# MOLECULAR VACCINES AGAINST PATHOGENS IN THE POST-GENOMIC ERA

EDITED BY : Angel Alejandro Oñate, Alberto Moreno and Yanmin Wan PUBLISHED IN : Frontiers in Cellular and Infection Microbiology

#### 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 978-2-88963-513-9 DOI 10.3389/978-2-88963-513-9

#### 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

# MOLECULAR VACCINES AGAINST PATHOGENS IN THE POST-GENOMIC ERA

Topic Editors:

Angel Alejandro Oñate, University of Concepcion, Chile Alberto Moreno, Emory University School of Medicine, United States Yanmin Wan, Fudan University, China

Citation: Oñate, A. A., Moreno, A., Wan, Y., eds. (2020). Molecular Vaccines against Pathogens in the Post-Genomic Era. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-513-9

# Table of Contents


Thomas Ebensen, Simon Delandre, Blair Prochnow, Carlos A. Guzmán and Kai Schulze

*46 Oral Vaccination With a Formulation Combining* Rhipicephalus microplus *Subolesin With Heat Inactivated* Mycobacterium bovis *Reduces Tick Infestations in Cattle*

Marinela Contreras, Paul D. Kasaija, Octavio Merino, Ned I. de la Cruz-Hernandez, Christian Gortazar and José de la Fuente


María Eugenia Zurita, Mieszko M. Wilk, Francisco Carriquiriborde, Erika Bartel, Griselda Moreno, Alicja Misiak, Kingston H. G. Mills and Daniela Hozbor

*83 A Multi-Stage* Plasmodium vivax *Malaria Vaccine Candidate Able to Induce Long-Lived Antibody Responses Against Blood Stage Parasites and Robust Transmission-Blocking Activity*

Jessica N. McCaffery, Jairo A. Fonseca, Balwan Singh, Monica Cabrera-Mora, Caitlin Bohannon, Joshy Jacob, Myriam Arévalo-Herrera and Alberto Moreno

*99 Chimeric Murine Polyomavirus Virus-Like Particles Induce* Plasmodium *Antigen-Specific CD8+ T Cell and Antibody Responses* David J. Pattinson, Simon H. Apte, Nani Wibowo, Yap P. Chuan, Tania Rivera-Hernandez, Penny L. Groves, Linda H. Lua, Anton P. J. Middelberg and Denise L. Doolan


# Editorial: Molecular Vaccines Against Pathogens in the Post-genomic Era

#### Angel Alejandro Oñate<sup>1</sup> \*, Yanmin Wan<sup>2</sup> and Alberto Moreno3,4

<sup>1</sup> Laboratory of Molecular Immunology, Department of Microbiology, Universidad de Concepción, Concepción, Chile, <sup>2</sup> Department of Infectious Disease, Huashan Hospital, Fudan University, Shanghai, China, <sup>3</sup> Emory Vaccine Center, Yerkes National Primate Research Center, Emory University, Atlanta, GA, United States, <sup>4</sup> Division of Infectious Diseases, Department of Medicine, Emory University, Atlanta, GA, United States

Keywords: vaccines, systems biology, multi-omics, immunoinformatics, modern vaccinology

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

#### **Molecular Vaccines Against Pathogens in the Post-genomic Era**

The development and implementation of vaccines are one of the greatest Public Health achievements in human history (Centers-for-Disease-Control-and-Prevention, 2011a,b). In the first decade of the twenty-first century, the Expanded Program on Immunizations averts more than 2.5 million deaths every year (WHO, 2009). Vaccines not only prevent deaths, disease, and disability but also provide community protection by reducing the spread of the disease within a population (Orenstein and Ahmed, 2017). In the US alone, it has been estimated that the prevention of clinical cases and deaths by vaccination for a single birth cohort represents a net savings of \$68.8 billion in total societal costs (Zhou et al., 2014). Although there are vaccines internationally available against 26 infectious diseases, nearly half of all deaths from infectious diseases are caused by pathogens for which no vaccine is available (Piot et al., 2019), including emerging and re-emerging pathogens (Williamson and Westlake, 2019). Interestingly, the majority of these vaccines have been developed empirically with limited information available on the mechanisms involved in protection (Pulendran and Ahmed, 2011). The development of high-throughput technologies and the advances in bioinformatics allow the massive generation and integration of datasets from multiple components of a biological system to understand in-depth physiological or pathological events (Pezeshki et al., 2019). This holistic approach of systems biology, when applied to studies of vaccine-induced immune responses, is known as system vaccinology (Pulendran et al., 2010). This research field will provide tools not only for the rational vaccine design but also for the development of novel adjuvants and vaccine delivery systems (Raeven et al., 2019). In this Frontiers Research Topic, some concepts of modern vaccinology are explored.

Edited and reviewed by: Nahed Ismail, University of Illinois at Chicago, United States

> \*Correspondence: Angel Alejandro Oñate aonate@udec.cl

#### Specialty section:

This article was submitted to Clinical Microbiology, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 12 November 2019 Accepted: 10 December 2019 Published: 22 January 2020

#### Citation:

Oñate AA, Wan Y and Moreno A (2020) Editorial: Molecular Vaccines Against Pathogens in the Post-genomic Era. Front. Cell. Infect. Microbiol. 9:443. doi: 10.3389/fcimb.2019.00443

#### INFLUENZA PANDEMIC PREPAREDNESS

In light of the new technological advances in systems biology, Short et al. reviewed host, pathogen, and environmental factors that determined the high mobility and mortality rates reported for the 1918 influenza pandemic. The progress in understanding the immune responses induced by influenza vaccines using system biology tools is reviewed by Sherman et al. The authors summarized the molecular signatures of B cell responses that have been reported to correlate with the response to vaccination. Signatures of long-lasting antibody responses and immunosenescence defined after influenza vaccination are also discussed. The availability of influenza challenge models combined with system biology tools has started to reveal differences in gene expression signatures in volunteers exposed either to H1N1 or H3N2. Furthermore, access to human challenge models has also shown the dynamics of the viral evolution within the host. Jang and Seong review Oñate et al. Vaccines in the Post-genomic Era

recent data on the advances and challenges on the development of universal influenza vaccines (UIVs). The search for a broadly protective vaccine capable of redirecting the immune responses from the variable immunodominant regions to conserved subdominant has identified potential candidates that can be delivered using novel vaccine platforms. Among the strategies, novel immunoinformatics tools to define consensus or ancestral sequences by phylogenetic analyses are discussed. Systems vaccinology approaches are needed to understand the mechanism of protections induced by UIVs.

### INTERACTOME MINING TO REVEAL PATHOGEN TARGETS

Characterizing the patterns of molecular interaction of proteins, using protein interaction networks or interactomes, is essential for understanding the cellular function and pathogen–host interactions. Mujawar et al. report the development of protein interactomes for Acinetobacter baumannii, a causative agent of nosocomial infections, to identify potential vaccine candidates and virulence factors for immunological or pharmacological targeting. The identified proteins are then mapped onto the whole genome protein interactome for in silico verification to generate a short list of proteins for future in vivo validation.

### ADJUVANT FORMULATIONS

The development of vaccines based on subunits is a promising strategy given its excellent safety profile. However, candidates are usually poorly immunogenic and require the formulation with potent adjuvants. In their article, Contreras et al. present new data on tick vaccination, using a combination of the recombinant subolesin from Rhipicephalus microplus described as protective for this arthropod in combination with heatinactivated Mycobacterium bovis and administered orally. The results of this study confirmed the efficacy of subolesin-based vaccines for the control of cattle tick infestations and expanded to oral vaccination using an immunostimulant. Ebensen et al. report the use of a combination of c-di-AMP, a STING agonist, and a promising adjuvant capable of stimulating an effective Th1/Th2 and cytotoxic immune response, with the well-known adjuvant alum. This adjuvant system was tested with a model antigen showing the induction of a balanced humoral and cellular immune responses.

## DELIVERY SYSTEMS

In the search for novel vaccine delivery systems, several strategies have been proposed. Zurita et al. address the need to develop more effective acellular pertussis vaccines. In eliciting tissueresident memory CD4+ T cells, a critical effector subset involved in protection, the authors show that an outer membrane vesicle (OMV)-based vaccine is more effective than a commercial acellular vaccine, having a better ability to induce protection and immunological memory. Chimeric virus-like particles (VLPs) using the murine polyomavirus are reported to be an effective platform to deliver subunit vaccines by Pattinson et al. In proof-of-concept studies, the authors produced chimeric VLPs genetically modified for surface expression of CD8+, CD4+, or B cell epitopes derived from the Plasmodium yoelii circumsporozoite protein. The vaccine platform was efficient to induce CD8+ T cell and antibody responses, but limited CD4+ T cell responses. The potential of using Porcine Circovirus Type 2 (PCV2) Chimeric VLPs for surface expression of exogenous peptides is presented by Wang et al.. The availability of threedimensional structural data of the PCV2 capsid protein allows the authors to use homology modeling to characterize surface displaying. The platform allows the insertion of foreign peptides without altering the virus assembly and its entrance to the host cell.

DNA vaccines consist of plasmid vectors that, after immunization, allows intracellular expression of the encoded antigens. Protective efficacy is achieved by the induction of a strong humoral and cellular immune response dependent on B and T cells. In a mini review by Mekonnen et al., the limitations and strategies for using DNA vaccines against human immunodeficiency virus (HIV)-1 and hepatitis C virus (HCV) in humans are discussed. The authors review the potential of DNA vectors for elicit protective compartmentalized CD8 + T cells in the liver for HCV and the genito-rectal mucosa for HIV.

Conjugation of poorly immunogenic antigens to carrier protein is the strategy used for glycoconjugate vaccines. McCaffery et al. reported the use of genetic conjugation to deliver a Plasmodium vivax sexual-stage vaccine candidate. The authors take advantage of a highly immunogenic chimeric protein that they designed targeting a blood-stage antigen to create a bifunctional vaccine by genetic linkage to the transmission-blocking vaccine candidate Pvs25. This approach addresses the need for the development of effective multi-stage malaria vaccines.

## CLOSING PERSPECTIVES

The wealth of information provided by systems biology approaches can be integrated into product design for the development of novel vaccines. This Research Topic offered a glimpse into some strategies in modern vaccinology. Although these tools are essential for the development of effective vaccines against agents with complex host–pathogen interactions such as HIV, tuberculosis, and malaria, from the global health perspective, it is also critical to identify and deal with factors associated with persistent social disparities. It has been estimated that only 5% of all children born worldwide receive all 11 vaccines recommended by the WHO (Mantovani and Santoni, 2018). It is therefore critical that in conjunction with research efforts to develop novel vaccines, global health initiatives such as the Global Alliance for Vaccines and Immunizations (GAVI) that promotes equal access to vaccines are strengthened (Ikilezi et al., 2019; Rappuoli et al., 2019; Zerhouni, 2019).

## AUTHOR CONTRIBUTIONS

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

#### FUNDING

AO received funded from the Fondo Nacional de Desarrollo Científico y Tecnológico, FONDECYT grants 1180122. AM

#### REFERENCES


received funded from the National Institutes of Health, NIAID grants R21 AI094402-01A1 and R21AI135711-01. YW received founded by project 81671636 and 81971559 from NSFC.


**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 Oñate, Wan and Moreno. 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.

# Structure-Based Design of Porcine Circovirus Type 2 Chimeric VLPs (cVLPs) Displays Foreign Peptides on the Capsid Surface

Dongliang Wang<sup>1</sup> , Sujiao Zhang<sup>1</sup> , Yawen Zou<sup>1</sup> , Wanting Yu<sup>1</sup> , Yifan Jiang<sup>1</sup> , Yang Zhan<sup>1</sup> , Naidong Wang<sup>1</sup> \*, Yanpeng Dong<sup>2</sup> and Yi Yang<sup>1</sup> \*

*<sup>1</sup> Hunan Provincial Key Laboratory of Protein Engineering in Animal Vaccines, Laboratory of Functional Proteomics, Research Center of Reverse Vaccinology, College of Veterinary Medicine, Hunan Agricultural University, Changsha, China, <sup>2</sup> Jiangsu Nannong Hi-Tech Co., Ltd, Jiangyin, China*

Although porcine circovirus-like particles can function as a vector to carry foreign peptides into host cells, displaying foreign peptides on the surface of virus-like particles (VLPs) remains challenging. In this study, a plateau, consisting of the middle portion of Loop CD (MP-Lcd) from two neighboring subunits of PCV2 capsid protein (Cap), was identified as an ideal site to insert various foreign peptides or epitopes and display them on the surface of PCV2 VLPs. One of the goals of this work is to determine if the surface pattern of this plateau can be altered without compromising the neutralizing activity against PCV2 infections. Therefore, biological roles of MP-Lcd regarding VLPs assembly, cell entry, and antigenicity were investigated to determine whether this was a universal site for insertion of foreign functional peptides. Three-dimensional (3D) structure simulations and mutation assays revealed MP-Lcd was dispensable for PCV2 Cap assembly into VLPs and their entry into host cells. Notably, substitution of MP-Lcd with a foreign peptide, caused surface pattern changes around two-fold axes of PCV2 VLPs based on 3D structure simulation, but was not detrimental to VLPs assembly and cell entry. Moreover, this substitution had no adverse effect on eliciting neutralizing antibodies (NAbs) against PCV2 infection in pigs. In conclusion, MP-Lcd of the PCV2 Cap was a promising site to accommodate and display foreign epitopes or functional peptides on the surface of PCV2 VLPs. Furthermore, chimeric VLPs (cVLPs) would have potential as bivalent or multivalent vaccines and carriers to deliver functional peptides to target cells.

Keywords: porcine circovirus type 2 (PCV2), chimeric virus-like particles (cVLPs), bivalent or multivalent vaccines, capsid protein, 3D structure prediction

## INTRODUCTION

Porcine circovirus (PCV), a member of the Circovirus genus in the Circoviridae family, has a small, single-stranded and circular DNA genome. There are two main genotypes, namely porcine circovirus type 1 (PCV1) and PCV2, both of which share high levels of nucleotide homology and a common genomic organization (Segales et al., 2013). PCV2, the key causative agent of porcine circovirus-associated diseases (PCVADs), causes severe economic losses worldwide. Based on genome or cap gene (it encodes the sole structural protein of PCV2 capsid) analysis of PCV2

#### Edited by:

*Alberto Moreno, Emory University School of Medicine, United States*

#### Reviewed by:

*Christian Nelson, SUNY Cortland, United States Karl William Boehme, University of Arkansas for Medical Sciences, United States*

\*Correspondence:

*Yi Yang yiyang@hunau.edu.cn Naidong Wang naidongwang@hunau.edu.cn*

Received: *23 April 2018* Accepted: *19 June 2018* Published: *09 July 2018*

#### Citation:

*Wang D, Zhang S, Zou Y, Yu W, Jiang Y, Zhan Y, Wang N, Dong Y and Yang Y (2018) Structure-Based Design of Porcine Circovirus Type 2 Chimeric VLPs (cVLPs) Displays Foreign Peptides on the Capsid Surface. Front. Cell. Infect. Microbiol. 8:232. doi: 10.3389/fcimb.2018.00232* isolates, PCV2 is divided into four main subtypes (PCV2a, PCV2b, PCV2c, and PCV2d) (Franzo et al., 2015). Furthermore, PCV2a is subdivided into five clusters (2A, 2B, 2C, 2D, and 2E), PCV2b into three clusters (1A, 1 B, and 1C), whereas PCV2c has only been reported in Denmark, with only three isolates in GenBank (An et al., 2007; Olvera et al., 2007). Recently, a novel PCV genotype was identified (Phan et al., 2016; Palinski et al., 2017; Zhang et al., 2017) and temporarily designated PCV3, as its genomic DNA only shared ∼30% nucleotide identities with PCV1 and PCV2 (Palinski et al., 2017). However, pathogenicity and potential lesions caused by PCV3 remain to be determined.

The circular genome of PCV2 contains 1766–1768 nucleotides, encoding two main viral proteins: the capsid protein (Cap) and virus replication associated proteins (Rep and its isoform of Rep', formed via alternative splicing) (Cheung, 2003). The PCV2 Cap, the sole structural protein of this virus, is composed of ∼233 residues, with 60 of the Caps capable of self-assembly into a virus-like particle (VLP) in vitro (Khayat et al., 2011). Furthermore, the VLP has been successfully used as the main antigen in commercial vaccines against PCV2 infection (Fachinger et al., 2008). In addition, PCV2 VLPs instead of virus, were exploited to study interaction of virus-host cells (Misinzo et al., 2006) and mechanism of virus entry into various swine host cell lines (i.e., PK15 and 3D/4 cell lines) (Misinzo et al., 2005, 2009).

The three-dimensional (3D) structure of the PCV2 Cap was elucidated by two research groups (Khayat et al., 2011; Liu et al., 2016). The Cap contains a typical jelly-roll fold composed of eight β-strands, connected via seven loops (Khayat et al., 2011). There are indications that these loops are responsible for assembly of VLPs and that they determine surface patterns of the PCV2 capsid (Khayat et al., 2011; Wang et al., 2016). Further, based on 3D structure analyses, loops exposed on the outer surface of the capsid have substantially diverged during PCV2 evolution (Wang et al., 2016). In addition, both the NH2- and carboxyl-terminus (NT & CT) of the Cap contain a non-structured stretch. The NT, rich in basic amino acids, has a typical nuclear localization signal (NLS), responsible for distribution of the PCV2 Cap in nucleus (Liu et al., 2001) and viral genome packaging (Khayat et al., 2011). Therefore, in a stable PCV2 capsid, the NT is located inside the capsid and interacts with viral genomic DNA, whereas the CT is exposed on the surface of the capsid, as illustrated in 3D structures (Khayat et al., 2011; Liu et al., 2016; Wang et al., 2016). Although both termini of the Cap have been engineered as targeted sites for insertions of or substitutions with foreign epitopes or functional peptides to generate chimeric VLPs (cVLPs) or chimeric PCV, foreign peptides located at the NT of the Cap may hide within PCV2 VLPs and consequently may not be displayed on the exterior surface of the capsid if the NT is not externalized. Chimeric PCV were constructed by insertion of various epitopes or tags into the 3'-terminal end of the cap gene (the CT of the Cap), and specific antibodies against these tags or foreign peptides were produced in response to these chimeric PCVs (Beach et al., 2011; Piñeyro et al., 2015a,b). Furthermore, maximal insertion size of foreign amino acid residues at the 3' terminal end of the cap gene has also been studied. Chimeric porcine circovirus (PCV) containing amino acid epitope tags in the C terminus of the capsid gene are infectious and elicit both anti-epitope tag antibodies and anti-PCV type 2 neutralizing antibodies in pigs (Beach et al., 2011).

Our laboratory previously designed and prepared PCV2 cVLPs to display a foreign epitope on the exterior surface of the cVLPs, based on 3D structural analyses of the PCV2 Cap and its assembly of the capsid (Hu et al., 2016). Several loops were considered for insertions of the GP5 epitope B of porcine reproductive and respiratory syndrome virus (PRRSV), although only Loop CD was an ideal site for insertion of the GP5 epitope B, since this insertion had no negative effects on either assembly or cell entry of cVLPs in vitro.

In this study, a plateau consisting of MP-Lcd around two-fold axes of PCV2 VLPs was used to determine, based on 3D structure analyses and simulations, its potential as a universal platform for insertions of various foreign peptides without compromising assembly and cell entry of PCV2 cVLPs. Notably, roles of MP-Lcd in PCV2 VLPs assembly, cell entry and antigenicity were evaluated.

## MATERIALS AND METHODS

Reagents, Genes, Plasmids, and Cell Lines Unless otherwise stated, all chemicals were purchased from Sigma Aldrich (St. Louis, MO, USA). PCV2 cap gene (GenBank accession number: JF504708) was optimized and synthesized, as we have described (Hu et al., 2016). Mutated PCV2 cap genes were constructed by overlapping polymerase chain reaction (PCR), as described (Hu et al., 2016). Resultant DNA fragments were subcloned into a pET100 vector (Invitrogen, Carlsbad, CA, USA; **Figure 2A**). After confirmation by DNA sequencing, recombinant plasmids were transformed into BL21 (DE3) competent cells (TransGen, Beijing, China) for protein expression.

### PCV2 Cap Expression, Purification, and Self-Assembly of VLPs in Vitro

Protein expression, purification, and VLPs assembly in vitro of PCV2 Cap wild type and its mutants were done as described (Zhang et al., 2016). For western blots, proteins from cell lysates were first separated by 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) (Bio-Rad, Hercules, CA, USA). Thereafter, proteins were transferred to a polyvinylidene difluoride membrane (PVDF; Life Technologies, Carlsbad, CA, USA) in a protein transfer device (Bio-Rad) at 80 V for 1 h. The membrane was then blocked with 3% bovine serum albumin (BSA; Roche, Basel, Switzerland) in phosphate-buffered saline (1×PBS, pH 7.4) for 1 h, and then primary antibody (rabbit anti-PCV2 Cap, 1:500) was incubated with the membrane overnight. A horseradish peroxidase (HRP) conjugated goat anti-rabbit IgG (1:5,000, Promega, Madison, WI, USA) was the secondary antibody and signals were visualized with High-sig ECL western blotting substrate (Tanon, Shanghai, China). Formation of VLPs was confirmed by transmission electron microscopy (TEM). The PCV2 VLPs were adsorbed on to carbon-coated copper grids for 10 min and stained with 1% phosphotungstic acid for 10 min. Subsequently, VLPs were

examined with TEM (CM100, Philips Electron Optics, Zurich, Switzerland).

## Cap Sequences Alignment and 3D Structure Simulations of PCV2 Cap and Capsid

Typical Cap sequences from PCV1, 2, and 3 were downloaded from GenBank (http://www.ncbi.nlm.nih.gov/). Subtype and cluster of each PCV2 isolate were confirmed via previous phylogenetic analyses (Zhan et al., 2016). Amino acid sequences were aligned using a web server (http://multalin.toulouse.inra. fr/multalin/). For each Cap mutant, a 3D structural model was generated via homology modeling using a crystal structure of PCV2 Cap (PDB accession number: 3R0R) as a template, with protein modelings done using Modeller (http://salilab. org/modeller/). Then, icosahedral structures (3D) of all capsid mutants were generated, based on the monomeric structure of the PCV2 Cap mutant, by applying a VMD 1.9 matrix transformation using Tcl script (Humphrey et al., 1996) and displayed with PyMol (Version 1.8.4.0, http://www.pymol.org/).

#### Preparation of Anti-PCV2 Cap Sera

Female New Zealand rabbits (n = 2), 2 mo old, were used to prepare anti-PCV2 Cap serum. Rabbits were initially immunized with 2 ml of PCV2 VLPs (wild type; 0.2 mg/ml) mixed with Alhydrogel adjuvant (InvivoGen, San Diego, CA, USA), and then similarly boosted twice more at 2-week intervals (Weeks 2 and 4). Blood samples were collected 6 wk after final immunization and serum was isolated and stored at −20◦C.

#### Indirect Immunofluorescence Assays (IFAs)

Cell entry of PCV2 VLPs were detected with IFAs, as described (Misinzo et al., 2005). PK15 cells cultured in 24-well plates were incubated with 2 µg of PCV2 VLPs in each well for 1 h. After incubation, PK15 cells were washed three times with PBS to remove unbound VLPs. As a control, PBS without VLPs was added to PK15 cells. For immunofluorescent staining, PK15 cells were fixed in 4% (wt/vol) paraformaldehyde in PBS at12 h post-inoculation of PCV2 VLPs. Cells were subsequently washed and permeabilized with 0.1 % Triton X-100 for 5 min at room temperature. Cells were washed three times with PBS, incubated with rabbit anti-PCV2 serum (1:1,000) for 1 h, and washed three times with PBS. Then, cells were incubated with fluorescein isothiocyanate (FITC) conjugated donkey antirabbit IgG (1:2,000, Life Technologies) for 1 h in the dark, and subsequently washed three times. Finally, cells were mounted with Prolong <sup>R</sup> Gold Anti-fade Reagent with 4′ ,6-diamidino-2 phenylindole (DAPI; Invitrogen) for nucleic acid staining and imaged with confocal microscopy (LSM 710 NLO & DuoScan System, Carl Zeiss, Germany).

#### Immunization and Serum Collection

Fifteen commercial 3-week-old pigs, serologically negative for PCV2 and porcine reproductive and respiratory syndrome virus (PRRSV), were used. These pigs were randomly allocated into three groups (five per group). Groups 1 and 2 were immunized with intramuscular injections of 200 µg of PCV2 wild type VLPs or cVLPs-M8 (**Figure 2A**) mixed with MontanideTM Gel 01 ST adjuvant (SEPPIC), respectively, whereas Group 3 was immunized with PBS only (negative control). Each pig was injected only once. Sera were collected 21 and 28 d postprimary immunization (dpi) and tested by indirect enzymelinked immunosorbent assay (ELISA) and PCV2 neutralization assays. All animal care and use protocols in this study were conducted in accordance with the guidelines of the Animal Care and Use Committee of Hunan Agricultural University and were approved by the Institutional Ethics Committee (IEC) of Hunan Agricultural University.

#### Indirect ELISA and PCV2 Neutralization Assay

PCV2 VLPs assembled from wild type PCV2 Cap (Cap WT) were coated as an antigen to detect specific anti-PCV2 antibodies by ELISA, as described (Zhang et al., 2016). The PCV2 neutralization assay was also performed as described (Fort et al., 2007). All serum samples from each group were heat-inactivated (56◦C for 30 min) and 50 µl of inactivated sera was serially twofold diluted with PBS, and then mixed with 50 µl of PCV2 virus (200 TCID50; GenBank accession number: KP112484) for 1 h at 37◦C. Then, serum-virus mixtures were inoculated to confluent PK-15 cells cultured in 96-well plates and incubated at 37◦C for 72 h. Cells were fixed with absolute ethyl alcohol for 30 min, 0.1% triton for 10 min and then blocked with 3% BSA for 1 h at room temperature. Thereafter, cells were incubated with rabbit anti-PCV2 serum (1:1,000) as the primary antibody and FITC-conjugated donkey anti-rabbit IgG (1:2,000 dilution, Life Technologies) as second antibody, each for 1 h. Cells were observed using a fluorescent microscope (Olympus IX73, Tokyo, Japan). The reciprocal of the highest dilution at >50% fluorescent focus reduction was the neutralizing antibody titer.

## RESULTS

### MP-Lcd From Two Neighboring Cap Subunits Formed a Plateau on the Two-Fold Axis of PCV2 Capsid

Loop CD, connecting β-strands C and D in PCV Cap, was a highly divergent loop among distinct genotypes of porcine circoviruses (PCV1, PCV2, and PCV3), based on amino acid sequence alignment (**Figure 1A**). There was no strictly conserved residue present in this loop among the three genotypes (**Figure 1A**). In contrast, 6 of 18 residues (81PP82, 84GG85, 87N, and 92P) between Caps of PCV1 and 2 were invariant (solid red dots at top of **Figure 1A**). In addition, 5 of the 6 conserved residues were located within the middle portion (eight residues from positions 80–87; red rectangle in **Figure 1A**) of Loop CD. Therefore, the portion composed of these eight residues was designated MP-Lcd (Middle Portion of Loop CD). In the 3D structure, Loop CD of the PCV2 Cap formed the second elevation on the exterior surface of the PCV2 capsid, with most residues in Loop CD located on surfaces of both the PCV2 Cap and the capsid (**Figures 1B–D**) (Khayat et al., 2011; Wang et al., 2016), whereas MP-Lcd formed a plateau on the surface of the two-fold

axes of the PCV2 capsid (red portions in **Figure 1D**). Of note, MP-Lcd was absent among PCV3 Caps (**Figure 1A**). Based on structural features and absence of MP-Lcd in PCV3 Cap, together with previous study that an epitope derived from PPRSV GP5 was successfully inserted Loop CD (between 85G and 86S of PCV2 Cap) (Hu et al., 2016). we inferred that MP-Lcd was an optimal target site for inserting foreign peptides and displaying them on the surface of the PCV2 capsid. To confirm this hypothesis, a series of PCV2 Cap mutants (Cap-M1-6 in **Figure 2A**) were designed to test tolerances of MP-Lcd to foreign peptides.

## MP-Lcd Had High Tolerances to Insertions of Various Foreign Peptides

Four peptides, including two epitopes derived from structural proteins of porcine epidemic diarrhea virus (PEDV) (Sun et al., 2008) and porcine parvovirus (PPV) (Sun et al., 2015), respectively (**Figure 2A**, Cap-M1 and -M2), and two artificial peptides (**Figure 2A**, Cap-M3 and -M4), were designed. Based on simulated 3D structures, these four peptides projected from a plateau formed by the MP-Lcd formed (orange in **Figure 3A**) after they were inserted into the PCV2 Cap. To test effects of insertions of various peptides in MP-Lcd on VLPs assembly and cell entry, Cap mutants (Cap-M1 to -M4, **Figure 2A**) were successfully expressed and purified (**Figures 2B,C**). Of note, these Cap mutants were capable of self-assembling into VLPs after in vitro dialysis (**Figure 3B**). Moreover, insertion of various peptides into MP-Lcd had no adverse effect on entry of these VLPs into PK15 cells (**Figure 3C**). Therefore, MP-Lcd tolerated insertions with various foreign peptides (Cap-M1 to 4), including those with highly charged residues (Cap-M3 and 4).

## Capacity for MP-Lcd to Accommodate Foreign Peptides

Having established that MP-Lcd allowed insertions of various foreign peptides without compromising in vitro assembly of VLPs and entry into PK15 cells, maximal insertion size was studied. Two new Cap mutants were designed. One contained an insertion of two distinct epitopes (derived from structural proteins of PEDV and PPV) spaced by a "GS" linker (Cap-M5, **Figure 2A**), whereas another had an insertion of three

FIGURE 2 | PCV2 Cap WT and mutants. (A) Residues within MP-Lcd of PCV2 Cap WT and mutants. Numbers indicate residue positions in PCV2 Cap and mutants. Sequence of MP-Lcd is shown in Cap-WT. Residues in blue represent foreign peptides inserted into MP-Lcd of PCV2 Cap, whereas residues in red indicate "GS" linkers between foreign peptides or foreign peptides and PCV2 Cap. Note: MP-Lcd is substituted by an epitope derived from PEDV spike protein in Cap-M8. (B) SDS-PAGE of purified Cap WT and Cap mutants via Ni-NTA affinity chromatography. Lane M, protein marker; expected recombinant Cap protein is indicated in each lane. (C) Western blots of supernatants from bacterial lysates containing expressed Cap-WT and various mutants. Lane M, protein marker; the expected specific band is present in each lane.

FIGURE 3 | PCV2 VLPs assembly and cell entry into PK15 cells. (A) Localizations of foreign peptides or epitopes on the surface of PCV2 capsid in simulated 3D models after assembly of PCV2 Cap-WT and mutants into VLPs. Images on bottom of (A) demonstrate (high magnification) 3D structures or orientations of foreign epitopes and artificial peptides on a plateau formed by two neighboring MP-Lcd. Foreign epitopes or artificial peptides were labeled in orange; MP-Lcd was labeled in red and Loop CD, except MP-Lcd was labeled in green and Loop BC-decorated five-fold axes of the icosahedra were labeled in blue on PCV2 capsid. (B) PCV2 VLPs formations observed by TEM. Negative staining of VLPs assembled from PCV2 Cap WT or mutants (bars = 100 nm). (C) Entry of PCV2 VLPs into PK-15 cells. Internalizations of PCV2 VLPs assembled from PCV2 Cap WT or Cap mutants (M1-M4) were confirmed by confocal microscopy. Inset in PCV2 Cap-WT represents a negative control in which PBS instead of VLPs was added into PK15 cell culture. Green fluorescence represents PCV2 Cap in PK15 cells. Nuclei (blue) of PK-15 cells were stained by DAPI.

FIGURE 4 | Insertions of large fragments had an adverse effect on PCV2 VLPs assembly. (A) Simulated 3D structures of foreign epitopes (Cap-M5 and M6) on the surface of the PCV2 capsid. Images on bottom of (A) show top views (high magnification) of foreign epitopes on a plateau formed by two neighboring MP-Lcd. Foreign epitopes were labeled in orange; MP-Lcd was labeled in red and Loop CD, except MP-Lcd was labeled in green and Loop BC-decorated five-fold axes of the icosahedra were labeled in blue on PCV2 capsid. (B) Morphologies of PCV2 Cap-M5 and -M6 after dialysis against assembly buffer. Morphologies were observed by TEM after samples were negatively stained.

repeats of the epitope derived from PPV, also spaced by two "GS" linkers (Cap-M6, **Figure 2A**). The simulated 3D structures demonstrated that both foreign peptides might be displayed on the outer surface of PCV2 VLPs (**Figure 4A**). In addition, both Cap mutants were also successfully purified after being expressed in bacteria (**Figures 2B,C**). After the two mutants were dialyzed against assembly buffer in vitro, morphologies were distinct from those of VLPs assembled from either the Cap WT or other Cap mutants (**Figure 4B**). Based on TEM results, we inferred that PCV2 VLPs assembly may have been adversely affected by 18 residues, in the absence of further optimization of linker or assembly conditions.

## MP-Lcd Was Dispensable for PCV2 VLPs Assembly and Cell Entry

To test functions of MP-Lcd on PCV2 VLPs assembly in vitro and cellular uptake, we designed two PCV2 Cap mutants (Cap-M7 and -M8 in **Figure 2A**), of which MP-Lcd was either deleted (M7) or replaced (M8) by a foreign epitope derived from a structural protein of PEDV spike protein (Sun et al., 2008). The 3D structure simulations revealed both PCV2 Cap mutants had a similar jelly-roll fold as wild type **(Figure 5A**). However, when Cap-M7 was assembled into a capsid, the plateaus composed of MP-Lcd around the two-fold axes had disappeared (**Figure 5A**). In contrast, a new surface pattern around the two-fold axes was formed in a capsid assembled by Cap-M8 (**Figure 5A**), due to replacement of MP-Lcd with the foreign epitope. Further, both mutants were expressed and purified via NTA-Ni2+ chromatography (**Figures 2B,C**). After dialysis in vitro, both were capable of assembling into VLPs with a diameter of ∼17 nm, as confirmed by TEM (**Figure 5B**), with very similar morphology as VLPs assembled from the WT of the Cap (**Figure 5B**). Based on IFA, VLPs assembled from both Cap mutants entered PK15 cells similar to wild type (**Figure 5C**). Therefore, MP-Lcd appeared dispensable for VLPs assembly in vitro and subsequent entry into PK15 cells.

### Replacement of MP-Lcd With a Foreign Epitope Had Minimal Effect on PCV2 Antigenicity and Production of PCV2-Specific NAbs in Pigs

Although replacement of MP-Lcd with the foreign peptide was not detrimental to PCV2 VLPs assembly and cell entry in vitro, effects of foreign peptide on immunogenicity of PCV2 VLPs had not been determined. Therefore, pigs were immunized with VLPs assembled either from the Cap-WT or the Cap-M8. Anti-PCV2 antibody was determined by indirect ELISA at 21 and 28 d post immunization (dpi). Antibody titers increased rapidly (P < 0.01) from 21 to 28 dpi, compared to the control group (**Figure 6**). Based on ELISA, PCV2 Cap-specific antibodies were successfully induced by PCV2 VLPs assembled from Cap-WT or from the Cap-M8, with no statistical difference in titers (**Figure 6A**). Further, PCV2 NAbs were detected at 21 and 28 dpi, and both anti-PCV2 Cap sera had neutralizing activity against PCV2 infections in vitro, whereas no NAbs were detected in the PBS group (**Figure 6B**). Notably, both anti-PCV2 Cap sera had consistent virus-neutralizing titers, which strongly indicated replacing MP-Lcd with the foreign peptide had no adverse effect on eliciting PCV2 Cap-specific NAbs in pigs.

## DISCUSSION

Inserting a foreign peptide into the middle of a protein often poses a huge challenge, as this may adversely affect protein folding, and thereafter affect protein functions. Although foreign peptides or tags are usually fused to the NT or CT of proteins to minimize effects on protein backbones, there is no assurance

that inserted foreign peptides or tags will be displayed on the surface of proteins or protein complexes (i.e., VLPs). However, PCV2 VLPs have been developed to carry foreign peptides, with fusion usually done on either the NT or the CT of the Cap (Li et al., 2013; Zhang et al., 2014). Since the NLS at the NT of the PCV2 Cap is not required for VLPs assembly (Khayat et al., 2011), several PCV2 cVLPs were successfully prepared by replacing the NLS with a T-cell epitope, B-cell epitope, or a T-cell epitope conjugated with a B-cell epitope of classical swine fever virus (CSFV). However, no neutralizing antibodies against CSFV were detected (Zhang et al., 2014). Perhaps these foreign epitopes fused at the NT of the PCV2 Cap were buried inside the PCV2 VLPs after assembly. However, an epitope located on the exterior surface of VLPs may be presented to B lymphocytes by binding to B cells receptor (BCR) (Bachmann and Jennings, 2010). Conversely, the CT of the PCV2 Cap is exposed on the surface of PCV2 VLPs (Khayat et al., 2011; Liu et al., 2016; Wang et al., 2016), and was used to fuse with foreign peptides, although no typical VLPs were observed by TEM (Hu et al., 2016). We also failed several times with fusion of distinct tags to the CT (data not shown). In our previous study, Loop CD exploited a target site to insert a B-cell epitope derived from PRRSV GP5 protein, and the recombinant Cap (rCap) was capable of self-assembling into VLPs in vitro (Hu

et al., 2016). Furthermore, cVLPs entered PK15 cells and induced NAbs against PCV2 and PRRSV infections. Loop CD of PCV2 Cap has several advantages over other loops of this protein for insertion of a foreign epitope (Hu et al., 2016). Recently, a novel genotype of PCV, named PCV3, was reported, although its pathogenesis and lesions remain to be determined (Phan et al., 2016; Palinski et al., 2017). Despite amino acid sequence identity between the Caps of PCV2 and PCV3 being only ∼30% (Palinski et al., 2017), the PCV3 Cap had a similar jelly-roll as the PCV2 Cap, present in many icosahedral viruses (Khayat and Johnson, 2011; Zhan et al., 2017). Surprisingly, MP-Lcd was missing in the PCV3 Cap (**Figure 1A**), which indicated this portion may not be necessary for Cap folding into a jelly-roll structure and VLPs assembly during PCV evolution. Loop CD of the PCV2 Cap is the third largest loop of the Caps with 18 residues and forms the second highest elevation on the capsid surface (Khayat et al., 2011). Furthermore, MP-Lcd in Loop CD contributes the entire plateau portion to the two-fold axis surface of the PCV2 capsid (**Figure 1D**). Simulated 3D structures of the PCV2 cVLPs assembled from the Cap mutants (M1-6) clearly demonstrated that MP-Lcd functioned as a platform to display foreign peptides outward (**Figures 3**, **4**, red). Of note, Loops CD of two neighboring Cap subunits were tightly aligned through their anti-parallel arrangement on the PCV2 capsid surface (Wang et al., 2016). Thus, two MP-Lcd were exactly side by side (**Figure 1D**, red), providing potential bivalent binding sites between foreign peptides (ligands) and their binding partners (receptors). Moreover, foreign peptides caused minimal alteration to the surface of PCV2 VLPs, since they projected from the plateau and were independently displayed. Thus, these foreign peptides might have no negative effect on assembly and antigenicity of PCV2 VLPs. Consequently, conformations of foreign peptides on the plateau also increased the likelihood of interactions with binding partners, e.g. BCR. By virtue of the MP-Lcd orientation within the PCV2 capsid, this is a promising target site for display of foreign peptides. In addition, retaining the plateau of MP-Lcd, instead of deleting it, projects foreign peptides further from the surface of the PCV2 cVLPs, based on the simulated 3D structures.

Further, we tested whether MP-Lcd was biased regarding amino acid composition of foreign peptides. Four PCV2 Cap mutants (Cap-M1 to M4) containing insertions with distinct foreign peptides were capable of self-assembly into VLPs in vitro (**Figure 3B**), and these VLPs retained their capacity to enter PK15 cells (**Figure 3C**). In particular, insertion of either poly-lysine or poly-glutamic acid residues into MP-Lcd (Cap-M3 and M4) had no effect on VLP assembly and cell entry (**Figures 3B,C**). Therefore, foreign peptides with highly charged residues may adopt very flexible conformations on MP-Lcd to minimize effects of charge repulsion on VLPs assembly. It also proves MP-Lcd exhibits high tolerance to insertions of various peptides. Indeed, 3D structure simulation also suggested both insertions independently protruded from the surface of MP-Lcd on PCV2 VLPs (**Figure 3**). Finally, maximal insertion size of MP-Lcd was tested. Two PCV2 rCap containing two distinct epitopes or three repeats of the same epitope in tandem, respectively, were successfully expressed and purified (**Figures 2B,C**). Morphologies of both assemblies had different sizes after rCap were dialyzed in assembly buffers (**Figure 4B**). Therefore, we concluded that morphology of PCV2 VLPs may be changed when the insertion size exceeded 18 residues in MP-Lcd. However, morphologies of PCV2 VLPs may also be changed in vitro through alterations of assembly conditions (e.g., temperatures and buffer), as reported for other VLPs (Tsukamoto et al., 2007).

Roles of MP-Lcd in PCV2 VLPs assembly, cell entry, and antigenicity were also investigated in the present study. Apparently, removal of MP-Lcd caused the plateau to disappear from the surface, but had no effects on VLPs assembly and cell entry. Therefore, we inferred that MP-Lcd may not be involved in recognition of virus-host cells for PCV2 internalization. Further, deleting MP-Lcd and replacing it with a foreign epitope (Cap-M8) introduced a new surface pattern around the two-fold axes of the PCV2 VLPs, since part of residues of the foreign epitope were present on the surface (orange in **Figure 5A**). Results from EM and IFA suggested that this Cap mutant was still capable of self-assembling into VLPs and entering PK15 cells. We also used this mutant to test effects of surface alteration around the two-fold axes on antigenicity of PCV2 VLPs in a swine model. Based on ELISA and neutralization experiments, we inferred that removal of MP-Lcd and alteration of the plateau around the two-fold axes of PCV2 VLPs had minimal effects on antigenicity and eliciting NAbs against PCV2 infection.

In conclusion, the MP-Lcd of Loop CD was identified as a very promising site within the PCV2 Cap to develop a functional carrier capable of displaying a variety of foreign peptides or epitopes on the surface of PCV2 VLPs. Of note, it was dispensable for assembly and cellular uptake of PCV2 VLPs. In addition, MP-Lcd was not required for eliciting NAbs against PCV2 infection. Therefore, development of PCV2 VLPs-based bivalent or multivalent vaccines against concurrent infections of PCV2 and other pathogens should be very feasible. Furthermore, based on structure simulation, MP-Lcd functioned as a plateau to project foreign peptides from the surface of PCV2 VLPs. Finally, although insertion size was tested, maximal size for MP-Lcd insertion still needs to be determined, as conditions may be further optimized for VLPs assembly in vitro.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

DW, NW, and YY are responsible for experimental design. DW, SZ, YZo, WY, YJ, YZh, and YD did experiments, collected and analyzed data. DW and YY wrote the manuscript. All authors read and approved the final manuscript.

#### ACKNOWLEDGMENTS

This project was supported by General Program of National Natural Science Foundation of China (Grants No. 31571432); Research Foundation of Hunan Provincial Education Department, China (Grant No. 15A086); Hunan provincial Natural Science foundation of China (Grant No.2018JJ2177); Lateral project funding (Grant No. 17430102000049 and 17430102001846); the Talent Input Project of Hunan Agricultural University (Grant No. 540490317006); and the Youth Science Foundation of Hunan Agricultural University (Grant No. 17QN10).


of porcine reproductive and respiratory syndrome virus. Virus Res. 213, 100–108. doi: 10.1016/j.virusres.2015.11.005


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

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

# Back to the Future: Lessons Learned From the 1918 Influenza Pandemic

Kirsty R. Short 1,2, Katherine Kedzierska<sup>3</sup> \* and Carolien E. van de Sandt 3,4 \*

*<sup>1</sup> School of Chemistry and Molecular Biosciences, The University of Queensland, Brisbane, QLD, Australia, <sup>2</sup> Australian Infectious Diseases Research Centre, The University of Queensland, Brisbane, QLD, Australia, <sup>3</sup> Department of Microbiology and Immunology, The Peter Doherty Institute for Infection and Immunity, University of Melbourne, Parkville, VIC, Australia, <sup>4</sup> Department of Hematopoiesis, Sanquin Research and Landsteiner Laboratory, Amsterdam, Netherlands*

2018 marks the 100-year anniversary of the 1918 influenza pandemic, which killed ∼50 million people worldwide. The severity of this pandemic resulted from a complex interplay between viral, host, and societal factors. Here, we review the viral, genetic and immune factors that contributed to the severity of the 1918 pandemic and discuss the implications for modern pandemic preparedness. We address unresolved questions of why the 1918 influenza H1N1 virus was more virulent than other influenza pandemics and why some people survived the 1918 pandemic and others succumbed to the infection. While current studies suggest that viral factors such as haemagglutinin and polymerase gene segments most likely contributed to a potent, dysregulated pro-inflammatory cytokine storm in victims of the pandemic, a shift in case-fatality for the 1918 pandemic toward young adults was most likely associated with the host's immune status. Lack of pre-existing virus-specific and/or cross-reactive antibodies and cellular immunity in children and young adults likely contributed to the high attack rate and rapid spread of the 1918 H1N1 virus. In contrast, lower mortality rate in in the older (>30 years) adult population points toward the beneficial effects of pre-existing cross-reactive immunity. In addition to the role of humoral and cellular immunity, there is a growing body of evidence to suggest that individual genetic differences, especially involving single-nucleotide polymorphisms (SNPs), contribute to differences in the severity of influenza virus infections. Co-infections with bacterial pathogens, and possibly measles and malaria, co-morbidities, malnutrition or obesity are also known to affect the severity of influenza disease, and likely influenced 1918 H1N1 disease severity and outcomes. Additionally, we also discuss the new challenges, such as changing population demographics, antibiotic resistance and climate change, which we will face in the context of any future influenza virus pandemic. In the last decade there has been a dramatic increase in the number of severe influenza virus strains entering the human population from animal reservoirs (including highly pathogenic H7N9 and H5N1 viruses). An understanding of past influenza virus pandemics and the lessons that we have learnt from them has therefore never been more pertinent.

#### Edited by:

*Yanmin Wan, Fudan University, China*

#### Reviewed by:

*Ali Ellebedy, Washington University in St. Louis, United States Daniel Angel Ortiz, Laboratory Corporation of America Holdings (LabCorp), United States*

#### \*Correspondence:

*Katherine Kedzierska kkedz@unimelb.edu.au Carolien E. van de Sandt cvandesandt@unimelb.edu.au*

Received: *10 August 2018* Accepted: *10 September 2018* Published: *08 October 2018*

#### Citation:

*Short KR, Kedzierska K and van de Sandt CE (2018) Back to the Future: Lessons Learned From the 1918 Influenza Pandemic. Front. Cell. Infect. Microbiol. 8:343. doi: 10.3389/fcimb.2018.00343*

Keywords: influenza, pandemic, 1918, host factors, viral factors, external factors, societal factors, prevention

## INTRODUCTION

In 1918 a mysterious and deadly disease spread around the world in three consecutive waves (spring 1918, autumn 1918, and winter 1918–19). This pandemic infected over one third of the world's population and killed an estimated 50 million people (Johnson and Mueller, 2002; Murray et al., 2006), with unusually severe clinical manifestations in previously healthy young adults (Collins, 1931; Hoffman, 2011). In 1918, the etiological agent that caused this disease was unknown (Hildreth, 1991). However, we now know that these events represented the largest influenza virus pandemic on record: the catastrophic 1918 influenza pandemic. Since 1918, the world has experienced three additional influenza pandemics: the 1957 "Asian" influenza pandemic, the 1968 "Hong Kong" influenza pandemic and the 2009 so-called "swine flu" pandemic. These pandemics, although mild in comparison to that of 1918, highlight the constant threat that influenza virus poses to human health. Given that almost 100 years have passed since 1918, it behooves us to ask: are we truly better prepared for the next influenza virus pandemic or are there still lessons to be learned? This review gives an overview of lessons learned from the 1918 influenza pandemic, highlighting new insights into our understanding of viral pathogenesis and their impact on our preparedness for the next outbreak of influenza.

## The Origins 1918 Influenza Virus

The 1918 influenza pandemic is often colloquially referred to as the "Spanish" influenza pandemic. However, it is unlikely that the 1918 influenza virus originated in Spain. Instead, influenza cases were widely reported in Spain due to the fact that, as a neutral country in World War I, Spain did not practice censorship in the press. In contrast, other countries involved in the war, such as Germany, Britain and France, most likely limited the news of this deadly pandemic, so as not to lower the moral of the troops and raise questions about their military readiness (Johnson, 2006). Today, the general consensus is that the 1918 influenza virus originated in the Midwest of the United States of America (Barry, 2004). Medical records reported the first cases of "influenza of a severe type" around March 1918 in military camps in Kansas (Barry, 2004). From here, the virus is thought to have spread throughout the United Stated and then transported by American troop ships to the battlefields of France, where it gradually spread throughout Europe and the rest of the world (Patterson and Pyle, 1991). The spread of the virus beyond port cities was further facilitated by local transport networks, predominately railways (Patterson and Pyle, 1991; Johnson, 2006). However, it is possible that the predecessor of this killer virus first entered human population prior to 1918 and became more virulent and/or more transmissible over time. Unusual influenza activity was already reported in the United States and several European countries before the first (spring) wave of the 1918 influenza outbreak (Frost, 1919; Johnson, 2006; Hoffman, 2011). Military camps in France already reported influenza infections accompanied with high mortality in the winter of 1916–17 (Hammond et al., 1917), which was followed 2 months later by a similar outbreak near London at Aldershot, one of Britain's biggest military camps (Oxford et al., 1999, 2002, 2005). Interestingly, no records of civilian influenza cases around that time exist, possibly because influenza cases were not recorded at the time or because they got lost with time. Alternatively, it is tempting to speculate that military camps, with their high population density, close proximity to livestock, high mobility, and large number of people with pre-existing lung conditions (due to exposure to toxic gasses in the trenches) served as the perfect breeding ground for the emergence of this catastrophic pandemic (Oxford et al., 2005).

Just as the geographic origins of the 1918 virus remain unclear, the original animal reservoir of the virus also remains controversial. As a segmented virus, influenza virus is capable of undergoing the process of reassortment. Reassortment occurs when two influenza virus strains co-infect the same cell, facilitating the emergence of a new "reassortant" virus which contains a novel constellation of genes. Reassortment between avian and human influenza viruses gave rise to the 1957 and 1968 influenza pandemics (**Figure 1**; Scholtissek et al., 1978; Kawaoka et al., 1989; Schäfer et al., 1993). In contrast, the 2009 influenza pandemic resulted from a reassortment event between avian, human and swine influenza viruses (**Figure 1**; Garten et al., 2009; Smith et al., 2009b). Unlike these more recent influenza pandemics, the 1918 virus is thought to have been directly introduced in the human population (i.e., in the absence of reassortment) from a single unidentified host (Taubenberger et al., 2005). This notion is supported by the fact that the 8 individual gene segments of the 1918 virus appear to have coevolved in the same host. However, the exact identity of this host remains unclear, as the nucleotide sequence of the virus is genetically distinct to all other known avian and mammalian influenza viruses (Reid et al., 2004a,b; Taubenberger et al., 2005). In contrast, others argue that the 1918 influenza virus could have indeed originated from a reassortment event between avian and mammalian, possibly swine and/or human, influenza viruses in the years prior to the 1918 pandemic (Smith et al., 2009a; Worobey et al., 2014). Unfortunately, in the absence of influenza virus sequence data in the years preceding the 1918 pandemic, this question may never be definitively answered.

## A Broad Spectrum of Clinical Disease

During the 1918 influenza pandemic, a broad spectrum of clinical illness was observed (Brundage and Shanks, 2008). In the first spring wave of the pandemic, disease was typically mild and mortality rates were not unusually high (Johnson and Mueller, 2002). However, there was a surprisingly large number of young adults who were affected by the outbreak (Ahmed et al., 2007). The second or autumn wave of influenza emerged in late August 1918 and by the end of 1918 almost no country was spared (Patterson and Pyle, 1991; Johnson, 2006). The striking feature of the autumn wave was its unprecedented virulence (Taubenberger et al., 2001). Patients typically suffered from a high fever, cyanosis, and fluid accumulation in the lungs (Johnson, 2006). In ∼5% of the fatal cases, death occurred rapidly after the onset of clinical symptoms (i.e., within 3 days), although for the majority of cases the time from clinical symptoms to death was ∼7–10 days (Brundage and Shanks, 2008). The third and final wave of influenza emerged at the start of 1919 (Beveridge, 1977). This

1918 H1N1 pandemic virus originated from multiple reassortment events between avian, swine and human influenza viruses, or if it was introduced by a direct zoonotic transmission event of an avian, swine or other influenza virus. The H1N1 virus continued to circulate, causing seasonal epidemics, until 1957 when it reassorted with an avian H2N2 virus. This virus circulated until 1968, when it reassorted again with the avian H3Nx virus, which has caused seasonal epidemics ever since. In 1977 the H1N1 virus was reintroduced in the human population and co-circulated with H3N2 viruses until the influenza pandemic of 2009 when it was replaced by another H1N1 virus which was the result of multiple reassortment events between avian, swine, and human influenza viruses.

wave was typically not as virulent as the fall wave and it did not affect every country (Beveridge, 1977; Taubenberger and Morens, 2006). During the course of the pandemic, ∼500 million people worldwide were infected, resulting in a case-fatality rate of >2.5% (Johnson and Mueller, 2002; Johnson, 2006; Taubenberger and Morens, 2006). While this fatality rate was certainly higher than other influenza virus pandemics (Taubenberger and Morens, 2006), these data indicate that influenza virus infection was not always fatal and that a large number of people survived the infection. These data raise two intriguing questions: (i) Why was the 1918 influenza pandemic more virulent than other influenza pandemics of the twentieth century; and (ii) Why did some people survive the pandemic and others succumb to the infection?

## VIRAL FACTORS ASSOCIATED WITH THE SEVERITY OF THE 1918 INFLUENZA PANDEMIC

It was not until 1933, more than a decade after the devastating pandemic of 1918–19, when the influenza virus was first isolated and demonstrated to be the causative agent of seasonal influenza virus infections (Smith et al., 1933). However, even then, an indepth understanding of the viral factors that contributed to the severity of the 1918 pandemic was thwarted by the absence of any available biological material from the virus in question. Finally, in the late 1990s the virus' genetic material was successfully isolated from formalin-fixed, paraffin-embedded lung tissue from 1918 influenza victims and from the lungs of a 1918 influenza victim buried in Alaska's permafrost (Taubenberger et al., 1997; Reid et al., 1999). These efforts unraveled a partial viral sequence from four viruses and the complete genomic sequence of one virus (Reid et al., 1999). The fully reconstructed 1918 influenza virus proved to be highly pathogenic in mice (Tumpey et al., 2005), ferrets (Memoli et al., 2009), and macaques (Kobasa et al., 2007). Interestingly, a recent study in ferrets demonstrated that the 1918 influenza virus could spread to, and induce cytokine responses in tissues outside the respiratory tract, which likely contributed to the severity of the infection (de Wit et al., 2018) and could explain the neurological complications observed during the 1918 influenza pandemic (Alexander, 1919; Ravenholt and Foege, 1982). Various reverse genetics experiments suggest that the high pathogenicity exerted by the 1918 influenza virus was most likely an interplay between different virulence factors, in which proteins encoded by the viral haemagglutinin (HA) and polymerase gene segments played a crucial role (Kobasa et al., 2004; Tumpey et al., 2005; Kash et al., 2006; Conenello et al., 2007; Pappas et al., 2008; Watanabe et al., 2009; Jagger et al., 2010).

One of the best-known virulence determinants of influenza virus is the presence of a multibasic cleavage site in the HA (Horimoto and Kawaoka, 1994; Subbarao et al., 1998). In avian species, influenza viruses without a multibasic cleavage site require the HA to be cleaved by host trypsin-like proteases for infection. Trypsin-like proteases are commonly found in the respiratory tract, thus limiting replication of these viruses to these tissues. However, the presence of the multibasic cleavage site means that these viruses can be cleaved by ubiquitously expressed proteases. The presence of a multibasic cleavage site in modern highly pathogenic avian influenza viruses can be associated with increased virulence in mammalian hosts (Schrauwen et al., 2012; Suguitan et al., 2012). However, none of the available 1918 HA sequences contained a multibasic cleavage site (Kawaoka and Webster, 1988; Taubenberger et al., 1997; Reid et al., 1999). Instead, analysis of the HA sequence of the 1918 viruses revealed that these viruses were adapted to bind to human epithelial cells. The influenza A virus HA protein requires only one amino acid change in order to switch binding to α-2,3-linked sialic acids (typically found on avian cells) to binding to the α-2,6 linked sialic acids (typically found on human epithelial cells in the upper respiratory tract) (Glaser et al., 2005). Compared to victims from the spring wave, an increased incidence of this mutation in the HA was observed in viruses isolated from victims of the more severe autumn wave (Reid et al., 2003; Glaser et al., 2005; Sheng et al., 2011). A second mutation, which strengthens the virus binding to the human receptor, could only be found in some of the 1918 HA sequences (Reid et al., 2003; Sheng et al., 2011). These data suggest that at least two H1N1 influenza viruses circulated in 1918, which differed in their binding affinity for the human receptor. Both viruses displayed a similar cell tropism in the respiratory tract of terminal stage human influenza victims (Sheng et al., 2011). However, this secondary adaptation is essential for effective transmission of the 1918 influenza virus between ferrets (Tumpey et al., 2007). Mutations in other gene segments, including PB2, PA, and PB1-F2 of the 1918 influenza virus can also play a role in host adaptation (Dunham et al., 2009; Jagger et al., 2010; Mehle et al., 2012; Mazel-Sanchez et al., 2018). In the absence of additional influenza virus sequences from 1918, it is hard to establish whether these or other mutations contributed to the dramatic increase in case-fatality seen during the autumn wave of the pandemic (Simonsen et al., 2018).

Gain- and loss-of-function experiments, such as those described above, have provided important insights as to how novel influenza viruses adapt to the human population (Subbarao et al., 1993; Mehle and Doudna, 2009; Herfst et al., 2012; Imai et al., 2012; Belser et al., 2013; Richard et al., 2013; Zhu et al., 2013; Watanabe et al., 2014). Specifically, this information is used to evaluate the pandemic potential of novel influenza viruses, including avian H7N9 and H5N1 viruses, which are frequently crossing the species barrier into the human population. The Global Influenza Surveillance and Response System (GISRS), a surveillance program that monitors which influenza virus strains circulate at a given time (Hay and McCauley, 2018; World Health Organization, 2018a), uses this information for a rapid risk assessment when a potentially pandemic virus is reported to circulate in animals (predominately birds or swine) or has crossed the species barrier into the human population. It is hoped that extensive surveillance activities, in combination with rapid clinical diagnosis, will afford us a "head start" in the case of a future influenza pandemic. However, the success of such surveillance programs is contingent upon their geographical breadth (Krammer et al., 2018). International cooperation and support for influenza surveillance will become even more pertinent in the future as climate changes continues to affect animal reservoirs and avian migration patterns, both of which could lead to the spread of influenza viruses to new locations and across a wider range of avian species (Klaassen et al., 2012; Shaman and Lipsitch, 2013; Audubon, 2018).

## HOST FACTORS ASSOCIATED WITH VARIATIONS IN INFLUENZA MORBIDITY AND MORTALITY IN 1918

The 1918 influenza pandemic is notorious for its high morbidity and mortality rates. However, it is important to recognize that there were substantial variations in mortality, both within and between countries (Mills, 1986; Johnson and Mueller, 2002; Johnson, 2006). General estimations assume an overall death rate of 2.5–5 per 1,000 individuals worldwide. Although this might be an accurate estimate for some countries [e.g., Australia (2.8/1000), Austria (3/1000), Demark (4.1/1000)], it represents an overestimation for some countries [e.g., Argentina (1.2/1000), Uruguay (1.4/1000), American Samoa (0/1000)], and a gross underestimation of others [e.g., Nauru (160/1000), Western Samoa (236/1000), Cameroon (445/1000)] (Johnson and Mueller, 2002; Johnson, 2006). These data indicate that in addition to viral factors, host factors had a major impact on the outcome of infection.

#### Age

An individual's age played a major role in determining one's risk of death during the 1918 influenza pandemic. Typically, when the mortality rates of seasonal influenza are graphed against the age of the population, a "U" shaped curve is produced, as the highest mortality occurs in the very young and old (Johnson, 2006). In contrast, pandemic outbreaks (to various degrees) are characterized by a shift in case-fatality toward younger age groups (Simonsen et al., 1998; Olson et al., 2005; Ahmed et al., 2007; Georgantopoulos et al., 2009). This was particularly pronounced during 1918 pandemic when young adults (15– 30 years) displayed such usually high mortality rate that a "W" shaped mortality curve was produced (Olson et al., 2005; Ahmed et al., 2007; Shanks and Brundage, 2012). The underlying mechanisms driving this mortality shift are not fully understood but are likely to be associated with the host's immune status.

#### Immunopathology

The high mortality rates observed in young adults during the 1918 pandemic has traditionally been attributed to the induction of an aberrant, dysregulated pro-inflammatory response (often referred to as a "cytokine storm"). This hypothesis is based upon experimental studies in various animal models using the reconstructed 1918 influenza virus. These experimental studies showed that the 1918 influenza virus triggered a potent, dysregulated pro-inflammatory response, which likely contributed to the severe lung lesions observed in victims of the 1918 influenza pandemic (Kobasa et al., 2004, 2007; Kash et al., 2006; Memoli et al., 2009; de Wit et al., 2018). Indeed, this dysregulated immune response has also been observed in natural and experimental infections with both the highly pathogenic avian H5N1 virus and the 2009 pandemic influenza virus (de Jong et al., 2006; To et al., 2010). However, it is important to note that all experimental 1918 influenza virus studies to date have been performed in immunologically-naïve animals. This is not necessarily indicative of the human situation in 1918, as influenza viruses caused epidemics and pandemics prior to 1918 (Dowdle, 1999; Johnson, 2006; Morens and Fauci, 2007; Morens et al., 2009; Valleron et al., 2010). It can therefore be assumed that a large proportion of the human population in 1918, with the possible exception of isolated countries/communities, would have encountered at least one previous influenza virus infection, resulting in pre-existing humoral and cellular immunity. It remains unclear whether such pre-existing immunity would cross-react with the 1918 H1N1 virus, and if so, whether it would enhance or dampen any dysregulated pro-inflammatory response in young adults.

### Humoral Immune Response

In contrast to young adults, older adults (aged 30–60 years) fared significantly better during the 1918 pandemic (Luk et al., 2001). This observation is likely to reflect the beneficial effects of pre-existing humoral immunity. It is theorized that an H1 and/or N1 influenza virus circulated in the human population prior to 1889, when it was replaced by a H3 influenza virus that caused the so-called "Russian" influenza pandemic (1889– 1892) (Ahmed et al., 2007). Accordingly, individuals born before 1889 (i.e., those 30 years or older during the 1918 pandemic) would have had cross-protective antibodies, while people born after 1889 would have been immunological naïve to the 1918 H1N1 pandemic virus (Dowdle, 1999; Ahmed et al., 2007). The lack of pre-existing 1918 influenza virus-specific or cross-reactive antibodies in children and young adults likely contributed to the high attack rate and rapid spread of the virus (Ahmed et al., 2007). Only people infected during the first "spring" wave of the pandemic acquired a protective immune response against the second, more virulent, "fall" wave of the 1918–19 pandemic (Gibbon, 1919; Shope, 1958; Palmer and Rice, 1992; Barry et al., 2008; Mathews et al., 2010; Shanks et al., 2010, 2011b; Fraser et al., 2011). Interestingly, unlike the majority of elderly populations worldwide, elderly populations in remote settings, including Indigenous Australians, Alaskan Natives, and Latin Americans, experienced high mortality during the 1918 pandemic. This most likely reflects the fact that these remote populations were not exposed to the previously circulating influenza viruses that conferred cross-protection (Ahmed et al., 2007).

Conclusive evidence that protective influenza virus-specific antibody responses are indeed long-lived came from the 2009 influenza pandemic. Here, elderly people who were exposed to the 1918 influenza virus (or its immediate descendant), 60– 90 years prior to the pandemic of 2009, were protected from infection and severe disease, as they maintained the antibody response that cross-reacted with the 2009 pandemic strain (Yu et al., 2008; Hancock et al., 2009; Ikonen et al., 2010; Reed and Katz, 2010).

Interesting, a recent study suggested that individuals medically-treated for influenza-like illness in the years prior to the 1918 pandemic (1916–1918) were actually at an increased risk of having clinically significant respiratory illness during the autumn wave of the 1918 pandemic (Shanks et al., 2016a). Similarly, the presence of cross-reactive but non-neutralizing antibodies, was associated with immune complex deposition and increased disease severity during the 2009 influenza pandemic (Monsalvo et al., 2011). These data suggest that protection against disease is dependent not just upon the presence of pre-existing antibodies, but rather their ability to neutralize the influenza virus strain in question.

## Cellular Immune Response

The issue of prior influenza virus exposure in the general population prior to 1918 raises the question as to why a preexisting cellular immune response, in particular cross-reactive CD8<sup>+</sup> T cells, offered so little protection to young adults during the 1918 influenza pandemic?

A robust CD8<sup>+</sup> T cell response plays an important role in protection against novel influenza virus strains and subtypes. Unlike antibodies, CD8<sup>+</sup> T cells can recognize the internal proteins of influenza virus. Since these internal proteins do not undergo rapid antigenic change, CD8<sup>+</sup> T cells are able to provide cross-protection against a broad range of different influenza virus strains. Accordingly, pre-existing influenza virus specific CD8<sup>+</sup> T cells provided protection against severe disease during the influenza pandemics of both 1957 and 2009 (Slepushkin, 1959; McMichael et al., 1983; Epstein, 2006; Sridhar et al., 2013; Hayward et al., 2015). In addition, seasonally induced influenza virus-specific CD8<sup>+</sup> T cells can cross-react with novel potentially pandemic avian influenza viruses (Kreijtz et al., 2008; Lee et al., 2008; van de Sandt et al., 2014) and facilitate more rapid recovery in patients following infection with low pathogenic H7N9 avian influenza virus (Wang et al., 2015). The presence of conserved CD8<sup>+</sup> T cell peptides in the viral protein sequences of the 1918 influenza virus (Quiñones-Parra et al., 2014) and the ability the of the 2009 H1N1 pandemic influenza virus to recall influenza virus-specific CD8<sup>+</sup> T cells, cross-reacting with the 1918 H1N1 influenza virus (Gras et al., 2010) suggest that pre-existing CD8<sup>+</sup> T cells should have been protective against severe infection with the 1918 H1N1 influenza virus, especially in the case of young adults. CD8<sup>+</sup> T cells may not have been optimal in very young children (age 0–4 years) due lack of exposure to previous influenza viruses (Bodewes et al., 2011a; Sauerbrei et al., 2014). Similarly, immunosenescence may have impaired CD8<sup>+</sup> T cell function in the elderly (>65 years) (Goronzy and Weyand, 2013). This may partially explain the high mortality observed in the youngest and oldest age groups during seasonal epidemics (Ushaped curve). However, individuals between 15 and 65 years of age, who suffered the greatest burden of disease during the 1918 pandemic, are thought to display the "gold-standard" immune response, with optimal cross-reactive CD8<sup>+</sup> T cell responses. The absence of protective immunity in this age group is unlikely to be due to the fact that heterosubtypic immunity is shortlived (Mathews et al., 2010), as the longevity of influenza virusspecific CD8<sup>+</sup> T cells in healthy individuals has recently been demonstrated (van de Sandt et al., 2015a). It is possible that the recall of pre-existing influenza virus-specific CD8<sup>+</sup> T cell responses was not rapid enough for the extremely virulent 1918 pandemic virus, causing rapid appearance of clinical disease and death within 3 days (Ahmed et al., 2007). Alternatively, pandemic H1N1 influenza viruses (1918 and 2009) may have suppressed immunogenic RIPK3-driven dendritic cell death needed for the induction of an effective CD8<sup>+</sup> T cell response (Hartmann et al., 2017). In addition, it is likely that ethnically defined genetic variations in HLA molecules influenced cross-reactive CD8<sup>+</sup> T cell responses in influenza virus infected individuals (Quiñones-Parra et al., 2014). This (combined with other socioeconomic factors) would leave some ethnicities, like Alaskan Natives and Indigenous Australians, more susceptible to severe influenza virus infections. Indeed, alarmingly high morbidity and mortality rates were observed amongst these populations during the pandemics of 1918 (Ahmed et al., 2007) and 2009 (La Ruche et al., 2009; Flint et al., 2010). Similarly, it is striking to note that the matrix protein of the 1918 virus already contained extraepitopic amino acid residues that were associated with evasion from the pre-existing influenza virus CD8<sup>+</sup> T cells (van de Sandt et al., 2015b), a phenomenon not observed in the comparatively mild 2009 pandemic influenza virus (van de Sandt et al., 2018a,b).

Finally, it is important to note that the highest influenza virus infection rates in 1918 were observed among children of school age (5–15 years). However, this increased infection rate occurred in the absence of high morbidity (Shanks and Brundage, 2012; Mamelund et al., 2016). Thus, school-aged children are thought to be in a "honeymoon period" of superior immunity, whereby they display increased resistance to a variety of different bacterial and viral pathogens (Ahmed et al., 2007). However, the mechanisms underlying such superior immunity are completely unexplored, despite the fact that it holds key information for inducing effective immune responses to influenza viruses.

In this review, we would like to propose an additional hypothesis that might have influenced the effectiveness of the cross-reactive cellular immune response and possibly contributed to the disproportional mortality amongst young adults in 1918; namely immune suppression as a resulting from recent measles infections (Moss et al., 2004; Griffin, 2010; de Vries et al., 2012; Mina et al., 2015). Measles epidemics were frequently reported at the end of the nineteenth and in the early twentieth century (Cliff et al., 1983; Duncan et al., 1997; Shulman et al., 2009; Shanks et al., 2011a, 2014), including a large measles outbreak in the US military camps in the winter of 1917–1918 (Shanks et al., 2014; Morens and Taubenberger, 2015). The elderly population would have experienced measles in their childhood and their immunity would have protected them from contracting measles in the years prior to the 1918 influenza pandemic. However, children and young adults, without prior measles infections, would have been immunologically susceptible to measles in the years preceding the 1918 influenza pandemic. Recent studies have demonstrated that the measles virus infects memory T lymphocytes, resulting in apoptosis and a prolonged state of immune suppression up to 3 years after the initial measles infection (Moss et al., 2004; Griffin, 2010; de Vries et al., 2012; Mina et al., 2015). Influenza virus-specific CTL responses could have been suppressed in young individuals who had to endure a measles infection in the years prior to the 1918 influenza pandemic, which may have increased their susceptibility to a severe influenza infection. The combination of recovering from immunosuppression and an infection with an unexpectedly highly virulent virus might have contributed to severe inflammatory related pathology in a mechanism better known as the immune reconstitution inflammatory syndrome (IRIS) (Hirsch et al., 2004; Morens and Fauci, 2007; Shulman et al., 2009; Barber et al., 2012). Whether recent measles infections indeed led to immunosuppression of the influenza virus-specific T cell responses, resulting in a higher susceptibility for severe influenza virus infections and potential IRIS or alternatively contributed to dampening CD8<sup>+</sup> T cell immunopathology remains an important area of future research. Fortunately, measles vaccines are now widely available and have greatly reduced the prevalence of measles worldwide (Moss and Griffin, 2012; Perry et al., 2014; Mina et al., 2015). However, the increased number of measles outbreaks in recent years and declining vaccination rates represent a key point of concern for future influenza virus pandemics.

Together, these data demonstrate that an individual's age (and the associated differences in their immune response) played an important role in determining disease outcome in the context of pandemic influenza virus infections. In 2009, age and pre-existing humoral immunity were taken into account when identifying priority individuals for vaccination. In 2009, the elderly population were less susceptible to severe influenza (Dawood et al., 2012) as they were protected through crossreactive antibodies and CD8<sup>+</sup> T cells acquired during previous seasonal infections, including the antigenically related A/H1N1 virus that circulated prior to 1957 (Yu et al., 2008; Hancock et al., 2009; Ikonen et al., 2010). Based on these findings, the first limited 2009 influenza vaccine stocks were administered to younger individuals, instead of being misdirected to the traditional high risk group: the elderly (National Center for Respiratory Diseases, CDC, and Centers for Disease and Prevention (CDC), 2009). Improving cross-reactive CD8<sup>+</sup> T cell responses to influenza vaccinations and natural infections remains a key research priority for the future (Clemens et al., 2018). This includes an understanding of CD8<sup>+</sup> T cell functionality in ethnically diverse populations and different age groups (Clemens et al., 2018).

#### Genetic Differences

In addition to the role of humoral and cellular immunity, there is a growing body of evidence to suggest that individual genetic differences contribute to differences in the severity of influenza virus infections. For example, during the 2009 influenza pandemic several single-nucleotide polymorphisms (SNPs) were strongly associated with severe pneumonia. These included SNPs in the genes for interferon response factor 7 (Ciancanelli et al., 2015), Fc fragment of immunoglobulin G, low-affinity IIA, receptor (Zúñiga et al., 2012), RPA interacting protein (Zúñiga et al., 2012), complement component 1q subcomponent binding protein (Zúñiga et al., 2012), CD55 (Zhou et al., 2012), IL-1α (Liu et al., 2013), IL-1β (Liu et al., 2013), surfactant protein B gene (To et al., 2014) and interferon induced transmembrane protein 3 (Everitt et al., 2012; Zhang et al., 2013), and IRF9 (Hernandez et al., 2018). Unfortunately, there is insufficient information available to conclude whether mortality variations in 1918 were influenced by any of the aforementioned SNPs. Defining which SNPs confer increased susceptibility to severe influenza remains an important aspect of influenza pandemic preparedness, as it will help to inform which populations are most at risk of severe disease.

#### Malnutrition

Host nutritional status has long been recognized as an important factor in the outcome of a variety of different infectious diseases (Cohen, 2000). In India in 1918, the effects of malnutrition and famine on influenza severity were particularly pronounced. The 1918 influenza pandemic hit India during a widespread drought, which affected the viability of many important food crops (Mills, 1986). Consequently, many Northern-Western, Western and Central Indian provinces experienced a famine during 1918 (Mills, 1986). It was these provinces which also experienced the highest 1918 influenza mortality rates (Mills, 1986). Due to the unusual age distribution of the pandemic, those who succumbed to the disease were typically young adults who formed the majority of the agricultural labor force (Mills, 1986). The resultant labor shortage only served to exacerbate the severity of the influenza pandemic (Mills, 1986). The exact mechanisms by which malnutrition and famine increase the severity of influenza remain to be defined. However, experimental studies suggest that not only does malnutrition suppress the host's immune response to influenza virus, but that it may also facilitate the emergence of novel viral variants, which display increased pathogenicity relative to the original parental strain (Beck et al., 2004).

Undernutrition (often exacerbated by ongoing civil conflicts) remains a problem for influenza pandemics of the twenty-first century and beyond. Indeed, chronic malnutrition was thought to have contributed to the high morbidity and mortality seen in Guatemalan children during the 2009 influenza pandemic (Reyes et al., 2010). Climate change may result in crop failures and exacerbate any food shortages in the future. However, in any future influenza virus pandemic, we will face a "double burden" of malnutrition whereby a proportion of the world's population will experience severe disease because of undernutrition and a proportion of the world's population will experience severe disease because of overnutrition. Specifically, it is now well accepted that obesity increases one's risk of being hospitalized with, and dying from, an influenza virus infection (Morgan et al., 2010; Louie et al., 2011; Van Kerkhove et al., 2011). Perhaps of even greater concern is the fact that obesity inhibits both virus-specific CD8<sup>+</sup> T cell responses and antibody responses to the seasonal influenza vaccine (Sheridan et al., 2012). The challenge for future influenza pandemics is therefore not only to protect those affected by undernutrition (in particularly in light of the growing problem of climate change), but also the growing number of people living with obesity.

## UNDERLYING INFECTIONS

## Co-infection With Bacterial Pathogens

Historical autopsy reports and examination of lung tissue sections from 1918 to 19 influenza case material indicated that for a significant number of patients, the cause of death was not primary viral pneumonia (Brundage and Shanks, 2008; Morens et al., 2008; Chien et al., 2009). Instead, these individuals succumbed to a secondary bacterial infection, most commonly pneumonia caused by bacteria such as Streptococcus pneumoniae, Haemophilus influenzae, Staphylococcus aureus, and Streptococcus pyogenes (Morens et al., 2008). H. influenzae was so frequently observed in influenza patients that it was often cited as the cause of the pandemic (and was thus named accordingly) (Hildreth, 1991). The role of secondary bacterial infections during the 1918 pandemic is consistent with epidemiological observations that while influenza virus attack rates in 1918 were similar among soldiers and civilians, mortality rates were much higher amongst newly arrived soldiers (Shanks et al., 2016b). The unhygienic circumstances in the army camps led to frequent bacterial infections, especially amongst immunologically naïve new army recruits. Thus, following an influenza virus infection, new army recruits were more likely to develop a lethal secondary bacterial pneumonia than civilians or long-serving soldiers (Shanks et al., 2010, 2016b). These observations have been echoed by numerous experimental animal studies, showing that coinfection with influenza virus and bacterial pathogens results in increased disease severity compared to infection with either pathogen alone (Brightman, 1935; Glover, 1941; Francis and de Torregrosa, 1945; Harford et al., 1946; Wilson et al., 1947; Short et al., 2012a, 2013). Different mechanisms have been proposed to explain this viral-bacterial synergism (McCullers, 2006; McAuley et al., 2007; Smith et al., 2013; Hrincius et al., 2015). These include, but are not limited to, reduced mucociliary clearance of inhaled bacteria following influenza virus infection, bacterial adhesion to the basement membrane (Morens et al., 2008; Taubenberger et al., 2012; Chertow and Memoli, 2013) and/or sialic acids exposed by influenza virus (McCullers and Bartmess, 2003; Peltola et al., 2005), viral alterations to the host immune response (Navarini et al., 2006; van der Sluijs et al., 2006; Ballinger and Standiford, 2010; Nakamura et al., 2011; Ellis et al., 2015; Lee et al., 2015) and the bacterial inhibition of epithelial cell repair following initial damage by influenza virus infection (Kash et al., 2011). Importantly, experimental studies suggest that influenza viruses not only increases the severity of secondary bacterial infections, but that it also increases the transmission of S. pneumoniae (Diavatopoulos et al., 2010; Short et al., 2012b).

In addition to co-infections with bacterial pathogens such as S. pneumoniae, chronic bacterial infections, such as those with Mycobacterium tuberculosis, contributed to variations in influenza mortality during the 1918 pandemic. For example, data from a Swiss sanatorium during the 1918 pandemic suggested that the risk of influenza death was higher among tuberculosis (TB) patients than non-TB controls (Oei and Nishiura, 2012). Similarly, individuals with TB were 2.2 times more likely to contract the 1918 influenza virus than non-TB individuals living in the same household (Noymer and Garenne, 2000; Noymer, 2011). A synergistic relationship between M. tuberculosis and influenza viruses has also been supported by experimental studies (Redford et al., 2014). The predominance of TB amongst young adults in 1918 may have contributed to the striking "W shaped" mortality curve associated with the 1918 influenza pandemic (Oei and Nishiura, 2012)**.**

Severe complications and morbidity as a result of bacterial co-infections were not unique to the 1918 influenza pandemic. Rather, bacterial co-infections were also observed in the influenza pandemics of 1957, 1968 and 2009, albeit to a lesser extent than in 1918 (Oswald et al., 1958; Robertson et al., 1958; Louria et al., 1959; Oseasohn et al., 1959; Chertow and Memoli, 2013; Joseph et al., 2013). In the 2009 influenza pandemic, TB was also identified as a risk factor for the development of severe disease (Morales et al., 2017). Thankfully, unlike in 1918, the severity of bacterial infections during these more recent influenza pandemics was likely minimized by the use of antibiotics, advanced medical care and the availability of bacterial vaccines [such as the pneumococcal polysaccharide and H. influenzae type b (Hib) vaccine] (Oswald et al., 1958; Robertson et al., 1958; Louria et al., 1959; Madhi et al., 2004; Wahl et al., 2018). However, as the rate of antibiotic resistance continues to rise and as pathogens such as methicillin-resistant S. aureus (MRSA) (Memoli et al., 2008) and multidrug-resistant M. tuberculosis (Zumla et al., 2013; Millard et al., 2015) become more common, we potentially face a future where antibiotics will be ineffectual in the treatment of bacterial infections. This will have direct and severe implications for any future influenza virus pandemic (Memoli et al., 2008). It must be considered an urgent priority to not only minimize antibiotic resistance, but also to invest in the discovery of new antibiotics and alternative treatment options for bacterial infections.

#### Malaria

In addition to individuals with bacterial co-infections, mortality during the 1918 influenza pandemic was considerably higher amongst malaria-infected individuals (Langford and Storey, 1992; Afkhami, 2003; Shanks, 2015). Although the underlying mechanism is not fully understood, a malaria-induced procoagulant state could play a role in increasing inflammation and subsequent clinical outcome (Shanks, 2015).

Today, chemopreventive strategies have lowered the disease burden associated with malaria and new eradication strategies are being developed. However, malaria still causes significant worldwide morbidity and mortality, there is ever increasing drug resistance and new malaria vaccines have yet to provide long-lasting benefits at a population level (Ashley et al., 2018). Until effective control measures have been developed and implemented, areas endemic for malaria remain at high risk for increased mortality during the next influenza pandemic.

## NON-PHARMACEUTICAL INTERVENTIONS

In 1918, a variety of different approaches were employed to limit the spread of influenza virus and treat infected patients. While many of these methods were of little or no avail, they contain important lessons for influenza pandemic preparedness in the twenty-first century.

#### Maritime Quarantine

In 1918, when the severity of the second wave of influenza became apparent, many countries imposed strict quarantine measures on all incoming ships to try and prevent the spread of influenza (Johnson, 2006). For the most part, these attempts were unsuccessful (Johnson, 2006). Quarantine measures were either implemented too late and the virus was already present within the country or quarantine was breached by infected individuals who were not yet symptomatic (Crosby, 1976; Tomkins, 1992). Thus, countries such as the U.K. and South Africa dismissed maritime quarantine as impractical and ineffectual (Blakely, 2006; Johnson, 2006). However, Australia imposed the maritime quarantine before any victims of the second wave were reported. All arriving vessels had to be cleared by Commonwealth Quarantine Officials before disembarking. This quarantine protected Australia from the second wave of the pandemic until December 1918 when the quarantine was finally breached. Maritime quarantine thus helped to protect Australia from the worst of the pandemic (Crosby, 2003; Johnson, 2006) and indirectly contributed to protecting certain Pacific Islands that depended on Australian supply ships (Shanks et al., 2018).

The most striking example of this was the mortality difference between American Samoa and Western Samoa. A strict maritime quarantine was imposed in American Samoa by the U.S. Governor in 1918 (Shanks and Brundage, 2013). This quarantine prevented influenza from entering the country, and no deaths from the 1918 influenza were ever recorded in America Samoa (Johnson, 2006; Shanks and Brundage, 2013). This was in sharp contrast to the nearby Western Samoa (located ∼100 km away), which did not practice strict maritime quarantine (Tomkins, 1992; Shanks and Brundage, 2013). As a result, Western Samoa was infected by the New Zealand supply ship, the Talune, and it is estimated that influenza killed more than a quarter of the population (Tomkins, 1992).

Global transportation has experienced a major transformation in the last century, with ships being replaced by the faster and more widely used air travel. The rise of commercial air travel helps explains the rapid global spread of the more modern influenza pandemics of 1957, 1968, and 2009 in the absence of major military troop movements (Rvachev and Longini, 1985; Hufnagel et al., 2004; Khan et al., 2009; Bajardi et al., 2011; Lemey et al., 2014). Accordingly, maritime quarantine is unlikely to play a role in limiting the spread of any future influenza pandemic. However, in 2009 authorities tried to limit the spread of influenza by using the modern-day equivalent of maritime quarantine: airport arrival screening. Unfortunately, analysis of arrival passengers at Sydney airport in 2009 suggested that airport screening had only a sensitivity of 6.67% for detecting influenzainfected patients, while costing ∼\$50,000 AUD per case detected (Gunaratnam et al., 2014). This limited efficacy likely reflects the fact that individuals infected with influenza virus can be contagious prior to becoming symptomatic (Hollingsworth et al., 2006). Airport arrival screening is therefore unlikely to control the spread of influenza by international air travel. Rather, advancing modeling (e.g., identifying which travel routes are most vulnerable to disease spread) and a variety of different education campaigns (e.g., raising awareness amongst the general public about the risks of traveling if they have been exposed to an infected individual) are likely to play a more significant role in future pandemic preparedness.

#### Mass Gatherings

In addition to limiting maritime travel, in 1918 most cities implemented simple non-pharmaceutical interventions to restrict the viral spread. These included imposing restrictions on social gatherings where person-to-person transmission could occur. As a result, schools, theaters, churches, and dance halls were closed, while mass gatherings such as weddings and funerals were banned in order to prevent overcrowding (Frost, 1919; Johnson, 2006; Bootsma and Ferguson, 2007; Hatchett et al., 2007). The peak death rate was lower in cities that rapidly implemented these non-pharmaceutical interventions within a few days after the first local cases were recorded, compared to those which waited a few weeks to respond (Bootsma and Ferguson, 2007; Hatchett et al., 2007). The timing when these interventions were lifted also affected the overall mortality (Bootsma and Ferguson, 2007; Hatchett et al., 2007). Thus, while restrictions on gatherings of people helped reduce influenza virus transmission, as soon as these restrictions were relaxed (typically within 2–8 weeks of their implementation) efficient viral transmission recommenced (Hatchett et al., 2007).

Following the outbreak of the 2009 pandemic influenza virus in Mexico, an 18-day period of mandatory school closure was implemented in the greater Mexico City area (Chowell et al., 2011). This was associated with a 29–37% reduction in influenza transmission (Chowell et al., 2011). Similarly, in Hong Kong there was approximately a 25% reduction influenza virus transmission following secondary schools closures from June 11 to July 10, 2009 (Wu et al., 2010). However, just as in 1918, the duration of these intervention strategies affected their efficacy, and there was a dramatic increase influenza activity in 32 Mexican states in the autumn of 2009, a period which coincided with schools opening for the autumn term (Chowell et al., 2011).

#### Facemasks and Hygiene

Facemasks were a popular preventative measure employed during the 1918 pandemic. While people were unsure of the etiological agent of the pandemic, the consensus was that it was an airborne disease and wearing a facemask would prevent infections (Crosby, 1976). Accordingly, many cities and regions, including Guatemala City, San Francisco, and certain prefectures of Japan, made wearing a facemask in public places obligatory, and special task forces and education campaigns were established to enforce this regulation (Crosby, 1976; Rice and Palmer, 1993; Rice, 2011). However, in order for a facemask to be at least partially effective against influenza virus it must be (i) worn at all times, (ii) properly made and fitted, and (iii) made of appropriate material. The surgical gauze masks of 1918 often failed to meet these criteria (Crosby, 1976). Thus, the mortality rate of Ontario, Canada (where wearing a mask was voluntary) was not significantly different from Alberta, Canada, (where mask wearing was enforced by law) (MacDougall, 2007). In fact, influenza deaths in Alberta continued to rise even after mask wearing was sanctioned by law, suggesting that in 1918 wearing a facemask was not sufficient to prevent deaths from influenza (World Health Organization Writing Group et al., 2006).

Proper hygiene (e.g., frequent hand washing) would also have helped limiting the spread of the influenza virus during the 1918 pandemic, as influenza viruses are transmissible via hand to face contact (World Health Organization Writing Group et al., 2006; Thomas et al., 2014). Thus, the Japanese traditional attitude to disease and illness might have contributed to a lower national pandemic mortality in 1918–19, as Japanese children are taught to remove their shoes and wash their hands upon re-entering the home (Rice and Palmer, 1993).

In the context of modern influenza pandemics, facemasks and handwashing/hand sanitizers have been used as preventative, non-pharmaceutical interventions. However, during the 2009 influenza pandemic for the most part, the use of facemasks was not obligatory (CDC, 2009). Rather, the CDC only recommended facemasks for individuals at increased risk of severe illness from influenza and/or individuals who were the direct careers of persons infected with the pandemic virus (CDC, 2009). Moreover, the effectiveness of facemasks in preventing the transmission of influenza virus remains unclear (Cowling et al., 2010) and just as was observed in 1918, low public compliance significantly limits the utility of facemasks in a modern pandemic setting (Cowling et al., 2010). Perhaps such interventions will be of greatest relevance to medical personal, who serve in the front line of a pandemic and are at high risk for infection. In contrast, handwashing and the use of hand sanitizers (whether or not in combination with wearing a facemask) had a clear protective effect during the influenza pandemic of 2009 (Larson et al., 2012; Suess et al., 2012; Wong et al., 2014).

These data suggest that non-pharmaceutical interventions such as social distancing, handwashing/hand sanitizers, and facemasks in any future influenza pandemic may buy valuable time before vaccines become widely available. However, the success of these interventions will depend upon their early and continuous implementation and also people's willingness to comply. The 1918 pandemic has shown that measures are most effective when they are voluntary, as people have low tolerance for mandatory health measures (Spinney, 2017). Indeed, a behavioral study showed that individuals were more likely to wear a facemask when they received autonomy-supportive advice as compared to controlled instructions (Chan et al., 2015).

### MEDICAL INTERVENTIONS, THERAPIES AND VACCINES: THEN AND NOW

The 1918 influenza pandemic occurred during a period in history when controlling infectious diseases had become a realistic goal of the medical profession (Tomkins, 1992; Johnson, 2006). Public health initiatives had already proven successful in limiting the spread of diseases such as cholera and TB (Hildreth, 1991; Tomkins, 1992; Tognotti, 2003). Thus, there was initially little to suggest that an influenza outbreak could

not be effectively controlled (Hildreth, 1991; Tomkins, 1992; Tognotti, 2003). However, despite the dramatic advances in microbiology in the previous decades, the etiological agent of the 1918 influenza pandemic remained a mystery. In the absence of clear information about the causative agent of the pandemic, a range of different therapeutic and preventative treatments were attempted. People experimented with medications (including Asprin) and homemade remedies such as mustard poultice, quinine, tobacco, beef tea, the inhalation of zinc sulfate, opium, salt water, and alcohol (Rice and Palmer, 1993; Johnson, 2006; Starko, 2009; Keeling, 2010). Traditional eastern medicine, like the Japanese Kanpo medicine (consisting of herbal remedies accompanied by green tea) may have had some beneficial effect by stimulating perspiration (helping to reduce fever), improving vitamin C levels and replacing lost fluids (Palmer and Rice, 1992; Rice and Palmer, 1993). Similarly, the use of Traditional Chinese Medicine may have reduced the severity of influenza infections in at least some individuals (Kobayashi et al., 1999; Cheng and Leung, 2007; Chen et al., 2011). However, for the most part little was available in terms of effective therapeutic and/or prophylactic treatments. Nursing care actually proved to have contributed to the recovery of patients, especially those suffering from secondary bacterial pneumonia (Robinson, 1990; Rice and Palmer, 1993). In addition, mortality rates were significantly higher in places deprived from nursing care, e.g., mining compounds (Phimister, 1973; Rice and Palmer, 1993). Unfortunately, during the 1918–19 pandemic many of those who typically performed these duties were either serving overseas or were sick themselves (Crosby, 2003; Keeling, 2010; Shanks et al., 2011b).

Today, the identification of the etiological agent of influenza has dramatically improved diagnostic speed and accuracy. Rapid and highly accurate molecular diagnostic techniques have largely replaced the labor intensive and time consuming "gold standard" cell culture method for diagnosing influenza virus infections (Ellis and Zambon, 2002), which allows for rapid isolation of infected individuals. Furthermore, risk assessment of potentially pandemic viruses has greatly improved by screening the viral genome of human and animal virus isolates for the presence of mutations that increase human adaptation and/or virulence.

In addition, we are able to deploy both anti-viral drugs and vaccines in the case of an influenza virus pandemic. Antivirals (such as the neuraminidase inhibitors oseltamivir and zanamivir) can be used as a therapeutic in severely ill patients, while also being employed prophylactically in outbreak situations (Cooper et al., 2003; Hayden et al., 2004; De Clercq, 2006; Zambon, 2014; Krammer et al., 2018). At present, potentially pandemic influenza viruses (such as avian H7N9 and H5N1 viruses) are sensitive to both oseltamivir and zanamivir (Herfst et al., 2012). However, acquired resistance to oseltamivir has been observed in several H5N1 isolates (De Clercq, 2006). Similarly, oseltamivir resistance is known to emerge in H7N9 viruses within just 2 days from the start of treatment (Hay and Hayden, 2013). These data suggest that in the case of any future influenza virus pandemic, antivirals should be used judiciously, and the emergence of drug-resistant viral variants closely monitored.

Influenza virus vaccines have also played a major role in reducing the morbidity and mortality associated with seasonal influenza. Unfortunately, antibodies elicited by seasonal influenza vaccines do not provide protection in the case of an antigenically distinct influenza virus of a novel subtype, such as A/H5N1 or A/H7N9 (De Jong et al., 2000). Furthermore, current inactivated seasonal influenza vaccines may even prevent the induction of cross-reactive CD8<sup>+</sup> T cell responses, which are our primary protection in case of a pandemic outbreak and may therefore prove to be a double-edged sword (Bodewes et al., 2009a,b, 2011b,c). Rapid vaccine production also remains a challenge for future influenza virus pandemics (World Health Organization, 2005; Rockman and Brown, 2010; Pada and Tambyah, 2011). This was particularly apparent during the 2009 pandemic when sufficient amounts of the vaccine against the pandemic virus were only available in October 2009, well and truly after the pandemic had spread globally (Butler, 2010). Vaccine production in a pandemic scenario may be further complicated by the fact that some avian influenza viruses can kill the embryonated chicken eggs needed for vaccine production (Tumpey et al., 2005). Novel vaccines strategies, in combination with alternative vaccine production platforms are needed to accelerate vaccine production and circumvent such problems (Schotsaert and García-Sastre, 2014). However, an influenza vaccine that offers long-lasting, broad-spectrum immunity remains the gold standard for pandemic preparedness. How basic fundamental humoral and cellular biology and human clinical data can be considered for the implementation of a universal influenza vaccine has recently been reviewed (Clemens et al., 2018).

#### CONCLUDING REMARKS

It is estimated that if a pandemic influenza virus were to reappear today, with a similar virulence and attack-rate as the 1918 influenza virus, mortality could rise to 21–147 million (Murray et al., 2006; Madhav, 2013). However, the high morbidity and mortality rates associated with the 1918 influenza pandemic resulted from a complex interplay between factors intrinsic to the 1918 virus itself, the host's immune response and the social context in which the pandemic struck. It is thus unlikely that this exact combination of factors would repeat itself in the future. Nevertheless, a comprehensive understanding of the factors that contributed to the severity of the 1918 pandemic plays an important role in our preparedness for the next influenza pandemic (**Figure 2**).

Today, we are better prepared for the next influenza virus pandemic than we were 100 years ago. Global influenza surveillance programs have been established to constantly monitor whether influenza viruses cross the species barrier into the human population (Hay and McCauley, 2018; World Health Organization, 2018a; Ziegler et al., 2018). This has already resulted in improved management strategies (Krammer et al., 2018) and the preventive slaughter of vast numbers of poultry that were infected with potentially pandemic viruses, such as H5N1 and H7N9 viruses (Oi, 2018). In addition, an improved understanding of the host-adaptation of influenza viruses and the existence of pre-existing immunity are likely to

However, lessons learned from the 1918 influenza pandemic will ensure that we are better prepared.

contribute to a more accurate predication of viral severity even before the influenza virus in question becomes established as a pandemic (Kreijtz et al., 2008; Lee et al., 2008; Herfst et al., 2012; Imai et al., 2012; Richard et al., 2013; Quiñones-Parra et al., 2014; van de Sandt et al., 2014; Wang et al., 2015). A better understanding of the human immune response against (pandemic) influenza viruses will eventually aid the development of broad-protective influenza vaccines (Clemens et al., 2018). However, in the interim, the majority of countries have established a pandemic preparedness program, which defines the precautionary measurements needed to be taken in case of an emerging viral pandemic (van Genugten et al., 2001; RIVM, 2018; World Health Organization, 2018b). These programs include surveillance, diagnostics, screening of passengers traveling from a potential outbreak region, quarantine procedures, stockpiling antibiotics, antivirals, bacterial and viral vaccines and the distribution of medical supplies (Brundage, 2006; Memoli et al., 2008; Mossad, 2009; World Health Organization, 2018b). We have also learnt from the 2009 H1N1 pandemic that it is important to have a somewhat flexible approach to pandemic preparedness, which allows countries to develop and implement their own risk assessments based on the global assessments provided by the WHO (Rudenko et al., 2015). However, good communication between countries and the WHO remains essential. In addition, it will be important for governments and those in authority to gain public trust before the next major pandemic outbreak. This will ensure that the public knows what to expect, how to act and is likely to improve compliance with preventative measures in a pandemic scenario. The internet represents a powerful and effective tool to disperse such information (Little et al., 2015).

Despite the advances that we have made in pandemic preparedness over the last 100 years, there are also several new challenges that we face in the context of twenty-first century (and later) influenza pandemics. Today's population demographic is dramatically different to that of 1918. Today, a large percentage of the world's population is either elderly (Morens et al., 2008; Mossad, 2009; Murray and Chotirmall, 2015) and/or living with one or more chronic medical conditions [such as heart disease, obesity, asthma, chronic obstructive pulmonary disease (COPD), and/or diabetes mellitus] (Morens et al., 2008; Jain et al., 2009; La Ruche et al., 2009; Flint et al., 2010; Hulme et al., 2017). The number of immunosuppressed individuals (due to untreated HIV infection, transplantation or/and chemotherapy) is also increasing (Jain et al., 2009; Kunisaki and Janoff, 2009; Sheth et al., 2011). This changing population demographic is of significance as each one of these host factors is known to increase the severity of even mild influenza virus infections. Mitigating the severity of future influenza pandemics will be further complicated by the prevalence of antibacterial resistance (Memoli et al., 2008), an increasing negative attitudes toward vaccination for other infectious diseases (such as measles and the pneumococcal polysaccharide and Hib vaccine) (Moss and Griffin, 2012; Perry et al., 2014) and an increase in seasonal

#### REFERENCES


influenza vaccinations of healthy individuals affecting the cross-reactive immune response otherwise induced by natural influenza virus infections (Bodewes et al., 2009b, 2011b). The high prevalence of underlying infections in less economically developed countries (such as TB and HIV), coupled with an underprepared health care system, places less economically developed countries at particularly high risk of severe morbidity during future influenza virus pandemics (Murray et al., 2006). These effects may even be compounded by the impacts of climate change which will lead to food shortages, famine and migration of climate refugees (van Schaik and Bakker, 2017).

At present, it is impossible to predict which influenza virus strain will cause the next pandemic. However, the growing number of human infections with the avian H7N9 virus represents a point of concern (especially in light of the approximate 40% mortality rate of this virus in humans). Like previous influenza pandemic viruses, human H7N9 virus infections have thus far displayed multiple waves of infections and shown signs of adaptation to human hosts (Zhu et al., 2017). Although this virus has yet to display efficient humanto-human transmission (Chen et al., 2013), it serves as a timely reminder that even though it has been 100 years since the 1918 pandemic, influenza pandemic preparedness remains of a paramount importance.

#### AUTHOR CONTRIBUTIONS

CvdS planned the review. KS, KK, and CvdS wrote the manuscript.

#### FUNDING

CvdS has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 792532. KS is supported by an Australian Research Council DECRA (DE180100512). KK is supported by a NHMRC Senior Research Fellowship Level B (GNT#1102792).


Beveridge, W. I. (1977). The start of pandemics: site, season and spread. Dev. Biol. Stand. 39, 443–444.


interferon amplification in human IRF7 deficiency. Science 348, 448–453. doi: 10.1126/science.aaa1578


recognition of influenza A viruses by M158-66 epitope-specific CD8<sup>+</sup> T cells is determined by extraepitopic amino acid residues. J. Virol. 90, 1009–1022. doi: 10.1128/JVI.02439-15


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

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

# The Combination Vaccine Adjuvant System Alum/c-di-AMP Results in Quantitative and Qualitative Enhanced Immune Responses Post Immunization

Thomas Ebensen† , Simon Delandre† , Blair Prochnow, Carlos A. Guzmán† and Kai Schulze\* †

Department of Vaccinology and Applied Microbiology, Helmholtz Centre for Infection Research, Braunschweig, Germany

#### Edited by:

Alberto Moreno, Emory University School of Medicine, United States

#### Reviewed by:

Elizabeth B. Norton, Tulane University, United States Yang Zhang, University of Pennsylvania, United States

#### \*Correspondence:

Kai Schulze kai.schulze@helmholtz-hzi.de

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Clinical Microbiology, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 16 August 2018 Accepted: 31 January 2019 Published: 19 February 2019

#### Citation:

Ebensen T, Delandre S, Prochnow B, Guzmán CA and Schulze K (2019) The Combination Vaccine Adjuvant System Alum/c-di-AMP Results in Quantitative and Qualitative Enhanced Immune Responses Post Immunization. Front. Cell. Infect. Microbiol. 9:31. doi: 10.3389/fcimb.2019.00031 The development of new effective vaccines strongly depends on adjuvants and formulations able to stimulate not only strong humoral responses against a certain pathogen but also effector as well as memory CD4+ and CD8+ T cells (Dubensky et al., 2013). However, the majority of vaccines licensed for human use or currently under clinical investigation fail to stimulate efficient cellular responses. For example, vaccines against hepatitis B virus (HBV), human papillomavirus (HPV), diphtheria, tetanus and influenza are usually administered by intramuscular (i.m.) injection and contain aluminum salts (alum) as adjuvant. Alum has been shown to stimulate Th2 immune cells resulting in increased production of antigen-specific antibodies but to be incapable of stimulating robust Th1 or cytotoxic responses. To overcome such limitations recent research has focused on the development of adjuvant combinations (e.g., MF59, AS03 or AS04) to not only further strengthen antigen-specific immune responses but to also allow their modulation. We have shown previously that bis-(3′ ,5′ )-cyclic dimeric adenosine monophosphate (c-di-AMP) constitutes a promising adjuvant candidate stimulating both effective Th1/Th2 and cytotoxic immune responses when included in mucosal or parenteral vaccine formulations. In the present work we demonstrate that c-di-AMP can be also combined with other adjuvants like alum resulting in increases in not only humoral responses but more striking also in cellular immune responses. This leads to improved vaccine efficacy against intracellular pathogens.

Keywords: adjuvant system, combination, c-di-AMP, alum, humoral, cellular

## INTRODUCTION

Today, infectious diseases represent the second leading cause of death worldwide (Global Health Observatory Data Repository, 2016). In order to prevent human illness and death, vaccination is currently the most effective tool. According to the World Health Organization (WHO), vaccination prevents 2 to 3 million deaths each year. Moreover, this number could even reach 6 million, if all children would receive the recommended vaccine schedule (Patil and Shreffler, 2018). Currently, vaccination allows us to control up to 10 major diseases and has resulted in the eradication of the smallpox virus in 1980. Although vaccination has demonstrated its strength to protect against infectious diseases, the emergence of new pathogens, as well as the increase of antibiotic resistance, reveals the necessity for the development of new vaccines. Presently, the vaccines in use are based on either live-attenuated pathogens, inactivated whole pathogens (virus or bacteria) or only pure microbial components. The latter, so-called subunit vaccines, constitute promising candidates for the development of vaccines showing increased safety profiles. Since subunit vaccines contain no living organism, these vaccines are especially useful for vaccination of immunocompromised individuals. However, the low complexity profile of subunit vaccines makes them less immunogenic. Booster immunizations and/or the inclusion of adjuvants are/is therefore required. In this context, adjuvants are not only used to enhance the stimulated antigen-specific immune responses but also to tailor the immune responses according to the specific clinical needs. In this regard, it is unlikely that a single adjuvant will be able to fulfill all the required properties to be implemented in all foreseeable vaccines. Presumably, different adjuvants are needed that stimulate the immune responses required following different vaccination strategies considering the pathogen, the type of antigen, the immune status and age of the vaccine and the application route. Recent approaches also address the possibility to combine different adjuvants in order to improve vaccine efficacy (Garcon and Di Pasquale, 2017). Especially the stimulation of a cell-mediated immune [T helper 1 (Th1) response and cytotoxic T lymphocytes (CTLs)] is of interest since actual vaccines stimulate predominantly humoral immune responses (Riese et al., 2013; Lee and Nguyen, 2015; Tandrup Schmidt et al., 2016). However, despite significant progress in adjuvant development during the last decades only a limited number of adjuvants are available for human use (Di Pasquale et al., 2015). Therefore, the aim of the present work was to evaluate the potential of the STING agonist c-di-AMP to increase vaccine efficacy when combined with the well-known adjuvant alum. In order to achieve this goal immunization studies were performed using the model antigen beta-galactosidase allowing an in depth dissection of the immune effector mechanisms stimulated by this system. Alum is the most used adjuvant worldwide and represents one of five adjuvants approved in the United States (FDA, 2011). However, alum stimulates only antigen-specific Th2 immune cells resulting in the secretion of IL-4, IL-5, and IL-10 and the subsequent improved antigenspecific antibody production (Brewer et al., 1999). In contrast, the cyclic di-nucleotide c-di-AMP, a second messenger in prokaryotes, exhibits strong immune modulatory properties stimulating antibody and mixed Th1/Th2 as well as cytotoxic responses when administered by either parenteral or mucosal routes (Ebensen et al., 2007a,b; Ebensen et al., 2017; Schulze et al., 2017a). This renders it very attractive for use in human vaccines, since most adjuvants supporting a Th1-dominated response lack the ability to induce humoral immunity (Libanova et al., 2010; Matos et al., 2017). Moreover, c-di-AMP is also able to promote the stimulation of CTL responses by induction of cross-priming (Lirussi et al., 2017). Therefore, an adjuvant system of alum/c-di-AMP could overcome the limitation of alum and not only further enhance the stimulated antigen-specific humoral response but at the same time promote the stimulation of Th1 and CTL responses.

## MATERIALS AND METHODS

#### Mice

Female BALB/c (H-2d) mice 6–8 weeks of age were purchased from Harlan Germany and kept at the animal facility of the Helmholtz Centre for Infection Research (Germany) under specific pathogen-free conditions as previously described (Schulze et al., 2017a). The animal experiments in this study have been reviewed for ethical compliance by the institutional ethical board and approved by the local government of Lower Saxony (Germany, No. 33.42502-13/1281). All experiments in this study were performed following standard biosecurity and institutional safety guidelines.

#### Immunization Protocol

Animals (n = 10) were immunized 3 times at day 0, 14, and 28 by intramuscular route. Each animal received a dose of 50 µl containing 15 µg of β-Gal protein (Sigma-Aldrich, Germany) as antigen. ß-Gal was either adsorbed to alum [1:1 v/v, aluminum hydroxyphosphate (Adju-Phos <sup>R</sup> ), Brenntag Biosector, Denmark] at pH 7.4 and 25◦C or co-administered with c-di-AMP (Biolog, Germany) at a concentration of 5 µg per dose. Fourteen days after the third immunization, spleens of vaccinated mice were collected, immune cells were extracted, pooled and restimulated with β-Gal. The cytokine concentration was measured by cytometric bead array (CBA). Results from one representative out of two independent experiments are shown.

#### Elisa

β-Gal-specific antibody titers in sera were investigated using ELISA assay as previously described (Schulze et al., 2017a). In brief, high binding protein plates were coated with β-Gal protein (2µg/ml in 0.05 M carbonate buffer). After blocking unspecific binding sites using 3% bovine serum albumin (BSA) in PBS serial 2-fold dilutions of sera in 3% BSA/PBS were added (100 µl/well). After 1 h incubation at 37◦C, plates were washed using 1% BSA/PBS/0.05% Tween 20 and the secondary antibodies were added: biotinylated goat anti-mouse IgG, IgG1, and IgG2a (Sigma, USA), respectively. After 1 h incubation at 37◦C, plates were washed and samples were incubated for 1 h at RT in the presence of peroxidase-conjugated streptavidin (BD Pharmingen, USA). Finally, reactions were developed using ABTS [2, 20-azino-bis(3- ethylbenzthiazoline-6-sulfonic acid)] in 0.1 M citrate-phosphate buffer (pH 4.35) containing 0.01% H2O2. Endpoint titers are expressed as absolute values of the last dilution giving an optical density (OD405 nm) being two times higher than the values of the negative control after 5 min incubation as previously described (Ebensen et al., 2007b).

#### ELISpot Assay

The quantity of β-Gal-specific cytokine-producing cells was investigated using an ELISpot assay as previously described (Lirussi et al., 2017; Schulze et al., 2017b). Flat bottomed 96 well plates with a 0.45µm hydrophobic High Protein Binding Immobilon-P-Membrane (BD Pharmingen) were coated with anti-IFN-γ, anti-IL2, anti-IL4 or anti-IL17 antibodies diluted in PBS and incubated overnight at 4◦C. Unspecific binding sites were blocked for 2 h at RT using 200 µl/well of complete

medium. Then, 4 × 10<sup>5</sup> and 2 × 10<sup>5</sup> spleen cells/well were added and incubated in the absence (blank, only media added) or presence of the β-Gal protein (5µg/ml) and the MHC-I immunodominant peptide TPHPARIGL of β-Gal (5µg/ml), respectively. For positive controls, splenocytes were stimulated with 5µg/ml of the mitogen concanavalin A. Samples were incubated for 16 (IFN-γ) or 48 h (IL-4) at 37◦C. Afterwards, plates were washed and further incubated for 2 h at RT in the presence of appropriate diluted biotinylated detection antibodies. Then, after another washing step samples were incubated for 1 h at RT in the presence of peroxidase-conjugated streptavidin. After a final wash, cytokine-secreting cells were detected by adding AEC substrate (diluted in 0.1 M acetate buffer pH 5.0) mixed with 0.05% H2O<sup>2</sup> (30%). After stopping the reaction with distilled water, plates were analyzed using the ImmunoSpot Image Analyzer software v3.2 (CTL-Europe GmbH). Results are expressed as Spot Forming Units (SFU) obtained from stimulated cells subtracted of background from non-stimulated cells (Ebensen et al., 2007b).

## Proliferation Assay

The ability of immune cells derived from spleen to proliferate upon restimulation with β-Gal as well as their cytokine profile were measured 96 h post restimulation. To this end, cell suspensions were seeded at 5 × 10<sup>5</sup> cells/well in flat-bottomed 96-well microtiter plates (Nunc) and incubated for 4 days in the presence of 1, 10, 20, and 40µg/ml of the β-Gal protein. During the final 18 h of culture, 1 µCi of [3H]-thymidine (Amersham International, Freiburg, Germany) was added to each well. Cells were harvested on paper filters (Filtermat A; Wallac, Freiburg, Germany) by using a cell harvester (Inotech, Wohlen, Switzerland) and the number of proliferating cells was indirectly determined by counting [3H]-thymidine events incorporated into the DNA of proliferating cells with a γ-scintillation counter (Wallac 1450, Micro-Trilux) (Lirussi et al., 2017; Schulze et al., 2017b). Results are expressed as ratio of values from stimulated and nonstimulated samples [stimulation index (SI)].

## Multiplex Flowcytomix (Cytometric Bead Array)

Supernatants of antigen-restimulated spelenocytes have been used to characterize the stimulated cytokine profiles using the Th1/Th2/Th9/Th17 FlowCytomix immunoassay from Biolegend according to the manufacturer's instructions (Ebensen et al., 2017).

## Statistical Analysis

Statistical significance of the observed differences was analyzed using the one-way ANOVA test (Tukey's multiple comparisons test) of the Graph Pad Prism 7 software for Windows (Version 7.04) as previously described (Mittal et al., 2015). Differences were considered significant at p < 0.05 (<sup>∗</sup> ), p < 0.01 (∗∗), p < 0.001 (∗∗∗), and p < 0.0001 (∗∗∗∗).

## RESULTS

## Adjuvant System C-di-AMP/Alum did not Induce any Signs of Unwanted Side Effects

The aim of this study was to investigate if c-di-AMP in combination with alum can further optimize the immune response against an antigen (β-Gal) when given by parenteral route. In a first attempt, the general behavior and the body weight development was investigated during the course of vaccination. In all vaccinated groups, no changes in spontaneous and provoked behavior, in grooming, feces character and body weight (**Figure 1)** were observed.

## Adjuvant System C-di-AMP/Alum Promotes Antigen-Specific Antibody Responses

The β-Gal-specific antibody titers post third immunization with or without the combination of alum and c-di-AMP are displayed as end-point-dilution titer in **Figure 2**. The combination of c-di-AMP with alum led to the stimulation of a stronger humoral immune response. The antigen specific

FIGURE 2 | the last sera dilution (end point dilution, e.p.d) showing the double value (OD 405 nm) of the control background. Each bullet symbol indicates a single animal, whereas horizontal lines represent the mean of animals. (C) β-Gal–specific IgG1/IgG2a ratio in the sera of immunized mice. Statistically significant differences were validated using the one-way ANOVA with Tukey's post hoc test with p < 0.001 (\*\*\*) and p < 0.0001 (\*\*\*\*) when compared with β-Gal alone (\*), or in combination with alum (o) or c-di-AMP (+), respectively.

antibody titers stimulated by c-di-AMP combined with alum are higher by a factor of 2.5 and 6.4, respectively, than with alum or c-di-AMP alone (**Figure 2A**). Moreover, only the adjuvant combination stimulated β-Gal-specific antibody titers that were statistically significantly higher compared to the titer stimulated by β-Gal alone or in combination with any single adjuvant. The humoral immune response can also give some indications regarding the polarization of the Tlymphocytes by looking at the ratio between IgG1/IgG2a antibody isotypes (**Figures 2B,C**). Taken individually, alum and c-di-AMP evoke the same IgG1 titer (3.6 × 10<sup>6</sup> and 3.3 × 10<sup>6</sup> , respectively), which was significantly higher compared to those observed in mice receiving β-Gal alone. However, the combination of alum and c-di-AMP resulted in significantly increased IgG1 titers compared to those obtained using single adjuvants (**Figure 2B**). Although not statistically significant, cdi-AMP alone induces an almost 10-fold higher titer of β-Galspecific IgG2a than alum alone. Again, the combination of the two adjuvants stimulated IgG2a titers significantly increased compared to those stimulated by β-Gal alone or adsorbed to alum. Thus, the addition of c-di-AMP with alum in the formulation somehow restores the ratio between IgG1/IgG2a compared to alum alone (ratio of 2.96 for the combination and 43.21 for alum alone; **Figure 2C**). Moreover, the ratio IgG1/IgG2a between c-di-AMP and the adjuvant combination is unchanged (3.56 and 2.96, respectively). Thus, the alum and c-di-AMP combination leads to not only a higher IgG1 titer but also a better IgG1/IgG2a balance, which could be translated into a better balance between the Th1/Th2 immune response (**Figure 2C**).

### Adjuvant System C-di-AMP/Alum Promotes Antigen-Specific Cellular Responses

The strongest proliferative capacity upon antigen restimulation was recorded in mice vaccinated with β-Gal and both adjuvants, as shown by the stimulation index of 13 (**Figure 3**). The proliferative capacity of the splenocytes is not only statistically significantly higher compared to those observed in mice vaccinated with β-Gal alone or in combination with a single adjuvant but also occurs already when restimulated with a very low concentration of antigen (1µg/ml of β-Gal), indicating a strong activation of antigen-specific immune cells. In contrast, nearly the same stimulation index is obtained only when splenocytes derived from mice immunized with β-Gal plus c-di-AMP were restimulated with 40µg/ml of β-Gal protein (**Figure 3**).

of immunized mice (n = 10). Results are displayed as average of

(Continued)

1µg/ml to 40µg/ml). Proliferation of the cells after 96 h was indirectly determined by counting [3H]-thymidine events incorporated into the DNA of proliferating cells. Results are average of quadruplicates and are expressed as stimulation index (SI). Statistically significant differences with respect to β-Gal alone (\*), β-Gal with alum (o) or β-Gal with c-di-AMP (+) were validated using one-way ANOVA with Tukey's post hoc test, with p < 0.05 (\*), p < 0.01 (\*\*), p < 0.001 (\*\*\*) and p < 0.0001 (\*\*\*\*).

Upon in vitro antigen restimulation, lymphocytes of vaccinated animals do not only proliferate, they also secrete cytokines which for some are the hallmark of their polarization. As represented in **Figure 4**, immunization with the combination of alum and c-di-AMP allows the stimulation of a significantly higher number of β-Gal-specific IFN-γ-producing cells than with alum alone. The combination seems also to be more potent than c-di-AMP alone, however, the observed difference wasn't significant. In contrast, when investigating the number of antigen-specific IFN-γ-producing CD8+ T lymphocytes, their number seems to decrease following vaccination with β-Gal using both adjuvants (the average number is comparable to alum alone) as indicated by the levels of β-Gal-specific IFN-γ-producing cells observed when splenocytes were restimulated with the CD8 peptide TPHPARIGL (**Figure 4A**). Thus, the elevated levels of IFN-γ-producing cells observed when splenocytes of mice receiving the adjuvant system c-di-AMP/alum were restimulated with β-Gal protein seem to be CD4+ T cells. The number of IL-2, IL-4 and IL-17 β-Gal-specific-producing cells is also statistically significantly increased in mice immunized with the adjuvant system c-di-AMP/alum with respect to those detected in mice receiving β-Gal alone. This supports the assertion of a more balanced T helper response with respect to the formulation encompassing only single adjuvants (**Figures 4B–D**).

The results obtained analyzing the number of cytokineproducing cells were in line with the levels of the corresponding cytokines detected in the supernatant of β-Gal restimulated splenocytes. Thus, only splenocytes of mice receiving β-Gal with the adjuvant system c-di-AMP/alum secreted statistically significantly higher levels of Th1 (IFN-γ, IL-2), Th2 (IL-4, IL-5, IL-10, IL-13), and Th17 (IL-17A, IL-22) cytokines compared to the group receiving β-Gal alone with p < 0.05 (**Figure 5**). However, no significant differences have been observed for TNF-α. Interestingly, the addition of c-di-AMP in the vaccine formulation with alum evokes a Th17 polarization of the lymphocytes following i.m. vaccination while c-di-AMP and alum stimulated only marginal IL-17 production if any at all (**Figure 5**). This is remarkable, since it is known already that cdi-AMP stimulates a strong IL-17 production only when applied by mucosal routes (Ebensen et al., 2011, 2017; Mittal et al., 2015). Taken together, the combination of c-di-AMP and alum seems to not only sustain the Th2 response stimulated by β-Gal but, in addition, strengthen the β-Gal-specifc Th1 and Th17 responses.

#### DISCUSSION

In contrast to most medications, vaccines are administered to large populations of more or less healthy persons. Moreover, also infants, children and immunocompromised individuals are target subpopulations, making no or only low potential risks or side-effects essential for vaccines (Di Pasquale et al., 2016). In order to increase vaccine safety, recent research has focused on so-called subunit vaccines. In contrast to common vaccine formulations, these vaccines usually consist only of microbial components. Unfortunately, the increased safety profile is at the expense of the vaccine efficacy, as pure antigens typically are only poorly immunogenic. Therefore, subunit vaccines essentially have to include antigen delivery systems or adjuvants (Ebensen and Guzman, 2008).

Adjuvants are key elements in both prophylactic as well as therapeutic vaccines since they can not only improve the strength of antigen-specific immune responses but also modify these responses according to the specific needs (Di Pasquale et al., 2016). However, one of the challenges in adjuvant development is to balance efficacy and safety in order to stimulate immunity (Batista-Duharte et al., 2018).

Besides a handful of adjuvants such as MF59, monophosphoryl lipid A (MPL A), AS03, AF03 or virosomes, only aluminum-based adjuvants (e.g., AS04) continue to be used worldwide since over 90 years. The main forms of aluminum adjuvant used in parenteral administered vaccines are aluminum hydroxide and aluminum phosphate (He et al., 2015). In addition to the beneficial aspects of alum, there are also some limiting effects such as induction of adverse reactions and preferential priming of Th2-type immune responses (He et al., 2015). In this regard, novel adjuvants including also mixtures of adjuvants (adjuvant systems) have opened the door for the development of vaccines with improved safety and efficacy profiles against emerging and/or re-emerging pathogens (Garcon and Di Pasquale, 2017). Thus, adjuvant combinations targeting

different pattern-recognition receptors (PRR), both endosomal and intracellular, enhance antigen-specific immune responses and/or direct them toward the response of need (e.g., cytotoxic or mucosal response) (Gutjahr et al., 2016). For example, the combination of the Toll-like receptor (TLR) agonist MPL A adsorbed on aluminum salts (AS04TM, GlaxoSmithKline) results in the stimulation of increased production of antigen-specific antibodies and an enhanced cell-mediated response by causing a local and temporary cytokine response (Reed et al., 2009; Garçon et al., 2011; Del Giudice et al., 2018). Nevertheless, no efficient antigen-specific Th1 and cytotoxic T cell responses are stimulated.

Therefore, the aim of the present work was the exploration of the great potential of c-di-AMP acting as a parenteral adjuvant in an adjuvant system combined with alum and the model antigen β-Gal to overcome the limitations of alum, since c-di-AMP was shown to stimulate strong Th1 and cytotoxic responses even when included in adjuvant combinations (Mittal et al., 2015; Ebensen et al., 2017; Matos et al., 2017; Sanchez Alberti et al., 2017; Schulze et al., 2017a; Temizoz et al., 2018). c-di-AMP binds to the transmembrane protein STING (stimulator of IFN genes) thereby activating the TBK3-IRF3 signaling pathway, subsequently triggering the production of type I IFN and TNF (McWhirter et al., 2009; Burdette et al., 2011; Shu et al., 2012). This in turn, results in strong adaptive immune responses.

In line with previously obtained results, co-administration of β-Gal with either alum or c-di-AMP alone stimulated strong humoral β-Gal-specific immune responses (Ebensen et al., 2011; McKee and Marrack, 2017). However, when immunizing mice with the combination of alum and c-di-AMP, the detected IgG titers were further increased compared to those observed in sera of mice receiving alum or c-di-AMP alone (factor 2.7 and 6.4, respectively). In order to evaluate if the combination of alum and c-di-AMP will also have a beneficial effect on antigen-specific cellular responses, we analyzed the proliferative capacity as well as the cytokine profiles of splenocytes of animals vaccinated i.m. with β-Gal alone or in combination with c-di-AMP, alum or the adjuvant system c-di-AMP/alum. Similar to the

observed humoral responses, the adjuvant system also stimulated increased antigen-specific cellular responses as indicated by the strong proliferative capacity already at relatively low doses of β-Gal antigen compared to the formulations encompassing only a single adjuvant. In addition, the observed cytokine profile stimulated following vaccination reveals the characteristic of cellmediated effector functions. Thus, splenocytes recovered from mice vaccinated with the adjuvant system c-di-AMP/alum plus β-Gal were found to produce enhanced levels of the Th1 cytokines IFN-γ and IL-2, the Th2 cytokines IL-4, IL-5, IL-10 and IL-13, and the Th17 cytokine IL-17, whereby the Th1/Th2 ratio was shifted toward Th1. In general, groups receiving either c-di-AMP or alum alone as adjuvant gave significantly lower numbers of cytokine-producing cells and cytokine titers compared to the group receiving the adjuvant combination. The observed cellular responses were confirmed by the obtained humoral immune responses, such as IgG titer and IgG subclass profiles. Hence, in line with previous studies, mice immunized with β-Gal coadministered with c-di-AMP showed a balanced production of IgG1 and IgG2a, correlating with a balanced Th1/Th2 response (Libanova et al., 2010; Ebensen et al., 2011; Mittal et al., 2015). The same is true for mice receiving the adjuvant system c-di-AMP/alum which showed a balanced IgG1/IgG2a ratio of approximately three, while alum alone promoted a Th2-biased response visualized by a IgG1/IgG2a ratio greater than 40.

Interestingly, the combination of alum with c-di-AMP also resulted in the stimulation of enhanced levels of IL-10, which was shown to block Th1 responses. Thus, when Oleszycka and coworkers immunized mice using alum as adjuvant, increased IL-10 titers were observed but only limited Th1 responses. In contrast, when they immunized IL-10 deficient mice using alum, increased Th1 responses have been obtained suggesting an inhibitory effect of IL-10 on Th1 cells (Oleszycka et al., 2018). However, previous findings also showed that induction of IL-10, which promotes IgA switch, displays broad anti-inflammatory properties (Lamm and Phillips-Quagliata, 2002) and is involved in self-regulation of Th1 responses (Jankovic et al., 2010). Thus, the strong IL-10 titers observed in mice vaccinated with β-Gal co-administered with the adjuvant system c-di-AMP/alum might simply reflect the necessity of self-regulation in order to restrain the stimulated antigen-specific Th1 immune response and prevent pathology (Ng et al., 2013). Likewise, vaccination using the adjuvant system c-di-AMP/alum also stimulated elevated levels of IL-17. This is in line with previous observations showing that alum-adjuvanted vaccines stimulated efficient immunity based on Th17 responses, which could even be improved when alum was combined with other adjuvants resulting in the induction of Th1/Th17 responses (Ross et al., 2013; Bagnoli et al., 2015). Interestingly, while c-di-AMP has been shown to facilitate Th17 polarization when administered by mucosal routes only marginal, if any at all, IL-17 production is stimulated when c-di-AMP is administered parenteral (Mittal et al., 2015; Ebensen et al., 2017). Therefore, the combination of alum with c-di-AMP applied by i.m. route seems to overcome this confinement.

Nevertheless, although the incorporation of c-di-AMP efficiently stimulates Th1 responses, the adjuvant system c-di-AMP/alum seems to be insufficient in compensating the inadequate stimulation of CD8+ T cells by alum as indicated by similar levels of IFNγ+ CD8-peptide-specific T cells in mice immunized using alum and the combination of alum and c-di-AMP, respectively. Instead, alum seems to inhibit the stimulation of CD8+ T cells by c-di-AMP. Further experiments need to be performed investigating whether higher concentrations of c-di-AMP would invert this effect.

Taken together, functionally distinct effector CD4+ T helper cell subsets are characterized by the secretion of distinct cytokine profiles. By this, our immune system is able to efficiently combat rapidly evolving and spreading pathogens infecting the host (Kara et al., 2014). Thus, while Th2 cells are critical for effective humoral immunity, by facilitating affinity maturation and class switch of the antigen-specific antibodies necessary for eradication of extracellular pathogens, Th1 cells are important for cell-mediated defense mechanisms helping to eliminate intracellular pathogens such as viruses (Chen and Cerutti, 2010). However, research in vaccine development seems to reveal that a single adjuvant will hardly be able to cope with all the foreseeable requirements in the field of infectious diseases. Thus, the development of new adjuvants increasing the portfolio of immunomodulatory molecules which allow the formulation of most effective vaccines tailored for the specific needs is essential. This tool box for vaccine developers can be even extended by combining adjuvants with different mechanisms of action. Moreover, adjuvant combination can also increase the safety profile of a certain vaccine, as in the case of alum, for example, the likelihood of connected adverse side effects is reduced by restraining immune reactions and diminishing the risk of immuno-pathological outcomes (Didierlaurent et al., 2009).

In this regard, our results demonstrate that the adjuvant system c-di-AMP/alum generates conditions sufficient for stimulating both humoral and cellular immune responses. Moreover, the combination of alum with c-di-AMP not only strengthens the stimulated antigen-specific immune responses but also modulates them in the direction of a Th1 response.

#### ETHICS STATEMENT

All animal experiments in this study were approved by the institutional ethical board and have been performed with ethical agreement by the local government of Lower Saxony (Germany) with the No. 33.42502 13/1281.

#### REFERENCES


#### DISCLOSURE

CAG and TE are named as inventors in a patent application covering the use of c-di-AMP as adjuvant (PCT/EP 2006010693). This does not alter our adherence to the Frontier Science policies on sharing data.

#### AUTHOR CONTRIBUTIONS

TE and KS conceived and designed the study. TE, KS, and SD performed the experiments. analyzed and interpreted the data, and wrote the manuscript. BP contributed to the writing and proofreading of the manuscript. CG gave scientific advice and supervised the work.

#### ACKNOWLEDGMENTS

The work was supported in part by the European Commission under the project Univax (contract No. 601738) and TRANSVAC (contract No. 228403). Furthermore, we are particularly grateful to Elena Reinhard for her outstanding technical help.


**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 Ebensen, Delandre, Prochnow, Guzmán and Schulze. 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.

# Oral Vaccination With a Formulation Combining Rhipicephalus microplus Subolesin With Heat Inactivated Mycobacterium bovis Reduces Tick Infestations in Cattle

Marinela Contreras <sup>1</sup> , Paul D. Kasaija1,2, Octavio Merino<sup>3</sup> , Ned I. de la Cruz-Hernandez <sup>3</sup> , Christian Gortazar <sup>1</sup> and José de la Fuente1,4 \*

#### Edited by:

Alberto Moreno, Emory University School of Medicine, United States

#### Reviewed by:

Riti Sharan, Texas A&M Health Science Center, United States Adriana Seixas, Federal University of Health Sciences of Porto Alegre, Brazil

\*Correspondence:

José de la Fuente jose\_delafuente@yahoo.com

#### Specialty section:

This article was submitted to Clinical Microbiology, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 16 November 2018 Accepted: 13 February 2019 Published: 01 March 2019

#### Citation:

Contreras M, Kasaija PD, Merino O, de la Cruz-Hernandez NI, Gortazar C and de la Fuente J (2019) Oral Vaccination With a Formulation Combining Rhipicephalus microplus Subolesin With Heat Inactivated Mycobacterium bovis Reduces Tick Infestations in Cattle. Front. Cell. Infect. Microbiol. 9:45. doi: 10.3389/fcimb.2019.00045 <sup>1</sup> SaBio, Instituto de Investigación en Recursos Cinegéticos IREC-CSIC-UCLM-JCCM, Ciudad Real, Spain, <sup>2</sup> National Livestock Resources Research Institute (NaLIRRI/NARO), Tororo, Uganda, <sup>3</sup> Facultad de Medicina Veterinaria y Zootecnia, Universidad Autónoma de Tamaulipas, Ciudad Victoria, Mexico, <sup>4</sup> Department of Veterinary Pathobiology, Center for Veterinary Health Sciences, Oklahoma State University, Stillwater, OK, United States

Vaccines are an environmentally friendly alternative to acaracides for the control of tick infestations, to reduce the risk for tick-borne diseases affecting human and animal health worldwide, and to improve animal welfare and production. Subolesin (SUB, also known as 4D8) is the functional homolog of Akirin2 involved in the regulation of development and innate immune response, and a proven protective antigen for the control of ectoparasite infestations and pathogen infection. Oral vaccination combining protein antigens with immunostimulants has proven efficacy with increased host welfare and safety, but has not been used for the control of tick infestations. Here we describe the efficacy of oral vaccination with a formulation combining Rhipicephalus microplus SUB and heat inactivated Mycobacterium bovis (IV) on cattle tick infestations and fertility. The levels of IgG antibody titers against SUB and M. bovis P22, and the expression of selected immune response genes were determined and analyzed as possible correlates of protection. We demonstrated that oral immunization with the SUB+IV formulation resulted in 51% reduction in the number of female ticks and 30% reduction in fertility with an overall efficacy of 65% in the control of R. microplus infestations by considering the cumulative effect on reducing tick survival and fertility in cattle. The akr2, IL-1β, and C3 mRNA levels together with antibody levels against SUB correlated with vaccine efficacy. The effect of the oral immunization with SUB+IV in cattle on tick survival and fertility is essential to reduce tick infestations, and extended previous results on the effect of R. microplus SUB for the control of cattle tick infestations. These results support the development of oral vaccines formulations for the control of tick infestations and the incidence of tick-borne diseases.

Keywords: subolesin, cattle tick, vaccine, oral vaccination, Rhipicephalus microplus, heat inactivated Mycobacterium bovis

## INTRODUCTION

Ticks are arthropod vectors of pathogens affecting human and animal health as well as animal welfare and production worldwide (Jongejan and Uilenberg, 2004; de la Fuente et al., 2008; Rashid et al., 2018). The cattle tick Rhipicephalus microplus Canestrini (Acari: Ixodidae) are economically important as parasites of a variety of livestock species with an impact on cattle industry in tropical and subtropical regions of the world (Rashid et al., 2018).

Despite the use of traditional cattle tick control methods such as the use of chemical acaricides, habitat management, and genetic selection of animals with higher resistance to ticks, tick prevalence continues to be a major economic problem for the cattle industry (de la Fuente et al., 2017; Rashid et al., 2018). This persistent problem is due to several factors including acaracide resistance in ticks and safety issues associated with these chemicals, which support the development of vaccines as an effective and environmentally sound approach for the control of tick infestations (de la Fuente and Contreras, 2015; de la Fuente et al., 2016b; 2017; de la Fuente, 2018). The commercial vaccines based on the R. microplus BM86 or BM95 recombinant antigens proved their efficacy for the control of cattle tick infestations and the reduction in the prevalence of certain tick-borne pathogens (de la Fuente et al., 2007, 2017; de la Fuente and Contreras, 2015; Rodríguez-Mallon, 2016; de la Fuente, 2018).

Tick Subolesin (SUB, also known as 4D8) is the functional ortholog of Akirin2 and is involved in the regulation of different biological processes including development and innate immune response (Artigas-Jerónimo et al., 2018). SUB was discovered as a tick protective antigen (Almazán et al., 2010), and since then it has shown vaccination efficacy for the control of infestations by different arthropod ectoparasite species and pathogen infection and transmission (recently reviewed by de la Fuente and Contreras, 2015; Artigas-Jerónimo et al., 2018).

Recent advances in tick vaccine research have resulted in the identification of new protective antigens for the control of tick infestations (recently reviewed by de la Fuente and Contreras, 2015; de la Fuente et al., 2016b, 2017; de la Fuente, 2018). However, research aimed at improving tick vaccine efficacy and safety by combining protective antigens and oral formulations is still to be done. Oral or intranasal vaccine formulations are easier to administer, and have proven efficacy with increased host welfare and safety by reducing stress and the risk of contamination or infection at the injection site and pathogen mechanical transmission (Wang et al., 2015; Lawan et al., 2018). However, orally delivered protein vaccines have a relatively low immunogenicity and antigen stability after immunization that require vaccine formulations with selected combinations of antigens and immunostimulants, and needleless delivery systems (Fry et al., 2012; Wang et al., 2015). In this context, the heat inactivated Mycobacterium bovis (IV) has been shown to activate the innate immune response-mediated trained immunity through complement component 3 (C3) to reduce mycobacterial infection and tuberculosis-like lesions in cattle, deer, pig, and zebrafish orally or systemically vaccinated with IV (Beltrán-Beck et al., 2014; de la Fuente et al., 2016a; Juste et al., 2016; Thomas et al., 2017; López et al., 2018, 2019; Risalde et al., 2018). Therefore, IV appears as a good immunostimulant candidate for oral vaccine formulations (de la Fuente et al., 2016a).

As a proof of concept of oral tick vaccine formulations, in this study we orally vaccinated cattle via needleless syringe using a formulation combining R. microplus SUB with IV for the control of cattle tick infestation. The results showed an effect of the oral vaccination on the reduction in the number of female ticks and fertility. Additionally, the akirin2 (akr2), interleukin-1beta (IL-1β), and C3 mRNA levels together with antibody levels against SUB correlated with vaccine efficacy. These results support research for the development of oral vaccines formulations for the control of tick infestations and the incidence of tick-borne diseases.

### MATERIALS AND METHODS

#### Ticks

The R. microplus (Susceptible Media Joya strain, CENAPA, Mexico) ticks were obtained from a laboratory colony maintained at the University of Tamaulipas, Mexico (Merino et al., 2011). Tick larvae were fed on cross-bred Bos taurus cattle and collected after repletion to allow for oviposition and hatching in humidity chambers at 12 h light:12 h dark photoperiod, 22–25◦C and 95% relative humidity (RH). Larvae were used for infestations at 15 days after hatching from eggs.

## Antigen Production and Vaccine Formulations

The synthetic R. microplus histidine-tag recombinant SUB (Genbank accession number GQ456170) with optimized codon usage for Escherichia coli was produced in E. coli BL21 and purified to >95% purity (**Figure 1**) by Ni affinity chromatography using 1 ml HisTrap FF columns mounted on an AKTA–FPLC system (GE Healthcare, Piscataway, NJ, USA) in the presence of 7 M urea lysis buffer as previously described (Almazán et al., 2010; Contreras et al., 2015). Protein concentration was measured using the Pierce <sup>R</sup> BCA protein assay kit (Thermo Scientific, Rockford, IL, USA).

The IV was prepared from M. bovis field isolate Neiker 1403 (spoligotype SB0339 originally isolated from a naturally infected wild boar) IV at Neiker-Tecnalia (Derio, Spain) under good manufacturing practices as previously described (Beltrán-Beck et al., 2014; Juste et al., 2016; Risalde et al., 2018). Briefly, the isolate was cultivated for 2–3 weeks in Middlebrook 7H9 medium enriched with OADC growth supplement (Sigma-Aldrich, St. Louis, MI, USA), and inactivated in a shaking water bath at 81–83◦C for 40 min. The inactivated inoculum was cultured in BACTEC Mycobacterial Growth Indicator Tubes (Becton Dickinson; Franklin Lakes, NJ, USA) and onto OADC agar solidified 7H9 plates in triplicate (100 µl each) to confirm the absence of viable mycobacteria. The final M. bovis IV preparation contained approximately the equivalent of 10<sup>7</sup> colony forming units (cfu) in 0.2 ml of PBS.

For vaccine formulation, purified recombinant SUB protein (200 µg) was mixed with 6 x 10<sup>6</sup> cfu IV in 18 ml PBS for SUB+IV group, or the same amount of IV was placed in 18 ml PBS

larvae on day 43 for the analysis of vaccine efficacy on the control of tick development and fertility, and the identification of possible correlates of protection and

immune mechanisms based on the host antibody response and the mRNA levels for selected immune response genes.

for control IV group. Vaccine formulations were stored at 4◦C until used.

#### Cattle Vaccination and Tick Infestation

Three and two 10-month-old European crossbred calves were randomly assigned to SUB+IV and IV vaccinated groups. Vaccines were administered via needleless syringe in the lateral region of the mouth followed by slightly raising the head of the calves (Jones et al., 2016). Cattle were vaccinated at days 0 and 22, and then infested at day 43 with 500 R. microplus larvae in single cells glued on the back of the calves. Adult engorged female ticks dropping from cattle were daily collected, counted, and weighted. All the collected adult female ticks were assessed for oviposition (egg mass weight/tick) and egg fertility (number of larvae/tick) to calculate vaccine efficacy (E) as previously reported (Merino et al., 2011, 2013) using only parameters with significant differences between groups (**Table 1**). The personnel vaccinating cattle and collecting the ticks were "blinded" as to which group animals belonged. Data were analyzed statistically to compare results between individuals fed on SUB+IV and IV vaccinated calves by Student's t-test with unequal variance (p = 0.05).

#### Analysis of Cattle IgG Antibody Response by ELISA

Serum samples were prepared from blood samples collected from each calf before each immunization (days 0 and 22),

#### TABLE 1 | Results of the oral vaccination on cattle tick infestations.


Results are shown for each vaccinated and infested cattle (N = 3 for SUB+IV and N = 2 for IV). Data was analyzed statistically to compare results between ticks fed on SUB+IV and IV vaccinated cattle (\*p = 0.02, \*\*p = 0.04; Student's t-test with unequal variance). Vaccine efficacy (E = 65%) was calculated as E = 100 x [1-(DT x DF)], where DT = No. female ticks in SUB+IV vaccinated calves/female ticks in IV vaccinated calves and DF = No. larvae/tick in SUB+IV vaccinated calves/No. larvae/tick in IV vaccinated calves.

tick infestation (day 43) and at the end of the experiment (day 65), and stored at −20◦C until analysis. An indirect ELISA test was performed to detect IgG antibodies against R. microplus SUB and M. bovis P22 proteins as described previously (Merino et al., 2011, 2013; Infantes-Lorenzo et al., 2018). High absorption capacity polystyrene microtiter plates were coated with 100 µl (0.01µg/ml solution of purified recombinant proteins) per well in carbonate-bicarbonate buffer (Sigma-Aldrich). After an overnight incubation at 4◦C, coated plates were blocked with 100 µl/well of blocking solution (5% skim milk in PBS). Serum samples or PBS as negative control were diluted (1:1000, 1:5000 and 1:10000, v/v) in blocking solution and 100 µl/well were added into duplicate wells of the antigen-coated plates. After an overnight incubation at 4◦C, the plates were washed three times with a washing solution (PBS containing 0.05% Tween 20) and then incubated with 1:20,000 rabbit anti-bovine immunoglubolin G (IgG)– horseradish peroxidase conjugate (Sigma-Aldrich) for 1 h at 37◦C. After three washes with washing solution, 100 µl/well of substrate solution (Fast OPD, Sigma-Aldrich) was added. Finally, the reaction was stopped with 50 µl/well of 2 N H2SO<sup>4</sup> and the optical density (OD) was measured in a spectrophotometer at 450 nm. Antibody titers in SUB+IV and IV vaccinated cattle were expressed as the O.D450nm (O.Dcattlesera – O.DPBScontrol) and compared between groups by ANOVA test (p = 0.05; https://www.socscistatistics.com/tests/ anova/Default2.aspx).

#### Analysis of Gene Expression by qRT-PCR

Total RNA was extracted from blood samples collected from each calf before each immunization (days 0 and 22), tick infestation (day 43) and at the end of the experiment (day 65) using TRI Reagent BD (Sigma-Aldrich, St. Louis, MI, USA) following the manufacturer's instructions. Gene expression profiles from selected genes involved in immunity C3 (NM\_001040469.2), tumor necrosis factor alpha (TNF-α; AF348421), IL-1β (NM174093), interleukin 2 (IL-2; NM180997), akr2 (NM\_001110087.1), and interleukin 12 (IL-12; U11815.1) were analyzed by qRT-PCR using the following gene-specific forward (F) and reverse (R) 5′ - 3′ primers and annealing temperature (C3, F: ATTGCCAGGTTCTTGTACGG and R: GTCACTGCCTGATTGCAAGA, 56◦C; TNF-α, F: CCTC ACCCACACCATCAG, and R: GCGATCTCCCTTCTCCAG, 54◦C; IL-1β, F: TCAGAATGGAAACCCTCTCTC and R: GCATGGATCAGACAACAGTG, 56◦C; IL-12, F: AAGT GAAGTCATTGCTGCTG and R: TGTCCATTGAATCCTTG ATCTC, 54◦C; akr2, F: CATTTATGGGCTGCCTTGTT and R: TGCACAGCTTCTACCACGAC, 54◦C; IL-12, F: TCTGG ACACTTCACCTGCTG and R: TGCACAGCTTCTACCA CGAC, 58◦C) and a Quantitect SYBR Green RT-PCR Kit in a Rotor Gene Q thermocycler (Qiagen, Inc. Valencia, CA, USA) following manufacturer's recommendations. A dissociation curve was run at the end of the reaction to ensure that only one amplicon was formed and that the amplicon denatured consistently in the same temperature range for every sample (Ririe et al., 1997). The mRNA Ct values were normalized against Bos taurus ß-actin (AY141970.1, F: GGCCGAGCGGAAATCG and R: GCCATCTCCTGCTCG AAGTC, 52◦C) using the genNorm ddCT method (Livak and Schmittgen, 2001). The normalized mRNA levels (mean of duplicated and normalized Ct values) were compared between SUB+IV and IV vaccinated groups by ANOVA test (p = 0.05; https://www.socscistatistics.com/tests/anova/ Default2.aspx).

## Correlation Analysis

Two different Spearman's Rho correlation analyses (p = 0.05; https://www.socscistatistics.com/tests/spearman/Default2. aspx) were conducted for the identification of putative correlates of vaccine efficacy. The analysis was conducted in individual SUB+IV and IV vaccinated cattle. The first analysis was conducted between the number of female ticks or larvae/tick and antibody titers (O.D450nm) or the normalized mRNA levels at the time of tick infestations (day 43) and at the end of the experiment (day 65). The second analysis was conducted between the normalized mRNA levels and antibody titers (O.D450nm) at the time of tick infestations (day 43) and the end of the experiment (day 65).

### RESULTS AND DISCUSSION

#### Rationale and Experimental Design

Despite the proven efficacy of R. microplus BM86 and BM95 based vaccines for the control of cattle tick infestations (de la Fuente et al., 2007, 2017; de la Fuente and Contreras, 2015; Rodríguez-Mallon, 2016; de la Fuente, 2018) and recent advances in the identification of new tick protective antigens (de la Fuente and Contreras, 2015; de la Fuente et al., 2016b, 2017; de la Fuente, 2018), research is needed for the development of vaccine formulations with higher efficacy and safety for the control of tick infestations and tick-borne diseases (de la Fuente, 2018). To address this challenge, the objective of this study was to provide a proof of concept for oral vaccine formulations for the control of cattle tick infestations, and the identification of candidate correlates of vaccine efficacy. Oral vaccine formulations are easier to administer and offer the possibility of reducing stress and vaccinationassociated risks while increasing protective efficacy with effective immunostimulants (Wang et al., 2015; Lawan et al., 2018).

The experimental design for this study included cattle vaccination and R. microplus infestations for the analysis of vaccine efficacy on the control of tick development and fertility (**Figure 1**). Cattle were orally vaccinated via needleless syringe using a formulation combining R. microplus SUB antigen with proven efficacy in the control of ectoparasite infestations (de la Fuente and Contreras, 2015; Artigas-Jerónimo et al., 2018) and IV with immunostimulatory activity (Beltrán-Beck et al., 2014; de la Fuente et al., 2016a; Juste et al., 2016; Thomas et al., 2017; López et al., 2018, 2019; Risalde et al., 2018) (SUB+IV group) or IV alone (IV group) for comparative analyses between both groups. An untreated group was not included because the study focused at comparing the effect of the SUB+IV combined formulation with any background of the IV immunostimulant alone. The immune markers selected to characterize possible correlates of vaccine protection included IgG antibodies, which have been shown to mediate SUB protective response (Merino et al., 2011, 2013), and the innate immune response-mediated trained immunity markers akr2, IL-1β, IL-2, IL-12, TNF-α, and C3 involved in protective response to IV oral or systemic vaccination in different species (Beltrán-Beck et al., 2014; de la Fuente et al., 2016a; Juste et al., 2016; Thomas et al., 2017; López et al., 2018, 2019; Risalde et al., 2018).

## Vaccine Efficacy for the Control of Cattle Tick Infestations

The results of the vaccination trial showed a 51% reduction in the number of engorged female ticks obtained in the SUB+IV group when compared to the IV control (p = 0.02; **Table 1**). This reduction in the number of ticks completing feeding was similar in all three cattle vaccinated with SUB+IV (**Table 1**). Additionally and despite animal-to-animal variations, a 30% reduction in tick fertility was observed in the eggs from ticks fed on SUB+IV vaccinated calves when compared to the IV control (p = 0.04; **Table 1**). The effect of tick vaccines on the reduction in the number of female ticks completing feeding and fertility has been identified as a critical parameter for the control of cattle tick infestations (Schetters et al., 2016). However, no differences were observed in the weight of engorged ticks or the oviposition (**Table 1**), a result that differs from previous vaccination trials with systemic SUB (Merino et al., 2011, 2013; Artigas-Jerónimo et al., 2018).

In a previous experiment (Contreras et al., 2015), sera from pigs orally immunized with SUB, SUB-Major surface protein 1a (MSP1a) E. coli membrane-bound chimera or PBS as control were used for R. microplus capillary feeding to evaluate tick weigh. The results of this preliminary trial showed no effect of anti-SUB antibodies but a 55% reduction in tick weight increase after feeding on SUB-MSP1a serum (Contreras et al., 2015). Bacterial membranes with the surface exposed SUB-MSP1a chimeric antigen were shown to enhance SUB immunogenicity and protective capacity, thus providing supporting evidences for the inclusion of bacterial-derived immunostimulants for SUBbased oral vaccine formulations (Contreras et al., 2015). The discrepancies on the effect of vaccination on tick weight may be attributable to different factors including differences in the immune response to various SUB antigens and between systemic and oral vaccine formulations.

#### Correlates of Vaccine Efficacy and Immune Mechanisms of Protection

To address the putative protective mechanisms of oral SUB+IV vaccination in cattle, the IgG antibody response against both SUB and IV (P22) were first characterized in all animals included in the study. The antibody response against SUB has been correlated before with vaccine efficacy in cattle (Merino et al., 2011, 2013). The P22 antigen is composed by various M. bovis proteins that have been validated before for the analysis of the antibody response in IV vaccinated hosts (Infantes-Lorenzo et al., 2017, 2018; López et al., 2018; Risalde et al., 2018).

Despite animal-to-animal differences in the IgG antibody response to both antigens (**Figure 2A**), the anti-SUB but not anti-P22 IgG antibody response was higher in SUB+IV vaccinated cattle when compared to IV vaccinated group (p = 0.01; **Figure 2B**). Differences in the animal-to-animal response could be attributed to variations in vaccine administration via

needleless syringe. Furthermore, a correlation analysis between vaccine efficacy on the reduction in the number of female ticks and anti-SUB or P22 IgG antibody levels was conducted as one of the proposed correlates of protection in tick vaccine trials (de la Fuente and Contreras, 2015). As in previous tick SUB vaccination trials (Merino et al., 2011, 2013), a negative correlation (r = −0.9; p = 0.04) was obtained between the number of engorged female ticks and IgG antibody titers against SUB but not P22 in vaccinated cattle at day 43 (**Figure 2C**), showing a correlation between the effect of oral SUB+IV vaccination (IgG antibodies) and vaccine efficacy on tick development.

The protective mechanism associated with IV vaccination has been attributed to the activation of the innate immune responsemediated trained immunity through C3 pathway (Beltrán-Beck et al., 2014; de la Fuente et al., 2016a; Juste et al., 2016). Therefore, C3 pathway components and other immune response genes were selected for the analysis of mRNA levels in vaccinated cattle. The results showed a tendency toward an increase in mRNA levels in response to vaccination/infestation in SUB+IV vaccinated cattle (**Figure 3A**). In IV-vaccinated cattle, the tendency was a decrease in mRNA levels at days 22 and 65 following first vaccination and tick infestation, respectively, and an increase at day 43 after second vaccination (**Figure 3A**). Nevertheless, a positive correlation (r = 0.9; p = 0.04) was obtained only between TNF-α mRNA levels and IgG antibody titers against SUB but not P22 at day 43 (**Figure 3B**). Additionally, for C3 pathway immune response akr2, IL-1β, and C3 genes a negative correlation (r = −0.9 to −1.0; p = 0.005 to 0.04) was observed between mRNA levels and the number of engorged female ticks at day 65 (**Figure 4**). Finally, TNF-α mRNA levels negatively correlated (r = −0.9; p = 0.04) with tick fertility (number of larvae per tick) at day 65 (**Figure 4**).

Recently, we proposed that the mechanism behind systemic SUB vaccine protective capacity is based on anti-SUB antibodies that could enter into tick cells by still unknown mechanisms where they can interact with cytosolic SUB to prevent its translocation to the nucleus and/or SUB-protein interactions necessary to exert its regulatory functions (de la Fuente et al., 2011) (**Figure 4**). SUB exerts its function through physical and/or functional interactions with other proteins that have not been fully characterized, but include some candidate proteins involved in the regulation of developmental processes potentially affecting tick feeding (Artigas-Jerónimo et al., 2018). Herein, the IgG antibody response against SUB also correlated with oral vaccination efficacy, therefore suggesting a similar protective mechanism in orally and systemically vaccinated cattle (**Figure 4**).

Considering the negative correlation between anti-SUB IgG antibody and C3 pathway gene mRNA levels and the number of female ticks, a protective mechanism for SUB+IV oral vaccination was proposed (**Figure 4**). Once the vaccine reaches the oral mucosa-associated lymphoid tissue underlying the epithelium, antigen uptake, processing and presentation occurs via microfold/membraneous cells (M) by antigen-presenting cells

FIGURE 3 | Effect of oral vaccination and infestation with R. microplus larvae on the cattle mRNA levels for immune response genes. Blood mRNA was obtained from samples collected before 1st (day 0) and 2nd (day 22) immunizations and tick infestation (day 43) with R. microplus larvae, and at the end of the experiment (day 65). (A) The mRNA levels for immune response genes C3, TNF-α, IL-1β, IL-2, akr2, and IL-12 were determined by qRT-PCR in vaccinated cattle, normalized against B. taurus ß-actin and normalized mRNA levels (mean of duplicated and normalized Ct values) compared between SUB+IV and IV groups by ANOVA test (p < 0.05). (B) The Spearman's Rho correlation analyses (p < 0.05; https://www.socscistatistics.com/tests/spearman/Default2.aspx) were conducted to correlate the IgG antibody response to mRNA levels at time of tick infestations (day 43). The normalized TNF-α mRNA levels were the only that positively correlated with anti-SUB but not P22 IgG antibody levels. The linear correlation coefficients (r) and p-value are shown (N = 5).

the innate and adaptive immune systems for the initiation of antigen-specific response by activated T cells (T), T helper cells (Th), cytotoxic T lymphocytes (CTLs) and B cells. Activated adaptive immune response cells secrete cytokines such as IL-12, IL-2, and TNF-α to realize the systemic immune protection. The C3 pathway was activated after oral vaccination with SUB+IV to mediate the production of IgG antibodies affecting SUB translocation/function after ingestion with blood meal for the reduction of tick infestations.

(APCs) such as macrophages (Mφ), dendritic cells (DCs) and B cells (B) (Wang et al., 2015) (**Figure 4**). The key APCs, DCs bridge the innate and adaptive immune systems for the initiation of antigen-specific response by activated T cells (T), T helper cells (Th), cytotoxic T lymphocytes (CTLs) and B cells (Holmgren and Czerkinsky, 2005; Peng et al., 2006; Lawson et al., 2011; Wang et al., 2015) (**Figure 4**). The C3 pathway mediates bridging of the innate and adaptive immune responses (Lubbers et al., 2017) (**Figure 4**). Activated adaptive immune response cells secrete cytokines such as IL-12, IL-2, and TNF-α, which mRNA levels positively correlated with anti-SUB antibody titers (**Figure 3B**), and then migrate to and multiply through the common mucosal immune system to realize the systemic immune protection (Seder and Hill, 2000; Wang et al., 2015). As also shown in this study, the production of neutralizing antibodies is vital for protective humoral immune response in reducing tick infestations (de la Fuente et al., 1998, 2011; Merino et al., 2011, 2013; Contreras and de la Fuente, 2016, 2017). In this process, the C3 pathway that was activated after oral vaccination with SUB+IV has been shown to mediate antibody response (Rutemark et al., 2011), a finding supported here by the positive correlation between akr2, IL-1β, and C3 mRNA levels and the number of female ticks (**Figure 4**). Additionally, the proinflammatory cytokine IL-1β has been reported to induce antigen-specific serum IgG and lymphocyte proliferative responses with intranasally administered with soluble protein antigens (Staats and Ennis, 1999).

Finally, the negative correlation between TNF-α mRNA levels and the number of larvae per tick suggested that T-mediated response after SUB+IV oral vaccination may have a role in the reduction of tick fertility. As discussed above and supported in this study, antigen-specific IgG antibody response plays a key role in the protective response to tick vaccines (de la Fuente et al., 1998, 2011; Merino et al., 2011, 2013; Contreras and de la Fuente, 2016, 2017). However, the possible role of the complement, T immune response and other effector mechanisms has been considered before as part of tick vaccine efficacy (Tellam et al., 1992). The development of specific antibody- or T-mediated immunologic responses and the activation of mucosally-induced immunity depend on complex sets of immunologic events, including the antigen-induced activation of different populations of B, T, and APC, and the expression of proinflammatory and immunoregulatory cytokines (Dinarello, 2000; Ogra et al., 2001). In particular, the proinflammatory cytokine TNF-α has been associated with DC maturation in mucosal immune response (Ogra et al., 2001). The results obtained here suggested that future experiments should address this component of the protective immune response to both systemic and oral vaccination with tick antigens. Additionally, other events and cytokines such as IL-4, IL-5, IL-6, IL-10, colony-stimulating factor (CSF), interferon gamma (IFN-γ) and transforming growth factor beta (TGF-β) reported to be involved in mucosal immune responses together with variations in APC, B, and T cell subsets should be addressed to provide additional support for the proposed protective mechanisms elicited by oral vaccination with SUB+IV.

#### CONCLUSIONS

The results of this study confirmed the efficacy of tick SUB for the control of cattle tick infestations, and expanded these results to oral vaccination with this antigen in combination with the IV immunostimulant. Furthermore, the identification of correlates of oral SUB+IV vaccine protection provided markers and targets for future experiments with larger number of animals aimed at conducting analyses to provide additional support for the protective immune mechanisms, the duration of the protective response, and optimizing vaccine formulation, delivery and efficacy. An optimal oral vaccine formulation requires the appropriate combination of antigens, immunostimulants and delivery carriers to induce a series of protective immune responses as shown here by the activation of the C3 pathway and the production of IgG antibodies and relevant cytokines (**Figure 4**). Research focused on oral tick vaccines would have a positive impact on the control of tick infestations and tick-borne diseases in humans, farm and companion animals through

#### REFERENCES


reducing stress and systemic vaccination-associated risks while increasing protective efficacy.

## DATA AVAILABILITY

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

#### ETHICS STATEMENT

The study was conducted in accordance with standards specified in the Guide for Care and Use of Laboratory Animals for the University of Tamaulipas (UAT), Mexico. The protocol was approved by the ethics committee of the UAT (No. CBBA\_17\_0).

## AUTHOR CONTRIBUTIONS

MC, OM, CG, and JF conceived the study and designed the experiments. MC, PK, OM, and NdlC-H performed the experiments. MC and JF performed data analysis. MC, PK, CG, and JF wrote the manuscript. All authors approved and contributed to the final version of the manuscript.

## FUNDING

This research was financially supported by the Ministerio de Economía, Industria y Competitividad (Spain) grant BFU2016- 79892-P and the Fondo Sectorial de Investigación para la Educación (SEP-CONACYT, Mexico) project CB-2015-01- 255205 (DICB/C1000/3284/2016).

#### ACKNOWLEDGMENTS

We thank Jenifer Naomi Gueta Spain and Justo Carrizales Zapata (UAT, Mexico) for technical assistance during vaccination trial.


challenges for the control of tick-borne diseases. Vaccine 35, 5089–5094. doi: 10.1016/j.vaccine.2017.07.097


**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 Contreras, Kasaija, Merino, de la Cruz-Hernandez, Gortazar and de la Fuente. 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.

# Toward DNA-Based T-Cell Mediated Vaccines to Target HIV-1 and Hepatitis C Virus: Approaches to Elicit Localized Immunity for Protection

#### Edited by:

*Alberto Moreno, Emory University School of Medicine, United States*

#### Reviewed by:

*Leo J. Swadling, University College London, United Kingdom Antonella Folgori, ReiThera Srl, Italy*

#### \*Correspondence:

*Danushka K. Wijesundara danushka.wijesundara@ adelaide.edu.au*

#### Specialty section:

*This article was submitted to Clinical Microbiology, a section of the journal Frontiers in Cellular and Infection Microbiology*

Received: *23 November 2018* Accepted: *14 March 2019* Published: *03 April 2019*

#### Citation:

*Mekonnen ZA, Grubor-Bauk B, Masavuli MG, Shrestha AC, Ranasinghe C, Bull RA, Lloyd AR, Gowans EJ and Wijesundara DK (2019) Toward DNA-Based T-Cell Mediated Vaccines to Target HIV-1 and Hepatitis C Virus: Approaches to Elicit Localized Immunity for Protection. Front. Cell. Infect. Microbiol. 9:91. doi: 10.3389/fcimb.2019.00091* Zelalem A. Mekonnen<sup>1</sup> , Branka Grubor-Bauk <sup>1</sup> , Makutiro G. Masavuli <sup>1</sup> , Ashish C. Shrestha<sup>1</sup> , Charani Ranasinghe<sup>2</sup> , Rowena A. Bull <sup>3</sup> , Andrew R. Lloyd<sup>3</sup> , Eric J. Gowans <sup>1</sup> and Danushka K. Wijesundara<sup>1</sup> \*

*<sup>1</sup> Virology Laboratory, Basil Hetzel Institute for Translational Health Research, Discipline of Surgery, University of Adelaide, Adelaide, SA, Australia, <sup>2</sup> Molecular Mucosal Vaccine Immunology Group, The John Curtin School of Medical Research, The Australian National University, Canberra, ACT, Australia, <sup>3</sup> Viral Immunology Systems Program, The Kirby Institute, The University of New South Wales, Sydney, NSW, Australia*

Human immunodeficiency virus (HIV)-1 and hepatitis C virus (HCV) are major contributors to the global disease burden with many experts recognizing the requirement of an effective vaccine to bring a durable end to these viral epidemics. The most promising vaccine candidates that have advanced into pre-clinical models and the clinic to eliminate or provide protection against these chronic viruses are viral vectors [e.g., recombinant cytomegalovirus, Adenovirus, and modified vaccinia Ankara (MVA)]. This raises the question, is there a need to develop DNA vaccines against HIV-1 and HCV? Since the initial study from Wolff and colleagues which showed that DNA represents a vector that can be used to express transgenes durably *in vivo*, DNA has been regularly evaluated as a vaccine vector albeit with limited success in large animal models and humans. However, several recent studies in Phase I-IIb trials showed that vaccination of patients with recombinant DNA represents a feasible therapeutic intervention to even cure cervical cancer, highlighting the potential of using DNA for human vaccinations. In this review, we will discuss the limitations and the strategies of using DNA as a vector to develop prophylactic T cell-mediated vaccines against HIV-1 and HCV. In particular, we focus on potential strategies exploiting DNA vectors to elicit protective localized CD8<sup>+</sup> T cell immunity in the liver for HCV and in the cervicovaginal mucosa for HIV-1 as localized immunity will be an important, if not critical component, of an efficacious vaccine against these viral infections.

Keywords: DNA vaccine, hepatitis C, human immunodeficiency virus, HIV/AIDS, HCV, tissue-resident memory, T cell immunity

## INTRODUCTION

Human immunodeficiency virus (HIV)-1 and hepatitis C virus (HCV) are significant contributors to the global disease burden with ∼36.9 million people living with HIV-1 and at least 71 million people persistently infected with HCV (WHO, 2017; UNAIDS, 2018). Anti-retroviral therapy (ART) and direct acting anti-virals (DAAs) have contributed significantly to prolonging the lifespan and curing of HIV-1- and HCV-infected individuals, respectively (Cihlar and Fordyce, 2016; Zhang, 2016), but the annual HIV-1 and HCV incidences are still rising by millions. Furthermore, only 17 million (<50%) people have access to ART (Cihlar and Fordyce, 2016) and only 20% of patients are diagnosed for HCV (WHO, 2017). Additional issues involving drug resistance, reactogenicity associated with life-long ART and the lack of universal access to testing and cost-subsidized therapies minimize the ability of effective anti-viral drugs to end the HIV-1 and HCV epidemics. Thus, there is an urgent need to develop effective prophylactic vaccines to control the number of new infections and reduce the burden of supplying ART and DAA therapies to patients (Shin, 2016; Stone et al., 2016).

HIV-1 and HCV are rapidly mutating RNA viruses that exhibit considerable genetic diversity (nine subtypes in the major group of HIV-1 (German Advisory Committee Blood SAoPTbB., 2016) and 8 genotypes (gt1-8) of HCV which include at least 67 subtypes (Borgia et al., 2018) making immunity that develops during natural infection mostly ineffective. The lack of immune correlates of protection and convenient animal models permissive to infection make vaccine design and testing extremely challenging, and have also contributed to the fact that there is still no licensed vaccine for either HIV-1 or HCV (Wang et al., 2015; Bailey et al., 2019). HIV-1 and HCV coinfections represent an additional obstacle (Platt et al., 2016) although a recent clinical study suggests that co-administration of HIV-1 and HCV vaccines in humans can elicit robust HIV-1- and HCV-specific T cell responses without perturbing the immunodominance hierarchies of T cells responding against the vaccine encoded HIV-1 or HCV antigens (Hartnell et al., 2018).

DNA vaccines have been investigated for nearly three decades and are essentially bacteria-derived plasmids genetically engineered to encode immunogens under the control of promoters that facilitate robust expression of DNA in mammalian cells to induce adaptive immunity (Ferraro et al., 2011). DNA vaccines are inexpensive, easily constructed, stable at room temperature, replication defective in transfected mammalian cells and have minimum side effects which simplifies handling and distribution such that even developing countries can benefit from DNA vaccines (Jorritsma et al., 2016). Furthermore, plasmid DNA can be more easily used in multi-dose regimens unlike recombinant virus vectors that suffer from anti-vector immunity (Frahm et al., 2012). Recent seminal studies described therapeutic DNA vaccination against human papillomavirus (HPV) which resulted in histological regression and/or eliminated persistent HPV infection and HPV-related cervical lesions (Kim et al., 2014; Trimble et al., 2015). More recently, a DNA vaccine was developed that induced protective neutralizing antibodies (NAb) to Zika virus (ZIKV) in mice (Larocca et al., 2016) and rhesus macaques (Abbink et al., 2017) leading to the development of safe and immunogenic ZIKV DNA vaccines for humans (Tebas et al., 2017; Gaudinski et al., 2018). Thus, the many advantages of using plasmid DNA to develop vaccines and the recent developments of DNA vaccines in eliciting protective immunity in humans and higher animal models warrant further examination as to how DNA vaccines can be harnessed in vaccination regimens to target HIV-1 and HCV.

## IMMUNE TARGETS FOR HIV-1 AND HCV PROPHYLACTIC VACCINE DEVELOPMENT

It is imperative that vaccines take into account the virus tropism, transmission routes, pathogenesis and immune responses that provide effective resistance against infections to elicit protective immunity against HIV-1 and/or HCV.

It is now established that mucosal tissues, mainly the gentiorectal tissues and gastrointestinal tract, are the major sites of HIV-1 entry and pathogenesis, respectively (Belyakov and Ahlers, 2012). Induction of robust HIV-specific immune responses at these sites will be necessary to prevent HIV-1 infection or at the very least control viraemia during the acute phase of infection thus reducing the viral set point (McMichael and Koff, 2014) and infection-induced microbial translocation which can result in diversion of immune responses to counteract dysbiosis (Vujkovic-Cvijin et al., 2013). Furthermore, a prophylactic HIV-1 vaccine will likely be delivered using an active immunization strategy and attempt to mimic immune responses reported to be protective in macaques against simian immunodeficiency virus (SIV) and/or provide resistance against natural HIV-1 infections (Pontesilli et al., 1998; Saez-Cirion et al., 2007; Hansen et al., 2011; Haynes et al., 2012; Barouch et al., 2015, 2018; Ackerman et al., 2016; Borducchi et al., 2016). In this regard, the most protective immune responses reported to date involve T cell-mediated immunity (CMI) (Pontesilli et al., 1998; Saez-Cirion et al., 2007; Hansen et al., 2011; Borducchi et al., 2016), polyfunctional antibody responses (Barouch et al., 2015, 2018; Ackerman et al., 2016), antibody-dependent cellular cytotoxicity (Haynes et al., 2012), and broadly neutralizing antibodies (bNAb) (Burton and Hangartner, 2016). Although potent bNAb represent a blueprint for HIV-1 vaccine design, these antibodies are unlikely to be as effective in preventing cell to cell transmission compared to neutralizing cell free virus (Parsons et al., 2017). Consequently, a highly effective prophylactic HIV-1 vaccine will likely also rely on CMI to target highly conserved viral proteins such as Gag and Pol (Rolland et al., 2007) and/or non-neutralizing antibodies to broadly target the virus Envelope to prevent cell-cell transmission of the virus.

Unlike HIV-1 which has a relatively broad tropism, HCV is a bloodborne virus that primarily infects and replicates in hepatocytes. In primary hepatitis C infection, ∼25% of patients naturally clear the virus and although reinfection occurs in many individuals (Grebely et al., 2012), it is evident that repeated infection is associated with a reduced magnitude and duration of viraemia, and a greater likelihood of clearance (Sacks-Davis et al., 2015). Thus, characterizing and eliciting the naturally-protective immune responses during primary infection and reinfection provide a rational path for the design of a prophylactic HCV vaccine (Grebely et al., 2012). The immune responses that correlate best with natural protection include robust and broad CMI to conserved HCV non-structural (NS) proteins (NS3, NS4, and NS5) (Smyk-Pearson et al., 2008; Baumert et al., 2014) and NAb targeting conserved regions of the viral envelope (E1E2) proteins (Houghton, 2011; Osburn et al., 2014; Bailey et al., 2017). Although CMI will not prevent infection, clinical data suggest that T cell responses could prevent the development of persistent infection in individuals who naturally clear the virus, which is an acceptable outcome given that primary infection is often asymptomatic and not associated with severe disease outcomes (Baumert et al., 2014). After two decades of unsuccessful pre-clinical studies and Phase I HCV vaccine trials, the current lead prophylactic candidate is in an NIH-sponsored Phase IIb, placebo-controlled trial (ClinicalTrials.gov Identifier: NCT01436357) in high risk people who inject drugs (PWID) (Swadling et al., 2014). The candidate vaccination regimen being tested utilizes a chimpanzee adenovirus (ChAd) prime and a modified vaccinia Ankara (MVA) boost to elicit systemic T cell immunity to gt1 NS antigens (Swadling et al., 2014). However, it is not clear if this vaccination can induce robust intrahepatic T cell immunity and sufficient multi-genotypic immunity to result in significant protection in vaccinated individuals especially given the increased prevalence of multiple genotypes in HCV endemic regions.

#### T CELL-MEDIATED DNA VACCINES AGAINST HIV-1 AND HCV IN THE CLINIC

DNA vaccines against HCV have been routinely tested in small and large animals including non-human primates (Latimer et al., 2014; Gummow et al., 2015; Grubor-Bauk et al., 2016; Wijesundara et al., 2018). Some candidates have also progressed in phase I/II clinical trials, but none have progressed to a large-scale efficacy trial in humans. A promising DNA vaccine that included a cocktail of four plasmids with each plasmid encoding codon optimized NS3/4A, NS4B, NS5A, or NS5B sequences from gt1a/b virus was used to prime/boost vaccinate macaques by electroporation (Latimer et al., 2014). In this study, the vaccine induced CD4<sup>+</sup> and CD8<sup>+</sup> T cells against each of the NS proteins encoded in the DNA cocktail which has resulted in the testing of the DNA cocktail in a phase I clinical trial (ClinicalTrials.gov Identifier: NCT02027116) although the results are yet to be disclosed.

A DNA vaccine has been tested for therapeutic vaccination against HCV. 12 hepatitis C patients suffering from chronic disease received three doses of a DNA vaccine encoding codon optimized NS3/4A from gt1a virus via electroporation on the deltoid muscle which induced NS3-specific CMI and a transient decrease in viral RNA levels (Weiland et al., 2013). The vaccine was also tested in eight patients who received interferon and ribavirin treatment of which six patients were completely cured of the infection (Weiland et al., 2013). Thus, DNA vaccines could be exploited in therapeutic settings against HCV, but this is unlikely to occur in the future given the success of using DAA to cure hepatitis C patients.

DNA vaccines against HIV-1 have been tested in different pre-clinical models and some have been tested in phase I/II clinical trials (Okuda et al., 1997; Cafaro et al., 2001; Tomusange et al., 2016). The first human clinical trial of a DNA vaccine, encoding env and rev genes, against HIV-1 was conducted in 1998 (MacGregor et al., 1998). Following vaccination of HIV-1 positive, treatment naïve individuals, no significant changes were observed in CD4<sup>+</sup> and CD8<sup>+</sup> T cell responses as well as in plasma HIV RNA. In another phase I clinical trial a DNA vaccine that encoded env and rev was shown to induce CD4<sup>+</sup> T cell and poor CD8<sup>+</sup> T cells responses in HIV-1 seronegative individuals (MacGregor et al., 2002). Similarly, low CD8<sup>+</sup> T cell responses were observed in another phase I clinical trial following prime/boost vaccination with a DNA vaccine that encoded gag and pol genes (Tavel et al., 2007). More robust HIV-specific T cell responses have been elicited when DNA vaccines are used to prime and recombinant viral vectors are used to boost immune responses (Kibuuka et al., 2010; Bakari et al., 2011; Churchyard et al., 2011; Hayton et al., 2014; Moyo et al., 2017). However, prime/boost vaccinations with DNA vaccines alone can be optimized to elicit robust immune responses in humans against HIV-1. For instance, a retrospective study evaluating the immunogenicity of 10 HIV-1 DNA vaccine trials that used DNA vaccines in the absence of viral vectors or adjuvants suggest that the use of DNA delivery devices (e.g., electroporators and biojectors), and increasing the number of vaccine doses and dosage could more reproducibly elicit CD4<sup>+</sup> and CD8<sup>+</sup> T cell responses (Jin et al., 2015).

The main limitation associated with DNA vaccines is their inability to induce long-term immune responses following a single or a few vaccinations (Abbink et al., 2017). Furthermore, DNA vaccines are poorly effective and not well-optimized in eliciting immunity in the liver, gut or genito-rectal mucosa which warrant further refinements of DNA-based vaccination regimens in order to elicit durable protection against HIV-1 and/or HCV.

### THE POTENTIAL OF TISSUE-RESIDENT MEMORY T CELLS FOR CONTROLLING HIV-1 AND HCV INFECTIONS

Since the initial discovery of highly cytotoxic memory T cells residing in tissues (Masopust et al., 2001), several studies have shown that CD8<sup>+</sup> tissue-resident memory T (TRM) cells residing in the female reproductive tract, the gut, the lung and the liver form a formidable frontline defense against various pathogen infections (Mueller and Mackay, 2016; Rosato et al., 2017). The protective role of CD8<sup>+</sup> TRM cells is primarily due to their ability to (1) maintain a stable and durable population following their formation in tissues even in the absence of cognate antigen encounter following their formation (Gebhardt et al., 2009; MacKay et al., 2012; Beura et al., 2018; Park et al., 2018), and (2) produce anti-viral cytokines and/or exert cytotoxic functions to reduce the number of pathogen-infected cells and to recruit other immune cells (e.g., circulating memory T cells) rapidly to the site of infection (Schenkel et al., 2013; Muruganandah et al., 2018; Park et al., 2018). Furthermore, CD8<sup>+</sup> TRM cells respond more rapidly, produce greater amounts of anti-viral/cytotoxic molecules (i.e., in the liver) and appear to be crucial for protection against liver tropic pathogens and pathogens exposed in the vagina and the female reproductive tract compared to circulating memory T cells (Cuburu et al., 2012, 2015; Shin and Iwasaki, 2012; Fernandez-Ruiz et al., 2016; Beura et al., 2018). The greater frequency of intrahepatic CD8<sup>+</sup> TRM cells (CD69<sup>+</sup> CD103+) amongst the total CD8<sup>+</sup> T cell population correlated with partial control of viraemia in Hepatitis B Virus (HBV)-infected patients (Pallett et al., 2017), providing further encouragement that intrahepatic HCV-specific CD8<sup>+</sup> TRM cells will likely be protective against HCV.

Despite HIV-1 and HCV being highly mutable with a complex and evolving quasispecies, several studies have revealed that only one or few variants, referred to as transmitted/founder (T/F) viruses, establish infection following transmission reflecting a strong genetic bottleneck (Bull et al., 2011; Joseph et al., 2015). T/F viruses will be exposed in the genito-rectal mucosa (i.e., the vagina and the rectum) during the vast majority (>80%) of HIV transmission and in the liver during HCV transmission. Thus, eliciting HIV- and HCV-specific CD8<sup>+</sup> TRM cells in the genitorectal mucosa and the liver, respectively, following vaccination is also an attractive strategy to circumvent issues associated with viral diversity and eliminate these viruses shortly after transmission/exposure. Several vaccine vectors such as radiation attenuated sporozoites (RAS), protein loaded nanoparticles (NP), adenovirus (Ad) vectors, adeno-associated virus (AAV), and HPV pseudovirus (HPV PsV) have been developed to elicit localized protection and in some instances elicit CD8<sup>+</sup> TRM cells in the liver or the vagina (**Figure 1**) (Cuburu et al., 2012, 2015, 2018; Fernandez-Ruiz et al., 2016; Ishizuka et al., 2016; Gola et al., 2018). This provides hope that a vaccine to elicit intravaginal or intrahepatic CD8<sup>+</sup> TRM cells can be developed to potentially provide protection against HIV-1 or HCV, respectively.

A recent study suggests that strategies that can induce interleukin (IL)-15 and/or inflammation in the liver can be effective in recruiting circulating effector CD8<sup>+</sup> T cells to differentiate into CD8<sup>+</sup> TRM cells in the liver (Holz et al., 2018). Systemic immunization strategies that promote up-regulation of gut homing molecules such as α4β7 on antigen-primed CD8<sup>+</sup> T cells in secondary lymphoid organs can be efficiently recruited to establish residency in the gut (Masopust et al., 2010). Although these studies and others suggest that local antigen deposition and antigen encounter are not essential to elicit CD8<sup>+</sup> TRM cells, it is evident that this process leads to the formation of greatest densities of CD8<sup>+</sup> TRM cells especially at sites such as the liver and the vagina (Cuburu et al., 2012; Fernandez-Ruiz et al., 2016; Davies et al., 2017). Intravenous delivery of vaccine vectors appears to be most efficient route to facilitate local, intrahepatic expression of vaccine encoded antigens (Nganou-Makamdop et al., 2012; Tay et al., 2014; Gola et al., 2018), to elicit high numbers of intrahepatic CD8<sup>+</sup> TRM cells, and protection against hepatotropic pathogens compared to intradermal and intramuscular vaccine delivery routes (**Figure 1**) (Epstein et al., 2011; Fernandez-Ruiz et al., 2016; Ishizuka et al., 2016; Gola et al., 2018). In the vagina, several studies suggest that intravaginal delivery of vaccine vectors [HPV PsV, and Ad serotypes 26 (Ad26) and 35 (Ad35)] is the most efficient route to express vaccine-encoded antigens in vaginal tissues and elicit cervicovaginal CD8<sup>+</sup> TRM cells (**Figure 1**) (Cuburu et al., 2015, 2018; Fernandez-Ruiz et al., 2016). Furthermore, topical application of chemokine ligands (CXCL9 and CXCL10) in the vagina have been reported to "pull" systemically primed effector CD8<sup>+</sup> T cells into the vagina and allow these cells to differentiate into CD8<sup>+</sup> TRM cells (Shin and Iwasaki, 2012).

### CAN WE EXPLOIT DNA VACCINES TO ELICIT TISSUE-RESIDENT MEMORY T CELLS FOR PROTECTION AGAINST HIV-1 OR HCV?

There has been much research and progress made to improve the immunogenicity of DNA vaccines with respect to the choice of adjuvants, route of vaccine delivery, codon optimization of genes, method of delivery (e.g., electroporation and gene gun), etc. These aspects have been reviewed extensively elsewhere (Nagata et al., 1999; Garmory et al., 2003; Jechlinger, 2006; Vanniasinkam et al., 2006; Jorritsma et al., 2016) and the resulting refinements have led to DNA vaccines being more effectively exploited for use in Phase I and II clinical trials especially in the context of cancer (Kim et al., 2014; Trimble et al., 2015). However, vast majority of the studies including those progressing to the clinic have delivered DNA vaccines using intradermal or intramuscular routes. These routes may not be as effective compared to intravenous route to elicit intrahepatic CD8<sup>+</sup> TRM cells or the intravaginal route to elicit cervicovaginal CD8<sup>+</sup> TRM cells (**Figure 1**).

As mentioned above, it is important that a vaccination regimen designed to elicit CD8<sup>+</sup> TRM cells facilitate local antigen presentation to naïve and antigen experienced precursors of CD8<sup>+</sup> TRM cells, which is best achieved by the local expression of vaccine-encoded antigens and/or promoting local inflammation (**Figure 1**). Manual massaging (Liu et al., 2004), hydrodynamic injections (Yu et al., 2014), and liposome complexes (Kawakami et al., 2000) are some commonly used techniques to transfect hepatocytes in vivo following intravenous delivery of DNA. The expression of vaccine-encoded antigens in hepatocytes is a common hallmark of studies that have elicited intrahepatic CD8<sup>+</sup> TRM cells (Fernandez-Ruiz et al., 2016; Ishizuka et al., 2016), but none of these delivery strategies have led to a licensed vaccine for use in humans. Furthermore, it not known whether any of these strategies can elicit intrahepatic CD8<sup>+</sup> TRM cells in humans mainly owing to the difficulties of isolating liver biopsies in healthy patients although fine needle aspirates may be used to less invasively sample liver-resident T cells (Gill et al., 2018a,b) and the lack of biomarkers (i.e., in the blood) that can accurately predict the formation of CD8<sup>+</sup> TRM cells in the liver and other tissues/organs. In the vagina, a proof of concept study has shown that DNA can be expressed following submucosal intravaginal delivery of DNA in mice (Sun et al., 2015). However, the same study

reported poorly immunogenic responses in mice and noted that electroporation was required to improve the immunogenicity of intravaginally delivered DNA which could be difficult to exploit in humans.

DNA can be used as a vector to prime high numbers of circulating antigen-specific T cells (Gummow et al., 2015; Wijesundara et al., 2018) which can then be recruited to the liver using vectors that efficiently enter cells in the hepatic tissues or the vagina using chemokine ligands or vectors that transduce vaginal epithelial cells (**Figure 1**). Furthermore, given the poor transfection efficiency and immunogenicity of DNA when delivered into the vagina or the liver, it is more feasible to exploit DNA as an immune priming agent in a vaccination regimen to elicit HIV-1- or HCV-specific CD8<sup>+</sup> TRM cells in the vagina or the liver, respectively. Furthermore, analogous strategies using protein-based T cell priming agents in prime/pull (Shin and Iwasaki, 2012), prime/trap (Fernandez-Ruiz et al., 2016), or prime/target (Gola et al., 2018) regimens have been used to elicit protective cervicovaginal or intrahepatic CD8<sup>+</sup> TRM cells. A caveat in this case is to determine whether the primed T cells express adequate levels of chemokine receptors (e.g., CXCR3 and CXCR6) necessary to home to the liver (Sato et al., 2005; Tse et al., 2014; Gola et al., 2018; Olsen et al., 2018) or the vagina (Shin and Iwasaki, 2012) following DNA immunization. Even if not obligatory, the expression of the relevant homing receptors could be required to ensure that high densities of primed CD8<sup>+</sup> T cells are recruited to the cervicovaginal mucosa or the liver following introduction of a vaccine vector or an agent (**Figure 1**) to facilitate the formation of CD8<sup>+</sup> TRM cells. Several studies have shown that the number of CD8<sup>+</sup> TRM cells is a crucial parameter that dictates the collective ability of these cells to confer protection against pathogens exposed in the skin, liver, or the vagina with greater numbers favoring protective outcomes (Cuburu et al., 2012; Shin and Iwasaki, 2012; Fernandez-Ruiz et al., 2016; Park et al., 2018).

## CONCLUDING REMARKS

DNA has recently re-emerged as an effective vaccination platform in humans, but its use in developing a T cell-based vaccine will likely rely on its ability to be exploited in a regimen that can elicit robust immunity in the vagina and the gut in the context of HIV-1, or the liver in the context of HCV. In this regard, we have highlighted the importance of eliciting cervicovaginal or intrahepatic CD8<sup>+</sup> TRM cells against these viruses and also reviewed strategies as well as caveats associated with using DNA to elicit localized CD8<sup>+</sup> TRM cells as a frontline defense against HIV-1 and HCV.

## AUTHOR CONTRIBUTIONS

DW and ZM conceived the initial drafts of the manuscript. BG-B, MM, AS, CR, RB, AL, and EG revised many parts of the manuscript and contributed to finalize the manuscript.

#### FUNDING

The following grants have supported the work conducted in our laboratories and cited in the manuscript: From the National Health and Medical Research Council (NHMRC): grants APP1026293 (EG), APP525431

#### REFERENCES


(CR), APP543139 (EG), and APP543143 (EG). From the Australian Centre for HIV and Hepatitis Virology Research, CR and EG have received an EOI grant. From The Hospital Research Foundation (THRF) and the Channel 7 Children's Research Foundation, DW received a project grant.

#### ACKNOWLEDGMENTS

THRF have provided early career fellowships for DW and AS. AL is supported by a Fellowship from the NHMRC (NHMRC; No. 1043067).


randomised, open-label, phase 1 clinical trials. Lancet 391, 552–562. doi: 10.1016/S0140-6736(17)33105-7


Jechlinger, W. (2006). Optimization and delivery of plasmid DNA for vaccination. Expert Rev. Vaccines 5, 803–825. doi: 10.1586/14760584.5.6.803


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

# The Future of Flu: A Review of the Human Challenge Model and Systems Biology for Advancement of Influenza Vaccinology

Amy Caryn Sherman\*, Aneesh Mehta, Neal W. Dickert, Evan J. Anderson and Nadine Rouphael

*Department of Medicine, Division of Infectious Diseases, Emory University, Atlanta, GA, United States*

Objectives: Novel approaches to advance the field of vaccinology must be investigated, and are particularly of importance for influenza in order to produce a more effective vaccine. A systematic review of human challenge studies for influenza was performed, with the goal of assessing safety and ethics and determining how these studies have led to therapeutic and vaccine development. A systematic review of systems biology approaches for the study of influenza was also performed, with a focus on how this technology has been utilized for influenza vaccine development.

#### Edited by:

*Yanmin Wan, Fudan University, China*

#### Reviewed by:

*Francesco Santoro, University of Siena, Italy Kevin Alby, University of North Carolina at Chapel Hill, United States*

> \*Correspondence: *Amy Caryn Sherman amy.sherman@emory.edu*

#### Specialty section:

*This article was submitted to Clinical Microbiology, a section of the journal Frontiers in Cellular and Infection Microbiology*

> Received: *31 January 2019* Accepted: *28 March 2019* Published: *17 April 2019*

#### Citation:

*Sherman AC, Mehta A, Dickert NW, Anderson EJ and Rouphael N (2019) The Future of Flu: A Review of the Human Challenge Model and Systems Biology for Advancement of Influenza Vaccinology. Front. Cell. Infect. Microbiol. 9:107. doi: 10.3389/fcimb.2019.00107* Methods: The PubMed database was searched for influenza human challenge studies, and for systems biology studies that have addressed both influenza infection and immunological effects of vaccination.

Results: Influenza human challenge studies have led to important advancements in therapeutics and influenza immunization, and can be performed safely and ethically if certain criteria are met. Many studies have investigated the use of systems biology for evaluating immune response to influenza vaccine, and several promising molecular signatures may help advance our understanding of pathogenesis and be used as targets for influenza interventions. Combining these methodologies has the potential to lead to significant advances in the field of influenza vaccinology and therapeutics.

Conclusions: Human challenge studies and systems biology approaches are important tools that should be used in concert to advance our understanding of influenza infection and provide targets for novel therapeutics and immunizations.

Keywords: human challenge model, influenza vaccination, systems biology, universal influenza vaccination, bioethics, transcriptomics, metabolomics

### INTRODUCTION

Although influenza virus was recognized as an important pathogen over a century ago, influenza continues to cause a significant burden of disease. In the United States alone, it's estimated that in the 2017–2018 season there were 959,000 hospitalizations related to influenza illness, and 79,400 deaths (CDC, 2018). Worldwide, WHO estimates that annual influenza epidemics cause 3–5 million cases of severe disease, with 290,000–650,000 of these severe cases resulting in death (Influenza (Seasonal), 2018). Although annual influenza immunizations are recommended and antivirals are available, both have several limitations. The efficacy of the seasonal influenza vaccine is compromised by several factors: antigenic changes over time (requiring a strain specific match each year), slow manufacturing processing, vaccine strain egg-adapted changes, short duration of protection, lack of cross-reactivity, and poor immunogenicity in certain populations (e.g., the elderly) (Goodwin et al., 2006; Soema et al., 2015; Raymond et al., 2016). Antiviral agents such as neuraminidase inhibitors are most effective if administered early in the disease course, and even then have only a modest impact upon the duration of clinical symptoms (McNicholl and McNicholl, 2001). Furthermore, data are inconclusive regarding the ability of neuraminidase inhibitors to reduce the risk of complications, such as hospitalizations or progression to pneumonia (Doll et al., 2017).

Novel platforms to understand influenza immunology are essential in order to address the burden of influenza disease and develop a more effective influenza vaccine that does not rely on annual updates. Combining old modalities—human challenge studies—with new technology—systems biology—has the potential to lead to exciting discoveries that can achieve this goal. There are many reasons why human challenge studies are essential for scientific progress, especially for the influenza field. Human challenge models have several benefits over traditional models such as animal and epidemiological models. Although mice and ferrets have been used in influenza challenges, animal models do not directly translate to humans in regards to their baseline immunity and subsequent immunological responses. Epidemiological or field studies have also been applied to study influenza vaccine efficacy, such as Petrie et al who followed a cohort of over 1,000 individuals from 2014 to 2015 to assess vaccine effectiveness (Petrie et al., 2017). However, these studies require a large sample size and often require numerous sampling points and several years to acquire sufficient data, with many conflicting and cofounding variables. In contrast, the human challenge model is efficient (relatively few participants are required to power a study), immunological responses of humans can be studied directly, and the exact timing of infection is known so that specific time points and measurements are precisely determined.

Human challenge studies for influenza are particularly attractive, with current national emphasis on development of a universal influenza vaccine. As outlined by NIAID's strategic plan, a universal flu vaccine would be at least 75% effective, maintain protection for at least 1 year, protect against group I (e.g., H1, H5) and II (e.g., H3, H7) influenza A virus strains, and be effective for all age groups (Erbelding et al., 2018). The strategic plan also elucidated how a human challenge model could offer unique benefits to better understand the concept of imprinting, determine correlates of protection against influenza, and evaluate different universal influenza vaccine candidates.

In this review, we will examine influenza human challenge studies that have been conducted and their safety, as well as review the ethical considerations in designing a challenge model. We will also review how the use of systems biology techniques in the context of human influenza challenge studies has great potential to advance our current understanding of the host response to acute influenza infection, and ultimately aid in the development of a universal influenza vaccine. The PubMed database was used to search for relevant clinical trials related to these subjects. We propose that successful integration of the right model (the human challenge study) in combination with systems biology approaches will help to better understand the immunological mechanisms of influenza infection and effects of vaccination, which will ultimately aid in the development of an improved vaccine (and perhaps even a universal influenza vaccine), and novel influenza therapeutics.

#### History and Safety of Influenza Challenge Models

In medical research, there is a long and complex history of human challenge studies, in which healthy participants are intentionally infected in order to study the natural history of a disease or to test experimental therapeutic and preventative measures. Perhaps the most famous challenge study in infectious disease was Edward Jenner's use of cowpox in 1796. Although he is not the first to use intentional infection to protect against disease—historical records show that similar practices were likely occurring in Africa, India and China (Gross and Sepkowitz, 1998) long before the eighteenth century—he was responsible for publishing and popularizing the idea. Jenner inoculated the 8-year-old boy James Phipps with cowpox derived from a lesion on the hand of a dairymaid, Sarah Nelms (Riedel, 2005). This successful challenge lead to the creation of the first vaccine, with widespread use in Europe by the year 1800, and eventual eradication of smallpox.

The first well described influenza challenge study was published by Smorodintseff et al. in 1937. The authors infected 72 volunteers via inhalation of a human influenza virus, that had been maintained through the passage of ferrets and mice (Smorodintseff et al., 1937). They discovered that only a small percent (∼20%) developed disease that was mild in intensity. The model was deemed safe and was utilized for several decades thereafter to understand immune responses to influenza and test preventative and therapeutic measures.

However, in the early 2000s, influenza human challenge studies came to a halt. This was a direct result of an adverse outcome associated with a human challenge study that was investigating the use of peramivir as a prophylactic agent (Ison et al., 2005). The subject was a 21-year-old, previously healthy individual with no prior cardiac history. Following the infection with mild influenza B infection and receipt of peramivir, he had asymptomatic ECG changes (described as new T wave inversions of leads II, aVF, and v4-6) on day 4 of the study. Repeat ECG at day 15 had returned to baseline, and he had no cardiac symptoms or enzyme elevation of CPK. He then traveled to Indonesia for 2 weeks, and became ill with an URI. When he returned to the US, he had an echocardiography performed 51 days after challenge since he had had initial ECG changes earlier. He had new reduced ejection fraction, with left ventricular enlargement and remained asymptomatic. Extensive work-up was unrevealing for infectious etiologies. Over the next 5 months, repeat echocardiograms showed gradual improvement and return to normal ejection fraction.

Despite the subject's return to baseline health, and despite the lack of direct causality linking myocarditis to the influenza challenge stock, no further influenza challenge studies were conducted in the US for nearly a decade later. Internationally, as well, only a few challenge studies for influenza were conducted in this time period. After the H1N1 pandemic, Memoli et al. at the NIH re-visited the human challenge model with their validation of a A/H1N1 challenge strain (Memoli et al., 2015). His group challenged 46 healthy participants with a virus that was rescued using reverse genetics (A/CA/04/2009) , with the goal to determine the dose needed to produce mild to moderate influenza infection in at least 60% of the participants who had baseline HAI titers of ≤1:40. The experimental influenza virus was delivered intranasally, and the participants were kept in isolation for at least 9 days following the challenge. The participants' symptoms were monitored and their immune response (serum cytokines, HAI and neuraminidase inhibition titers) documented at specific pre- and post-challenge time points. Discharge from isolation occurred only after the participant had two negative nasal washes on consecutive days. All of the participants demonstrated clinical symptoms of infection (as intended), and 70% of participants had both viral shedding and symptoms. At the dose of 10<sup>7</sup> TCID50, 85% of the participants who received that dose had a ≥4 fold rise in HAI titer by week 8. No significant adverse effects or complications of the influenza infection occurred. Carrat et al. also performed a large and thorough review of 56 different influenza challenge studies, and confirmed that infection from a challenge stock induced only mild disease, with one third of participants having a fever and one fifth of participants developing lower respiratory symptoms (Carrat et al., 2008). This extensive review, in addition to further studies by Memoli's team (Memoli et al., 2016; Hunsberger and Memoli, 2017; Han et al., 2018), demonstrated that influenza challenges can be implemented safely and used in influenza research.

#### Ethical Considerations for Human Challenge Studies

Human challenge studies raise specific ethical concerns because they place research subjects at risk with no potential for direct benefit. The main considerations that must be weighed include the level of risk to the participants, the overall social value of the study, the absence of good alternative study designs, and informed consent.

In human subject research, the risk-to-benefit ratio is usually favorable to the subject, before the risk-to-benefit ratio is weighed for the broader society (Pollard et al., 2012). For example, an individual with a rare and life-threatening disease may choose to participate in a trial that tests a novel therapy, because they personally may benefit from a cure for an otherwise untreatable condition. However, for the human challenge model, the participant is not directly gaining anything for their health, and the risk-to-benefit ratio leans toward the side of risk. Therefore, an acceptable challenge model must have risks that are reasonable to the participants, because the challenge model can be justified only by the benefits to society and not to the individuals. To mitigate risk to the participants, certain criteria should be evaluated. The study hypothesis must only be answerable by challenging human subjects; if the question and endpoint of the study could be determined by animal models or in vitro techniques, the human challenge model should not be used. Infected subjects should also have therapy available (in the case of influenza, antivirals such as oseltamivir or peramivir) in the event that they develop severe influenza infection during the experimental challenge. Participants should receive compensation for their time and effort; however, the more controversial topic is how much participants should receive. Is there a certain amount that is too much, and thus coercive? And likewise, is there an amount that is too little based on the risk inherent in the challenge study? Several authors, such as Miller and Grady, have offered a framework to evaluate the ethical considerations of an infection-inducing challenge study (Miller and Grady, 2001). To maintain ethical integrity, other specific considerations must be addressed in designing an influenza human challenge study, as outlined in **Table 1**.

If these criteria and considerations are carried out in a thoughtful manner, influenza challenge studies can be implemented safely and ethically.

A limitation of the human challenge model is the requirement to have healthy volunteers with no significant comorbidities. As stated above, healthy volunteers must be recruited in order to be ethically sound, in order to reduce the overall risk and complications to the individual. However, this limits our understanding of influenza pathogenesis and vaccine performance in high-risk populations who have the greatest likelihood of severe influenza complications, such as the very young and elderly, immunocompromised, and those with comorbidities. Another inherent limitation of this model is the goal to produce only mild to moderate influenza disease, not severe disease. Again, the efficacy of new therapeutics would be most important in the population suffering severe disease, which cannot be tested in this model for ethical reasons.

TABLE 1 | Ethical considerations in the design of an influenza challenge model.

Selection of the appropriate strain and dose of influenza challenge stock to achieve mild to moderate influenza illness.

Strict inclusion and exclusion criteria to ensure healthy volunteers with minimal comorbidities.

Full review of the proposed study by a third-party ethics committee.

Selection of appropriate clinical and microbiological endpoints to minimize risk to participants.

Transparent informed consent and fair compensation for participants.

Facilities and trained staff that can ensure close and careful monitoring of infected participants.

Proof of decreased infectivity (e.g., undetectable virus by molecular testing) upon discharge to eliminate possibility of transmission to general public.

Adequate clinical follow-up and evaluation for adverse events or sequelae of influenza infection.

\**Loosely adapted from Darton et al. (2015) and Miller and Grady (2001).*

## Advancing Influenza Therapeutics With the Influenza Human Challenge Model

The influenza challenge model has proven useful in advancing the development of several current antivirals to the market, while also useful in terminating the clinical development of others. Two large randomized controlled trials were conducted in 1997 by Hayden et al. which investigated the use of oseltamivir, an oral neuraminidase inhibitor, for both prophylaxis and treatment (Hayden et al., 1999). Healthy participants (N = 117) were inoculated with influenza A/Texas/36/91 (H1N1), and given oseltamivir (at two different doses) or placebo. The results showed that prophylaxis and early treatment with oseltamivir significantly reduced symptoms and had anti-viral effects. This trial is the basis for the standard of care practiced today—oseltamvir continues to be recommended for influenza prophylaxis and treatment by clinical treatment guidelines (Uyeki et al., 2019). Peramivir, another neuraminidase inhibitor agent, was also tested using a similar design. Four randomized, double-blind, placebo-controlled trials were conducted, with 288 healthy volunteers inoculated with either A/Texas/36/91 (H1N1) or B/Yamagata/16/88 (Barroso et al., 2005). At the time of the study, oseltamivir was the only oral neuraminidase inhibitor available. Therefore, the authors sought to determine the tolerability and antiviral efficacy of oral peramivir for treatment and prophylaxis of influenza A and B, using the experimental influenza virus. The results showed that an oral dose of 200 mg twice daily or 400 mg once a day was effective for influenza A, and that prophylaxis with peramivir did not significantly reduce viral shedding. The authors also found relatively low blood peramivir concentrations, and recommended pursuing further study with parental dosing. In both the oseltamivir and peramivir studies, the challenge model was essential for understanding the exact timing of infection in order to test the efficacy of the new agents and identify the appropriate dosing schedule.

Testing of therapies that did not work were equally important in advancing the field. For example, a topical interferon inducer was tested in subjects experimentally infected with an influenza A H3N2 strain (Douglas et al., 1975). They found no difference in frequency or severity of illness, or in quantitative viral shedding between the placebo group and the group who received the interferon inducer. Another study investigated a new antiviral (agent ICI 130,685) that was similar to amantadine for both prophylactic and therapeutic use (Al-Nakib et al., 1986). The authors found that that the higher dosage of 200 mg/day of the new agent did have protective efficacy as compared to placebo when used for prophylaxis; however, the number of side effects in this group were double the side effects reported for the placebo group. For the therapeutic group, the new agent did show a reduced mean daily clinical score and decreased virus in the nasal washings as compared to the placebo arm. Yet, these reductions were not statistically significant for viral concentration until day 3, and for symptom score until day 4. Therefore, this product did not forward toward licensure since the risks (increased side effects) were greater than the benefits (only minor reductions in viral shedding and late symptom improvement).

## Advancing Vaccine Development With the Influenza Challenge Model

Human challenge studies have also been critical for the development of influenza vaccines. The first influenza vaccines were developed and distributed in the US in 1938, and provided to soldiers in World War II (Hannoun, 2013). Thomas Francis, who was a leader in the field of influenza and the author of "Doctrine of Original Antigenic Sin," published a small trial in 1940 in which he inoculated active influenza viral cultures into the nares of 11 human subjects (Francis, 1940). He reported that none of the subjects experienced significant signs or symptoms of infection, and one individual showed a rise of influenza antibodies. He proposed that this technique may have potential for vaccine development. Another study in 1942–1943 challenged individuals with active influenza virus who had received an allantoic fluid vaccine at least 4 months before (Henle et al., 1943). Although it was a small study, the investigators were able to determine the efficacy of the vaccine. Of their controls, ten of twenty-eight individuals developed clinical influenza after inhalation of the isolated and active virus. In contrast, only one of forty-four individuals became ill with influenza after receiving the vaccination. Other early investigations had similar goals, and tested vaccine efficacy for both influenza A and B (Francis and Magill, 1937; Francis et al., 1945; Henle et al., 1946). In 1971, Couch et al. investigated the use of a recombinant influenza A vaccine (X-31 influenza A2, Hong Kong variant) in comparison to the standard vaccine (Couch et al., 1971). Two groups of healthy volunteers were given either the recombinant or the standard vaccine, and then challenged with the live virus strain (Hong Kong variant) used in the vaccine a month later. They measured neuraminidase-inhibiting antibody in the serum, neutralizing antibody in nasal secretions, influenza virus in the nasal secretions, and degree of clinical illness. The results demonstrated that the two vaccines were equally effective. Treanor et al. later challenged healthy adults with wild-type influenza strains to compare the efficacy of a trivalent, live, cold-adapted influenza vaccine (CAIV-T) against the trivalent inactivated influenza vaccine (TIV) (Treanor et al., 1999). After challenging the immunized individuals, they measured influenza illness (defined as respiratory symptoms with wild-type influenza virus isolated from nasal passages or >4-fold increase of HAI antibody from baseline). The efficacy of the immunizations was calculated as 85% for CAIV-T and 71% for TIV, although the difference was not statistically significant.

## IMMUNOLOGY AND APPLYING SYSTEMS BIOLOGY TO INFLUENZA VACCINOLOGY Pathogenesis and Immunology of Influenza

Infection Several studies have used experimental influenza inoculation of humans to investigate the pathogenesis and immune mechanisms associated with infection. Many sought to better understand viral infectivity by route of transmission (e.g., aerosolization), or determine if influenza could be transmitted across different host species (Kasel et al., 1965). For example, one study sought to determine if equine influenza virus could produce clinical symptoms if given to a human (Alford et al., 1966).

Others investigated various human immune responses following experimental infection. The mechanism of systemic and local antibody responses to infection was examined in a study conducted by Murphy et al. Their volunteers were children, aged 1.5 to 4.5 years old, and were inoculated intranasally with either A/Alaska/7/77 (H3N2) or A/Hong Kong/123/77 (H1N1) viruses (developed as candidate vaccine viruses) (Murphy et al., 1982). They found that there was a correlation between the IgA HA antibodies recovered from the nasal passageways and the serum IgG response. Thus, they proposed that intranasal inoculation stimulates both systemic and local immune responses and could be used for immunization purposes. Brown et al. sought to further explore the distinction between serum and secretory IgA antibodies in response to infection. By challenging 13 human volunteers with intranasal A/Peking/2/79(H3N2) wildtype virus, they measured serum and nasal IgA antibodies and their subclasses pre- and post- inoculation with the influenza virus (Brown et al., 1985). They found that IgA1 accounted for most of the increase in IgA anti-HA levels after infection, and determined that the origin of serum IgA antibodies to HA were from the mucosa.

Other investigations examined cell-mediated immune responses, challenging healthy volunteers with different strains of influenza A and measuring serum antibody, viral shedding, and other peripheral blood parameters (such as white blood cell counts), as well as local and systemic cytokine responses. In 1977, Dolin et al administered influenza A virus to 19 volunteers in order to assess cell-mediated immune responses up to 4 weeks after the challenge (Dolin et al., 1977). Eight of the nineteen volunteers had clinical symptoms and 4-fold increases of serum antibody, and lymphopenia was described in this cohort. The authors found that depression of lymphocytes and decreased functionality of the lymphocytes were present even at 4 weeks post-challenge. Another study by Hayden et al challenged volunteers with a H1N1 influenza strain and examined the relationship between clinical symptoms and cytokine responses (Hayden et al., 1998). They collected nasal lavage fluids, plasma and serum levels from the participants and analyzed various cytokines (IL-1 β, IL-2, IL-6, IL-8, IFN-α, TGF-β, and TNF-α) over time. The authors described an association between nasal fluid IFN-α, IL-6, and TNF-α levels and fever on day 2 postchallenge, and IFN-α and IL-6 levels with lower respiratory symptoms on days 5 and 6. They found that IL-6 levels were associated with total, systemic, and upper respiratory symptom scores on days 2 and 4. They concluded that IL-6 is likely the main factor involved in causing fever in influenza (not TNF-α, which is usually described as contributing to fever symptoms in other infections) based on the larger magnitude of IL-6 response early in the course of infection, and IFN-α is responsible for early systemic and local symptoms experienced in influenza infection. The authors hoped that their description of the cytokine response to influenza could be utilized to either (a) develop therapeutic agents that would target specific cytokine responses, or (b) use cytokine levels to more accurately measure the impact of an antiviral intervention. An interesting study by Gentile et al. utilized the influenza human challenge model to control for inflammatory responses seen in patients with allergic rhinitis, which they hypothesized would cause more severe disease (Gentile et al., 2001). Gentile et al enrolled 27 participants who had a history of allergic or nonallergic rhinitis, and then inoculated the participants with an influenza A H1N1 strain and measured anti-IgE-induced leukocyte histamine release, plasma histamine levels, and serum IgG, IgA, IgM, and IgE. The results showed no significant enhancement of systemic immune or inflammatory responses after inoculation in the allergic rhinitis group. The authors speculated that perhaps no difference was observed because the experimental infection was so mild in nature.

## Systems Biology and Influenza Immunization

The field of systems biology applied to vaccines uses mathematical modeling, networking, and other measurements to describe and predict the human immune response to vaccines. In addition, systems biology approaches have the potential to better explain the host-viral interaction and elucidate specific immunological pathways and mechanisms in regards to both influenza natural infection and influenza vaccination. Using systems biology methodologies may be especially useful in the context of influenza human challenge studies.

Nakaya et al. have done extensive work describing specific molecular signatures in individuals who have received seasonal influenza vaccines (Nakaya et al., 2011). They compared immune responses in individuals who had received either the trivalent inactivated vaccine (TIV) or live attenuated influenza vaccine (LAIV). They identified molecular signatures that correlated with B cell responses at day 7 and 28 after immunization, and demonstrated how these could be used to predict an individual's later immune response to the vaccine. Several specific genes were identified that corresponded with HAI response, many of which were regulated by XBP-1 (transcription factor). XBP-1 has been shown to be necessary for the differentiation of plasma cells and the unfolded protein response (Iwakoshi et al., 2003). Another important feature of an effective vaccine is its ability to produce and maintain a long-term durable antibody response to provide long-lasting protection. In a separate study, Nakaya et al. investigated antibody responses after influenza vaccination at day 28 vs. 180, with emphasis on identifying molecular pathways associated with either a persistent or waning antibody response with time (Nakaya et al., 2015). One potentially important pathway associated with persistent antibody response involves P-selectin (SELP), which affects mobility of leukocytes to the vaccine administration site from the peripheral blood stream. The authors were also able to identify transcriptional responses to vaccination, with particular interest in micro RNA expression of miR-424. MiR-424 is a regulator of the interferon response that occurs post vaccination. Another contribution was the finding of specific signatures associated with different age groups. The authors demonstrated that in the elderly population, modules TABLE 2 | Using systems biology to study immune responses to influenza vaccination.


associated with antiviral and IFN-related genes were impaired in the early innate response as compared to the younger population. In the elderly population, there was an enhanced NK-cell related expression and higher proportions of monocytes at baseline and post vaccination. It is hypothesized that perhaps these persistent inflammatory responses seen in the elderly may actually be having a negative effect by inhibiting appropriate vaccine immune responses, and perhaps explains the underlying mechanism of immunosenescence.

Several other studies have examined patterns of gene expression in response to influenza vaccination. The results of key experiments are summarized in **Table 2**.

Systems biology approaches have also been used to study human influenza infection, mostly in human cell culture, in order to better define pathogenesis, virulence factors, and immune responses. For example, transcriptomic responses to a human tissue culture cell line infected with various strains of influenza were examined by Josset et al. (2014). By analyzing transcriptomic responses to Influenza A Virus (IAV), the authors described that the avian H7N9 had further adapted to the human host. They also were able to test various therapeutics in vitro, and demonstrated antiviral responses associated with gene expression alterations. Other authors have studied proteomics in macrophages and monocytes of the lung after IAV infection. These studies showed interferon and TNF-alpha expression in response to infection, as well as pathways leading to secretions of specific proteins and antiviral cytokines (Lee et al., 2009; Lietzén et al., 2011; Cypryk et al., 2017). Ultimately, these findings demonstrate that macrophages could be used as a biomarker to determine the severity of influenza infection.

#### Applying Systems Biology to the Human Influenza Challenge Model

To continue to advance our understanding of influenza, the techniques of systems biology should be furthered applied to the influenza challenge model. Very few studies have been published that use both techniques in concert. An influenza human challenge study was conducted by Woods in collaboration with Retroscreen Virology (London, UK), with prevaccination and post-vaccination RNA extracted from volunteers to examine gene expression, with the purpose of exploring new diagnostic options (Woods et al., 2013). Twenty-four healthy volunteers were inoculated with A/Brisbane/59/2007 (H1N1), and 17 were inoculated with A/Wisconsin/67/2005 (H3N2). The peripheral blood transcriptome was then analyzed at several time points over the course of 7 days. The results showed that peripheral blood gene expression after infection had a distinct signature that was specific to either H1N1 or H3N2 infection. Importantly, these genomic signatures were able to identify infected individuals before they manifested symptoms or had only mild, non-specific symptoms. This early recognition could lead to earlier administration of antivirals and have a more significant impact on lessening the severity of influenza illness. Sobel Leonard et al. used the same 17 participants inoculated with H3N2 from Woods' trial to evaluate evolution of the influenza virus at the time of transmission (Sobel Leonard et al., 2016). Using deep sequencing techniques, the nasal washings of the participants inoculated were compared with the viral stock and the reference strain (A/Wisconsin/67/2005) used to create the stock virus. They found that in acute infection in the human host, the influenza virus populations can undergo rapid viral evolution and changes, which largely occurs during a transmission bottleneck effect that is highly selective. Sobel Leonard et al. were able to further characterize these findings by describing viral evolution within the host and identifying genetic selection factors (Sobel Leonard et al., 2017).

A recent study by Jochems et al investigated the role of influenza infection leading to secondary bacterial pneumonia, using a "double" experimental human challenge model and systems biology approaches (Jochems et al., 2018). Subjects were first inoculated with human type 6B Streptococcus pneumoniae (Spn) to simulate carriage, which is an important pre-requisite of clinical pneumococcal pneumonia. The authors then gave the participants live attenuated influenza virus (LAIV) to simulate influenza infection. Using systems biology approaches of nasal secretions, they were able to identify specific immune mediators that control Spn carriage, and determine how influenza infection can affect these pathways. The results showed that LAIV increased the carriage density of Spn by impairing degranulation of nasal neutrophils and decreasing the recruitment of monocytes, which are two mechanisms essential for bacterial clearance. Although the study has limitations—only

#### REFERENCES


one pneumococcal serotype (6B) was used, and the influenza challenge was with LAIV, not wild-type influenza—the findings are important and highlight how the innate immune system is involved in Spn carriage and clearance, and how pre-existing viral infections can negatively affect immune-mediated control of infection. This study also underscores the power of applying systems biology approaches to the human challenge model. The granular details and elucidation of immune mechanisms were only able to be determined by integrating these two techniques.

#### CONCLUSIONS

There is an urgent need to utilize novel platforms that can lead to the development of more effective vaccines and therapeutics for influenza, which continues to cause a significant burden of disease. Human challenge models have been successfully used for centuries. With advancing technologies and new methods to investigate the host-pathogen interaction, human challenge studies will be essential for progress, and can be performed in a safe and ethical manner. Furthermore, systems biology (e.g., transcriptomics, metabolomics, proteomics, lipidomics, etc.) allows for fundamental changes and patterns of the human immune system to be dissected. Harmonizing these two modalities is very promising; future studies should address using systems biology in a human challenge model to identify important gaps in our knowledge of influenza pathogenesis, and identify essential pathways involved in producing effective immune responses to vaccination. The ultimate goal would be to use these methods in concert to discover novel therapeutics, and potentially even lead to the development of a universal influenza vaccine.

#### AUTHOR CONTRIBUTIONS

AS and NR drafted the article. AS contributed to the collection and assembly of data, and the analysis and interpretation of the data. ND, AM, EA, and NR critically revised the article. ND contributed to the revision of the article and provided intellectual content for the ethics section.

#### FUNDING

This work was supported by the NIH NIAID (Vaccinology T32, Award No. T32AI074492).

human influenza: randomized, controlled trials for prophylaxis and treatment. Antivir. Ther. 10, 901–910.


influenza vaccine and live attenuated influenza vaccine correlate with the kinetics and magnitude of interferon signaling in children. J. Infect. Dis. 210, 224–233. doi: 10.1093/infdis/jiu079


nonadjuvanted trivalent seasonal influenza vaccines in early childhood. Proc. Natl. Acad. Sci. U.S.A. 113, 1853–1858. doi: 10.1073/pnas.1519690113


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

# A Pertussis Outer Membrane Vesicle-Based Vaccine Induces Lung-Resident Memory CD4 T Cells and Protection Against *Bordetella pertussis,* Including Pertactin Deficient Strains

#### *Edited by:*

Alberto Moreno, Emory University School of Medicine, United States

#### *Reviewed by:*

Riti Sharan, Texas Biomedical Research Institute, United States Gideon Kersten, Intravacc, Netherlands

#### *\*Correspondence:*

Daniela Hozbor hozbor.daniela@gmail.com; hozbor@biol.unlp.edu.ar

#### *Specialty section:*

This article was submitted to Clinical Microbiology, a section of the journal Frontiers in Cellular and Infection Microbiology

*Received:* 08 December 2018 *Accepted:* 09 April 2019 *Published:* 26 April 2019

#### *Citation:*

Zurita ME, Wilk MM, Carriquiriborde F, Bartel E, Moreno G, Misiak A, Mills KHG and Hozbor D (2019) A Pertussis Outer Membrane Vesicle-Based Vaccine Induces Lung-Resident Memory CD4 T Cells and Protection Against Bordetella pertussis, Including Pertactin Deficient Strains. Front. Cell. Infect. Microbiol. 9:125. doi: 10.3389/fcimb.2019.00125 María Eugenia Zurita<sup>1</sup> , Mieszko M. Wilk <sup>2</sup> , Francisco Carriquiriborde<sup>1</sup> , Erika Bartel <sup>1</sup> , Griselda Moreno<sup>3</sup> , Alicja Misiak <sup>2</sup> , Kingston H. G. Mills <sup>2</sup> and Daniela Hozbor <sup>1</sup> \*

<sup>1</sup> Laboratorio VacSal, Facultad de Ciencias Exactas, Instituto de Biotecnología y Biología Molecular (IBBM), CCT-CONICET La Plata, Universidad Nacional de La Plata, La Plata, Argentina, <sup>2</sup> School of Biochemistry and Immunology, Trinity Biomedical Sciences Institute, Trinity College Dublin, Dublin, Ireland, <sup>3</sup> Facultad de Ciencias Exactas, Instituto de Estudios Inmunológicos y Fisiopatológicos (IIFP), CCT-CONICET La Plata, Universidad Nacional de La Plata, La Plata, Argentina

Pertussis is a respiratory infectious disease that has been resurged during the last decades. The change from the traditional multi-antigen whole-cell pertussis (wP) vaccines to acellular pertussis (aP) vaccines that consist of a few antigens formulated with alum, appears to be a key factor in the resurgence of pertussis in many countries. Though current aP vaccines have helped to reduce the morbidity and mortality associated with pertussis, they do not provide durable immunity or adequate protection against the disease caused by the current circulating strains of Bordetella pertussis, which have evolved in the face of the selection pressure induced by the vaccines. Based on the hypothesis that a new vaccine containing multiple antigens could overcome deficiencies in the current aP vaccines, we have designed and characterized a vaccine candidate based on outer membrane vesicle (OMVs). Here we show that the OMVs vaccine, but not an aP vaccine, protected mice against lung infection with a circulating pertactin (PRN)-deficient isolate. Using isogenic bacteria that in principle only differ in PRN expression, we found that deficiency in PRN appears to be largely responsible for the failure of the aP vaccine to protect against this circulating clinical isolates. Regarding the durability of induced immunity, we have already reported that the OMV vaccine is able to induce long-lasting immune responses that effectively prevent infection with B. pertussis. Consistent with this, here we found that CD4 T cells with a tissue-resident memory (TRM) cell phenotype (CD44+CD62LlowCD69<sup>+</sup> and/or CD103+) accumulated in the lungs of mice 14 days after immunization with 2 doses of the OMVs vaccine. CD4 TRM cells, which have previously been shown to play a critical role sustained protective immunity against B. pertussis, were also detected in mice immunized with wP vaccine, but not in the animals immunized with a commercial aP vaccine. The CD4 TRM cells secreted IFN-γ and IL-17 and were significantly expanded through local proliferation following respiratory challenge of mice with B. pertussis. Our findings that the OMVs vaccine induce respiratory CD4 TRM cells may explain the ability of this vaccine to induce long-term protection and is therefore an ideal candidate for a third generation vaccine against B. pertussis.

Keywords: *Bordetella pertussis*, pertussis, outer membrane vesicles, TRM cells, pertactin deficient strains, protection

#### INTRODUCTION

Although vaccination is the most cost-effective strategy to prevent life threatening infectious diseases, these diseases continue to be the main cause of global morbidity and mortality (Plotkin, 2005). Vaccine preventable diseases (e.g., measles, tetanus, diphtheria, polio, etc.) are responsible for ∼25% of the 10 million deaths that occur annually among children under 5 years of age. In addition, around 25% of adult deaths (15–59 years) are still attributed to infectious diseases. Therefore, there is an urgent need to improve some traditional vaccines of the Expanded Program on Immunization and to design new vaccines for emerging, re-emerging or resurgent pathogens (WHO, 2010a,b). One of the traditional vaccines that needs to be improved is that designed to prevent the severe respiratory disease whooping cough or pertussis. During the last decades the incidence of this disease has increased in adults, adolescents, as well as in children despite good vaccine coverage (Berti et al., 2014; Clark, 2014). Individuals at the extreme ends of life may be the most vulnerable to severe Bordetella pertussis (the etiologic agent of the disease) infection, though hospitalization can be necessary across all age groups (Mbayei et al., 2018).

The incidence of pertussis was drastically reduced following introduction of global immunization with the whole cell pertussis (wP) vaccine, consisting in a suspension of the causative agent B. pertussis killed by heat and chemically detoxified (Cherry, 1984). The wP vaccine was introduced in the 1940 and 1950s and it is still in use, in developing countries for the pediatric population. However safety concerns with wP vaccines (Desauziers et al., 2004; Klein, 2014) and its acceptance diminished in different countries (Romanus et al., 1987; Klein, 2014). This lead to development of acellular pertussis (aP) vaccines containing purified antigenic protein components of B. pertussis (2, 3, or 5 immunogens) (Sato and Sato, 1985; Edwards and Karzon, 1990). The aP vaccines have a better safety profile and gradually replaced wP vaccine in many industrialized countries (Zhang et al., 2012).

During the last two decades the epidemiology of pertussis has changed (Clark, 2014; Tan et al., 2015), with major outbreaks in many developing countries but also in developed countries (Hozbor et al., 2009; Clark, 2012), even in those with high rates of vaccination (He and Mertsola, 2008; Anon, 2010; Clark, 2014; Mbayei et al., 2018). There have been a number of explanations for the resurgence of pertussis, including waning of immunity induced by vaccines, in particular aP vaccines (Koepke et al., 2014; McGirr and Fisman, 2015), pathogen adaptation to escape vaccine induced immunity (Mäkelä P. H., 2000; King et al., 2001; Mooi et al., 2001; He et al., 2003; David et al., 2004; Gzyl et al., 2004; Bottero et al., 2007; Bowden et al., 2016), and the failure of pertussis vaccines, in particular aP vaccines, to prevent infection and spread of B. pertussis.

Regarding pathogen evolution, the first reports were related to polymorphism in genes coding for proteins included in the vaccine [pertactin (PRN) and pertussis toxin (PTx) among others] and later in the pertussis toxin promoter (ptxP) (Advani et al., 2011; Kallonen et al., 2012). Recently, there has been an increase in B. pertussis isolates that do not produce some of the vaccine antigens (Bodilis and Guiso, 2013; Hegerle and Guiso, 2014; Lam et al., 2014). In particular in US, Canada and Australia it was reported that PRN-deficient isolates [PRN(-)] increased substantially in the last years (Lam et al., 2014; Pawloski et al., 2014; Tsang et al., 2014). These isolates are expected to be resistant to the phagocytosis mediated by anti-pertactin antibodies (Hellwig et al., 2003). It has been proposed that the loss of this vaccine antigen probably provides a selective advantage for bacterial survival in populations vaccinated with aP vaccines. Commercial aP vaccines containing PTx, PRN, and filamentous hemagglutinin (FHA) are not as effective as expected in controlling the infection caused by the recent circulating bacteria that do not express PRN (Hegerle et al., 2014). Moreover, recently it was demonstrated in a mixed infection mouse model that PRN(-) B. pertussis colonizes the respiratory tract of aP immunized mice more effectively than the PRN(+) strain, outcompeting the PRN(+) strain (Safarchi et al., 2015).

Regarding waning immunity, it is well known that while wP vaccines induce potent Th1 and Th17 responses, the current aP vaccines are inefficient at promoting Th1 responses, but do induce potent antibody and Th2-polarized responses and weak Th17 responses (Ross et al., 2013; Brummelman et al., 2015). Furthermore, immunization with wP vaccines appear to be more effective than current aP vaccines at inducing immunological memory and in conferring long-term protection against pertussis (Brummelman et al., 2015). Recent data has demonstrated that wP but not aP vaccines induced CD4 T memory cells that reside in the lungs (Allen et al., 2018; Borkner et al., 2018). These respiratory tissue-resident memory CD4 T cells that express CD44+CD62LlowCD69<sup>+</sup> confer long-term protective immunity against B. pertussis.

Possible solutions to improved control of pertussis disease, in the short term, include the return to the use of the wP vaccine in the primary dose and/or add more boosters. However, adding many boosters implies having both, resources to sustain the vaccination schedule and pharmaceuticals capable of responding to the demands of all countries that require them. Regarding those countries that switched to aP, it seems difficult to reintroduce the same wP vaccine since the acceptability of this vaccine by the population had been lost (Plotkin, 2014).

We have designed a new multi-antigen aP vaccine formulation that shares the beneficial properties of current aP vaccines in terms of biosafety and those of wP vaccines in terms of immunogenicity and protective capacity (International patent granted in the USA and in process in other countries, Application Number: PCT/IB2014/060143) (Roberts et al., 2008; Asensio et al., 2011; Gaillard et al., 2014). This new acellular formulation has been obtained from membrane components of B. pertussis (outer membrane vesicles, OMVs) in which antigens are presented in their native conformation, with membraneassociated PAMPs acting as immunostimulatory molecules, such as in the commercial wP vaccines. We have reported that the OMVs-based vaccine was capable of inducing a more robust immune response than current aP vaccines with a Th1/Th17 and Th2 cellular profile (Bottero et al., 2016), that confers long lasting protection against B. pertussis (Gaillard et al., 2014).

In this study we have evaluated whether our OMVs vaccine is capable of overcoming the deficiencies of commercial vaccines in both controlling infections caused by PRN(-) isolate/strain and inducing memory immunity. We found that our OMVsbased formulation has a higher protective capacity against the PRN(-) bacteria than that induced with a commercial aP vaccine. We found that CD4 T cells with a tissue-resident memory (TRM) cell phenotype (CD44+CD62LlowCD69<sup>+</sup> and/or CD103+) accumulated in the lungs of mice after the second OMVs vaccine immunization. CD4 TRM cells were also detected in mice immunized with wP vaccine, but not in the animals immunized with a commercial aP vaccine. The CD4 TRM cell population was significantly expanded through local proliferation following respiratory challenge of mice with B. pertussis. These CD4 TRM cells secreted IFN-γ and IL-17 that have previously been shown to play a critical role in adaptive immunity against B. pertussis infection. Our findings suggest that the OMVs-vaccine is an ideal candidate for the development of a third generation pertussis vaccine.

## MATERIALS AND METHODS

#### Animals

C57BL/6 (8-week-old) mice were obtained from Harlan Laboratories U.K. or the Comparative Medicine Unit, Trinity College Dublin and housed in a specific pathogen–free facility. All animal experiments performed in Dublin were conducted in accordance with the recommendations and guidelines and under licenses approved by the Health Products Regulatory Authority of Ireland in accordance with protocols approved by the Trinity College Dublin Animal Research Ethics Committee. Animal experiments using female BALB/c mice with 3–4 weeks of age, obtained from the Instituto Biológico Argentino (Biol. SAIC, Argentina) were also performed in Argentina. The studies have been approved by Ethical Committee for Animal Experiments of the Faculty of Science at La Plata National University (Argentina, approval number 004-06-15 and 003-06-15).

#### Bacterial Strains and Growth Conditions

Bordetella pertussis Tohama phase I strain, its isogenic mutant strain defective in PRN (Roberts et al., 1991), and PRN-deficient clinical isolate (Bp935) were used throughout this study. Bp935, which was kindly provided by CDC (Atlanta) is a representative PRN(-) isolate obtained from the 2012 Washington (US) outbreak (Pawloski et al., 2014). PRN deficiency in this isolate was caused by IS481 insertion into prn gene, and its MLVA-MLST type is the most prevalent among the isolates from that outbreak (ptxP3, ptxA1, prn2). Bacteria were grown on Bordet-Gengou agar (BGA, Difco) supplemented with 10% defibrinated sheep blood at 36.5◦C for 72 h and plated again on the same medium for 24 h before each infection.

#### Isolation and Characterization of Outer Membrane Vesicles (OMVs)

OMVs were produced and characterized as previously described (Hozbor et al., 1999; Asensio et al., 2011). Briefly, culture samples from the decelerating growth phase were centrifuged and the bacterial pellet obtained was resuspended in 20 mM Tris–HCl, 2 mM EDTA pH 8.5. The suspension was sonicated (ultrasonic bath) in cool water for 20 min. After two centrifugations at 10,000 × g for 20 min at 4◦C, the supernatant was pelleted at 100,000 × g for 2 h at 4◦C. This pellet was resuspended in Tris buffer (20 mM pH 7.6). The samples obtained were negatively stained for electron microscope examination. Protein content was estimated by the Bradford method using bovine serum albumin as standard. The presence of the main immunogenic proteins in the OMVs was corroborated by immunoblot assays using specific antibodies as we previously described (Roberts et al., 2008).

## Formulation of OMV-Based Vaccine

The characterized OMVs that range in size from ∼50 to 200 nanometers in diameter were used to formulate the vaccine with tetanus (5–7 Lf/ dose with a power ≥2 UIA/ml serum) and diphtheria (1–3 Lf / dose with an output of 0.1 UIA/ml serum) toxoids as we previously described (Gaillard et al., 2014). The LPS concentration determined by GC-MS ranged from 0.275 to 0.352 µg per dose of OMV based vaccine. The safety of this vaccine was evaluated by a mouse weight-gain test (WHO, 2007) and the murine and human whole-blood IL-6–release assays (Stoddard et al., 2010; Bottero et al., 2016). To perform the experiments described below we verified that the OMV based vaccine prepared by us fulfilled the WHO criteria for safety in the weight-gain test. The safety of OMV-based vaccines was also confirmed by human whole-blood assays.

#### Immunization of Mice

Groups of 4–5 female C57BL/6 or BALB/c mice (discriminated in the legends to the figures) were immunized with OMVsbased vaccine formulated as previously described (3 µg total protein per dose formulated with alum as adjuvant) (Asensio et al., 2011), 1:50 or 1:10 human dose of aP vaccine BOOSTRIX <sup>R</sup> [GlaxoSmithKline, with composition per human dose: pertussis toxoid (8 µg), pertactin (2.5 µg), FHA (8 µg), tetanus toxoid (20 IU), diphtheria toxoid (2 IU), and alum as adjuvant], or 1:40 human dose of wP vaccine (National Institute of Biological Standards and Control, South Mimms, UK; NIBSC batch 41S) in 200 µl PBS via intraperitoneal (i.p.) injection using a two-dose schedule. The i.p route, although not translatable to humans, was chosen as it allowed us to compare our results with previous studies on the immunogenicity and protective OMVs-based pertussis vaccines that used this route of immunization (Gaillard et al., 2014; Bottero et al., 2016). Two weeks after the second immunization, mice were sacrificed, and the immune response was evaluated in lungs and spleen. For protection assays mice were challenged with B. pertussis by exposure to an aerosol of 5 × 10<sup>8</sup> bacteria per ml or by intranasal inoculation (sublethal dose 10<sup>7</sup> -10<sup>8</sup> CFU 40 µl −1 ) as is described below. In both case, mice were killed 1 and/or 2 weeks after challenge.

## *B. pertussis* Respiratory Challenge

For protection assays, mice were challenged with B. pertussis by exposure to an aerosol of 5 × 10<sup>8</sup> bacteria per ml or by intranasal inoculation (sublethal dose 10<sup>7</sup> -10<sup>8</sup> CFU 40 µl −1 ). For aerosol challenge, B. pertussis bacteria were grown from a frozen stock on Bordet Gengou plates containing glycerol and horse blood (Cruinn) at 36.5◦C. After 3 d of culture, the bacteria were collected in supplemented Stainer–Scholte medium and cultured overnight at 36.5◦C in a shaking incubator at 220 rpm. Bacteria were centrifuged and resuspended in 1% casein solution, and the OD was measured at 600 nm. B. pertussis infection of C57BL/6 mice was performed by aerosol challenge (BP338 strain; 1 × 108 CFU/ml) administered using a nebulizer (PARI TurboBOY SX) over 10 min (Zhang et al., 2012). The course of infection was followed by performing CFU counts on lung homogenates at intervals post-infection (p.i.), as described (Zhang et al., 2012). For intranasal challenge, mice were infected with a sublethal dose (10<sup>7</sup> -10<sup>8</sup> CFU 40 µl −1 ) of B. pertussis clinical isolate/strain. Bacterial counts were performed 7 days after the challenge as described previously (Asensio et al., 2011; Gaillard et al., 2014).

### Intravascular Staining for Discriminating Circulating Cells From Lung-Retained Cells

To discriminate blood-borne circulating cells from lung-localized cells, we used a well-described approach in which antimouse PE-CD45 Ab (eBioscience) was administered i.v. to mice 10 min before they were euthanized and lungs were harvested (Gzyl et al., 2004).

### Isolation and FACS Analysis of Cells From Lung or Spleen Tissue

Lung and nasal tissue mononuclear cell suspensions were prepared by mechanical (chopping with a scalpel) followed by enzymatic disruption of tissue for 1 h at 37◦C with Collagenase D (1 mg/ml; Sigma-Aldrich) and DNAse I (20 U/ml; Sigma-Aldrich). Next, lungs or spleens were passed through a 40 mm cell strainer to a obtain single-cell suspension, followed by RBC lysis. The cells were incubated with CD16/CD32 FcgRIII (1:100) to block IgG Fc receptors. Cells were incubated with LIVE/DEAD Aqua (Invitrogen), followed by surface staining with fluorochrome-conjugated anti-mouse Abs for various markers. To detect cytokines, cells were stimulated with PMA (50 ng/ml) and ionomycin (500 ng/ml) in the presence of brefeldin A (5 mg/ml) for 4 h at 37◦C. The following surface Abs were used: CD45R-PE, CD3-BV421, CD44-BV605 (Biolegend), CD62L-PE-CF594 (BD), CD103-BV786, CD4-APCeF780, CD69-FITC (eBioscience). For detection of intracellular cytokines, cells were fixed in 2% PFA and permeabilized with 0.5% saponin (Sigma-Aldrich, Ireland), followed by staining with IL-17A–PerCP-Cy5.5 and IFN-γ-BV650 (eBiosciences). Fluorescence minus one or non-specific isotype Abs were used as controls. Flow cytometric analysis was performed on an LSR Fortessa, and data were acquired using Diva software (BD Biosciences). The results were analyzed using FlowJo software (TreeStar).

## Ag-Specific IL-17, IL-5, and IFN-γ Production by Spleen Cells

Spleens from untreated and immunized mice were passed through a 40-mm cell strainer to obtain a single-cell suspension. Spleen cells were cultured with sonicated B. pertussis (sBp; 5 µ/ml), or medium only. After 72 h of incubation, IFNγ, IL-5, and IL-17A concentrations were quantified in supernatants by ELISA. sBp was obtained by sonication of the bacterial suspension in PBS (10<sup>10</sup> CFU/ml) on ice using ultrasonic homogenizer (10 × 15 s. pulse; Sonopuls Bandelin). The sonicated bacterial suspension was then centrifuged at 10,000 rpm × 15 min. Supernatant was collected and protein concentration was determined as described above.

## Statistical Analysis

For the analysis of CFU counts in animal lungs, before applying the statistical methods described below we evaluated the normality of the data by using Shapiro-Wilk test (http:// scistatcalc.blogspot.com.ar/2013/10/shapiro-wilk-test-

calculator.html). After verifying that our CFUs data follows a normal distribution, we statistically analyzed them by using one-way analysis of variance (ANOVA) followed by Bonferroni's multiple comparison test (GraphPadPrism <sup>R</sup> ). Differences were considered to be significant when p < 0.05. For fluorescence values analysis, we used Mann-Whitney statistical analysis (p < 0.05). All statistical analysis of data was performed using GraphPadPrism <sup>R</sup> version 6.00 for Windows, GraphPad <sup>R</sup> Software.

## RESULTS

## OMVs Vaccine Protects Against PRN(-) *B. pertussis* Clinical Isolate

We compared the protective capacity of the OMVs vaccine with that induced with a commercially available aP vaccine against infection with B. pertussis Tohama phase I strain [ptxP1 variant and prn1 allele, PRN(+)] or PRN(-) clinical isolate of B. pertussis Bp935. The OMVs vaccine and the commercial aP vaccine conferred protective immunity against the B. pertussis Tohama phase I strain. The bacterial counts in the lungs were significantly lower in immunized compared with non-immunized mice (**Figure 1A**). In contrast, the protective capacity of OMVs vaccine against PRN(-) isolate was significantly higher than that of the

commercial aP vaccine (**Figure 1B**). Mice immunized with the OMVs had 3 Log<sup>10</sup> reduction in colonies 7 days after challenge with B. pertussis PRN(-) compared with the non-immunized mice, whereas the CFU counts in the lungs of mice immunized with the commercial aP vaccine was only reduced by 1 log<sup>10</sup> (**Figure 1B**). Since there may be other unidentified features other than the absence of PRN in the clinical isolate, we also evaluated the efficacy of the studied acellular vaccines against B. pertussis Tohama phase I isogenic bacteria that only differ in PRN expression. We found that immunization with the OMVs vaccine significantly (p < 0.001) reduced the bacterial counts in the lung to near the detection limit of the assay after challenge with the PRN-defective B. pertussis Tohama mutant (**Figure 1C**). Importantly, the protective capacity of commercial aP vaccine was impaired (at least in 1 log10, p < 0.001) against PRN defective mutant strain (**Figure 1C**) compared with that observed again the wild type strain (**Figure 1A**). The findings demonstrate that the current aP vaccine has impaired capacity to protect against mutant strain or certain circulating strains of B. pertussis that lack PRN.

#### CD4 TRM Cell Accumulate in the Lungs of Mice Immunized With the OMVs Vaccine

We used the murine model to evaluate whether the OMVsbased candidate pertussis vaccine induced systemic antigenspecific immune responses and CD4 T cells with a TRM cell phenotype in lungs. The Mills group had already demonstrated that natural infection or immunization with an wP vaccine primed respiratory TRM cells, whereas an aP vaccine did not (Wilk et al., 2017; Allen et al., 2018). Therefore, we performed experiments comparing the responses with that induced with the OMV-based vaccine with aP or wP vaccines, with the wP vaccines acting as positive control. First, we used the murine respiratory challenge model to confirm that the immunization with the different vaccines tested in this study were capable of inducing protection against infection (**Figure 2**). We found that two parenteral immunizations with the OMVs vaccine was highly effective at preventing lung infection; the CFU counts were at or close to background 7 and 14 days post-challenge (**Figure 2**). The protection induced with the OMVs vaccine was similar to that generated with an wP vaccine (**Figure 2** upper panel) and was significantly better than induced with the aP vaccine (**Figure 2**, lower panel).

To examine antigen-specific immune responses to B. pertussis, spleen cells from immunized mice 14 days after the second dose were stimulated with sonicated B. pertussis (sBp), cytokines were quantified in the supernatants by ELISA. The results revealed that spleen cells from mice immunized with the OMVs vaccine produced significantly higher concentrations of IFN-γ and IL-17 than spleen cells from mice immunized with the aP vaccine (**Figure 3**). Bordetella pertussis-specific IFN-γ and IL-17 was also detected in spleen cells from mice immunized with the wP vaccines. However, antigen-specific IL-17 production was significantly stronger in mice immunized with the OMVs vaccine. In contrast, the aP vaccine predominantly induced antigen-specific IL-5. These data demonstrate that OMVs and wP vaccine induce a mixed Th1/Th17 response while the current aP vaccines induce Th2-polarized responses.

We next assessed TRM cell in the lungs (the gating strategy is shown in **Figure S1**). In order to discriminate blood borne circulating leukocytes from lung-retained leukocytes, we used a validated approach (Gzyl et al., 2004) in which a fluorescently labeled anti-CD45 antibody (Ab) was administered i.v. to mice 10 min prior to sacrifice. Circulating cells become labeled with the Ab (CD45+), whereas the Ab cannot penetrate the tissue to stain the lung resident cells and therefore remain unstained (CD45−). We found that the OMVs vaccine and the wP vaccine (used as a positive control) induced cytokine-secreting respiratory CD4 TRM cells. There were significant higher number of lung-resident CD4 T cells that expressed the TRM cell markers CD69 and/or CD103 between in the lungs of mice immunized with OMVs compared with aP immunized or non-immunized mice (14.02 × 10<sup>4</sup> ± 2.09 ×10<sup>4</sup> TRM for OMVs, 14.99 × 10<sup>4</sup> ± 4.70 × 10<sup>4</sup> TRM for wP, 4.44 × 10<sup>4</sup> ± 0.76 ×10<sup>4</sup> TRM for aP, 1.59 × 10<sup>4</sup> ±

FIGURE 2 | Protection of mice against B. pertussis infection induced with the OMVs vaccine compared with wP and aP vaccines. C57BL/6 mice were immunized (i.p.) twice, 4 weeks apart, with either wP, aP, or OMVs vaccines or were non-immunized. Mice were challenge 2 weeks after the second immunization by aerosol with B. pertussis BP338 strain (1 × 10<sup>8</sup> CFU/ml) administered using a nebulizer (PARI TurboBOY SX) over 10 min. Results are shown as the number of bacteria recovered from the mouse lungs at different time points, expressed as the log10 of the means ± SEM (error bars) of the CFUs per lung, plotted on the ordinate for the samples from the lungs of the experimental groups indicated on the abscissa. The experiment involved three biological replicates with the results from a representative one being presented. The dotted horizontal line marks the lower limit of detection. \*\*\*p<0.001 (ANOVA followed by the Bonferroni post-hoc test).

0.26 × 10<sup>4</sup> TRM for PBS; **Figure 4**). To evaluate the functionality of the induced TRM cell we used intracellular cytokine staining for IFN-γ or IL-17. The results showed that the immunization with the wP vaccine or the OMVs vaccine induced IFN-γ or IL-17-secreting CD4 TRM cells in the lungs. Although the numbers were highest in mice immunized with the wP vaccine, the number of cytokine-secreting CD4 TRM cells significantly (p < 0.05) higher in the lungs of mice immunized with the OMVs vaccine when compared with the aP vaccine (**Figure 5**). Furthermore, CD4 T cells with a tissue-resident memory phenotype expanded significantly (5-fold) after B. pertussis challenge (**Figure 6**) in mice vaccinated with the wP or OMVs-based vaccine, but not in mice immunized with the aP vaccine (**Figure 6**). These findings demonstrate that immunization of mice with the OMVs vaccine, like the wP vaccine but not the aP vaccine, is capable of generating TRM cells that expand in the lung after B. pertussis challenge.

#### DISCUSSION

In this study we demonstrated that the OMVs vaccine, like the wP vaccine, promoted induction of protective immunity against B. pertussis lung infection in mice and that while both vaccines were effective at inducing B. pertussis-specific INFγ (marker of Th1 cells) and IL-17 (marker of Th17 cells) by spleen cells, the aP vaccine mainly induced IL-5 (marker of Th2 cells) (Ryan et al., 1998; Bottero et al., 2016). Though antibodies are known to be involved in protection against pertussis, Th1 and Th17 responses are also crucial for mediating adaptive immunity and clearance of B. pertussis from the respiratory tract (Mills et al., 1993; Ryan et al., 1997).

Consistent with previous studies showing that OMV-based vaccine was capable of conferring both long-lasting immunity and protection against pertussis (Gaillard et al., 2014), here we demonstrated that OMV-based vaccine was also effective at inducing respiratory INF-γ- and IL-17-secreting TRM cells. TRM cells were also expanded in the lungs of mice immunized with the wP vaccine, not in mice immunized with a commercial aP vaccine. The failure of aP vaccine to induce TRM cells, even using the i.p. route of vaccination, which induces the strongest immune response in mice, may explain the waning immunity reported in populations immunized with this vaccine (Klein et al., 2013; McGirr and Fisman, 2015). The growing evidences that pertussis affects all age groups suggests that pertussis vaccines, particularly aP vaccines, do not provide long-lasting immunity. The induction of respiratory TRM cells in immunized mice with the OMVs vaccine is an important finding since this memory CD4 T cell population not only provide much more immediate protection than the anti-pathogen responses conferred by naïve T-cells, they also could provide more effective immune protection to the host as demonstrated in other models of infection (Teijaro et al., 2011; Glennie et al., 2017). CD4 TRM cells have been shown to play a crucial role in protective immunity against re-infection with B. pertussis (Wilk et al., 2017) and following vaccination with aP vaccine formulated with a novel adjuvant, comprising agonists for TLR2 and intracellular receptor stimulator of interferon genes (STING) (Allen et al., 2018). Our experiments with the OMV focused on lung TRM cells, but we do not rule out the possibility that OMV induces TRM cells in upper respiratory tract or in the gut. Thus, the immunization with the OMVs as well as wP vaccine induces long-term memory T cells that could spread to different parts of the body and a fraction of these cells would constitute the respiratory TRM cell population. We found that the CD69+CD4<sup>+</sup> TRM cells induced in the lungs by immunization with OMVs or wP vaccines secreted INF-γ and IL-17. Moreover, these CD69+CD4<sup>+</sup> TRM cells were expanded in the lungs after B. pertussis challenge in mice immunized with OMVs or wP vaccine, but not in mice immunized with a commercial aP vaccine.

FIGURE 3 | Cytokine production by spleen cells from immunized mice. C57BL/6 mice were immunized as described in Figure 2. Fourteen days after the last immunization, mice were sacrificed and their spleen cells stimulated with sonicated B. pertussis (sBp) or medium alone (negative control). After 72 h of culture, the concentrations of IFN-γ, IL-17, and IL-5 were determined in the culture supernatant by ELISA. The results are expressed as mean values (± SEM) of three biological replicates with each having 4 mice per group. Significant differences were analyzed for each cytokine between non-immunized and immunized mice. \*p < 0.05, ANOVA followed by the Bonferroni post-hoc test.

The long-lived IL-17-producing Th17 memory T cells have also been detected in the respiratory mucosa of B. pertussisinfected baboons (Warfel and Merkel, 2013). The studies in this non-human primate model showed that Th17 cells persist long after pertussis infection and suggest that these cells play an important role in adaptive immunity to B. pertussis (Warfel and Merkel, 2013). Our findings suggest that, in contrast to the aP vaccine, immunization with the OMVs vaccine is an effective

approach for inducing lung INF-γ and IL-17-secreting TRM cells (present study) and for conferring long lasting protection against B. pertussis colonization (Gaillard et al., 2014). Although the used i.p. route of immunization is not translatable to humans, several previous studies, including our own had used this route as useful and highly reproducible route of systemic immunization as proof-of-principle in mice. However, we have reported that intranasal immunization with an aP vaccine formulated with a novel adjuvant is much more effective than the i.p. route for the generation of B. pertussis-specific TRM cells (Allen et al., 2018). Furthermore, studies with B. pertussis OMV have demonstrated that pulmonary immunization is more effective than the s.c. route of immunization for induction of CD103<sup>+</sup> TRM cells (Raeven et al., 2018). Therefore, a "prime and pull" strategy of systemic priming followed by nasal boosting might be an interesting approach for induction of systemic and local memory immune response against B. pertussis (Shin and Iwasaki, 2012).

The OMVs derived from B. pertussis represent an attractive aP vaccine candidate (Roberts et al., 2008; Asensio et al., 2011; Ormazabal et al., 2014; Hozbor, 2016) not only because of its safety and ability to induce protective Th1, Th17 cells (Mills et al., 1993; Ryan et al., 1997) and TRM cells, but because it contains a greater number of immunogens in conformations close to those found in pathogen, when compared with the current aP vaccines (Hozbor, 2016). This broad immunogenic composition is an important characteristic of our vaccine candidate since is expected to exert a lower selection pressure on the circulating bacterial population than that exerted by the commercial aP vaccines consisting of only a few antigens. The prevalence of bacteria that do not express vaccine antigens in regions that only use aP vaccine provides indirect evidence of selection pressure being exerting by the aP on the circulating bacterial population (Bodilis and Guiso, 2013; Hegerle and Guiso, 2014; Lam et al., 2014). In particular in United States, Canada and Australia it was reported that PRN(-) strains have increased substantially in recent years (Lam et al., 2014; Pawloski et al., 2014; Tsang et al., 2014). The polymorphism in PRN described first and the spread of PRN-deficient isolates later, have elicited deep concerns in the healthcare systems since it was hypothesized that these changes might represent a selective advantage of the bacteria against immunity induced by the aP vaccines. In particular PRN-deficient clinical isolates may have an advantage in an aP-vaccine primed immunity (Martin et al., 2015). It has been reported that PRN-deficient clinical isolates are able to overcome the anti-PRN mediated inhibition of macrophage cytotoxicity in vitro. Moreover, a recent study revealed that recent PRNdeficient B. pertussis clinical isolates harboring ptxP3 variant and prn2 allele remain at higher CFUs/lung and are capable of sustaining infection longer than isolates still producing this adhesin, in mice immunized with a 3-component aP vaccine

mice (Hegerle et al., 2014). The authors of such study speculated that these particular isolates might thus be capable of infecting immunized individuals at an earlier stage of waning immunity post-aP vaccine immunization or post-infection, presenting an advantage when compared to isolates producing PRN. These findings (Hegerle et al., 2014) are consistent with those on the higher fitness of PRN negative strain in the immunized mice recently reported by Safarchi et al. (2015).

Consistent with previous reports (Hegerle et al., 2014; Safarchi et al., 2015), we found that immunization with commercial aP vaccine does not protect against PRN(-) isolate as effectively as against B. pertussis Tohama strain (PRN+). However, the PRN(-) strain used in this study was a clinical isolate that is not isogenic to B. pertussis Tohama strain (PRN+) and contains polymorphisms at other loci that may affect the fitness of these bacteria. Therefore, we also examined protection against a PRN defective mutant derived from B. pertussis Tohama strain. Consistent with previous results (Roberts et al., 1991), we showed that in addition to PRN, other virulence factors and key antigens are equally expressed in parental and derived strain. We found that the commercial aP vaccine exhibits lower level of protection against the PRN(-) strain when compared with the parental PRN(+) positive strain. These results clearly showed the impact of the absence of PRN expression in the effectiveness of aP vaccine against B. pertussis when comparisons are made on strains that contain the same genetic background.

The results obtained here clearly showed that the OMVs vaccine is more effective than a current commercial aP vaccine against PRN(-) strains. Therefore the OMV formulation appears as an attractive vaccine candidate that could replace the current aP without causing concern on the reactogenicity associated with wP vaccines because of the proven safety of the OMVs vaccines (Bottero et al., 2016). Since major limitations of the current aP are their strong selection pressure exerted on the circulating bacterial population and their failure to induce sustained protective immunity, the OMV-based vaccine, that contains high number of antigens and that induces INF-γ and IL17-secreting TRM cells, has the potential to replace the current aP vaccine.

#### ETHICS STATEMENT

This study was conducted in accordance with the recommendations and guidelines and under licenses approved by the Health Products Regulatory Authority of Ireland and Argentina. The protocol was approved by the Trinity College Dublin Animal Research Ethics Committee and the Ethical Committee for Animal Experiments of the Faculty of Science at La Plata National University (Argentina, approval number 004-06-15 and 003-06-15).

#### AUTHOR CONTRIBUTIONS

DH and KM planned the study, interpreted data, and wrote the manuscript. MZ and MW planned the study, interpreted data, and edited the figures and manuscript. FC, EB, GM, and AM performed experiments and laboratory analyses. All authors approved the final manuscript.

#### REFERENCES


#### FUNDING

This study was funded by a grant from the ANCPyT (PICT 2014- 3617), CONICET and FCE-UNLP (Argentina) grants to DH and by a Science Foundation Ireland grant (16/IA/4468) to KM. DH, GM, and MZ are members of the Scientific Career of CONICET. FC and EB are fellows from CONICET. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2019.00125/full#supplementary-material


**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 Zurita, Wilk, Carriquiriborde, Bartel, Moreno, Misiak, Mills and Hozbor. 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.

#### *Edited by:*

Ricardo Ataide, Burnet Institute, Australia

#### *Reviewed by:*

Rhea Jessica Longley, Walter and Eliza Hall Institute of Medical Research, Australia Takafumi Tsuboi, Ehime University, Japan

*\*Correspondence:*

Alberto Moreno alberto.moreno@emory.edu

†These authors have contributed equally to this work

#### *‡Present Address:*

Jairo A. Fonseca, Division of Leukemia and Lymphoma, Department of Pediatrics, Emory University School of Medicine, Atlanta, GA, United States

#### *Specialty section:*

This article was submitted to Clinical Microbiology, a section of the journal Frontiers in Cellular and Infection Microbiology

> *Received:* 30 January 2019 *Accepted:* 15 April 2019 *Published:* 01 May 2019

#### *Citation:*

McCaffery JN, Fonseca JA, Singh B, Cabrera-Mora M, Bohannon C, Jacob J, Arévalo-Herrera M and Moreno A (2019) A Multi-Stage Plasmodium vivax Malaria Vaccine Candidate Able to Induce Long-Lived Antibody Responses Against Blood Stage Parasites and Robust Transmission-Blocking Activity. Front. Cell. Infect. Microbiol. 9:135. doi: 10.3389/fcimb.2019.00135

# A Multi-Stage *Plasmodium vivax* Malaria Vaccine Candidate Able to Induce Long-Lived Antibody Responses Against Blood Stage Parasites and Robust Transmission-Blocking Activity

Jessica N. McCaffery 1†, Jairo A. Fonseca1,2†‡, Balwan Singh<sup>1</sup> , Monica Cabrera-Mora<sup>1</sup> , Caitlin Bohannon<sup>1</sup> , Joshy Jacob1,3, Myriam Arévalo-Herrera<sup>4</sup> and Alberto Moreno1,2 \*

<sup>1</sup> Emory Vaccine Center, Yerkes National Primate Research Center, Emory University, Atlanta, GA, United States, <sup>2</sup> Division of Infectious Diseases, Department of Medicine, Emory University, Atlanta, GA, United States, <sup>3</sup> Department of Microbiology and Immunology, Emory University School of Medicine, Atlanta, GA, United States, <sup>4</sup> Caucaseco Scientific Research Center, Malaria Vaccine and Drug Development Center, Cali, Colombia

Malaria control and interventions including long-lasting insecticide-treated nets, indoor residual spraying, and intermittent preventative treatment in pregnancy have resulted in a significant reduction in the number of Plasmodium falciparum cases. Considerable efforts have been devoted to P. falciparum vaccines development with much less to P. vivax. Transmission-blocking vaccines, which can elicit antibodies targeting Plasmodium antigens expressed during sexual stage development and interrupt transmission, offer an alternative strategy to achieve malaria control. The post-fertilization antigen P25 mediates several functions essential to ookinete survival but is poorly immunogenic in humans. Previous clinical trials targeting this antigen have suggested that conjugation to a carrier protein could improve the immunogenicity of P25. Here we report the production, and characterization of a vaccine candidate composed of a chimeric P. vivax Merozoite Surface Protein 1 (cPvMSP1) genetically fused to P. vivax P25 (Pvs25) designed to enhance CD4<sup>+</sup> T cell responses and its assessment in a murine model. We demonstrate that antibodies elicited by immunization with this chimeric protein recognize both the erythrocytic and sexual stages and are able to block the transmission of P. vivax field isolates in direct membrane-feeding assays. These findings provide support for the continued development of multi-stage transmission blocking vaccines targeting the life-cycle stage responsible for clinical disease and the sexual-stage development accountable for disease transmission simultaneously.

Keywords: malaria, *Plasmodium vivax*, multi-stage, chimeric vaccine, transmission-blocking, P25

## INTRODUCTION

Malaria remains one of the most serious threats to global health. In 2017, there were an estimated 219 million malaria cases resulting in 435,000 deaths worldwide (World Health Organization, 2018). Of the five Plasmodium species that cause malaria in humans, P. vivax is the most widely distributed with ∼2.8 billion people at risk of infection (Guerra et al., 2010). Its wide geographical range is mainly due to the ability of P. vivax to develop within the Anopheles mosquito vector at lower temperatures, allowing for its survival at higher altitudes and temperate climates (World Health Organization, 2014). Furthermore, P. vivax has the ability to produce hypnozoites, dormant liver-stage parasites present in P. vivax but not in P. falciparum (Krotoski et al., 1982), causing relapse infections weeks to months after the initial infection. Effective malaria control programs, therefore, require comprehensive measures that involve targeting both Plasmodium species (Battle et al., 2012; Gething et al., 2012).

Current malaria control efforts have mainly been focused on the use of vector-based interventions, including long-lasting insecticide-treated nets (LLIN), indoor residual spraying (IRS), and preventative therapies. Preventive therapies include intermittent preventative treatment in pregnancy (IPTp) with sulfadoxine-pyrimethamine and seasonal malaria chemoprevention (SMC) in children aged 3–59 months living in areas of high seasonal malaria transmission (World Health Organization, 2018). While these interventions have resulted in a significant reduction in P. falciparum cases (Mendis et al., 2009), P. falciparum vector-based interventions are less efficacious against P. vivax (Bockarie and Dagoro, 2006). Anomalous climate patterns, as well as the emergence of mosquito resistance to insecticides (Corbel et al., 2007; Dondorp et al., 2009) and parasite resistance to antimalarial treatments (Thomas et al., 2016; Haldar et al., 2018; World Health Organization, 2018), pose additional challenges to the prevention and treatment of malaria despite improved malaria control coverage.

Due to the numerous challenges faced by traditional malaria control methods, the development of novel intervention tools is essential. One potential strategy is the use of transmission-blocking vaccines as they are considered one of the best alternatives to achieve malaria control. Since the life cycle of Plasmodium requires the female Anopheles mosquito to ingest gametocytes during a blood meal from an infected human host to reach the mosquito midgut and begin the next stage of development outside the human red blood cells, this transition could be interrupted by anti-parasite antibodies present in the blood meal (Saxena et al., 2007).

There are two kinds of transmission blocking antigens that can be targeted by vaccines: pre-fertilization and post-fertilization antigens. Pre-fertilization antigens are expressed by gametocytes and gametes; antibodies against these antigens can block the formation of zygotes by binding to the gametes (Sauerwein and Bousema, 2015). Post-fertilization antigens are expressed by zygotes and ookinetes, antibodies that recognize these forms prevent the mosquito midgut invasion (Saxena et al., 2007; Sauerwein and Bousema, 2015). Under natural conditions, the human host is not exposed to post-fertilization antigens. However, transmission-blocking vaccination can be used to elicit antibodies targeting post-fertilization antigens that the mosquito will be exposed to during the blood meal.

Of the post-fertilization antigens described to date, the P25 protein present on the surface of ookinetes and oocysts, first described by Tsuboi et al. (1998), is one of the best characterized (Blagborough et al., 2016). P25 mediates several functions including promoting the clustering of the ookinetes and allowing them to survive the midgut proteolytic environment (Gass and Yeates, 1979). P25 also mediates the attachment and invasion of the mosquito midgut by damaging the midgut epithelium (Han et al., 2000; Zieler and Dvorak, 2000; Vlachou et al., 2004), and binding to laminin and collagen IV in the basal membrane which serves as the starting signal for the ookinete to oocyst development (Vlachou et al., 2001; Arrighi and Hurd, 2002).

While previous phase I clinical trials using the P. vivax P25 protein (Pvs25) have demonstrated that humans can produce antibodies against this antigen, an ideal formulation has not been reported. The first clinical trial of a protein-based Pvs25 vaccine candidate formulated with alum as an adjuvant showed poor immunogenicity and no transmission blocking effect (Malkin et al., 2005). A subsequent clinical trial using a protein-based Pvs25 formulated with Montanide ISA 51 as an adjuvant, showed that low doses of the formulation were able to induce transmission blocking immunity, but higher doses were associated with systemic adverse events (Wu et al., 2008). However, pre-clinical and clinical studies aimed at improving the suboptimal immunogenicity observed by immunization with the Plasmodium P25 proteins, Pfs25, and Pvs25 have suggested that the addition of a carrier protein could potentially enhance the immunogenicity of this protein (Qian et al., 2007; Parzych et al., 2017; Radtke et al., 2017).

A vaccine targeting only a transmission-blocking antigen faces challenges in maintaining an antibody response to parasite antigens to which there would be no bosting effect by natural exposure. Furthermore, this type of vaccine would not provide the human host with protection against infection and would likely have low compliance especially if multiple vaccinations are required. We hypothesize that the development of a bifunctional P. vivax vaccine able to target both a blood stage antigen and a sexual stage antigen could provide protection against infection to the vaccinated individual as well as reduce transmission. A multi-stage transmission-blocking vaccine is particularly relevant for P. vivax given the fact that most relapses are asymptomatic (Van den Eede et al., 2011). These individuals are less likely to receive treatment to clear the infection, resulting in longer periods where the parasite can be transmitted to mosquitoes. In addition to targeting a reservoir of malaria transmission, a P. vivax bifunctional blood stage and transmission blocking vaccine may also improve vaccine uptake due to its potential to provide clinical immunity, as well as a reduction in transmission.

Our group has previously defined several CD4<sup>+</sup> T cell epitopes within the erythrocytic stage antigen Merozoite Surface Protein 1 (MSP1) of P. vivax. These epitopes contain features that define them as promiscuous T cell epitopes (i.e., able to bind a broad range of MHC class II alleles) (Caro-Aguilar et al., 2002). Synthetic peptides representing these P. vivax MSP1 T cell epitopes are recognized by lymphocytes from individuals naturally infected with P. vivax (Caro-Aguilar et al., 2002). We have designed and expressed a chimeric P. vivax MSP1 (cPvMSP1) by genetically linking five of these promiscuous T cell epitopes arrayed in tandem conformation to an extended version of the carboxyl-terminal 19kDa fragment of the P. vivax MSP1 Merozoite Surface Protein 1 (PvMSP119) (Fonseca et al., 2016). We have shown that immunization with cPvMSP1 induced significantly increased cellular and humoral immune responses in the murine model when compared to the native protein (Fonseca et al., 2016). Here we report the design, production, and characterization of a chimeric bifunctional protein composed of the previously described cPvMSP1 (Fonseca et al., 2016), now genetically fused to recombinant Pvs25 (cPvMSP1-Pvs25). We hypothesize that cPvMSP1 will serve both as a carrier protein that can improve Pvs25 immunogenicity while also inducing robust anti-blood stage protective immune responses. Here we assessed the cellular and humoral immunogenicity of cPvMSP1-Pvs25 in mice and its ability to induce long-lived plasma cells, as well as the ability of antibodies elicited by vaccination with cPvMSP1-Pvs25 to reduce transmission when tested in functional assays.

## METHODS

#### Ethics Statements

This study including human samples was carried out in accordance with the recommendations of the ICH/GCP guidelines, Comité de Etica para Investigación con Humanos, Centro Internacional de Vacunas (CECIV, Cali, Colombia), and the protocol approved by the CECIV. All subjects gave written informed consent in accordance with the Declaration of Helsinki.

All animal protocols that include experimental animal procedures using mice and NHP were carried out in accordance with the US Animal Welfare Act and approved by the Emory University's Institutional Animal Care and Use Committee and followed accordingly.

#### Design and Biochemical Characterization of the *P. vivax* Chimeric Pvs25-MSP1 Protein

The 861 bp synthetic gene encoding the chimeric P. vivax merozoite surface protein 1 protein (cPvMSP1) used for these studies has been previously described (Fonseca et al., 2016). A 546 bp synthetic gene encoding Pvs25 (codon optimized for expression in E. coli) was produced by Geneart (Regensburg, Germany). The sequence for the synthetic gene was derived from the Salvador I strain (XP\_001608460; A23 to L195), which does not include its signal peptide and the GPI anchor. A sequence encoding the peptide MAVD was added upstream of the amino-terminal A23 for protein expression. The synthetic gene was subcloned into a pET24d(+) vector. For the production of the synthetic gene encoding the bifunctional chimeric protein, the chimeric PvMSP1 plasmid construct was digested with XhoI and the Pvs25 plasmid amplified by PCR using XhoI/NcoI specific primers. The fragments were annealed and then ligated with T4 DNA ligase. The proper configurations of the Pvs25 and cPvMSP1-Pvs25 genes were verified by enzyme restriction analysis and the sequence confirmed using an automatic sequencer. The recombinant pET plasmids were transformed into BL21 (DE3) cells with kanamycin selection. The sequence of the recombinant bifunctional erythrocytic stagetransmission blocking chimeric protein, designated cPvMSP1- Pvs25 (**Figure 1A**), includes: (i) MAVD amino terminus to reduce degradation during synthesis in E. coli and to provide a start signal; (ii) Five promiscuous T cell epitopes derived from P. vivax MSP1 capable of binding to a broad range of MHC class II alleles, arranged in tandem interspaced with GPGPG spacers: PvT4 (N78-L97), PvT6 (F118-H137), PvT8 (L158-D177), PvT19 (L378-S397), and PvT53 (S1058-N1077); (iii) An extended version of the P. vivax MSP1<sup>19</sup> kD protein fragment, which includes two promiscuous T cell epitopes derived from MSP1<sup>33</sup> protein fragment; (iv) A (NANP)<sup>6</sup> tag from the original chimeric PvMSP1 derived from the P. falciparum circumsporozoite protein included for biochemical characterization of antigenic integrity of the chimeric protein and to provide an optional affinity purification tag; and (v) The Pvs25 sequence derived from the P. vivax Salvador I strain, without the signal peptide and the GPI anchor, but including the MAVD sequence derived from the plasmid Pvs25.

Protein expression was induced with 1 mM IPTG, and the soluble Pvs25 was purified with a Ni-NTA affinity column. cPvMSP1-Pvs25 was expressed in inclusion bodies and refolded as previously described (Singh et al., 2001) using 4M concentration of urea in the refolding solution. After refolding, the protein was purified using gel filtration chromatography. The integrity of the proteins was analyzed by western blot using the anti-Pvs25 monoclonal antibody (mAb) N1-1H10 (MRA-471, BEI Resources), an anti-His tag mAb, or the mAb 2A10 that recognize the (NANP)<sup>6</sup> carboxyl terminal tag of the cPvMSP1-Pvs25 (**Figure 1** and **Supplementary Figure 1**). Additionally, endotoxin levels of the purified protein product were determined using the E-Toxate Limulus amebocyte lysate kit (Sigma), according to the manufacturer's instructions, and were determined to range between 25 and 42 EU/mg of protein.

#### Synthetic Peptides

A library of 61 15-mer synthetic peptides overlapped by 11 residues and spanning the complete cPvMSP1-Pvs25 chimeric protein sequence was commercially synthesized by the multiple solid-phase technique (Sigma-Aldrich). Peptide pools were used to characterize cellular reactivity, with the cPvMSP1 peptide pool 1 representing the sequence of the cognate T cell epitopes included in our chimeric construct and the cPvMSP1 pool 2 representing the complete amino acid sequence of the MSP1<sup>19</sup> kD protein fragment. Pvs25 pool 1 and pool 2 represent the amino acid sequence of Pvs25 (**Table 1**).

#### Mice Immunizations

Groups of 10 female CB6F1/J (H-2d/<sup>b</sup> ) mice, 6–8 weeks of age, were purchased from The Jackson Laboratory. The animals were immunized subcutaneously on days 0, 20, and 40, in

the cPvMSP1 C terminal tag. The molecular weight markers (BioRad) are indicated. (D) Polyclonal anti-cPvMSP1-Pvs25 and anti-Pvs25 elicited in rabbits compete for binding with the transmission blocking monoclonal antibody N1-1H10. Fixed amounts of N1-1H10 (1 µg) were tested with 2-fold dilutions of purified rabbit IgG using Pvs25 as antigen. O.D. values (y-axis) are shown for anti-Pvs25 in closed circles and anti-cPvMSP1-Pvs25 in open triangles using polyclonal antibodies ranging from 0.078 ng/ml to 20µg/ml (x-axis). Data are presented as geometric mean values.

the base of the tail and the interscapular area, using 20 µg of either the cPvMSP1-Pvs25 or Pvs25 proteins emulsified in the adjuvant Montanide ISA 51 VG (Seppic). As a control, a group of mice received PBS emulsified in the same adjuvant. A summary of the immunization regimens and groups can be found in **Table 2**. All animal protocols were approved by the Emory

University's Institutional Animal Care and Use Committee and followed accordingly. Rabbits were immunized four times either with cPvMSP1-Pvs25 or Pvs25 at twenty days intervals, and sera were obtained prior to the first immunization and after each immunization. Rabbit immunization and sera collections were performed by Convance Inc.

TABLE 1 | Chimeric PvMSP1 and Pvs25 peptide pools.


Peptide pools used for stimulation of T cells for assessment of cytokine production after the second immunization. cPvMSP1 Pool 1 represents MSP1 T cell epitopes present in cPvMSP1. Pool 2 represents the extended MSP1<sup>19</sup> kD protein fragment. Pvs25 pools 1 and 2 cover the full sequence of the Pvs25 protein.

#### ELISA Assays

The procedures for the assessment of IgG antibody titers, subclasses, and avidity have been previously described (Fonseca et al., 2016). Antibody titers elicited by immunization of mice were determined by ELISA using Immulon 4HBX plates (Thermo Scientific) coated with 1µg/ml of cPvMSP1-Pvs25, Pvs25, or PvMSP1<sup>19</sup> diluted in carbonate buffer as described (Singh et al., 2010).

Briefly, plates were allowed to incubate overnight with 100 µl of the 1µg/ml protein solution. The solution was removed, and plates were washed 3 times with wash buffer consisting of PBS 1X with 0.05% Tween 20. 200 µl of blocking solution, BSA (KPL) diluted 1:10 in distilled water, was added to each well and plates were incubated again for 2 h at 37◦C. Blocking solution was removed without washing. Sera were diluted in a dilution solution composed of 1:20 BSA (KPL) in distilled water at a starting dilution of 1:320. Cutoffs for positive titers were set at the highest dilution of sera where the O.D. was greater than that of the mean plus three standard deviations above the optical densities obtained using pre-immune sera**.** Following a 1-h incubation with 100 µl of the diluted mouse sera at 37◦C, the plates were washed five times with wash buffer before the addition of 100 µl of peroxidase labeled goat anti-mouse IgG antibody (KPL) at 1:1,000 in dilution solution. Plates were again incubated for 1 h at 37◦C before washing five times with wash buffer. ABTS solution (KPL) was used as a substrate following a 1-h incubation. Optical densities were determined using a VERSAmax ELISA reader (Molecular Device Corporation) with a 405 nm filter. Results are presented as the reciprocal of the end-point dilution.

IgG1 and IgG2a subclass profiles of vaccine-induced antibodies were also determined. ELISA assays were performed as described for the determination of antibody titers, except that after incubation with sera the plates were washed and incubated with biotinylated rat anti-mouse mAbs IgG1 or IgG2a, (BD Pharmingen) for 2 h. After washing, the bound antibodies were detected using horseradish peroxidase (HRP)-streptavidin (KPL) and the SureBlueTM TMB Microwell Peroxidase Substrate (KPL). The peroxidase reaction was stopped with the TMB Stop Solution (KPL). Optical densities were determined using a VERSAmax ELISA reader (Molecular Device Corporation) with a 450 nm filter.

The avidity indices of the antibodies were assessed by ammonium thiocyanate elution-based ELISA using sera samples obtained at day 60, corresponding to 20 days after the final immunization, and day 730, (2 years after the first immunization). The avidity ELISA was conducted similarly to the total IgG titer ELISA, with slight modifications. Briefly, serial dilutions of the sera were assayed in the absence and presence of 1M NH4SCN (Sigma Aldrich) in PBS. The plates were incubated for 15 min at room temperature before washing and proceeding with the assay as described above. The avidity index was calculated as the ratio between the antilog of the absorbance curves obtained with (x1) and without (x2) NH4SCN, as previously described (Perciani et al., 2007).

For ELISA competition assays, purified polyclonal anticPvMSP1-Pvs25 and anti-Pvs25 elicited in rabbits (Covance)

#### TABLE 2 | Immunization regimens.


CB6F1/J mice received subcutaneous immunizations at days 0, 20, and 40 with 20 µg of either cPvMSP1-Pvs25 or Pvs25, both emulsified at a 1:1 volume ratio with the adjuvant Montanide ISA 51 VG. An equivalent volume of PBS was emulsified at a 1:1 volume ratio for subcutaneous injections in the adjuvant control group at the same intervals.

were used. Fixed amounts of the monoclonal antibody N1- 1H10 (1 µg) were tested with 2-fold dilutions of purified rabbit IgG using the recombinant Pvs25 protein as antigen. The concentration of polyclonal antibodies required for 50% inhibition of Pvs25-N1-1H10 interaction was then estimated using linear regression.

#### Indirect Immunofluorescence Assays (IFA)

Sera obtained from 10 C5B6F1/J mice after the third immunization with cPvMSP1-Pvs25 were pooled, and antibody reactivity against native P. vivax PvMSP1 protein was evaluated by indirect immunofluorescence. For assessment of antibody reactivity, an aliquot of blood was collected from a P. vivax infected Saimiri boliviensis monkey (kindly provided by Dr. Mary Galinski) into CPD tubes and washed twice using RPMI 1640 medium before the cells were adjusted to 1% hematocrit. Ten microliters of the cell suspension were added to wells of 12-well slides (ICN Biomedicals Inc) and air-dried before storage at −20◦C. To evaluate reactivity, parasites slides were air-dried at RT for 30 min. Afterward, slides were incubated 90 min with mouse sera, diluted at 1:500 in PBS with 0.2% BSA in a dark, moist chamber. After the incubation, slides were washed 3 times with PBS containing Tween 20 (PBS-T), to minimize non-specific binding. Parasites were stained for 30 min at RT in a dark, moist chamber with goat anti-mouse Alexa Fluor 488 (Invitrogen) at a 1:500 dilution in Evans Blue 0.4% in PBS 1X. After staining, microscope slides were washed 3 times and allowed to dry completely. Parasite nuclei were stained using 4′ ,6-diamidino-2-phenylindole dihydrochloride (DAPI) included in the anti-fade mounting medium ProLong Gold (Life Technologies).

For assessment of reactivity to Pvs25, young oocysts were derived in vitro as described (Janse et al., 1985) and produced using the P. berghei transgenic parasite expressing Pvs25 (MRA-904, pv25DR, BEI Resources). Culture smears were stored at −80◦C until need. Slides were allowed to air dry at room temperature for 30 min before being fixed for 10 min in 4% PFA/PBS. Slides were then washed 3 times with PBS 1X and blocked for 1 h in blocking buffer (10% v/v FBS, 1% w/v BSA in PBS). Slides were allowed to incubate overnight with sera from individual rabbits immunized with either Pvs25 or cPvMSP1- Pvs25 at 1:500 in 1% (w/v) BSA in PBS at 4◦C in a wet chamber. The following day, slides were washed 3 times in PBS and then incubated with Alexa Fluor 488-conjugated goat anti-rabbit (H+L) IgG (ThermoFisher) at 1:500 in 0.4% Evans Blue in PBS 1X for 60 min. After washing 3 times, slides were allowed to dry completely (8 h) and mounted with ProLong Gold anti-fade reagent with DAPI (Life Technologies).

#### ELISpot Assays

Ninety-six-well plates were coated with 5µg/ml Pvs25, cPvMSP1-Pvs25, and cPvMSP1 and blocked with complete RPMI (10% FBS, 1% penicillin/ streptavidin, 1% HEPES, and 50µM 2-mercaptoethanol). Bone marrow and splenic cells were then isolated from the CB6FJ/1 mice 2 years post-immunization with Pvs25, cPvMSP1-Pvs25, or the Montanide adjuvant control. To isolate bone marrow cells from the mice, femurs were removed after the mice were euthanized following Emory IACUC approved procedures. Femurs were then placed in complete RPMI, and the ends of the bones were clipped off with sterile surgical scissors. The bone marrow was then flushed from the femur with RPMI using a syringe into a new sterile conical tube. The bone marrow was then passed through the syringe needle several times to resuspend the cells. Similarly, spleens were removed post-mortem and then mashed through a cell strainer using the plunger end of a syringe. Cells were washed and used immediately with no further processing. Cells were serially diluted on prepared plates and incubated for 16 h at 37◦C. The plates were then treated with anti-IgGbiotin (Southern Biotechnology) followed by incubation with streptavidin-alkaline phosphatase (Sigma). Plates were then developed with 5-bromo-4-chloro-3-indolylphosphate (Sigma) until spots appeared, and spots counted with CTL ImmunoSpot software. Results were then normalized to adjuvant control mice.

#### Flow Cytometry Assays

Flow cytometry analyses of cPvMSP1-Pvs25 or Pvs25 specific T cells were conducted to simultaneously analyze IFN-γ at the single-cell level in T cells derived from splenocytes obtained 5 days after the final boosting immunization. Mice were euthanized according to the Emory IACUC approved protocols and spleens were removed. Spleens were transferred into complete media, composed of DMEM, 1% non-essential amino acids, 2 mM L-glutamine, 5% inactivated FBS, 50µM 2-mercaptoethanol, 10 mM HEPES, 100 U/ml penicillin, and 100µg/ml streptomycin. Spleens were then homogenized under sterile conditions, and the homogenized fluid was passed through 200µm nylon strainer to remove clumps and large pieces of tissue. Red blood cells were lysed using 2 ml of BD PharmLyse buffer incubated with for 3 min before centrifugation at 400 g for 5 min and washing with 5 ml of flow cytometry buffer. Cells were counted and the concentration was adjusted to 10<sup>7</sup> cells/ml. 100 µl was then placed in individual round bottom tubes. Cells were stimulated for 6 h with peptide pools at 2µg/ml at 37◦C, in the presence of GolgiPlug (BD Biosciences). Cells were then incubated with Live/Dead Aqua Stain (Life Technologies) followed by surface staining with α-CD3 (PerCP Cy5.5), α-CD4 (Alexa Fluor 700), and α-CD8α (APC-Cy7) for 30 min. The cells were then fixed, permeabilized and stained with antibodies against IFN-γ (APC). All the monoclonal antibodies were obtained from BioLegend. Flow cytometry analyses were performed using an LSRII flow cytometer (BD Biosciences), and data were analyzed using FlowJo software version 10.1. The lymphocytes were initially gated on the Live/Dead channel, and then CD3<sup>+</sup> , and then CD4<sup>+</sup> and CD8<sup>+</sup> populations. Antigenspecific cytokine-secreting T cells were identified within both the CD4<sup>+</sup> and CD8<sup>+</sup> populations. The frequency of antigenspecific cytokine-producing cells was determined by subtracting the percentage of cytokine-producing T cells after incubation with medium alone from the percentage of cytokine-producing T cells after incubation with the peptide pools. Samples that did not meet this requirement were set to zero.

#### Transmission-Blocking Assays

The transmission-blocking activity of sera derived from rabbits immunized with Pvs25 or cPvMSP1-Pvs25 was measured by direct membrane feeding assays as described elsewhere (Arevalo-Herrera et al., 2005, 2015; Vallejo et al., 2016). Briefly, 150 µl of infected RBCs from P. vivax infected patients were washed twice with RPMI 1640 medium (Sigma Aldrich) and diluted in a 150 µl of sera from rabbits that were not heatinactivated obtained after three immunizations with 20 µg of cPvMSP1-Pvs25 or Pvs25 to feed 100 adults (2–3 day old) An. albimanus mosquitoes. Pre-immune sera from the same rabbits were used as a negative control. After 30 min of feeding, unfed mosquitoes were removed from the cages, and fed mosquitoes maintained at 27◦C and 80–90% relative humidity. All procedures were performed at 37◦C. Seven days after feeding, 30–40 mosquitoes were dissected, midguts were stained with 2% mercurochrome, and the numbers of oocysts per mosquito midgut were counted.

## Statistical Analysis

Statistical analysis and graphs were made using GraphPad Prism 5.0 software (GraphPad Software Inc.). For analysis of the antibody responses, all ELISA titers were log-transformed to conform to the normality and variance requirements of parametric testing and compared using Student's t-test for comparison of antibody titers between groups. Student's ttest was used for the comparison of antibody avidity between groups. Mann Whitney was used for the comparison of antibody subclass ratios between immunization groups. Differences in the numbers of antibody secreting cells obtained via ELISpot were analyzed using the Mann-Whitney test to compare responses obtained from the Pvs25 and cPvMSP1-Pvs25 immunization groups. Levels of cytokine IFN-γ production obtained from flow cytometry were analyzed using unpaired t-tests to compare immunization groups. The transmission-blocking activity of the anti-cPvMSP1-Pvs25 or anti-Pvs25 sera compared against naïve sera or between the groups was analyzed using one-way ANOVA.

## RESULTS

## Design, Expression, and Characterization of the Chimeric Protein cPvMSP1-Pvs25

The chimeric PvMSP1 protein used to create a bifunctional vaccine construct by genetic fusion to the Pvs25 protein has previously been reported by our group (Fonseca et al., 2016). Briefly, the protein consists of five experimentally defined promiscuous T cell epitopes (Caro-Aguilar et al., 2002) derived from several regions along the native P. vivax MSP1. These epitopes are arrayed in tandem, interspaced with GPGPG spacers, and genetically fused to an extended form of the PvMSP1<sup>19</sup> kD fragment that contains two T helper epitopes derived from MSP1<sup>33</sup> kD fragment. Initial immunogenicity studies showed that this protein was able to elicit robust cytophilic antibody and CD4<sup>+</sup> and CD8+ T cell responses when delivered either alone as a recombinant protein formulated in water-in-oil emulsion using homologous prime-boost immunization regimens (Fonseca et al., 2016) or via heterologous prime-boost immunization regimens using recombinant adenoviral vectors for priming immunization (Fonseca et al., 2018). To test the feasibility of developing a hybrid vaccine with the potential of inducing blood and mosquito stagespecific immunity, we expressed cPvMSP1-Pvs25 and Pvs25 as recombinant proteins in E. coli (**Figures 1A,B**).

To confirm the biochemical identity of the cPvMSP1-Pvs25 protein, western blot analysis using monoclonal antibodies targeting specific regions of the cPvMSP1 and Pvs25 protein components was conducted (**Figure 1C**). The monoclonal antibody N1-1H10 was used for assessment of both the cPvMSP1-Pvs25 and recombinant Pvs25 proteins, as it recognizes a conformational-dependent epitope (Hisaeda et al., 2001) present in the second epidermal growth factor-like domain of Pvs25 (Saxena et al., 2006) and is able to inhibit oocyst development in the vector (Ramjanee et al., 2007). Anti-His tag antibodies were used to identify the His-tag present at the C terminal end of the Pvs25 protein segment of the cPvMSP1-Pvs25 protein and the recombinant Pvs25. The monoclonal antibody 2A10 targets the (NANP)<sup>n</sup> repeat region derived from P. falciparum circumsporozoite protein, which is present in the cPvMSP1-Pvs25 protein at the carboxylterminal end of the chimeric PvMSP1 protein. We observed recognition of the cPvMSP1-Pvs25 protein as a single band with the expected size of ∼52 kDa by the monoclonal antibody targeting the NANP repeats (2A10), the anti-His tag monoclonal antibody, and the monoclonal antibody targeting the Pvs25 protein (N1-1H10). As expected, we also observed binding of the anti-Pvs25 monoclonal antibody (N1-1H10) and the anti-His-tag antibody to the recombinant Pvs25 protein, shown as a ∼25 kDa band. These experiments confirm that at least the epitopes of these monoclonal antibodies remained correctly folded.

To determine if the Pvs25 functional domains are conserved within cPvMSP1-Pvs25, we used ELISA competition assays (**Figure 1D**). Purified polyclonal antibodies elicited in rabbits by immunization with the recombinant proteins were tested at different concentrations and co-incubated with fixed amounts of the monoclonal antibody N1-1H10 (Ramjanee et al., 2007). The polyclonal antibodies inhibited the binding of N1-1H10 with a distinct competition pattern. The concentration of the polyclonal antibodies required for 50% inhibition of antigen-monoclonal antibody interaction was estimated in 32 µg for anti-Pvs25 and 13 µg for anticPvMSP1-Pvs25. These results suggest that antibodies elicited by immunization with the protein expressed in E. coli have transmission-blocking potential.

### Antibody Response Induced by cPvMSP1-Pvs25

Hybrid CB6F1/J mice were immunized with Pvs25 or cPvMSP1- Pvs25 on days 0, 20, and 40 (**Table 2**). Sera were obtained 20 days after each immunization. When we assessed the anticPvMSP1-Pvs25 antibody titers at day 20 and 40, we observed significantly higher titers in the cPvMSP1-Pvs25 immunized group than in the Pvs25 group at both time points (P = 0.0017 and 0.0262, respectively, **Figure 2A**). At day 60, 20 days after the last immunization, mice immunized with cPvMSP1-Pvs25 had mean antibody titers of 3.4 × 10<sup>6</sup> against the chimeric recombinant protein, significantly higher (P = 0.0414) than those in mice immunized with Pvs25 which had mean antibody titers of 2.4 × 10<sup>6</sup> (**Figure 2B**). At day 730, 2 years after the first immunization, the group immunized with cPvMSP1-Pvs25 had mean antibody titers of 1.7 × 10<sup>6</sup> against cPvMSP1-Pvs25, a reduction of 26.8%. This observation is in sharp contrast with the group of mice immunized with Pvs25 that had mean antibody titers of 3.7 × 10<sup>5</sup> against cPvMSP1-Pvs25, a reduction of 89.2% (P = 0.0107, **Figure 2B**). The data suggest that unlike Pvs25 the antibody response induced by cPvMSP1-Pvs25 is long-lasting in mice.

To assess if antigenic competition occurs in mice immunized with the chimeric protein, antibody titers against the individual components of the bifunctional chimeric protein (Pvs25 and cPvMSP1) were also measured (**Figure 3**). The cPvMSP1- Pvs25 immunization group was able to recognize Pvs25 at day 60 with mean antibody titers of 2.4 × 10<sup>6</sup> , which were similar to levels in mice immunized with Pvs25 (P = 0.9617) (**Figure 3A**). Assessment of anti-Pvs25 titers at day 730 in mice immunized with cPvMSP1-Pvs25 revealed a titers reduction of 47.2%, while the antibody titers in mice immunized with Pvs25 alone were reduced by 86.0%, a significant difference between the groups (P = 0.0402) (**Figure 3A**). As expected, mice immunized with Pvs25 alone did not produce antibodies against PvMSP1, showing only minimal reaction at the highest concentration of sera used. In contrast, mice immunized with cPvMSP1-Pvs25 had mean anti-PvMSP1 antibody titers of 6.6 × 10<sup>6</sup> at 20 days after the final immunization, and 1.0 × 10<sup>6</sup> two years later (P = 0.0027) (**Figure 3D**); the later titers at day 730, remained significantly higher than the Pvs25 immunized group, with a reduction of 84.8% (P = 0.003).

### Characteristics of Anti-Pvs25 and Anti-MSP1 Antibodies

The average avidity index of anti-Pvs25 antibodies induced by immunization with cPvMSP1-Pvs25 at 20 days after the final immunization was significantly lower than that induced by immunization with Pvs25 (0.79 and 0.95, respectively, P = 0.0038) (**Figure 3B**). In contrast, by day 730 the respective values were 0.68 and 0.50 (P = 0.0108), indicating a significantly higher avidity index for antibodies induced by cPvMSP1-Pvs25 compared with Pvs25 alone.

The characteristics of the antibodies against chimeric PvMSP1 were only analyzed in the group immunized with cPvMSP1- Pvs25 since only these mice recognize this antigen. The average avidity index of the cPvMSP1-Pvs25 induced anti-MSP1 antibodies 20 days after the final immunization was 0.74 and remained similar after 2 years (0.71, P = 0.6381) (**Figure 3E**).

The IgG subclasses induced by vaccination with Pvs25 and cPvMSP1-Pvs25 was assessed in order to determine antibody quality, as cytophilic antibodies against MSP1, which correspond to IgG2a in mice, have previously been reported to be associated with protection (Stanisic et al., 2009). Measurement of the Pvs25-specific IgG2a and IgG1 subclasses, expressed as the IgG2a/IgG1 ratio, revealed a higher ratio for the cPvMSP1- Pvs25 immunized mice at both day 60 and day 730, however only at day 730 was the IgG2a/IgG1 subclass ratio of the cPvMSP1-Pvs25 group significantly higher than that of Pvs25 immunized mice (P = 0.0131) (**Figure 3C**). Assessment of the IgG subclass ratios for anti-PvMSP1 responses induced by vaccination with cPvMSP-Pvs25 revealed a different pattern, with the IgG2a/IgG1 ratio of the anti-MSP1 antibodies significantly higher at 60 days than at 730 days after the first immunization (P = 0.0010) (**Figure 3F**), showing a shift toward a Th2 phenotype over time.

#### Anti-cPvMSP1-Pvs25 Induced Antibodies Recognize the Native Parasite Proteins

The ability of the antibodies induced by vaccination with cPvMSP1-Pvs25 or Pvs25 to recognize the native antigens on the surface of oocysts and schizonts was assessed by immunofluorescence (**Figure 4**). Transgenic P. berghei parasites (MRA-904 parasites) expressing Pvs25 on the surface of young oocysts were used to assess the ability of antibodies from cPvMSP1-Pvs25 or Pvs25-immunized rabbits to recognize native Pvs25. As expected, sera from the cPvMSP1-Pvs25 or Pvs25-immunized rabbits were able to bind to the surface of P. berghei Pvs25 transgenic parasites (**Figures 4A,B**). Thin smears made from infected red blood cells from a P. vivax infected Saimiri monkey were used to assess the ability of antibodies from cPvMSP1-Pvs25 immunized mice to recognize

are shown in (C) for Pvs25 and (F) for cPvMSP1. Avidity indices and subclass results are presented as the mean values obtained from sera pooled from the 10 mice at day 60 or the 7 surviving mice at day 730, responses were averaged from four technical replicates. Statistical analysis was conducted using unpaired t-tests. Statistically significant differences are denoted by \*p < 0.05 and \*\*p < 0.01.

the native PvMSP1 protein. Pooled sera from the cPvMSP1- Pvs25 immunized group obtained at day 60 recognized P. vivax schizonts (**Figure 4C**). Combined, these data indicate that chimeric cPvMSP1-Pvs25 immunization elicited antibodies capable of binding the native structure of Pvs25 and the native P. vivax MSP1.

#### cPvMSP1-Pvs25 Plasma Cell Induction

To confirm that the long-lasting antibody responses observed at 2 years post-immunization were related to the induction of longlived plasma cells (LLPCs), antigen-specific IgG plasma cells were measured in CB6F1/J mice immunized with Pvs25, cPvMSP1- Pvs25, or Montanide via ELISPOT 2 years post-immunization. Mice immunized with cPvMSP1-Pvs25 generated a robust, longlived IgG plasma cell response not only to the cPvMSP1-Pvs25 protein itself but also to Pvs25 and cPvMSP1 (**Figure 5**). This response lasts for the lifetime of the animal. The number of longlived IgG plasma cells in the bone marrow specific to cPvMSP1- Pvs25 was significantly higher (P = 0.0456) in mice immunized with cPvMSP1-Pvs25 than those immunized with Pvs25 alone (**Figure 5A**). Furthermore, the group immunized with Pvs25 was unable to generate a plasma cell response significantly above that of the adjuvant-only control. Mice immunized with cPvMSP1- Pvs25 also generated significantly greater numbers of IgG ASCs specific to Pvs25 (**Figure 5B**), and PvMSP1 (**Figure 5C**) compared to mice immunized with Pvs25 (P = 0.0022 and

P = 0.0092, respectively). This increase in long-lived IgG plasma cells was most pronounced in the bone marrow, though antigenspecific IgG plasma cells persist in the spleen at lower numbers. Overall, immunization with cPvMSP1-Pvs25 was more efficient, as it was able to stimulate long-lived, Plasmodium-specific IgG plasma cells when compared to standard Pvs25 immunization.

## Cellular Response Induced by cPvMSP1-Pvs25

Following the assessment of the humoral response induced by vaccination with cPvMSP1-Pvs25 as compared to Pvs25, we sought to determine if cPvMSP1-Pvs25 was able to induce a cellular response able to recognize individual components of the bifunctional chimeric protein. CB6F1/J mice were immunized 3 times with 20 µg of either the chimeric cPvMSP1-Pvs25 or the recombinant Pvs25 on day 0 and 20, as described in **Table 2**. Mice were euthanized 5 days after the second immunization and splenocytes were stimulated with peptide pools representing the recombinant Pvs25 and cPvMSP1 proteins (**Table 1**) to analyze the production of IFN-γ by both CD4<sup>+</sup> and CD8<sup>+</sup> T cells (**Figure 6**). A sample gating strategy is shown in **Supplementary Figure 2**. We observed no difference in the production of IFN-γ between the immunization groups in either the CD4<sup>+</sup> or CD8<sup>+</sup> T cell populations in response to stimulation with the Pvs25 peptide pools (**Figures 6A,B**). Following stimulation with Pool 1 of cPvMSP1, we found that CD4<sup>+</sup> T cells from mice immunized with cPvMSP1- Pvs25 produced significantly higher levels of IFN-γ than those immunized with Pvs25 (P = 0.0010) (**Figure 6C**). Similarly, we found that following stimulation with Pool 2 of PvMSP1, both CD4<sup>+</sup> and CD8<sup>+</sup> T cells from mice immunized with the cPvMSP1-Pvs25 protein produced significantly more IFN-γ than the Pvs25 immunization group (P = 0.0043 for CD4+, P = 0.0138 for CD8+, **Figures 6C,D**).

### Transmission-Blocking Activity of Anti-cPvMSP1-Pvs25 Antibodies

Sera samples, obtained from rabbits after three immunizations with either cPvMSP1-Pvs25 or Pvs25, were tested for transmission-blocking activity using three different P. vivax isolates in independent direct membrane feeding assays. The transmission blocking activity and the number of oocysts counted were assessed for both groups (**Table 3**). The activity of immune sera was compared to pre-immune control sera

FIGURE 5 | IgG-producing long-lived plasma cells. Antigen-specific IgG plasma cells present in the bone marrow or spleen of CB6F1/J mice immunized with Pvs25 (closed circles, n = 7), cPvMSP1-Pvs25 (open triangles, n = 8) were analyzed at two years post first immunization. IgG long-lived plasma cells specific to (A) cPvMSP1-Pvs25, (B) Pvs25, and (C) cPvMSP1 are shown. Statistical analysis was conducted using Mann Whitney. Statistically significant differences are denoted by \*p < 0.05 and \*\*p < 0.01.

(**Table 3**). Upon dissection of mosquitoes, we found that sera obtained from both Pvs25 and cPvMSP1-Pvs25 resulted in significantly lower percentages of infected mosquitoes when compared to the pre-immune sera samples. Similarly, we found significantly lower numbers of oocysts present in the mosquitoes that had fed on the Pvs25 and cPvMSP1-Pvs25 samples as compared to the pre-immune rabbit sera controls. Overall, our results indicate that antibodies induced by cPvMSP1-Pvs25 are able to block the infection in 90% of exposed mosquitos, and in infected mosquitos, the parasite load is 98% lower than in mosquitoes feeding on a non-immunized source.

unpaired t-tests. Statistically significant differences are denoted by \*p < 0.05, \*\*p < 0.01, and \*\*\*p < 0.001.

#### DISCUSSION

Development of novel malaria intervention tools, such as transmission-blocking vaccines (TBVs), is essential to address the increased reports of insecticide and drug resistance (Corbel et al., 2007; Dondorp et al., 2009), as they could increase the efficiency and sustainability of these existing malaria control methods (Sauerwein, 2007). TBVs rely on the generation of antibodies that block the development of Plasmodium parasites within the mosquito midgut. We propose that the ideal TBV formulation should consist of a transmission-blocking component combined with a prophylactic vaccine in order to simultaneously provide protection against disease to the recipient and reduce transmission. Under natural conditions, the human host is not exposed to the post-fertilization antigens. Since there would not be a natural boosting effect during infection, a vaccine targeting a post-fertilization antigen should be highly immunogenic to produce effective antibody responses (Saul, 2007). Here we present evidence that a chimeric protein designed based on one of the best-characterized post-fertilization antigens, P25, conjugated to a chimeric P. vivax MSP1 (cPvMSP1-Pvs25), elicits long-lasting antibody responses against both proteins, without immune interference, while inducing robust cellular responses to PvMSP1 in CB6F1/J, a F1 hybrid of BALB/c, and


 

bAverage calculated as the total of oocyst /total of mosquitoes dissected±standard deviation.

 cPercent inhibition of mean mosquitoes with rabbit sera post-third immunization compared to pre-immune rabbit sera in each independent assay calculated as (1 – [mean mosquitoes in normal rabbit sera/mean mosquitoesimmunizationsera]x100).ApoolofABnormalhumanserawasusedasanegativecontrol.

 post 3rd

 patients as

 

dPre-immune serum from this rabbit was not tested for TBA, antibody titer was 1280.

eStatistical significance P < 0.05. Sera samples from immunized rabbits (cPvMSP1-Pvs25 and Pvs25) were tested in direct membrane-feeding assays (three different experiments) using blood samples collected from naturally infecteddescribed(Vallejoal.,2016). C57BL/6 mice, comparable to those we have previously reported for cPvMSP1 alone in the parental strains (Fonseca et al., 2016).

Assessment of the quality and longevity of the antibody response induced revealed that immunization with cPvMSP1- Pvs25, as compared to Pvs25, resulted in the induction of an antibody response against the immunogens as well as against the individual components, which lasted for the lifetime of the animal with a much lower reduction in antibody titers at 2 years post-immunization. Most critically, we observed a lower reduction in anti-Pvs25 antibody titers at 2 years post-immunization in groups of mice immunized with the bifunctional chimeric protein as compared to the Pvs25 group. These results are significant as an ideal transmission-blocking vaccine candidate must be capable of inducing functional and durable antibodies using a simplified immunization regimen for use in mass administration (Nunes et al., 2014).

For the proof-of-concept studies reported here, we selected Montanide ISA 51 VG, a water-in-oil based adjuvant (Aucouturier et al., 2002) that has been tested before in clinical trials (Baumgaertner et al., 2016; Caballero et al., 2017; Saavedra and Crombet, 2017). Our results are consistent with the reported high efficiency of this adjuvant to induce robust antibody responses given its ability to promote polarization of naïve T cells into T follicular helper cells (Riteau et al., 2016). However, safety concerns have arisen from a Phase I clinical trial of Pfs25 and Pvs25 formulated in Montanide 51 showing systemic adverse events compatible with erythema nodosum (Wu et al., 2008). Recent comparisons of P. falciparum Pfs25 formulated in novel oil-in-water based adjuvants compared with alum-based adjuvants found that oil-in-water-based adjuvants EM081 and EM082 were more efficient in eliciting high titers of anti-Pfs25 IgG antibodies than the alum adjuvants (Patra et al., 2015). Interestingly, this study also found that an adjuvant composed of oil-in-water-glucopyranosyl lipid A-induced high-affinity antibodies that effectively blocked infection of mosquitoes with P. falciparum and demonstrated that avidity could provide a surrogate measure of efficacy beyond the antibody titers (Patra et al., 2015). The strong adjuvant effect was also confirmed by a Phase I clinical trial of a Pfs25 vaccine candidate paired with the aluminum adjuvant Alhydrogel, which observed a rapid decline in antibody responses after vaccination (Talaat et al., 2016). Overall, these experiments corroborate the adjuvant-dependency on the magnitude and durability of the antibody response of P25-based vaccines (Mlambo et al., 2008; Radtke et al., 2017). The long-lasting immunity induced by cPvMSP1-Pvs25 is therefore very encouraging compared to previous studies and warrant further investigation regarding the best adjuvant system to deliver the bifunctional chimeric protein.

Due to the importance of long-lived plasma cells (LLPC) in maintaining protective antibody levels for years (Radtke et al., 2017), LLPC induced by vaccination with cPvMSP1-Pvs25 were assessed. We were able to confirm high levels of bone marrow and spleen LLPCs at 2 years post-immunization. We observed that immunization with cPvMSP1-Pvs25 was more efficient at generating long-lived Plasmodium-specific IgG plasma cells when compared to Pvs25, with significantly more Pvs25-specific

 et

LLPCs also observed in the bone marrow. This is in contrast with the poor immunogenicity reported for the native P25 (Tomas et al., 2001; Radtke et al., 2017). Chemical conjugation to carrier proteins, a strategy used for glycoconjugate vaccines to enhance the immunogenicity of bacterial polysaccharides (Rappuoli, 2018), has been tested to improve the immune responses induced by post-fertilization antigens. Clinical trials of the conjugated Pfs25-exoprotein A (EPA) vaccine showed excellent safety profile, but the antibody responses induced by immunization were short lasting with poor responses elicited in volunteers living in endemic areas compared to non-endemic areas (Talaat et al., 2016; Sagara et al., 2018). More recently, chemical conjugation of Pfs25 to tetanus toxoid (TT), a carrier protein used in existing polysaccharide vaccines, resulted in a significant improvement in immunogenicity and longevity (Radtke et al., 2017). TT contains CD4+ T cell epitopes that enhance humoral immunity compensating for the lack of CD4 T cells epitopes within Pfs25 resulting in the generation of T follicular helper (Tfh) cells (Radtke et al., 2017). Tfh cells are required for germinal center B cell differentiation into LLPC essential for long-lasting protection. Similar findings were subsequently reported by Parzych et al. with the P. falciparum merozoite surface protein 8 (MSP8) tested as a carrier to deliver Pfs25 (Parzych et al., 2017). These reports highlighted the relevance of T cell epitopes present in the carrier proteins to promote robust immune responses. The likely mechanism of action of the cPvMSP1 within cPvMSP1-Pvs25 is to serve as a carrier protein to the recruitment of Plasmodium-specific CD4+ T cells, which ultimately promote a robust humoral immune response to the hybrid vaccine and induction of the anti-Pvs25 LLPC response.

Avidity of the antibodies induced by vaccination with either Pvs25 or cPvMSP1-Pvs25 was also assessed as a measure of immunogenicity. At 20 days after the final immunization, anti-Pvs25 antibody avidity indices were 0.95 for Pvs25 and 0.79 for cPvMSP1-Pvs25 immunization groups. By 2 years postimmunization, the anti-Pvs25 avidity index of the Pvs25 group had fallen to 0.50, significantly lower than the avidity index of 0.68 at 2 years induced by immunization with cPvMSP1-Pvs25. Consistent with the maintenance of antibody avidity against Pvs25 by the cPvMSP1-Pvs25 immunization regimen, the avidity indices of the anti-cPvMSP1 response were not significantly different at 60 days than at 2 years post-immunization. The high avidity antibodies elicited by cPvMSP1-Pvs25 are, therefore, encouraging given their relevance as a biomarker of efficacy of TBVs beyond the antibody titers (Patra et al., 2015). We can also conclude that this outcome is likely an effect of the improved LLPC response induced by the cPvMSP1-Pvs25 vaccination regimen, allowing for the maintenance of higher avidity antibodies over time.

Cytophilic IgG antibody subclass responses directed against MSP119, as well as other merozoite surface antigens, have previously been found to be associated with control of parasitemia and protection from symptomatic illness in children in P. falciparum endemic areas (Stanisic et al., 2009). The cytophilic antibody subclasses correspond to IgG1 and IgG3 in humans and IgG2a in mice. We observed a shift in the IgG2a/IgG1 ratio of the anti-cPvMSP1 antibody response between day 60 and 2 years post-prime, with a significant reduction in the IgG2a subclass response. In contrast, there was a significantly higher IgG2a/IgG1 subclass ratio for anti-Pvs25 antibodies in cPvMSP1-Pvs25 immunized mice compared to Pvs25 immunized mice at the 2-year time point. Although we observed higher levels of IgG1 than IgG2a in the anticPvMSP1 response, the significantly higher IgG2a/IgG1 ratio observed for the anti-Pvs25 response elicited by cPvMSP1-Pvs25 vaccination is promising and may have biological significance due to differences in the ability of IgG subclasses to activate the classical complement pathway. Parasites present in the mosquito midgut after the female Anopheles mosquito ingests infected red blood cells are exposed to multiple plasma elements present in the mosquito blood meal including complement, granulocytes, and the host-derived antibodies (Sauerwein, 2007; Saul, 2007). These components can all affect parasite development within the mosquito and ultimately reduce transmission. Studies have shown that human sera with anti-Pvs25 antibodies have reduced killing activity after heat inactivation indicating that complement may be necessary to block transmission with Pvs25 vaccine-induced antibodies (Malkin et al., 2005). The classical complement pathway involves the binding of complement molecule C1q to the Fc region of the antigen bound antibody. However, IgG subclasses differ in their ability to activate complement, as human cytophilic antibodies IgG3 and IgG1 have been found to more effectively activate the classical complement pathway than the IgG2 subclass. This study suggests that cPvMSP1-Pvs25 immunization is more effective at inducing cytophilic IgG2a subclass antibodies in mice.

The ability of antibodies induced by individual components of the bifunctional chimeric protein in this study to bind the native structures expressed by young oocysts and schizonts is essential for their functional activities. We confirmed by IFAs that antibodies induced by immunization with cPvMSP1-Pvs25 recognized the native structure on the surface of transgenic oocysts expressing Pvs25 as well as the native structure of PvMSP1 in P. vivax blood stage schizonts. We have previously shown that passive immunization using antibodies elicited by the orthologous P. yoelii chimeric MSP1 protein, based on sequences from the P. yoelii 17X strain, protect naïve mice against heterologous challenge with the P. yoelii nigeriensis N67 isolates (Singh et al., 2010). Although we were unable to repeat similar passive transfer experiments in this study, future studies using transgenic parasites expressing P. vivax MSP1 are warranted. Additionally, the binding capability of anti-cPvMSP1- Pvs25 antibodies and the robust transmission blocking activity elicited by immunization as determined by direct membrane feeding assays, support further studies on the functionality of the anti-Pvs25 and anti-cPvMSP1 antibodies induced by vaccination with cPvMSP1-Pvs25. Furthermore, while no differences in the IFN-γ response were observed between the immunization groups following stimulation with the Pvs25 peptide pools, as expected, stimulation with cPvMSP1 peptide pools revealed that cPvMSP1- Pvs25 induced high levels of IFN-γ CD4<sup>+</sup> and CD8<sup>+</sup> T cells as compared to Pvs25-immunized mice. The ability of cPvMSP1- Pvs25 to maintain robust IFN-γ responses to cPvMSP1 when conjugated to the Pvs25 protein is encouraging for the blood stage component of this vaccine candidate, as several reports have indicated that high frequencies of IFN-γ secreting CD4<sup>+</sup> T cells provide protection from malaria in humans (Roestenberg et al., 2009; White et al., 2013; King and Lamb, 2015).

Direct membrane feeding assays allow the assessment of transmission-blocking activity to determine the functionality of antibodies elicited by vaccination (Bompard et al., 2017). Currently, transmission-blocking assays are considered the most epidemiologically important method of evaluating transmission blocking vaccine candidates (Bompard et al., 2017). We found that sera obtained from both Pvs25 and cPvMSP1- Pvs25-immunized rabbits had significantly high transmission blocking activity, as shown by the lower percentages of infected mosquitoes and the number of oocysts compared with pre-immune sera. Unlike P. falciparum membrane feeding experiments, which can be done from parasite cultures, these results are promising as they came from wild P. vivax isolates, allowing for better prediction of the transmission blocking potential these antibodies might achieve if tested in clinical trials in endemic areas. These results support further development of cPvMSP1-Pvs25 as an effective transmission-blocking vaccine.

The development of licensed vaccines that reduce malaria transmission of P. falciparum and P. vivax is one of two main strategic goals of the 2030 Strategic Goals of Malaria Vaccine Technology published by the WHO; the other is a malaria vaccine with at least 75% efficacy against clinical malaria (Moorthy et al., 2013). While there is controversy concerning the use of transmission-blocking vaccines, including the fact that transmission-blocking antigens alone will only reduce transmission but will not protect vaccinated populations from disease (Jamrozik et al., 2015), the combination of transmission blocking antigens with antigens targeting the clinical stages of malaria could be used to prevent both disease and transmission. Studies evaluating the impact of vector control measures have demonstrated that the reduction of transmission in medium to high transmission areas induces a decrease in all-cause mortality, with the youngest age groups of the population benefiting the most (Smith et al., 2001; Rowe and Steketee, 2007; Larsen et al., 2011). We report here that the addition of an anti-erythrocytic stage antigen in the form of the chimeric PvMSP1 protein genetically conjugated to Pvs25 improves the immunogenicity of the TBV candidate while preserving the functionality of Pvs25 induced antibodies. Our results provide support for the continued development of cPvMSP1-Pvs25 and other multi-stage malaria vaccine candidates to address the need for effective vaccines targeting both the parasite stage responsible for the clinical manifestations and simultaneously the sexual stage responsible of transmission.

#### REFERENCES

Arevalo-Herrera, M., Solarte, Y., Zamora, F., Mendez, F., Yasnot, M. F., Rocha, L., et al. (2005). Plasmodium vivax: transmissionblocking immunity in a malaria-endemic area of Colombia. Am. J. Trop. Med. Hyg. 73, 38–43. doi: 10.4269/ajtmh.2005.73.5\_suppl. 0730038

### AUTHOR CONTRIBUTIONS

AM conceived the study. JF, BS, JJ, MA-H, and AM designed the experiments. JM, JF, BS, MA-H, MC-M, and CB performed the experiments. JM, JF, and AM performed data analysis. JM, JF, and AM wrote the manuscript. All the authors approved the final version of the manuscript.

### FUNDING

This research was supported by the National Institutes of Health, NIAID grants R56-AI103382-01A1, R21 AI094402-01A1, and R21AI135711-01. This project was funded in part by ORIP/OD P51OD011132.

#### ACKNOWLEDGMENTS

The following reagents were obtained through BEI Resources Repository, NIAID, NIH: Plasmodium berghei, Strain pv25DR (15cy1Pb25-/PD28-), MRA-904, contributed by Robert Sinden, and Hybridoma N1-1H10 anti-Plasmodium vivax Ookinete Surface Protein 25 (Pvs25), MRA-471 contributed by Carole A. Long. The authors would like to thank Dr. Oskar Laur for the cloning of the hybrid cPvMSP1-Pvs25 protein gene, Kelly Rubiano and Andres Amado from Caucaseco Scientific Research Center at Cali, Colombia for supervision of P. vivax transmission blocking assays and Dr. Mary Galinski for providing our group with P. vivax infected Saimiri boliviensis blood aliquot for IFA slides.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2019.00135/full#supplementary-material

Supplementary Figure 1 | A full scan of the entire original Western blot analysis of the purified cPvMSP1-Pvs25 (left) and the purified Pvs25 proteins (right). Samples were incubated with the specified antibodies: monoclonal antibody N1-1H10 which targets Pvs25; anti-His-Tag monoclonal antibody targeting the C terminal tags of the cPvMSP1-Pvs25 and Pvs25 proteins; or the monoclonal antibody 2A10 which targets the cPvMSP1 C terminal tag. The molecular weight markers (BioRad) are included.

Supplementary Figure 2 | Gating strategy for CD4<sup>+</sup> and CD8<sup>+</sup> IFN-γ-producing murine T cells. The gating strategy for identifying IFN-γ CD4<sup>+</sup> and CD8<sup>+</sup> T cells from murine splenocytes isolated 5 days post boosting includes: FSC singlets, viability gating, CD3<sup>+</sup> cells, CD4<sup>+</sup> vs. CD8<sup>+</sup> T cells, and IFNγ <sup>+</sup> cells. Data shown represent IFN-γ gating in CD4<sup>+</sup> T cells of mice receiving Pvs25 or cPvMSP1-Pvs25.


Pvs25 and Pvs28 do not show synergism. Vaccine 20, 763–770. doi: 10.1016/S0264-410X(01)00402-9


**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 McCaffery, Fonseca, Singh, Cabrera-Mora, Bohannon, Jacob, Arévalo-Herrera and Moreno. 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.

# Chimeric Murine Polyomavirus Virus-Like Particles Induce Plasmodium Antigen-Specific CD8<sup>+</sup> T Cell and Antibody Responses

David J. Pattinson1,2 \*, Simon H. Apte<sup>1</sup> , Nani Wibowo<sup>3</sup> , Yap P. Chuan<sup>3</sup> , Tania Rivera-Hernandez <sup>3</sup> , Penny L. Groves <sup>1</sup> , Linda H. Lua<sup>4</sup> , Anton P. J. Middelberg<sup>3</sup> and Denise L. Doolan1,2 \*

1 Infectious Diseases Programme, QIMR Berghofer Medical Research Institute, Brisbane, QLD, Australia, <sup>2</sup> Centre for Molecular Therapeutics, Australian Institute of Tropical Health and Medicine, James Cook University, Cairns, QLD, Australia, <sup>3</sup> Australian Institute for Bioengineering and Nanotechnology, University of Queensland, Brisbane, QLD, Australia, <sup>4</sup> Protein Expression Facility, University of Queensland, Brisbane, QLD, Australia

#### Edited by:

Alberto Moreno, Emory University School of Medicine, United States

#### Reviewed by:

Giampietro Corradin, Université de Lausanne, Switzerland Kai Huang, University of Texas Medical Branch at Galveston, United States Sean C. Murphy, University of Washington, United States

#### \*Correspondence:

David J. Pattinson david.pattinson@jcu.edu.au Denise L. Doolan denise.doolan@jcu.edu.au

#### Specialty section:

This article was submitted to Clinical Microbiology, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 20 February 2019 Accepted: 03 June 2019 Published: 19 June 2019

#### Citation:

Pattinson DJ, Apte SH, Wibowo N, Chuan YP, Rivera-Hernandez T, Groves PL, Lua LH, Middelberg APJ and Doolan DL (2019) Chimeric Murine Polyomavirus Virus-Like Particles Induce Plasmodium Antigen-Specific CD8<sup>+</sup> T Cell and Antibody Responses. Front. Cell. Infect. Microbiol. 9:215. doi: 10.3389/fcimb.2019.00215 An effective vaccine against the Plasmodium parasite is likely to require the induction of robust antibody and T cell responses. Chimeric virus-like particles are an effective vaccine platform for induction of antibody responses, but their capacity to induce robust cellular responses and cell-mediated protection against pathogen challenge has not been established. To evaluate this, we produced chimeric constructs using the murine polyomavirus structural protein with surface-exposed CD8<sup>+</sup> or CD4<sup>+</sup> T cell or B cell repeat epitopes derived from the Plasmodium yoelii circumsporozoite protein, and assessed immunogenicity and protective capacity in a murine model. Robust CD8<sup>+</sup> T cell responses were induced by immunization with the chimeric CD8<sup>+</sup> T cell epitope virus-like particles, however CD4<sup>+</sup> T cell responses were very low. The B cell chimeric construct induced robust antibody responses but there was no apparent synergy when T cell and B cell constructs were administered as a pool. A heterologous prime/boost regimen using plasmid DNA priming followed by a VLP boost was more effective than homologous VLP immunization for cellular immunity and protection. These data show that chimeric murine polyomavirus virus-like particles are a good platform for induction of CD8<sup>+</sup> T cell responses as well as antibody responses.

Keywords: malaria, vaccine, circumsporozoite protein, virus-like particle, murine polyomavirus, cellular immunity, T cell responses, Plasmodium yoelii

## INTRODUCTION

Although annual malaria related mortality rates have decreased by approximately 25% since 2010, improvements in morbidity and mortality have stabilized in recent years (World Malaria Report, 2017) and rebounds have been reported in some countries (Alonso et al., 2011). This is despite an annual financial investment of approximately US\$2.7 billion, which includes malarial preventative measures such as residual spraying, insecticide-treated mosquito nets and preventative therapies (World Malaria Report, 2017). An effective vaccine is considered by many to be an essential tool for malaria disease control and eradication (Alonso et al., 2011). The ideal vaccine would target the pre-erythrocytic stage to induce sterile infectionblocking immunity which either prevents sporozoite invasion of the hepatocyte and/or halts the development of the parasite at the liver-stage; this would stop the development of clinical symptoms of malaria which manifest during the bloodstage, as well as the transmission of malaria which occurs during the sexual stage. A partially effective vaccine which reduced liverstage parasite burden and thus lowered blood-stage parasitemia to potentially below a threshold associated with clinical disease would, however, also be a useful tool.

Recently, GlaxoSmithKline achieved a significant milestone with MosquirixTM (also known as RTS,S) by receiving a positive scientific opinion from the European Medicines Agency as the first malaria vaccine for the immunization of children aged 6 weeks to 17 months. This sub-unit virus-like particle (VLP) based vaccine targets the P. falciparum circumsporozoite protein (PfCSP) by combining a protein which includes multiple PfCSP B cell repeats and T cell epitopes fused with recombinant hepatitis B surface antigen (RTS), and recombinant wild-type hepatitis B surface antigen (S) (Gordon et al., 1995). These recombinant proteins combine to form stable VLPs, which are co-administered with the AS01 adjuvant (Agnandji et al., 2012). Disappointingly, clinical studies have shown that although RTS,S provided some protection in the first year after vaccination, this efficacy was very low and waned quickly (Tinto et al., 2015; Olotu et al., 2016) with negative efficacy and a rebound in later years demonstrated in a long-term follow-up study where RTS,S immunized 5–17 month children were more likely to be infected 5 years following vaccination (Olotu et al., 2016). These data emphasize the urgent need to identify new vaccine targets and vaccine delivery platforms that enhance immunogenicity and provide durable protection.

The CSP remains the most advanced subunit vaccine candidate, as a target for neutralizing antibodies prior to liver infection (Charoenvit et al., 1991; Mishra et al., 2012) and for T cell responses whilst in the hepatocyte (Grillot et al., 1990; Weiss et al., 1990, 1992; Franke et al., 1997, 2000). The induction of T cell or antibody mediated protection against sporozoite challenge has been demonstrated in murine and non-human primate models as well as humans by various vaccine platforms based on the CSP (Sedegah et al., 1994; Walsh et al., 2006; Mettens et al., 2008; Tewari et al., 2010; Tamminga et al., 2011; Noe et al., 2014; Janitzek et al., 2016; Collins et al., 2017; Yoshida et al., 2018). Thus, the CSP is widely considered the antigen of choice for evaluating novel vaccine delivery platforms targeting the pre-erythrocytic stage of the parasite life cycle.

Whilst antibody responses generated by chimeric VLP immunizations using various VLP platforms with antigens from various pathogens in clinical studies are well-documented (reviewed in Kushnir et al., 2012), little research has been done to evaluate their efficacy at generating cellular responses. Chimeric VLPs using hamster and simian virus 40 polyomavirus structural proteins have shown antigen-specific cytotoxic T lymphocyte (CTL) responses with viral epitopes (Mazeike et al., 2012; Kawano et al., 2014), and a murine polyomavirus VLP with large HER-21−<sup>683</sup> oncogene insert mouse study showed rejection and inhibition of antigen expressing tumor cells in vivo (Tegerstedt et al., 2005). Furthermore, both HPV16 L1 VLPs and their subunit component capsomeres are potent inducers of CTL responses capable of tumor regression without adjuvants (Ohlschläger et al., 2003).

In this study, we produced chimeric VLPs using the murine polyomavirus (MuPyV) (Salunke et al., 1986) by genomic insertion of either Plasmodium yoelii CSP CD8<sup>+</sup> (280−288) (Weiss et al., 1992) or CD4<sup>+</sup> (59−79) (Grillot et al., 1990) T cell epitopes, or the B cell repeat epitope(QGPGAPx2) into a surface-exposed region of the VP1 structural protein (Middelberg et al., 2011). We selected the MuPyV-VP1 platform which has been extensively developed by our group, and can be produced in bacterial expression systems which can be purified at gram per liter levels (Liew et al., 2010). The proteins form pentameric capsomeres (Salunke et al., 1986) which can be chemically induced in vitro to self-assemble into VLPs (Chuan et al., 2008a; Middelberg et al., 2011; Rivera-Hernandez et al., 2013). We comprehensively evaluated antibody and T cell responses as well as protection from Plasmodium yoelii 17XNL sporozoite challenge using chimeric PyCSP surface exposed on murine polyomavirus VLP constructs delivered separately or pooled.

Consistent with previous results with the murine polyomavirus VLPs (Anggraeni et al., 2013; Rivera-Hernandez et al., 2013; Wibowo et al., 2014; Seth et al., 2016; Tekewe et al., 2017), the Plasmodium B cell VLP was a strong inducer of antigen-specific antibody responses. Additionally, we showed that the VLP platform was capable of inducing robust VLPinduced CD8<sup>+</sup> T cell responses. However, the VLPs did not stimulate a good CD4<sup>+</sup> T cell response, and there was no apparent help provided by the CD4 VLP to the VLP-induced CD8<sup>+</sup> T cell responses nor to the induced antibody titers. The VLP-induced CD8<sup>+</sup> T cell responses were however enhanced using a heterologous DNA prime-boost regimen.

## MATERIALS AND METHODS

#### Plasmid Construction

The plasmid pGEX-4T-1 (GE Healthcare Biosciences, UK) containing the murine polyomavirus VP1 sequence (M34958) was provided by Professor Robert Garcea (University of Colorado, USA). The VP1 sequence was modified by inserting an AfeI restriction enzyme site flanked with Glycine4-Serine linker sequences at position 293 and designated pGEX-VP1- S4-G4S (Middelberg et al., 2011). This allows for antigens to be inserted into a surface-exposed region of the VP1 protein with the linker sequences added to reduce structural interference from the insert on VLP formation. Codon optimized sequences representing defined PyCSP CD8<sup>+</sup> (280−288) (Weiss et al., 1992) and CD4<sup>+</sup> (59−79) (Grillot et al., 1990) T cell epitopes and the B cell repeat epitope (QGPGAPx2) (Charoenvit et al., 1991) were separately inserted into the AfeI site in pGEX-VP1-S4- G4S using standard molecular biology techniques to generate chimeric CD8<sup>+</sup> or CD4<sup>+</sup> T cell or B cell VLP constructs. All constructs were confirmed by Sanger sequencing.

## Protein Expression, Purification, and VLP Assembly

Wild-type pGEX-VP1-S4-G4S, or the chimeric CD8+, CD4+, or B cell constructs were individually transformed into chemically competent E. coli Rosetta DE3 pLysS bacteria (Novagen, CA, USA). The GST-tagged VP1 proteins were expressed and purified as previously described (Chuan et al., 2008a; Middelberg et al., 2011). Briefly, bacteria inoculated Terrific Broth was incubated at 37◦C and 180 RPM until the OD<sup>600</sup> reached approximately 0.5, after which the culture was cooled to 26◦C and induced by adding 0.2 mM Isopropyl β-D-1-thiogalactopyranoside (IPTG) and incubated overnight at 26◦C and 180 RPM. The cultures were centrifuged and pellets lysed by sonication then filtered lysate was passed through a GSTrap HP affinity column (GE Healthcare, UK) and the purified protein was cleaved from the GST tag using thrombin (GE Healthcare, UK) and polished using a Superdex 200 10/300 GL column (GE Healthcare, UK). Endotoxin levels were reduced (<5 EU/ml) using Vivapure Q maxi H ion exchange columns (Sartorius Stedim, Germany) and confirmed with an Endosafe PTS reader (Charles River Laboratory, USA) as previously described (Middelberg et al., 2011). VP1 capsomeres were assembled into VLPs by dialysis against an assembly buffer then against PBS (Chuan et al., 2008a; Liew et al., 2010; Middelberg et al., 2011). The resultant VLPs were analyzed by asymmetric flow field-flow fractionation coupled to multi angle light scattering (AF4-MALS) using an Eclipse 2 AFFFF system coupled with a Dawn EOS MALS system (Wyatt Technology Corporation, Santa Barbara, USA), and transmission electron microscopy (TEM) using a JEOL 1010 (JEOL Ltd., Tokyo, Japan) to assess size distribution as previously described (Chuan et al., 2008b; Lipin et al., 2008).

## Immunization of Mice

All animal experiments were approved by the QIMR Berghofer Animal Ethics Committee and were conducted in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes (2004). BALB/c mice (n = 10/group) aged 7–8 weeks (Animal Resources Center, WA, Australia) were immunized with the chimeric VLP constructs three times at 3-week intervals by subcutaneous injection at the base of the tail. VLPs were administered at two doses, 10 or 30 µg in independent experiments, and individually or pooled. VLPs were administered without adjuvant as this platform is considered self-adjuvating (Chackerian et al., 2002; Stanley, 2006). Synthetic peptides corresponding to the epitopes presented by the chimeric VLP co-administered with 50 µg of high molecular weight poly (I:C) adjuvant (Invivogen, USA) were included as comparator groups in each experiment. In heterologous prime-boost regimens, mice received two doses of 100 µg of PyCSP plasmid DNA (pVR2516, Vical Inc, CA, USA) at 3-week intervals by intramuscular injection into the tibialis anterior muscle, followed by a booster dose of pooled VLP or synthetic peptide in poly(I:C) adjuvant as previously described. In the second experiment, one group received three homologous immunizations with PyCSP plasmid DNA at 100 µg per dose. Control groups included wild-type VLPs, ovalbumin CD8<sup>+</sup> <sup>257</sup>−<sup>264</sup> and CD4<sup>+</sup> <sup>323</sup>−<sup>339</sup> pooled peptides with 50 µg poly(I:C), and PBS only (naïve). Immunized mice were split into two groups (n = 5/group) to evaluate either immunogenicity using splenocytes for T cell assays or protective efficacy by sporozoite challenge (**Figure 1**).

#### Sporozoite Challenge and Protection

Two weeks after the final immunization, mice were challenged with 1,000 cryopreserved P. yoelii 17 XNL sporozoites (Sanaria Inc., MD, USA) in 200 µl PBS with 2% naive mouse serum administered intravenously into the tail vein. Livers were harvested 42 h post-challenge to assess liver-stage parasite burden as previously described (Schussek et al., 2013). Briefly, livers were homogenized in 5 ml of RLT buffer (Qiagen, Netherlands) supplemented with 1% β-2-mercaptoethanol (Sigma-Aldrich), then RNA was extracted using an RNeasy Mini kit (Qiagen) following the manufacturer's protocol. Using 2.5 µg of RNA, cDNA was synthesized using a Super Script VILO reverse-transcriptase cDNA synthesis kit (Life Technologies, USA) according to manufacturer's guidelines. Then the P. yoelii 18S rRNA and murine housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) cDNA were quantified using quantitative reverse transcriptase-PCR (qRT-PCR) using a Rotor-gene 3000 (Corbett Research, Mortlake, Australia) with acquisition on the FAM channel. For Py18S cDNA quantification, a Taqman Fast Advanced master mix (Applied Biosystems, Australia) with a custom made Taqman probe [250 nm] (6FAM-CTGGCCCTTTGAGAGCCCACTGATT-BHQ-1) and primers [1 um] (5′′- CTTGGCTCCGCCTCGATAT and 3′′ - TCAAAGTAACGAGAGCCCAATG) (Applied Biosystems). A GAPDH kit (Applied Biosystems) combined with platinum Taq polymerase, PCR buffers (Invitrogen) was used for GAPDH quantification. The data are presented as a ratio of copies of Py18S rRNA per 10<sup>5</sup> copies of GAPDH.

#### Splenocyte Harvesting and Stimulation for T Cell Assays

Ten days after the final immunization, spleens were removed and single cell suspensions were generated by mechanical disruption followed by red blood cell lysis. Splenocytes were then coincubated with gamma irradiated (16,666 cGy) mouse B cell lymphoma A20 cells (ATCC TIB-208) which were transfected, peptide stimulated, or untreated. Transfections with PyCSP encoded plasmid DNA pVR2516 (Vical, USA) or empty vector pVR1020 (Vical, USA) plasmid DNA was achieved the AMAXA Nucleofector system (Lonza, Switzerland) using Kit V and program C-25 with 5 x 10<sup>6</sup> A20 cells per cuvette, following the manufacturer's protocol. Peptide stimulation used either PyCSP CD8<sup>+</sup> (280−288) or CD4<sup>+</sup> (59−79) T cell epitope synthesized peptides, or these peptides combined with the B cell repeat epitope (QGPGAPQGPGAP) peptide at 10µg/ml. For this assay, cells were incubated in KD-MEM media comprised of Dulbecco's Modified Eagle's Medium (SAFC Global, USA) supplemented with folic acid (136 nM), L-asparagine (32 mM), L-arginine (67 mM), sodium bicarbonate (24 mM), HEPES (10 mM), β-2 mercaptoethanol (5 nM), L-glutamine (1.5 mM), penicillin (100 Units/L), streptomycin (100 mg/L) and 10% fetal calf serum. Each well-contained 5 x 10<sup>5</sup> splenocytes and 1.5 × 10<sup>5</sup> A20 cells in 200 µl of media. Background responses were removed by subtracting transfected pVR1020 wells from pVR2516 wells, and untreated A20 wells from peptide-pulse wells for ELISpot, cytometric bead array (CBA) and intracellular cytokine staining (ICS) assays.

## IFN-γ ELISpot Assay

IFN-γ ELISpot assays were conducted as previously described (Schussek et al., 2017). Briefly, MSIPS4510 multiscreen ELISpot plate (Merck Millipore, Germany) well were pre-coated with 100µg/ml anti-mouse IFN-γ (BD Biosciences, USA) then blocked with KD-MEM supplemented with 10% FCS. Quadruplicate wells were used for each stimulation. Plates were incubated for 40 h at 37◦C and 5% CO2. After plate washing, IFN-γ secreting cells were stained with 2µg/ml biotinylated anti-mouse IFN-γ (BD Biosciences, USA) followed by 1µg/ml streptavidin-HRP (BD Biosciences, USA). The assay was developed using AEC substrate (BD Biosciences, USA), and spots counted using the AID ELISpot reader system (Autoimmun Diagnostika GmbH, Germany).

#### Cytometric Bead Array

Splenocyte/A20 cultures were incubated at 37◦C and 5% CO<sup>2</sup> for 72 h. Culture supernatant was collected and secreted IFNγ, TNF, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12p70, and IL-13 cytokines assayed using the mouse cytometric bead array flex kit (BD Biosciences, USA) following the manufacturer's protocol. Samples were analyzed using a FACSArray instrument (BD Biosciences, USA) using the CBA array software (BD Biosciences, USA).

## Intracellular Cytokine Staining

To detect monofunctional or polyfunctional T cell responses, splenocyte/A20 cultures were incubated with 0.1% Golgi Plug (BD Biosciences, USA) for 6 h at 37◦C and 5% CO2. Cells were stained with anti-CD8<sup>+</sup> (53-6.7) and anti-CD4<sup>+</sup> (RM4.5) antibodies before being fixed with 1% paraformaldehyde. Cells were then stained with anti-IFN-γ (XMG1.2), anti-IL-2 (JES6- 5H4), and anti-TNF-α (MP6-XT22) diluted in permwash buffer (BD Biosciences, USA). All antibodies were purchased from Biolegend. Flow cytometric analysis was performed on a Fortessa 4 (BD Biosciences, USA). Post-acquisition data analysis was performed using FlowJo software version 10 (Treestar, USA).

## ELISA

Antigen-specific antibody responses were detected by ELISA using the PyCSP B cell repeat peptide linked to a polystyrene binding tag (Kumada et al., 2010) with a glycine<sup>4</sup> spacer (Kogot et al., 2012) (PST-B cell, RIIIRRIRGGGG-QGPGAPx3) (Mimotopes, Australia). The PST-tag with a glycine spacer was incorporated to enhance attachment of the peptide to the plate and expose the B cell repeat for antibody recognition. Nunc maxisorp plates (Thermo Fisher Scientific, USA) were coated overnight with the PST-B cell peptide (5µg/ml) in a carbonate coating buffer, then blocked with PBS containing 2% BSA. Triplicate wells of sera from mice were titrated by 2-fold dilutions in PBS-BSA 0.1%, and then incubated with biotinylated donkey α-mouse IgG (Jackson ImmunoResearch Laboratories, USA) diluted in PBS-BSA 0.1%, followed by an incubation with streptavidin-HRP (BD Biosciences, USA) diluted 1:1,000 in PBS-BSA 0.1 and 0.2% Tween20. Plates were developed with tetramethylbenzidine (TMB) and stopped using TMB stop reagent (Sigma Aldrich, USA). Absorbance was measured at 450 nm using a VersaMax microplate reader (Molecular Devices, USA). Positive results were recorded if OD<sup>450</sup> values were > 3 standard deviations above the mean blank (no serum) values.

#### Statistical Analysis

Statistical analysis was performed using GraphPad Prism version 6.0 (GraphPad, CA, USA). Logarithmic transformed data of groups were compared by one-way analysis of variance (ANOVA) and Bonferroni's multiple comparison test. Statistical significance is displayed as <sup>∗</sup>p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

#### RESULTS

#### VLP Construction

In-frame genomic insertion of antigen-derived epitopes into the MuPyV-VP1 protein was confirmed by sequencing. After the VLPs were assembled in vitro, analysis using AF4-MALS showed only small amounts of protein aggregation and TEM images confirmed that the chimeras had a similar morphology to wildtype VLPs (**Figure 2**). The mean radius of each of the chimeric VLPs was similar to that of the wild type VLPs [Wild type- 21.00 ± 1.27 nm; CD8 VLPs- 20.51 ± 0.67 nm; CD4 VLPs- 21.07 ± 0.61 nm; B cell VLPs- 20.85 ± 0.67 nm (mean radius ± SD)].

#### VLP-Induced T Cell Immune Responses

Mice were immunized with chimeric CD8 or CD4 VLPs to assess their capacity independently to induce CD8<sup>+</sup> or CD4<sup>+</sup> T cell responses, and to evaluate if pooling of the VLPs would result in an additive or synergistic response. Pooled VLPs were evaluated in independent experiments at 10 µg per dose and at 30 µg per dose to dose-match the independent VLPs (**Figure 1**).

IFN-γ responses detected by ELISpot showed that homologous immunization with the CD8 VLP induced significant IFN-γ spot forming cells (SFCs) when stimulated with pooled peptide or PyCSP transfected A20 cells (p < 0.001) (**Figures 3A**, **4A**). The magnitude of VLP-induced CD8<sup>+</sup> T cell response was similar to that induced by CD8 peptide immunization when stimulated by peptide (but not transfected A20 cells). In contrast, the CD4 VLP failed to induce a significant IFN-γ response (**Figures 3A**, **4A**); robust responses against peptide (p < 0.0001) and transfected stimulation (p < 0.001) were observed in mice immunized in parallel with CD4 peptides in adjuvant (**Figure 3A**) and similarly in the subsequent experiment (**Figure 4A**).

Immunization with pooled CD8 plus CD4 VLPs resulted in a similar IFN-γ profile to that induced by the individual VLPs, with a robust response to the CD8 peptide and negligible response to the CD4 peptide (**Figure 4A**).

Heterologous prime-boost immunization with DNA prime/VLP boost induced a stronger response than homologous VLP immunization (**Figures 3A–C**). This was attributed to the plasmid DNA prime with the homologous DNA regimen inducing the most robust response (**Figures 4A–C**). The reduction in responses between the DNA only and DNA prime/VLP boost could be due to an inhibition caused by VLPs, but more likely that the third dose of DNA was more immunogenic than the VLPs.

The profile of IFN-γ responses detected with both CBA and ICS readouts was similar to that shown by ELIspot for the CD8 and CD4 VLPs (**Figures 3A–C**, **4A–C**). Robust IFN-γ responses

FIGURE 2 | Structural analysis of in vitro assembled chimeric murine polyomavirus virus-like particles. Wild and chimeric MuPyVP1-G4S proteins were expressed in E. coli and purified by liquid chromatography. Proteins were dialyzed against an assembly buffer to form VLPs and then against PBS. Post-assembly solutions were analyzed to detect the formation of VLPs using asymmetrical flow field- flow fractionation coupled with multi-angled lights scattering (AF4-MALS) and transmission electron microscopy of (A) wild-type VLPs, (B) CD8 VLP chimeras, (C) CD4 VLP chimeras, and (D) B cell VLP chimeras. UV280 absorbance is displayed relative to peak absorbance (solid line) for each sample, and particle size is shown as root-square-radius (open circles). Scale bar is 100 nm.

induced by the CD8 VLP were similar to that induced by peptide in adjuvant. Negligible CD4<sup>+</sup> T cell responses were induced by CD4 VLP immunizations, as opposed to the robust CD4<sup>+</sup>

FIGURE 3 | Cellular immune responses induced by individual (30 µg) or pooled (10 µg each) chimeric VLP immunizations. BALB/c mice (n = 5/group) were immunized with a prime/boost regimen using two i.m., priming immunizations with PyCSP plasmid DNA (100 µg) followed by a single s.c. boost using either pooled PyCSP VLPs (10 µg each) or pooled PyCSP peptides (30 µg each) with poly(I:C); or with three homologous s.c doses of pooled PyCSP VLPs (10 ug each) or individual PyCSP VLPs (30 ug); or pooled or individual PyCSP peptides (30 µg each) with poly(I:C); wild type VLPs (30 µg); pooled ovalbumin peptides (30 µg each) with poly(I:C); or PBS only. Ten days after the final boost, splenocytes (5 × 10<sup>5</sup> /well) were cultured with irradiated A20 cells pulsed with pooled PyCSP CD8280−288, CD459−79, or B cell peptides (1.5 <sup>×</sup> <sup>10</sup><sup>5</sup> /well), or with irradiated A20 cells transfected with pVR2516 PyCSP plasmid DNA (5 × 10<sup>4</sup> /well), for assessment of T cell specific immune responses. (A) IFN-γ ELISpot was used to determine IFN-γ spot forming cells/million splenocytes following 40 h incubation with pooled CD8280−288 plus CD459−<sup>79</sup> peptides (left) or DNA transfected A20 cells (right). (B) ICS was used to determine CD8<sup>+</sup> and CD4<sup>+</sup> T cell, IFN-γ, IL-2, and TNF cytokine responses following 18 h stimulation with pooled peptides, with the final 6 h supplemented with GolgiPlug. (C) Cytometric Bead Array (CBA) was used to quantify cytokine concentrations in 72 h culture supernatant following pooled peptide stimulation. All data are presented as group means + SEM (media only subtracted). Statistical comparisons are made to the PBS control group, and between homologous VLP and peptide immunization groups using log-transformed data with significance determined using one-way ANOVA followed by Bonferroni's post-hoc test. \*p < 0.05, \*\*p < 0.01, \*\*\*p < 0.001, and \*\*\*\*p < 0.0001.

T cell responses which were evident following CD4 peptide immunization. This differentiation is likely due to the reduced actual peptide load in the CD4 VLPs of approximately 0.52 µg.

TNF was induced in mice immunized with the CD4 peptide individually or in pools (**Figures 3C**, **4C**). TNF was also significantly induced by DNA immunizations in response to pooled peptide (p < 0.01) and CD4 peptide stimulation (p < 0.0001). No VLP immunizations resulted in significant TNF or IL-2 production from stimulated splenocytes. Our CBA assessed a range of cytokines specifically: IFN-γ, L-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12p70, IL-13 and TNF. However, with the exception of IFN-γ and TNF, we only detected a very low level of cytokines induced by our VLP immunizations (data not shown).

#### VLP-Induced Antibody Responses

The B cell VLP was predicted to induce strong antibody responses since it comprised the dominant PyCSP B cell linear epitope exposed on the surface of the VLP in a repetitive array. Responses to the B cell VLP were assessed in parallel individually as well as in a pool with the CD8 and CD4 VLPs since it was anticipated that the inclusion of the CD4 VLP would enhance the epitope-specific antibody titer.

Robust anti-PyCSP B cell peptide IgG responses were induced by all groups containing the B cell VLP (p < 0.0001 relative to naive) (**Figure 5**). In the homologous VLP immunizations, there were no significant differences between any groups indicating no additive effect by including the CD4 VLP. VLP-induced antibody responses were significantly higher than homologous DNA immunizations (p < 0.0001). In the DNA prime/VLP boost regimen, the response induced by the 10 µg/construct boost was significant higher as compared to the 30 µg/construct boost (p < 0.01); conversely in the homologous VLP only groups, the 30 µg per dose group higher than the 10 µg per dose, although not significantly different.

#### Protection From Sporozoite Challenge

To evaluate the protective efficacy of chimeric VLPs, VLPimmunized mice (n = 5/group) were challenged with 1,000 P. yoelii 17XNL sporozoites. In the first experiment, 40% (2/5) mice immunized with DNA prime / VLP boost had no parasite RNA detected in their liver and there was a 38% reduction in liver-stage parasite burden relative to the PBS control group overall in the group; these differences were not significant given the relatively small group sizes. When the experiment was repeated using the higher dose pooled VLPs (30 µg/construct), a significant reduction in parasite burden was noted in mice immunized with the heterologous DNA prime / VLP pooled VLP regimen (p < 0.01) (**Figure 6**). This equates to a mean 72% reduction in the liver parasite burden relative to the PBS immunized group. In the pooled VLP immunized mice, there was a mean 57% reduction in liver-stage parasite burden relative to the PBS control, although this did not reach statistical significance.

#### DISCUSSION

Development of an efficacious malaria vaccine has been challenging, with even the most advanced candidate MosquirixTM showing very poor efficacy despite very

extensive refinement during preclinical development and early stage clinical testing (Tinto et al., 2015; Olotu et al., 2016). MosquirixTM is a chimeric VLP construct expressing the central repeat region plus the entire C-terminal flanking region (amino acids 207–395) of the P. falciparum CSP fused to the hepatitis B virus surface antigen, HBsAg. Protection has been associated with antibodies or CD4<sup>+</sup> T cell responses, but this vaccine fails to induce CD8<sup>+</sup> T cell responses which are thought to be required for protection against intracellular pathogens such as the Plasmodium spp. parasite (Seder and Hill, 2000). Various Plasmodium targets have been evaluated using a range of subunit vaccine delivery platforms including recombinant protein, recombinant viruses, synthetic peptides, plasmid DNA, nanoparticles, and VLPs [reviewed in Draper et al. (2015)]. However, none of these platforms have induced robust CD8<sup>+</sup> and CD4<sup>+</sup> T cell responses as well as antibody responses sufficient for protection against malaria. In other disease models, VLP vaccines show potential for induction of broad antibody and cellular immune responses.

Protection directed against the pre-erythrocytic stage of the Plasmodium lifecycle following sporozoite challenge can be achieved primarily by antibody responses which inhibit sporozoite infection of the hepatocyte (Majarian et al., 1984; Charoenvit et al., 1991) or CD8<sup>+</sup> T cell responses which act directly or indirectly upon the liver-stage parasite (Weiss et al., 1988, 1990, 1992; Doolan and Hoffman, 2000); moreover, there is evidence that induction of CD4<sup>+</sup> T cell responses is required for optimal efficacy (Weiss et al., 1993). There is increasing interest in immunization regimens which induce liver-resident memory T cells as the parasite-infected liver is considered the primary site

\*\*\*\*p < 0.0001 (nt, not tested).

for protective T cell mediated immune responses which act to reduce liver-stage parasite burden (Fernandez-Ruiz et al., 2016; Gola et al., 2018; Olsen et al., 2018). It is possible that immune responses measured in the periphery or in splenocytes may not accurately reflect the vaccine-induced responses in the liver, so future studies should consider evaluation of liver-resident memory T cells in addition to the more traditional readouts from splenocytes.

Many VLP vaccine platforms have been developed, using structural proteins from various viral species. VLPs based on the murine polyomavirus are considered a promising VLP for vaccine development because of their reported humoral immunogenicity and their ability to be produced reproducibly with high yields (Chuan et al., 2008a; Liew et al., 2010, 2012; Middelberg et al., 2011; Anggraeni et al., 2013; Rivera-Hernandez et al., 2013; Wibowo et al., 2014; Seth et al., 2016; Tekewe et al., 2017). However, their capacity to induce robust antibody plus CD4<sup>+</sup> and CD8<sup>+</sup> T cell responses has not been thoroughly investigated. Accordingly, we evaluated both cellular and antibody responses using chimeric murine polyomavirus VLPs incorporating the CSP as a vaccine delivery platform, in the P. yoelii murine model.

It is well-established that VLPs are very effective at inducing antigen-specific antibody responses [reviewed in Grgacic and Anderson (2006)] likely due to their ability cross-link B cell surface receptors (Thönes et al., 2008) acting in a T cell

independent manner (Snapper and Mond, 1993; Szomolanyi-Tsuda and Welsh, 1998). Consistent with that, herein we showed that our B cell VLP induced high IgG titres against the PyCSP B cell epitope. There was a significant decrease in antibody titers in the DNA prime/VLP boost group when the pooled VLP component was increased from 10 to 30 µg per VLP (**Figure 5**). This variation was not seen in the VLP only immunized mice, but those mice received three VLP doses and the additional doses may have normalized the responses.

We established that immunizations with either chimeric CD8 VLPs alone, or as pooled with a B cell VLP and CD4 T cell VLP were able to induce moderate levels of IFN-γ directed against the CD8<sup>+</sup> T cell epitope presented as a synthetic peptide or in the context of whole antigen transfectant, with both ELISpot and ICS (**Figures 3A,C**, **4A,C**). This antigen-specific recognition confirms that the CD8 chimeric VLP presented the incorporated CD8<sup>+</sup> T cell epitope in vivo to naive CD8<sup>+</sup> T cells, and that this presentation was via MHC class I molecules (Heath et al., 2004). These results are consistent with previous studies reporting that chimeric VLPs incorporating a single CD8<sup>+</sup> T cell epitope from viral or ovalbumin targets induced CTL activity mediated by CD8<sup>+</sup> T cells (Sedlik et al., 1997; Crisci et al., 2009). It is possible that our immunizations with pooled VLPs using 30 µg of each construct failed to protect mice from challenge as the induced CD8<sup>+</sup> T cell response was insufficient to protect (Schmidt et al., 2008) with a mean frequency of 0.59% of CD8<sup>+</sup> T cells expressing IFN-γ (**Figure 4B**).

We had anticipated that immunizing with a pool of VLPs which included the CD4 T cell peptide might result in an additive or synergistic response, as noted for immunization with a pool of the synthetic peptides. However, there was no evidence of any additive or synergistic effects on CD8<sup>+</sup> T cell responses when VLPs were administered as a pool with the CD4 VLP; this was likely due to the very low responses induced by the CD4 VLP. The carrier MuPyV VP1 protein is likely to contain CD4<sup>+</sup> T helper epitopes similar to those in the HBV core protein (Schödel et al., 1992; Bickert et al., 2007) and may provide carrier-related T cell help for antibody responses, but those would provide no antigen-specific help against challenge by the Plasmodium parasite. Poly (I:C) was selected to adjuvant the synthetic peptides in vivo because it is a known inducer of type I (Salem et al., 2005; Qiu and Cui, 2007; Coffman et al., 2010; Schneider-Ohrum et al., 2011) and II (Tewari et al., 2010; Kastenmüller et al., 2013) cellular responses. However, VLPs are generally considered self-adjuvating due to their viral structure (Stanley et al., 2006; Chackerian, 2007). There are however contradictory reports, with some reports showing that adjuvants are essential (Storni et al., 2002), beneficial (Freyschmidt et al., 2004; Ding et al., 2009) or not necessary (Mazeike et al., 2012; Rivera-Hernandez et al., 2013; Kawano et al., 2014). In our studies, we elected to exclude adjuvants so as to better understand the immunogenicity of our chimeric VLPs. We cannot however rule out the possibility that administration of the CD4 VLP with poly(I:C) could have enhanced the VLP-induced T cell response, but other data generated in our laboratory showed no enhancement of CD4<sup>+</sup> T cell when pooled VLPs were co-administered with poly(I:C) (manuscript in preparation).

In our VLP constructs, each T cell epitope was inserted into surface-exposed regions of the VP1 protein, since we predicted that epitopes inserted into this region would be less likely to interfere with VLP formation. It is possible that this location may not have been optimal for the induction of cellular immune responses and that placing the epitopes on terminal regions may have been beneficial, as reported in one study in a chimeric calicivirus-like VLP system with ovalbumin CD8 epitopes showing significantly higher IFN-γ responses, targetspecific lysis and protection (Crisci et al., 2009). It is also possible that the lack of responses to the CD4 epitope may have been a dose related issue; the CD4 peptide component of the 30 µg VLP dose was approximately 0.52 µg peptide epitope equivalent, whereas the peptide immunized mice received nearly 60 times the epitope per dose. Our immunogenicity assessments used lymphocytes taken from spleen and not draining lymph nodes, but the strong responses detected in splenocytes from the peptide-immunized mice would be expected to accurately reflect a response in the draining lymph nodes; if the VLPs induced a robust CD4<sup>+</sup> T cell response then this should also have been detected in splenocytes.

To evaluate the protective capacity of our chimeric VLPs, we harvested livers 42 h post- sporozoite challenge, at the late liver-stage of the P. yoelii lifecycle before parasites were expected to release into the blood (Baer et al., 2007). This timepoint maximized the opportunity for VLP-induced protective immune responses to clear parasite-infected cells. Liver-stage protection has previously been achieved by adoptively transferring a CD8<sup>+</sup> <sup>280</sup>−<sup>288</sup> T cell clone (Weiss et al., 1988), or by passive transfer of monoclonal antibodies specific to the PyCSP central repeat (Charoenvit et al., 1991). These protective antigenic epitopes were incorporated within our VLP chimeras, and generated epitope-specific CD8<sup>+</sup> T cell and antibody responses. In our study, there was evidence that VLPs could induce protection against sporozoite challenge since sterile protection was seen in 2 of 5 mice and there was a reduction in parasite load; however, this protection was suboptimal. Our immunological data supports the induction of the type of immune response required to protect against Plasmodium sporozoite challenge, but the magnitude of those VLP-induced responses may have been insufficient to reach a protective threshold (Schmidt et al., 2010). This is consistent with other studies where protection correlated with the number of adoptively transferred clonal CD8<sup>+</sup> <sup>280</sup>−<sup>288</sup> T cells where 20 × 10<sup>6</sup> cells were required for sterile protection (Weiss et al., 1992), or with the number of memory CD8<sup>+</sup> T cells stimulated by immunizations (Schmidt et al., 2008, 2010), or with the amount of passive transferred monoclonal antibodies (Charoenvit et al., 1991).

Indeed, we found that more robust antibody and cellular responses were induced by a prime/boost regimen as compared with homologous VLPs, and we anticipate that this would be further enhanced with the inclusion of a second VLP boost.

Overall, our data showed that the MuPyV chimeric VLP platform is capable of inducing antigen-specific antibody responses, and established for the first time for this vaccine platform, their capacity to induce CD8<sup>+</sup> T cell immune responses. Future studies could evaluate alternative epitope insertion sites within the MuPyV VP1 protein to determine if other sites enhance cellular immune responses, or the inclusion of adjuvants in the vaccination regimen. Additional regions of the target antigens (e.g., repetitive regions targeted by antibody responses, together with T helper epitopes) could also be included. Alternative target antigens could be also evaluated since studies have suggested that multiple antigen targets may be required for protection from Plasmodium challenge (Doolan et al., 1996; John et al., 2005; Osier et al., 2008). With approximately 2,000 active genes, and 816 identified proteins present during Plasmodium liver-stage of infection (Tarun et al., 2008) part of the challenge is to identify the most optimal target antigens for induction of vaccine-induced protective immunity (Schussek et al., 2017).

### DATA AVAILABILITY

The datasets generated for this study can be obtained from the corresponding author upon reasonable request.

#### ETHICS STATEMENT

All animal experiments were approved by the QIMR Berghofer Animal Ethics Committee and were conducted in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes (2004).

## AUTHOR CONTRIBUTIONS

DP, SA, DD, and AM contributed conception and design of the study. DP, NW, YC, TR-H, LL, and AM contributed to construction of virus-like particles. DP, SA, and PG conducted mouse experiments. DP performed

## REFERENCES


the statistical analysis. DP wrote the first draft of the manuscript. DD and SA edited the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

## FUNDING

This research was funded by the National Health and Medical Research Council (NHMRC) Program grant #1037304, and by an Australian Infectious Diseases UQ-QIMR Seed funding grant. DD was supported by a NHMRC Principal Research Fellowship #1023636. DP was supported by an Australian Postgraduate Award.

## ACKNOWLEDGMENTS

The authors thank Professor Robert Garcea (University of Colorado, CO, USA) for the plasmid containing the MuPyV-VP1 protein, and Dr. Stephen Hoffman and colleagues (Sanaria Inc., Rockville, MD, USA) for providing cryopreserved sporozoites. TEM images were taken by Alice Lei Yu from the University of Queensland, Australia.


falciparum circumsporozoite protein in mice and monkeys induced by a novel formulation of RTS,S vaccine antigen. Vaccine 26, 1072–1082. doi: 10.1016/j.vaccine.2007.12.018


screen induce protection associated with polyfunctional T cell responses. Sci. Rep. 7:18. doi: 10.1038/s41598-017-15354-0


**Conflict of Interest Statement:** The University of Queensland (UQ) filed patents on the use of MuPyV as a vaccine platform. LL and AM contributed to those patents and, through their employment with UQ, hold an indirect interest in this intellectual property.

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 Pattinson, Apte, Wibowo, Chuan, Rivera-Hernandez, Groves, Lua, Middelberg and Doolan. 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.

# Delineating the Plausible Molecular Vaccine Candidates and Drug Targets of Multidrug-Resistant *Acinetobacter baumannii*

#### Shama Mujawar 1†, Rohit Mishra2†, Shrikant Pawar 3,4, Derek Gatherer <sup>5</sup> and Chandrajit Lahiri <sup>1</sup> \*

#### *Edited by:*

*Alberto Moreno, Emory University School of Medicine, United States*

#### *Reviewed by:*

*Gyanendra Prakash Dubey, Institut Pasteur, France Muhammad Ammar Zafar, Wake Forest School of Medicine, United States*

> *\*Correspondence: Chandrajit Lahiri chandrajitl@sunway.edu.my*

*†These authors have contributed equally to this work*

#### *Specialty section:*

*This article was submitted to Clinical Microbiology, a section of the journal Frontiers in Cellular and Infection Microbiology*

*Received: 01 February 2019 Accepted: 23 May 2019 Published: 20 June 2019*

#### *Citation:*

*Mujawar S, Mishra R, Pawar S, Gatherer D and Lahiri C (2019) Delineating the Plausible Molecular Vaccine Candidates and Drug Targets of Multidrug-Resistant Acinetobacter baumannii. Front. Cell. Infect. Microbiol. 9:203. doi: 10.3389/fcimb.2019.00203* *<sup>1</sup> Department of Biological Sciences, Sunway University, Petaling Jaya, Malaysia, <sup>2</sup> Department of Bioinformatics, University of Mumbai, Mumbai, India, <sup>3</sup> Department of Computer Science, Georgia State University, Atlanta, GA, United States, <sup>4</sup> Department of Biology, Georgia State University, Atlanta, GA, United States, <sup>5</sup> Division of Biomedical and Life Sciences, Lancaster University, Lancaster, United Kingdom*

Nosocomial infections have become alarming with the increase of multidrug-resistant bacterial strains of *Acinetobacter baumannii*. Being the causative agent in ∼80% of the cases, these pathogenic gram-negative species could be deadly for hospitalized patients, especially in intensive care units utilizing ventilators, urinary catheters, and nasogastric tubes. Primarily infecting an immuno-compromised system, they are resistant to most antibiotics and are the root cause of various types of opportunistic infections including but not limited to septicemia, endocarditis, meningitis, pneumonia, skin, and wound sepsis and even urinary tract infections. Conventional experimental methods including typing, computational methods encompassing comparative genomics, and combined methods of reverse vaccinology and proteomics had been proposed to differentiate and develop vaccines and/or drugs for several outbreak strains. However, identifying proteins suitable enough to be posed as drug targets and/or molecular vaccines against the multidrug-resistant pathogenic bacterial strains has probably remained an open issue to address. In these cases of novel protein identification, the targets either are uncharacterized or have been unable to confer the most coveted protection either in the form of molecular vaccine candidates or as drug targets. Here, we report a strategic approach with the 3,766 proteins from the whole genome of *A. baumannii* ATCC19606 (AB) to rationally identify plausible candidates and propose them as future molecular vaccine candidates and/or drug targets. Essentially, we started with mapping the vaccine candidates (VaC) and virulence factors (ViF) of *A. baumannii* strain AYE onto strain ATCC19606 to identify them in the latter. We move on to build small networks of VaC and ViF to conceptualize their position in the network space of the whole genomic protein interactome (GPIN) and rationalize their candidature for drugs and/or molecular vaccines. To this end, we propose new sets of known proteins unearthed from interactome built using key factors, KeF, potent enough to compete with VaC and ViF. Our method is the first of its kind to propose, *albeit* theoretically, a rational approach to identify crucial proteins and pose them for candidates of vaccines and/or drugs effective enough to combat the deadly pathogenic threats of *A. baumannii*.

Keywords: *Acinetobacter baumannii*, nosocomial infection, vaccine candidates, drug targets, network analysis

## INTRODUCTION

Nosocomial or hospital-acquired infections are among the multitude of diseases caused by the opportunistic pathogen Acinetobacter baumannii, one of the world's six most important multidrug-resistant (MDR) microorganisms identified in hospitals (Talbot et al., 2006; Lin and Lan, 2014). While critically ill patients in the intensive care unit (ICU) accounted for up to 20% of ventilator-associated pneumonia and bloodstream infections with A. baumannii (Fournier and Richet, 2006; Vincent et al., 2009), the mortality rates might reach up to 35% along with endocarditis, meningitis, skin and wound sepsis, and even urinary tract infections for immunosuppressive patients (Lin and Lan, 2014; Darvishi, 2016). The cases for treatment are complicated due to the fact that the severity of the infection depends on the site and the patient's susceptibility to such diseases (Antunes et al., 2014). Moreover, there exist numerous natural reservoirs for A. baumannii including natural and agricultural soil, vegetables, aquaculture, and other inanimate objects outside the hospital environment (Eveillard et al., 2013).

The complications, resulting in an array of diseases caused by A. baumannii, arise from a plethora of virulence factors used by the pathogen to access and colonize the host system. These include, but are obviously not limited to, porins, capsular polysaccharides, lipopolysaccharides, phospholipases, outer membrane vesicles (OMVs), metal acquisition systems, and protein secretion systems (Lee et al., 2017). Besides these, β-lactamases acquisition, efflux pumps up-regulation, aminoglycosides modification, target sites alteration, and permeability defects are the crucial factors guiding the mechanism of antibiotic resistance conferred by this pathogen (Lee et al., 2017). Further threats of infection arise from colonization outside the human host, mainly on medical devices, through the mechanism of biofilm production involving the associated pathways, proteins, secretion systems, and quorum sensing (Perez et al., 2007). With such a robust antibiotic resistance mechanism entailing a barrage of proteins comprising the host invading machinery, A. baumannii has been able to confer extensive drug resistance (XDR). In fact, such ability has gone to the extent of evading almost every new-generation antibiotic, including carbapenems, which used to be prescribed to treat MDR organismal infections (Viehman et al., 2014).

To cater to the need of addressing the urgent and pressing issues of antibiotic threats, vaccine development has been resorted to as one of the cost-effective and most promising strategies to prevent infections. This is due to the fact that inactivated whole cells and attenuated strains are able to elicit antibodies against multiple surface proteins, which can be utilized in the form of vaccines, to combat the antibiotic threats. In fact, efforts have been tested to utilize only parts of the pathogen without the administration of whole organisms. Thus, vaccines comprising multiple proteins of the bacterial outer membrane complex (OMC), OMVs, OmpA, auto-transporter (Ata), biofilm-associated protein (Bap), K1 capsular polysaccharide, and poly-N-acetyl-β-(1-6)-glucosamine (PNAG) have been shown to elicit antibodies and to induce protective immunity against infection, thereby giving promising results in early human clinical trials (Bertot et al., 2007; Fransen et al., 2007; Chiang et al., 2015). Moreover, recent studies to determine potential vaccine targets have delineated a combinatorial approach of in silico prediction tools with reverse vaccinology through comparative genome analysis and in vitro proteomics (Moriel et al., 2013; Singh et al., 2017). For instance, in silico analysis in A. baumannii helped the identification of a highly conserved outer membrane protein with β-barrel assembly, BamA, as the potential target for vaccine (Singh et al., 2017). However, the advent of new and emerging XDR strains of A. baumannii, which possibly arises from immune selection, might lead to antigen sequence variability and a down-regulation of the target antigens, thereby conferring poor "cross-protective efficacy" (Chiang et al., 2015). Therefore, the identification of potential antigens, expressed by the new and emerging A. baumannii strains during infections, still keeps the issue quite complex to be addressed.

Simplifying the complexity of diseases, caused by such XDR pathogens like A. baumannii, is thus challenging. Therefore, to unearth plausible antigenic proteins, potential enough to elicit an antibody response, a detailed analysis of the complex interaction of the proteins, involved in the disease phenomenon, might be helpful. Earlier attempts to computationally analyze such protein interaction networks or interactomes (PIN) of infectious diseases mostly focused on network centrality parametric values for the identification of candidate drug targets (Lahiri et al., 2014; Pan et al., 2015). Generally, in biological networks, centrality measures of degree (DC), betweenness (BC), closeness (CC), and eigenvector (EC) have been used extensively (Jeong et al., 2001; Lahiri et al., 2014; Pan et al., 2015). While DC gives a very basic understanding of the number of interacting partners of a particular protein, EC relates to the essential proteins interacting with other crucial partners in the disease phenomenon (Lahiri et al., 2014). Besides BC, EC has been shown to be a good target for drugs, albeit theoretically for infectious diseases and utilized in the identification of side effect-free drug targets of idiopathic diseases like cancer (Lahiri et al., 2014; Pan et al., 2015; Ashraf et al., 2018). However, in order to gain an insight into the global scenario of the disease complexity, the PIN needs to be

inspected thoroughly for an effective analysis, potential enough to be translational in nature. Thus, the whole genome protein interactome (GPIN) has been utilized to prune and decompose to obtain a core of highly interacting proteins through the kcore analysis approach (Seidman, 1983). This coupled with the functional module-based cartographic analyses of the global network (Guimerà and Nunes Amaral, 2005a) has already been adopted to theoretically identify potential role players in bacterial infectious diseases (Pawar et al., 2017, 2018).

In this study, we have similarly delineated the relevance of the aforementioned centrality parametric measures for PIN of the claimed vaccine candidates and virulence factors (Moriel et al., 2013) of A. baumannii, namely, VaCAB and ViFAB. To this end, we have analyzed KeFAB, the PIN of key factors responsible for virulence and pathogenicity of A. baumannii (Chen et al., 2015). The top rankers of these three PINs were mapped onto the GPIN of A. baumannii to unravel their position in the network space and rationalize their candidature for drugs and/or molecular vaccines compared to our own sets of proposed candidates for vaccines. We consolidate our findings by the antigenic potential of these proteins along with their active sites for drug targets. In summary, we analyze the PIN of different relevant pathogenic proteins of A. baumannii to identify the plausible potential candidates for vaccines and/or drugs targets.

## MATERIALS AND METHODS

#### Acronyms and Terms Utilized

The present study on A. baumannii comprises a conglomerate of several terms pertaining to graph theory in general and network biology in particular. **Table 1** lists them all in a comprehensive manner with an expansion of the acronym, where applicable, followed by a short description of the terms for the ease of reference of a broad interdisciplinary range of readers.

#### Dataset Collection and Processing

Datasets for A. baumannii proteins were collected in different ways. The focus was given for either the strains AYE or ATCC17978, having either substantial reports or evidences for causing MDR. For the strain AYE, 168 proteins mentioned in the **Supplementary Data 2** by Moriel et al. (2013) were collected and categorized as vaccine candidates (VaC, **Supplementary Data 1**) while 124 from **Supplementary Datas 3**–**5** reported by the same group were collated as virulent factors (ViF, **Supplementary Data 2**). Other proteins were collected using keywords used by default for the available strains of ATCC17978 and AYE in VFDB (Chen et al., 2015). To this end, using the same keywords, further proteins of these strains were collected through direct literature search and the sequences of all these were retrieved from UniProt database (Apweiler et al., 2004). The last category proteins from VFDB, UniProt, and literature were parked under key factors (KeF, **Supplementary Data 3**). The counts of identifiers (IDs) for the various KeF proteins were 973 for ATCC17978, 92 for AYE, and 24 for no IDs found in the literature. Nine of these IDs, belonging to t-RNA, were eliminated, leaving a final total of 1,078 KeF (**Figure 1**).

The datasets of the aforementioned VaC, ViF, and KeF proteins were converted into the counterparts of the corresponding A. baumannii strain ATCC19606, due to the reason of non-availability of ATCC17978 and AYE strains in the STRING 10.5 biological meta-database of protein interaction (Szklarczyk et al., 2016). A FASTA file containing a total of 3,376 protein sequences, downloaded from STRING for A. baumannii strain ATCC19606, was used as a database for standalone BLAST during mapping of the sequences of strain AYE onto strain 19606. Standalone BLAST is a non-graphic user interface version of BLAST that runs on command lines in Linux operating system (OS) and allows execution of BLAST locally on such OS for sequence alignment. The execution of such BLAST, returned 2,111 and 1,676 hits for VaC and ViF, respectively. A filtering of the topmost ones with the highest percentage of identities from these hits returned 168 and 123 proteins, respectively, of the strain 19606. A further threshold cutoff of 99–100% identity was set to select out the identical proteins of strain AYE in 19606 for the next set of analysis, leaving out other ambiguous ones (duplicates and/or percentage-wise less identical) to obtain 79 VaC and 78 ViF proteins of strain 19606. Similar approaches were adopted for selecting out the KeF proteins of the strain 19606 from those of the strains ATCC17978 and AYE. For this, 1,078 KeF protein sequences were executed in standalone BLAST to yield a total of 15,793 hits, from which the top-ranked 1,075 proteins, with highest percentage identity, were selected out. Final cutoff of 99–100% identity was used to remove ambiguous ones and duplicates to obtain 640 IDs for further processing (**Figure 1**).

Submission of these mapped proteins of ATCC19606 into STRING helped the retrieval of protein interaction datasets having the default medium (0.4) level confidence upon the interaction (period of access: August to September 2018). The detailed protein links file containing the interaction datasets for the whole genome of A. baumannii strain ATCC19606 was retrieved for the accession number 575584 in STRING. All interaction datasets for each category of VaC, ViF, and KeF proteins, hereafter used for PIN construction, have been listed in **Supplementary Datas 1**–**3**.

#### Interactome Construction

After the removal of duplicate interactions, all individual interaction data obtained as above were imported for construction and visualization of the small PINs (SPINs), which were named VaCAB, ViFAB, and KeFAB, and the whole genome PIN named GPIN, using Cytoscape version 3.6.0 (Shannon et al., 2003) and Gephi 0.9.2 (Bastian et al., 2009) (**Figure 2**; **Supplementary Datas 1**–**3**: Sheets 5–7). All interactomes were considered to be non-directional in nature to represent undirected graphs as G = (V, E), where V are finite set of vertices and E are edges in which e = (u,v) connecting two vertices (nodes), u and v, or proteins in the present context. Thus, the degree, d (v), indicates the number of interactions (physical and functional) a protein has with other proteins (Diestel, 2000).

#### TABLE 1 | The acronyms and terms used in the present study.


#### PIN Analyses SPIN

Each of the constructed three SPINs were subsequently analyzed utilizing BC, CC, DC, and EC measures of centrality, commonly applied to biological networks (Pavlopoulos et al., 2011) (**Supplementary Datas 1**–**3**). This was done with the Cytoscape integrated java plugins, namely, Network Analyzer and CytoNCA (Assenov et al., 2007; Tang et al., 2015), utilizing the edge weights as the combined scores obtained from different parameters in STRING. These combined scores, which range from 0 to 1, generally convey the confidence of the protein's interaction with level of parametric evidences from gene neighborhood, gene fusion, gene co-occurrence, gene co-expression, experiments, annotated pathways, and text mining. Using the top 15 proteins appearing for each measures of analysis as mentioned above, a commonality of the proteins for all centrality measures was observed through Venny 2.1 (Oliveros, 2007–2015) (**Figure 3**).

#### GPIN

The whole genome protein interaction network, GPIN, was analyzed by MATLAB version 7.11 (MATLAB Statistics Toolbox Release, 2010). An idea of the simplest of the technical aspects of the GPIN, i.e., the distributions of network degree (k), was obtained by plotting it against the Complementary Cumulative Distribution Function (CCDF) (**Figure 4A**). A kcore analysis of the whole genome was done by a network decomposition method that produces a gradually increasing cohesive sequence of sub-graphs and reveals the number of proteins having at least k-degree classified in K-shell as per their interacting patterns (**Figures 4B,C**) (Seidman, 1983). Further knowledge of the connectivity and participation of each protein, with respect to their functions, was obtained through analysis of the cartographic representation of the network topology that plots the within-module degree z score of the protein against its participation coefficient, P (**Figure 5**) (Guimerà and Amaral, 2005b). The participation of each protein in a modular network, sharing common biological function, gives rise to a concept of functional module having high intra-connectivity and sparse inter-connectivity (Vella et al., 2018). Calculations of such functional modules as per Rosvall method (Rosvall and Bergstrom, 2011) led to a major classification of the proteins of GPIN into

non-hub and hub nodes, each having further sub-classes. These are ultra-peripheral (R1), peripheral (R2), non-hub connector (R3), and non-hub kinless (R4) for the former while provincial (R5), connector (R6), and kinless (R7) for the latter categories (Guimerà and Amaral, 2005b) (**Figure 5**; **Supplementary Data 4**).

## Vaccine and/or Drug Candidature Prediction

The network analyzed and shortlisted VaC, ViF, and KeF proteins along with their interactors from the three SPINs, namely, VaCAB, ViFAB, and KeFAB, were subjected to further analyses for predicting the plausible vaccine and/or drug candidates. To this end, the indispensable proteins of the GPIN were also taken into consideration as per the KFC method described by Ashraf et al. (2018). All such proteins were explored for their cellular localization, signal peptide prediction, COG classification, antigenic site prediction, followed by active site prediction. Cellular localization was analyzed by PSORTb v3.0.2 (Yu et al., 2010). Location of signal peptides was predicted using the server called SignalP 4.1 Server (Petersen et al., 2011). Lipoprotein signal peptides were predicted using the LipoP 1.0 Server (Juncker et al., 2003). For uncharacterized proteins, functional annotation with classification was done through the WebMGA server (Wu et al., 2011). Such COG classification from WebMGA was performed for all proteins, however, even if they are characterized, to maintain unanimity of comparison. To predict the antigenic potential of the candidate proteins, epitope prediction was done through the immune epitope database (IEDB) resource utilizing Bepipred Linear Epitope Prediction and maintaining a threshold cutoff of 0.75 for increased sensitivity (Vita et al., 2014). Furthermore, without any solved X-ray crystallographic or NMR 3D structures for the selected proteins, they were homology modeled and validated to pursue active site prediction studies. We have used Phyre2 (Kelley et al., 2015) and SWISS MODEL (Schwede et al., 2003) protein modeling servers to generate the structures and the integrated Procheck server in the latter, to evaluate them through Ramachandran plot, Q mean score, and z score. To identify the best structure of the different models generated by the abovementioned servers, we have performed consensus studies. These consensus models were finally utilized to determine the active sites or binding pockets of the selected proteins by the CASTp server (Computer Atlas of Surface Topology of protein) (Dundas et al., 2006). A comparative account of the results obtained from this section of the analysis is shown in **Table 7**.

FIGURE 3 | Venn diagram representation for the top-ranking network centrality measures of SPIN and GPIN of *A. baumannii*. (A,B) VaCAB, (C,D) ViFAB, (E,F) KeFAB, and (G) GPIN. Measures of four types of centrality are from CytoNCA and three types are from Network Analyzer. BC, CC, DC, and EC denote betweenness centrality, closeness centrality, degree centrality, and eigenvector centrality, respectively.

### Interactor-Free Candidature Analyses

In order to determine the important vaccine candidates among a barrage of proteins proposed by Moriel et al. (2013), filtration was used to remove the influence of the interactors on the actual candidates upon their ranking. For the same, we have initially merged the VaCAB and ViFAB together, having their individual centrality-based analyzed data. Thereafter, the VaC and ViF interactors, of the respective PIN, were removed, leaving behind the actual candidates. As the PINs (VaCAB and ViFAB) were constructed from STRING data using the strain ATCC19606, we mapped back the centrality measures of the total VaC and ViF proteins onto the AYE strain of A. baumannii. This was followed

by assigning the localization status of the total set of candidate proteins through the usage of PSORTb. Sorting was done based on localization followed by descending order of the centrality measures. Top five rankers, in each case, were used for creating a Venn diagram to find out the most promising candidates from the intersection of the centrality measures (**Figure 1**).

#### RESULTS

#### The Candidate Proteins

In order to the identify the potential molecular vaccine candidates and drug targets of MDR A. baumannii, we have started with the available pathogenic strains, namely, AYE and ATCC17978, to accumulate the proclaimed proteins in categories of VaC, ViF, and KeF with counts of 168, 124, and 1,078, respectively (**Supplementary Datas 1**–**3**: Sheet 1). As these strains are not enlisted in the protein interaction database, STRING, we had to find the real counterparts of these proteins, from the only listed strain ATCC19606. An initial processing of the data through standalone blast from the total protein sets of ATCC19606 yielded 168, 123, and 1,075 VaC, ViF, and KeF candidates, respectively (**Supplementary Datas 1–3**: Sheet 3), which were further filtered for an exact match of 99–100% to result in 79, 78, and 640 proteins, respectively (**Supplementary Datas 1**–**3**: Sheet 4). Thus, all the candidate proteins were mapped onto ATCC19606 from other pathogenic strains. All data, hereafter, refer to strain ATCC19606 only, unless otherwise stated.

### The Three Individual SPINs

To identify the crucial candidates among the three aforementioned sets of VaC, ViF, and KeF proteins of A. baumannii, we have taken an approach of building three different SPINs, each with its respective set of interaction data (**Figure 2**; **Supplementary Datas 1**–**3**: Sheets 5–7). Upon analyzing these SPINs through the Cytoscape plugins CytoNCA and Network Analyzer, the top rank holders as per each centrality measures are recorded (**Supplementary Datas 1**–**3**: Sheets 8–11), indicating their importance. A combined result from both the plugins are presented in **Table 2**, wherein protein entries are represented by their NCBI gene accession numbers with the suffix "HMPREF0010\_" being replaced by "AB\_" for the ease of further use. It is to be noted that there are overlaps of these protein entries across the different centrality measures and/or the different categories of VaCAB, ViFAB, and KeFAB. An overlap of the proteins across the measures within a particular category is represented by a "caret" (∧) notation while that from cross-categories is denoted with an "asterisk" (<sup>∗</sup> ). For instance, the protein AB\_03493 is associated with the VaCAB category and is represented with V1<sup>∗</sup> , due to its presence in the category of ViFAB (BC measure), despite its presence in two measures of BC and DC of VaCAB. On the contrary, AB\_1947 is present in the BC, CC, and DC categories of ViFAB only and is represented by F5∧. The numbers of the proteins to denote such overlaps are put randomly, however, and do not indicate the actual ordering of the rank as depicted in **Table 2**. Each one of such overlapping proteins has been indicated in bold to stand out from the rest in the same measures and/or categories and has been utilized for further mapping in the whole genome context later (**Figure 5**). It is important to note that there are more overlaps between the VaCAB and ViFAB categories of proteins as indicated by the asterisk in the entries. On the contrary, proteins of KeFAB tend to have much fewer overlaps among other categories. The numbers of the top-ranking proteins from each of the categories and measures of centralities and their overlaps are reflected with Venn diagrams (**Figure 3**). The functions of these proteins, taken from UniProt, reflect those that have been mostly characterized, with some having proven and eminent roles as vaccine and drug targets (**Supplementary Datas 1**–**3**: Sheet 12).

Despite the overview obtained about the importance of the proteins from a comparative account within and between each of the aforementioned categories of VaCAB, ViFAB, and KeFAB, it is to be noted that these PINs are a mixture of the actual VaC, ViF, and KeF proteins along with their interactors that were extracted from STRING. Interestingly, of the 15 top rank holders of each category of PIN, the actual proteins were 9, 3, 7, and 1 as per BC, CC, DC, and EC measures of VaCAB, and the rest were their interactors. Similarly, for ViFAB, the actual VaC proteins were 6, 6, 3, and 1 for the same measures, respectively, as mentioned above. For KeFAB, however, the numbers of KeF proteins were almost the same throughout, being 4, 5, 5, and 5 for BC, CC, DC, and EC, respectively (**Table 3**). However, to decipher the effective drug or a vaccine candidate from the global perspective of the whole genome, the total sets of proteins interacting in the whole genome of A. baumannii (GPIN) were analyzed as described in the next section.

## The Complete GPIN

In an attempt to project the most probable candidates for vaccines and drug targets, we have further explored the GPIN. GPIN, built from the theoretically predicted and empirically found physical and functional interactions, was analyzed for its degree distribution in general and k-core distribution in particular. With an exponential curve for the cumulative degree distribution frequency (CCDF), the non-linear preferential attachment nature was evident (**Figure 4A**) (Vázquez, 2003). To this end, an idea of the important proteins forming the core of the genome was obtained through the k-core network decomposition. The shell size of the 165th innermost core had a conglomerate of 494 proteins (**Figures 4B,C**).

We have started analyzing the GPIN with the four common centrality measures as mentioned for SPIN (**Table 2**). A closer look at the top 15 from each measure reflects proteins mostly from the categories of KeF, two of which have overlaps with the ViFAB category (F1 and F2, **Table 2**). Around two-thirds of the top-ranking positions in DC are occupied by KeFAB proteins, while 99% of the top 15 EC measures comprise the KeFAB categories.

A detailed k-core analysis of the different proteins as per the centrality measures revealed other important facts. The numbers of VaCAB proteins belonging to the innermost core are 2, 3, 7, and 12 as per the top rank holders of BC, CC, DC, and EC measures (**Table 3**). Of these, the numbers of VaC proteins are 1, 1, 2, and 1 as per the same measures while the rest are their interactors (**Table 3**). This analysis filters out HMPREF0010\_01334 as per the BC, CC, and DC measure overlaps and HMPREF0010\_02233 as per EC, to be in the innermost core of the whole genome from the proclaimed VaC proteins. For the ViFAB proteins, 6, 10, 14, and 14 proteins, respectively, for BC, CC, DC, and EC measures belong to the 165th core, of which only 1, 2, 2, and 1, respectively, are the VaC proteins while the rest are their interactors (**Table 3**). This filters out HMPREF0010\_03351 from BC, CC, and DC overlaps, while the new candidates HMPREF0010\_01731, HMPREF0010\_02683, and HMPREF0010\_01948 stand out from CC, DC, and EC, respectively. The picture becomes different when it comes to the KeFAB where CC, DC, and EC all had only 165th core proteins from the top rankers while BC reflected 14 of them. Moreover, the actual KeF candidates are more in this PIN compared to the VaC and ViF proteins. The analysis filters out 4, 5, 5, and 5 KeF proteins from the BC, CC, DC, and EC measured top rankers (**Table 3**). This brings forth the candidates HMPREF0010\_03512 from BC and CC, HMPREF0010\_00239 from CC and DC, and HMPREF0010\_00091 from DC and EC overlaps. Moreover, three candidates, namely, HMPREF0010\_02437, HMPREF0010\_01461, and HMPREF0010\_02576, were unique in having overlaps across CC, DC, and EC measures, while HMPREF0010\_01789 from EC and three candidates from BC, namely, HMPREF0010\_03553, HMPREF0010\_01114, and HMPREF0010\_00382, were unique. It is important to note that TABLE 2 | The topmost proteins of *A. baumannii* SPIN and GPIN as per BC, CC, DC, and EC network centrality measures.


*The bold cased proteins are overlapping either within same categories (*∧*) or between others (*\**) in SPIN. Numbers are arbitrary.*

TABLE 3 | The innermost core proteins of VaCAB, ViFAB, KeFAB, and GPIN as per the four centrality measures.


*(Continued)*


*Shades of green in light are for interactors while those in dark are the candidates from each category.*

sorting of the whole genome proteins as per the four centrality measures yielded 2, 0, 14, and 15, respectively, of which 0, 0, 4, and 2 proteins, respectively, belong to either the VaC, ViF, or KeF category while the rest are their interactors (**Table 3**).

Furthermore, to identify the candidate VaC, ViF, and KeF proteins in the network topological space of A. baumannii, we have classified the protein sets of GPIN and represented them cartographically (**Figure 5**; **Supplementary Data 4**). Essentially, one part of such classification, namely, z score, is based on their regional connectivity with other similar proteins having a similar biological function, which is referred to as the functional module. The other part, namely, the participation coefficient, P, deals with the participation of these proteins with other functional modules, either related or non-related. Thus, there are seven such quadrants that are formed and termed R1 to R7. Noticeably, the proteins of VaCAB and ViFAB are spread throughout R1–R3 in the network space, while those of KeFAB are mostly concentrated in R5 (**Figure 5**; **Supplementary Data 4**). Interestingly, R4 and R6 did not show any mapping of the VaC, ViF, or KeF proteins or their interactors, represented by V, F, and K comprising the full VaCAB, ViFAB, and KeFAB SPIN (**Figure 5**; **Supplementary Data 4**). Furthermore, there were no proteins from the GPIN that occupied the R7 quadrant, having the highest P or z scores. With a more focused analysis to find out the most indispensable proteins of GPIN, we have performed the KFC method mentioned by Ashraf et al. (2018). These KFC proteins all belong to the innermost 165th core and had R5 as their classifying P vs. z quadrant and high EC scores (**Table 3**; **Supplementary Data 4**).

#### The Candidature Prediction

To ascertain the plausible candidature of the proteins of VaCAB, ViFAB, KeFAB, and GPIN as either vaccine or drug targets, we have reclassified them as VaC, ViF, and KeF, along with their interactors and KFC proteins, as mentioned in the previous section. This projects 2 VaC with 13 interactors, 4 ViF with 29 interactors, 19 KeF with 41 interactors from the 165th core and 10 KFC proteins for further analyses (**Tables 3**, **4**). Among these, the cellular localization could not be determined for one VaC, two ViF, five KeF, and two R6 proteins, respectively. Notably, one candidate from VaC, namely, AB\_02233, belongs to the outer membrane, while the periplasmic and extracellular comprise the ViF category. The remaining categories of KeF, KFC, and R6 mostly comprise cytoplasmic proteins having few cytoplasmic membrane proteins for KeF and R6 as well (**Table 7**). To this end, only one KeF (AB\_01691) did not reflect any COG classification for the aforementioned categories. On the contrary, 50% of R6 proteins did not reflect any COG, being uncharacterized. Interestingly, only one protein from each of the VaC, ViF, and KeF categories, namely, AB\_02233, AB\_03351, and AB\_01758, respectively, predicted a signal peptide while KFC and R6 did not show any. Moreover, only one lipoprotein cleavage site was predicted for ViF, KeF, and R6, namely, AB\_03351, B\_01758, and AB\_00641, respectively, while VaC has two proteins (AB\_01334 and AB\_02233) showing the same compared to KFC, having none. Furthermore, about 50% of the proteins of the category KeF and KFC indicated an overlap of the antigenic site with the active pocket site while proteins of VaC, ViF, and R6 had no such overlap for candidature prediction (**Table 7**).

### The Final Selection

Finally, to assess for the relevance of the final set of proteins (**Table 4**), in virulence and pathogenicity of Acinetobacter, we have cross-examined through UniProt, PDB, and PubMed, for their role, either predictive or empirical, in Acinetobacter and/or other gram-negative bacterial pathogens. Two unique VaC proteins, namely, AB\_01334 and AB\_02233, are predicted to encode tetratricopeptide repeat protein and peptidase M16 inactive domain protein, respectively (**Table 7**). Annotations for other ViF proteins like thiol:disulfide interchange protein (AB\_03351), malate dehydrogenase (AB\_02683), and 50S ribosomal protein L7/L12 (AB\_01948) were inferred from homology while AB\_01731 was experimentally verified to have the function of a nucleoside diphosphate kinase (**Table 7**). The KeF proteins are shown to be a conglomerate of different types engaged in different biological process starting from carbohydrate, amino acid, and DNA metabolism to even those involved in signal transduction, cell wall synthesis,

#### TABLE 4 | The candidate proteins of VaC, ViF, KeF, and KFC along with their interactors in *A. baumannii* ATCC19606 strain.


*VaC, ViF, KeF, and KFC denote Vaccine candidates, Virulent Factors, Key Factors, and K-core-Functional modularity-Centrality, respectively. The prefix of the accession numbers for the proteins has been replaced by AB.*

ribosomal and translational machineries, as well as some uncharacterized proteins (AB\_02008 and AB\_01691; **Table 7**). On the contrary, proteins of the KFC class are only concentrated on carbohydrate, amino acid, and fatty acid metabolism with the inclusion of one (AB\_01641) having DNA polymerase with 5′ - 3 ′ exonuclease activity. Again, proteins of the R6 category are mostly either uncharacterized or belong to the transcriptional regulator family with the inclusion of one CRISPR-associated protein (AB\_01430).

All the corresponding candidates from ATCC19606 have been mapped onto ATCC17978 and AYE strains of A. baumannii as reflected in **Table 5** for the ease of use by future researchers.


TABLE 6 | The human counterparts of the corresponding VaC, ViF, KeF, KFC, and R6 candidates of *A. baumannii* ATCC19606 strain.


Moreover, to determine the orthologous presence of these proteins in the human host, their pairwise identities (PI) and query coverages (QC) are reflected in **Table 6**. Notably, the VaC proteins have considerably low PI and QC, which is exactly opposite in nature to all but one of the proteins of KFC, having high QC and PI. Intriguingly, ViF, KeF, and R6 comprise a mixture of unique proteins having no human counterpart as well as those having moderate to high PI. Of these, however, only proteins of ViF have very high QC as well (**Table 6**).

#### *VaC, ViF, KeF, and KFC denote Vaccine candidates, Virulent Factors, Key Factors, K-core-Functional modularity-Centrality and Function-R6, respectively. The prefix of the accession numbers for the proteins has been replaced by AB.*

AB\_00406 – – – AB\_00641 – – – AB\_01430 NP\_005236.2 12% 36% AB\_01223 – – – AB\_01974 – – –

#### DISCUSSION

The sole aim of this study is to look out for plausible vaccine and/or drug candidates among a plethora of proteins from the whole genome of A. baumannii. In this context, the work proposed by Moriel et al. (2013) has mentioned an array of proteins as candidates for vaccines to test out in reallife scenario. Besides these, several researchers have proposed different virulence factors crucial for the XDR A. baumannii, probably potential enough to be targeted as drugs. Moreover, the VFDB presents lists of several such factors as well. Considering a filtration to shortlist just few of these would probably be a good idea to save the time and money of future researchers in this field of study. Thus, network analysis is being considered in order to sort and identify, albeit theoretically, the most probable candidates among them.

With the target being set out for the pathogenic strains like AYE and ATCC17978, the main hindrance was the lack of protein interaction datasets for these strains in the STRING database. This led to the mapping of the proteins from these strains onto ATCC19606. We started with an initial categorization of the protein sets proposed by Moriel et al. (2013) based on their results. One set, representing the A. baumannii antigenic proteins identified through reverse vaccinology approach, was denoted as Vaccine Candidates (VaC). The other set was named Virulence Factors (ViF) and comprised all other proteins listed by the same group including OMVs and secretome, potentially insoluble proteins and periplasmic proteins found in OMV and secretome of A. baumannii. Initial filtration through standalone BLAST led to almost the same number of proteins in ATCC19606. A stringent filtering approach with 99–100% identity cutoff threshold, however, was adopted to rule out any ambiguity of the protein functions upon such conversion from the former strains to the latter. Similar strategies were adopted for databaseand literature-retrieved proteins searched through keywords and were named Key Factors (KeF).

As the proteins would always be interacting with others to manifest their functions, a PIN construction was the next move for the VaC, ViF, and KeF protein sets to yield the three SPINs namely, VaCAB, ViFAB, and KeFAB. These were analyzed for their importance through the four centrality parameters BC, CC, DC, and EC, often utilized for biological network analysis (Jeong et al., 2001; Lahiri et al., 2014; Pan et al., 2015; Pawar et al., 2017, 2018; Ashraf et al., 2018). Among them, DC reflects the simple network connectedness of any protein, while for a virulent phenotype, CC might bring out the functional proximity of a protein with others. Moreover, BC might help in reflecting the bridging of different functionally important groups of virulent proteins, thereby posing its importance to be targeted for therapeutic purposes. EC, however, might reflect the connectivity of the most important proteins with other important proteins in a virulent network, thereby posing them to be indispensable for therapeutic targets. Notably, there were overlaps between VaC and ViF categories of proteins across different centrality measures (**Table 2**), probably indicating a faint line of difference between them, which actually were set by this study and not by Moriel et al. (2013). Moreover, there were very few overlaps of VaC and ViF proteins with those of KeF sets, probably indicating the uniqueness of the former groups compared to the common KeF proteins already reported in the literature and database searched through keywords (**Table 2**).

The functional aspects of the candidate proteins would be best put forward through the whole genome global scenario for which the GPIN was constructed (**Figure 2D**) and analyzed by the network centrality and other topological parametric measures (**Supplementary Data 4**). The initial characterization of the GPIN was done through the exponential decay of the degree distribution, P(k), of a particular node upon connecting to k other nodes, for large values of k. Such construction at least confirms the non-random (Erds and Rényi, 1960) or nonsmall-world nature (Watts and Strogatz, 1998) of the GPIN, if not completely following the power law (Albert et al., 2000) and becoming scale free (**Figure 4A**). Hereafter, the constructed GPIN is analyzed with the four centrality measures. Proteins of KeFAB categories occupying most of the EC and DC measures probably indicate the importance of either of these measures in bringing out the top rankers of KeFAB proteins (**Table 2**). Moreover, with no appearance among the top 15 important categories, the VaCAB and ViFAB proteins might not be so essential from the whole genome perspective (**Table 2**).

The actual set of proteins important from the GPIN perspective are probably reflected by the innermost core of the proteins brought about by the k-core/K-shell topological parameters (**Figures 4B,C**). The concept of the importance of the innermost k-core lies in the fact that k-core is a subnetwork with a minimum number of k-links such that the 165th innermost core would have 165 connections of each of those proteins lying in that core (**Figure 4B**; **Supplementary Data 4**). Essentially, the number of proteins in this case is 498 (**Figure 4C**; **Supplementary Data 4**). With such a large inner core member proteins having high connectivity, the core tends to be highly interactive and, thus, robust in nature (Alvarez-Hamelin et al., 2005). This could be indicative of the tight control resistance mechanism of this MDR/XDR species of A. baumannii. It is important to note that the relation between a K-shell and k-core is that the former is the part of the latter but not of the (k + 1) core, such that the former is a set of nodes having exactly k-links. This brings out the fact that there are a lower number of proteins (interacting among their partners) with lower k-core that belongs to the outer shell, and thus, the innermost core would have the maximum number of proteins needed to be decomposed to affect the global network of A. baumannii. Interestingly, only limited VaC and ViF proteins (2 and 4, respectively) belong to the 165th innermost core identified through all different centrality measures (**Table 3**). These numbers are 12–13% of the total proteins of VaCAB and ViFAB including the VaC and ViF proteins along with their interactors. In comparison to these, ∼32% of the total proteins of KeFAB are important as KeF (**Table 3**). These probably tells us that only few proteins are important from VaC and ViF compared to KeF categories. Thus, considering just these few proteins as either vaccines or drug targets may not be sufficient to break the robust inner core compared to a large number of options available for KeF proteins.

With some preliminary idea about the centrality and kcore measures, we moved on to delve deep into the functional connectivity, R, of the modules formed in the GPIN. Such connectivity of the proteins within and between the functional modules is represented cartographically by P-values and z score, respectively, across the x- and y-axis. This results in the lowest values of P and z for R1 and the highest for R7. The classifications are thus named ultra-peripheral proteins (R1) and peripheral proteins (R2), which can be detached with convenience from TABLE7|Protein analyses of selected VaC, ViF, KeF, and KFC proteins of*A. baumannii* for cellular localization, COG classification, antigenicity, and active site predictions.


Mujawar et al.

*(Continued)*


and Drug Targets of *A. baumannii*

*(Continued)*

Vaccine Candidates

Mujawar et al.


**127**


Mujawar et al.

*(Continued)* Vaccine Candidates

and Drug Targets of

*A. baumannii*


the network. Moreover, the non-hub connectors (R3) might be involved in connecting fundamental sets of interactions while the non-hub kinless proteins (R4) connect other proteins evenly distributed across the modules without forming hubs themselves. Furthermore, the provincial hub proteins, R5, have many withinmodule connections, whereas the connector hubs, R6 proteins connect most of the other modules, and thus are probably most conserved in terms of decomposition as well as evolution. Finally, the kinless hubs, R7 proteins, show the highest connection within and between the members of the GPIN such that they could be the most essential ones to be maintained by the pathogen for its very survival. Taking into account the importance of such functional modules, we have sorted the GPIN as per the KFC method adopted by Ashraf et al. (2018). We found the 165th core getting aligned with the functional module R5 proteins having the highest EC values (**Table 3**; **Supplementary Data 4**: Sheet 2).

With a consolidated set of VaC, ViF, KeF, and KFC proteins, we have set our ultimate goal to determine their antigenic potential and predicted some probable active sites. As cellular location plays an important role in conferring the potential of proteins as vaccines and/or drugs, we have attempted for the same and identified the COG classes for them, keeping in view the uncharacterized proteins taken as reference for the R6 categories. Interestingly, most of the proteins of VAC, ViF, KeF, and KFC could be classified as per COG, suggesting their likelihood to share similar functions. To this end, the antibody epitope site prediction shared 50% similarities with the predicted active sites of the validated homology modeled structures of KeF and KFC proteins, suggesting their plausibility to be used either as vaccine candidates or as drug candidates. On the contrary, VaC, ViF, and R6 proteins were unique enough and would probably have to be experimentally verified for their actual candidature. All such VaC, ViF, KeF, and KFC proteins were mapped back from ATCC19606 strains to the strains of AYE or ATCC17978 for the ease of quick referral for future researchers (**Table 5**). Besides the aforementioned list in **Table 4**, a comparison of the proteins with their Homo sapiens counterpart has been made as reflected in **Table 6**. This was done to ensure that the proteins of A. baumannii are different from their host, either fully or largely, and thus could be used as targets.

A comparison of the data from **Tables 6**, **7** shows that the protein with the accession ID AB\_02233 is a good candidate for vaccine, residing on the outer membrane, having a signal peptide, and bearing much less similarity with its human counterpart. Similarly, the protein AB\_03351 is a good candidate for drugs, having periplasmic location, a signal peptide, and no match at all with the human host proteins. A very difficult comparison is posed for the KeF proteins, which largely present themselves as either cytoplasmic or cytoplasmic membrane. The protein with the accession number AB\_01758 could have some potential though its similarity with the human counterpart rules out such possibility. Proteins of the KFC category, surprisingly, do not present themselves as good candidates at all, all being cytoplasmic, having no signal peptides, and bearing huge similarity with their human homologs. The protein AB\_02522 in this category having much less query coverage (3%), however, does not stand a chance either, owing to its cellular localization in the cytoplasm. Similarly, most cytoplasmic proteins of the R6 category present themselves as poor candidates for vaccines and/or drugs. One uncharacterized protein, viz., AB\_00210 of unknown cellular localization in this category could be of some potential due to the absence of any match with human counterpart.

A completely different approach of interactor-free centralitybased ranking of the different classes of the candidates proposed by Moriel et al. (2013) unanimously pulls out AB\_00353 from proteins of the outer membrane class, AB\_01731, from those of the extracellular region and a set of five proteins from the periplasmic region category (**Supplementary Data 5**). Among these, the protein AB\_00353 or BamA has already been found by some in silico approach earlier and reported to elicit high IgG antibody titer with the production of opsonizing antibodies against a virulent MDR clinical isolate using a murine pneumonia model (Singh et al., 2017). The other protein, AB\_01731, coding for nucleoside diphosphate kinase, has also been reported by another group through reverse vaccinology approach (Chiang et al., 2015). Of the five periplasmic proteins, HMPREF0010\_03351 (**Supplementary Data 5**) (reflected as AB\_03351 in **Table 7**) has been found out by our detailed interactome-based approach as well. These proteins have the potential to be used as either vaccine candidates, for the outer membrane proteins, or drugs, for other periplasmic or cytoplasmic proteins. Unknown or uncharacterized proteins from different aforementioned categories (**Table 7**), however, can be worked upon by future researchers for more prospective candidates as well. Thus, a mixed-bag result has come out of the analyses done through our two approaches. Of the interactome-based approach, where the interactors have been considered as well, the actual candidates were sorted through network centrality and protein signature analyses and the results need to be further experimentally validated as most of the candidates are novel and not reported earlier. The other, interactor-free approach, considered the candidates from their cellular localization and network centrality analyses unanimously bring out two candidates already reported earlier. The success of future researchers in targeting a protein for vaccine candidates and/or drugs would, thus, depend on further experimental validation for any category of proteins.

## CONCLUSION

The study is based on the concept of utilizing the already proclaimed vaccine candidates and virulent and other key pathogenic factors to sort and filter them to a useful list of most probable final candidates. Essentially, the work revolved around a network biological approach of analyzing the conceived networks of the aforementioned proteins and mapping them onto the whole genome perspective to shortlist the candidates. To this end, established methods of antigenicity and active site prediction have been added to produce the final list for further experimental validation.

## AUTHOR CONTRIBUTIONS

The analyses and the study were conceptualized, planned, and designed by CL. Data generated by RM and SM were analyzed by CL and supported by SM and RM with tabulation. Artwork was done by RM. CL wrote the manuscript aided by inputs from SP, SM, and DG in general and RM and SM for the Materials and Methods section.

#### ACKNOWLEDGMENTS

The authors acknowledge the support of Sunway University, Selangor, Malaysia, and IMSc, Chennai, India, for providing the computational facilities. The authors immensely acknowledge the contribution made by Md. Izhar Ashraf in providing some initial network data for further analyses. They also wish to thank Indhuja Thirumudi, Ng Jia Wei, and Vaibhav Kandale for their preliminary initiatives to earlier forms of this work which metamorphosed to the current state.

### REFERENCES


Diestel, R. (2000). Graph Theory. Heidelberg: Springer-Verlag.

Dundas, J., Ouyang, Z., Tseng, J., Binkowski, A., Turpaz, Y., and Liang, J. (2006). CASTp: computed atlas of surface topography of proteins with structural and

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2019.00203/full#supplementary-material

Supplementary Data 1 | Excel sheets showing the data collection, processing, interactome construction, analysis, and functions of vaccine candidates, VaC of *A. baumannii*.

Supplementary Data 2 | Excel sheets showing the data collection, processing, interactome construction, analysis, and functions of virulent factors, ViF of *A. baumannii*.

Supplementary Data 3 | Excel sheets showing the data collection, processing, interactome construction, analysis, and functions of key factors, KeF of *A. baumannii*.

Supplementary Data 4 | Excel sheets showing the data collection, processing, interactome construction, analysis, and functions of whole genome proteins of *A. baumannii*.

Supplementary Data 5 | Excel sheets showing the strain mapping, cellular localization, and centrality measures along with Venn diagram for the total set of candidate proteins of *A. baumannii*.

topographical mapping of functionally annotated residues. Nucleic Acids Res. 34(Suppl.2), W116–W118. doi: 10.1093/nar/gkl282


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

# The Quest for a Truly Universal Influenza Vaccine

#### Yo Han Jang1† and Baik Lin Seong1,2 \*

*<sup>1</sup> Molecular Medicine Laboratory, Department of Biotechnology, College of Life Science and Biotechnology, Yonsei University, Seoul, South Korea, <sup>2</sup> Vaccine Translational Research Center, Yonsei University, Seoul, South Korea*

There is an unmet public health need for a universal influenza vaccine (UIV) to provide broad and durable protection from influenza virus infections. The identification of broadly protective antibodies and cross-reactive T cells directed to influenza viral targets present a promising prospect for the development of a UIV. Multiple targets for cross-protection have been identified in the stalk and head of hemagglutinin (HA) to develop a UIV. Recently, neuraminidase (NA) has received significant attention as a critical component for increasing the breadth of protection. The HA stalk-based approaches have shown promising results of broader protection in animal studies, and their feasibility in humans are being evaluated in clinical trials. Mucosal immune responses and cross-reactive T cell immunity across influenza A and B viruses intrinsic to live attenuated influenza vaccine (LAIV) have emerged as essential features to be incorporated into a UIV. Complementing the weakness of the stand-alone approaches, prime-boost vaccination combining HA stalk, and LAIV is under clinical evaluation, with the aim to increase the efficacy and broaden the spectrum of protection. Preexisting immunity in humans established by prior exposure to influenza viruses may affect the hierarchy and magnitude of immune responses elicited by an influenza vaccine, limiting the interpretation of preclinical data based on naive animals, necessitating human challenge studies. A consensus is yet to be achieved on the spectrum of protection, efficacy, target population, and duration of protection to define a "universal" vaccine. This review discusses the recent advancements in the development of UIVs, rationales behind cross-protection and vaccine designs, and challenges faced in obtaining balanced protection potency, a wide spectrum of protection, and safety relevant to UIVs.

#### Keywords: influenza virus, universal influenza vaccine, cross-protection, HA stalk, M2e, T cell, live attenuated influenza vaccine

## INTRODUCTION

Influenza viruses present a high level of antigenic diversity and variability due to their segmented RNA genome. These viruses are classified into four major types, A, B, C, and D, based on their nucleoprotein (NP) and matrix (M) genes. Human infecting type A and B viruses are further classified into multiple subtypes or lineages, respectively, depending on the antigenicity of viral surface proteins, hemagglutinin (HA), and neuraminidase (NA) genes (Paules and Subbarao, 2017). Influenza A and B viruses co-circulate in every season and, thus, c represents the primary targets of seasonal influenza vaccines (Sridhar et al., 2015). In addition to seasonal epidemics, influenza viruses have caused pandemics at the intervals

#### Edited by:

*Yanmin Wan, Fudan University, China*

#### Reviewed by:

*Peter Palese, Icahn School of Medicine at Mount Sinai, United States Florian Krammer, Icahn School of Medicine at Mount Sinai, United States*

#### \*Correspondence:

*Baik Lin Seong blseong@yonsei.ac.kr*

#### †Present address:

*Yo Han Jang, Department of Biological Sciences and Biotechnology Major in Bio-Vaccine engineering, Andong National University, Andong, South Korea*

#### Specialty section:

*This article was submitted to Clinical Microbiology, a section of the journal Frontiers in Cellular and Infection Microbiology*

Received: *24 May 2019* Accepted: *24 September 2019* Published: *10 October 2019*

#### Citation:

*Jang YH and Seong BL (2019) The Quest for a Truly Universal Influenza Vaccine. Front. Cell. Infect. Microbiol. 9:344. doi: 10.3389/fcimb.2019.00344* of ∼10–40 years since the 1918 Spanish flu H1N1, the 2009 pandemic H1N1 being the last outbreak (Saunders-Hastings and Krewski, 2016). While vaccination remains the most cost-effective measure to prevent influenza virus infections, antigenic drift in the surface antigens allows these viruses to escape antibody-mediated neutralization (Kim et al., 2018a). In addition, the sudden occurrence of pandemics is often accompanied by zoonotic spillover of the surface genes into the human-infecting viruses, rendering preexisting vaccines ineffective to newly emerging viruses. The variation caused by genetic drift and shift is unpredictable, posing a serious challenge to the management of influenza outbreak. Based on the amino acid sequences of HAs, influenza A viruses (IAVs) are divided into two phylogenetic groups. The IAV HA group 1 viruses include H1, H2, H5, H6, H8, H9, H11, H12, H13, H16, H17, and H18, while the group 2 viruses comprise H3, H4, H7, H10, H14, H15 (**Figure 1**). The NAs of IAVs are also antigenically diverse, presenting two distinct groups. Influenza B viruses (IBVs) are not divided into subtypes but circulate as two distinct Yamagata-like and Victoria-like lineages. Influenza C viruses (ICVs) generally cause a mild respiratory disease in humans and do not cause epidemics (Dykes et al., 1980). Contemporary influenza epidemics are caused by the H1N1 and H3N2 of the IAVs and one or two lineages of the IBVs, dictating trivalent (TIV) or quadrivalent influenza vaccine (QIV) containing two IAV antigens and one or two IBV antigens, respectively (Ambrose and Levin, 2012).

Many strategies have been undertaken on the pursuit of developing a universal influenza vaccine (UIV) (Paules et al., 2017). The induction of cross-protective immune responses directed toward conserved B cell or T cell epitopes is a major principle underlying broad protection (**Figure 2**). The direct binding of antibodies to the viral surface proteins interferes with their functions and results in virus neutralization before cell entry (**Figure 2A**). Alternatively, the antibodies may bind to viral antigens displayed on the surface of virus-infected cells and mediate effector functions to remove the infected cells. A cytotoxic T lymphocyte (CTL) can kill the virus-infected cells in an independent manner (**Figure 2B**). Several conserved viral antigens such as M2 extracellular domain (M2e), HA stalk and receptor binding site, NA, and T cell epitopes in the internal proteins such as polymerase basic protein 1 (PB1), NP, and M1 were defined as targets for eliciting cross-reactive immune response. A variety of vaccine platforms were tested for effective exposure of those antigens to the immune system (Wang et al., 2018a; Estrada and Schultz-Cherry, 2019). While these vaccines showed broader cross-protection than the classical inactivated influenza vaccines (IIVs), their protection potency was weak. They provided partial protection against the antigenically distant viruses, and the protection breadth is usually limited to the same group (subtype-specific or HA group-specific) and not effective in another HA group. Thus, there exists a considerable gap between the current status, and the ultimate goal, of a truly universal vaccine.

Some studies have discovered rare antibodies specific to the conserved HA stalk in animals and humans with prior exposure to influenza viruses. Some of these antibodies show extremely broad specificity, encompassing both HA groups of IAVs (Ekiert et al., 2011) or even both IAVs and IBVs (Dreyfus et al., 2012), representing an optimistic prospect of developing a UIV. However, no studies so far have reported a successful vaccine approach that induces such antibodies to a protective level, reflecting considerable difficulty in selectively inducting particular antibodies in the context of vaccination. While antibody-dependent effector mechanisms and T cell immunity have emerged as potential correlates of broad protection against influenza infections (Krammer and Palese, 2015), the exact molecular mechanisms of their protective action are not completely understood, and immunopathology by T cell responses still remains a challenge (Peiris et al., 2010; Duan and Thomas, 2016; Erbelding et al., 2018). Moreover, understanding how preexisting immunity (immune imprinting or antigenic sinlike phenomenon) shaped by prior exposure to influenza viruses affects the magnitude, hierarchy, and sustainability of antibody response to vaccination is suggested as critical for designing a UIV in humans (Henry et al., 2018). The involvement of multiple factors in eliciting broad protection and the influence of pre-existing immunity on the subsequent vaccination pose a dilemma on establishing the correlate of cross-protection against heterologous or hetero-subtypic viruses, representing a critical obstacle for the licensure of a UIV by the regulatory authorities.

This review updates the recent advances in UIV development and focuses on the critical issues to be addressed in designing a 'truly' UIV. Alternative vaccine antigens and vaccine strategies for durable and broader protection are also discussed in detail. This review is meant to help the readers to acquire general information on the cutting edge of UIV researches and to gain wide perspectives on rational design of a truly UIV with improved potency, breadth, and safety.

## CRITERIA FOR A UIV

### Definition of Universal Protection

Although there is a clear consensus about the urgent need for a vaccine that provides durable and broad protection against multiple strains of influenza virus, the definition of the "universality" of UIV is still debated (Krammer et al., 2018b). Recently, the National Institute of Allergy and Infectious Diseases (NIAID) held a meeting to identify and develop the criteria to define a UIV. The participants from multiple disciplines agreed that a reasonable UIV should be at least 75% effective against symptomatic influenza virus infection caused by group 1 and group 2 IAVs, with the protection lasting over 1 year for all age groups (Paules et al., 2017; Erbelding et al., 2018) (**Table 1**). Similar criteria have also been suggested as preferred product characteristics (PPC) and target product profiles (TPP), describing the desired characteristics of a UIV, by the World Health Organization (WHO) and the Bill and Melinda Gates Foundation (BMGF), respectively (**Table 1**). Despite the priority given to IAVs in WHO PPC (**Table 1**), the IBVs represent a "low hanging fruit" because of their low antigenic diversity and the lack of animal reservoir, presenting credence to potential eradication from humans (Tan et al., 2018). The proposed consensus to define universal protection

FIGURE 1 | Phylogenetic trees representing HA and NA diversity among influenza viruses. The 18 subtypes of HAs of IAV are divided into two phylogenetic groups according to their amino acid sequences similarities. The HAs of IBVs are divided into Victoria-like and Yamagata-like lineages but they are closer to each other than any of two different subtypes of IAVs. The HAs of ICVs are antigenically distant from those of IAVs and IBVs. The NAs of IAVs also show high levels of antigenic variability and are divided into two groups. Phylogenetic trees were constructed based on amino acid sequence comparisons among influenza viruses. Multiple alignments were carried out using the representative sequence of each HA or NA subtype or lineage. The phylogenetic trees were constructed by the ClustalW algorithm using neighbor joining (N-J) method and visualized by FigTree v1.4.4. The scale bars represent amino acid change (%).

offers a valuable guideline to develop an effective and safe UIV. Ideally, the spectrum of protection may covers both IAVs and IBVs, considering the existence of extremely broadly protective antibodies and T cell epitopes across IAVs and IBVs from animals and humans (Corti et al., 2011; Koutsakos et al., 2019). However, as the antigenic difference between a vaccine and a target virus becomes larger, the number of conserved B cell or T cell epitopes between the two strains decreases. The limited availability of target epitopes may result in compromised protection robustness due to (1) decreased clonal diversity of cross-reactive antibodies

#### TABLE 1 | Consensus criteria for a UIV.


*NIAID, National Institute of Allergy and Infectious Diseases (https://www.niaid.nih.gov/diseases-conditions/universal-influenza-vaccine-research); BMGF, Bill and Melinda Gates Foundation (https://gcgh.grandchallenges.org/challenges). WHO, World Health Organization (https://apps.who.int/iris/bitstream/handle/10665/258767/9789241512466-eng.pdf; https://jsessionid=37D9E056C3EA58A90EE237432AA7D65F?sequence=1).*

and T cells, or (2) occurrence of escape mutants by even small amount of genetic mutations in the epitopes. Therefore, a balanced breadth and robustness of cross-protection should be considered when developing a reliable UIV (**Figure 3**).

The NIAID and the researchers have proposed three research areas to address the knowledge gaps in developing a UIV; (1) understanding of influenza transmission, natural history, and viral pathogenesis, (2) characterization of correlates of protection, and (3) rational design of a UIV to improve potency and breadth of protection (Erbelding et al., 2018). In particular, the correlates of protection elicited by a UIV may vary considerably in quantity and quality, depending on the vaccine type used and the target influenza virus tested. For instance, while the protection potency of HA stalk-based vaccine may easily be evaluated by measuring the neutralizing activity or indirect effector mechanisms by the stalk-reactive antibodies (Jegaskanda et al., 2017b), such correlates of protection cannot be used to evaluate the protective efficacy of a T cell epitope-based vaccine containing non-HA epitopes. Moreover, when using HA stalk-based vaccines, high levels of stalk-reactive antibodies represent a good protective efficacy against the same group IAVs (**Figure 1**). However, the binding affinities of HA stalk antibodies are variable among different viruses, and therefore may require different antibody titers to exhibit sufficient protection against the viruses. This speculation is supported by the observations that broadly reactive HA stalk antibodies show considerably different neutralizing abilities and binding affinities to the viruses within the same HA group (Ekiert et al., 2011, 2012). Furthermore, HA stalk-based vaccine and M2e-based vaccine may demonstrate a different balance of protective abilities to each other, between direct neutralization and indirect effector mechanisms (ADCC, for instance) by the respective antibodies. It has been shown that M2e antibodies provide protection via FcR-dependent effector functions rather than direct virus neutralization (Deng et al., 2015a), whereas HA stalk antibodies exert both direct neutralization and indirect effector functions (Krammer and Palese, 2015). The antibody-mediated inhibition of virus attachment measured by hemagglutinin inhibition assay or microneutralization assay is the gold standard for seasonal influenza vaccines (Ohmit et al., 2011). However, the assay cannot be applied simply to reflect the cross-protection by a UIV against heterologous/heterosubtypic influenza viruses. Therefore, it is essential to develop mechanistically distinct in vitro or in vivo assays to measure potential correlates of protection in order to evaluate the protection potency and breadth of the vaccine.

#### Mode of Protection by a UIV

The cornerstone of developing a UIV is the determination of the precise protection mechanisms of immune response against influenza viruses. Influenza HA recognizes sialic acid on the cellular receptors and initiates infection by entering the cell via receptor-mediated endocytosis (**Figure 4**). While HA inhibitory (HAI) antibodies have long been considered as the gold standard for strain-specific protection, very few of them were shown to elicit a broad protection by binding to the conserved receptor-binding site (RBS) of HA, thereby preventing viral entry to the cell (Krause et al., 2011; Ekiert et al., 2012). Recently, multifunctional protection mechanisms have been described for HA stalk-reactive antibodies. It has been shown that HA stalk antibodies may inhibit membrane fusion, the release of viral genome into the cytoplasm of the cell, and maturation of the HA precursor (Krammer and Palese, 2015). Furthermore, HA stalk antibodies can induce antibodydependent effector functions such as antibody-dependent cellular cytotoxicity (ADCC), antibody-dependent cellular phagocytosis (ADCP), and complement-dependent cytolysis (CDC), resulting in clearance of virus-infected cells by the immune cells or the complement system (Jegaskanda et al., 2017b). During viral budding, NA cleaves the sialic acid from HA and supports multiple infection cycles by release of the newly assembled viral particles. NA inhibitory (NAI) antibodies specific to the conserved regions have shown an exceptional breadth, inhibiting divergent influenza viruses (Chen et al., 2018). In addition to the broadly protective antibodies, T cell immunity against conserved viral internal proteins also provides a broad protection. Crossreactive cytotoxic T lymphocytes (CTLs) recognize the viral epitopes presented on MHC molecules and kill the infected cells. It is noteworthy that the cross-reactivity of T cell immunity has been recently shown to cover both IAVs and IBVs, and even the ICVs (Koutsakos et al., 2019), although its protective role in vivo has not been confirmed.

vaccine provides only strain-specific protection against well-matched strains. Recently, many efforts have been put in order to improve the protection breadth of a vaccine from subtype-specific protection to, ideally, pan-influenza universal protection [adopted from Erbelding et al. (2018)].

FIGURE 4 | Protection mode of action afforded by a UIV. Antibodies against the HA globular head domain inhibit viral attachment via HA-mediated receptor binding to the sialic acid on cellular receptors (a). HA stalk antibodies have multiple protective functions. As the virus enters the cell, pre-bound stalk antibodies prevent the fusion of viral and endosomal membranes and block the viral genome release into cytoplasm of the cell (b). Binding of stalk antibodies can also limit the access of cellular proteases to the cleavage site located in the stalk domain and inhibit the cleavage and subsequent conformational change of HA that is an essential step for acquiring viral infectivity (c). Different antibodies against HA stalk and also other viral proteins such as NA, M2, and NP are shown to mediate antibody-dependent effector functions such as antibody-dependent cellular cytotoxicity (ADCC), antibody-dependent cellular phagocytosis (ADCP), and complement-dependent cytotoxicity (CDC), leading to the lysis of the virus-infected cells by immune cells or complement system (d–f). NA antibodies inhibit receptor destroying activity of NA and prevent the budding of newly formed viral particles from the cells (g). Cytotoxic T lymphocytes (CTLs) recognize the viral peptide presented on MHC-I molecule and kill the virus infected cell by the secretion of cytotoxic granules and cytokines (h).

## CURRENT STATUS OF M2e-BASED UIV APPROACHES

IAVs have two major surface proteins, HA and NA, and one minor surface protein, the M2 ion channel. During the infection cycle, the M2 ion channel is responsible for acidification of the viral interior, facilitating virus uncoating, and unloading of viral ribonucleoproteins (RNPs) into the host cytoplasm (Pinto et al., 1992). The extracellular domain of M2 protein (M2e) consists of 24 amino acids, among which 9 N-terminal amino acids are completely conserved among all IAVs (H1–H18) and minor mutations are observed in the distal portion (Deng et al., 2015a; Tsybalova et al., 2018). Therefore, M2e is considered as a promising target for eliciting broadly reactive antibodies. However, due to poor immunogenicity of the small M2e region, UIV approaches targeting M2e required carriers, vectors, and adjuvants to enhance immune responses (Deng et al., 2015a; Lee et al., 2015). One of the most efficient approaches is to make virus-like particles (VLPs) which display M2e on their surface. It has been shown that the hepatitis B virus core (HBc) protein fused with M2e self-assembled into VLPs that resemble wild type virus particles, expressing M2e on the surface (Neirynck et al., 1999). Following this work, substantial efforts were made to improve the immunogenicity and protective efficacy of M2e-HBc VLP vaccine constructs. Major strategies include the use of multiple copies of M2e by tandem repeats (Ravin et al., 2015; Sun et al., 2015; Tsybalova et al., 2015) and combination with adjuvants such as cholera toxin A1 (De Filette et al., 2006) or B subunit of Escherichia coli heat labile enterotoxin (LTB) (Zhang et al., 2009). The M2e-HBc VLP was further harnessed with T cell immunity by combining conserved T cell epitope from NP (Gao et al., 2013). The arginine-rich domain of HBc was shown to induce Th1-biased immune response of M2e-HBc VLP by binding to RNA, leading to improved protection (Ibanez et al., 2013). Besides the HBc, M2e was fused with various coat or capsid proteins derived from other viruses including Malva mosaic virus (Leclerc et al., 2013), tobacco mosaic virus (Petukhova et al., 2013), Papaya mosaic virus (Denis et al., 2008), T7 bacteriophage (Hashemi et al., 2012), and RNA phage Qβ (Bessa et al., 2008). Moreover, the enveloped VLPs displayed M2e in a membrane-anchored manner by combining influenza matrix protein and transmembrane (TM) domain of HA fused into M2e. The Sf9 insect cells infected with baculoviral vectors expressing influenza M1 and M2e-TM of influenza HA produced influenza VLPs displaying much higher levels of M2e than the wild type virions (Kim et al., 2015, 2017). Other vaccine types, including DNA vaccines or recombinant protein vaccines expressing M2e with various carriers to enhance immune responses, have also been advanced (Deng et al., 2015a).

Many studies conducted on animals and humans have elucidated the mechanisms of cross-protection conferred by M2e-based vaccines. A common observation is that antibodies specific to M2e cannot directly neutralize the viruses but confer cross-protection by eliciting several mechanisms of antibodymediated and cell-mediated immune responses. The most wellcharacterized protection mechanisms include ADCC, ADCP, and CDC (**Figure 4**). Studies have shown that multiple types of immune cells including the natural killer (NK) cells, neutrophils, dendritic cells, or macrophages have the ability to sense the virusinfected cells by interaction with the Fc receptors (FcR) and the Fc region of M2e antibodies. This results in killing (ADCC) or phagocytosis (ADCP) of the infected cells (Huber et al., 2001; Jegerlehner et al., 2004; Hashimoto et al., 2007; Guilliams et al., 2014). The complement is also known to bind to the Fc of M2e antibodies and triggers the complement cascade, leading to formation of the membrane attack complex and target cell lysis (CDC) (Wang et al., 2008; Kim et al., 2018b). These studies together suggest that M2e-specific non-neutralizing antibodies play a crucial role in broad protection against IAVs through the clearance of the virus-infected cells. Besides non-neutralizing effector functions, it has also been shown that M2e-displaying recombinant bacteriophages induce HAI antibodies in a mouse model (Deng et al., 2015b). In addition, the T cell responses directed to M2e have been shown to contribute to crossprotection. Several studies have shown the presence of CD4 and CD8 T cell epitopes in M2e region in mice and humans (Jameson et al., 1998; Gianfrani et al., 2000; Mozdzanowska et al., 2003; Eliasson et al., 2008). In line with this, it has been demonstrated that CD4 T cell and CD8 T cell-mediated immunity are critical for cross-protection elicited by M2e-based UIVs including VLPs or recombinant peptide vaccines (Lee et al., 2014; Zhang et al., 2016; Arevalo et al., 2017; Pendzialek et al., 2017; Eliasson et al., 2018).

It remains a challenge to overcome the low protective efficacy of M2e-based vaccines, due to intrinsic low immunogenicity of the vaccine antigen and the low abundance of M2 proteins on influenza virions and infected cells. From the clinical and practical standpoints, multiple immunizations with a high dose of vaccine antigens combined with potent adjuvants may pose safety concerns, rendering M2e-based vaccine approaches inadequate as a stand-alone UIV. Consequently, M2e antigen was combined with other influenza viral antigens (mostly HA) or supplemented with strain-specific vaccines, resulting in improved potency, and broader cross-protection. For instance, several studies combined M2e with influenza HA head or stalk domain to elicit broadly protective immunity against heterologous influenza infections with various vaccine platforms including nanoparticles, recombinant proteins, and recombinant influenza viruses (Guo et al., 2017; Bernasconi et al., 2018; Deng et al., 2018; Tsybalova et al., 2018). Supplemented vaccination with M2e was examined to overcome the strain-specific protection or poor immunogenicity of classical inactivated or split vaccines (Music et al., 2016; Song et al., 2016). Besides the approaches described above, a vast amount of research on M2e-based UIVs have accumulated in the last decade, which are described in specialized reviews (Deng et al., 2015a).

## HA STALK-BASED UIV APPROACHES

## General Principles of HA Stalk-Based Approaches

The HA stalk-based approach currently leads the mainstream of UIV development. Influenza HA is the primary target antigen of the currently licensed seasonal vaccine, and HAI antibodies serve as the "gold standard" when evaluating the protective Jang and Seong Universal Influenza Vaccine

efficacy of HA-based vaccines (Ohmit et al., 2011). However, HA-based vaccines provide only strain-specific protection to antigenically homologous viruses, necessitating annual update of the surface antigens of seasonal vaccines to match the circulating viral strains. HA comprises two distinct domains; globular head, and stalk domains. The head domain is highly variable and immunologically dominant and thus the majority of HI antibodies are directed to the head domain that harbors the RBS (**Figure 5**). In contrast, the HA stalk domain is relatively conserved among the viruses, and therefore eliciting antibodies specific to the stalk is considered as a key to developing a UIV. However, the stalk is generally regarded as immunologically subdominant, due to masking effect by bulky head domain and close proximity to the viral membrane (Krammer and Palese, 2013; Crowe, 2018; Krammer, 2019), which necessitates rational design of a HA antigen and novel vaccination strategies to increase stalk-reactive antibodies. The major protective function of HA stalk antibodies is to lock the HA trimer in a prefusion state. This prevents pH-dependent conformational change of HA that triggers the membrane fusion and release of viral genome into cytoplasm of host cell (Ekiert et al., 2009). The membrane fusion inhibitory activity of the stalk antibodies may lead to neutralization of influenza viruses, although the neutralizing potency is weaker than head antibodies that directly prevent the receptor binding. Besides the neutralizing activity, stalk antibodies have multiple indirect mechanisms that contribute to broad protection, including antibody-dependent effector mechanisms (Jegaskanda et al., 2017b) and the inhibition of the NA enzymatic activity (Wohlbold et al., 2016; Chen et al., 2019). To overcome the low immunogenicity of HA stalk, genetic modifications were required for efficient exposure of stalk to the host immune system. Prime-boost vaccination with chimeric HAs or "headless" HAs were designed to boost stalk-reactive antibodies (**Figure 5**). Alternatively, hyperglycosylation in HA head resulted in the redirection of immune response from head to stalk (Krammer and Palese, 2015) (**Figure 5**). These approaches have paved the way to increase the breadth of protection against influenza viruses.

## Protection Mechanisms of HA Stalk-Based Vaccine

A primary function of stalk-reactive antibodies is membrane fusion inhibition. Following the influenza virus entry into the cells via endocytosis, a pH-dependent conformational change of HA proteins triggers the membrane fusion between the virus and endosome, leading to release of the viral genome into the cytoplasm. Binding of the antibody to the stalk interrupts the conformational change of HA and prevents subsequent membrane fusion. This leads to the trapping of the virus in the endosome, eventually aborting the infection (Ekiert et al., 2009; Sui et al., 2009). Moreover, binding of the stalk antibody blocks the access of proteases to the cleavage site located in stalk region and prevents protease-dependent cleavage of the HA precursor into HA1 and HA2 subunits, which is a prerequisite for the conformational change of HA (Ekiert et al., 2009; Brandenburg et al., 2013). These two overlapping functions of stalk-reactive antibodies result in inhibition of membrane fusion and direct neutralization of the virus. In addition to these mechanisms that operate at the early stage of infection, the stalk antibody may also bind to the newly expressed HA proteins on the cell surface, and interfere with the viral budding or release, at the later stage of infection (Tan et al., 2014).

In addition to direct neutralization, stalk antibodies have indirect mechanisms that involve diverse innate immune cells to clear the virus-infected cells from the host. The NK cells sense virus-infected cells via interaction between FcR and Fc portion of antibody bound to HA stalk expressed on the cell surface, and kill the infected cells by releasing cytotoxic granules and antiviral cytokines, via ADCC mechanism (DiLillo et al., 2014) (**Figure 4**). The FcR-Fc interaction is also required for ADCP, by which macrophages or neutrophils recognize and engulf antibodybound influenza viral particles or infected cells (Ana-Sosa-Batiz et al., 2016). In addition, stalk antibodies bound to virus-infected cells are also able to activate the complement system, leading to the lysis of virus-infected cells (CDC) (Terajima et al., 2011). Animal challenge models have demonstrated broad protection by passive immunization with HA stalk monoclonal antibodies (DiLillo et al., 2014, 2016). The stalk-reactive antibodies also inhibit NA enzymatic activity through steric hindrance, limiting NA access to the HA-bound sialic acid, thereby preventing viral egress (Wohlbold et al., 2016; Chen et al., 2019; Kosik et al., 2019). The failure to release newly assembled virions from the infected cells prevents the viruses from entering into multiple cycles of infection. Overall, the stalk antibodies confer a broad protection from influenza virus infection through multiple mechanisms affecting both HA and NA throughout the infection cycle of influenza virus.

#### HA Stalk-Based Vaccine Constructs

The first reported HA stalk-based UIV construct is the headless HA, a deletion mutant of HA lacking the globular head region (Graves et al., 1983; Sagawa et al., 1996; Steel et al., 2010) (**Figure 5**). While the initial strategies demonstrated broadly protective potentials in animal models, the removal of the head region often led to misfolding of the stalk immunogen, rendering the critical neutralizing epitopes ineffective. Subsequent studies were conducted to enhance the structural integrity of the stalk immunogens by rational designs to mimic the native trimeric conformation (Lu et al., 2014; Mallajosyula et al., 2014; Impagliazzo et al., 2015). Other studies designed and evaluated nanoparticle structures consisting of the stalk antigens for broad protection in mice and ferret models (Yassine et al., 2015; Corbett et al., 2019). The second approach to preferably induce stalkreactive antibodies is based on chimeric HA that consists of stalk domain derived from the major target strains such as H1, H3, or B viruses fused with head domain from irrelevant avian strains such as H5, H6, or H8 (Krammer and Palese, 2015) (**Figure 5**). Multiple immunization was required to increase the stalk antibodies sufficient to provide protection from viruses within the same HA group (Hai et al., 2012; Krammer et al., 2013, 2014). The chimeric HA approach involves a full-length functional HA protein and, therefore, it may be expressed in genetically engineered viruses of the wild type or live attenuated

influenza viruses (Krammer et al., 2013; Isakova-Sivak et al., 2018). Third approach relies on hyperglycosylation in the HA head domain (**Figure 5**). The introduction of additional Nglycosylation sites into the immunodominant head domain shields the major antigenic sites in the head and redirects the host immune responses toward the immune-subdominant stalk. The hyperglycosylated HA antigens induced higher stalk antibodies and provided better protection than the wild type HA in a mouse model (Eggink et al., 2014). Similar approaches were tested for avian H5N1 vaccines, in which hyperglycosylated HA antigens were delivered in various vaccine formats such as DNA, recombinant protein, VLP, and adenoviral vector (Lin et al., 2012, 2014). A recent study has shown that glycan shielding of HA head resulted in immune focus on a conserved epitope occluded

(HA of A/swine/Iowa/15/1930, PDB ID 1RUY) and the final image was produced by PyMOL program).

at the head interface, with Fc-dependent protection activity in mice (Bajic et al., 2019a). These results together suggest that hyperglycosylation in the HA head is a promising strategy to find novel target epitopes hiding in the head and stalk in HA.

#### Current Issues in HA Stalk-Based UIVs Vaccine Efficacy

As discussed above, the protective efficacy of HA-stalk based vaccines is relatively weak, and multiple boost immunizations are required for efficient protection (Krammer et al., 2013, 2014). In addition, the breadth of protection is within the same HA group rather than both the groups (Krammer et al., 2013). To complement the low efficacy, prime-boost approaches that entail sequential immunization by LAIV followed by inactivated virus

antigenic sites in head domain results in the redirection of antibody responses toward stalk domain. The HA structure was downloaded from the Protein Data Bank

containing chimeric HAs are being evaluated. As examined by a ferret challenge model, this approach provided potent cross protection (Nachbagauer et al., 2017). It was observed that there is a "disconnect" between stalk antibody titers in serum and the protection level (Nachbagauer et al., 2017). The results show that the broad protection observed in the ferret model was primarily mediated by multiple-level immune responses by LAIV and rather than the stalk antibodies (see section Live attenuated influenza vaccine as an alternative strategy below). This observation in preclinical model leads to the question of the relevance of HA stalk antibodies to cross-protection in humans. Studies with human infection model have led to a somewhat different interpretation on the protective role of the stalk antibodies, depending on the study design. A human challenge study showed that the naturally occurring stalk antibody titers were predictive of lowering viral shedding, but demonstrated poor correlation with the reduction of clinical symptoms upon the pandemic A/H1N1 challenge (Park et al., 2018). In contrast, a more recent study suggested the preexisting HA stalk antibodies as an independent correlate of protection against natural infection with the pandemic A/H1N1 virus (Ng et al., 2019). Thus, further studies are required for a conclusive demonstration of the breadth and efficacy of cross-protection offered by HA stalk antibodies in human challenge models.

#### Potential Vaccine Safety Issues

Although the HA stalk based strategies have shown promising broad protection in animal models, potential safety issues have been raised (Crowe, 2018; Khurana, 2018). It is suggested that the stalk antibodies are responsible for antibody-dependent enhancement of viral infectivity in a swine model. In a study, pigs vaccinated with inactivated H1N2 vaccine developed more severe respiratory diseases upon heterologous H1N1 challenge as compared to the non-vaccinated control, which was ascribed to infectivity-enhancing effects of stalk antibodies (Khurana et al., 2013). Similarly, the HA subunit vaccine resulted in vaccineassociated enhanced respiratory disease (VAERD) in pigs after heterologous challenge (Gauger et al., 2011; Rajao et al., 2014). This effect cannot be generalized as the enhancement was noticed only in the adjuvanted antigen in a swine model, but not seen in similar studies with non-adjuvanted vaccines in a ferret model (Krammer et al., 2014). In addition, it has been recently shown that head-reactive, non-neutralizing monoclonal antibodies increase the stalk flexibility and membrane fusion kinetics, resulting in enhanced respiratory disease in a mouse model (Winarski et al., 2019). These results indicate that head-reactive antibodies may also induce antibody-dependent enhancement. However, similar observations in humans are debatable. For instance, humans immunized with 2008–2009 inactivated TIV exhibited an increase in illness following infection with 2009 H1N1 pandemic virus (Janjua et al., 2010; Skowronski et al., 2010) due to the presence of cross-reactive antibodies (Monsalvo et al., 2011). Similar fatal cases were observed during 1957 H2N2 pandemic, although the relation of HA stalk antibodies to these observations is not known. It is noteworthy that NAI antibodies could reduce VAERD caused by mismatched heterologous HA, suggesting that vaccines which target HA protein alone may be prone to VAERD and cross-protective NAI antibodies may counteract VAERD (Rajao et al., 2016). Although there is no report on VAERD associated with the headless or chimeric HA vaccines, careful monitoring is advised as these approaches rely heavily on stalk antibodies. The inclusion of NA antigens into HA stalk-based vaccine merits further evaluation. Another issue raised by a recent study is the auto-reactivity of stalk antibodies to human proteins, which is significantly higher than the head antibodies, as confirmed by multiple in vitro assays (Bajic et al., 2019b). Although biological implications of this observation have not been studied in vivo, it was speculated that vaccine strategies focused exclusively on the stalk, or any single conserved epitope may fail to induce adequate antibody titers due to negative selection of the auto-reactive B cell clones (Bajic et al., 2019b). However, clinical observations showed that, despite under potential negative selection, preexisting stalk antibodies conferred protection against the 2009 pandemic A/H1N1 infection in humans (Ng et al., 2019).

#### Stalk Variability

Although conserved among the influenza viruses, the HA stalk is not antigenically identical, showing sequence variability even within the same HA subtype. In line with this, a study demonstrated that H1 HA stalk could undergo mutations in vitro by immune pressure, although the variability was less than the head region (Anderson et al., 2017). On the other hand, another study show that while the stalk domain does evolve over time, this evolution is slow and, historically, is not directed to aid in evading neutralizing antibody responses (Kirkpatrick et al., 2018). Several studies have isolated mutant influenza viruses showing resistance to stalk antibodies (Ekiert et al., 2011; Friesen et al., 2014; Chai et al., 2016). Notably, all isolated resistant viruses were seen to harbor three different mutations (Gln387Lys, Asp391Tyr, and Asp391Gly) in stalk region. While Gln387Lys mutation completely abolished the binding of the antibody to stalk region, the other two mutations rarely affected the antibody binding but enhanced the fusion ability of HA, representing two independent resistance mechanisms of the virus to escape stalk-reactive antibodies (Chai et al., 2016). These reports together show that the stalk may undergo natural or directed antigenic changes, asking important considerations in the context of developing stalk-based UIVs. Especially, the occurrence of mutations that enhance the membrane fusion of HA in presence of stalk antibodies presents a potential concern to vaccine safety issues. It should, however, be noted that the stalk antibody escape mutants tend to lose a viral fitness and are highly attenuated in vivo (Henry Dunand et al., 2015; Chai et al., 2016), alleviating the safety concerns.

#### Rational Design of Stalk

It is well-recognized that the stalk is less immunogenic than head and that the stalk antibodies are less potent in virus neutralization than the head antibodies. Therefore, multiple immunizations (three or four times) with stalk-based vaccines are required for inducing a sufficiently protective level of antibody response. More importantly, most humans have a diverse history of exposure to influenza antigens, by natural

infection, or vaccination. The established immune memory influences the subsequent immune response to a UIV, posing a great challenge to rendering qualitatively uniform and protective immune responses in humans. As discussed earlier, the HA stalk-based vaccines elicit broad-spectrum protection within the same HA group, but usually fail to provide protection to strains belonging to different HA group (Krammer et al., 2013; Margine et al., 2013). These results apparently dissent from the finding that some of the stalk monoclonal antibodies isolated from humans recognize all IAV subtypes, neutralizing both group 1 and group 2 viruses, presenting a promising prospect of developing pan-influenza A therapeutic solution (Corti et al., 2011). Isolation of rare antibodies with extremely broad neutralization potency from humans with prior vaccinations or infections (Corti et al., 2011) indicates that our immune system is able to find the cryptic and conserved epitopes and generate specific antibodies to the regions. However, inducing such antibodies to a protective level by vaccination remains a big challenge. Some recent reports encourage other options such as the activation of the conserved "cryptic" epitopes via antigen processing mechanisms (Lee et al., 2018b) or deliberately downregulating surface antigens (Yang et al., 2013). Moreover, Fcengineering technologies developed to enhance the therapeutic efficacy of antibodies may be harnessed to modulate FcR-Fc mediated effector functions (DiLillo et al., 2014) to achieve broad protection. A comprehensive understanding of immune response to broadly neutralizing epitopes and structure-based antigen design is required for the rational design of a pan-influenza A vaccine covering both group 1 and group 2 IAVs, as a consensus criterion of UIV (**Table 1**).

#### CONSERVED TARGETS IN HA OTHER THAN STALK

The inability to induce a complete protection by HA stalkbased approach led to a search for alternative targets for UIVs. Human monoclonal antibody CH65, mimicking the interaction with sialic acid, was shown to bind to the RBS in HA and neutralize multiple H1N1 influenza strains (Whittle et al., 2011). A caveat is that this antibody demonstrated stringent structural requirements for neutralization and mutation at the binding region led to generation of escape mutants. A panel of headreactive monoclonal antibodies was also isolated and shown to recognize conserved region in the RBS and neutralize multiple influenza viruses (Krause et al., 2011, 2012; Ekiert et al., 2012; Benjamin et al., 2014).

The HA head domain was also reported to contain conserved epitopes outside the RBS. Different broadly neutralizing antibodies recognized the conserved regions located in HA head domain and neutralized multiple influenza viruses without detectable HI activity (Iba et al., 2014; Raymond et al., 2018). Generally, neutralization breadth of head antibodies was considerably variable to each other, ranging from pan-subtype (covering the same subtype) to pan-type (covering both group 1 and 2) coverage, indicating the presence of variably conserved regions. Additionally, a novel class of cross-reactive antibodies was discovered in humans vaccinated with seasonal TIV (Lee et al., 2016). These antibodies were shown to bind to a highly conserved region located on the HA RBS that was occluded in the HA trimer, conferring protective immunity against H1N1 and H3N2 strains in vivo, without neutralizing activity in vitro. Furthermore, a novel class of antibodies targeting vestigial esterase (VE) domain in HA has been characterized. The VE domain consists of two non-continuous sequences in HA head domain, which together forms a structurally distinct subdomain from the RBS and HA stalk domain (Zheng et al., 2018). The VE domain of the HEF protein of ICV is responsible for cleaving the host receptor to facilitate viral release, whereas in IAVs and IBVs the same receptor cleaving function is provided by a separate NA. The VE domains are found in both IAVs and IBVs, although their functions are not well-defined. The VE domains are highly conserved within a subtype of IAV but are variable among different subtypes (Ha et al., 2002). Monoclonal antibodies to the VE domain of H5N1 virus demonstrated broad neutralization against multiple clades of H5N1 subtype by preventing viral entry into cells (Oh et al., 2010). To date, different monoclonal antibodies have been isolated and they bind to different epitopes in the VE domain of H5N1 viruses, suggesting the presence of multiple neutralization epitopes in the VE domain (Paul et al., 2017). The VE-binding antibodies are reported to mediate ADCC for in vivo protection via FcR-Fc interaction (Wang et al., 2017). Besides H5N1, several monoclonal antibodies neutralizing H3 or H7 of IAVs or IBVs have also been recently characterized (Tan et al., 2016; Chai et al., 2017; Bangaru et al., 2018). Taken together, possibility remains to identify conserved neutralizing epitopes in the head domain in HA, in addition to extensively characterized HA stalk. Activation of these epitopes via antigen processing machineries (Lee et al., 2018b) may offer an option for enhancing the potency of cross-protection.

Despite the presence of conserved epitopes, the head domain is not a feasible vaccine antigen because of the immunodominance of the surrounding variable regions in the head that compete and prevent effective induction of antibodies toward the conserved regions. Therefore, several strategies were designed to enhance the breadth of protection by HAbased vaccines. The centralized HA was reconstructed such that whole HA contained consensus amino acids derived from diverse strains within a subtype (Weaver et al., 2011). This engineered HA antigen induced stronger immune response and provided better protection against heterologous influenza viruses, as compared to natural wild type HA antigen. Another approach, conceptually similar to the centralized HA, is based on a computationally optimized broadly reactive antigen (COBRA), in which the HA was designed to carry consensus sequences. The COBRA strategy was tested against H1N1, H3N2, and H5N1 influenza viruses, and demonstrated broad protection within a subtype (Giles and Ross, 2011; Crevar et al., 2015; Carter et al., 2016). A third strategy is the use of ancestral sequences as vaccine antigens to widen the window of cross-protection against diversified lineages or clades. Through phylogenetic tree analysis, putative ancestral HA and NA sequences have been determined and used as vaccine antigens, showing broadened cross-protection against multiple clades of H5N1 viruses in animal models (Ducatez et al., 2011, 2013). Collectively, the UIV candidates using HA head or full-length antigen are based on reconstructed HA containing consensus sequences. Although the protection breadth of those strategies appears to be restricted to a subtype or a specific clade, similar concepts may be applied to other antigens such as HA stalk or NA to substantially improve the protection breadth.

## NA AS A NOVEL TARGET FOR A UIV Multiple Function of NA in Infection Cycle

NA is a tetrameric type II transmembrane glycoprotein and the second major surface protein of influenza viruses. The role of NA in influenza infection cycle is classically known as an expedited release of virus particle from infected cells by cleaving off the sialic acid residue present in host cell membrane, thus enabling multiple rounds of infection by the newly generated viral progeny. In addition to the canonical role that operates at later stage of infection, other functions of NA relevant to the infection cycle are being recognized. For instance, the sialidase activity of NA is critical for viral entry into a host cell at early stage of infection. At the entry site in the mucus, influenza virus meets mucosal defense proteins such as mucins that are highly glycosylated and capture viral HA. NA is able to release the captured viral particles via sialidase activity, allowing them to reach the host cells successfully (Cohen et al., 2013; Yang et al., 2014). Furthermore, with the same sialidase activity, NA facilitates HA-dependent membrane fusion and enhances the viral infectivity by removing the sialic acid residues from the virion-expressed HAs (Su et al., 2010). Additionally, NA cooperates with HA to enable crawling and gliding motions of influenza virus on cell surface to enhance viral entry into a cell (Sakai et al., 2017). More interestingly, some of the H3N2 viruses use their NA for receptor binding instead of HA, suggesting a novel role of NA other than receptor-destroying activity (Lin et al., 2010; Mogling et al., 2017). These observations show that NA performs multiple functions in the entire infection cycle, suggesting that NA antibodies may represent an important means of protection against influenza viruses.

## NA Antibodies as Important Correlate of Protection

The 1968 H3N2 pandemic gave us important lessons pertaining to NA-mediated protection. The antigenic drift of NA is independent of HA; the pandemic involved a shift in HA, but NA remained close to the previous influenza A/H2N2 viruses (Schulman and Kilbourne, 1969). Notably, it has been shown that individuals with higher N2 antibody titers are less likely to be infected with the H3N2 pandemic, depicting the contribution of NA antibodies to broad protection (Schulman, 1969; Murphy et al., 1972; Monto and Kendal, 1973). However, NA has been largely ignored in the formulation of influenza vaccines due to the general beliefs that NA antibodies do not inhibit viral entry and that NA is less abundant than HA on a virion. Furthermore, the lack of a convenient assay to measure functional NA antibodies has rendered the NA forgotten antigens in influenza vaccines for decades (Eichelberger and Monto, 2019). Most of the current vaccine approaches focus on the induction of HA antibodies, both in trivalent/quadrivalent seasonal influenza vaccines and in the recent UIV candidates. However, it has been increasingly acknowledged that NA antibodies are important and independent correlates of protection and that NA immunity should be considered when evaluating vaccine potency. Clinical studies have shown a correlation between vaccination-induced or preexisting NAI antibody levels and decreased frequency of influenza infection and illness (Couch et al., 2013; Monto et al., 2015; Park et al., 2018). Further, a human challenge model depicted that NAI titers correlated more significantly with protection and disease severity than HAI titers (Memoli et al., 2016), or even HA stalk antibodies (Park et al., 2018). The observations in the human challenge models suggest that NA should be given full consideration as a vaccine antigen for better protection.

Several animal studies have identified NAI monoclonal antibodies that show protective effects against heterologous influenza infection. The breadth of NAI antibodies varied from subtype-specific to pan-influenza, depending on the conserved epitopes (Doyle et al., 2013a,b; Wan et al., 2013). Recently, it was reported that influenza infection in humans induces a variety of broadly reactive antibodies directed to the NA (Chen et al., 2018). In this study, it was shown that among the total influenza-specific antibodies induced by infection, the NA-reactive antibodies accounted for 23% and HA-reactive antibodies 35%. By contrast, the subunit or split vaccine resulted in antibody response directed predominantly to HA (87%), with only 1% for NA. The poor ability of the seasonal vaccine to induce NA antibodies was apparently due to insufficient content or structural integrity of NA antigen used in current vaccine formulation. This research suggests that correctly folded and immunologically relevant NA antigen is capable of inducing broadly protective antibody responses.

### NA-Based Vaccine as Low-Hanging Fruit for a UIV?

Although the importance of NA-immunity in protection against homologous and heterologous influenza infections is clearly established, only a few literatures have demonstrated the crossprotection of NA-only vaccine constructs. One recent study in a mouse model has reported that computationally engineered recombinant NA proteins containing consensus sequences show broad protection within the H1N1 subtype (Job et al., 2018). Some other studies reported that VLPs expressing NA provided cross-protection against heterologous challenge in mice and ferrets (Easterbrook et al., 2012; Walz et al., 2018; Kim et al., 2019), and recombinant NA in a mouse model (Liu et al., 2015; Wohlbold et al., 2015). However, co-administration of H7 HA and N3 NA in modified vaccinia virus Ankara (MVA) vectors did not demonstrate enhanced protection efficacy as compared to the efficacy of MVA-HA or MVA-NA vaccine alone (Meseda et al., 2018). A predominant immune response in favor of HA over NA, when presented in an influenza virion, is already welldocumented (Johansson et al., 1987), and the dissociation of HA and NA eliminates this antigenic competition (Johansson and Kilbourne, 1993, 1996). These observations together suggest that NA-specific immunity may be dwarfed by competition with highly immunogenic HA in the final vaccine formulation. It could be argued that if the controlled ratio of HA and NA (dwarfing NA) is the strategy adopted by successful virus infection to minimize the host immune surveillance, then a deliberate perturbation of their ratio (increasing NA) in the vaccine formulation may be a promising strategy for effective protection. It was shown that the ratio of HA/NA could vary widely (up to 50 fold) without affecting viral fitness by a single mutation in the viral promoter (Lee and Seong, 1998). It remains to be seen if such a reverse-genetic approach could be harnessed to enhance the potency of NA-based vaccines.

Currently, we have very limited knowledge about anti-NA immunity. To develop a broadly protective vaccine based on NA, there are several critical questions that need to be answered. Firstly, although the NAI antibodies have been determined as an independent correlate of protection in humans (Couch et al., 2013; Monto et al., 2015), this needs to be further validated by the NA-only vaccine constructs in animal and human models. Secondly, very little is known about the breadth of NA immunity. The literature discussed earlier has consistently demonstrated a subtype-specific protection (e.g., within N1 or N2) of NA-based vaccines in animal models. Considering the repertoire of influenza viruses infecting humans and animals (including N1, N2, N3, N7, and N9 encompassing both NA group 1 and 2) (**Figure 1**), a successful NA-based vaccine should be designed to elicit broad protection covering both NA groups. Hence, the determination of conserved regions or epitopes hidden in NA is urgently needed. Thirdly, the mechanisms by which NA antibodies contribute to protection are not completely understood. Many NAI antibodies inhibit its enzymatic activity and thus prevent the release of newly formed viral particles. However, the extent to which NAI antibody titers may be considered protective has not been determined yet. Evaluation of cross-protection against mismatched or heterologous strains may be even more complicated. While ADCC was shown to be involved in protection by nonneutralizing NA antibodies (Jegaskanda et al., 2014, 2017a; Wohlbold et al., 2017), other protective mechanisms are yet to be further elucidated. Further isolation and characterization of broadly protective NA antibodies is required for better design of NA-based vaccines. Comprehensive reviews on NA-based immunity and the perspectives on current knowledge gaps to be addressed may be found in specialized reviews (Wohlbold and Krammer, 2014; Krammer et al., 2018a).

#### UIV AGAINST INFLUENZA B VIRUSES

Besides IAVs, ∼25% of all human influenza virus infections in each epidemic season is caused by IBVs that are classified into two distinct lineages, Victoria-like and Yamagata-like lineages (Ambrose and Levin, 2012). The current seasonal influenza vaccine is prepared in a trivalent or quadrivalent formulation, depending on the inclusion of one or two lineages of IBV antigens. Although priority is given to IAVs owing to a greater impact, IBVs may be more vulnerable targets against which to develop a UIV (**Table 1**) because of their low antigenic diversity and lack of animal reservoir (**Figure 1**) (Tan et al., 2018). Indirect evidence is being accumulated by the isolation of cross-protective antibodies against IBVs. Several broadly protective antibodies binding to head or stalk domain of influenza B HA have been isolated in humans. Overall, these monoclonal antibodies show lineage-specific neutralizing activity in vitro. Further, in vivo protection against both lineages was also demonstrated in mice by passive transfer, through non-neutralizing antibodydependent effector functions (Shen et al., 2017; Hirano et al., 2018; Vigil et al., 2018; Asthagiri Arunkumar et al., 2019; Liu et al., 2019b). Notably, some of B HA stalk antibodies demonstrated extremely broad binding ability (Hirano et al., 2018) or protection against both IAVs and IBVs (Dreyfus et al., 2012). Influenza B NA-reactive broadly neutralizing antibodies were also isolated in animals and humans. Seasonal QIV induced NA antibodies with broad and potent antiviral activity against both lineages in humans (Piepenbrink et al., 2019). Additionally, murine NA antibodies also showed broad protection against both lineages of IBV (Wohlbold et al., 2017).

In line with these observations, chimeric HA strategy has also been tested for a UIV against IBVs. Chimeric HAs consisting of HA head domain from IAV and stalk domain from IBV, delivered as a DNA vaccine (prime), followed by two boosting immunizations with protein vaccines into mice, provided broad protection against both the lineages as well as an ancestral strain of IBV (Ermler et al., 2017). Mosaic HA is an advanced version, in which major antigenic sites of type B HA head domain were replaced by those of type A HA head so that antibodies directed to conserved regions both in the head and stalk domains of type B HA could be induced (Sun et al., 2019). The mosaic B HA provided broad protection against both lineages of IBV, through non-neutralizing ADCC-mediating antibody responses. There are only a few studies reporting B NA-based vaccine offering cross-lineage protection. A study showed that mice immunized with recombinant B NA protein of B/Yamagata/16/88 were protected from homologous Yamagatalike and Victoria-like lineages (Wohlbold et al., 2015). Another study demonstrated that a B NA-based vaccine inhibited the transmission of both homologous and heterologous IBVs in Guinea pig model (McMahon et al., 2019). As compared to IAVs, very little effort has been made so far to develop a UIV against IBVs. However, considering the limited diversity and variability (**Figure 1**) compared to IAVs, further identifications of broadly protective B cell and T cell epitopes would make it possible to develop a pan-influenza B vaccine in the near future (Tan et al., 2018).

## T CELL IMMUNITY AS AN ESSENTIAL FACTOR FOR TRULY UNIVERSAL INFLUENZA PROTECTION

A vast majority of current efforts to develop a UIV are focused on inducing antibody response toward surface glycoproteins, M2e, HA, and NA. However, T cell immunity has been acknowledged as a potential immune correlate of broad protection against influenza infections (Sridhar, 2016; Clemens et al., 2018). T cell immunity may not provide sterilizing or neutralizing immunity against influenza viruses but improves the standard of care by reducing the disease severity and duration of infection, facilitating recovery from illness (Sridhar, 2016). It, therefore, seems that multiple immune arms including both antibodies and T cell immunity are critical to provide a truly universal protection against highly variable influenza viruses. The influenza-specific T cell immunity is known to be highly cross-reactive by recognition of conserved peptides between different subtypes of IAV (Assarsson et al., 2008; Kreijtz et al., 2008; Lee et al., 2008; van de Sandt et al., 2014). Extensive studies have proven the protective role of vaccination or infection-induced cross-reactive CD8+ T cells in various animal models (Kreijtz et al., 2007, 2009; Bodewes et al., 2011; Hillaire et al., 2011). Additionally, in humans, CD8+ T cells offered cross-protection across seasonal, pandemic, avian IAVs, and both lineages of IBVs (Gras et al., 2010; Hayward et al., 2015; van de Sandt et al., 2015; Wang et al., 2018b). The majority of cross-reactive T cell epitopes of IAVs are derived from internal proteins; among >70 T cell epitopes identified between H5N1 and H3N2 viruses, only one derived from HA and other from internal proteins (Lee et al., 2008). This is not surprising given that the conservation rate of internal proteins is >90%, whereas that of surface HA and NA is only 40– 70% among IAVs (Lee et al., 2008), which shows that inducing T cell immunity directed to internal proteins of influenza virus may provide a basis of developing T cell-based UIVs.

Despite poor sequence homology between the HAs of IAVs and IBVs, HA stalk harbors not only extremely broad B cell epitopes but also T cell epitopes encompassing both types of influenza viruses. The fusion peptide in HA stalk contains a cross-reactive CD4+ T cell epitope conserved in both IAVs and IBVs, although its protective role has not been examined in vivo (Babon et al., 2012). A number of CD4+ and CD8+ T cell epitopes are highly conserved in internal proteins (Terajima et al., 2013). A recent study has discovered a universal human CD8+ T cell epitope in PB1 (NMLSTVLGV PB1413−412) that is identical across IAVs, IBVs, and ICVs (Koutsakos et al., 2019). The preexisting memory PB1413−412+CD8+ T cells were detected in the blood and lung tissues of healthy donors and clonally expanded upon infection with IAV or IBV. This report not only showed the existence of heterotypic memory CD8+ T cells in humans that could be induced by exposure to influenza viruses, but also presents the prospect of designing a T cell-based UIV. However, these cross-reactive T cells were not induced in HHD-A2 mouse model despite multiple influenza infections or vaccinations and the protective role of the T cells could not be assessed in the study. Nonetheless, the existence of a number of cross-reactive T cell epitopes between IAVs and IBVs provides an avenue to address to a UIV.

Several T cell-based vaccine candidates are in different stages of clinical development, the major underlying strategy of which is to deliver multiple T cell epitopes derived from different viral antigens including internal as well as surface antigens (Sridhar, 2016). Delivery platforms include replicating or nonreplicating viral vectors derived from vaccinia or adenovirus, recombinant VLPs, recombinant protein or peptide vaccines, and DNA vaccines (Sridhar, 2016; Clemens et al., 2018). Modified vaccinia Ankara (MVA) vector encoding NP and M1 was shown to induce robust T cell responses and provide cross-protection against multiple subtypes in animals and humans (Antrobus et al., 2012; Powell et al., 2013; Hessel et al., 2014; Folegatti et al., 2019). The baculovirus VLPs carrying influenza HA/NA and M1 offered cross-protection where T cells played a significant role in protection in mice (Hemann et al., 2013; Keshavarz et al., 2019). Synthetic peptides or fusion proteins harboring multiple conserved T cell epitopes have also been evaluated for immunogenicity and protective efficacy in animal models (Adar et al., 2009; Rosendahl Huber et al., 2015). While the vaccine approaches described above deliver exogenous antigens and induce CD4+ T cells as well as CD8+ T cells by crosspresentation, DNA vaccines are designed to predominantly activate cytotoxic CD8+ T cells to recognize endogenously expressed antigens presented on MHC class I molecules. In fact, the first report on DNA vaccines was targeted to influenza virus (Cohen, 1993; Ulmer et al., 1993), but initial success in a mouse model did not well-translate into higher animal models due to poor performance (Porter and Raviprakash, 2017). To date, much progress has been made to improve the efficacy of DNA vaccine against influenza virus, encompassing rational design of antigens and expression vectors and the development of novel adjuvants and delivery methods (Lee et al., 2018a). Candidate universal DNA vaccines encoding NP, matrix proteins, or PB1 were shown to decrease viral load and cross-protect against heterologous challenges in diverse animal models including mice, pigs, ferrets, and macaques (Ulmer et al., 1993; Tompkins et al., 2007; Price et al., 2009; Bragstad et al., 2013; Wang et al., 2015; Koday et al., 2017). Further studies are required for to refine DNA vaccine approaches as a stand-alone UIV. Recent studies have indicated that DNA vaccines may serve an attractive component of prime-boost strategy, considering it as a very effective means to priming immune system when preexisting immunity is low (Ledgerwood et al., 2013; DeZure et al., 2017). Despite the potential for broad protection, the safety issues need to be monitored closely, especially because of the documented rise in pathological consequences associated with CTL responses (Peiris et al., 2010; Duan and Thomas, 2016).

#### LIVE ATTENUATED INFLUENZA VACCINE AS AN ALTERNATIVE STRATEGY

## Cross-Protective Immunogenicity of LAIV

LAIV has been approved for clinical use in humans since 2003 and is proven effective in preventing influenza infections (Mohn et al., 2018). The protection of LAIV is superior to IIVs due to multifaceted immune arms including antibody responses and cell-mediated responses that operate systemically and locally (Jang and Seong, 2013a,b; Sridhar et al., 2015). Further, displaying a whole set of viral antigens in a native conformation into the host immune system presents a definite advantage of LAIV to generate better protective immunity than the other strategies relying on a limited set of antigens. As discussed above, T cell immunity directed to the conserved viral epitopes constitutes the cornerstone of cross-protection. A large number of researches have shown that T cell responses induced by LAIV are critical for broad protection against heterologous and heterosubtypic influenza infections in animal models (Cheng et al., 2013; Jang and Seong, 2013a; Rekstin et al., 2017). In young children, only LAIVs were shown to induce durable and potent T cell responses, while developing similar levels of antibody response as compared to IIVs (Belshe et al., 2000; Hoft et al., 2011; Mohn et al., 2015, 2017). Despite well-documented cross-protection, LAIV has received little attention to develop a UIV. This may be attributed to the general belief that LAIV is not effective at inducing broadly neutralizing antibodies against conserved domains in surface antigens (e.g., M2e or HA stalk). However, close attention is recently being given to LAIV as an alternative platform as a potent and cross-protective vehicle than previously thought, through inducing multifaceted immune correlates including T cell response and antibody-mediated effector functions (Jang and Seong, 2013a, 2014).

#### UIV Approaches Using LAIV

In the UIV approaches reported so far, LAIV was used either as a component in prime-boost regimens with other different vaccine formats such as IIV, DNA vaccine, or recombinant protein vaccine. Alternatively, LAIVs were also studies as a standalone vaccine given in single or multiple doses. A reassortant LAIV expressing a chimeric HA was constructed under the genetic background of Russian strain (A/Leningrad/134/17/57 cold-adapted virus) and used as a boosting vaccine in a ferret model (Nachbagauer et al., 2017). Notably, the LAIV-IIV regimen showed greater protective efficacy against the pandemic H1N1 challenge than the IIV-IIV regimen in terms of viral loads in the respiratory tissues, despite 32-fold lower stalk antibody titers in serum. Several factors were presumed to account for this disconnect between stalk antibody titers and protection efficacy, including anti-NA immunity, mucosal IgA antibodies, cellmediated immunity, and non-specific innate immune responses offered by the LAIV. Similar results were obtained when using a different LAIV strain (A/Ann Arbor/6/60 cold-adapted strain) (Nachbagauer et al., 2018). In a subsequent study performed by the same group, chimeric HA-containing the LAIV-LAIV (A/Leningrad/134/17/57 cold-adapted strain) vaccine regimen was compared with the LAIV-IIV combination in terms of protection efficacy in a ferret model, in which the LAIV-LAIV vaccine regimen conferred superior protection against pandemic H1N1 and H6N1 challenges than the LAIV-IIV (Liu et al., 2019a). Another group tested a vaccination regimen comprising only LAIV (A/Leningrad/134/17/57 cold-adapted strain) as primeboost vaccination in a mouse model (Isakova-Sivak et al., 2018). To enhance the breadth of protection, an internal gene of cold-adapted virus was replaced with the wild type. This study compared the immunogenicity and cross-protection between chimeric HA-containing LAIVs and natural HA-containing LAIVs. The chimeric HA-containing LAIVs induced higher HA stalk antibody titers and showed better cross-protection against heterologous infection with various group 1 IAVs. Thus, a cooperative role of cell-mediated immunity and HA stalk antibodies was suggested, although their individual contribution to protection were not assessed directly. It would be interesting to investigate if the cross-protection could be extended to group 2 influenza viruses such as H3 or H7 strains.

Besides the Leningrad and Ann Arbor strains, an independent cold-adapted vaccine strain (X-31ca) that was previously used for trivalent seasonal vaccine (Jang et al., 2014), H5N1 pre-pandemic vaccine (Jang et al., 2013c), and 2009 pdmH1N1 vaccine (Jang et al., 2012, 2013a), was recently tested as a UIV in a mouse model (Jang et al., 2018). Mice vaccinated with single or two doses of X-31 ca LAIVs were completely protected against lethal challenge of heterosubtypic strains encompassing both HA group 1 and 2 IAVs. Interestingly, boosting with heterosubtypic LAIVs carrying different HA and NA surface antigens showed more potent cross-protection than homologous boosting. T cell immunity and NK cell-mediated ADCC activity was demonstrated to contribute significantly to the observed cross-protection. As the first report of pan-influenza A protection covering both HA groups, these results merit further studies in a ferret model for clinical relevance. Hence, the LAIVs appear to be a powerful tool to develop a UIV that confers broad and potent cross-protection as a stand-alone vaccine or in combination with other strategies.

## Future Prospects of LAIV-Based UIVs

While the LAIV presents a remarkable prospect for a broadly protective influenza vaccine, several important issues need to be addressed for it to serve as a reliable vaccine modality. The protection efficacy of a LAIV substantially differs with age. The estimated protection efficacy of a LAIV is 80% in young children and only 40% in adults, to the matched strains (Jefferson et al., 2008, 2010). As for T cell immune responses, clinical studies indicate that LAIVs induce better T cell response than IIVs in both children and adults (He et al., 2006; Subbramanian et al., 2010; Hoft et al., 2011). However, clinical studies have reported that LAIVs are not effective at inducing T cell responses in adults and the elderly, perhaps due to preexisting "vector" immunity, which limits the replication of LAIVs (He et al., 2006; Forrest et al., 2008; Hoft et al., 2011). Considering that both humoral and cell-mediated immunity contribute to broad protection, it will be important to elucidate how preexisting immunity or immunologic imprinting affects B cell and T cell immune response induced by LAIVs in humans (Gostic et al., 2016; Henry et al., 2018). For this purpose, animal models mimicking preexisting immunity and controlled human challenge studies will be needed. Considering that LAIVs mimic natural infection, a fundamental question remains to be answered: if infection (or vaccination with LAIV) is effective for conferring protection, why are humans vulnerable to repeated infections with homologous or heterologous strains? At the population level, currently used cold-adapted LAIVs provide a relatively low protection rates (∼40% in adults) even against matched strains (Jefferson et al., 2010). However, little is known whether individuals who successfully acquired protective immunity by a cold-adapted LAIV were protected against other heterologous strain(s) in the next epidemic. This may be directly addressed by a wellcontrolled longitudinal cohort study using human challenge models. Some studies showed that LAIVs were able to generate cross-reactive T cell responses in children for up to 1 year after vaccination, as a basis of long-term cross-protection in humans (Mohn et al., 2015, 2017).

Most of the LAIV-based approaches are based on coldadapted attenuated strains (Nachbagauer et al., 2017, 2018; Jang et al., 2018; Liu et al., 2019a). These strains acquire multiple attenuation mutations in the internal genes during the coldadaptation process, which contribute to genetic stability to overall attenuation (Jang et al., 2016). A common concern for live vaccines is safety issues, especially those associated with potential reversion of attenuating mutation(s) into virulence during vaccination. LAIVs acquired multiple attenuation mutations among various internal genes and proven safe as seasonal influenza vaccines (especially for A/Ann Arbor strain as the master strain for FluMist). However, genetically engineered LAIVs with a limited set of attenuation mutations, e.g., NS1 deletion or elastase-dependent HA cleavage (Talon et al., 2000; Stech et al., 2005), may require additional monitoring on safety.

Defining the precise correlates of protection represents the most challenging step in the development of a UIV (Erbelding et al., 2018). Significant efforts were made to identify the protection mechanisms of HA stalk-based vaccines, suggesting that direct neutralization in combination with Fc-dependent indirect effector mechanisms mediated by stalk antibodies were the primary correlates of protection. In contrast to HA stalkbased vaccines, LAIVs elicit multiple immunological factors including serum IgG antibodies and mucosal IgA antibodies to surface antigens (HA, NA, and M2) and cell-mediated immunity to internal antigens, synergistically contributing to protection. However, their quantitative relationship to protection has not been determined, not even for homologous protection (Sridhar et al., 2015; Mohn et al., 2018), let alone for cross-protection against heterologous infection. The development of standardized assays to quantitatively measure T cell-mediated protection is particularly challenging, as the magnitude and the subset of T cells critical for protection is likely to differ according to strains of LAIV and challenge viruses. Further, mucosal IgA antibodies are believed to correlate with cross-protection, but it is still challenging to measure the neutralizing activity or effector functions of mucosal IgA antibodies. The complicated nature of LAIV-induced immunity, including non-neutralizing antibodies and diverse subsets of T cells, present a bottleneck to identifying precise correlates of protection.

Another important aspect of LAIV-based strategies lies on the LAIV strains. During the past century, H1N1 and H3N2 subtypes of influenza A viruses were the most prevalent strains in humans, causing annual epidemics as well as occasional pandemics, except for the temporal circulation of H2N2 during 1957–1968 (Kilbourne, 2006) (**Figure 6**). Accordingly, seasonal influenza vaccines are recommended to include H1N1 and H3N2 vaccine antigens for more than 40 years since 1977. Therefore, it is likely that most contemporary population has preexisting T cell immunity to H1N1 and H3N2 strains through natural infections or vaccinations. It should be remembered that currently licensed LAIVs (A/Ann Arbor/6/60 ca and A/Leningrad/134/17/57 ca) are of H2N2 subtypes. Probably, the nature of strain itself does not really matter for seasonal influenza vaccine, for which strain-specific immunity is focused on the surface HA antigen. However, for eliciting cross-protection, the role of conserved region become important (see section HA stalk-based UIV approaches, Conserved targets in HA other than stalk, NA as a novel target for a UIV for HA/NA and section T cell immunity as an essential factor for truly universal influenza protection for internal proteins). It is likely that human populations under 50 years of age (born after 1968 when H2N2 became extinct) has little preexisting immunity against H2N2, but predominantly against H1N1 and H3N2 viruses. It will therefore be worthwhile to examine whether cold-adapted LAIVs of H1N1 (Jang et al., 2018) or H3N2 origin (non-existent, to our knowledge) offer a beneficial effects on boosting the preexisting cross-reactive T cell immunity and antibody effector functions (section Mode of protection by a UIV; **Figure 3**).

As for IBVs, two distinct lineages diverged and circulated after 1985, which necessitates the incorporation of Victoria-like and Yamagata-like lineages in seasonal influenza vaccines. This leads us to raise a possibility of using an ancestral influenza B strain (such as B/Lee/40) before divergence into two different lineages as a UIV candidate for IBVs. Establishing new cold-adapted LAIV strains is a time-consuming and laborious. However, recent advances in reverse genetics and rational approaches to attenuate viral virulence have enabled the rapid conversion of a wild type virus into a novel LAIV strain. These approaches include NS1 truncation, elastase-dependent HA cleavage, caspase-dependent NP and NS1 cleavage, microRNA-mediated silencing, and codon deoptimization (Talon et al., 2000; Stech et al., 2005; Coleman et al., 2008; Perez et al., 2009; Jang et al., 2013b).

The redirection of host immune responses from surface proteins toward internal proteins may be achieved by rational vaccination strategies with LAIVs. For example, the down regulation of expression levels of HA or NA in a LAIV (Yang et al., 2013) is likely to result in preferable induction of T cell immunity to internal proteins. Alternatively, vaccination with LAIV carrying HA from non-human influenza viruses such as H5 or H9 may be effective at boosting preexisting T cell immunity to internal proteins in humans. Given the cross-reactivity of T cell immunity between IAVs and IBVs, A type LAIVs and B type LAIVs may be administered as a bivalent formulation or by sequential vaccination to induce improved protection.

#### OTHER STRATEGIES FOR UIVs

A number of alternative strategies are being investigated for their potential to serve as a UIV platform. First approach is to use infection-competent but replication-defective viruses. For instance, the M2 knock-out influenza virus (M2SR) rescued from M2-producing cells is able to infect cells but does not produce progeny virus due to the lack of functional M2 protein. The M2SR was shown to induce strong cell-mediated and humoral immunity and to provide broad protection against both homologous and heterologous influenza virus challenge in mice and ferrets (Sarawar et al., 2016; Hatta et al., 2017, 2018). A conceptually similar to M2SR, PB2 knock-out influenza virus could be produced in PB2-expressing cells as a vaccine.

The protection efficacy of PB2-KO vaccine was tested against diverse influenza strains in mice (Victor et al., 2012; Uraki et al., 2013; Ui et al., 2017). Pseudotyped influenza A virus also presents a similar replication-incompetent virus approach. The pseudotyped influenza virus lacking HA produced in HAexpressing cells can infect cells and express all the viral proteins except for HA. Studies have shown that the vaccination of mice and ferrets with the pseudotyped influenza virus generated a vigorous T cell response and reduced lung viral loads and weight loss after challenge with homologous and heterologous influenza viruses (Powell et al., 2012; Baz et al., 2015). These approaches based on replication-defective viruses should be rigorously evaluated for safety, considering potential reversion into virulence by acquiring the wild type gene by reassortment with circulating viruses (Lowen, 2017).

Secondly, mRNA vaccines have emerged as a promising alternative to conventional vaccine platforms against various infectious diseases including the influenza virus (Pardi et al., 2018a). Initially, the mRNA vaccines successfully induced protective B cell and T cell immune response in mice, ferrets, and pigs (Petsch et al., 2012). In subsequent studies, mRNA vaccines encoding HA induced neutralizing antibodies and T cell immunity essential for heterologous protection in mice and ferrets (Brazzoli et al., 2016). Notably, mRNA vaccines encoding HA elicited HA stalk antibodies in mice, rabbits, and ferrets, and the stalk antibody responses were associated with protection against homologous and heterologous influenza virus infection in mice (Pardi et al., 2018b), depicting promising potential as a UIV platform.

## CONCLUSIONS AND PROSPECTS

During the past few years, several critically important issues have emerged in developing a UIV. While identification of a number of broadly protective antibodies presents an optimistic prospect for pan-influenza therapeutics, induction of such antibodies at a sufficiently protective level by vaccination has not been accomplished yet. A variety of viral targets have been identified to induce broad protection, including M2e and the conserved regions in the HA, such as stalk, RBS, and VE, and the longneglected NA has emerged as an essential target for durable and broad protection. In virtually all cases, the rationale for broad protection is to redirect the immune responses from the variable immunodominant regions to conserved immunosubdominant regions. While effective in eliciting cross-protection, concerns were raised to non-neutralizing antibodies that potentially trigger the enhancement of disease. T-cell immunity has also been considered as an important correlate of cross-protection. LAIVs are able to induce multifaceted immune response and thus have increasingly received close attention as a promising vaccine platform, either alone, or in boosting format. Preexisting immunity or immunologic imprinting, established by prior exposure to influenza by infection or vaccination, may influence or thwart the desired breadth of immune response by UIVs. Therefore, disparities may exist between pre-clinical evaluation of a UIV without due recognition of immune imprinting and human challenge studies. Lastly, the correlates of protection and their precise molecular mechanisms have not been determined yet, which remains a significant gap between development and licensing of a UIV. Given the limited antigenic diversity of IBVs among humans and the lack of animal reservoir, paninfluenza B virus vaccine may be attainable in the near future. On the other hand, IAVs continually change their antigenicity via antigenic drift and shift and zoonotic spillover from animal reservoirs, requiring multi-faceted immune arms to increase the breadth of protection. Judicious choice of antigens and their efficient designing to confer a broad protection for a prolonged period are required to counterfeit the immune evasion and

#### REFERENCES


provide a truly universal protection against the ever-changing influenza viruses.

#### AUTHOR CONTRIBUTIONS

BS designed and conceived the review. YJ and BS investigated the literature and wrote the manuscript.

#### FUNDING

This work was supported by grants from National Research Foundation of Korea (NRF-2018R1D1A1B07048881; NRF-2018M3A9H4079358) and also supported by grant from the Ministry of Food and Drug Safety of the Republic of Korea (19172MFDS255).


MVA-NP+M1 mnufactured on the AGE1.CR.pIX avian cell line. Vaccines 7:33. doi: 10.3390/vaccines7010033


attenuated pre-pandemic H5N1 influenza vaccines. Vaccine 31, 3339–3346. doi: 10.1016/j.vaccine.2013.05.080


of influenza virus-like particles in a mouse model. J. Cell Physiol. 226, 1110–1117. doi: 10.1002/jcp.28339


that contains ectodomains of matrix protein 2. Vaccine 21, 2616–2626. doi: 10.1016/S0264-410X(03)00040-9


by Tobacco mosaic virus-based vectors. Curr. Pharm. Des. 19, 5587–5600. doi: 10.2174/13816128113199990337


cross-reactive human monoclonal antibodies. J. Virol. 85, 13463–13467. doi: 10.1128/JVI.05193-11


**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 Jang and Seong. 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.

digital media

of impactful research

article's readership