: Is There Significant News?

Nicola Principi <sup>1</sup> , Barbara Camilloni <sup>2</sup> , Anna Alunno<sup>2</sup> , Ilaria Polinori <sup>3</sup> , Alberto Argentiero<sup>3</sup> and Susanna Esposito<sup>3</sup> \*

<sup>1</sup> Università degli Studi di Milan, Milan, Italy, <sup>2</sup> Department of Medicine, Università degli Studi di Perugia, Perugia, Italy, <sup>3</sup> Department of Surgical and Biomedical Sciences, Pediatric Clinic, Università degli Studi di Perugia, Perugia, Italy

Vaccines remain the best measure to reduce total influenza burden. However, presently available influenza vaccines have some limitations that cause a reduced efficacy compared to immunization practices with other respiratory pathogens. This paper shows the clinical roles of antiviral drugs against influenza that have been licensed in at least one country and the potential roles of compounds that are in development. Several attempts have been made to develop new agents against influenza viruses to overcome the supposed or demonstrated limitations of neuraminidase inhibitors (NAIs). Antibodies against the highly conserved stem region of the haemagglutinin molecule of influenza A viruses and drugs that target different stages of the influenza virus life cycle than NAIs in human cells have been developed and tested. Among these preparations, baloxavir marboxil (BAM), and favipiravir (FP) (i.e., polymerase inhibitors) are the only drugs that have reached the market (the first in Japan and the USA, and the second only in Japan). Other antiviral compounds and monoclonal antibodies are in advanced stage of development, but none of these new drugs and monoclonal antibodies in development have adequate characteristics to substitute for NAIs at present. However, although NAIs remain the drug of choice for influenza treatment, their overuse has to be avoided. Accurate selection of patients for whom treatment is truly needed is required.

### Edited by:

Ville Peltola, Turku University Hospital, Finland

### Reviewed by:

Elena A. Govorkova, St. Jude Children's Research Hospital, United States Esteban A. Hernandez-Vargas, Frankfurt Institute for Advanced Studies, Germany

> \*Correspondence: Susanna Esposito

susanna.esposito@unimi.it

### Specialty section:

This article was submitted to Infectious Diseases - Surveillance, Prevention and Treatment, a section of the journal Frontiers in Medicine

> Received: 14 February 2019 Accepted: 02 May 2019 Published: 28 May 2019

### Citation:

Principi N, Camilloni B, Alunno A, Polinori I, Argentiero A and Esposito S (2019) Drugs for Influenza Treatment: Is There Significant News? Front. Med. 6:109. doi: 10.3389/fmed.2019.00109 Keywords: antiviral drugs, baloxavir marboxil, favipiravir, influenza, monoclonal antibodies, neuraminidase inhibitors

### INTRODUCTION

Vaccines represent the best way to reduce the influenza impact (1–3). However, influenza viruses vary continuously through antigenic drifts and occasional antigenic shifts that help the virus evade pre-existing immunity (4). Therefore, continuous reformulation of vaccine compositions and annual immunization are needed. In the case of viral shifts, licensed vaccines are completely ineffective, and ad hoc vaccine preparations are generally available only several weeks after the emergence and spread of a pandemic influenza virus (5). Finally, the immune responses induced by the influenza vaccines are suboptimal in a number of subjects, especially in younger children and the elderly, who are at risk of severe influenza, which further reduces the protection offered by influenza vaccination (6).

In addition to the intrinsic limitations of influenza vaccines, a second problem limits the vaccineinduced prevention of influenza. Universal immunization against influenza in pediatric age is recommended only in a minority of countries (2). Healthy children and adults frequently are not included in the list of patients for whom official health authorities strongly suggest influenza Principi et al. Antiviral Drugs Against Influenza

immunization (7). Moreover, even when vaccines are recommended worldwide, for example, in the elderly, influenza vaccination coverage remains suboptimal (8, 9). The World Health Organization estimates that 5–10% of the global population suffers from influenza every year, 3–5 million people develop severe influenza and 290,000–650,000 people die (10); thus, developing safe and effective alternatives for prophylaxis and treatment is critical.

In this paper, the clinical roles of antiviral drugs against influenza that have been licensed in at least one country will be discussed. Additionally, the potential roles of the anti-influenza compounds in development are evaluated.

### CURRENTLY LICENSED ANTI-INFLUENZA DRUGS

### Traditional Anti-influenza Virus Drugs

Antiviral drugs have been developed for a long time in an attempt to overcome the abovementioned problems and reduce the influenza-related risks. For years, the adamantane derivatives rimantadine and amantadine and the neuraminidase inhibitors (NAIs) oseltamivir, zanamivir (used worldwide) and, more recently, laninamivir and peramivir (used first in Japan and subsequently in China, Japan, South Korea, and the USA) have been the only drugs licensed for influenza prevention and control. However, these drugs have differences in their pharmacokinetic characteristics, routes of administration and ages of the targeted patients (11).

Starting from the 2004–2005 influenza season, use of adamantane derivatives was no longer recommended, mainly due to the emergence of resistance in most circulating influenza viruses. However, their activity was limited to influenza A viruses, and they showed poor tolerability, which could be considered sufficient reasons per se to avoid prescription of these drugs (12). In practice, only NAIs have been prescribed for influenza prevention and treatment since that time period. The emergence of influenza virus strains resistant to NAIs has been reported. Resistance to oseltamivir emerged only during the 2007–2008 and 2008–2009 influenza seasons, with up to 90% of circulating strains exhibiting resistance to this NAI (13– 15). Fortunately, the influenza virus strains circulating during the 2009 pandemic and in the following years rarely contained the mutations in the neuraminidase viral surface glycoprotein that conferred resistance to oseltamivir. Localized clusters of oseltamivir-resistant influenza virus have been reported (16) and resistance to NAIs is increasing (17).

However, generally, an influenza virus resistant to oseltamivir is sensitive to the other NAIs, because cross resistance among oseltamivir and other NAIs has not been observed (18, 19). Patients with influenza due to an oseltamivir-resistant virus can be successfully treated with other NAIs, such as zanamivir (15). In patients undergoing treatment, the NAI-resistant viruses are found to be NA subtype–specific and drug-specific (16, 19, 20). These clinically-derived NAI-resistant variants of influenza A viruses of N1 NA subtype most frequently carry H274Y and N294S NA amino acid substitutions. Viruses of N2 NA subtype carry E119Vand R292K NA mutations, and NAI-resistant variants of influenza B viruses harbor R152K and D198N NA mutations. Another important point is that NAI-resistant viruses can emerge either under drugselection pressure or naturally in the course of influenza virus evolution (without drug intervention). The high prevalence of oseltamivir-resistant A/Brisbane/59/2007(H1N1)-like viruses (subclade 2B) carrying the NA H274Y–resistance marker was reported worldwide during the 2007–2009 influenza seasons (16, 19, 20). Epidemiologic studies did not show evidence of an association between the development of resistance and oseltamivir use. Thus, H274Y NA amino acid substitution occurred naturally, and viruses with this mutation acquired remarkable transmissibility and superior fitness compared to their drug-sensitive counterparts (19, 20).

Moreover, all NAIs have been found to be safe and well tolerated. In particular, oseltamivir, which is the most frequently used NAI, can be administered not only to otherwise healthy adults and subjects with severe underlying disease but also to neonates, younger infants (18) and pregnant women (21) with age- and weight-appropriate doses without significant risks of severe adverse events.

Based on the results of several clinical trials, NAIs have been considered effective for the prevention and control of influenza infection (22, 23). Most available studies in this regard were carried out with oseltamivir and zanamivir, which were marketed before the other NAIs (24–28). Moreover, all of the data collected in younger children were derived from studies in which oseltamivir was used instead of zanamivir, which was licensed for use only in school-age patients. The effectiveness of NAIs has been documented in some meta-analyses, including those by Jefferson et al. (24) and Dobson et al. (25). A global evaluation of all of the studies analyzed in these meta-analyses suggests the conclusion that NAIs can limit the severity and duration of influenza in patients of any age with uncomplicated disease. However, the efficacy is time-dependent, because it is seen mainly in subjects who receive these drugs within the first 48 h of symptom onset. Jefferson et al. (24) and Dobson (25) calculated that administration of oseltamivir in adults with uncomplicated influenza decreased the time to first alleviation of symptoms of influenza-like illness (ILI) by 16.8 h [95% confidence interval [CI], 8.4–25.1] and 17.8 h (95% CI, 27.1 to −9.3), respectively. When only laboratory-confirmed influenza cases were considered, the advantage was slightly higher, because the time to perceived benefits was decreased by 25.2 h (95% CI, 16.0–36.2) (25). Similar results were reported for zanamivir, with the difference in time to alleviation of ILI symptoms reported as 14.4 h (95% CI, 9.36–19.44). No substantial differences were noted in the results obtained in children treated with oseltamivir (24). Regarding prophylaxis, it was calculated that oseltamivir could reduce the absolute risk for laboratory-confirmed influenza among community members and nursing home residents by 3.05% [relative risk [RR] 0.45; 95% CI, 0.30–0.67] and in a household setting by 13.6% (RR 0.20; 95% CI, 0.09–0.44) (21).

However, the available data do not permit firm conclusions regarding the effect of NAIs on severe cases (26–28). Although administration of NAIs was associated with a reduction of mortality during the recent 2009 pandemic (29), the impact on the development of serious influenza complications, hospitalization and mortality was not definitively demonstrated. The results of studies specifically planned to evaluate these problems were conflicting (30). A good example in this regard is given by studies that have measured the prevention of acute otitis media (AOM), which is a common complication of influenza in children. Jefferson et al. (24) did not find any statistically significant effect of oseltamivir on this disease (RR 0.8; 95% CI, 0.62–1.02), whereas Wang et al. (31) reported a statistically significant reduction in the AOM incidence with oseltamivir treatment in children 1–12 years of age (risk difference [RD] −0.09; 95% CI, −0.16 to −0.03). Moreover, in some cases, such as the meta-analysis by Muthuri et al. (32), NAI treatment was shown to be associated with an increased risk of severe outcomes, including pneumonia (OR 2.29; 95% CI, 1.16–4.53). Because laninamivir (33) and peramivir (34) do not seem to have substantially different efficacies compared to that of oseltamivir, these findings explain why the present possibility of treating influenza with NAIs cannot be considered completely satisfactory and the development of new drugs against influenza viruses has been deemed mandatory.

Several new approaches have been attempted to achieve influenza prevention and control. Antibodies against the highly conserved stem region of the haemagglutinin (HA) molecule of influenza A viruses and drugs that target different stages of the influenza virus life cycle than NAIs in human cells have been developed and tested (35). Among these preparations, baloxavir marboxil (BAM) and favipiravir (FP) are the only drugs that have reached the market (the first in Japan and the USA and the second only in Japan).

### RECENTLY-LICENSED ANTI-INFLUENZA VIRUS DRUGS

### Baloxavir Marboxil (BAM)

BAM was licensed in 2018 in Japan and the USA for the treatment of uncomplicated influenza in subjects aged ≥ 12 years with influenza clinical manifestations for ≤48 h (36). BAM is a prodrug that is given by mouth and is hydrolysed in the intestine, blood, and liver to form baloxavir acid (BXA), which is the active compound (37). The drug acts by inhibiting the cap-dependent endonuclease activity of the influenza A and B virus polymerase acidic protein (PA) to prevent the so-called cap snatching (i.e., the mechanism used by viruses to deviate the host mRNA transcription system and allow synthesis of viral RNAs) (37). In practice, BMA, in contrast to NAIs that reduce viral release from infected cells (38), inhibits viral replication. In vitro studies have shown that this activity is exerted without cytotoxicity even in cells infected with NAI-resistant influenza viruses. Moreover, BMA was shown to be effective not only against influenza viruses that usually infected humans but also against avian subtypes (39).

Phase 1 clinical trials were carried out in a total of 55 healthy subjects aged 20–59 years and were mainly directed to evaluate the safety, tolerability, pharmacokinetics and food effects (40). BAM was well tolerated, as no serious adverse events or deaths were reported. Moreover, treatment-emergent adverse events, such as headache, stomatitis, an increased eosinophil count, and elevation of alanine aminotransferase (ALT) and aspartate aminotransferase (AST), emerged in a very small number of subjects, were mild and resolved spontaneously within a few days. Only one subject had an increased lactate dehydrogenase (LDH) serum concentration that did not disappear during the follow-up period (22 days). Pharmacokinetic data confirmed that the drug could be administered as a single dose because its half-life was very long (more than 70 h). Moreover, even the lowest dose of the drug used in these trials (6 mg of BAM) was associated with high BXA serum concentrations 24 h after the single dose. Finally, food could influence drug metabolism, since BXA concentrations were lower in subjects who were fed or when measured before meal states. However, the levels remained high enough to ensure significant viral inhibition (40).

Phase 2 and 3 studies involving subjects aged 12–64 years confirmed the efficacy and safety of BAM (41). Adverse events, such as diarrhea, bronchitis, nausea, common cold symptoms (nasopharyngitis), and headache, were reported in ∼2% of patients and were not considered severe. Moreover, in otherwise healthy subjects with fever and mild to moderate uncomplicated respiratory disease for <48 h, administration of the drug could lead to a reduction of the mean time to alleviation of symptoms of ∼50–65 h compared to the 77.7–80 h required for patients who received a placebo. However, the clinical efficacy was not different from that evidenced in patients receiving oseltamivir, although BAM administration was associated with a more rapid decline in the viral load and a shorter duration of infectious virus detection (41). Similar results were obtained with a doubleblind, randomized, placebo-, and oseltamivir-controlled trial carried out in patients >12 years old suffering from influenza for <48 h with an underlying disease that was considered a risk factor for influenza-related complications. In this study, a shorter time to improvement of symptoms was also evidenced in the BAM group than in the placebo group (median 73.2 h vs. 102.3 h, p < 0.0001), whereas the efficacy of oseltamivir was quite similar (81.0 h, p = 0.8347). However, when the efficacy against the different types of influenza viruses was tested, BAM and oseltamivir demonstrated similar efficacies against the A/H3N2 virus, but the new drug was superior to oseltamivir against B viruses (time to symptom improvement 74.6 h vs. 101.6 h; p = 0.0251) (41). Moreover, despite the presence of severe underlying diseases, adverse events remained uncommon and did not differ from those found in patients receiving placebo or oseltamivir.

BAM has at least two important advantages over other available anti-influenza virus drugs. It is administered as a single dose by mouth, and it can overcome the problem of influenza viruses resistant to NAIs (39). Oseltamivir is also given by mouth, but to be effective it needs 5 days of administration, which can reduce compliance. Peramivir is given as a single dose by the intravenous route, requires administration by a nurse or a doctor and is more expensive than BAM. Zanamivir and laninamivir are administered by inhalation and are difficult to use in younger children.

However, despite obtaining a Food and Drug Administration (FDA) license, some unsolved problems seem to indicate that the use of BAM in clinical practice as an alternative to NAIs, particularly oseltamivir, for treatment of uncomplicated influenza cases can be debated. First, BAM is more expensive than oseltamivir. The cost of treatment with BAM has been calculated to be ∼3 times higher than that of treatment with oseltamivir (42). Moreover, the epidemiological and clinical importance of resistance to BAM is not precisely defined. Before obtaining the license, almost all isolated influenza virus strains were susceptible to BAM. A study carried out in the USA showed that the frequency of genetic mutations (amino acid 38 substitutions for isoleucine in the endonuclease domain of the viral RNA polymerase PA subunit) that caused reduced susceptibility of influenza viruses to BAM was very low during the 2016/2017 and 2017/2018 seasons (43). Moreover, no mutated virus was detected in 2018/2019 before BAM was licensed (43). However, treatment was associated with a rapid and substantial emergence of resistant strains. Some of the studies used to obtain the license evidenced that mutated viruses could be identified in 2.2–9.7% of cases (41). Resistance to BAM was also evidenced in Japan after its introduction in clinical practice. A pediatric study showed that ∼20% of treated children developed mutations (44). Recently, this problem was confirmed in strains isolated from adult patients (45). Although the mutated viruses remained sensitive to NAIs and had an impaired replicative capacity in vitro (44), patients with mutations shed the viruses and remained symptomatic for a longer period than those without the mutations (41), suggesting a potential negative effect of resistance to BAM.

These findings seem to suggest that emergence of resistance to BAM should be strictly monitored and that the clinical effects of BAM should be more carefully evaluated. Because definitive data in this regard are lacking, the routine use of BAM for treatment of uncomplicated influenza in adults seems premature. On the other hand, development of BAM is not complete. BAM is not licensed for complicated influenza cases. Moreover, data from children, pregnant women and patients at risk are rare or completely lacking, as are data concerning influenza prophylaxis.

### Favipiravir (FP)

FP is a prodrug. FP is a nucleoside analog that after oral ingestion requires intracellular phosphoribosylation to be transformed into its active form, FP ribofuranosyl-5′ -triphosphate (FRTP). Similar to BAM, FRTP acts by reducing viral replication through inhibition of the RNA-dependent RNA polymerase (RdRp) of RNA viruses (46). However, its activity is not limited to influenza viruses but instead is extended to several other RNA viruses, including arenaviruses, phleboviruses, hantaviruses, flaviviruses, enteroviruses, respiratory syncytial virus, and noroviruses (47). In vitro studies have shown that FP is effective against all of the influenza virus subtypes, including the avian viruses and those that are poorly sensitive or insensitive to NAIs (48–50). In experimental animals, FP was found to be significantly more effective than oseltamivir in preventing influenza or reducing its severity in mice infected with lethal doses of influenza viruses, even when the treatment was started 72 h after infection (51). Moreover, when used in combination with oseltamivir, protection due to the NAI was increased and the treatment efficacy window was extended to 96 h after symptom onset (52).

The safety and tolerability of FP were found to be good, although some data regarding the potential teratogenicity of FP were collected in all animal species assessed (53). Interestingly, the exposure causing teratogenicity was quite similar to that found in humans receiving the dosage of the drug that was useful for influenza treatment. Finally, although in vitro studies have shown that viral mutations associated with non-viable viral phenotypes rapidly develop in influenza viruses exposed to this drug, in vivo use of FP is only rarely associated with influenza virus mutations (54).

Some phase 2 and 3 studies in adults have been carried out in Japan, the USA and Europe (NCT01068912, NCT02026349, NCT02008344, NCT01728753, and NCT03394209). Generally, the drug was administered by mouth two times per day for 5 days even when different dosages were used. Most studies were completed several months ago, but the results are not available with the exception of those from NCT01068912, which aimed to identify the most effective and best tolerated drug dosage. The reasons for the unavailability of the study results are unknown. However, FP is licensed for use in humans in Japan. Its use is authorized only in patients infected by a novel or re-emerging influenza virus (i.e., a pandemic virus?) that is resistant to all other available influenza drugs. To avoid irrational prescriptions, FP is distributed only upon request by the Minister of Health, Labor and Welfare of Japan. Moreover, the guidelines clearly establish that pregnant women are excluded from use of FP, and females of childbearing potential have to avoid pregnancy in the 7 days following drug use to obtain the complete elimination of the FP concentration from the plasma (55).

# ANTIVIRAL PREPARATIONS IN ADVANCED STAGES OF DEVELOPMENT

### Pimodivir

Pimodivir is an oral drug that inhibits the polymerase basic protein 2 (PB2) subunit of the influenza A virus polymerase complex. Consequently, it inhibits viral replication (56). In vitro and experimental animal studies have shown that pimodivir is effective against influenza A virus, even when strains resistant to NAIs are tested. Compared to oseltamivir, it was more effective in improving body weight and reducing the severity of lung infection. However, its activity can significantly vary according to the influenza A virus subtype, probably as a consequence of differences in the PB2 structure among various A viruses (57). Positive results were also obtained when the drug was tested in humans with uncomplicated influenza, for whom pimodivir reduced viral shedding and influenza symptoms. Generally, adverse events (mainly diarrhea) were mild and spontaneously resolved within a few days (58). These findings were confirmed by a double-blinded phase 2b study involving adults with uncomplicated influenza A in whom different pimodivir dosages were given 2 times per day for 5 days alone or in combination with oseltamivir (59). This study showed that 600 mg of the drug per dose was adequate to obtain the best results. In the TABLE 1 | Antiviral drugs against influenza available on the market.


treated groups, both the viral load titer and the time to symptom resolution were reduced compared to those of the placebo group. Moreover, combined therapy was more effective than administration of pimodivir alone (59, 60).

However, the use of pimodivir has been associated with the emergence of mutations leading to viruses with a several fold decrease in susceptibility to the drug. In particular, emergence of PB2 substitutions or phenotypic resistance to pimodivir was evidenced in 11 of 172 patients (6.4%) (59, 60). The mutations included S324K/N/R, F325L, S337P, K376N/R, T378S, and N510K (59, 60). Although the mutated viruses were not associated with a deterioration of clinical conditions and the patients did not shed the virus after treatment, this finding deserves attention.

### Monoclonal Antibodies

Studies of natural immune responses to influenza virus infection evidenced that N-linked glycosylation sites in the haemagglutinin stem region of influenza A viruses were relatively well conserved and that antibodies against these sites were effective against a large number of influenza viruses (61–64). Starting from this evidence, a number of monoclonal antibodies targeting the HA conserved regions was developed and tested for the treatment of influenza A.

Preparations identified as MHAA4549A, MEDI8852, and VIS 410 have reached the phase II stage of development. All of these preparations were found to be safe and well tolerated and were able to reduce the peak viral load, duration of viral shedding, and influenza symptom scores compared to those of the placebo group (65–67). Moreover, no mutated virus was evidenced after monoclonal antibody administration (65–67). Compared to those of oseltamivir, all of these preparations have superior pharmacokinetic properties and a longer therapeutic window. The half-life is ∼3 weeks, one dose can be sufficient to control infection and efficacy can be obtained even if administration occurs after 48 h from symptom onset (65, 68, 69). However, two clinical trials (NCT02293863 and NCT02603952) in which the efficacy of oseltamivir alone was compared to that of the combination oseltamivir/MHAA4549A or MEDI8852 in adults with influenza A infection revealed that addition of the monoclonal antibody did not add any significant clinical advantage to those offered by the old NAI. When these results are considered together with the fact that the activity is limited to influenza A viruses, this finding seems to be a relevant limitation that may preclude extensive use of monoclonal antibodies. However, a definitive conclusion in this regard can only be drawn when studies in patients with severe influenza illnesses and those at high risk, including children and pregnant women, are performed.

### CONCLUSIONS

**Table 1** summarizes antiviral drugs against influenza available on the market. Several attempts have been made to develop new agents against influenza viruses that are able to overcome the supposed or demonstrated limitations of NAIs. These drugs, mainly oseltamivir, are discussed in terms of the risk of emergence of resistant strains and doubts regarding their efficacy in severe influenza cases and prevention of bacterial complications or death. Some of the new drugs have been licensed or are in the advanced stages of development. However, none of them has been completely developed, and the lack of data regarding patients for whom oseltamivir is recommended is discussed. Moreover, in some cases, use of the new drugs has been associated with emergence of resistance. In practice, none of these new drugs and monoclonal antibodies that are in development has adequate characteristics to substitute for NAIs at present. However, although NAIs (especially oseltamivir due to its greater ease of administration) remain the drug of choice for influenza treatment, their overuse has to be avoided. Accurate selection of patients for whom treatment is truly needed is necessary. Finally, the limitations of antivirals can be overcome with more extensive use of influenza vaccines.

### AUTHOR CONTRIBUTIONS

NP wrote the first draft of the manuscript. BC and AAl gave scientific contributions. IP and AAr performed the literature review. SE co-wrote the manuscript and supervised all activities. All authors approved the final submitted version of the manuscript.

# FUNDING

This work has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No. 701088.

### REFERENCES


**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 © 2019 Principi, Camilloni, Alunno, Polinori, Argentiero and Esposito. 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.

# Adaptive Evolution of Human-Isolated H5Nx Avian Influenza A Viruses

Fucheng Guo<sup>1</sup> , Yiliang Li<sup>1</sup> , Shu Yu<sup>1</sup> , Lu Liu<sup>2</sup> , Tingting Luo<sup>1</sup> , Zhiqing Pu<sup>1</sup> , Dan Xiang<sup>2</sup> , Xuejuan Shen<sup>1</sup> , David M. Irwin3,4, Ming Liao1,5 and Yongyi Shen1,2,5 \*

<sup>1</sup> College of Veterinary Medicine, South China Agricultural University, Guangzhou, China, <sup>2</sup> Joint Influenza Research Centre (SUMC/HKU), Shantou University Medical College, Shantou, China, <sup>3</sup> Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, ON, Canada, <sup>4</sup> Banting and Best Diabetes Centre, University of Toronto, Toronto, ON, Canada, <sup>5</sup> Key Laboratory of Zoonosis Prevention and Control of Guangdong Province, Guangzhou, China

Avian influenza A viruses (AIVs) H5N1, first identified in 1996, are highly pathogenic in domestic poultry and continue to occasionally infect humans. In this study, we sought to identify genetic changes that occurred during their multiple invasions to humans. We evaluated all available H5Nx AIV genomes. Significant signals of positive selection were detected in 29 host-shift branches. 126 parallel evolution sites were detected on these branches, including 17 well-known sites (such as T271A, A274T, T339M, Q591K, E627K, and D701N in PB2; A134V, D154N, S223N, and R497K in HA) that play roles in allowing AIVs to cross species barriers. Our study suggests that during human infections, H5Nx viruses have experienced adaptive evolution (positive selection and convergent evolution) that allowed them to adapt to their new host environments. Analyses of adaptive evolution should be useful in identifying candidate sites that play roles in human infections, which can be tested by functional experiments.

### Edited by:

Ville Peltola, Turku University Hospital, Finland

### Reviewed by:

Ming Xiu Peng, Zhejiang University, China Ahmed Mostafa, National Research Centre, Egypt

> \*Correspondence: Yongyi Shen shenyy@scau.edu.cn

### Specialty section:

This article was submitted to Virology, a section of the journal Frontiers in Microbiology

Received: 01 February 2019 Accepted: 28 May 2019 Published: 12 June 2019

### Citation:

Guo F, Li Y, Yu S, Liu L, Luo T, Pu Z, Xiang D, Shen X, Irwin DM, Liao M and Shen Y (2019) Adaptive Evolution of Human-Isolated H5Nx Avian Influenza A Viruses. Front. Microbiol. 10:1328. doi: 10.3389/fmicb.2019.01328 Keywords: H5Nx, Avian influenza A virus, convergent evolution, human infection, host shift

### INTRODUCTION

Avian influenza viruses (AIVs) pose a continuous threat to public health due to their pandemic potential (Widdowson et al., 2017). The H5N1 subtype of highly pathogenic avian virus (HPAIV) is highly pathogenic in domestic poultry, since its initial detection in China in 1996 (Peiris et al., 2007). Since its first discovery in humans during 1997 in Hong Kong (Yuen et al., 1998), H5N1 HPAIVs continue to result in occasional human infections with an high fatality rate of more than 60% (Yu et al., 2008; Cowling et al., 2013). H5 viruses evolved into different clades, and have reassorted with different NA subtypes, including N1, N2, N3, N5, N6, and N8, resulting in outbreaks in poultry and lethal human infections (Smith and Donis, 2015). H5 is the predominate AIV subtype that infect human populations, and thus, pose great threat to public health (Peiris et al., 2007).

Human infections of H5Nx are mainly through direct avian-to-human transmission. The human-isolated H5Nx viruses are distributed into different phylogenetic clades, that is, these hostshift events occurred independently multiple times. During the adaptation of an organism to a new environment, adaptive evolution occurs. When similar morphological or physiological changes

are observed on multiple evolutionary lineages, convergent or parallel amino acid changes in key genes occur (Zhang, 2006; Shen et al., 2012). Viruses face great challenges when they emerge in new hosts. Our previous study showed that during the multiple invasions of humans by H7N9 AIVs, convergent evolution occurred to allow these human-isolated viruses to adapt their new hosts (Xiang et al., 2018).

Although H5Nx viruses do not have the ability to be transmitted efficiently in a sustained manner from personto-person, these HPAIVs, which are panzootic in poultry, continue to spread, and their interspecific transmission poses a major challenge to human health. In this study, we conducted a comprehensive evolutionary analysis of H5Nx viruses by collecting all available sequence data to examine molecular mechanisms used by H5Nx viruses to frequently infect humans.

# MATERIALS AND METHODS

### Data Source and Preliminary Treatment

All available sequences of H5Nx viruses were downloaded from three databases: the Influenza Virus Resource at the National Center for Biotechnology Information (NCBI)<sup>1</sup> , the Global Initiative on Sharing Avian Influenza Data<sup>2</sup> , and the Influenza Research Database (IRD)<sup>3</sup> . Redundant sequences, laboratory strains and short (<80% of the corresponding gene) sequences were removed. Sequences from egg isolations from human hosts were excluded, as these sequences might carry additional in vitro adaptive mutations (Bush et al., 2000). Our final dataset contains 9945, 6719, 6845, 7966, 6454, 6401, 6466, and 6423 HA, NS, M, NA, NP, PA, PB1, and PB2 sequences, respectively (**Supplementary Table 1**). The sequences in each dataset were aligned by MAFFT v7.221, separately (Katoh and Toh, 2010). Initial phylogenetic trees for the eight genes were constructed separately, using the maximum likelihood method RAxML v.8.0.14 (Stamatakis, 2006). Best-fit evolutionary models for the sequences in each datasets were identified using ModelTest (Posada and Crandall, 1998).

### Selection Analyses

The CODEML program in the PAML package (Yang, 2007) was used to identify signals of potential positive selection. The branch-site model, which was used to determine whether a gene had undergone positive selection on a foreground branch, was used to assess selective pressure. Bayes Empirical Bayes (BEB) analysis was used to calculate the Bayesian posterior probability of any positively selected site or branch. Finally, LRT statistics were calculated between the branch-site model and the branchsite model with fixed ω<sup>0</sup> = 1. The significance of the difference between the models was determined using twice the difference in the log-likelihood values of LRTs (21lnL) between the two models, which follows a chi-squared (χ2) distribution with degrees of freedom equaling the difference in the number of parameter estimated (Zhang et al., 2005).

### Convergent Evolution Analyses

Ancestral amino acid sequences for target nodes of each dataset were inferred using PAML4.0 (Yang, 2007). The statistical significance of the number of convergent/parallel evolutionary substitutions between pairs of branches was tested using the method of Zhang and Kumar (1997). Candidate substitutions were defined if (i) the topology of each lineage consisting of human isolate and its genetically related isolates had high bootstrap support values (≥90), and (ii) the posterior probabilities of the character states at each ancestral node was ≥0.90. The corresponding sites in HA protein were mapped onto a published three-dimensional (3-D) structure of A/duck/Egypt/10185SS/2010 (H5N1) virus (Protein Data Bank code: 5E2Y) using PyMOL (Molecular Graphics System, version 2.0.7.0 Schrödinger, LLC, accessed on 19-Jan-2018)<sup>4</sup> (Delano, 2002).

# RESULTS

### Phylogenetic Analyses

The HA phylogeny reconstructed using RAxML v.8.0.14 (Stamatakis, 2006) revealed that the H5 sequences are grouped into 10 clades (clades 0–9), and that the humanisolated sequences distribute to clades 0 (16 human-isolated sequences), 1 (101 human-isolated sequences), 2 (360 humanisolated sequences), 3 (one human-isolated sequences), and 7 (two human-isolated sequences) (**Figure 1**, **Supplementary Figures 1**, **2**, and **Supplementary Table 2**). Similarly, phylogenetic trees were reconstructed, separately, for other genes.

In order to simplify the calculations that focused on the human-isolated viruses, we divided the HA, NS, M, NA, NP, PA, PB1, and PB2 sequences into 132, 101, 98, 114, 87, 80, 90, and 92 datasets, respectively, based on the initial phylogenetic trees. Each dataset contains the human isolates and their closely related avian isolates (**Supplementary Figures 3**–**10**). These HA, NS, M, NA, NP, PA, PB1, and PB2 gene datasets contained 266, 147, 206, 260, 186, 155, 164, and 171 host-shift branches, respectively.

### Positive Selection on Host-Shift Branches

We used the CODEML program from the PAML package (Yang, 2007) to identify signals of positive selection on host-shift (avianto-human) branches. In total, 29 branches with 38 sites (H5 numbering) were identified as having experienced significant positive selection, including branches HA-107b, HA-18c (473R), HA-6a, HA-64b (11N, 15Q, 20M, 314K, 315T, 522T, 529L, 546L, 547Q, and 548C), HA-68a (212R and 500R), HA-72b, HA-74a, HA-75a, HA-76a, HA-77a, HA-83a, and HA-107b in HA; PB2- 14d and PB2-74b in PB2; MP-46e, MP-50a (5T, 6E, 7V, 8E, 257T, 258E, 259V, and 260E), and MP-85a (277P, 279V, 282A, 283N, 284I, 285I, 287I, 292L, 328Y, 330Q, 336V, 339D, 340D,

<sup>1</sup>www.ncbi.nlm.nih.gov/genomes/FLU

<sup>2</sup>www.gisaid.org

<sup>3</sup>www.fludb.org/brc/home.spg?decorator=influenza

<sup>4</sup>http://www.pymol.org/

and 344V) in MP; NA1-15b (188N) in NA1, NA6-2b in NA6; NP-32a, NP-65a (486S and 487Y) and NP-66c in NP; PA-11a, PA-25d, and PA-72b in PA; NS-31a, NS-72c, and NS-89a in NS, and PB1-31b in PB1 (**Supplementary Table 3**). Other host-shift branches did not show significant signals of positive selection (**Supplementary Table 4**).

### Convergent/Parallel Evolution of Human-Isolated H5Nx Viruses

To determine whether convergent/parallel evolution occurred during the multiple avian-to-human transmissions of H5Nx HPAIVs, ancestral amino acid sequences for the target nodes that lead to host-shift branches were reconstructed for convergent evolutionary analyses. In total, we identified 126 parallel evolution substitutions (34 in HA, 20 sites in PB2, seven in PB1, 13 in PA, eight in NP, 20 in NA, six in MP, and 18 in NS) that occurred on the host-shift branches (**Table 1**).

For the HA gene, the parallel amino acid substitutions S223N (H5 numbering) occurred on seven host-shift branches, A134V on four branches, and six mutations (D31N, D154N, T/N195I, V219I, I375M, and E502G, H5 numbering) were each discovered on three branches. The remaining 26 parallel mutations occurred each on pairs of branches. Of these convergent/parallel amino acid mutations, four substitutions (A134V, D154N, S223N, and R497K) had been identified in earlier studies as having functional roles and were mapped to the three-dimensional (3-D) structure of the HA protein (**Figure 2**).

For the PB2 gene, the well-known mutations E627K and D701N (Hatta et al., 2001; Li et al., 2005; Song et al., 2014; Zhu et al., 2015) were observed on 30 and 10 host-shift branches, respectively. N456D and G727R were detected on five and four branches, respectively. An additional four substitutions (T339M, I451V, S471F, and R369K) occur in parallel on three sets of branches and the remaining 12 parallel substitutions were shared by pairs of host-shift branches.

For the PB1 gene, there were seven parallel-evolved variations (I57T, E172D, M179I, S361G, N375S, K387R, and L598P) that occur on pairs of host-shift branches.

For the PA gene, three mutations K142E, N321K, and K615R were each detected on three host-shift branches and ten mutations (F4C, M12I, M86V, T97I, F105L, L226F, E237K, P275L, T369A, and V387I) parallelly occurred on pairs of hostshift branches.

For the NA gene, 20 parallel-evolved variations in the NA1 subtype, including three mutations I/V16A, V33I, and I243V occurred on each of three host-shift branches, and 17 mutations (V16I, N39S, P45T, K55R, A58T, K241R, N305T, G318S, P323S, S364N, G365E, I380V, V404I, N430S, N430D,


TABLE 1 | Continued


(Continued)

(Continued)

### TABLE 1 | Continued

fmicb-10-01328 June 11, 2019 Time: 18:0 # 5


G435S, and T441D) were shared by pairs of branches. No parallel substitutions were detected in the other NA subtypes.

For the NP gene, two mutations A284T and V343I were shared by four and three human-isolated branches, respectively, and five mutations (R100I, R384K, S413L, P419S, R452K, and P453S) were shared by pairs of human-isolated branches.

For the MP gene, six mutations were detected as parallelevolved variations, including T137A, A239T, C269Y, V280I, S283N, and D340N. Each of these was separately shared by two human-isolated branches.

For the NS gene, 18 parallel-evolved variations were detected, 15 of which were in NS1 and three in NS2. Of the 15 NS1 mutations, four (R67Q, T81I, L212F, and L185F) were each shared by three branches, and 11 (G47S, N48S, R59H, E70K, R88C, V136A, I137V, D139N, S205N, D209N, and V209I) were identified in pairs of human-isolated branches. For the NEP/NS2 genes, three mutations M/A14V, A48T, and T/V115A were shared by pairs of branches.

### DISCUSSION

Since their first detection in 1996, H5 AIVs have had a substantial impact on veterinary and human health (Harfoot and Webby, 2017). Although the direct transmission of H5Nx viruses from

avian species to humans remains a relatively rare event, they still pose a serious pandemic threat due to their high virulence and mortality, and their increasingly expanding host range, as well as the significant ongoing evolution toward efficient transmission in mammals (Guan and Smith, 2013). Sporadic human infections continue to occur in countries where H5Nx have become endemic in birds, providing a persistent threat to global health due to the possibility of virus adaptation to humans. Thus, the study of their genetic mechanisms of human adaptation remains essential.

Adaptive evolution of functional important genes is essential for the invasion of new niches (Zhang and Kumar, 1997; Zhang, 2006; Shen et al., 2012). AIVs should face great challenges when they emerge in humans from their avian sources. During the host-shift process of the multiple human H7N9 AIV invasions, convergent evolution occurred for the human-isolated viruses to adapt to their new hosts (Xiang et al., 2018). Similar to H7N9, human-isolated H5 sequences are distributed in multiple phylogenetic positions (**Figure 1** and **Supplementary Figures 1**, **2**), suggesting that there were multiple independent invasions of H5Nx AIVs into humans. Here, we found that adaptive evolution (positive selection, and convergent/parallel evolution) occurred on the independent host-shift branches in H5Nx AIVs, and that some of the adaptive sites have functional importance (**Table 1**).

Avian influenza viruses preferentially bind α2-3 sialic acid receptors. In contrast, human-adapted influenza viruses preferentially bind α2-6 sialic acid receptors. The switch of preference, from avian to human type sialic acid receptors, is considered to be a key element necessary for AIVs to cause human pandemics (Matrosovich et al., 2000; Parrish and Kawaoka, 2005). The receptor-binding domain (RBD) of HA is formed by four loops and one helix in the RBS, which contact the base and potentially the extension region of the human receptor. Among the 34 parallel amino acid substitutions detected in the HA gene (**Table 1**), two of them A134V and S223N, have previously been shown to associate with the switch of preference from avian to human type sialic acid receptors (Yamada et al., 2006; Auewarakul et al., 2007; Imai et al., 2010; Chen et al., 2012). In addition, three others, S141P (130-loop), R162I (150 loop), and T/N195A/I(190-loop), are in or around the RBD and may dramatically alter the receptor binding preference of the H5Nx influenza viruses. Furthermore, S223N has previously been detected by in silico prediction and experimentally confirmed to enhance human receptor specificity of H5N1 influenza A viruses (Schmier et al., 2015), indicating that mutations detected by several methods might have greater potential functional effects and that the combination of multiple methods is recommended for selecting potential functional mutations for experimental studies. S223N is located close to the sialic-acid-binding site (**Figure 2**), and this mutation in H5N1 AIVs has weakened affinity toward α-2, 3 and an increased affinity toward α-2, 6 sialic acid receptors (Yamada et al., 2006; Chen et al., 2012). Alanine at position 134 (A134) is in the 130-loop of the receptor binding domain (**Figure 2**). This site is highly conserved in avian H5N1 viruses. The Ala to Val substitution at position 134 could change the receptor-binding preference of H5-HA from α-2,3 to both α-2,3 and α-2,6-sialic acid binding (Auewarakul et al., 2007; Imai et al., 2010). To initiate influenza virus infection, hemagglutinin (HA), which is the major surface glycoprotein of influenza viruses, binds to the host cell surface complex glycans via a terminal sialic acid. The preference of HA for particular sialic acid moieties on host cells is a key determinant of host range and tissue tropism (Matrosovich et al., 2000). Parallel evolution of amino acid sites that play important roles in the change of receptor-binding preference in host-shift branches suggests that during human invasion by AIVs that they have adapted to the common challenge of a difference in the surface glycoprotein between birds and humans.

After binding and entering human cells, efficient replication (efficient production of new viruses) is a critical factor that influences viral infection. A series of parallel evolution sites with roles in increasing polymerase activity and replication efficiency in mammals were identified, such as T271A (Finkelstein et al., 2007), A274T (Leung et al., 2010), Q591K (Yamada et al., 2010), E627K (Hatta et al., 2001; Song et al., 2014), and D701N (Li et al., 2005; Zhu et al., 2015), in PB2 (**Table 1**). The E627K and D701N mutations might be especially important as 17.5% (30 of 171) and 5.85% (10 of 171) of the host-shift lineages in the PB2 gene share these two mutations. In addition, some parallel sites in other genes are also suggested to be associated with enhanced polymerase activity and/or increased virulence in mammals (**Table 1**), such as L598P in PB1 (Xu et al., 2012), T97I and N321K in PA (Cheng et al., 2014; Yu et al., 2014; Wu et al., 2016; Zhao et al., 2016; Nam et al., 2017), and A284T in NP (Zhao et al., 2016). Parallel evolution at these sites suggests that during the host-shift of H5Nx AIVs, these changes have allowed more efficient replication in human cells. Compared with the adaptive evolution of human-isolated H7N9 AIVs (Xiang et al., 2018), human-isolated H5Nx also have the E627K and D701N mutations in PB2 genes in the host-shift branches. This suggests that these two mutations play important roles in human adaptation of AIVs in both subtypes. Other adaptive sites are not shared by these two viruses, suggesting that there are some differences in the adaptation of H7N9 and H5Nx to humans.

Positive selection is the force that drives an increase in the prevalence of advantageous new mutations. In this study, 29 host-shift branches had significant signals of positive selection, with 12 in HA, two in PB2, three in MP, one in NA1, one in NA6, three in NP, three in PA, three in NS, and one in PB1. The frequency of positive selection in host-shift branches is quite low, indeed, only a few adaptive events were detected by the PAML package. This is an unavoidable limitation of the current approaches for positive selection analyses. A total of 38 positively selected amino acid sites were identified. The 340D mutation in the MP gene had significant signals in both the positive selection and the convergent evolution analyses, implying that this site might be important for adaptation to new host environments. More attention on this positive selection site, and its functional roles, is needed.

To determine whether the human-adaptive sites were actually part of human adaptation following zoonosis, we calculated the prevalence of the adaptive mutations (positive selection and convergent/parallel evolution sites) in the

human- and avian-isolated strains (**Supplementary Table 5**). It is expected that the adaptive-evolution substitutions should be more prevalent in the human viruses than in the avian viruses (Arai et al., 2016). For most of the adaptive mutations, their proportions in the human isolates are higher, with many tending to be fixed, than in the avian isolates, suggesting that these mutations have higher fitness in humans (**Supplementary Table 5**).

Host barriers restrict interspecies transmission of AIVs. Factors that contribute to AIVs infection and transmission in humans are complex. Although some amino acid changes associated with receptor affinity, temperature tolerance, viral replication, and mammalian adaptation have been found to play a role (Widdowson et al., 2017), the genetic basis for host shifts is not fully understood. In this study, we identified a series of adaptive changes at sites during the multiple human invasions by H5Nx viruses. Some of the adaptive mutations are known to have functional importance in cross-species transmission from avian to humans, while others are useful candidates for further experimental studies, especially those located in critical domains. Analyses of adaptive evolution should identify useful additional candidate sites that might play roles in human infections for functional studies.

# REFERENCES


### AUTHOR CONTRIBUTIONS

YS conceived, designed, and supervised the study. FG, YL, SY, LL, TL, ZP, DX, and XS collected and analyzed the data. YS and DI wrote the drafts of the manuscript. ML commented on and revised the drafts of the manuscript. All authors read and approved the final draft of the manuscript.

### FUNDING

This work was supported by the National Natural Science Foundation of China (31822056) and Guangdong Natural Science Funds for Distinguished Young Scholar (2014A030306046).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2019.01328/full#supplementary-material



virus infection in China. PLoS One 3:e2985. doi: 10.1371/journal.pone.000 2985


**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 © 2019 Guo, Li, Yu, Liu, Luo, Pu, Xiang, Shen, Irwin, Liao and Shen. 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.

# Bigger and Better? Representativeness of the Influenza A Surveillance Using One Consolidated Clinical Microbiology Laboratory Data Set as Compared to the Belgian Sentinel Network of Laboratories

### Edited by:

Kelvin To, The University of Hong Kong, Hong Kong

### Reviewed by:

Pedro Xavier-Elsas, Federal University of Rio de Janeiro, Brazil Anthony Tam, Queen Mary Hospital, Hong Kong

\*Correspondence:

Olivier Vandenberg olivier.vandenberg@ulb.ac.be

†The last three authors act as equivalent co-senior authors

### Specialty section:

This article was submitted to Infectious Diseases - Surveillance, Prevention and Treatment, a section of the journal Frontiers in Public Health

Received: 25 December 2018 Accepted: 23 May 2019 Published: 18 June 2019

### Citation:

Van den Wijngaert S, Bossuyt N, Ferns B, Busson L, Serrano G, Wautier M, Thomas I, Byott M, Dupont Y, Nastouli E, Hallin M, Kozlakidis Z and Vandenberg O (2019) Bigger and Better? Representativeness of the Influenza A Surveillance Using One Consolidated Clinical Microbiology Laboratory Data Set as Compared to the Belgian Sentinel Network of Laboratories. Front. Public Health 7:150. doi: 10.3389/fpubh.2019.00150 Sigi Van den Wijngaert <sup>1</sup> , Nathalie Bossuyt <sup>2</sup> , Bridget Ferns 3,4, Laurent Busson<sup>1</sup> , Gabriela Serrano<sup>5</sup> , Magali Wautier <sup>1</sup> , Isabelle Thomas <sup>6</sup> , Matthew Byott <sup>7</sup> , Yves Dupont <sup>2</sup> , Eleni Nastouli 3,8, Marie Hallin1†, Zisis Kozlakidis 7,9† and Olivier Vandenberg5,7,10 \* †

<sup>1</sup> Department of Microbiology, LHUB-ULB, Pole Hospitalier Universitaire de Bruxelles, Université Libre de Bruxelles, Brussels, Belgium, <sup>2</sup> Sciensano, SD Epidemiology and Surveillance, Service 'Epidemiology of Infectious Diseases', Brussels, Belgium, <sup>3</sup> Department of Clinical Virology, University College London Hospitals NHS Foundation Trust, London, United Kingdom, <sup>4</sup> UCLH/UCL Biomedical Research Centre, NIHR, London, United Kingdom, <sup>5</sup> Research Centre on Environmental and Occupational Health, School of Public Health, Université Libre de Bruxelles, Brussels, Belgium, <sup>6</sup> National Influenza Centre, Sciensano, Brussels, Belgium, <sup>7</sup> Division of Infection and Immunity, Faculty of Medical Sciences, University College London, London, United Kingdom, <sup>8</sup> Department of Population, Policy and Practice, Great Ormond Street Institute of Child Health, University College London, London, United Kingdom, <sup>9</sup> International Agency for Research on Cancer, World Health Organization, Lyon, France, <sup>10</sup> Innovation and Business Development Unit, LHUB-ULB, Pole Hospitalier Universitaire de Bruxelles, Université Libre de Bruxelles, Brussels, Belgium

Infectious diseases remain a serious public health concern globally, while the need for reliable and representative surveillance systems remains as acute as ever. The public health surveillance of infectious diseases uses reported positive results from sentinel clinical laboratories or laboratory networks, to survey the presence of specific microbial agents known to constitute a threat to public health in a given population. This monitoring activity is commonly based on a representative fraction of the microbiology laboratories nationally reporting to a single central reference point. However, in recent years a number of clinical microbiology laboratories (CML) have undergone a process of consolidation involving a shift toward laboratory amalgamation and closer real-time informational linkage. This report aims to investigate whether such merging activities might have a potential impact on infectious diseases surveillance. Influenza data was used from Belgian public health surveillance 2014–2017, to evaluate whether national infection trends could be estimated equally as effectively from only just one centralized CML serving the wider Brussels area (LHUB-ULB). The overall comparison reveals that there is a close correlation and representativeness of the LHUB-ULB data to the national and international data for the same time periods, both on epidemiological and molecular grounds. Notably, the effectiveness of the LHUB-ULB surveillance remains partially subject to local regional variations. A subset of the Influenza samples had their whole genome sequenced so that the observed epidemiological trends could be correlated to molecular observations from the same period, as an added-value proposition. These results illustrate that the real-time integration of high-throughput whole genome sequencing platforms available in consolidated CMLs into the public health surveillance system is not only credible but also advantageous to use for future surveillance and prediction purposes. This can be most effective when implemented for automatic detection systems that might include multiple layers of information and timely implementation of control strategies.

### Keywords: Belgium, influenza A, surveillance, clinical microbiology laboratory, sequencing

### INTRODUCTION

Infectious diseases remain a serious public health concern in industrialized and low- and middle-income countries. Due to their considerable effect on global human demographics (1, 2) and the economy (3, 4), the public health community has developed many surveillance strategies and systems to improve infectious disease surveillance around the world. Even if diagnostic and computer resources have expanded considerably, infectious disease surveillance remains challenging. The 2009 H1N1 influenza pandemic and the recent Ebola outbreak in West Africa are few examples showing that infectious diseases cannot easily be predicted and modeled reliably in real-time (5, 6). Therefore, the need for reliable and representative surveillance systems remains as acute as ever. One of the more established surveillance strategies, known as traditional public health surveillance of infectious diseases, is the use of reported positive results from sentinel clinical laboratories or laboratory networks to survey the presence of specific microbial agents known to constitute a threat to public health in a given population (7).

According to Colson et al. a total of 31 laboratory-based surveillance systems have been implemented in Europe for the purpose of global surveillance (8). Two such characteristic examples of surveillance systems implemented successfully using the above reporting strategy are the Health Protection Agency in England and Wales which counts infectious pathogens detected by hospital and specialist laboratories (9), and the surveillance systems of the Netherlands Reference Laboratory for Bacterial Meningitis (10).

In Belgium, such a strategy has been launched and implemented in 1983 by Sciensano, formerly known as Scientific Institute of Public Health (WIV-ISP), with the establishment of the sentinel laboratory network collecting information on the epidemiology of infectious diseases (11, 12). With 79–88 participating clinical microbiology laboratories (CMLs) across different geographical areas, the main objective of the Belgian Sentinel Network of Laboratories (BSNL) is to monitor the emergence and evolution of different infectious diseases over time, based on a representative fraction of the microbiology laboratories nationally (13, 14). With a ratio of 60% of participating laboratories to the total number of laboratories, the BSNL is able to describe trends and monitor changes in 12 groups of pathogens both at national and regional levels (15). However, in most European countries such laboratory surveillance is made on a voluntary base and is often not financially covered.

In parallel, the rationalization of public health costs has led to the development of novel strategies for laboratories' cost containment. In this perspective, a number of CMLs have undergone a process of consolidation involving a shift toward laboratory amalgamation and closer real-time informational linkage. Through this consolidation activity, an operational model has emerged with large centralized clinical laboratories performing on one central platform and one or several distal laboratories dealing locally only with urgent analyses. The increasing centralization of diagnostic services over a large geographical region has given rise to the concept of "microbiology laboratories network" (16). The reduction in the number of small clinical laboratories and the aggregation of the remaining ones, may condition the ability to detect epidemiological changes. The sensitivity and representativeness of national surveillance systems should be therefore carefully monitored using coverage measures which indicate the proportion of the target population included within the surveillance system (15).

It is conceivable that the consolidated CMLs could become a cornerstone of public health models in the near future akin to the regional healthcare hospital networks with interactive surveillance for AMR control in France and cross-border regions (17). Due to the adoption of a 24/7 working scheme and improved automation, consolidated CMLs are also able to analyse a large influx of samples in the context of an outbreak investigation. In addition to the volume capacities and the large range of diagnostic tools, the ability of consolidated CMLs to access multiple different partners, geographies and clinical specialities can enhance their capabilities to provide advanced systems for disease surveillance and early recognition.

As there is the potential that such CMLs merging might have a consequence on infectious diseases surveillance, we used the availability of Influenza data to evaluate whether influenza infection trends could be estimated effectively from only one CML serving the Brussels area. The obtained data was compared to the available laboratory surveillance data provided by the directorate Epidemiology and public health of Sciensano in Belgium for the same time period.

### MATERIALS AND METHODS

### Location

The study took place in the Department of Microbiology of the Laboratoire Hospitalier Universitaire de Bruxelles-University Laboratory of Brussels (LHUB-ULB) serving 5 University Hospitals located on three geographical poles i.e., Center, North and West of Brussels and representing close to 3,000 beds. Their catchment population covers a population of 700,000 inhabitants according to the Belgian statistical office, where the inhabitants of Brussels area are described for the year 2017 (18). Altogether, the LHUB-ULB performs annually more than 1,200,000 microbiology analysis including viral culture (12,000) and molecular diagnosis (26,000).

### Data Collected

In our routine surveillance perspective, all patients diagnosed with flu infection by either ImmunoChromatographic Test (ICT), Lateral Flow Chromatography (LFC), molecular diagnostic tests or by cell culture methods are considered as notifiable cases of flu infection. These laboratory diagnosed cases of flu infection are transferred on a weekly basis to the BSNL by mail. The encoded variables include the diagnosed infectious disease, some patients demographic data allowing the identification of duplicates i.e., date of birth (or previously age), gender, and postal code. In addition the specimen and its sample identification number, the diagnostic method and the date of diagnosis are recorded as well. To protect the patients' identity, all data transfers are anonymous.

To assess the representativeness of the LHUB-ULB flu surveillance, LHUB-UB's data collected from 1st January 2014 to 31st December 2017, were compared with all flu cases notification of the BSNL according to the geographical coverage and time distribution. In addition to BSNL epidemiological data's, weekly reports on the incidence of clinical influenza-like illness (ILI) and virological data collected by sentinel network of general practitioners (SNGPs) were also considered to better describe the epidemiological trends of flu at the regional and national level.

The SNGPs comprises about 120 general practices spread throughout Belgium who weekly report data about eight different health problems (infectious and non-infectious diseases). The coverage of the network is estimated at 1.1–1.5% of the Belgian population. Since 2007, the SNGPs has continuously recorded the general medicine consultations for ILI and acute respiratory tract infections. For each episode age group, vaccination status, outcome and hospitalization are recorded. In a subset of these patients, a clinical sample is collected and virologically tested by the NRC Influenza (https://www.sciensano.be/en/projects/ network-general-practitioners).

According to Flu News Europe, the influenza season spent from week 40 to 20 of the following year. The yearly epidemic threshold was derived from historical surveillance and laboratory data using the moving epidemic method (19).

Data from both sources were collected retrospectively and anonymized before analysis in a routine surveillance perspective. Ethics approval was granted by the Ethics Committee of the Saint-Pierre University Hospital. No written informed consent was collected.

To test the validity of LHUB-ULB data as a source for flu surveillance, the flu notification trends of the LHUB-ULB were compared with the trends of the overall BSLN network, BSLN network minus LHUB-ULB notification (BSLN-) and ILI consultation rate by Spearman's rank correlation coefficient TABLE 1 | Coverage of the LHUB-ULB notification by municipalities, surface and inhabitants.


Belgium area: 30,528 km<sup>2</sup> .

Belgian population: 11,776,158 inhabitants.

within each epidemic season and for the corresponding data over 4 years investigated at the national and regional level. All statistical tests were two-sided with the alpha set at 0.05. Analyses were done with the STATA/IC, version 14.

### Microbiology Methods

To better assess the input of the LHUB-ULB as early warning lab for molecular surveillance, the genomes of 39 randomly selected Influenza A(H1N1)pdm09 isolates obtained from patients attending the emergency room of Saint-Pierre University Hospital, located in Brussels (01/02/2016 to 15/03/2016), were whole genome sequenced and compared to equivalent Influenza A(H1N1)pdm09 sequences deposited at the Epiflu database for the same location and period and against France, Germany, Netherland and UK-derived sequences from the same time period (Table of all isolate IDs used in this study are included in the **Supplemental Digital Content 1**).

All samples were initially screened for influenza A virus by reverse transcription-PCR targeting the matrix gene. A total of 39 influenza A virus-positive samples had their whole genome sequenced. RNA was amplified using a modified eight-segment method. Library preparations were generated as previously described (20) and short reads were assembled de novo (21). A neighbor joining phylogenetic tree was constructed using the Multiple Sequence Alignment Software v.7 (MAFFT) (22).

# RESULTS

From 2014 to 2017, 31,809 flu cases were declared by the sentinel laboratories to the BSNL. Among them, 3,186 infections were reported by the LHUB-ULB representing 38.75% (2,746/7,088) and 10.1% (3,186/31,809) of all influenza cases reported by the BSNL in the Brussels region and at the national level, respectively. During the study period, the number of municipalities in which the LHUB-ULB represents at least 50% of coverage ranges from 21 (2017) to 32 (2016) representing 28.1% (21.6–36%) of the Belgium territory and 9.1% (8.1–10.4%) of the Belgium population (**Table 1**). Even if most of the municipalities are located in the Brussels region area; we note that each year the LHUB-ULB represents 100% of the notification of 5 to 10 municipalities located outside of the Brussels area either in Flanders or in Wallonia underscoring the variation in BSNL coverage at provincial level. In another hand, the LHUB-ULB's coverage was globally lower in municipalities located in Flanders, and for Walloon region in the provinces of Namur, Liege and Luxembourg. **Supplemental Digital Content 2** describes the geographical representativeness of the flu infection's notification by the LHUB-ULB in the BSNL notification for the year 2016.

The representativeness of the LHUB-ULB to describe the trends of the flu notification at the regional and national level is described in **Figure 1**. In order to better describe the flu trends, the incidence ILI based on data's provided by the network of SGPs is also represented.

Overall for the 4 years, the notifications provided by the BSLN closely follow the trends of the incidence of consultations by the SGP for ILI per 100,000 inhabitants with coinciding start, peak and end of the epidemics. In addition, the three sources of data show a comparable course over years and a high correlation within each season, i.e., Spearman's rank correlation for each epidemic season ranged from 0.70 to 0.98 for the Brussels region and at the national level. However, LHUB-ULB's data only were not able to reflect the trends of flu infection for the Flanders and Walloon regions, i.e., the lowest correlation (although still statistically significant) was observed in w40/14 to w39/15 for the Walloon Region (**Supplemental Digital Content 3**).

The representativeness of the LHUB-ULB at the microbiological level was also assessed by comparing the microbiological data's with those of National Influenza Centre for the three consecutive flu seasons (2014–2017). Among the 2,309 respiratory samples sent by the network of SGPs during the study period and analyzed at the National Influenza Centre (NIC), 1,197 (51.8%) were positive for influenza (70.0% influenza A, 23.1% influenza B). Thirty samples were co-infected with influenza A and B. Among the influenza A samples that were subtyped, 25.6% (236/922) were A(H1N1)pdm2009 and 70.4% (649/922) were A(H3N2). Thirty-six samples (7.6%) could not be subtyped due to their low viral load. Of the 277 influenza B samples analyzed, 70.8% (196/277) belonged to the Victoria lineage and 28.8% (80/277) to the Yamagata lineage. During the same period, 72.5% (1,553/2,141) and 28.1% (601/2,141) of influenza viruses detected by the LHUB-ULB overall were type A and type B, respectively. Difference in proportion with the NIC globally and over time were not significant (p > 0.05) (**Supplemental Digital Content 4**).

The representativeness of the LHUB-ULB in terms of molecular epidemiology was assessed by the phylogenetic analysis of the sequenced isolates (**Figure 2**). The LHUB-ULB derived Hemagglutinin (HA) sequences (n = 39) were compared to all sequences available from GISAID from Belgium for the same time period (n = 17) and an equivalent number of sequences from the United Kingdom (n = 56), France (n = 126), and the Netherland (n = 36). Most of the LHUB-ULB samples constitute distinct phylogenetic clusters, co-located with other samples with Belgium as a place of origin, within a background of seasonal Influenza A phylogeny. Preliminary results show that no significant genomic differences were observed within the LHUB-ULB samples or against deposited genomes from the same location and time period in Belgium, the Netherlands and France. However, most LHUB-ULB derived sequences form clusters distinct to the UK-derived samples. This observation would require larger sample numbers than the ones currently available and to be repeated for a number of seasons in order to be further validated. **Figure 2** shows HA sequence analyses, though the LHUB-ULB samples did have the Neuraminidase (NA) genomic information available due to the whole genome sequencing, the availability of NA sequences in the public databases from the same samples was almost entirely unavailable.

### DISCUSSION

In the twenty-first century, laboratory-based surveillance benefits from increased use of rapid diagnostic testing including ICT or LFT and multiplex PCR assay and increasingly rapid pathogen identification. The use of such advanced detection tools have dramatically cut the time to accurate diagnosis of infected patient but also increased the knowledge of epidemiological trends in comparison with older diagnostic methods based on culture and serology. The gradual integration of new detection and typing data into the European surveillance and alert systems represents one of the most exciting and challenging developments that could revolutionize the understanding of communicable diseases in the coming years (23).

It is clear that both individual contributors (the ultimate source of payment in fully privatized settings) as well as collective contributors (e.g., governmental interfaces) show a keen interest to such information on infectious disease occurrence. Individual payments (directly or through health insurance coverage) relate to an individual's health status and well-being; collective payments relate to the population-level health and the strategic knowledge of whether a pandemic is beginning. Hence, the interests of both individual and collective contributors are linked as an infection may be indicative of global trends of strategic importance. Therefore, "bigger" might indeed be "better" in the view that the more one knows about an epidemiological threat (of which influenza is the prototype), the better one can take care of the population, even though one may not be able to do much for the individual (as in Ebola outbreaks). If geographical coverage or microbiological tools are limited, detection of these threats may fail, thus CMLs which cover a large area and respond to many microbiological questions can represent an efficient solution (16). However, it is important to stress that one may be big yet in the wrong place or looking at the wrong threat, demonstrating volume in itself is not enough. LHUB-ULB is both centralized and well-located within highly connected and populated areas. As such it is effective in detecting Influenza, spreading very fast due to its high contagiousness and aided by the high population density and fast mobility of large numbers of people. It also concentrates well-focused subject expertise. Therefore, positioning and focus complement the "bigger" aspect.

However, such laboratory surveillance activity is often not financially covered and/or mandatory. Therefore, the cost containment leading to privatization of laboratory medicine may have a significant impact on this laboratory activity which is not directly centered on patient care. For example, in Belgium, the consolidation process leads to a decrease of clinical laboratories from 496 in 1996 to 148 in 2017 with a shift toward large

private laboratory structure. According to data on reimbursed microbiology tests obtained from the Belgian National Institute for Health and Disability Insurance (INAMI-RIZIV) for the period 2010–2015, on average 23.4% (range 22.5–24.3%) of all microbiological analysis (10,616,599 out of 45,355,005) performed were processed by private laboratories in regards to the 76.5% (range 77.4–75.7%) (34,696,723 out of 45,355,005) performed by university or hospital CMLs (Muyldermans G, unpublished data). The latter are more traditionally involved in the non-profit activities, such as research and development and/or public health surveillance.

The use of data from large CMLs may fill this gap that the privatization of the laboratory medicine may beget. One of the CMLs advantage is that more data are obtained from the processing of fewer samples, due to the higher analytical capacity of the consolidated lab, and since the data can be very representative, an increase in information can be achieved without necessitating a proportional increase in costs, even in cost-restricted settings (24). In this study, the handling of the flu data from the LHUB-ULB reveals its attractive features that can facilitate an early detection of seasonal influenza epidemics. These results illustrate that data are not only credible but also advantageous to use for surveillance and prediction purposes, especially for an automatic detection system. Despite, the LHUB-ULB catchment area represents a small geographical area; its representativeness for the nation-wide data is striking. In the future, the extent of representation will be further improved when data are collected from more consolidated laboratories or for a larger microbiological diversity.

In addition, this study confirms the lack of covering by BSLN in some municipalities located in Flanders or in Wallonia (15). In our study, the LHUB-ULB represented each year 100% of the notification in several municipalities located in these regions. The use of laboratory information's provided by one consolidated CML thanks to new molecular tools show to be a good complement of the information's provided by the SGPs and Hospital network and use by the NRC influenza (25, 26). The real-time integration of consolidated CMLs into the public health surveillance system would help the monitoring of influenza activity (intensity, duration, severity, . . . ) all over the year, the determination of type and subtypes of circulating strains and their antigenic and genetic characterization. As such integration could also contribute to the annual determination of the influenza vaccine content, the monitoring of resistance to antivirals and the detection of new potentially pathogenic influenza viruses (27).

In this frame, the use of high-throughput whole genome sequencing platforms available in large CMLs network, such as met in the LHUB-ULB demonstrates its potential for molecular epidemiological surveillance. Because, early detection

of epidemics is a key element to prevent loss of (quality of) life and its economic and material impact, such molecular surveillance would gain in efficiency through automated realtime monitoring and reporting to public health authorities from the regional to the European levels. Thus, governmental support of such CMLs seems a necessity, in view of the depth of real-time population-level information that can be obtained. Furthermore, other European countries have started demonstrating a clinical benefit from such data collected initiatives, as is the case in the UK at the National Mycobacterial Reference Service in Birmingham (28) and at UCLH with the integration of near real-time, whole genome sequencing utilized for the purposes of HIV and Influenza surveillance (20, 29). In the latter example, such integrated real-time surveillance was not only able to detect the emergence of novel subclade of influenza A(H3N2) virus in London, but also demonstrated its powerful to supplement traditional infection control procedures in the investigation and management of nosocomial outbreaks (19, 30).

However the availability of new microbial typing and detection techniques and culture-independent diagnostic methods, brings about a fundamental change in the way data has to be handled. These approaches are high-throughput and data-rich and create systematic stresses in the collection, analyses and safe handling of the generated data (31). For example one current obstacle toward clinical translation is that most algorithms in use need some programming expertise, together with specialized servers to handle and store all of the data. The generation of user-friendly informatics tools to effectively analyse high-throughput genomic data will be essential to the successful clinical application of genomic technology. In addition there are systemic aspects that also need to be addressed, such as the number of additional molecular or genomic testing parameters, regardless of the method, which can be supported routinely by the existing electronic health records that provide the architectural framework. These aspects can include the ordering of the test, the receiving of a document that summarizes the clinical interpretation, and storage of the interpretation (32). The integration of molecular genetic data to clinical and/or epidemiological data creates a challenge that requires interactive, information-sharing workspaces rather than uni-directional centralized reporting, such as those deployed in the TYPENED approach (33).

# CONCLUSION

Our results suggest that consolidated CMLs represent a wealth of information, including data usable for public health surveillance. The advent of real-time sequencing of organisms, their direct integration in surveillance tools at the regional, national and European levels would lead to real-time detection and alert, allowing the rapid prioritization of public health threats and the timely implementation of control strategies

### AUTHOR CONTRIBUTIONS

SV wrote the initial draft and supervised the manuscript revisions. NB and MW analyzed data and contributed to writing. BF performed the next generation sequencing experiments in the laboratory. LB and IT supervised the validity of virological results at the clinical level and National Reference Centre, respectively. GS, YD, and MB analyzed data and co-wrote the manuscript. EN designed the study, analyzed data, and cowrote the manuscript. MH coordinates the reliability of all laboratory results provided by LHUB-ULB, wrote substantial paragraphs, and provided critical review and commentary. ZK coordinated the whole genome sequencing pipeline for ICONIC. OV had the rationale for this work. All three co-senior authors contributed constructively in the writing of the manuscript and offered equally valuable advice for the discussion part of the manuscript.

### REFERENCES


### ACKNOWLEDGMENTS

We wish to thank the personnel of the Department of Microbiology of LHUB-ULB for its daily technical assistance. This work has been supported in part by a UCL Global Engagement Fund award to OV and ZK, which is gratefully acknowledged. The authors would like to thank Gaetan Muyldermans for his help in the preparation of the manuscript. This work was partly presented at the 28th European Congress of Clinical Microbiology and Infectious Diseases, 2018 April 21–24, Madrid, Spain. p.112: P0001.

Where authors are identified as personnel of the International Agency for Research on Cancer/WHO, the authors alone are responsible for the views expressed in this article and they do not necessarily represent the decisions, policy or views of the International Agency for Research on Cancer/WHO.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpubh. 2019.00150/full#supplementary-material


London, December 2016 to January 2017. Euro Surveill. (2017) 22:30466. doi: 10.2807/1560-7917.ES.2017.22.8.30466


**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 © 2019 Van den Wijngaert, Bossuyt, Ferns, Busson, Serrano, Wautier, Thomas, Byott, Dupont, Nastouli, Hallin, Kozlakidis and Vandenberg. 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.

# Soluble Recombinant Hemagglutinin Protein of H1N1pdm09 Influenza Virus Elicits Cross-Protection Against a Lethal H5N1 Challenge in Mice

*Shinya Yamada1 , Atsuhiro Yasuhara1 and Yoshihiro Kawaoka1,2,3 \**

*1 Division of Virology, Department of Microbiology and Immunology, Institute of Medical Science, University of Tokyo, Tokyo, Japan, 2 Department of Pathobiological Sciences, School of Veterinary Sciences, Influenza Research Institute, University of Wisconsin-Madison, Madison, WI, United States, 3 Department of Special Pathogens, International Research Center for Infectious Diseases, Institute of Medical Science, University of Tokyo, Tokyo, Japan*

Currently, influenza vaccines are produced using embryonated chicken eggs. Recently, recombinant influenza vaccines have been developed as a potential alternative to egg-grown vaccines. In this study, we evaluated the efficacy of soluble recombinant hemagglutinin (HA) protein produced in human cell culture (Expi293F cells) as an influenza vaccine against homosubtypic and heterosubtypic influenza virus challenges in mice. Mice were immunized intramuscularly with purified soluble HA protein of H1N1pdm09 virus and then challenged with a lethal dose of H1N1pdm09, seasonal H3N2, or highly pathogenic avian influenza (HPAI) H5N1 virus. Vaccinated mice showed better morbidity than mock-vaccinated mice following H1N1pdm09 challenge. By contrast, all mice died following H3N2 challenge. Interestingly, all vaccinated mice survived challenge with H5N1 virus, whereas all mock-vaccinated mice died. These results suggest that intramuscular immunization with recombinant HA proteins produced in Expi 293F cells could be of value in influenza vaccine strategies.

Keywords: influenza, vaccines, recombinant hemagglutinin protein, intramuscular immunization, cross-protection

# INTRODUCTION

Vaccination is the main public health strategy used to prevent and control influenza. Influenza vaccines are currently produced by propagating selected vaccine seed viruses in embryonated chicken eggs. However, recent vaccine seed viruses have changed their antigenicity through egg adaptation, causing reduced vaccine effectiveness (Katz and Webster, 1989; Wang et al., 1989; Rajakumar et al., 1990; Skowronski et al., 2014). Cell-based vaccine production may offer a solution to this problem. Recently, recombinant proteins have been produced in mammalian cells, insect cells, plant cells, and *E. coli* as an alternative strategy for vaccine development (Lin et al., 2008; Wei et al., 2008; Biesova et al., 2009; Chiu et al., 2009; Shoji et al., 2009; Song et al., 2009; Cornelissen et al., 2010; Kalthoff et al., 2010; Du et al., 2011; Khurana et al., 2011; Chen et al., 2013; Prabakaran et al., 2013; Wohlbold et al., 2015; Ge et al., 2016; Sim et al., 2016; Dunkle et al., 2017). As an added advantage, the absence of egg proteins in cell-based vaccines eliminates the potential for egg-related allergic reactions. In addition, cellbased vaccines can be stably produced with a shorter lead time than the 6 months required

### *Edited by:*

*Ville Peltola, Turku University Hospital, Finland*

### *Reviewed by:*

*Masatoshi Okamatsu, Hokkaido University, Japan Rabeh El Shesheny, St. Jude Children's Research Hospital, United States*

### *\*Correspondence:*

*Yoshihiro Kawaoka yoshihiro.kawaoka@wisc.edu*

### *Specialty section:*

*This article was submitted to Virology, a section of the journal Frontiers in Microbiology*

*Received: 11 January 2019 Accepted: 19 August 2019 Published: 04 September 2019*

### *Citation:*

*Yamada S, Yasuhara A and Kawaoka Y (2019) Soluble Recombinant Hemagglutinin Protein of H1N1pdm09 Influenza Virus Elicits Cross-Protection Against a Lethal H5N1 Challenge in Mice. Front. Microbiol. 10:2031. doi: 10.3389/fmicb.2019.02031*

**80**

for egg-based vaccine production, and production would not be affected by the supply of eggs (Milian and Kamen, 2015). A recent report showed that a recombinant influenza vaccine provided 30% greater efficacy than a standard inactivated influenza vaccine against influenza H3N2 in a clinical trial with a total of 9,003 recipients 50 years of age or older (Dunkle et al., 2017). However, in many of these studies, the recombinant proteins were produced by using a baculovirus expression system. A previous report showed that hemagglutinins (HAs) produced in 293T cells induce higher hemagglutination inhibition (HI) antibody titers compared with those produced in insect cells (de Vries et al., 2012). Expi293F cells, which were developed to produce a large amount of recombinant protein, are derived from the 293 cell line (Jain et al., 2017). We, therefore, used this cell line for our study to produce a recombinant HA protein. However, the vaccine efficacy of a recombinant HA produced in Expi293F cells has not been fully explored.

In many previous protection studies involving immunization with recombinant proteins, only the efficacy against the homologous challenge was examined. Although some studies showed cross-protection in mice against heterologous H5 virus challenge, the results were not obtained through intramuscular immunization of the recombinant HA protein but *via* intranasal infection with cold-adapted H1N1pdm09 influenza virus or baculoviruses that possessed influenza HA proteins (H1N1pdm09) (Jang et al., 2012; Sim et al., 2016). In addition, attenuated H5 viruses (not highly pathogenic H5 viruses) were used in those studies. One study has shown that intramuscular immunization with recombinant headless HA protein protects mice from heterologous virus challenge with H5N1 virus, but, again, attenuated H5N1 virus was used in the challenge experiment (Wohlbold et al., 2015). Therefore, the efficacy of the recombinant HA protein of seasonal influenza virus against heterologous highly pathogenic H5N1 virus challenges as a vaccine has not been fully studied. Accordingly, in this study, we sought to evaluate the efficacy of a soluble form of the recombinant HA protein of seasonal H1N1pdm09 virus, expressed in mammalian Expi293F cells, in terms of its cross-protection of intramuscularly immunized mice against heterologous virus challenges with highly pathogenic H5N1 virus and seasonal H3N2 virus.

# MATERIALS AND METHODS

# Ethics and Biosafety Statements

Human blood was collected in accordance with protocols that were approved by the Research Ethics Review Committee of the Institute of Medical Science, the University of Tokyo. Written informed consent was obtained from all participants. All experiments with H5N1 viruses were performed in biosafety level 3 (BSL3) laboratories at the University of Tokyo, which are approved for such use by the Ministry of Agriculture, Forestry, and Fisheries, Japan. All experiments with mice were performed in accordance with the University of Tokyo's Regulations for Animal Care and Use and were approved by the Animal Experiment Committee of the Institute of Medical Science, the University of Tokyo.

### Cells

Madin-Darby canine kidney (MDCK) cells were maintained in Eagle's minimal essential medium (MEM) containing 5% newborn calf serum (NCS). Human embryonic kidney 293 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 10% FCS. Expi293F cells (Thermo Fisher Scientific), maintained in Expi293 expression medium (Thermo Fisher Scientific), were incubated on an orbital shaker platform rotating at 125 rpm at 37°C under 8% CO2.

### Viruses

Mouse-adapted A/California/04/2009 (MA-CA04; H1N1pdm09) (Sakabe et al., 2011), mouse-adapted A/Aichi/2/1968 (MA-Aichi; H3N2), and A/Vietnam/1203/2004 (VN1203; H5N1) were propagated in MDCK cells or eggs and titrated in MDCK cells.

### Cloning, Expression, and Purification of Recombinant Hemagglutinin Protein

A gene fragment encoding the HA protein of MA-Ca04 without the transmembrane or cytoplasmic tail domain was amplified by PCR and cloned into the pCAGGS vector with a six-His tag coding sequence at the C-terminus. Primer sequences are available upon request. The plasmid was transfected into Expi293F cells by using ExpiFectamine 293 (Thermo Fisher Scientific) according to the manufacturer's protocol. At 4–6 days posttransfection, supernatants were cleared by low-speed centrifugation (900 rpm, 4°C, 5 min) and incubated with Ni-nitrilotriacetic acid (NTA) resin (Invitrogen) according to the manufacturer's instructions. Subsequently, SDS-PAGE and western blot analysis with an anti-His-tag mouse monoclonal antibody (MBL, Medical & Biological Laboratories) were performed. The concentration of the purified soluble MA-Ca04 HA protein (sMA-Ca04HA) was measured by using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific) according to the manufacturer's instructions.

### Immunofluorescence Assay

Twenty-four hours after transfection with plasmids, 293 cells were washed with PBS and fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X100. Cells were reacted with a mixture of anti-H1N1pdm09 HA mouse monoclonal antibodies (3E11C8, 5G10F, 7C2G7, 9C4C11), which were prepared in our laboratory, and with an anti-His-tag monoclonal antibody (MBL, Medical & Biological Laboratories). The cells were then incubated with Alexa Fluor 546 goat anti-mouse immunoglobulin G (Invitrogen).

### Enzyme-Linked Immunosorbent Assay

Ninety-six-well microtiter plates were coated with 2 μg/ml purified sMA-Ca04HA and then incubated with serially 5-fold diluted human or rabbit antibodies (starting at 5 μg/ml). We used two pdmH1-specific human monoclonal antibodies (10-5-64/6 and R4-5-72/1); an anti-H3HA human monoclonal antibody (1429D88/2), which was screened and established from human peripheral blood mononuclear cells (PBMCs) in our laboratory; CR9114, which broadly recognizes a conserved epitope in the stalk region; and an H5HA-specific rabbit monoclonal antibody (clone ID89, Sino Biological Inc.). After a 2-h incubation at 4°C, the plates were washed three times with ice-cold PBS, and then incubated with horseradish peroxidase (HRP) conjugated rabbit anti-human IgG or mouse anti-rabbit IgG (Jackson Immuno Research) (1:2,000 dilution) at 4°C for 1 h. The plates were then incubated with OPD (*O*-phenylenediamine, Sigma) in PBS containing 0.01% H2O2 for 10 min at room temperature, and the reaction was stopped with 0.05 ml of 1 N HCl. Absorbance was determined at 490 nm.

### Enzyme-Linked Immunosorbent Assay for Mouse Sera

Serum samples were collected from vaccinated mice and mockvaccinated mice 2 weeks after the second and third immunizations. Ninety-six-well microtiter plates were coated with 2 μg/ml purified sMA-Ca04HA or recombinant HA derived from either A/Perth/16/2009 (H3N2) or A/Vietnam/1203/2004 (H5N1) (Sino Biological). After being blocked with 5-folddiluted Blocking One (Nakarai), the plates were incubated with serially 4-fold diluted serum samples. After a 2-h incubation at 4°C, the plates were washed three times with ice-cold PBS, and then incubated with HRP-conjugated rabbit anti-mouse IgG (1:2,000 dilution) at 4°C for 1 h. The plates were then incubated with OPD in PBS containing 0.01% H2O2 for 10 min at room temperature, and the reaction was stopped with 0.05 ml of 1 N HCl. Absorbance was determined at 490 nm.

### Immunization and Virus Challenge

Five 6-week-old female BALB/c mice (Japan SLC) per group were intramuscularly immunized with 10 μg of purified HA protein adjuvanted with 50 μl of AddaVax (InvivoGen) three times with a 10- to 14-day interval between vaccinations. As a control, mice in the non-immunized group were injected with the same volume of PBS. Serum samples were collected 2 weeks after the second and third immunizations. To assess the protective efficacy of vaccination with sMA-Ca04HA, mice were intranasally infected with 10 mouse median lethal doses (10 MLD50) of mouse-adapted A/California/04/2009 (MA-CA04; H1N1pdm09), mouse-adapted A/Aichi/2/1968 (MA-Aichi; H3N2), or A/Vietnam/1203/2004 (VN1203; H5N1) under anesthesia 2 weeks after the final boost immunization. Morbidity and mortality were monitored for 13 days after challenge. Mice with body weight loss of more than 25% of their baseline body weight were euthanized.

# Virus Neutralization Assay

Serum samples collected from three vaccinated mice or three mock-vaccinated mice, which were pretreated with a receptordestroying enzyme (RED II; Denka Seiken, Tokyo, Japan), in duplicate, were serially two-fold diluted with MEM containing 0.3% bovine serum albumin (BSA-MEM) prior to being mixed with 100 TCID50 (50% tissue culture infectious doses) of mouseadapted A/California/04/2009 (H1N1pdm09), mouse-adapted A/Aichi/2/1968 (H3N2), or A/Vietnam/1203/2004 (H5N1) at 37°C for 30 min. The mixtures were then inoculated into MDCK cells and incubated for 1 h at 37°C. BSA-MEM containing TPCK-treated trypsin was added to each well and the cells were incubated for 3 days at 37°C. The cytopathic effect (CPE) was examined, and the neutralization titer was determined as the reciprocal of the highest serum dilution.

# RESULTS

### Generation and Characterization of Soluble Recombinant H1N1pdm09 Hemagglutinin Protein

To express soluble recombinant H1N1pdm2009 HA (pdmH1HA) in mammalian cells, the ectodomain of mouse-adapted A/ California/04/2009HA (MA-Ca04HA) (Sakabe et al., 2011) and a His-tag were cloned into the pCAGGS expression vector (**Figure 1A**), which was then transfected into Expi293F cells. The secreted HA was purified using Ni-NTA agarose chromatography. The expression and secretion of the soluble MA-Ca04HA protein with the His tag (sMA-Ca04HA) was examined by means of gel electrophoresis followed by western blotting using an antibody against the His-tag. The results showed that sMA-Ca04HA was successfully secreted into the supernatant with an expected molecular mass of 75 kDa (**Figure 1B**). Immunofluorescence analysis showed antipdmH1HA monoclonal antibody binding to the expressed HA protein, suggesting that the HA protein had folded properly (**Figure 1C**). We also examined the binding properties of several other antibodies to the soluble MA-CA04HA by using an ELISA (**Figure 1D**). We used two pdmH1HA-specific human monoclonal antibodies (10-5-64/6 and R4-5-72/1), which were screened and established from human PBMCs in our laboratory (Kubota-Koketsu et al., 2009); CR9114, which broadly recognizes a conserved epitope in the stalk region (Dreyfus et al., 2012); an anti-H3HA human monoclonal antibody (1429D88/2), which was also screened and established from human PBMCs in our laboratory (Kubota-Koketsu et al., 2009); and an H5HA-specific rabbit monoclonal antibody (clone ID89, Sino Biological Inc.). The two pdmH1HA-specific human monoclonal antibodies and CR9114 bound to sMA-Ca04HA, but the anti-H3HA and anti-H5HA monoclonal antibodies did not. These results suggest that sMA-Ca04HA retained the conformation and antigenicity of the HA protein of A/California/04/2009.

# Efficacy of sMA-Ca04HA as a Vaccine Against Lethal Infection in Mice

To evaluate the immunogenicity of sMA-Ca04HA, naïve 6-weekold BALB/c mice were vaccinated intramuscularly with 10 μg of purified sMA-Ca04HA protein adjuvanted with 50 μl of AddaVax (InvivoGen) (Goff et al., 2013) three times with a 10- to 14-day interval between vaccinations. As a control, mice in the non-immunized group were injected with the same volume of PBS. Serum samples were collected 2 weeks after the second and third immunizations. We examined the seroreactivity induced by sMA-Ca04HA by using an ELISA against purified HA proteins (**Figures 2A–C**). Mice vaccinated with sMA-Ca04HA elicited

high antibody titers against sMA-Ca04HA (**Figure 2A**). To determine whether immunization with sMA-Ca04HA could increase the breadth of reactivity, we analyzed heterologous seroreactivity against HA proteins from H5N1 and H3N2 viruses. The sera from sMA-Ca04HA-vaccinated mice showed reactivity in the ELISA to both H3HA and H5HA protein (**Figures 2B,C**). In contrast, mice injected with PBS elicited no antibodies reactive to the test HA proteins. To determine whether immunization with sMA-Ca04HA could induce neutralizing antibodies against

homosubtypic H1N1pdm09 virus and heterosubtypic H3N2 and H5N1 viruses, we performed neutralization assays using sera obtained from three mice immunized with the recombinant H1N1pdm09 HA (**Table 1**). In the assay, we detected neutralizing activity against homosubtypic H1N1pdm09 virus but not against H3N2 or H5N1 viruses. To assess the protective efficacy of sMA-Ca04HA as a vaccine, we performed a viral challenge study (**Figures 3A–C**). Mice were infected with 10 MLD50 of mouse-adapted A/California/04/2009 (MA-CA04; H1N1pdm09),

### TABLE 1 | Virus neutralization by serum of mice immunized with sMA-Ca04HA.


expressed as mean changes in body weight ± SD (*n* = 5).

mouse-adapted A/Aichi/2/68 (MA-Aichi; H3N2), or A/ Vietnam/1203/2004 (VN1203; H5N1) 2 weeks after their final boost vaccination. Upon challenge with MA-CA04, all mice vaccinated with sMA-Ca04HA showed no weight loss and survived, whereas mock-vaccinated mice lost more than 20% of their initial body weight and all of the mice died by day 7 post-infection (**Figure 3A**). In contrast, viral challenge with MA-Aichi/2/68 (H3N2) caused all vaccinated and mock-vaccinated mice to succumb to their infections within 9 days of challenge (**Figure 3B**). Interestingly, upon challenge with Vietnam/1203/04 (H5N1), all vaccinated mice survived and showed moderate weight loss, whereas the mock-vaccinated mice began to succumb to their infections at 8 days post-infection and all of these mice died within 12 days of challenge (**Figure 3C**). Our results indicate that this soluble recombinant HA protein mediates limited cross-protection against heterosubtypic influenza virus infection.

# DISCUSSION

In this report, we showed that intramuscular inoculation of a recombinant Expi293F cell-based influenza vaccine candidate that comprised the ectodomain of H1N1pdm09 virus HA induced antibodies against the HAs of H1N1pdm09, H3N2, and highly pathogenic H5N1 viruses and provided protection against homosubtypic H1N1pdm09 and heterosubtypic H5N1 challenge in mice. Although sMA-Ca04HA immunization induced antibodies against H3HA protein, vaccinated mice were not protected from H3N2 challenge. H1 and H5 belong to group 1, whereas H3 belongs to group 2 of the phylogenetic groupings of hemagglutinin. Therefore, sMA-Ca04HA might induce antibodies against epitopes in the stalk region conserved among group 1 members that efficiently protect mice from challenge with viruses belonging to group 1. Interestingly, in neutralization assays using sera obtained from mice immunized with the recombinant H1N1pdm09 HA, we did not see any neutralizing activity against H5N1 virus in the sera, suggesting that the protection from the lethal H5N1 virus challenge afforded by the recombinant H1 HA is mediated by mechanisms other than virus neutralization.

Although our sMA-Ca04HA contained a His-tag, previous studies have shown that antibodies against His-tag are not generated in mice immunized with His-tagged recombinant proteins (Delaney et al., 2010; Shoji et al., 2011). We, therefore, assumed that antibodies against His-tag were not produced in our experiments and our results were not affected by the His-tag.

Recently, recombinant HA proteins have been produced on various platforms as an alternative strategy for influenza vaccine production. The glycosylation state of HA proteins differs depending on the cells in which they are expressed and this difference can affect the antigenicity of the proteins (de Vries et al., 2012). The report showed that HAs produced in 293T cells induce higher HI antibody titers compared with those produced in insect cells. In our study, we used Expi293F cells, which originated from the same 293 cell line as 293T cells. Therefore, we can assume that HAs expressed in Expi293F cells possess a similar glycosylation state and antigenicity to 293T cell-expressing HAs. Consequently, HAs produced in Expi293F cells may induce more effective antibodies compared with those produced in insect cells. The glycosylation state of the recombinant HA protein expressed in Expi293F might act advantageously to induce antibodies that target the conserved region of the HAs of H1N1pdm09 and H5N1 viruses and provide cross-protection mediated by mechanisms such as antibody-dependent cellular cytotoxicity. However, further studies are needed to better understand the influence of the glycosylation state of HA on its antigenicity and to understand which expression system is best suited for the development of a recombinant HA protein vaccine.

Previous reports have shown that intranasal infection of mice with cold-adapted H1N1pdm09 virus or baculovirus displaying the HA protein of H1N1pdm09 virus protected mice from H1N1pdm09 and H5 virus challenge (Jang et al., 2012; Sim et al., 2016). However, in these studies, attenuated viruses were used. By contrast, here we showed that intramuscular immunization with recombinant H1N1 pdm09 HA protein alone successfully protected mice from heterosubtypic highly pathogenic H5N1 challenge. Future mechanistic studies to understand how intramuscular immunization with recombinant HA could be sufficient to induce immunity to protect mice from heterosubtypic highly pathogenic H5N1 challenge will be important for vaccine development.

In conclusion, intramuscular immunization with Expi293F cell-based soluble recombinant HA protein induced protection against homosubtypic and heterosubtypic challenge in the same phylogenetic group. Our human Expi293F cell-based vaccine candidate thus offers a new strategy for influenza vaccine development. Further studies using other HA proteins from different subtypes would enhance the development of Expi293F cell-based vaccines. Further studies to understand the mechanistic basis of the immunity induced by the intramuscular administration of recombinant HA protein will aid in vaccine development.

# AUTHOR CONTRIBUTIONS

SY designed the study and performed the experiments. SY, AY, and YK analyzed the data. SY and YK wrote the manuscript. All authors reviewed and approved the manuscript.

# FUNDING

This research was supported by Strategic Basic Research Programs from the Japan Science and Technology Agency (JST); by Leading Advanced Projects for medical innovation (LEAP) from the Japan Agency for Medical Research and Development (AMED) (JP18am001007, JP18am001008); by a Grant-in-Aid for Scientific Research on Innovative Areas from the Ministry of Education, Culture, Science, Sports, and Technology (MEXT) of Japan (No. 16H06429, 16K21723, and 16H06434); by the Japan Initiative for Global Research Network on Infectious Diseases (J-GRID) from the MEXT of Japan; by e-ASIA Joint Research Program from AMED (JP17jm0210042); by Research Program on Emerging and Re-emerging Infectious Diseases from AMED (JP19fk0108104, JP19fk0108066); by fund for the Promotion of Joint International Research [Fostering Joint International Research (B)] from Japan Society for the Promotion of Science (JSPS) (JP18KK0225); by MEXT KAKENHI Grant Number (15K08491, 18K07139); and

### REFERENCES


by the NIAID-funded Center for Research on Influenza Pathogenesis (CRIP) (HHSN272201400008C).

# ACKNOWLEDGMENTS

We would like to thank Dr. Susan Watson for editing the manuscript.


Wohlbold, T. J., Nachbagauer, R., Margine, I., Tan, G. S., Hirsh, A., and Krammer, F. (2015). Vaccination with soluble headless hemagglutinin protects mice from challenge with divergent influenza viruses. *Vaccine* 33, 3314–3321. doi: 10.1016/j.vaccine.2015.05.038

**Conflict of Interest Statement:** YK has received speaker's honoraria from Toyama Chemical and Astellas Inc.; has received grant support from Chugai Pharmaceuticals, Daiichi Sankyo Pharmaceutical, Toyama Chemical, Tauns Laboratories, Inc., Otsuka Pharmaceutical Co., Ltd., and Denka Seiken Co., Ltd.; and is a co-founder of FluGen.

The remaining 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 © 2019 Yamada, Yasuhara and Kawaoka. 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.*

# Characteristics of Hospitalized Rhinovirus-Associated Community-Acquired Pneumonia in Children, Finland, 2003–2014

Maria Hartiala<sup>1</sup> \*, Elina Lahti 1,2, Ville Forsström<sup>1</sup> , Tytti Vuorinen<sup>3</sup> , Olli Ruuskanen<sup>1</sup> and Ville Peltola<sup>1</sup>

<sup>1</sup> Department of Pediatrics and Adolescent Medicine, Turku University Hospital, University of Turku, Turku, Finland, <sup>2</sup> Child and Adolescent Clinic, City of Turku Welfare Division, Turku, Finland, <sup>3</sup> Department of Virology and Clinical Virology, Turku University Hospital, University of Turku, Turku, Finland

Background: Rhinovirus (RV) is the most common cause of respiratory tract infections in children but, still, the clinical characteristics of RV-associated pneumonia have not been sufficiently investigated.

Edited by: Olivier Vandenberg,

LHUB-ULB, Belgium

### Reviewed by:

Peng Yang, Capital Medical University, China Ana Afonso, University of São Paulo, Brazil Marijke Bernadette Reynders, AZ Sint-Jan Brugge-Oostende AV, Belgium

> \*Correspondence: Maria Hartiala mkhonk@utu.fi

### Specialty section:

This article was submitted to Infectious Diseases - Surveillance, Prevention and Treatment, a section of the journal Frontiers in Medicine

> Received: 13 February 2019 Accepted: 07 October 2019 Published: 22 October 2019

### Citation:

Hartiala M, Lahti E, Forsström V, Vuorinen T, Ruuskanen O and Peltola V (2019) Characteristics of Hospitalized Rhinovirus-Associated Community-Acquired Pneumonia in Children, Finland, 2003–2014. Front. Med. 6:235. doi: 10.3389/fmed.2019.00235 Methods: We identified children and adolescents younger than 18 years of age treated for community-acquired pneumonia as inpatients at the Turku University Hospital from 2003 to 2014 and analyzed for RV by PCR of a respiratory tract specimen. We collected the data from medical records and compared RV-positive children with RV-negative children.

Results: Of the study population of 313 children with pneumonia who were studied for RV, it was detected in 82 (26%). RV-positive children were younger (median age 2.6 years, interquartile range [IQR] 1.1–4.6 vs. 3.5 years, IQR 1.7–8.3, p = 0.002) and they had more often a history of preterm birth (16% vs. 5%, adjusted odds ratio 2.89, 95% confidence interval 1.21–6.92, p = 0.017) than RV-negative children. RV-positive children had a higher median white blood cell count than RV-negative children at presentation with pneumonia. The signs, symptoms, and severity of pneumonia were mostly similar in RV-positive and RV-negative children.

Conclusions: RV was frequently detected in young children hospitalized with community-acquired pneumonia. We identified premature birth as a factor associated with RV-positive pneumonia. The clinical features of pneumonia did not clearly differ between RV-positive and RV-negative children. Further studies are needed to clarify the clinical significance of detection of RV in children with pneumonia.

Keywords: children, pneumonia, respiratory tract infection, rhinovirus, white blood cell count

# INTRODUCTION

Pneumonia is a common cause of hospitalization in children. Rhinovirus (RV) is the most common cause of respiratory tract infections in children worldwide, and frequently detected in community-acquired pneumonia (CAP) (1–6). According to global estimates, 120 million episodes of pneumonia in children younger than 5 years of age are recorded annually (7). RV is detected in

**88**

up to 46% of children with CAP (5, 8, 9). RVs are small non-enveloped RNA viruses classified into three species (A, B, and C) with approximately 160 different types (10, 11). Simultaneous circulation of several RV types in populations year-round explains the high frequency of RV infections.

The scientific literature on asthma exacerbation and wheezing illnesses caused by RV is extensive (12), but the clinical characteristics of RV-associated pneumonia in children have not been thoroughly investigated. RV is commonly present in mixed viral-bacterial and viral-viral infections (13) and a substantial proportion of asymptomatic children are positive for RV by PCR of upper respiratory tract specimens (14). The role of RV as the causative agent of pneumonia is unclear.

The aim of this study was to assess the risk factors for, clinical characteristics and prevalence of RV-associated pneumonia in children. We compared the medical record data of RV-positive and RV-negative children hospitalized with CAP.

### METHODS

### Participants and Data Collection

This retrospective study involved children and adolescents younger than 18 years of age treated as inpatients at the Department of Pediatrics and Adolescent Medicine, Turku University Hospital (Turku, Finland), during a 12-year period from 2003 to 2014. To identify children who were hospitalized with CAP, we searched the Electronic Registry of the Turku University Hospital for International Classification of Diseases (ICD-10) codes related to pneumonia (J12–18, J10.0, J11.0, J85, J86, J90) with an age limit of 18 years. Of this patient population, we identified those with a diagnostic polymerase chain reaction (PCR) test for RV performed during the hospitalization for CAP. The medical records of these children were reviewed to collect the clinical and background data. Excluding preterm birth, only currently present underlying conditions were considered. Tachypnea was defined as respiratory rate >60/min in infants younger than 2 months of age, >50/min in infants from 2 to 12 months, >40/min in children from 1 to 5 years and >30/min in children 6 years of age or older.

The study was approved by the Institutional Review Board at the Clinical Research Centre of the Turku University Hospital.

### Laboratory Detection

The diagnostic tests for RV were in house qualitative reverse transcription (RT) -PCR assays and commercial multiplex PCR tests for respiratory viruses including RV, which were in routine use in the diagnostic laboratory during the study period. The first in house PCR used detected RV and enterovirus (15). It was later replaced by a triplex test for RV, enterovirus and respiratory syncytial virus (16). The analytical procedures, sensitivities and specificities of the in house tests are described in the abovecited references. Since 2008 we also used commercial multiplex PCR kits, first a Seeplex RV12 Ace detection kit and since 2013 Anyplex RV16 detection kit (both from Seegene, Seoul, Korea). The commercial multiplex PCR methods may have slightly lower sensitivities for RV than the in house tests (17). A child was considered as a RV-positive case if RV was detected either by the in-house PCR or the multiplex PCR or both.

### Data Analysis

RV-positive children were compared with RV-negative children. To test whether the results were affected by the presence of other viruses, we conducted a sensitivity analysis of children with a sole RV finding (no other viruses detected) compared with those who had no viruses detected. Data were presented as proportions, or medians with interquartile ranges (IQR). Univariate comparisons were performed for continuous data by use of the Wilcoxon ranksum test and for categoric data by use of the χ 2 test or Fisher's exact test. All tests were two-sided. The significance level was P < 0.05. A multivariate logistic regression analysis was conducted to examine the independent risk factors for RV-positive CAP. The final model included age, sex and presence of the following prior diseases or conditions: asthma or reactive airway disease, premature birth, neurological condition, cardiovascular disease, and atopic eczema or sensitization to aeroallergen. Statistical analyses were performed using SAS system for Windows, version 9.4. (SAS Institute Inc., Cary, NC, USA) or SPSS version 23.0 (IBM SPSS Statistics, IBM Corp., Armonk, NY, USA).

# RESULTS

### Study Population, Characteristics and Underlying Conditions

Of a total of 2484 children with CAP, 1270 (51%) were treated as inpatients and 1214 (49%) as outpatients. Hospitalization was needed for 81 to 143 children with CAP per year (**Figure 1**). Inpatients were younger than outpatients (median age 2.88 [IQR 1.49–5.63] years vs. 3.38 [1.77–7.15] years, p < 0.001). Of 1270 inpatients 313 (25%) had PCR diagnostics for RV done during the hospitalization, and 82 (26% of 313) had RV detected. Children treated as outpatients for pneumonia were not tested for RV.

The final study population (n = 313) consisted of 171 males (55%) and 142 females (45%) with a median age of 3.09 (IQR 1.53–7.35) years (**Table 1**). All patients had radiologically confirmed pneumonia. The monthly peak occurrence of RV pneumonia was in October (**Figure 2**). RV-positive patients were younger (median age 2.59 [IQR 1.08–4.59] years) than RV-negative patients (median age 3.51 [IQR 1.68–8.26] years) (p = 0.002).

Preterm birth was more frequent in RV-positive (16%) compared to RV-negative children (5%, p = 0.002 in univariate analysis). This association remained significant in the multivariate logistic regression analysis (OR 2.89, 95% CI 1.21–6.92, p = 0.017). Other underlying conditions were not significantly associated with RV-positive CAP. Fifteen percent of RV-positive and 12% of RV-negative children had a diagnosis of asthma at the time of admission. Atopic eczema or sensitization to aeroallergen was present in 27% of RV-positive and in 19% of RV-negative children.

### Clinical Findings

The clinical profiles of pneumonia were in general similar in children with or without RV, the most common symptoms

TABLE 1 | Demographic characteristics and underlying conditions of children with community-acquired pneumonia requiring hospitalization; RV-positive patients compared to RV-negative patients.


CI, confidence interval; IQR, interquartile range; OR, odds ratio; RV, rhinovirus.

<sup>a</sup>Wilcoxon rank-sum test.

bχ 2 test.

<sup>c</sup>Fisher's exact test.

being fever and cough in both groups (**Table 2**). However, the frequency of documented fever was lower in RV-positive than in RV-negative children (84% vs. 97%, p < 0.001). Dyspnea or shortness of breath was present in 35% of RV-positive patients and in 28% of RV-negative patients, and tachypnea in 30% and 29%, respectively. The duration of symptoms before referral

was shorter in RV-positive (median 3 [IQR 1–7] days) than in RV-negative children (median 5 [IQR 2–8] days) (p = 0.006).

### Laboratory Results

The white blood cell (WBC) count determined on admission was higher in RV-positive than in RV-negative children (median 16.4 [IQR 12.2–25.0] × 10<sup>9</sup> /L vs. 14.1 [8.5–20.4] × 10<sup>9</sup> /L, p = 0.002), whereas the C-reactive protein (CRP) concentrations were not significantly different (**Table 2**).

At least one virus other than RV was detected by PCR in 13 (16%) of 82 children positive for RV and in 45 (19%) of 231 children negative for RV. Five children had pathogenic bacteria in the blood culture: 1 RV-positive and 3 RV-negative children had Streptococcus pneumoniae and 1 RV-negative child had Streptococcus pyogenes bacteremia.

### Treatment and Outcome

Thirteen percent of RV-positive and 21% of RV- negative patients received antimicrobial therapy before referral to the hospital, and all patients received antimicrobial therapy during hospitalization. Of RV-positive children, 34% received oxygen supplementation, 20% were admitted to the intensive care unit, and 10% required invasive mechanical ventilation, whereas the corresponding rates were, respectively, in RV-negative children 27, 15, and 5%. These differences were not statistically significant. The median duration of hospitalization was 2.0 days in both groups. Complicated pneumonia (defined as lung abscess, necrotising pneumonia, or empyema) was documented in 5 (6%) of RV-positive and in 20 (9%) of RV-negative patients. A 1-year-old boy with no underlying conditions, negative for RV but positive for influenza B and adenovirus, died of pneumonia after 5 days of hospital treatment. Thus, the mortality rate in the study population was 0.3%.

# Sensitivity Analysis

As a test of sensitivity of our results to effects of other viruses, we compared children with a sole RV finding (no other virus detected) (n = 69) with those who had no virus detected (n = 186). The findings remained essentially similar. Children with RV only were younger than virus-negative children (median age 2.92 [IQR 1.11–4.67] vs. 4.52 [1.81–9.16] years, p = 0.002) and they had more often a history of preterm birth (14% vs. 5%, univariate p = 0.009, multivariate analysis OR 2.98 [95% CI 1.10– 8.05], p = 0.032). Similar to results of all subjects, the duration of symptoms before referral was shorter, fever was documented less frequently, and the median WBC count was higher in sole RV-positive children compared to virus-negative children.

# DISCUSSION

In this study covering a 12-year period, RV was detected in 26% of children hospitalized with radiologically-confirmed CAP in whom testing was performed. RV-associated CAP was particularly common in young children and in children born prematurely. Clinically, RV-associated pneumonia did not clearly differ from RV-negative pneumonia. CRP and WBC levels were in most cases high and patients were treated with antibiotics because of suspected bacterial pneumonia.

Hospitalization was required in about half of all children evaluated at the emergency department for CAP, which is similar TABLE 2 | Clinical features, laboratory results and outcomes of children with community-acquired pneumonia requiring hospitalization; RV-positive patients compared to RV-negative patients.


CRP, C-reactive protein; IQR, interquartile range; RV, rhinovirus; WBC, white blood cell.

aχ 2 test.

<sup>b</sup>Fisher's exact test.

<sup>c</sup>Wilcoxon rank-sum test.

<sup>d</sup>Data available on 80 RV-positive and 228 RV-negative children.

<sup>e</sup>Data available on 81 RV-positive and 229 RV-negative children.

<sup>f</sup>Data available on 79 RV-positive and 228 RV-negative children.

<sup>g</sup>Data available on 81 RV-positive and 230 RV-negative children.

to earlier data from Finland (18, 19). The yearly number of children hospitalized with CAP was variable and had a downward trend during the latter part of the study period (**Figure 1**). Worldwide, introduction of pneumococcal conjugate vaccines has had a substantial impact on children's hospitalizations with all-cause pneumonia and, also, on hospitalizations with viruspositive pneumonia (20). In Finland, a 10-valent pneumococcal conjugate vaccine was included in the national immunization program for all children in 2010. As pneumococcal pneumonia cases decrease due to vaccinations, viruses are presumably becoming even more substantial cause of pneumonia. This was not seen for RV in this study, which did not include a long period of time after introduction of pneumococcal vaccinations; the yearly proportion of RV-positive CAP of those tested for RV did not systematically change. The RV prevalence documented in our study is in concordance with previous literature reporting the detection rate of RV from 14 to 46% in children hospitalized with pneumonia (2, 3, 5, 8, 9, 13). RV infections occur year-round but most commonly during autumn and spring (21), which was seen also in our study.

In our study, RV-positive pneumonia patients were younger than RV-negative patients. Children with RV-positive and RVnegative pneumonia were largely similar in respect of the severity of illness and the response to treatment. The need of oxygen supplementation, treatment at the intensive care unit, and invasive mechanical ventilation were non-significantly more common in those infected with RV. We found only a few earlier studies considering the clinical characteristics of RV pneumonia in children. Annamalay et al. compared clinical features as well as laboratory and microbiology findings of RV-positive and RVnegative children hospitalized with pneumonia in Mozambique, without finding any significant between-group differences (22). In the study of Ahn et al., RV-positive children hospitalized for acute lower respiratory tract infections were younger, had shorter fever duration, and higher frequencies of chest retraction and wheezing than RV-negative children (23). Other studies compared children with CAP or with an unspecified lower respiratory tract infection caused by different RV types and found only marginal differences (23–25).

Wheezing illnesses (bronchiolitis, recurrent wheezing, or exacerbation of asthma) often associate with RV (11, 12, 23, 26– 28). Our study focused on pneumonia, which may be sometimes difficult to differentiate from wheezing illnesses. Furthermore, pneumonia and wheezing may have common risk factors. An association between childhood pneumonia and asthma has been noticed for long (29–31). Parental asthma, earlier wheezing, and exposure to tobacco smoke are associated with more severe clinical course of pneumonia (32). In our study, asthma was slightly more common in RV-positive patients than others, but, contrary to our expectation, wheezing in auscultation of lungs was not more prevalent among RV-positive than RV-negative CAP patients. The clinical picture of RV-associated pneumonia seems to be clearly different from RV-associated wheezing illness.

We found that RV-positive patients were considerably more often prematurely born than others. Accordingly, in the study of Miller et al., prematurely born children were found to be particularly susceptible to RV infections (33). Kennedy et al. have highlighted that it is the nature and extent of the immune response to the virus that determines the symptom profile (34). Numerous pathophysiological mechanisms, such as diminished immune responses and lung function but also inflammatory and airway re-modeling pathways activated by viruses, are proposed to influence the increased risk of respiratory disorders, including RV infections, following preterm birth (35).

The role of RV as a true pneumonia pathogen is unclear (36– 41). Life-threatening cases of pneumonia caused by only RV have been reported (42), and RV has been detected directly from the lung tissue from a child with pneumonia (43). RV viremia is particularly common in CAP patients with RV-C (44, 45). In our study, as well as in most other studies concerning the etiology of pneumonia, RV was detected by PCR from upper respiratory tract specimens, and the presence of the virus at the site of infection, the lung, remains unknown. As RV is frequently (approximately in 15%) detected also in asymptomatic children (14), its role as a causative agent vs. a bystander in children with pneumonia can be questioned. It should be noted, however, that the mean virus shedding time after RV infection is as short as 11 days in immunocompetent children (46). Persistent shedding of RV is not known to occur in otherwise healthy subjects. The detection of RV in asymptomatic children may reflect previous infection, ongoing asymptomatic or mild infection (possibly by less virulent RV types), or an incubation period preceding the onset of symptoms.

The identification of RV directly from the lung is not feasible in common clinical settings. In severe pneumonia or in immunocompromised subjects, obtaining invasive samples like bronchoalveolar lavage for microbiological evaluation including RV PCR would be critically informative (47). RV loads in upper respiratory tract specimens do not clearly correlate with the clinical course of the infection. Children with wheezing or only rhinitis have been reported to have similar RV loads in their nasal washes (28). On the contrary, in a recent study, higher RV viral load was associated with more severe respiratory symptoms (48). Serologic assays for RV infection could help to prove acute infections but such tests are not in routine use. Recently, despite high phylogenetic diversity of RV, the development of RV speciesspecific antibody test has been successful (49). Integrating studies of host response, microbe detection, and airway microbiome is a modern approach in pneumonia diagnostics (50). Development of effective vaccines and drug treatments for RV would ultimately make it possible to show the impact of RV on childhood pneumonia. Recently, Toll-like receptor 3 blockage has been studied as one of the possible options for treatment (51).

RV has been associated with a severe course of pneumonia in children, similar to our findings (52). Children with RV-positive pneumonia may have a concomitant bacterial pneumonia, or a secondary bacterial pneumonia following a RV infection. In our children with RV-associated CAP, CRP and WBC levels were elevated, and the median WBC counts were even higher in the RV-positive than in the RV-negative group. Increased levels of WBC have been earlier reported in RV-related lower respiratory tract infections (53). CRP and WBC are not highly specific in differentiation between viral and bacterial infections and better biomarkers or microbiologic methods would be needed in order to confirm or exclude bacterial co-infection in RV-associated CAP. Mixed infections are increasingly recognized and possibly associated with a more severe course of pneumonia, particularly the combination of RV and S. pneumoniae (13, 54–57). In vitro, RV infection has been shown to stimulate adhesion of S. pneumoniae to airway epithelial cells via increases in the plateletactivating factor receptors (58). The impairment of immune response to bacterial products and phagocytosis of bacteria in human macrophages in response to RV exposure has been documented (59, 60). Moreover, the seasonality of pneumonia

coinciding with viral lower respiratory tract infections has been observed (61).

In the absence of effective antiviral drugs for other respiratory viruses than influenza A and B virus, the clinical significance of diagnosing viral etiology of pneumonia has not been firmly established. Avoidance of the unnecessary use of antibiotics and excessive laboratory or other tests, shortening of the length of stay in the hospital, prediction of the clinical course of illness, and prevention of transmission to other patients by isolation or other methods are potential benefits of virus detection, in addition to surveillance of local epidemiology of seasonal viruses. In our study, the characteristics of RV-associated pneumonia were closely similar to RV-negative pneumonia, and identification of RV did not result in the withholding of antibiotic treatment. The matter could be different in milder pneumonia cases not needing hospitalization.

Our study has notable limitations related to the retrospective setting. First, our study population was somewhat selected as RV tests were not routinely performed for all CAP patients. However, our study population can be considered to be a representative sample of overall CAP inpatients as key figures in patients undergoing RV detection in the present study and in CAP inpatients in our previous prospective CAP study are comparable (13). Second, data on the types of RV and viral loads were not available, and other viruses and bacteria were not comprehensively analyzed. In our previous CAP study with induced sputum as a diagnostic sample and patient population overlapping with our present study, 64% of RV findings belonged to RV A species and 36% to RV C species, and viral-bacterial coinfections were frequent (13). Similarly, other researchers have reported that RV A and C species are frequent and RV B species rare in children with pneumonia (23, 24). Third, the clinical data was collected from the medical records, which might be incomplete. Fourth, diagnostic methods and clinical practices varied during the 12-year study period. Fifth, our study included only hospitalized CAP patients. Sixth, the lack of a control group of healthy children is an obvious limitation in this study.

In conclusion, RV is frequently present in childhood pneumonia, particularly in vulnerable patients such as young children and those with a history of preterm birth. Among children hospitalized with CAP in this study, RV-positive pneumonia was a rather severe disease with high levels of inflammatory biomarkers and a clinical course that often

### REFERENCES


necessitated intensive care, with no justification for withholding antibiotic treatment. Other studies are needed to establish the clinical characteristics of RV pneumonia in outpatients. Considering the large global burden of pneumonia and the high prevalence of RV in children with pneumonia, development of a diagnostic marker indicating RV as a true cause of the disease, and drugs and vaccines for RV-specific treatment and prevention could have major significance in the future.

### DATA AVAILABILITY STATEMENT

The datasets for this manuscript are not publicly available because: Participant privacy prevents public sharing of individual-level data. Requests to access the datasets should be directed to Maria Hartiala, mkhonk@utu.fi.

# ETHICS STATEMENT

This study involved only retrospective review of medical record data that had been collected during routine patient care. The study was approved by the Institutional Review Board at the Clinical Research Centre of the Turku University Hospital with a statement that an evaluation by the Ethics Committee was not needed.

# AUTHOR CONTRIBUTIONS

MH and VF collected the clinical data. TV was responsible for virus diagnostics. MH analyzed data and wrote the manuscript with support of VP, EL, VF, TV, and OR. VP supervised the project.

### FUNDING

This study was supported by the Turku University Foundation, the Väinö and Laina Kivi Foundation, the Research Foundation for the Pulmonary Diseases, and the Foundation for Pediatric Research.

# ACKNOWLEDGMENTS

The authors thank M. Koivisto for help with data analyses.


with hypogammaglobulinaemia. Clin Microbiol Infect. (2013) 19:E322–7. doi: 10.1111/1469-0691.12193


**Conflict of Interest:** 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 © 2019 Hartiala, Lahti, Forsström, Vuorinen, Ruuskanen and Peltola. 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.