**Vaccines against RNA Viruses**

Editor

**Juan Carlos Saiz**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Editor* Juan Carlos Saiz Instituto Nacional de Investigacion y Tecnologia Agraria y Alimentaria Spain

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Vaccines* (ISSN 2076-393X) (available at: https://www.mdpi.com/journal/vaccines/special issues/ RNA Viruse).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Article Number*, Page Range.

**ISBN 978-3-03943-623-1 (Hbk) ISBN 978-3-03943-624-8 (PDF)**

c 2020 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

## **Contents**


## **About the Editor**

**Juan Carlos Saiz** received his Ph.D. in Biochemistry and Molecular Biology from Universidad Autonoma de Madrid (U.A.M.), Spain, in 1987. Then, he was trained in Virology as a post-doctoral ´ at the PIADC/ARS in New York (1987-91), at the Instituto Nacional de Investigacion Agraria y ´ Alimentaria, INIA, in Madrid (1991-92), and at the Centro de Biolog´ıa Molecular "Severo Ochoa", CBMSO (1992-96), also in Madrid. From 1996 to 2002 he was in charge of the Laboratory of Viral Hepatitis at the Hospital Cl´ınic (Barcelona). Since 2002, Professor Saiz is the Research Leader of the "ZOOVIR" laboratory at INIA-CSIC, where he was director of the Dpt. Biotechnology (2010-2015). Professor Saiz's group study different aspects of Flavivirus (West Nile, Dengue, Zika, and Usutu viruses) biology, focusing in the prevention and control of these worldwide threatening infectious pathogens. More specifically, the main group's objectives are i) the identification of viral and cellular targets for antiviral intervention; ii) the development of vaccines candidates and its testing in animal model (mice and birds); and iii) the development of new diagnostic systems. Professor Saiz (Researcher ID: L-4638-2014; Orcid: 0000-0001-8269-5544) is co-author over 160 publications with over 3500 citations in more than 2760 articles (h-index 35), and has been granted with 37 national and international open call projects.

## *Editorial* **Vaccines against RNA Viruses**

#### **Juan-Carlos Saiz**

ZOOVIR, Department of Biotechnology, Instituto Nacional de Investigaciones Agrarias (INIA), 28040 Madrid, Spain; jcsaiz2012@gmail.com

Received: 25 August 2020; Accepted: 26 August 2020; Published: 27 August 2020

RNA viruses cause animal, human, and zoonotic diseases that affect millions of individuals, as is being exemplified by the devastating ongoing epidemic of the recently identified SARS-Cov-2. For years vaccines have had an enormous impact on overcoming the global burden of diseases. Nowadays, a vast number of different approaches, from purified inactivated and live attenuated viruses, nucleic acid (DNA or RNA) based candidates, virus-like particles, subunit elements, and recombinant viruses are been employed to combat diverse diseases caused by RNA viruses. However, although for some of them efficient vaccines are available, in most instances, and for different reasons (technologic o economic restrictions, etc.), they are scarcely used in the field, and even for many of them no licensed vaccines exist. This will probably change dramatically with the current Covid-19 pandemic, as a vast variety of vaccinology approaches are being tested against it, with hundreds of candidates under development, dozens of them already in clinical trials phase III and IV, a fact that is breaking records in vaccine development and implementation. This is becoming possible thanks to the enormous work carried out during years to have the bases for a quick response, even against unknown pathogens, in an impressive short time.

In this *Vaccines* Special Issue "Vaccines against RNA viruses", results obtained with different vaccine methodological approaches against human, animal, and zoonotic viruses are presented by field experts.

Human Immunodeficiency Virus (HIV), for which no vaccine is available, continues to be a major global public health issue that has already cost almost 33 million lives so far, and with an estimated 38 million people living with it. Even though antiretroviral therapy considerably prolongs the lifespan of patients making the disease a manageable chronic health condition, treatment is expensive and is not available for all infected patients. Thus, vaccines are needed to control HIV spread. So, in this Issue, Rudometov et al. [1] describe the development of a human immunodeficiency virus-1 (HIV-1) immunogen using a polypeptide recombinant non transferrin bound iron (nTBI) protein carrying four T- and five B-cell epitopes from HIV-1 Env and Gag proteins recognized by broadly neutralizing HIV-1 antibodies, combined with Th-epitopes. Immunization of rabbits with this nTBI protein elicited antibodies that recognize HIV-1 proteins, supporting its possible use as immunogen.

Another relevant human virus is the human rotavirus (HRV) that is a leading cause of severe, dehydrating gastroenteritis, mainly in children. Current live rotavirus vaccines are efficient, but costly, and they may present increased risk of intussusception due to vaccine replication in the gut of vaccinated children. Here, Ramesh and coworkers [2] assess the immunogenicity and protective capacity of a novel P24-VP8\* nanoparticle vaccine using the gnotobiotic (Gn) pig model of human rotavirus. Three doses (200 μg/dose) of the vaccine candidate intramuscularly administered with Al(OH)3 (600 μg) as an adjuvant conferred significant protection against infection and diarrhea after challenged with virulent heterologous rotavirus strains. Vaccinated animals showed a significant reduction in the mean duration of diarrhea, virus shedding in feces, and significantly lower fecal cumulative consistency scores in comparison with mock immunized control animals. Vaccinated animals also elicited strong VP8\*-specific serum neutralizing antibody responses, but, consistent with the methodological approach (immunization route, adjuvant, and lack of replication), no serum or

intestinal IgA antibody responses, or strong effector T cell responses, were induced. Authors indicate that their results open the option of the initiation of clinical trials using the new designed P24-VP8\* nanoparticle vaccine candidate.

For his part, Duncan and coworkers [3] update the current status of the hepatitis C virus (HCV) vaccination, describing and discussing the different ongoing research in the field, and emphasizing that, despite the great advances in HCV therapeutics achieved lately, treatments sometime fail to prevent reinfection and are quite expensive and, thus, efficient prophylactic HCV vaccines are still needed.

Pigs are one of the main livestock species, comprising 36% of the worldwide meat consumption, with hundreds of tons produced yearly. Pigs can be infected by a wide variety of RNA viruses that cause huge losses for the food industry. In addition, they can serve as bridge between wild host animals and humans. Thereby, availability of efficient vaccines is crucial. Among the viral infections affecting pigs, a very important one is that of the porcine reproductive and respiratory syndrome virus (PRRSV) that causes reproductive failure in pregnant sows and respiratory disease in young pigs, accounting for huge economic losses to industry across the world. One critical drawback for vaccine development against PRRSV is its high genetic diversity. Here, Ciu and coworkers [4] show their results after testing a glycoprotein GP5-Mosaic DNA vaccine and a recombinant GP5-Mosaic vaccinia virus (rGP5-Mosaic VACV) vaccine in pigs, and compare their immunogenicity and protective capacity against heterologous virulent strains (VR2332 or MN184C) with that of a rGP5-wyld-type, WT (VR2332) DNA and a rGP5-WT (VR2332) VACV vaccine, using as controls groups inoculated with empty vector DNA or empty VACV. The authors show that vaccination with the GP5-Mosaic-based vaccines resulted in cellular reactivity and higher levels of neutralizing antibodies to both VR2332 and MN184C PRRSV strains, whilst vaccination with the GP5-WT vaccines only induced response against the heterologous challenging virus (VR2332). After infection with either strain, viral titers in sera and tissues, as well as lung lesions, were lower in the GP5-Mosaic vaccinated pigs. These results indicate that, using a DNA-prime/VACV boost regimen, the developed GP5-Mosaic candidates confer protection in pigs against heterologous viruses and, thus, are feasible vaccine candidates.

Another important disease in pigs is that caused by the porcine epidemic diarrhea virus (PEDV), a coronavirus responsible of highly contagious intestinal infections that may result in the death of newborn piglets and weight loss in pigs of all ages, and that seriously damages the swine industry. From 2010, novel highly virulent PEDV genotype 2 variants have spread through China, resulting in high mortality of newborn piglets and huge economic losses. Current vaccines against the classical CV777 strain of genotype 1 do not provide effective protection against Chinese highly virulent PEDV variant infections. Thus, Shi and colleagues [5] using a RNA interference (RNAi) approach, show that three short hairpin RNA (shRNA) expressing plasmids directed against the viral nucleocapsid (N) present antiviral activities in intestine epithelial cells infected with a both classical CV777 and LNCT2 strains, and that these shRNAs markedly reduce viral replication of both strains upon downregulation of N protein production; thus, these strategies, based on targeting viral processivity factors, may be feasible vaccine alternatives.

By a different approach, Cañas-Arranz et al. [6] deepen their many years of previous research with peptides as vaccine candidates against foot-and mouth disease virus (FMDV). The virus is responsible for a highly contagious transmissible infection of cloven-hoofed animals (mainly pigs and cows) that may cause huge economic impact and whose control relies on efficient vaccination using a conventional chemically inactivated virus. However, these current vaccines present limitations, as the need for a strict maintenance of the cold chain and for a constant update of vaccine strains due to the virus's antigenic diversity, as well as the fact that the manufacturing process poses significant biosafety concerns, as it has been related to occasional escape episodes. Therefore, vaccines incorporating specific outbreak-relevant epitopes capable of eliciting protective and quick responses can become an invaluable emergency resource for FMD containments during outbreaks. With this background, the Spanish group go one step further by testing in pigs a dendrimer B2T peptide with two copies of a B-cell VP1 epitope linked through maleimide units to a T-cell 3A epitope, and show that a single

dose of B2T evokes a specific protective immune response with high neutralizing titers, activates the T cell response, induces IFN production, and fully protects 70% of vaccinated pigs that did not present clinical signs of the disease. Their results strengthen the potential of B2T as a safe, cost-effective candidate vaccine, which can be of particular interest in emergency scenarios.

Rift Valley fever virus (RVFV), a mosquito-borne bunyavirus widely distributed in Sub-Saharan countries, Egypt, and the Arabian Peninsula, is another important pathogen causing disease in humans, in which severe cases can end in encephalitis or hemorrhagic fever, and in ruminant livestock, characterized by an increased incidence of abortion or fetal malformation. At present, there are a few veterinary vaccines available for use in endemic areas, but there is no licensed human vaccine. López-Gil and coworkers [7], by using an approach based on the modified vaccinia Ankara (MVA) virus encoding the RVFV glycoproteins (rMVAGnGc), extend their previous observations that a single inoculation was sufficient to induce a protective immune response in mice after a lethal viral challenge, which was related to the presence of glycoprotein specific CD8+ cells and a low-level detection of in vitro neutralizing antibodies. They tested the efficacy and immune response in mice immunized with recombinant MVA viruses expressing either glycoprotein Gn (rMVAGn) or Gc (rMVAGc) and suggest that, in the absence of serum neutralizing antibodies, protection is strain-dependent and mainly due to the activation of the cellular response against Gc epitopes. Even more, their data point to the induction of a suboptimal humoral immune response, since disease was exacerbated upon the virus challenge in the presence of rMVAGnGc or rMVAGn immune serum. These results support that Gc-specific cellular immunity is an important component that contributes to efficient protection against RVFV infection.

The fishing industry is increasingly relevant, and the number of farms has grown exponentially, being a very important source of food. Viral hemorrhagic septicemia virus (VHSV), a novirhabdovirus, is one of the worst viral threats to fish farming. VHSV has been isolated from more than 50 fish species across the world, including farmed and free-living marine species. Detection of even a single positive sample in a farm has to be notified to the Office International des Epizooties, and implies the sacrifice of all the farmed fish, thus leading to serious economic losses. Non-virion (NV) gene-deleted VHSV (dNV-VHSV) has been postulated as an attenuated virus, as its absence leads to lower induced pathogenicity. In the study by Chinchilla et al. [8] the immune transcriptome profiling in trout infected with dNV-VHSV and wt-VHSV and the pathways involved in immune responses were analyzed in the context of infection. The authors show that dNV-VHSV upregulates more trout-signaling immune related genes and pathways, whereas wt-VHSV maintains more non-regulated genes. Therefore, wt-VHSV impairs the activation at short stages of infection of pro-inflammatory, antiviral, proliferation, and apoptosis pathways, delaying innate humoral response and cellular crosstalk, whereas dNV-VHSV promotes the opposite effects, supporting the use of dNV-VHSV as a potential live vaccine candidate.

Finally, Jiménez de Oya and coworkers [9] deeply update current knowledge and available data about the vaccination of birds against the West Nile virus (WNV), the worldwide most distributed mosquito-borne flavivirus. Although humans and equids can sporadically be infected, birds are the natural host of WNV. When clinical signs arise in birds it is due to multi-organ invasion, mainly in the central nervous system, which can lead to death 24–48 h later. Nowadays, vaccines are only available for use in equids; thus, availability of avian vaccines would benefit bird populations, both domestic and wild ones. Such vaccines could be used in endangered species housed in rehabilitation and wildlife reserves, and in animals located at zoos and other recreational installations, but also in farm birds, and in those that are grown for hunting and restocking activities. Even more, controlling WNV infection in birds can also be useful to prevent its spread and limit outbreaks. In their review, Jimenez de Oya and colleagues comprehensively present the results obtained with commercial and experimental vaccines in domestic and wild avian species, and the possible benefits and drawbacks of bird vaccination against WNV are discussed.

The world remains burdened by high morbidity and mortality diseases and, as exemplified by the current devastating pandemic of SARS-Cov-2, and new emerging or re-emerging pathogens are

likely to spread in the future. Hereby, a more comprehensive understanding of the current trends in vaccine development and assessment of the molecular mechanisms and immune responses involved in the elicited responses are essential. In this line, the articles in this Issue highlight recent advances in the development of efficient vaccines against RNA viruses infecting animals and humans, some of which are zoonotic. Different approaches are described from attenuated and recombinant viruses, to peptides, and DNA and RNA-based candidates, which hopefully will contribute to a better and quick preparedness against RNA virus infections.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


© 2020 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **Artificial Anti-HIV-1 Immunogen Comprising Epitopes of Broadly Neutralizing Antibodies 2F5, 10E8, and a Peptide Mimic of VRC01 Discontinuous Epitope**

**Andrey P. Rudometov 1,\*, Anton N. Chikaev 2,\*, Nadezhda B. Rudometova 1, Denis V. Antonets 1, Alexander A. Lomzov 3, Olga N. Kaplina 1, Alexander A. Ilyichev <sup>1</sup> and Larisa I. Karpenko 1,\***


Received: 26 June 2019; Accepted: 30 July 2019; Published: 6 August 2019

**Abstract:** The construction of artificial proteins using conservative B-cell and T-cell epitopes is believed to be a promising approach for a vaccine design against diverse viral infections. This article describes the development of an artificial HIV-1 immunogen using a polyepitope immunogen design strategy. We developed a recombinant protein, referred to as nTBI, that contains epitopes recognized by broadly neutralizing HIV-1 antibodies (bNAbs) combined with Th-epitopes. This is a modified version of a previously designed artificial protein, TBI (T- and B-cell epitopes containing Immunogen), carrying four T- and five B-cell epitopes from HIV-1 Env and Gag proteins. To engineer the nTBI molecule, three B-cell epitopes of the TBI protein were replaced with the epitopes recognized by broadly neutralizing HIV-1 antibodies 10E8, 2F5, and a linear peptide mimic of VRC01 epitope. We showed that immunization of rabbits with the nTBI protein elicited antibodies that recognize HIV-1 proteins and were able to neutralize Env-pseudotyped SF162.LS HIV-1 strain (tier 1). Competition assay revealed that immunization of rabbits with nTBI induced mainly 10E8-like antibodies. Our findings support the use of nTBI protein as an immunogen with predefined favorable antigenic properties.

**Keywords:** artificial protein; polyepitope B- and T-cell HIV-1 immunogen; epitopes of broadly neutralizing HIV-1 antibodies; peptide mimic of discontinuous epitope; immunogenicity

#### **1. Introduction**

Although Human Immunodeficiency Virus (HIV-1) is one of the best-characterized viruses, there is no efficient vaccine against this pathogen so far. Giving credit for notable progress in approaches to antiretroviral therapy that considerably prolongs the lifespan of HIV-infected patients, it should be noted that these are only palliative means to control the virus which cannot stop the HIV-1 pandemic [1,2]. For the most effective control of HIV-1 spread, a prophylactic vaccine should be used widely [3,4]. However, vaccine development is associated with particular well-known issues. First of all, HIV-1 genetic and consequent antigenic drift allows for evasion of the protective effects of the immune system. Therefore, traditional vaccine strategies have failed to protect against the virus [5–7].

Development of artificial polyepitope HIV-1 immunogens using a broad range of protective B- and T-cell epitopes from the viral antigens that can induce broadly neutralizing antibodies and responses of cytotoxic (CD8+ CTL) and helper (CD4+ Th) T-lymphocytes is one of the promising strategies for antiviral vaccine design [6,8–13].

There are a number of efforts developing artificial polyepitope T-cell immunogens [10,14–21]. Some of them have proven successful in inducing CD4+ T-cell and CD8+ T-cell responses of much greater breadth and magnitude in non-human primates compared to the vaccines containing full-length HIV protein genes [6,10]. Several polyepitope T-cell vaccine candidates have undergone phase I clinical trials [22–24].

The development of artificial B-cell HIV-immunogens, including those constructed using epitopes of broadly neutralizing HIV-1 antibodies (bNAbs), is the most complicated problem, since the majority of them recognize conformational epitopes and, significantly more rarely, linear epitopes. Furthermore, conformational B-cell epitopes on HIV surface glycoproteins are formed by lipids and glycans and their combinations [25,26], which further complicates the design of immunogens capable of inducing the required B-cell response. This task is believed to be solved using peptide mimics of conformational epitopes that can be obtained using combinatorial biology (the phage display technique) [27].

Concerning studies related to the development of artificial B-cell immunogens, a protein scaffold approach should be mentioned. Such scaffolds can expose one or several epitopes of broadly neutralizing antibodies to provide the most efficient exposure of the desired epitopes to the immune system [28–32]. Epitope scaffolds developed by rational design were able to elicit 4E10 and 2F5-like antibodies in laboratory animals [28,29]. Zhu et al. proposed computationally designed epitopes that mimic carbohydrate-occluded neutralization epitopes (CONEs) of Env through 'epitope transplantation', in which the target region is presented on a carrier protein scaffold. Although a tested anti-CONE serum demonstrated a modest magnitude of inhibitory activity on HIV-1 infectivity, the consistency of the effect against multiple isolates of HIV-1 Env pseudoviruses allows us to suggest that this approach could provide a broad neutralizing antibody response [33].

Another HIV vaccine strategy is based on the use of soluble stabilized Env trimer spikes for inducing broadly neutralizing antibodies. These trimeric antigens are comprised of cleavage products of gp120 and gp41 subunits forming a native-like Env conformation exposing vulnerable sites recognized by bNAbs [3,34–36]. However, as well as scaffolds, trimers could contain undesirable epitopes, diverting the protective humoral immune response [36,37]. To date, several approaches are used to decrease the immunogenicity of such epitopes [38–41].

Considering all of the above, it seems reasonable to create an immunogen which contains only the HIV-specific epitopes crucial for inducing a protective immune response. This approach focuses the immune response specifically on protective antigenic determinants and excludes the undesirable vaccine epitopes that could induce autoreactive antibodies or antibodies intensifying viral infectivity.

The paper represents the results of a study on constructing and investigating immunogenic properties of an artificial nTBI molecule comprising epitopes recognized by bNAbs 2F5, 10E8 [42,43], and a linear peptide mimic of a conformational epitope recognized by VRC01 [44] (Figure 1).

**Figure 1.** Schematic presentation of the experimental strategy for development of the nTBI protein and studying its immunogenic properties. A spatial model of the nTBI protein structure was obtained using the I-TASSER (Iterative Threading ASSEmbly Refinement) method [45].

#### **2. Materials and Methods**

#### *2.1. Monoclonal Antibodies, Peptides, Bacterial Strains, Cell Lines, Plasmids, Media, and Bu*ff*ers*

VRC01, 10E8, and 2F5 monoclonal antibodies were kindly provided by the NIH AIDS Research and Reference Reagent Program (cat. # 12033, 12294, 1475). Murine monoclonal antibody (mAb) 29F2, *E. coli* strain BL21(DE3) pLysS (Invitrogen) and HEK 293T/17 cells (cat. # 103) as well as pTBI plasmid, encoding TBI gene (T- and B-cell containing immunogen) [46,47] were found in the collection of the State Research Center of Virology and Biotechnology Vector (SRC VB Vector, Koltsovo, Russia). TZM-bl cells (cat. # 8129) were also provided by the National Institutes of Health AIDS Research and Reference Reagent Program (USA). Synthesis of *nTBI* and *TBI\_tag* genes and further cloning into pET21a expression vector (Novagen) were performed by Evrogen Lab, Ltd. (Moscow, Russia). 10E8 [NWFNITNWLWYIK], 2F5 [NEQELLELDKWASLWNK], and VRC01 [VSWPELYKWTWS] epitope peptides were synthesized by Synpeptide Co., Ltd. (Shanghai, China).

#### *2.2. Expression and Purification of Recombinant Proteins nTBI and TBI\_tag*

The pET-nTBI and pET-TBI\_tag plasmids were maintained and transformed into *E. coli* competent cells, single colonies were picked up and grown overnight in 2xYT medium with ampicillin (100 μg/ml). An overnight culture was diluted to 1:100 and grown at 37◦C and 160 rpm until an OD600 = 0.5. Protein expression was induced with 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG). Cells were centrifuged at 6000 rpm for five minutes at 4 ◦C; pellets were resuspended in lysis buffer (0.05% tween 20, 50 mM monosodium phosphate, 300 mM sodium chloride, 30 mM imidazole) and sonicated on ice using Soniprep 150 Plus cell disruptor (16 times per minute at 13.2 μm amplitude). Inclusion bodies were pelleted by centrifugation, washed in Tris-HCl buffer (pH 8.3), and dissolved in 6 M urea for two hours. Insoluble fraction was removed by centrifugation, dissolved fraction was loaded onto a nickel nitrilotriacetic acid (Ni-NTA) column (Qiagen), washed with buffer containing 20 мM Tris-HCl, 0.5 M NaCl, 6 M urea and 20 мM imidazole (pH 7.9), and eluted with the same buffer with an increased concentration of imidazole up to 0.5 M. The proteins were then sequentially dialyzed for five hours at 4 ◦C against phosphate-buffered saline (PBS) buffer (pH 7.5) containing 6, 4, 2 and 1 M urea, respectively,

followed by final dialysis in PBS. Purity and identity of refolded soluble proteins were estimated by SDS-PAGE and Western blotting.

#### *2.3. SDS-PAGE, Western Blot, and New Lav Blot 1 Analysis*

Protein samples were analyzed by SDS-PAGE on a 15% gel using Coomassie brilliant blue staining method. Western blot analysis was performed using SNAP i.d. system (Millipore). Proteins were transferred onto a nitrocellulose membrane (Amersham), blocked with 3% BSA for 15 minutes at room temperature, and washed three times with wash buffer (PBS with 0.1% Tween 20). After washing, the membrane was probed with mAbs VRC01, 10E8, 2F5, and 29F2 diluted to 1:10,000 in wash buffer for 10 minutes at room temperature. Antibodies were washed three times with wash buffer. Secondary antibodies (anti-human or anti-mouse) conjugated with alkaline phosphatase were used at 1:10,000, diluted in wash buffer. After 10 minutes, incubation membranes were washed five times. Proteins were visualized using nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) substrate solution (Sigma, USA). HIV-1-specific antibodies were detected using a New Lav Blot 1 test kit (Bio-Rad, France), according to the manufacturer's instructions.

#### *2.4. Circular Dichroism*

Circular dichroism (CD) assays were carried out at room temperature (25 ◦C) in normal saline and in 20% trifluoroethanol using a Jasco J-600 spectropolarimeter. Spectra were recorded in the range a 195–260 nm, using a 1 mm path-length quartz cell. Each spectrum was obtained by averaging three scans with 1 nm step and 2 nm spectral bandwidth. The samples TBI\_tag and nTBI have the same optical absorption at 214 nm.

The fractions of the secondary structure elements were calculated by minimizing the difference between the theoretical and experimental curves by varying of the impacts of the α-helixes, β-sheets, turns and non-structured forms. Theoretical values at every wavelength were the linear combination of the basis spectra of every type of secondary structure [48].

#### *2.5. Immunization*

Male Chinchilla rabbits of four weeks of age (~2 kg body weight) were purchased from Vector's animal breeding facilities and housed in a certified animal facility managed by the Animal Center of SRC VB Vector. All experiments were made to minimize animal suffering and carried out in line with the principles of humanity described in the relevant Guidelines of the European Community and Helsinki Declaration. The protocol was approved by the Institutional Animal Care & Use Committee (IACUC) of the SRC VB Vector (# 03-02.2017).

Animals were randomly divided into two groups (three rabbits per group). The first group was immunized with the nTBI protein, the second with the TBI\_tag. Immunization was carried out three times (1, 14, and 28 days). Rabbits were primed with 500 μg of corresponding proteins subcutaneously with complete Freund's adjuvant (Sigma, USA), for the second immunization animals received 500 μg of protein subcutaneously with incomplete Freund's adjuvant. For the third immunization rabbits were injected 500 μg of protein without adjuvant. Serum samples were collected prior to the first immunization (pre-immune) and two weeks after the third immunization.

#### *2.6. Purification of Rabbit IgG*

To obtain antigen-specific antibodies and to eliminate non-specific effects of other serum components pooled rabbit sera were purified using protein A chromatography (BioVision, USA). Briefly, samples (9 ml) of rabbit sera were mixed with 9 ml binding buffer (1 × TBS, 0.15 M NaCl, in 50 mM sodium borate, pH 8.0). Diluted serum samples were added to a column containing Protein A agarose equilibrated with binding buffer and passed through a column. The column was washed with 10 volumes of binding buffer. IgG was eluted with 10 ml of 0.1 M citric acid, pH 2.75, and the eluate was immediately neutralized with 1.5 M Tris-HCl, pH 8.8 (150 μl per 1 ml of eluate). For each

sample, IgG fractions with the highest protein concentration were pooled and dialyzed three times at 4 ◦C against 1 × phosphate-buffered saline. IgG purity was assessed by the Coomassie staining of 14% SDS-PAGE gel and was found to be equal to or greater than 90%. The antibody concentrations were quantified by measuring the absorbance at 280 nm using a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, USA), and then stored at – 20◦C.

#### *2.7. ELISA*

ELISA was used to identify Ab responses of individual rabbits to nTBI and TBI\_tag. MaxiSorp 96-well plates (Thermo Fisher Scientific, USA) were coated with 5 μg/ml of nTBI and TBI\_tag overnight at 4 ◦C in PBS. Plates were blocked for two hours with 0.5% casein in PBS at 37 ◦C and washed three times with 0.05% Tween 20/PBS (PBS-T). Eight serial dilutions of sera samples in PBS-0.1% casein (1/200, 1/1000, 1/5000 ... 1/15,625,000) were added to the wells (100 μl per well) and incubated for two hours at 37 ◦C with shaking. Plates were washed three times with 0.05% PBS-T. Goat anti-rabbit IgG conjugated with alkaline phosphatase (Sigma, USA) were then added to the wells (100 μl/well) at 1:5000 dilution in PBS. Plates were incubated for one hour at room temperature and washed five times with 0.05% PBS-T. To visualize immunogen-specific antibodies BCIP/NBT substrate (Sigma, USA) was added to each well and incubated overnight in the dark at 37 ◦C. Plates were read at 405 nm (Model 680 Microplate reader, Bio-Rad, USA). Optical density (OD) values were calculated for each group by subtracting two times the average background of pre-immune rabbit serum.

#### *2.8. Production of Env-Pseudoviruses*

The reference panel of recombinant plasmids containing full-length Env genes of HIV-1 subtype B [49], HIV-1 Env clone SF162.LS [50], 6535 clone 3 (Cat # 11017), QH0692 (Cat # 11018), TRO.11 (Cat # 11023), PV04 (Cat # 11022), backbone Env-deficient plasmid pSG3Δenv was obtained through the NIH AIDS Reagent Program, Division of AIDS, NIAID, NIH. Plasmids phGPMΔMD-1 encoding GP of Marburg virus were constructed at SRC VB Vector.

HIV-1 Env-pseudotyped were prepared as described in [51]. Briefly, 293T cells in a T-75 culture flask were cotransfected with 10 μg of plasmids encoding heterologous envelope variants and 10 μg of an Env-deficient pSG3 env backbone plasmid using MATra reagent (PromoKine, Germany). Pseudoviruses were collected after a 48-hour incubation, filtered through a 0.45 μm pore size filter (TPP, Switzerland) and stored at –80 ◦C until use. Lentiviral particles pseudotyped with Marburg envelope glycoprotein (MGP) were used as a specificity control virus in the neutralization assay. The median tissue culture infective doses (TCID50) of each pseudovirus was calculated by the Reed and Muench method to be used in a neutralization assay.

#### *2.9. Neutralization Assay*

Neutralization activities of immunogen-specific rabbit-derived antibodies were measured in the decrease of Tat-induced *Luc* reporter gene expression after a single round Env-pseudotyped HIV-1 infection of TZM-bl cells, as described previously [14,51]. Briefly, antibodies were serially diluted in a four-fold stepwise manner (initial concentration 4 mg/ml) and mixed with 50 μL of 200 TCID50 of pseudovirus in 96-well flat-bottom culture plates (Corning-Costar) in triplicate and incubated for 30 minutes at 37 ◦C. Purified IgG (initial concentration 4 mg/ml) from the pool of pre-immune sera were used as a negative control. mAbs 4E10, 2F5, and VRC01 (initial concentration 30, 25, and 10 μg/ml, respectively) were used as a positive control. The mixture was transferred to TZM-bl cells (104 per well) and incubated for 48 hours at 37 ◦C. Following a 48-hour incubation, 150 μl of culture medium was removed from each well and replaced with a *Luc* reporter gene assay system reagent. After a two-minute incubation at room temperature to allow cell lysis, 150 μl of cell lysate were transferred to 96-well black solid plates for measurements of luminescence using a LuMate 4400 microplate luminometer (Awareness Technology Inc., Palm City, F L, USA) (expressed as relative luminescence units (RLUs) equivalent). The percentage decrease expressed in RLU (% neutralization) was determined

by calculating the difference in average RLU between test wells (cells + sample + virus) and cell control wells (cells only), dividing this result by the difference in average RLU between virus control (cell + virus) and cell control wells, subtracting from 1 and multiplying by 100. To quantify neutralizing activities of rabbit antisera, the concentration of Ab required to obtain 50% neutralization (the IC50) was calculated. All the assays were carried out in triplicate, statistical analysis was performed using GraphPad Prism 6 software (GraphPad, USA) to calculate concentrations of antibodies (μg/ml).

#### *2.10. Competition Neutralization Assay*

10E8 [NWFNITNWLWYIK], 2F5 [NEQELLELDKWASLWNK], and VRC01 [VSWPELYKWTWS] peptides were synthesized by Synpeptide Co., Ltd. (Shanghai, China) with >80% purity and tested for their ability to compete with HIV-1 Env-pseudotyped virus SF162 for binding to anti-nTBI IgG, anti-TBI\_tag IgG, and mAbs 10E8, 2F5, and VRC01. Competition assay was carried out as described previously [52]. Briefly, serial dilutions of Ab were tested and IC 90 values for SF162 were determined by linear regression. Next, 1:1 (v/v) mixtures of each peptide (2 mg/ml) and corresponding antibodies were incubated for 30 minutes at 37 ◦C, added to SF162 pseudovirus (200 TCID50) and coincubated for one hour at 37 ◦C. Further, a mix of the peptide, antibodies, and SF162 was added (1:1 v/v) to the TZM-bl indicator cell line. Luciferase activity was detected 48 hours after infection assay reagent II (Promega, USA). To control for nonspecific inhibition of the interaction between SF162 and target cells by the peptides, a neutralization assay was performed in the absence of antibodies (negative control). The inhibiting effect of peptides was detected in regards to log10 RLU data obtained in the neutralization of SF162 pseudovirus with respective antibodies without the addition of peptides and log10 RLU data obtained with the addition of peptides.

#### *2.11. Statistical Analysis*

A non-parametric Mann–Whitney U-test was used for multiple comparisons of the interactions between corresponding peptides and antibodies. Multiple testing correction was performed according to the Benjamini–Hochberg procedure (FDR). Statistical analysis and plotting were carried out using R, the open source statistical analysis language and environment (v. 3.3.2) [53].

#### **3. Results**

#### *3.1. Design and Construction of Plasmids Encoding Immunogens TBI\_tag and nTBI*

In order to assemble the polyepitope nTBI protein, previously designed TBI (T- and B-cell immunogen) was used as a frame [46,47]. Several steps were made to improve the immunogenic properties of this modified version. Firstly, to enhance protein expression, the *TBI* gene sequence was codon optimized using the GenScript Rare Codon Analysis Tool [54]. Next, we added an *E. coli infB* gene fragment encoding N-terminal expressivity tag into the *TBI* gene's structure [55]. The codon-optimized and expressivity tag-fused *TBI* gene was cloned into pET21a expression vector in-frame with C-terminal 6x-HIS tag, which allowed the purification of the corresponding protein by Ni-NTA chromatography. The genetic construction modified in this manner was named TBI\_tag.

Next, we decided to modify the TBI\_tag peptide composition by replacing three B-cell epitopes that have weak capacity to elicit virus neutralizing antibodies (according to the Los Alamos HIV Database) with the epitopes recognized by bNAbs. For this purpose, Env (255–266) and Gag (92–109) epitopes that have a predominantly alpha-helical structure were replaced with the 10E8 bNAb-specific epitopes NWFNITNWLWYIK and NEQELLELDKWASLWNK, recognized by bNAb 2F5. Both of these sequences were predicted as alpha-helical; besides, Env (255–266) and Gag (92–109) epitopes are located at the N- and C- termini of TBI, so we assumed that their replacement would not dramatically disrupt the protein backbone.

The remaining Gag (351–361) epitope had an irregular secondary structure and was surrounded by two alpha-helical regions in the context of TBI protein. Instead of this epitope, we decided to put the phage display-selected linear peptide mimic of the epitope recognized by the VRC01 antibody [52]. Similar to Gag (351–361), this peptide has an irregular structure, therefore it was assumed that incorporation of the peptide between alpha-helical motifs could give it a certain rigidity and the ability to be exposed on the protein's surface.

The proteins' structures were predicted using the I-TASSER [45] and PSSpred [56] prediction tools, proteins models were shown in Figures 1 and 2, respectively. According to modeling data, the constructed immunogens TBI\_tag and nTBI have a structure of mainly α-helical regions and flexible interhelical loops.

**Figure 2.** Depicting secondary structure of TBI\_tag and nTBI proteins. The structures were predicted using the PSSpred tool available at [56]. bNAbs epitopes that were appended into the nTBI protein are highlighted with blue; T-h epitopes are marked with red; α-strings—light green cylinders; β-coils—grey arrows; unshaped structures—dashed lines. All proteins comprised a C-terminal sequence from histidine amino acid residues.

#### *3.2. Expression, Purification, and Antigenic Properties of Recombinant nTBI and TBI\_tag Proteins*

The designed and synthesized *TBI\_tag* and *nTBI* genes were cloned in plasmid pET21a and expressed in BL21(DE3) pLysS E. coli cells. Expression of recombinant proteins encoded by the cloned genes was characterized by SDS-PAGE, followed by Coomassie blue staining (Figure 3A).

It was revealed that recombinant proteins TBI\_tag and nTBI were mainly expressed as inclusion bodies; therefore, we used a purification scheme including the sonication of bacterial cells, solubilizing the inclusion bodies by urea solution, affinity chromatography using Ni-NTA resin and refolding. It was established that the modified TBI\_tag can be successfully produced and purified in soluble form, while. nTBI tended to form aggregates after purification step (the solution was opalescent). The of isolated TBI\_tag and nTBI proteins were further checked by SDS-PAGE using Coomassie blue staining (Figure 3B).

To ensure that Coomassie-stained bands correspond to the target proteins, Western blot analysis was performed using the mouse monoclonal antibody 29F2 that recognizes the peptide from HIV-1 p24 [EPFRDYVDRFYKTLR], which is a part of both nTBI and TBI-tag. It was demonstrated that both proteins specifically bind to 29F2 (Figure 3C).

To check if all the substituted epitopes retained their antigenic properties in the context of a recombinant protein, we performed Western blot analysis using 10E8, 2F5, and VRC01 mAbs as primary antibodies. It was shown that TBI\_tag selectively binds to 29F2 mAb as expected, whereas nTBI reacts with all the mAbs used in screening except for VRC01 (Figure 3C).

**Figure 3.** (**A**) SDS PAGE analysis of *E. coli* BL21 (DE3) pLysS cell lysate after IPTG induction. 1) molecular weight marker; 2) lysate of E. coli BL21/pET-nTBI cells; 3) lysate of E. coli BL21/pET-TBI\_tag cells; 4) lysate of non-transformed E. coli BL21 cells. (**B**) SDS PAGE analysis of purified samples of nTBI and TBI\_tag; 1) molecular weight marker; 2) purified nTBI protein; 3) purified TBI\_tag protein. (**C**) Western blot analysis of expressed nTBI and TBI\_tag. Purified proteins were separated by SDS-PAGE in 15% gel and transferred onto a nitrocellulose membrane (nTBI at lane 1 and TBI\_tag at lane 2). Monoclonal antibodies 29F2, 2F5, 10E8, and VRC01 were used for immunodetection.

#### *3.3. Circular Dichroism*

Experimental analysis of the polypeptides' secondary structures was performed using circular dichroism spectroscopy. CD spectra of TBI\_tag and nTBI were both registered in saline and in 20% trifluoroethanol in saline (Figure 4). The quantitative analysis of the secondary structure of TBI\_tag shown the same fraction of α-helices and β sheets in the saline. Significant increases of α-helices and decreases of β sheets and non-structured forms were observed after the addition of 20% trifluoroethanol, which stabilizes the secondary structure of the protein (Table 1). The CD signal magnitude was close to zero for nTBI in the studied conditions.


**Table 1.** Circular dichroism spectra of the TBI\_tag protein.

Unordered structures 70% 37%

**Figure 4.** CD spectra of TBI\_tag protein: (**A**) saline; (**B**) 20% trifluoroethanol in saline.

#### *3.4. Assessment of Immune Response in Rabbits*

To assess the immunogenicity of recombinant proteins, ELISA was performed using sera from immunized rabbits. Pre-immune sera were used as a negative control. It was found that sera from both groups of immunized rabbits contain antigen-specific antibodies. Average antibody titers in both groups of animals immunized with TBI\_tag as well as nTBI were more than 1:3,125,000 after the third immunization. Furthermore, there was no significant difference in serum level between rabbits from one group. At the same time, the signal of pre-immune sera in ELISA was no more than the background level.

Further, sera were tested for the presence of antibodies specific to HIV-1 proteins (Figure 5). For the purpose, we used a commercially available test system, New Lav Blot 1 (Bio-Rad), including nitrocellulose stripes coated with HIV-1 proteins. Pre-immune sera were used as a negative control, while inactivated human serum containing anti-HIV-1 antibodies (from the New Lav Blot 1 kit) served as a positive control.

**Figure 5.** Western blot of sera from animals immunized with TBI\_tag and nTBI using the New Lav Blot 1 test system. 1) Positive control from the New Lav Blot kit. 2) Pooled sera of rabbits immunized with TBI\_tag. 3) Pooled sera of rabbits immunized with nTBI. 4) A pool of pre-immune sera of rabbits from both groups.

It was shown that antisera raised in nTBI-immunized rabbits contain antibodies against gp120, gp160, gp41, and p24 (p55), whereas TBI\_tag-induced antibodies were additionally recognized p17. These results are consistent with our expectations, as nTBI doesn't bear the p17 HIV-1 epitope which was replaced with the 2F5 epitope from the membrane-proximal external region (MPER).

#### *3.5. Neutralizing Activity of Purified Specific Antibodies*

IgGs purified from pool sera of rabbits immunized with TBI\_tag and nTBI were tested for the capability of neutralizing HIV-1 pseudoviruses. Env-pseudotyped viruses from the global panel of reference Env clones (NIH) were packaged and used to perform a neutralization assay.

At the first step, we used easy-to-neutralize (tier 1) SF162.LS pseudovirus. It was shown that both anti-nTBI and anti-TBI\_tag antibodies efficiently neutralized the tier 1 pseudotyped HIV-1 strain, while IgGs from pre-immune sera showed no neutralization activity (Figure 6). In addition, we found out that IC50 anti-nTBI antibodies was 0.44 μg/mL, approximately six times lower than the IC50 for anti-TBI\_tag antibodies (2.76 μg/mL). mAbs 2F5 and VRC01 (Figure 6) were used as positive controls. Next, we performed a neutralization assay using tier 2 and tier 3 Env-pseudotyped HIV strains (6535, QH0692 иPV04). A similar trend was observed for these pseudoviruses with percent neutralization being higher for anti-nTBI IgG compared to the anti-TBI\_tag IgG. However, since the mean IC50 values were more than 50 μg/mL, we cannot assume virus neutralizing activity relating to the used pseudoviruses.

**Figure 6.** Neutralizing activity of IgGs isolated from sera of immunized animals against SF162.LS pseudovirus. The figure depicts neutralization curves for anti-TBI\_tag (**A**), anti-nTBI (**B)** rabbit IgGs, as well as for human mAbs 2F5 (**C**) and VRC01 (**D**). 2F5 and VRC01 were used as positive controls. A pool of IgGs purified from pre-immune sera of rabbits (IgGs-pre-immune) was used as a negative control. Lentiviral particles pseudotyped with Marburg envelope glycoprotein (MGP) were used as a specificity control virus in the neutralization assay.

#### *3.6. Competition Neutralization Assay*

Next, we decided to test whether bNAb epitopes that were included into nTBI are able to interfere with the neutralization activity of corresponding antibodies. We carried out competitive inhibition of the neutralization of the Env-pseudotyped SF162.LS HIV strain by 10E8, VRC01 and 2F5 mAbs, and purified IgG from immune sera using synthetic peptides 10E8 [NWFNITNWLWYIK], 2F5 [NEQELLELDKWASLWNK], and VRC01 [VSWPELYKWTWS].

Preliminary experiments revealed that these peptides are non-toxic for TZM-bl cells and induce no non-specific inhibition of the interaction between pseudovirus and target cells.

10E8 peptide showed the inhibition of neutralizing activity of anti-nTBI antibodies against SF162.LS (*p* < 0.01). Additionally, inhibition of neutralization of TBI\_tag-specific antibodies by this peptide was not observed (Figure 7A). Predictably, 10E8 peptide epitope blocked neutralization of SF162.LS by 10E8 (*p* < 0.01) (Figure 7A).

A peptide mimic of the VRC01 epitope inhibited neutralization of SF162 by VRC01 (*p* < 0.01), but weakly blocked neutralization of SF162 by anti-nTBI antibodies (*p*<0.20), and not inhibit TBI\_tag-specific antibodies (Figure 7C). As expected, the 2F5 epitope interfered with 2F5 mAb and decreased its neutralizing activity against SF162, whereas we observed no inhibition of neutralizing activity of both anti-nTBI and anti-TBI\_tag antibodies (Figure 7B).

**Figure 7.** Inhibition of virus-neutralizing activity of antibodies by synthesized peptides: (**A**) 10E8 [NWFNITNWLWYIK], (**B**) 2F5 [NEQELLELDKWASLWNK], (**C**) VRC01 [VSWPELYKWTWS]. The following combinations were tested in the assay: cell + virus; cell + virus + monoclonal mAb 10E8 or 2F5, or VRC01; cell + virus + monoclonal mAb 10E8 or 2F5, or VRC01 + corresponding synthetic peptide; cell + virus + anti-nTBI IgG; cell + virus + anti-nTBI IgG + synthetic peptide; cell + virus + anti-TBItag IgG; cell + virus + anti-TBItag IgG + synthetic peptide. Relative luminescence intensity is indicated on the Y-axis (log10 RLU). Statistical differences between groups cell + virus + mAb 10E8, or 2F5, or VRC01 and cell + virus + mAb 10E8, or 2F5, or VRC01 + synthetic peptide is *p* < 0.01; statistical differences between groups cell + virus + anti-nTBI IgG and cell + virus + anti-nTBI IgG + synthetic peptide is *p* < 0.01.

#### **4. Discussion**

Eroshkin et al. have previously designed an artificial polyepitope TBI protein (T- and B-cell epitopes containing immunogen) composed of conservative epitopes from Env and Gag HIV-1 and based on a well-known protein structural motif, i.e., a four-helix bundle [47]. TBI included four Th-cell epitopes (amphipathic α-helix) and five B-cell epitopes (regions with flexible hydrophilic loops) [46,47]. The rationale for the TBI design was that combining T- and B-cell epitopes in one construct would stimulate both proper B-cell and T-cell responses and the necessary interplay between B- and T-cells. The TBI protein had a CD spectra similar to α-helical proteins and showed crystal-yielding capacity, which was demonstrated for the first time in an artificial protein with a predicted tertiary structure [57].

It was shown that TBI induced both cellular and humoral responses to HIV-1 in immunized mice and rhesus macaques, and TBI-induced antibodies showed virus-neutralizing activity to HIV-1 [58]. Lately, TBI was included into the composition of the CombiHIVvac candidate vaccine that had undergone phase I clinical trials [24,58].

Based on its ability to crystallize, we assumed that the TBI protein structure was similar to that of the natural protein. We decided to design a modified protein based on TBI with an enhanced ability to induce HIV-neutralizing antibodies.

First, in order to increase the yield of recombinant protein, we changed the expression vector to pET21a, which led to an increase of gene expression level and allowed the use of the 6xHis/Ni-NTA system for protein purification. Additionally, we added an *E. coli infB* gene fragment encoding an N-terminal expressivity tag to its structure [55]. This protein was referred to as TBI-tag and used for further modifications. At the next step, we replaced three B-cell epitopes from TBI-tag with epitopes recognized by bNAbs. We substituted the Env (255–266), Gag (99–109), and Gag (351–361) epitopes (Figure 2). The first two epitopes were replaced with peptides from the MPER region, i.e., (NEQELLELDKWASLWNK) and (NWFNITNWLWYIK), recognized by bNAbs 2F5 and 10E8, respectively. Instead of Gag (351–361), we added a phage display-selected linear peptide mimic of the epitope recognized by the VRC01 antibody, into the protein structure [52]. According to the PSSpred tool modeling data, the epitopes for 10E8 and 2F5 have retained their peculiar α-helical structure, which was shown to be required for their antigenic and immunogenic properties (Figure 2).

T-helper epitopes and spacer sequences between epitopes remained unchanged, since they form amphipathic α-helices which seem to stabilize protein structure (Figure 2). Significance of the helical amphipathicity of epitopes for the recognition of T-cells is evident, since 70% of T-helper epitopes have an α-helical conformation. Previously, it was shown that TBI elicited HIV-specific T-cell response in mice, which supported the idea that TBI processing as well as presentation of T cell epitopes were adequate [46]. Since antigen-presenting cells present T-cell epitopes as linear peptides after antigen processing, we suppose that the functional activity of T-helper epitopes in the context of nTBI should have remained at about the same level as in TBI.

Modeling the secondary structure enabled us to assume that accomplished modifications didn't significantly affect the recombinant protein organization, since the predicted secondary structures of TBI\_tag and nTBI were found to be similar (Figures 1 and 2).

To evaluate the secondary structure of the TBI\_tag and nTBI proteins, we used a circular dichroism spectroscopic method. CD spectra analysis of the proteins in a 20% trifluoroethanol solution revealed that TBI\_tag contains 54% α-helices and 4% β-sheets, which is consistent with theoretical data predicted with PSSpred; 52% and 2%, respectively. The corresponding predicted values for nTBI were the following: 59% of α-helices and 3% of β-sheets. However, we failed to determine its secondary structure using CD. The probable reason is that the inclusion of the peptide mimic of the VRC01 conformational epitope into an internal area of the molecule TBI\_tag resulted in a destabilizing of the protein structure and changing of its physico-chemical properties, since it became prone to aggregation. Western blot analysis revealed that the mAb VRC01 poorly interacted with the peptide mimic as part of nTBI as compared to the same peptide as part of protein p3 of bacteriophage M13 [52]. Two other

peptides, 2F5 and 10E8, included in the nTBI molecule are effectively recognized by their corresponding MPER bNAbs (2F5 and 10E8) (Figure 3A).

Antibody titers of rabbits' antisera derived after three-fold immunization with both TBI-tag and nTBI were more than 1:3,125,000, which indicates the high immunogenicity of both proteins. New Lav Blot 1 analysis showed that induced antibodies in rabbits by immunization with TBI\_tag and nTBI were capable of binding to HIV-1 proteins (Figure 5). Consequently, both immunogens were able to elicit HIV-specific antibodies.

The analysis of virus neutralizing activity of sera from animals immunized with both TBI\_tag and nTBI revealed that both proteins elicit antibodies capable of neutralizing tier 1 SF162.LS. Additionally, anti-nTBI IgG neutralizing activity proved to be higher than that of the antibodies induced with TBI\_tag. The IC50 of anti-nTBI antibodies was six times lower than the IC50 for anti-TBI\_tag antibodies (Figure 6). To test whether the neutralizing activity of anti-nTBI IgGs from rabbits' immune sera is mediated by the antibodies which were elicited against substituted epitopes, we performed a peptide competition neutralization assay.

The 10E8 peptide was the most immunogenic, since its inhibition capacity of anti-nTBI IgG significantly differed from control (Figure 7A). Thus, enhancement of virus neutralizing activity of antibodies induced by nTBI seems to be connected to the inclusion of peptide 10E8 in the compound of the TBI protein.

VRC01 mimotope VSWPELYKWTWS contributes poorly to the induction of neutralizing antibodies. When immunized with the nTBI protein, the neutralizing activity of the antibodies induced against this mimotope decreased after the addition of the VSWPELYKWTWS peptide; although the difference between the control group was not significant (Figure 7C). Probably, this may be explained by a shielding of the peptide mimic by flanking amino acid sequences in the nTBI compound. Thereby, placing the peptide mimetic of the VRC01 epitope into the internal region of nTBI could alter the immunogenicity of this peptide. Importantly, the peptide mimic of the VRC01 epitope efficiently inhibited the binding of bNAbs VRC01 to the SF162 pseudovirus, which is consistent with the literature [52].

Concerning 2F5 peptide, we failed to demonstrate inhibition of SF162 neutralization by IgGs from rabbits immunized with TBI\_tag and nTBI. The epitope recognized by the bNAb 2F5 was designed at the protein C-terminus, and was followed by an additional six histidine residues. Inclusion of this hexahistidine tag helped streamline protein purification, yet it could also affect the structure and accessibility of the adjacent epitope. This could be one of the possible reasons why the 2F5 epitope retained antigenicity in the context of the nTBI, but failed to induce 2F5-like Abs (as shown by the competition analysis using the synthetic 2F5 peptide). It seems that 2F5 epitope heavily relies on its natural environment to expose its immunogenic features [29]. The obtained findings will be considered in the further enhancement of TBI immunogen. It is possible that removing the histidines from the C-terminus of nTBI will result in efficient presentation of the 2F5 peptide.

Another possible explanation of low neutralizing activity of the immune sera could be the animal model that we used for the experiment. Although rabbits can be used for characterization of HIV-1 immunogenicity, they are not a valid model for the analyses of novel immunogens driving an anti-HIV response. One reason is that they are naturally resistant to HIV infection, and more importantly, sequences of their immunoglobulin genes are too diverged from those of primates. In our future experiments testing the elicitation of neutralizing antibodies, common marmosets will be used as the model animals.

#### **5. Conclusions**

Approach to the development of artificial proteins constructed using conservative T- and B-cell epitopes and their mimics is believed to be promising for the development of vaccines against variable viruses, including HIV. In such constructs, all of the component helper T cell epitopes may be expected to provide T cell help for antibody production against all of the component B cell epitopes, thus overcoming the limitation of genetic restriction [46]. In theory, this approach makes it possible to evade the antigenic variability of the virus, focus the immune responses on protective determinants, and exclude from the vaccine compound undesired determinants capable of inducing autoantibodies or antibodies increasing virus infectivity.

In this study, we used a polyepitope-based HIV immunogen design strategy to develop an artificial nTBI protein exposing epitopes recognized by bNAbs 2F5, 10E8, and a phage display-selected peptide mimic of the VRC01 discontinuous epitope. Antigenic properties of the incorporated peptides were saved in the compound of nTBI, except for the peptide mimic VRC01. nTBI became less soluble as compared to the initial TBI. This is probably because substitution of the initial TBI B-cell epitopes by new ones negatively affected its conformation, despite the prediction results. However, compared to the initial TBI, nTBI demonstrated a better capacity to induce virus neutralizing antibodies in comparison with the initial TBI, at least for the SF162 pseudovirus the neutralizing activity of nTBI-induced IgGs outperformed that of the TBI\_tag-induced IgGs (the IC50 of anti-nTBI antibodies was six times lower than the IC50 of anti-TBI\_tag antibodies (Figure 6)). Competition assay revealed that immunization of rabbits with nTBI induced mainly 10E8-like antibodies, whereas no 2F5 epitope and VRC01 mimotope-induced antibodies were detected. We considered several strategies for boosting the antigenicity and immunogenicity of nTBI. The peptide mimetic targeted by VRC01 can be placed at the C- or N- terminus of TBI, which may translate into the induction of neutralizing antibodies. Our earlier results [52] indicate that the peptide mimetic selected using phage display was able to induce neutralizing antibodies, either in the context of the M13 phage or as individual synthetic peptide. Furthermore, following affinity selection, we obtained many more VRC01 mimetics, and these may be considered viable alternatives for future modifications of TBI.

As for the 2F5 epitope, testing whether the C-terminal hexahistidine tag affects its immunogenicity remains an attractive experiment to do. Also, one may think of increasing the spacers between the epitopes, as this may translate into greater accessibility of the epitopes to B-cell receptors. Establishing a broader range of TBI derivatives may therefore be highly productive, as this may help identify the variants showing greater activity. Finally, the lead immunogen should be thoroughly tested in various prime-boost immunization schemes. The ultimate goal of this approach is to obtain an anti-HIV vaccine component that would provide the desired neutralizing immune response. It is possible that such an immunogen will comprise a cocktail of several lead TBI versions.

**Author Contributions:** Conceptualization, A.P.R., A.N.C., A.A.I., and L.I.K.; Data curation, A.P.R. and N.B.R.; Formal analysis, D.V.A.; Investigation, A.P.R., N.B.R., A.A.L. and O.N.K.; Methodology, A.N.C.; Project administration, L.I.K.; Visualization, A.P.R., A.A.I., and L.I.K.; Writing—Original Draft, A.P.R., A.N.C., A.A.I., and L.I.K.

**Funding:** The work was supported by RFBR Grant N 18-29-08051 mk and RFBR and the government of the Novosibirsk region of the RF, grant N 18-44-543017.

**Acknowledgments:** The authors are grateful to Alexey Eroshkin for his helpful advice. The authors also thank Dmitrii Shcherbakov and Nadezhda Shcherbakova for help with discussion the results.

**Conflicts of Interest:** The authors declare no conflict of interest, financial or otherwise.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## **Parenterally Administered P24-VP8\* Nanoparticle Vaccine Conferred Strong Protection against Rotavirus Diarrhea and Virus Shedding in Gnotobiotic Pigs**

**Ashwin Ramesh 1, Jiangdi Mao 1, Shaohua Lei 1, Erica Twitchell 1, Ashton Shiraz 1, Xi Jiang 2,3, Ming Tan 2,3 and Lijuan Yuan 1,\***


Received: 16 August 2019; Accepted: 24 October 2019; Published: 6 November 2019

**Abstract:** Current live rotavirus vaccines are costly with increased risk of intussusception due to vaccine replication in the gut of vaccinated children. New vaccines with improved safety and cost-effectiveness are needed. In this study, we assessed the immunogenicity and protective efficacy of a novel P24-VP8\* nanoparticle vaccine using the gnotobiotic (Gn) pig model of human rotavirus infection and disease. Three doses of P24-VP8\* (200 μg/dose) intramuscular vaccine with Al(OH)3 adjuvant (600 μg) conferred significant protection against infection and diarrhea after challenge with virulent Wa strain rotavirus. This was indicated by the significant reduction in the mean duration of diarrhea, virus shedding in feces, and significantly lower fecal cumulative consistency scores in post-challenge day (PCD) 1–7 among vaccinated pigs compared to the mock immunized controls. The P24-VP8\* vaccine was highly immunogenic in Gn pigs. It induced strong VP8\*-specific serum IgG and Wa-specific virus-neutralizing antibody responses from post-inoculation day 21 to PCD 7, but did not induce serum or intestinal IgA antibody responses or a strong effector T cell response, which are consistent with the immunization route, the adjuvant used, and the nature of the non-replicating vaccine. The findings are highly translatable and thus will facilitate clinical trials of the P24-VP8\* nanoparticle vaccine.

**Keywords:** rotavirus nanoparticle vaccine; gnotobiotic pigs

#### **1. Introduction**

Human rotavirus (HRV) is a leading cause of severe, dehydrating gastroenteritis in children under five years of age. Although two live-attenuated oral vaccines, RotaTeq® and Rotarix®, have been implemented as part of national vaccination programs in over 98 different countries [1], their vaccine efficacy was reported to be lower in low- and middle- income countries (LMICs [39%–70%]) compared to that in high-income countries (80%–90%) [2–7]. A combination of factors have been suggested to be responsible for the lower efficacy, which have been reviewed and discussed in detail by Arnold [8] and Desselberger [9]. Both these live vaccines have also been linked to a low risk of intussusception (1–7 cases per 100,000 vaccinated infants) as a result of the vaccine rotavirus (RV) strains replicating in the intestines [10,11], and have remained expensive even after Gavi, the Vaccine Alliance, subsidization [12], particularly in resource-deprived countries. These scenarios have increased the demand for a safer, more cost-effective, and more efficacious vaccine, especially in LMICs that can be easily administered.

Parenteral intramuscular (IM) vaccines have been preferred to oral vaccines due to their increased immunogenicity. They are also not directly affected by microbiome composition or gut enteropathy, both of which have been known to affect the efficacy of oral vaccines in LMICs [13]. Non-replicating rotavirus vaccines (NRRV) have been proposed as safer alternatives to live-attenuated vaccines as they do not lead to intussusception due to their parenteral immunization route [14].

The VP4 region of RV constitutes surface spike proteins that are cleaved by host intestinal proteases into two fragments, VP5\* and VP8\*. VP8\* forms the distal portion of the VP4 spikes, interacting with glycan receptors to facilitate viral attachment and entry [15,16]. VP8\* expressed in various culture systems has been explored as an immunogen in rotavirus vaccine development [17–22]. VP8\*-containing vaccine candidates have been shown to induce rotavirus-specific neutralizing antibodies and/or protection in mouse, guinea pigs, and gnotobiotic (Gn) pig models [23–28]. Among those, the leading candidate is the P2-VP8 (VP8\* fused to a universal tetanus toxin CD4+ T cell epitope P2) vaccine, adsorbed with aluminum hydroxide for IM administration, which has progressed to phase 1/2 human clinical trials [19,29].

The norovirus (NoV) P particle, referred to as the P24 particle, is an octahedral nanoparticle (≈840 kDa) composed of 24 copies of the protrusion (P) domain of the NoV capsid protein. It can be easily produced in large quantities using an *E. coli* expression system. The distal surface of each P domain, corresponding to the outermost surface of the P particle, contains three surface loops, which can tolerate large sequence insertions. Based on this concept, a nanoparticle vaccine was developed by inserting the HRV VP8\* antigen into the loop sections of the P domains. The P24-VP8\* nanoparticle consists of a 24-valent core of NoV P particle and 24 surface-displayed HRV VP8\*s. The P24-VP8\* nanoparticle shares the features of the P24 particle in self-formation, easy production, and high stability over a wide range of temperatures [30]. Efficacy studies in mice revealed that the P24-VP8\* nanoparticle vaccine is highly immunogenic and capable of inducing a significantly higher VP8\* specific antibody response as compared with free VP8\* particles even without adjuvant [30].

The main objectives of this study were to assess the immunogenicity and protective efficacy of a novel P24-VP8\* nanoparticle vaccine using the gnotobiotic (Gn) pig model of human rotavirus infection and disease. The Gn pig model of HRV (Wa, G1P [8]) infection and diarrhea has been well established and used in the pre-clinical evaluation of HRV vaccine efficacies [31]. No other conventional lab animals develop diarrhea after HRV inoculation [32]. Pigs are genetically, physiologically, anatomically, and immunologically similar to humans [33–35], allowing data from Gn pigs to be translated to humans. The immunogenicity and protective efficacy of the P24-VP8\* nanoparticle vaccine were determined using the Gn pig model of HRV infection and disease. High serum IgA, IgG, and virus-neutralizing (VN) antibody titers, as well as HRV-specific IFN-γ producing T cells, have been correlated with protection from HRV infection and disease, and data has been demonstrated to be comparable in Gn pigs and human studies [24,36,37].

#### **2. Materials and Methods**

#### *2.1. Human Rotavirus*

The virulent HRV (VirHRV) inoculum consisted of a pool composed of intestinal contents collected from the 27th passage in Gn pigs of the Wa strain HRV based on successive passages carried out in Gn pigs. A total of 1 <sup>×</sup> <sup>10</sup><sup>5</sup> fluorescent focus-forming units (FFUs) of VirHRV were diluted in 5 mL of Diluent #5 [minimal essential media (MEM, ThermoFisher Scientific); 100 IU of penicillin per mL, 0.1 mg of dihydrostreptomycin per ml; and 1% HEPES] for the inoculation of each Gn pig. The median infectious dose (ID50) and median diarrhea dose (DD50) of the VirHRV in Gn pigs were determined as approximately 1 FFU [38].

The cell culture-adapted HRV Wa strain (AttHRV), derived from the 35th passage in African green monkey kidney cells (MA104, ATCC# CRL-2378.1) [38,39], were used as the positive control for the assessment of RV antigens in feces using enzyme-linked immunosorbent assay (ELISA). The origination and passage history of the VirHRV and AttHRV have been explained by Wentzel et al. [40].

#### *2.2. Vaccine*

The P24-VP8\* vaccine was comprised of 200 μg of P24-VP8\* proteins and 600 μg Aluminum hydrogel (Al(OH)3) adjuvant. The vaccine was stored at 4 ◦C (up to 8 months) until administered to Gn pigs (Supplementary Figure S1). The dosage of the P24-VP8\* vaccine was selected based on a similar VP8\* molar amount of the P2-VP8 subunit vaccine used in the clinical trial [19,27,29]. The VP8\* region used in this vaccine was designed based on the sequence of Wa HRV. As the negative control, the Al(OH)3 adjuvant (G-Biosciences, St. Louis, MO, USA) was diluted in sterile PBS to form a final concentration of 600 μg/mL and stored at room temperature, as per manufacturer instructions, until administered.

The purified P24-VP8\* proteins were used as the detector antigen in ELISA for the detection of serum IgA and IgG antibody responses [41] and as stimulating antigen in the intracellular IFN-γ staining assay [39,42].

#### *2.3. Gn Pigs and Treatments*

Pigs (Large white cross breed) used in this study were surgically derived by hysterectomy and maintained in sterile isolators, as described previously [43]. The sterility status of the pigs housed in the gnotobiotic isolators was confirmed by culturing isolator swabs and pig rectal swabs on blood agar plates and in thioglycolate broth first at 3 days after derivation and then repeated once a week until the end of the study. Pigs were fed on a diet that solely consisted of commercial UHT sterile whole cow's milk (The Hershey Company, Hershey, PA, USA) until post-inoculation day (PID) 21, and were switched over to Similac® baby formula (Abbott Laboratories, Chicago, IL, USA) until the end of the study.

A total of 25 pigs were assigned to two groups, and a subset of pigs in each group were euthanized either pre-challenge (PID 28) or at post-challenge day (PCD) 7 (Table 1).


**Table 1.** Assignment of treatment groups for gnotobiotic (Gn) pigs.

Pigs were administered IM with an equal volume (1 mL) of either P24-VP8\* vaccine formulated with adjuvant or adjuvant alone at 5 days of age (PID 0), followed by two booster doses at PID10 and PID21. The Phase I and Phase II clinical trials carried out to evaluate the effects of P2-VP8\* vaccine demonstrated that participants who received a 3-dose vaccination regime shed fewer attenuated rotavirus in feces as compared to trial participants who received two doses [19,29]. Based on this rationale, we opted to use the 3-dose regimen in this current study. The timing of 3 injections in Gn pigs are established in previous studies [27,37] based on the time needed to prime and boost immune responses in Gn pigs. Serum was collected at PID 0, PID 10, PID 21, PID 28, and PCD 7 for the detection of VP8\*-specific IgA, IgG, and Wa HRV-specific neutralizing antibody responses.

One subset of pigs (*n* = 3–7) from each group was euthanized before the challenge at PID 28. Another subset of pigs (*n* = 7–8) was orally challenged with 1 <sup>×</sup> 10<sup>5</sup> FFU of VirHRV Wa strain and monitored from PCD 0 to PCD 7 to assess the protection against virus shedding and diarrhea conferred by the vaccine before euthanasia on PCD 7. The pathogenesis of the Wa VirHRV infection has been studied in detail in Gn pigs; diarrhea and virus shedding persisted between 4 to 7 days post infection [33,38,44,45]. Based on these observations, we limited the study duration to 7 days

post-challenge in order to assess the immediate protection conferred by the vaccine against VirWa challenge. Four milliliters of 200 mM NaHCO3 were given orally 15–20 min before the VirHRV challenge to reduce stomach acidity to allow for rotavirus inoculum to pass through the stomach without being degraded due to low pH in the stomach.

At euthanasia, small and large intestinal contents (SIC and LIC) were collected from all pigs and processed, as described [46], for the detection of intestinal antibody responses by ELISA. Ileum, blood, and spleen were collected, and mononuclear cells (MNCs) were isolated from them for the detection of effector T cell responses by flow cytometry as described [42].

#### *2.4. Assessment of Diarrhea and Detection of RV Shedding in Feces by Rotavirus Antigen ELISA and CCIF*

The pigs were on a milk-based diet throughout the duration of the study, making their fecal consistency resemble that of a newborn infant. For the assessment of diarrhea, fecal consistency was recorded daily from PCD 0–7 and categorized as follows: 0: normal; 1: pasty; 2: semi-liquid; 3: liquid. The fecal scoring system used here has been well established and used for multiple Gn pigs studies [32,33,38,44,45,47,48]. Pigs were considered to be having diarrhea when their daily fecal consistency scores were recorded to be 2 or greater (≥2).

Rectal swabs were collected daily to monitor virus shedding by ELISA (for the detection of RV antigens) and cell culture immunofluorescence (CCIF; for the detection of infectious virions) from PCD 0–7. Rectal swabs were processed, as reported previously [49]. ELISA and CCIF assays for the detection and titration of VirHRV antigen in rectal swabs were carried out as previously described [33,38,44,45,47,50–52]. CCIF titers [fluorescent focus units (FFU)/mL)] were determined by Equation (1):

$$\frac{\frac{\partial f\dot{u}}{\partial mL}}{\frac{\partial L}{\partial mL}} = \frac{\text{\# Plaques counted}}{d \ge V} \tag{1}$$

where d = dilution factor, and *V* = volume of virus added.

#### *2.5. RV-Specific Serum VN, and VP8\*-Specific Serum and Intestinal IgA and IgG Antibody Titration*

VN antibody titers in serum samples were determined based on methods described previously [47]. MA104 cells were cultured in 96-well plates until an even monolayer was formed (≈3–4 days). Cells were washed once with sf-EMEM, and enriched with 100 μL of sf-EMEM and incubated at 37 ◦C for 2 h. The media was then discarded, and the cells were inoculated with trypsin-activated AttHRV (4 <sup>×</sup> 103 FFU in 10 μg/mL trypsin) in the absence or presence of 4-fold decreasing concentrations of Gn pig serum samples. The inoculum was discarded, and the plates were incubated at 37 ◦C for 18 h in 5% CO2 containing fresh sf-EMEM. The remainder of the steps has been described in detail in a previous publication [47]. The VN antibody titer was expressed as the reciprocal of the serum dilution, which reduced the number of fluorescent cell-forming units by >80% compared to the negative control serum. VP8\*-specific IgA and IgG antibody titers in serum and intestinal contents were measured by using isotype-specific antibody ELISA with purified P24-VP8\* as detector antigen at the plate coating concentration of 6.63 μg/mL, following methods described elsewhere [46,47,53]. When loading the testing samples on ELISA plates, four-fold serial dilutions of each sample started from 1:4 to 1:16384 for IgA, SIC, and LIC and 1:256 to 1:1048576 for IgG.

#### *2.6. Flow Cytometry*

Mononuclear cells (MNCs) collected from the ileum, blood, and spleen were diluted to a concentration of 2 <sup>×</sup> 106 cells/mL and were seeded into 12-well plates and stimulated with 12 <sup>μ</sup>g/mL of P24-VP8\* antigen for 17 h at 37 ◦C in 5% CO2 as determined previously [42]. CD3+CD4+ and CD3+CD8+ cell surface marker staining and IFN-γ intracellular staining have been described in previous publications [42,47,54,55]. All samples were stored in 0.05 mL of stain buffer and were maintained at 4 ◦C. A minimum of 100,000 events were acquired using a FACSAria flow cytometer (BD Biosciences, San Jose, CA, USA). Flow cytometry data were analyzed using FlowJo X software (Tree Star, Ashland, OR, USA).

#### *2.7. Statistical Analysis*

Gn pigs were randomly assigned into treatment groups upon derivation regardless of gender and body weight by animal care technicians. Student's *t*-test was used for comparisons of virus shedding and diarrhea data between the treatment groups. One-way analysis of variance (ANOVA) (General linear model) was used to compare rotavirus-specific IgA, IgG, virus-neutralizing (VN) antibody titers between the treatment groups. Tukey-Kramer HSD was used for the comparison of different time points within the same treatment group. Two-way ANOVA, followed by a Multiple *t*-test, was used for comparisons of frequencies of IFN-γ producing T cells between treatment groups. ANOVA analyses were carried out using JMP 14.0 (SAS Institute, Kerry, NC, USA), and all other statistical analyses were performed using GraphPad Prism 7.0 (GraphPad Software, San Diego, CA, USA). A *p* value lower than 0.05 was accepted to be statistically significant.

#### **3. Results**

#### *3.1. Protection against Diarrhea and Virus Shedding upon Challenge with VirHRV*

Vaccinated and control Gn pigs were challenged with VirHRV at PID 28 and were monitored daily for clinical signs (diarrhea) and virus shedding from PCD 1 to PCD 7. Gn pigs that were administered with P24-VP8\* vaccine had a significantly delayed onset of diarrhea (from 1.6 to 4.4 days), a significantly reduced duration of diarrhea (from 6.0 to 3.3 days), significantly lower mean diarrhea scores on PID 1 and 2, and a significantly lower cumulative fecal consistency score (from 14.3 to 9.1) as compared to the mock-vaccinated control group (Table 2 and Figure 1A). A delayed onset of virus shedding, a reduced peak titer, a reduced cumulative virus titer (presented as the area under the curve, AUC), and a significantly reduced duration (from 5.9 to 2.5 days) of virus shedding were observed in P24-VP8\* vaccinated pigs when compared to the control group (Table 2). In addition, the mean daily virus shedding titer in the vaccinated pigs was significantly reduced at PCD 2 (Figure 1B), and the reduction of total virus shed (AUC) was 2.27-fold compared to the control pigs (Table 2). However, the vaccine did not significantly reduce the incidence (%) of diarrhea and virus shedding (Table 2).



for clinical signs and virus shedding. Pigs with daily fecal scores of ≥2 were considered diarrheic. Fecal consistency was scored as follows: 0, normal; 1, pasty; 2, semi-liquid; and 3, liquid. Fecal virus shedding data was determined by ELISA and/or CCIF. b An arbitrary designation of Day 8 was assigned to pigs that did not develop diarrhea or shed virus in feces for calculating the mean days to onset. c For the purposes of calculating diarrhea and virus shedding duration, if no diarrhea or virus shedding was observed in pigs until euthanasia day (PCD 7), the duration days were recorded as 0. d Standard error of the mean. § Student's *t*-test was used for comparison between vaccine and control groups. Asterisk indicates statistical significance between the groups (*<sup>n</sup>* = 7–8; \*, *p* < 0.05).

**Figure 1.** P24-VP8\* vaccine protected against VirHRV diarrhea and reduced overall virus shed among vaccinated pigs. Fecal consistency (**A**) and virus shedding (**B**) were monitored daily from post challenge day (PCD) 1 to PCD 7 after the challenge with VirHRV. Fecal consistency scores ≥2 were considered to be diarrheic (dashed line indicates the threshold of diarrhea). Statistical significance between vaccinated and control groups, determined by multiple t tests, are indicated by asterisks (\*, *p* < 0.05; \*\*, *p* < 0.01).

#### *3.2. Strong VP8\*-Specific IgG and Virus Neutralizing, but Lack of IgA, Antibody Responses in Serum*

In order to monitor the development of VP8\* specific humoral immunity, serum samples were collected during the time of vaccine administration (PID 0, PID 10, and PID 21) at the VirHRV challenge (PID 28) and upon euthanasia (PCD 7). Serum IgG and IgA antibody responses were quantified using ELISA and depicted in Figures 2A and 2B, respectively. P24-VP8\*-specific IgG antibody titers in serum were significantly higher (*p* < 0.001) in vaccinated pigs at PID 10, PID 21, PID 28, and PCD 7 when compared to pigs in the control group (Figure 2A). However, serum IgA titers were only detectable after challenge (PCD 7) with VirHRV (Figure 2B).

HRV neutralizing antibodies were detected in the serum of P24-VP8\* vaccinated pigs starting from PID 21 and were observed to increase similarly with VP8\*-specific IgG titers until euthanasia at PCD 7. In control pigs, VN antibodies were only detectable after challenge with VirHRV and were at significantly lower levels compared to the vaccinated pigs (Figure 2C).

**Figure 2.** Geometric mean VP8\*-specific IgG (**A**) and IgA (**B**) and Wa-HRV neutralizing (**C**) antibody titers in serum collected from Gn pigs at PID 0, 10, 21, 28, and PCD 7. Pigs were vaccinated with P24-VP8\* vaccine or Al(OH)3 adjuvant only. Each serum specimen was tested at an initial dilution of 1:4. Negative samples were assigned an arbitrary value of 2 for calculation and graphical illustration purposes. Comparisons between groups at the same time points were carried out using Student's *t*-test and significant differences are identified by \*\*\* (*n* = 10–15; *p* < 0.001). Tukey-Kramer HSD was used for the comparison of different time points within the same group, where different capital letters (A, B, C, D) indicate a significant difference, *p* < 0.01, and shared letters indicate no significant difference.

#### *3.3. Lack of P24-VP8\* Specific Antibody Responses in the Intestines*

P24-VP8\*-specific IgA and IgG antibody titers in SIC and LIC, collected at the time of euthanasia (PID 28 and PCD 7), were measured by ELISA. The P24-VP8\* vaccine did not induce any detectable intestinal IgA or IgG antibody responses before the challenge at PID 28. After the challenge, among the eight vaccinated and challenged pigs, only VP8\*-specific IgG antibodies were detected (ELISA titers ranging from 256 to 1024) in the SIC of three pigs at PCD 7 (Supplementary Figure S2). However, the SIC IgG titers were not associated with the severity of diarrhea or the amount of virus shed in the three pigs throughout the challenge period.

#### *3.4. P24-VP8\* Vaccine did not Induce Strong VP8\*-Specific E*ff*ector T Cell Responses in Intestinal and Systemic Lymphoid Tissues*

Frequencies of IFN-γ+CD8+ and IFN-γ+CD4+ T cells in ileum, peripheral blood (PBL), and spleen at PID 28, and PCD 7 are summarized in Figure 3. At PID 28, slightly higher (not statistically significant) IFN-γ producing CD4+ and CD8+ T cell responses to the P24-VP8\* antigen was detected in vaccinated pigs as compared to control pigs (Figure 3A). P24-VP8\* vaccinated pigs had higher frequencies of IFN-γ+CD4+ T cells in ileum and blood and higher IFN-γ+CD8+ T cells in ileum, blood, and spleen compared to the mock-vaccinated control pigs. Upon the VirHRV challenge, there was still no significant difference in the frequencies of IFN-γ secreting CD4+ and CD8+ T cells between the two groups in the intestinal (ileum) or the systemic tissues (PBL and spleen) (Figure 3B).

**Figure 3.** Frequencies of IFN-γ+CD8+ and IFN-γ+CD4+ T cells in ileum, peripheral blood (PBL), and spleen at PID 28 (**A**) and PCD 7 (**B**). Two-way ANOVA followed by Multiple *t*-tests were carried out for comparisons. (*n* = 3–8; *p* < 0.05). There were no significant differences.

#### **4. Discussion**

In this study, the immunogenicity (antibody and T cell responses) and protective efficacy of the P24-VP8\* nanoparticle vaccine were evaluated in Gn pigs. We first demonstrated that the IM P24-VP8\* vaccine conferred significant protection against infection and diarrhea when challenged with the homotypic virulent strain Wa of HRV. This was indicated by the significant reduction in the mean duration of diarrhea, virus shedding in feces, and significantly lower fecal cumulative consistency scores recorded from PCD 1–7 in vaccinated pigs compared to the controls. However, the vaccine did not significantly reduce the incidence (%) of diarrhea and virus shedding, indicating that there was a lack of protective immune effectors at the site of infection (small intestine) at the time of challenge, which is consistent with the observed intestinal immune responses. The IM P24-VP8\* vaccine with Al(OH)3 adjuvant was highly immunogenic in Gn pigs. It induced strong VP8\*-specific serum IgG and virus-neutralizing antibody responses from PID 21 to PCD 7 but did not induce serum or intestinal IgA antibody responses or a strong effector T cell response. These results are consistent with the IM immunization route, the Al(OH)3 adjuvant, and the nature of the non-replicating vaccine. Non-replicating vaccines are typically ineffective in inducing effector T cell responses. The Al(OH)3 adjuvant is characteristic for its ability to enhance a Th2 type immune response, promoting strong humoral responses and suppressing effector T cell responses [56].

The observed protection and immune responses data together suggest that the protection conferred by the P24-VP8\* vaccine against diarrhea and virus shedding upon challenge with the virulent Wa HRV was mediated by the vaccine-induced antibodies in the serum. Although there were no antibodies present at the lumen of the small intestine, the site of HRV infection, at the time of challenge to totally prevent the initiation of RV infection, the viruses disseminated into blood from the infected small intestinal epithelial cells could have been neutralized by the high titers of VP8\*-specific IgG and virus-neutralizing antibodies during the phase of viremia. Such mechanisms can reduce the chance of infection of more epithelial cells by the virus from the basolateral side [45]. Studies showing that passively transferred serum antibodies can suppress or delay viral infection in RV-challenged pigtailed macaques [57], and an inactivated IM HRV vaccine (CDC-9) reduced virus shedding in Gn pigs upon challenge with Wa VirHRV [48] likely share the same protection mechanism with the P24-VP8\* vaccine. The serum IgG and virus-neutralizing antibody responses induced by the P24-VP8\* IM nanoparticle vaccine had similar dynamics and magnitude as the aluminum phosphate adjuvanted inactivated CDC-9 and P2-VP8\* IM vaccines in Gn pigs [27,48]. The P24-VP8\* vaccine demonstrated a similar degree of protection against diarrhea but a stronger protection against virus shedding in Gn pigs as compared to the P2-VP8\* vaccine [27].

There was a trend of inverse correlation between serum VP8\*-specific IgG titers at PID 28 and cumulative diarrhea scores post-challenge in the vaccinated pigs (Pearson's rank correlation, r = −0.6699 and *p* = 0.0691), suggesting that vaccinated pigs with higher serum VP8\*-specific IgG responses are more likely to be protected against severe diarrhea, which is in agreement with the study of serum IgG antibody in human adults showing that VP4-specific IgG titer was correlated with resistance to HRV infection [58]. The presence of high preexisting IgG titers was also correlated with less severe or shorter duration of diarrhea among children under three years of age [59]. As reviewed by Jiang et al., serum antibodies, if present at critical levels, are either protective themselves or are an important and powerful correlate of protection against rotavirus disease [60].

Additional investigations are required to explore the full potential of P24-VP8\* vaccine efficacy. First, P24-VP8\* is a candidate dual-vaccine against both NoV and RV, but we only examined the immune responses and protection against HRV, not human norovirus (HuNoV). Further studies in the Gn pig model of HuNoV infection are needed to evaluate its efficacy against NoV. Second, we only examined the protection against challenge with a homotypic HRV, and it remains to be determined whether the P24-VP8\* vaccine would be effective in protecting against heterotypic HRV, as the monovalent P[8] HRV vaccine Rotarix showed significant efficacy against P[4] (70.9%) and P[6] (55.2%) HRV associated gastroenteritis in African infants [61]. One of the important potential advantages of the novel P24-VP8\* nanoparticle dual vaccine is that the vaccine can be formulated as a cocktail vaccine to cover multiple types of RVs and NoVs for broad protection. Thus far, the Gn pig model of HRV infection and diarrhea has only been evaluated using the P [8] Wa HRV, requiring the need to test the effectiveness of Gn pigs as a suitable model for additional HRV P types to evaluate the broadness of protection of the novel P24-VP8\* nanoparticle vaccine.

#### **5. Conclusions**

The P24-VP8\* vaccine candidate is a typical nanoparticle vaccine with 24 copies of the major RV surface neutralizing antigen VP8\* displayed on the self-assembled norovirus P24 particles. The P24-VP8\* nanoparticles are easily produced in *E. coli* with a high yield and simple purification procedures at a low cost. Significant enhancement of the immunogenicity of both VP8\* and P domain backbone have been demonstrated in mouse immunization studies. In this current study, the usefulness of the P24-VP8\* vaccine was assessed in a Gn pig model, followed by the challenge of HRV. Three doses of IM immunization of Gn pigs demonstrated the nanoparticle vaccine's effectiveness to significantly shorten the duration of HRV diarrhea and virus shedding, reduce the severity of diarrhea, and lower the amount of virus shed when challenged. Immune responses associated with protection include high titers of VP8\*-specific serum IgG antibodies and virus-neutralizing antibodies induced by the

vaccine after the second and third booster doses. These findings will facilitate clinical trials of this vaccine candidate into a useful, safe, non-replicating, parental vaccine against RVs.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-393X/7/4/177/s1, Figure S1: Negative stain TEM images of P24-VP8\* particles. Figure S2: Geometric mean VP8\*-specific IgG titers in SIC of Gn pigs.

**Author Contributions:** Conceptualization, L.Y. and X.J.; Validation, A.R. and L.Y.; Formal Analysis, A.R.; Investigation, A.R., J.M., S.L., E.T. and A.S.; Resources, X.J. and M.T.; Writing – Original Draft Preparation, A.R. and L.Y.; Writing – Review & Editing, A.R., J.M., S.L., E.T., A.S., X.J., M.T. and L.Y.; Visualization, A.R. and L.Y.; Supervision, L.Y.; Project Administration, A.R. and L.Y.; Funding Acquisition, L.Y.

**Funding:** This research received no external funding.

**Acknowledgments:** We thank Enqi Huang for preparation of the vaccines, Sherrie Clark-Deener, Kevin Pelzer, Julie Settlage and Amy Rizzo for veterinary services, Karen Hall, Rachel McNeil for animal care. We thank Melissa Makris for flow cytometry service.

**Conflicts of Interest:** The authors declare that there are no conflicts of interest.

**Ethical Approval:** All animal experiments were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee at Virginia Tech (IACUC protocol: 16-214-BIOL).

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Review* **Hepatitis C Virus Vaccine: Challenges and Prospects**

**Joshua D. Duncan 1,2,3,\*, Richard A. Urbanowicz 1,2,3, Alexander W. Tarr 1,2,3 and Jonathan K. Ball 1,2,3**


Received: 30 December 2019; Accepted: 4 February 2020; Published: 17 February 2020

**Abstract:** The hepatitis C virus (HCV) causes both acute and chronic infection and continues to be a global problem despite advances in antiviral therapeutics. Current treatments fail to prevent reinfection and remain expensive, limiting their use to developed countries, and the asymptomatic nature of acute infection can result in individuals not receiving treatment and unknowingly spreading HCV. A prophylactic vaccine is therefore needed to control this virus. Thirty years since the discovery of HCV, there have been major gains in understanding the molecular biology and elucidating the immunological mechanisms that underpin spontaneous viral clearance, aiding rational vaccine design. This review discusses the challenges facing HCV vaccine design and the most recent and promising candidates being investigated.

**Keywords:** hepatitis C virus; vaccines; neutralising antibodies; animal models; immune responses

#### **1. Introduction**

First discovered in 1989 [1], the hepatitis C virus (HCV) is a major global health burden. Current estimates of HCV prevalence state that approximately 1% of the world's population are infected [2]. Chronic infection with HCV leads to cirrhosis of the liver and is associated with the development of hepatocellular carcinoma (HCC). Annually, 400,000 deaths are attributed to HCV and in the US, deaths from HCV have now overtaken those attributed to the human immunodeficiency virus (HIV) [3]. The extensive damage resulting from chronic infection makes this virus the leading cause for liver transplantation, a procedure that ultimately results in reinfection of the transplanted organ [4,5]. In recent years, the growing problem of HCV prompted the World Health Organisation to set a target to eliminate HCV as a public health burden by 2030. However, in the absence of a vaccine against HCV, this will prove challenging.

Therapeutic treatment of HCV has been vastly improved over the past decade due to the development of direct acting antivirals (DAAs). These compounds act as inhibitors of either the NS3/4a serine protease, NS5a or the NS5b RNA-dependent polymerase [6] and can achieve a 95% cure rate [7]. However, there are limitations to this strategy. Firstly, in order to treat an infection, a diagnosis must be made which may not occur in asymptomatic cases. It has recently been reported that the diagnosis rate in 2014 in the United Kingdom was estimated at capturing less than half of infected individuals [8]. Secondly, the cost of these therapeutics limits their use in developed countries, and all but excludes their use in low and middle-income countries with high HCV burdens. Thirdly, the ability of HCV to rapidly respond to selective pressures means that the emergence of DAA-resistant strains is a major risk [9,10]. Indeed, resistance-associated substitutions (RAS) have been detected in circulating HCV strains in treatment naive patients [11,12]. One such example of a RAS that has been reported in treatment-naïve patients is the C316 mutation which has been associated with resistance to non-nucleotide NS5B inhibitors such as Nesbuvir [13]. Another example is the S282T RAS which is associated with sofosbuvir resistance in vitro [14] although at the time of writing this polymorphism is not prevalent in clinical cohorts. The risk of DAA-resistance is mitigated by using combinational therapies of DAAs that target different HCV proteins.

The ability to effectively treat HCV infection has been a major achievement. However, there is growing evidence that HCV can leave lasting impacts upon its host post infection. For example, during infection extensive liver fibrosis can occur, which can persist for several years after viral clearance. Furthermore, persistent hyperfunctional CD8+ T cell phenotype has been reported following successful treatment with DAAs, suggesting continued immunological impairment [15]. There are also conflicting reports about the risk of hepatocellular carcinoma after virus clearance [16,17]. Understanding the long-term effects following virus clearance will take years more research as more data become available from successful cases. However, it is now apparent that therapeutics alone are unlikely to achieve the 2030 elimination target and thus a vaccine is urgently needed.

Modelling based on viral kinetics in reinfected individuals with pre-existing immunity has shown that transmission risk can be greatly reduced when an immune response occurs, despite detectable virus RNA titres [18]. This is of particular importance when considering vaccination of people who inject drugs, a high-risk group for HCV infection and reinfection. This suggests that a successful vaccine could be used to reduce viral titres rather than inducing sterilising immunity. This is a key point as it provides a realistic goal for HCV vaccine research.

#### **2. Host Immune Responses to Hepatitis C Virus**

Virus-host interactions determine the outcome of acute HCV infection. This interplay is complex and includes components of both the adaptive and innate host immune system. The most common scenario is progression to chronic infection. However, 25%–40% of individuals undergo spontaneous viral clearance (SVC) [19,20] within 12 months of infection [21,22]. Approximately 80% of these individuals will achieve SVC a second time following reinfection [23] with a marked decrease in viral RNA titres and reduced infection times compared to the first infection [23,24]. This indicates that initial infection can lead to the establishment of an immunological memory which can control subsequent HCV infection. This protective response consists of both humoral and cellular adaptive immune responses that do not result in sterilising immunity but prevent chronic infection. This phenomenon provides a benchmark for HCV vaccine research and thus is vital to elucidate.

#### *2.1. Innate Immune Reponses*

Innate immunity provides a first line of defence against viral infection. The liver presents a unique microenvironment that is enriched with cells that participate in this response, namely Kupffer cells, natural killer (NK) cells, hepatic stellate cells (HSCs) and macrophages. Initial innate responses are triggered by HCV-derived pathogen-associated molecular patterns (PAMPs) that are recognised by pattern recognition receptors (PRRs). Virion-associated PAMPs are the E1 and E2 glycoproteins [25,26], while intracellular detection of viral PAMPs, including viral proteins and RNA, is mediated by toll-like receptors (TLRs), nucleotide-binding oligomerisation domain (NOD)-like receptors (NLRs) and retinoic acid-inducible gene-I-like receptors (RLRs) [27,28]. Activation of signalling cascades downstream of these receptors results in the production of proinflammatory cytokines including interleukin-1β (IL-1β), IL-18 and type I and type III interferons (IFNs). These IFNs mediate upregulation of interferon stimulated genes (ISGs) in an autocrine and paracrine manner leading to an antiviral response in the liver [29,30]. However, the HCV viral proteins core, E2, NS3/4a and NS5a all impair the expression of ISGs through disruption of signalling cascades, allowing the virus to overcome the host innate response [31,32]. This results in an innate immune response that is incapable of clearing HCV.

Successful SVC is influenced by the outcome of innate immune responses. A major example is the association of single nucleotide polymorphisms (SNP) rs12979860 present in the type III IFN gene *IFN*λ*4* with SVC. In this case approximately 50% of individuals with a C/C genotype achieve SVC [33,34]. Additionally, NK inhibitory receptor, killer immunoglobulin receptor 2DL3 (KIR2DL3) and human leukocyte antigen C group 1 (HLA-C1) are associated with SVC [35] due to a reduced inhibition of cytotoxic NK activity [36,37].

#### *2.2. Ceullar Immune Responses*

Cellular immunity has long been associated with spontaneous HCV clearance [38] and is mediated through two main T cell subsets, the cytolytic CD8+ T cells and CD4+ helper T cells. CD8+ T cells destroy infected cells in a manner restricted by MHCI presented epitopes [39]. In contrast, recognition by CD4+ helper T cells is MHCII restricted and their role is to aid the function of CD8+ T cells and the establishment of T cell memory through the secretion of cytokines such as IFN-γ [40,41]. CD4+ T cells also aid B cell activation and a CD4+ T cell subset, follicular helper T cells (TFH) are required to establish a long-term antibody response [42,43]. HCV-specific T cells are detectable within the first 12 weeks of infection and target a broad range of HCV epitopes present on both structural and non-structural viral proteins [44]. During the progression to chronicity, the HCV-specific CD4+ T cells display an exhausted phenotype and the population collapses [45,46]. The decrease in CD4+ T cell function leads to a dysregulated CD8+ T cell response in which these cells become exhausted and dysfunctional with reports of continued IFN-γ secretion but an absence of cytolytic activity [47]. The reasons for this reduction in effective cellular responses are incompletely understood. The loss of functional HCV-specific T cells could be the result of host regulation of the immune system since persistent antigen stimulation could lead to the prolonged production of proinflammatory cytokines which in turn contributes to hepatic tissue damage.

The importance of T cells to SVC was first demonstrated in experimentally infected chimpanzee in which HCV persistence was observed in the absence of either a CD4+ or CD8+ T cell response [48,49]. Interestingly, when CD4+ T cells were depleted HCV persisted alongside functional CD8+ T cell responses. HCV-specific CD4+ T cells and CD8+ T cells are detectable during acute infection [45]. This provides strong evidence that the T cell responses have a major role in the outcome of HCV infection. HCV-specific CD4+ T cells are broadly targeting with the most common epitopes being found in the core, E2, NS3, NS4a, NS4b, NS5a and NS5b HCV proteins [44].

#### *2.3. Humoral Immune Responses*

Neutralising antibodies (nAbs) in the context of HCV infection were first described by Farci et al. [50], although their role in spontaneous clearance was disputed for many years due to reports of cell mediated clearance in seronegative individuals [51–53], suggesting that nAbs are not essential to achieve SVC. However, analysis of sera from individuals who cleared HCV has shown the presence of nAbs and these are detectable at earlier time points compared to acute infections that proceed to chronicity and are subsequently lost following viral clearance [54] suggesting that a rapid, short-lived humoral response is required for clearance [55–58]. It has recently been shown that nAbs generated within the first 100 days of infection often have a narrow neutralising capacity directed towards the founder virus [58]. The selective pressure exerted by nAbs upon the circulating strains can also drive the evolution of HCV towards escape mutations that compromise viral fitness further aiding clearance of the infection [59,60]. The delayed appearance of cross-reactive nAb responses are apparent in chronically infected individuals as isolated sera can neutralise circulating strains from previous infection time points with greater potency than the current dominant virus [56] and cross-reactive nAbs have been isolated from chronically infected individuals [61–64]. Although these nAbs cannot clear the infection, they have been associated with reduced liver fibrosis [65] and patients that experience hypogammaglobulinemia have a more severe disease progression [66].

Further insight into the humoral response has been obtained through characterising nAbs derived from patients. Bailey et al. [67] sequenced cross-reactive nAbs isolated from two individuals that cleared HCV and showed these nAbs shared >90% similarity with germline heavy chain variable (VH) genes and >92% similarity with germline light chain gene sequences revealing that generation of cross-reactive nAbs required limited somatic hypermutation. A common feature of these nAbs was the *VH1-69* gene which is found in potent cross-reactive nAbs that target antigenic region 3 (AR3) and is also present in nAbs targeting HIV and Influenza [68–71]. Structural investigation has also shown that cross-neutralising activity is a result of long complementarity-determining region H3 (CDRH3), typically 18–22 residues, that forms a β-hairpin structure that is stabilized by a disulphide bond allowing for interaction with conserved E2 epitopes [72,73]. This level of insight into how potent nAbs work can inform rational design of B-cell immunogens to favour the production of these types of antibodies in the host.

#### **3. Hepatitis C Virus Envelope Proteins as Vaccine Targets**

The HCV envelope glycoproteins, E1 and E2, are located at residues 192–746 of the polyprotein and are the targets of the humoral immune response, making them an attractive vaccine target [74] (Figure 1A). Both E1 and E2 are type I transmembrane proteins that form an intracellular non-covalent heterodimer that form higher order covalent structures on the mature virion [75,76]. The E2 ectodomain contains an immunoglobulin-like β-sandwich that is flanked by α-helices which form a front and back layer [77,78]. There are also regions that exhibit high levels of variability which are referred to as hypervariable regions (HVRs) 1 and 2, and a third intergenotypic variable region (IgVR; Figure 1A). The E1 protein is smaller, more conserved than E2 and less well characterized with partial crystal structures resolved for fragments encompassing residues 192-271 [79] and 314–324 [80].

The E1E2 heterodimer mediates entry into the hepatocytes through interactions with four essential host receptors, CD81 [81,82], scavenger receptor B1 (SRB1) [83], claudin [84] and occludin [85,86]. The first step in viral entry is the interaction between SRB1 and the HVR1 (Figure 1A) located at the N-terminus of E2 which induces a conformational change that exposes the conserved E2 core region and the CD81 binding loop (residues 519–535) [77]. The interaction between E2 and CD81 results in recruitment of claudin to CD81 which leads to clathrin-mediated endocytosis [87]. Membrane fusion occurs in low pH endosomes, which is thought to induce a conformational change in the E1E2 heterodimer [88]. This leads to membrane fusion, possibly mediated by E1 via the action of a putative fusion peptide located at residues 272–285 (Figure 1A) [89].

Antibody-mediated neutralisation of HCV is achieved through targeting the E1E2 heterodimer on the surface of the virus. To date, most characterized nAbs target the E2 protein. The are several linear and discontinuous regions of E2 that are targeted by Abs and these have varying nomenclature, being referred to as ARs1-5 [69], Epitopes I-III (Figure 1) [90], or domains A-E [91]. Importantly, AR4 and 5 are reliant on the presence of the E1E2 heterodimer for binding [69]. The epitope I, II and AR3 regions form the neutralising face of E2 which is targeted by some of the most potent cross-reactive nAbs described to date [92] arguing for their inclusion in a nAb-eliciting vaccine.

**Figure 1.** (**A**) Schematic diagrams of the hepatitis C virus envelope glycoproteins E1 and E2, showing N-linked glycosylation sites (N), transmembrane domains (TMDs), and the E1 fusion peptide (FP). E2 hypervariable regions (HVRs) 1 and 2, and the intergentypic variable region (IgVR) are also depicted. Linear epitopes I, II and III are highlighted in red, dark blue and light blue, respectively. (**B**) E2 structure (PDB: 6MEH). Linear epitopes I, II and III are highlighted in the corresponding schematic. Antigenic regions (ARs) are also shown.

#### **4. Challenges to Hepatitis C Virus Vaccine Design**

#### *4.1. Genetic Diversity*

As a species, HCV exhibits extensive genetic diversity is driven by a mutation rate in the order of 10−<sup>4</sup> substitutions per site of the genome [93,94]. Mutations are acquired through the activity of the error prone NS5b RNA-dependent RNA polymerase and this coupled with high levels of virus production and selective pressures exerted by the host immune response has driven the diversification of HCV. There are currently eight genotypes (Gt1-8) reported which are defined by 30% difference in nucleotide sequence [95]. Gt1 HCV isolates have the highest prevalence, accounting for 49% of diagnosed cases globally followed by Gt3 accounting for 17.9% of cases [96], and are more prevalent in developed countries [96]. In contrast, Gt4 and Gt5 isolates are more prevalent in lower income countries in Africa and the Middle Eastern regions [96]. Genotypes are further classified into subtypes of which there are currently 90 confirmed groups. These exhibit a 15% variation in nucleotide sequence [95]. Gt1-7 contain multiple subtypes. However, the recently reported Gt8 group currently contains a single subtype isolated from four individuals [97]. Rapid evolution of HCV during the course of an infection leads to the establishment of a heterogeneous population [98]. The diversification of this population is driven by host immune selective pressure and the degree to which variation occurs in acute infection correlates with a progression to chronicity [99].

The extent of genetic diversity gives rise to genotype-specific immune responses. The E1E2 sequence shows the greatest level of variation as a result of the selective pressure exerted by the host immune response [100], leading to humoral responses that can have reduced heterologous neutralising activity [101]. However, cross-reactive nAbs that are capable of targeting isolates from different genotypes have been described in several studies [102], thus highlighting that this challenge can be overcome. Epitope variability also leads to genotype-specific cellular responses for example Luxenburger et al. [103] have recently shown CD8+ T cells from individuals chronically infected with a Gt4 HCV isolate failed to recognize Gt1 derived epitopes. Limited intergenotypic cross-reactivity of Gt3 HCV-specific T cells has also been reported in patients that successfully cleared Gt3 infections [104]. Limited cross-reactive immune responses to distantly related HCV isolates will be a key challenge to produce an effective vaccine

#### *4.2. Evading the Host Adaptive Immune Response*

The ability of HCV to establish a chronic infection highlights the efficiency in which this virus can subvert the host immune response. This is achieved through multiple strategies. Firstly, immunodominant epitopes such as the HVR1 of E2 elicit a response of non-neutralising antibodies. Currently it is thought that HVR1 acts as a shield of the more conserved epitopes in the E2 core that contain CD81 binding residues, since removal of HVR1 increases virus susceptibility to neutralisation [105,106]. Evidence also shows that HVR1 induces greater homologous nAbs and deletion induces broader heterologous nAbs [107].

Epitope masking is also achieved through the glycosylation of envelope proteins to form a glycan shield [108]. This mechanism diminishes the binding of nAbs and is used by other viruses such as HIV and Influenza although by comparison, HCV exhibits a reduced variability in the position of each glycan group. The E1 and E2 proteins contain 4 and 11 N-linked glycosylation sites which are highly conserved across different genotypes (Figure 1A) [109]. Deletion of E2 glycosylation sites increases HCV susceptibility to nAbs showing that glycosylation acts via steric hinderance to nAbs [109,110]. Interestingly, removal of E2 glycan sites or expression of this protein in systems that use smaller, less complex glycan groups enhance the immunogenicity compared to mammalian expressed E2 [111–113]. Another mechanism in which nAb resistance can be mediated through glycosylation is glycan shift. This arises through non-synonymous mutations that result in the deletion of a glycosylation site and the appearance of a new glycosylation site in a different position in the protein. Pantua et al. [114] reported this phenomenon after incubating cell cultured HCV (HCVcc) with the murine nAb, AP33 which targets residues within the E2 epitope I region [115]. After 5 days of incubation with AP33, escape mutants could be detected which contained either N417S or N417T residue variations coupled with a new glycosylation site at position 415 [114].

The evasion of nAbs can also be attributed to the presence of host derived factors. One such factor is high-density lipoproteins which are present in human serum and has been shown to increase HCV infectivity via SR-BI interactions which in turn decreases the time available for nAbs to bind to their target [116,117]. Additionally, the presence of host derived apolipoprotein E on the mature HCV virion has been shown to have a role in epitope masking the E2 protein abrogating nAb activity [118]. Another evasion method is through the use of decoys. Recently it has been reported that in a HCVcc system, infected cells produce lipid droplets coated with the E2 protein [119]. This may also be a strategy to subvert the immune system by using these E2-coated droplets to interact with nAbs, thereby reducing the availability of nAbs to target E2 present on the surface of the virus particle. Finally, evasion from nAbs has recently been shown to be associated with resistance to antiviral interferon-induced transmembrane (IFITM) proteins [120] showing that the innate immune response can also be a driving force for nAb evasion.

#### *4.3. Models for Hepatitis C Virus Infection*

#### 4.3.1. In Vitro Models

Since the discovery of HCV, research was hampered by the lack of infection models both in vivo and in vitro. The use of models in vaccine research is essential for the assessment of sera and monoclonal antibodies arising from natural infection, experimental immunizations or vaccine trials. In the case of in vitro models, the key challenge was generating a cell culture-based system that could produce infectious HCV at titres suitable for further experimentation. This problem was solved with the discovery of two Gt2a isolates, JFH1 and J6 which produce high titres of infectious HCV from cultured Huh 7.5 cells [121]. Following this, it was found that intergenotypic recombinant viruses consisting of core-NS2 of an isolate of interest fused to NS3-NS5b of JFH-1 could produce viable chimeric viruses [121]. The developments in HCVcc techniques have allowed panels with representatives of the seven major genotypes to be set up [122–124]. Our group has recently described a method in which E1E2 patient derived sequences could be incorporated into chimeric HCVcc expression vectors [125,126], further aiding the ability to characterize isolates that have different neutralisation phenotypes.

Viral pseudotyping has also been used to study the entry mechanism of HCV. This approach utilizes the ability of retroviral gag/pol proteins to self-assemble into enveloped virus particles. Simultaneous expression of foreign viral envelope proteins in the same cells leads to the incorporation on to the surface of the retroviral particle whose entry properties are dictated by the envelope glycoprotein. Infectivity can be measured by the incorporation of a reporter gene, usually luciferase, which is introduced into target cells and expressed. Methods to generate HCV pseudoparticles (HCVpps) commonly use murine leukemia virus or HIV derived vectors [127,128]. Our group and others have utilized these systems to characterize extensive panels of patient-derived E1E2 sequences allowing important characterisation of these unique isolates [57,129–131]. These studies have shown that neutralisation sensitivity of patient derived isolates to both sera and monoclonal nAbs is markedly varied and not associated with genotype but is isolate dependent. Furthermore, it has been shown that the HCV reference isolate H77, a Gt1a isolate that was first adopted due to its ability to infect and cause disease in chimpanzees [132–134], is more susceptible to antibody-mediated neutralisation than many patient-derived isolates [102]. This finding is of significance since H77 neutralisation has been used to validate multiple vaccine candidates and therefore these data provide limited insight into how effective a candidate may be towards clinically relevant isolates. H77 has also been used in the design of immunogens with moderate success, although it could be argued that immunogen design based on more nAb-resistant isolates may elicit more potent nAbs.

The ability to generate both HCVcc and HCVpp provide essential tools for the in vitro study of HCV and assessing the neutralising breadth of sera and monoclonal nAbs. The importance of isolate panels for the validation of vaccine candidates is exemplified in the field of HIV vaccine research, in which candidates have been tested against an extensive panel of 109 isolates developed by Seaman et al. [135], which are grouped based on neutralisation susceptibility. In this way humoral responses following vaccination can be assessed in a standardized manner allowing for relevant comparisons between candidates and vaccines that induce responses capable of neutralising the most resistant isolates can be identified for further development. At the time of writing a consensus panel of isolates is yet to be widely adopted but doing so will greatly enhance our understanding of which candidates are promising and those that are not.

#### 4.3.2. In Vivo Models

Humans are the natural host of HCV, and therefore it is imperative that suitable in vivo models are used in order to test the efficacy and safety of preclinical vaccine candidates. Arguably, the most successful model has been the chimpanzee, which is permissive to HCV infection under experimental conditions. However, ethical concerns over the use of this species has led to a ban on its use in experimental research. An alternative has been the use of chimeric humanized or transgenic mouse models which have humanized livers such as Alb-uPA/SCID mice [136,137] or those engineered to express human CD81 and occludin [138,139]. There are limitations to using chimeric mouse models, most notably that they do not exhibit cirrhosis or HCC, and are technically difficult to produce [140].

In light of the limited options available for the in vivo modelling of HCV, there has been increasing interest in other members of the *Hepacivirus* genus for use as analogues. At the time of writing there are two species of particular interest, the Rodent Hepacivirus (RHV) and the Equine Hepacivirus (EqHV). RHV was initially discovered in the species *Rattus norvegicus* in 2014 [141]. Experimental infection of different lab rat breeds leads to the establishment of permanent infection with features observed during HCV infection in humans such as hepatic fibrosis, steatosis and evidence of SVC in the Holtzman rat strain [142]. There are key contrasts between RHV and HCV infections, notably the lack of IFN-γ+ CD8+ T cell responses during RHV infection [143].

EqHV was initially discovered in canines and subsequent serology testing identified equines as the natural host [144,145]. This virus is the closest relative of HCV and there are shared features of both species such as similar levels of glycosylation of the E1E2 proteins and a conserved seed site in the 5 -UTR for the liver-specific microRNA-122 (miR-122) which contributes to stability, translation, and replication of the viral RNA [146]. Seroprevalence of EqHV is in the region of 30% of surveyed horses, with approximately 3% testing positive for viral RNA [147]. This discrepancy between serology and viral RNA may be indicative of a high viral clearance rate. Despite this potential high clearance rate, EqHV acute infection may proceed to chronicity after 6 months [148], further validating the relevancy of this virus as a model for HCV.

The use of analogous Hepacivirus species will allow for the testing of vaccine strategies and the experimental challenge in the natural host of that species, a feat that cannot currently be achieved for HCV. Whilst these approaches are likely to prove highly valuable to the field of HCV vaccine research it is important to acknowledge the differences in mammalian immune systems and how this will impact on our interpretation of experimental data. For example, the *IgVH 1-69* gene with an extended CDRH3 region has previously been described as a feature of anti-HCV specific nAbs that are elicited in SVC and therefore it is logical that a vaccine candidate could be assessed based on its ability to elicit these types of nAbs. With this in mind it is worth considering how comparable the antibodies of animal model species are to those of humans. One such difference is the length of the CDRH3 regions which are typically 16 residues long [149], however these are reduced to an average of 12 to 14 residues in rodent [149] and equine species [150], respectively. The CDRH3 region can be critical for the formation of the antigen-antibody complex and extended CDRH3 loops are effective at penetrating the glycan shield of viral envelope proteins, a favorable characteristic of anti-HCV nAbs. This raises an important question about the relevance and suitability of current animal models as predictors of the human antibody response. Similarly, the paradigm of vaccine research has been to test novel candidates in small animals followed by larger animal species before clinical testing in humans. This approach enables evaluation of immunogenicity and safety prior to exposure of an immunogen in humans. However, questions should be asked as to whether this approach has hindered the progress of at least some HCV vaccine candidates because of differences between the model and human antibody repertoires.

#### *4.4. Rational Design of Immunogens to Elicit Cross-Reactive Neutralising Antibodies*

The informed, rational design of immunogens will be vital in order to produce a successful HCV vaccine. As advances in the analysis of antibody repertoires are allowing deeper insight into the types of nAbs that are associated with SVC, there is potential to use this information to develop immunogens that favour the production of these types of antibodies. An example of this can be seen with the E2 epitope I region or antigenic site 412 (AS412). This linear epitope is structurally flexible and when in complex with nAbs and adopts three distinct conformations. Firstly epitope I forms an extended conformation in complex with the rat nAb 3/11, a β-hairpin conformation is observed when bound to AP33 or the human nAb HCV-1, and an intermediate conformation occurs when in complex with the

human nAb HC33.1 [151]. As previously discussed, HCV may escape epitope I targeting nAbs by the mechanism of glycan shift. However, this mechanism does not abrogate and instead enhances the potency of HC33.1 [152]. With this in mind, rational design of an immunogen could strive to present the epitope I region in an intermediate conformation to bias the humoral response to elicit HC33.1-like nAbs. To date this has not been investigated however, efforts have been undertaken to present epitope I in the β-hairpin conformation using cyclic peptides based on θ-defensin which adopts a similar conformation [153]. This approach induced nAb responses when tested in mice however these were lower compared to mice immunized with E2 [153]. The rational design of HCV epitopes in this way is technically challenging and is limited to targeting linear epitopes. Additionally, delivery of cyclic peptides will require further development in order to enhance the immunogenicity and stimulate greater titers of nAbs.

#### **5. Vaccine Prospects**

Despite the challenges faced in HCV vaccine design, there have been a variety of different vaccine approaches investigated with a small number of these candidates reaching human trials (summarized in Table 1). Currently, DNA and peptide-based vaccine candidates are actively being investigated in murine models [154–158]. One recently reported peptide candidate consisted of overlapping peptides derived from the p7 protein which induced antigen-specific CD4+ T cells and cytotoxic CD8+ T cells capable of targeting p7 expressing hepatocytes in vivo [154]. This study has been the first to show the immunogenicity of p7 when used as the sole target in a vaccine. Additionally, a DNA-based vaccine has been shown to induce CD4+ T cell and CD8+ T cell responses and elicit T cell memory in mice; however, a non-neutralising Ab response was also observed [156]. These vaccine candidates are in an early stage of development and therefore the following sections will discuss candidates that have been more extensively studied.

#### *5.1. Recombinant Subunit Vaccines*

Recombinant subunit vaccines are an attractive technology as they have previously been utilized for a variety of pathogens [159] and are economical to produce on a commercial scale. In the field of HCV vaccine research, attempts to generate a subunit vaccine have focused exclusively on targeting the HCV E proteins. The most successful candidate has been the recombinant E1E2 protein (rE1E2) derived from a Gt1a isolate. Initially, HCV-1 derived rE1E2 was used to vaccinate seven chimpanzees which resulted in a humoral response [160]. Following homologous HCV challenge, five out of seven vaccinated chimpanzees exhibited sterilising immunity with the remaining two developing acute infections that were subsequently cleared [160,161]. This was further validated in small animal models and that immunisation rE1E2 derived from a single isolate could induce cross-reactive nAbs [161]. Following this work, a phase I human trial was performed in which rE1E2 derived from the Gt1a HCV-1 isolate was administered at different doses (4 μg, 20 μg and 100 μg of rE1E2) with the MF59 adjuvant to healthy volunteers [162]. Importantly this study showed that rE1E2 can be safely tolerated in humans, a finding of significance given that a successful HCV vaccine candidate will likely require these proteins or derivatives of them. Polyfunctional CD4+ T cell responses were detected and the magnitude of this response was inversely related to dosage. Humoral responses and nAbs were initially reported [162,163] and further testing using a diverse HCVcc panel with isolates from Gt1-7 showed that 3/16 volunteers had elicited cross reactive nAbs [164]. Additionally, post vaccination sera from 5 volunteers competed with nAbs AP33, AR3B, AR4A, AR5A and IGH526, showing that the humoral response targeted both E1, E2 and the E1E2 heterodimer [165]. The ability for rE1E2 from a single isolate to induce nAbs in humans is significant and encouraging although work should be undertaken to enhance the efficacy at which nAbs are generated. A key issue with rE1E2 was purification which relied on lectin-based affinity methods that were low affinity and non-specific [166]. Furthermore, efforts to improve rE1E2 production were made by removing the transmembrane domains of the E protein. However, this truncated E1E2 was found to be a weak immunogen that elicited low levels of

nAbs [167]. In light of these issues, both flag-tagged and Fc-tagged rE1E2 have been investigated and elicit nAbs in mice at comparable levels to wild type rE1E2 [166,168].

Another subunit vaccine approach has been the use of truncated soluble E2 (sE2) which results in a secreted form of the E2 ectodomain. A candidate of particular note is the H77 derived sE2384-661 construct with deleted HVR1, HVR2 and IgVR (sE2Δ123) which can be expressed in a correct conformation which interacts with CD81 and conformation sensitive antibodies [169–171]. Vietheer et al. [170] reported that the expression of sE2Δ<sup>123</sup> in mammalian cells results in biochemically and immunogenically distinct isoforms of this protein that exist as monomeric, dimeric, pentameric and a high molecular weight aggregate (HMW1; estimated at 51 protomers). Immunisation of guinea pigs with different isoforms showed that HMW1 elicited greater titres of cross-reactive nAbs that could neutralise isolates from Gt2a, 4a, 5a, 6a and 7a expressed in a HVCcc system [170]. Additionally, HMW1 had exhibited a reduced activity with non-neutralising antibodies, suggesting that this aggregate focuses the humoral immune response away from these epitopes, allowing for greater focusing to nAb epitopes [170]. Whilst this initial report is promising, it was noted that sE2Δ<sup>123</sup> reduced homologous neutralisation activity, a likely result of the removal of HVR1 which was also observed by Law and colleagues [172].

#### *5.2. Virus-Like Particles*

Virus-like particles (VLPs) are a promising area of vaccine research and have been investigated extensively to design a HCV vaccine [173]. To date, there are two widely used commercial vaccines that utilize VLP technology, the hepatitis B virus (HBV) and human papilloma virus vaccines. VLPs are formed when the structural proteins of a virus assemble in a genome-independent manner to produce a particle that resembles the native virus whilst lacking the ability to replicate and are more immunogenic than soluble protein due to their repetitive structure and ability to drain into tissue lymph nodes [107,174].

HCV-LPs were first described by Baumert et al. [175] in which VLPs were expressed in the insect cell line Sf9 following transduction with a recombinant baculovirus containing the core E1 and E2 HCV structural proteins. These HCV-LPs elicited HCV-specific IgG production, IFN-γ secreting CD8+ T cell and CD4+ T cell mediated cellular responses in mice and baboons [176–178]. Immunisation of four chimpanzees with four doses of HCV-LPs resulted in CD4+ and CD8+ T cell responses in all animals [179]. Subsequent challenge with homologous HCV resulted in a reduced viremia in vaccinated animals compared to controls with two animals testing negative for HCV two years post challenge [179]. Interestingly, humoral responses in chimpanzees were not detectable in three of these animals and appear to have been absent during the reduction of viremia. Further work on this candidate has not been progressed since this first study in chimpanzees. However, it is an important example for future HCV vaccine work, as it showed that viral clearance can be achieved using an experimental VLP vaccine candidate. Despite work on this candidate ceasing, HCV-LPs continued to be investigated. Recently, a HCV-LP system has been established through the transduction of the human hepatoma cell line Huh 7 with recombinant adenovirus encoding the structural genes of HCV [180]. Initially designed using the H77 isolate, further development has expanded by generating HCV-LPs using the structural sequences of different isolates representing Gt1b, 2a and 3a subtypes to produce HCV-LPs with a conformation close to that the wild type virus [180–183]. Trials of this candidate in mice showed the induction of nAbs and activation of both CD4+ and CD8+ T cell responses and this has recently been reported in a vaccinated Landrace pigs [184]. To date, assessment of the humoral response has demonstrated that this vaccine elicits nAbs with activity towards homologous strains. Heterologous neutralisation data have not been reported for this candidate. Testing the sera from animals immunized with this quadrivalent vaccine against a diverse panel of HCV isolates would be of particular interest not only for this candidate but to inform future vaccine design.

#### *5.3. Viral Vector Vaccines*

Viral vectors have the capacity to induce high level cellular immune responses [185], making them an attractive approach for HCV candidates. Viral vectors of particular interest are human adenovirus 6 (HuAd6), chimpanzees Ad 3 (ChAd3) and modified vaccinia ankara (MVA) [186]. The only HCV vaccine candidates to be tested in phase I and phase I/II human trials has been ChAd3 and MVA which each encoded HCV NS3, NS4a, NS4b, NS5a and inactivated NS5b (NSmut) derived from the Gt1b BK isolate [187,188]. These candidates were tested together in a heterologous primer/boost strategy in which healthy human volunteers were primed with ChAd3-NSmut and boosted with MVA-NSmut 8 weeks later [188]. The results of this phase I study showed that prime/boost ChAd3/MVA-NSmut induced both CD4+ T cell and CD8+ T cell responses directed towards epitopes on all five HCV non-structural proteins, importantly the effect of the MVA-NSmut enhanced the T cell responses and these were detectable at week 70 [188]. Given the promising cellular responses observed in the phase I trial, this approach was tested in patients with chronic HCV infections although this was unable to reverse T cell exhaustion [189], ruling out the use of ChAd3/MVA-NSmut as a therapeutic vaccine. Recently ChAd3/MVA-NSmut was tested in a large phase I/II trial involving 548 high risk individuals was concluded (clinicaltrials.gov identifier: NCT01436357). At the time of writing, this trial has concluded, and data have yet to be published. However, a press release statement from the National Institute of Allergy and Infectious Disease stated that 14 vaccinated individuals became chronically infected with HCV, suggesting that this candidate does not provide protection against HCV infection [190]. Understanding the reasons for this failure will be vital to inform our understanding of vaccine design. Given that the ChAd3/MVA-NSmut candidate exclusively targeted T-cell epitopes, the failure of this vaccine candidate could further emphasize the need for both humoral and cellular immune responses in clearance of acute infections.



 (not applicable) for vaccine candidates that are not designed to elicit HCV-specific B cell responses. N.D (not determined) in the study.

#### **6. Conclusions**

In the 30 years since the discovery of HCV, it has been apparent that this virus is highly complex and presents a major public health challenge. Encouragingly, efforts toward a vaccine continue and investigation into a range of approaches have yielded interesting results. The vast improvements in in vitro analysis of HCV have greatly aided our understanding of how nAbs work, informing rational vaccine design. In this review, we have highlighted the potential advantages of adopting reference panels of patient-derived E1E2 sequences in order to assess vaccine candidates for their potential to induce nAbs capable of targeting circulating HCV strains. Adopting this approach will provide a more rigorous tool for evaluating the nAb response produced by novel HCV vaccine candidates a characteristic that will likely be essential in a successful HCV vaccine.

**Author Contributions:** Writing—original draft preparation, J.D.D.; review and editing, R.A.U., A.W.T. and J.K.B.; funding acquisition, J.K.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by the Medical Research Council [grant number MR/R010307/1]; and the NIHR Nottingham Biomedical Research Centre. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **Broad Protection of Pigs against Heterologous PRRSV Strains by a GP5-Mosaic DNA Vaccine Prime**/**GP5-Mosaic rVaccinia (VACV) Vaccine Boost**

#### **Junru Cui, Caitlin M. O'Connell, Connor Hagen, Kim Sawicki, Joan A. Smyth, Paulo H. Verardi, Herbert J. Van Kruiningen and Antonio E. Garmendia \***

Department of Pathobiology and Veterinary Science, College of Agriculture, Health and Natural Resources, University of Connecticut, Storrs, CT 06269, USA; junru.cui@uconn.edu (J.C.); caitlin.o'connell@uconn.edu (C.M.O.); connor.hagen@uconn.edu (C.H.); kimitt@gmail.com (K.S.); joan.smyth@uconn.edu (J.A.S.); paulo.verardi@uconn.edu (P.H.V.); herbert.vankruiningen@uconn.edu (H.J.V.K.) **\*** Correspondence: antonio.garmendia@uconn.edu; Tel.: +1-860-486-3945

Received: 22 January 2020; Accepted: 23 February 2020; Published: 28 February 2020

**Abstract:** Background: Porcine reproductive and respiratory syndrome (PRRS) viruses are a major cause of disease and economic loss in pigs worldwide. High genetic diversity among PRRSV strains is problematic for successful disease control by vaccination. Mosaic DNA and vaccinia (VACV) vaccines were developed in order to improve protection against heterologous PRRSV strains. Methods: Piglets were primed and boosted with GP5-Mosaic DNA vaccine and recombinant GP5-Mosaic VACV (rGP5-Mosaic VACV), respectively. Pigs vaccinated with rGP5-WT (VR2332) DNA and rGP5-WT VACV, or empty vector DNA and empty VACV respectively, served as controls. Virus challenge was given to separate groups of vaccinated pigs with VR2332 or MN184C. Necropsies were performed 14 days after challenge. Results: Vaccination with the GP5-Mosaic-based vaccines resulted in cellular reactivity and higher levels of neutralizing antibodies to both VR2332 and MN184C PRRSV strains. In contrast, vaccination of animals with the GP5-WT vaccines induced responses only to VR2332. Furthermore, vaccination with the GP5-Mosaic based vaccines resulted in protection against challenge with two heterologous virus strains, as demonstrated by the significantly lower viral loads in serum, tissues, porcine alveolar macrophages (PAMs), and bronchoalveolar lavage (BAL) fluids, and less severe lung lesions after challenge with either MN184C or VR2332, which have only 85% identity. In contrast, significant protection by the GP5-WT based vaccines was only achieved against the VR2332 strain. Conclusions: GP5-Mosaic vaccines, using a DNA-prime/VACV boost regimen, conferred protection in pigs against heterologous viruses.

**Keywords:** PRRSV Mosaic T-cell DNA vaccine VACV; PRRS; cross protection; heterologous virus challenge

#### **1. Introduction**

Porcine reproductive and respiratory syndrome (PRRS) is a major disease in pigs that causes significant economic losses to industry across the world. In the United States alone, PRRS causes over \$600 million in losses a year [1]. PRRS causes reproductive failure in pregnant sows and respiratory disease in young pigs. The causal virus, PRRSV, is a positive-sense, single-stranded RNA virus with a lipid envelope, and belongs to the Family *Arteriviridae,* Order *Nidovirales* [2–4]. The genome is approximately 15 Kb long and encodes at least 22 different viral proteins including fourteen non-structural and eight structural proteins [5]. PRRSV induces both humoral and cellular immune responses in pigs and some of the viral proteins inducing such responses have been identified; however, protection against reinfection is incomplete [6–9]. Killed virus (KV) vaccines and modified

live virus (MLV) vaccines for PRRS have been licensed for more than two decades. These vaccines generally reduce the severity of clinical signs and the transmission of virus; however, they do not induce sterilizing immunity. In general, the efficacy of MLV vaccines is superior to that of KV vaccines [10–13]. Importantly, however, highly variable and generally sub-optimal levels of protection against heterologous PRRSV strains have been reported [10,14–18]. The potential for reversion to the virulence of MLV vaccines and subsequent transmission to susceptible pigs is a concern [19–21].

The major challenge for the development of efficacious, broadly protective PRRS vaccines is the extraordinary genetic variation among disease producing PRRSV strains. Currently, two major PRRSV genotypes are recognized, Genotype 1 (European) and Genotype 2 (North American), which exhibit nearly 40% sequence dissimilarity [22,23]. Genotype 1 and Genotype 2 PRRSVs can be further divided into four subgroups and nine different lineages, respectively, based on phylogenetic analysis of ORF5 [24]. The co-existence of multiple variants within one farm, or even within individual pigs, possibly indicates the occurrence of quasispecies variation of PRRSV [25]. Thus, the development of vaccines that can induce protection against different subgroups and lineages is highly desirable. Multi-subunit vaccines [26], consensus vaccines [27], molecular breeding of different viral structural proteins [28], Mosaic T-cell vaccines [29,30], a polyvalent vaccine containing five different live-attenuated PRRSV strains [31] and DNA prime-MLV boost [32] are just some of the vaccines that are or have been under evaluation recently.

Studies on potential vaccines against human immune deficiency virus type 1 (HIV-1), have shown that Mosaic sequences designed from naturally occurring HIV sequences could effectively address genetic diversity when used as vaccines [33,34]. Mosaic vaccines have been shown to elicit broader immune responses successfully and to confer cross protection in non-human primates and these Mosaic HIV vaccines have been trialed in humans [34–38]. We previously reported that a PRRSV Mosaic vaccine based on 748 GP5 sequences of the Genotype 2 PRRSV strain was immunogenic [29] and induced reactivity to four divergent Genotype-2 PRRSV strains that had at least 10% or more difference in their GP5 sequence—as shown by the expression of interferon gamma (IFN-γ) by peripheral blood mononuclear cells—and, further, that vaccinated pigs were protected against challenge with VR2332 [30]. The present study demonstrates further that vaccination of pigs with the GP5-Mosaic by using a DNA prime and rVaccinia boost approach protects them against strains that are more than 10% divergent in their sequences (VR2332 and MN184C).

#### **2. Materials and Methods**

#### *2.1. Viruses and Cells*

The viruses used in the study included the VR2332 NA reference strain (ATCC VR-2332) and the MN184C strain provided kindly by Drs. Kelly Lager and Kay Faaberg at U.S. Department of Agriculture-Agriculture Research Service (USDA-ARS). The viruses were propagated in MARC-145 cells [39] grown in Dulbecco's modified Eagle's medium (DMEM) containing 2 mM L-glutamine, 100 U penicillin/mL, 100 μg streptomycin/mL and 10% fetal bovine serum (FBS). The Reed–Muench formula was used to calculate virus titers [40]. Viruses were purified over continuous cesium chloride gradients, quantified using a NanoDrop 1000 spectrophotometer (Thermo Scientific, Waltham, MA, USA) and stored at −80 ◦C for later use. Titrated viruses were used for ex vivo recall immune response assay and challenge experiments.

#### *2.2. Transfer Vector Construction*

A combination of gene synthesis (DNA 2.0, Menlo Park, CA, USA) and standard sub-cloning was utilized to generate three transfer vectors. Each transfer vector contained a cassette with the tetR gene (based on GenBank: X00694), and either a natural or Mosaic version of Genotype 2 PRRSV ORF 5 [29,30], followed by the EMCV IRES (based on GenBank: NC\_001479.1) and the EGFP gene (based on plasmid pEGFP-1, GenBank: U55761). The tetR gene and the ORF 5–IRES-EGFP genetic segment were placed under back-to-back synthetic VACV early/late promoters [41]. The tetO2 operator sequence was inserted after the putative VACV D6R promoter, as described in [42]. The cassettes were flanked by 600 bp of the VACV D5R gene to the left and 600 bp of the VACV D6R gene to the right (based on GenBank: NC\_006998.1). Restriction endonuclease analysis was used to confirm plasmid identity.

#### *2.3. Generation of Recombinant VACVs and Preparation of High-Titer Stocks*

Standard homologous recombination was used to generate Tetracycline-inducible recombinant VACVs after the transfection of transfer vectors with FuGene HD Transfection Reagent (Promega, Madison, WI, USA) into BS-C-1 cells, infected 1 h before with an IPTG-inducible VACV strain (based on the WR clone 9.2.4.8.) in the presence of 0.1 mM IPTG and 1 μg/mL doxycycline. Recombinant EGFP-positive tetracycline-inducible VACVs were plaque purified in the absence of Isopropyl βd-1-thiogalactopyranoside (IPTG) and in the presence of 1 μg/mL doxycycline. The elimination of the parental virus was confirmed by PCR analysis of viral DNA that was purified using a Nucleospin® Blood kit (Macherey-Nagel, Düren, Germany). High-titer stocks were generated by infecting HeLa S3 cells with recombinant virus at a multiplicity of infection (MOI)of 0.1 in the presence of 1 μg/mL doxycycline. Infected cells were harvested and homogenized 4 days post-infection. Homogenates were clarified by centrifugation at 750× *g* for 10 min, purified over a sucrose cushion [43], and resuspended in 1X PBS pH 7.3 (with no doxycycline).

#### *2.4. Protein Sequence Alignment and Phylogenetic Analysis*

MEGA 7 (Molecular Evolutionary Genetics Analysis) (www.megasoftware.net) (NIH, Bethesda, MD, USA) was utilized to evaluate protein sequence alignments between GP5 sequences of VR2332, MN184C, Mosaic 1 and Mosaic 2 and to create a phylogenetic tree.

#### *2.5. Vaccination and Collection of Samples*

Cross-bred three- to four-week-old male and female piglets, which were free of PRRSV and porcine circovirus-2, were used in this study. The experimental design is summarized in Table 1. GP5-Mosaic, GP5-WT and vector-control vaccines were administered by both intradermal and intramuscular injection. Briefly, 1 mL vaccine containing 500 μg DNA and 100 μg Quil-A® as adjuvant was injected intradermally (0.1 mL) on the back of ear and intramuscularly (0.9 mL) in the neck region at day 0 and again on day 14. The animals were further boosted at day 28 with a 1 mL suspension containing 108 PFU VACV expressing GP5-Mosaic, GP5-WT or VACV and 100 μg Quil-A® as adjuvant. Pigs were challenged at day 35 with VR2332 and at day 37 with MN184C to their respective groups. A total of 10<sup>6</sup> TCID50 of each virus was administered by intranasal and intramuscular challenge to each animal. Blood was collected at days 0, 7, 14, 21, 28, the challenge day, and 4, 7, 10, and 14 days after challenge.



<sup>a</sup> One mL vaccine containing 500 μg DNA and 100 μg Quil-A® as adjuvant was injected intradermally (0.1 mL) on the back of ear and intramuscularly (0.9 mL) in the neck region at day 0 and again at day 14. VACV 108 PFU and 100 μg Quil-A® as adjuvant was administered at day 28. <sup>b</sup> Pigs were challenged at day 35 with 106TCID50 VR2332 (*n* = 4/group) and day 37 with 106TCID50 MN184C (*n* = 4/group) both intranasally and intramuscularly.

The pigs were euthanized at 14 days post-challenge. The lungs were evaluated macroscopically, weighed, and bronchoalveolar lavage (BAL) was then performed. Tissue samples were collected from each lung lobe, tracheobronchial lymph nodes (TBLN), spleen, inguinal lymph nodes (ILN) and tonsils: portions of each were either fixed in 10% neutral buffered formalin or kept frozen. Fixed tissues were routinely processed to paraffin, and 4–5 μm sections were stained with hematoxylin and eosin and evaluated histologically. Lung lesion scores were calculated as reported before [44]. All animal work was performed under a protocol approved by the University of Connecticut Institutional Animal Care and Use Committee, the animal protocol number under which the study was conducted is IACUC A17-003.

#### *2.6. Measurement of IFN-*γ *Immune Response to PRRSV*

Peripheral blood mononuclear cells harvested from blood collected at days 0 and on the challenge day, were seeded in 24-well flat-bottom plates (5 <sup>×</sup> 105 cells/well). They were then stimulated, in duplicate, by adding 200 TCID50 VR2332 or MN184C/well, or mock treated for 48 h at 37 ◦C in a 5% CO2 atmosphere. The cells were then collected and total RNA was extracted for quantitative real-time PCR analysis.

#### *2.7. Serum Neutralization Test*

Virus neutralization testing was carried out as previously reported [30]. Briefly, each serum was mixed with equal volumes of DMEM containing 100 TCID50 of VR2332 or MN184C and the mixtures were incubated at 37 ◦C for 1 h. The serum–virus mixtures were then added to 96-well plates containing 80–90% confluent MARC-145 cells and incubated for 48 h at 37 ◦C in a 5% CO2 atmosphere (final serum dilution 1:4). The VR2332 or MN184C virus, negative serum, and uninfected cells served as the virus control, negative serum control and cell control, respectively. The neutralizing capacity of serum was quantified by measuring the viral copy numbers by RT-qPCR in supernatants 48 h after the addition of pre-incubated serum–virus mixtures to the cells.

#### *2.8. Quantitative Real-Time PCR*

TRIzol LS Reagent or TRIzol Reagent (Invitrogen, Grand Island, NY, USA) was used to extract total RNA from serum/ culture supernatants or tissues, respectively. The extracted RNA was quantified in a NanoDrop 1000 spectrophotometer (Thermo Scientific, Waltham, MA, USA). The cDNA was synthesized from each sample RNA using a 20 μL reaction mixture and random primers (Invitrogen, Grand Island, NY, USA) and the following reaction conditions: 26 ◦C for 10 min, 42 ◦C for 45 min and 75 ◦C for 10 min. The cDNA was then amplified by SYBR Green real-time PCR. To this effect, SYBR qPCR Master Mix (Bimake, Huston, TX, USA), the cDNA template and 5 -ATG ATG RCC TGG CAT TCT-3 and 5 -ACA CGG TCG CCC TAA TTG-3 were utilized as the forward and reverse primers for ORF7, respectively. Real time PCR was then performed as follows: 2 min at 95 ◦C, followed by 40 cycles at 95 ◦C for 15 s and 61 ◦C for 1 min using Bio-Rad CFX96 Touch System (Bio-Rad, Hercules, CA, USA). For use in quantification, a standard curve was generated using serial dilutions of viral RNA which contained 102–107 copies/μL. Both positive and negative reference samples were run concurrently with the test samples. Viral loads were determined by plotting the Ct values against the standard curve. Melting curves were analyzed to verify the specificity of the PCR.

To test for IFN-γ expression, total RNA was extracted by TRIzol Reagent from virus-stimulated or mock-treated PBMCs; this was then used as template for cDNA synthesis and this, in turn, was used for real-time PCR following the protocol described above. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (forward primer: 5 -CGT CCC TGA GAC ACG ATG GT-3 and reverse primer: 5 -CCC GAT GCG GCC AAA T-3 ) was used as internal control to calculate the fold changes in the expression of IFN-γ (forward primer: 5 -TGG TAG CTC TGG GAA ACT GAA TG-3 and reverse primer: 5 -GGC TTT GCG CTG GAT CTG-3 ) by the delta–delta method [44].

#### *2.9. Lung Lesion Scoring*

The scoring of lung lesions was done systematically, as previously described [45], by a board-certified veterinary pathologist, blinded to the treatment groups. Nine sections, representing all six lung lobes, were scored, with one section from each division of the left cranial lung lobe and two sections from each diaphragmatic lobe.

#### *2.10. Statistical Analysis*

Two-way ANOVA or Student's t-test was used to evaluate the differences in measurements between the samples within or between groups. The data were analyzed using GraphPad Prism (version 7.0) (GraphPad Software, San Diego, CA, USA).

#### **3. Results**

#### *3.1. Sequence Alignment and Analysis of GP5*

The GP5 amino acid (aa) sequences in Mosaic 1, Mosaic 2, MN184C and VR2332 (the prototype of Genotype 2 PRRSV) were all the same size, with no deletions or insertions. Sequence alignments showed aa identity from 85% to 93%. To further investigate the antigenic relationship of these strains, a phylogenetic tree was constructed using the neighbor-joining method b. The four sequences clustered into two subgroups (Figure 1), with the two most distant being Mosaic 1 and MN184C. Mosaic 1 and Mosaic 2 were, relatively, more closely related to VR2332 and MN184C, respectively.

**Figure 1.** Phylogenetic analysis of the GP5 Amino acid sequence of the two Mosaic sequences, MN184C and VR2332. The analysis was done via the neighbor-joining method using MEGA7.0.

#### *3.2. GP5-Mosaic Vaccines Induced Both Humoral and Cellular Response*

GP5-Mosaic-vaccinated pigs had significantly higher levels of neutralizing antibodies in their serum compared to the vector-control animals, both to VR2332 (*p* < 0.001) and MN184C (*p* < 0.05) The virus neutralizing capability of serum from GP5-WT-vaccinated animals was greater against VR2332 (*p* < 0.01) and against MN184C (*p* > 0.05) when compared to serum from vector control animals (Figure 2A); vaccination with the GP5-Mosaic vaccine resulted in broader recall cellular responses than those induced with the GP5WT vaccine. Thus, a significantly higher (*p* < 0.05) relative fold-change in IFN-γ mRNA expression was detected upon stimulation of PBMCs derived from GP5-Mosaic-vaccinated pigs, with either VR2332 or MN184C strains, when compared to those in equally stimulated PMBCs from vector control pigs (Figure 2B). In contrast, the expression of IFN-γ mRNA in GP5-WT-vaccinated pigs was significantly higher only if their PBMCs were stimulated with

VR2332, when compared to those in the equally stimulated PBMCs of the control pigs at the same time points (Figure 2B). No changes were detected with the PBMCs of vector-control pigs with any type of stimulation.

**Figure 2.** Vaccination-induced humoral and cellular responses. (**A**) The virus neutralization was expressed as viral copy numbers (log10 scale), as measured by RT-qPCR in cell supernatants after the infection of MARC-145 cells with pre-incubated serum–virus mixtures. (**B**) The expression of IFN-γ mRNA as fold changes, by either VR2332 or MN184C-stimulated PBMCs collected on the challenge day. Each dot represents the value of one animal. The variation is expressed as standard error of the means. There were three independent replications. Significant differences were calculated by a two-way ANOVA or Student's *t* test (\* *p* < 0.05, \*\*\* *p* < 0.0001).

#### *3.3. GP5-Mosaic Vaccination Induced Cross-Protection in Pigs*

Viral loads in serum of both GP5-WT and GP5-Mosaic-vaccinated groups were significantly lower than those in the vector-control group at 7, 10 and 14 days after challenge with VR2332 (\* *p* < 0.05, \*\* *p* < 0.01) (Figure 3A). Therefore, the capacity of reducing VR2332 virus loads in serum was relatively similar between the GP5-Mosaic and the GP5-WT vaccines. Furthermore, there was a steady decrease in viral loads from 4 days to 14 days after challenge in both GP5-WT and GP5-Mosaic vaccinated groups, while viral loads increased during the course of challenge and reached their peak at 10 days post-challenge (DPC), then declining slightly at 14 DPC in the vector-control group (Figure 3A). In contrast, the viral loads in serum of animals receiving the GP5-Mosaic-vaccine were significantly lower than those of animals receiving the GP5-WT vaccine or those of vector-control animals at 7 and 14 DPC with MN184C (\* *p* < 0.05) (Figure 3B), which was different compared to VR2332-challenged groups. Viral load levels in GP5-WT vaccinated animals, increased during the infection with MN184C and reached their peak at 10DPC, which is comparable to the vector-control group. The viral loads in GP5-Mosaic vaccinated animals decreased steadily from 4 to 10 DPC and remained unchanged until 14 DPC (Figure 3B). Furthermore, the viral loads in lung, TBLN, spleen, ILN and tonsils were significantly lower in GP5-Mosaic-vaccinated animals than those in the vector-control group (\* *p* < 0.05) after challenge with MN184C, while those of the GP5-WT group were similar or slightly lower compared to the vector-control group (Figure 3C). Viral loads in the lungs, TBLN, and tonsil were significantly lower in GP5-Mosaic-vaccinated animals than those in the vector-control group (\* *p* < 0.05) after challenge with VR2332, while those of the GP5-WT group were significantly lower in the lungs, spleen and ILN compared to the vector-control group (\* *p* < 0.05). There was no significant difference in VR2332 virus loads between the GP5-Mosaic and GP5-WT groups (Figure 3D).

**Figure 3.** Virus clearance in sera and tissues. (**A**) Viral copy numbers in serum from 0 to 14 DPC upon VR2332 challenge. (**B**) The viral copy numbers in serum from 0 to 14 DPC upon MN184C challenge. Each dot represents the mean value of each group. The variation bars are expressed as the standard error of the mean. Three separate experiments were performed for each. A two-way ANOVA or Student's *t* test (\* *p* < 0.05, \*\* *p* < 0.01, was used to calculate significant differences; days post-challenge (DPC). (**C**) The viral copy numbers in tissues at necropsy upon VR2332 challenge. (**D**) The viral copy numbers in tissues at necropsy upon MN184C challenge. Each bar represents the mean value of each group. The bars are the standard error of the mean. Three separate experiments were performed for each. Significant differences were calculated by Student's *t* test (\* *p* < 0.05).

Tests in PAMs and BAL fluids showed a similar pattern in which the viral loads in both the GP5-Mosaic and GP5-WT groups were lower than those in the vector-control group after challenge with VR2332 (Figure 4A,C), meanwhile only the viral loads of GP5-Mosaic-vaccinated animals were significantly lower than those in the vector-control animals after challenge with MN184C (\* *p* < 0.05 and \*\* *p* < 0.01, respectively), furthermore there was no apparent difference in MN184C viral loads between GP5-WT and vector-control animals (Figure 4B,D).

**Figure 4.** Virus clearance in bronchioalveolar lavage fluids (BAL) and PAMs. (**A**) The viral copy numbers in BAL fluids at necropsy upon challenge with VR2332. (**B**) The viral copy numbers in BAL fluids at necropsy upon challenge with MN184C. Each bar represents the mean value of each group. Bars represent the standard error of the mean. Three independent experiments were performed for each. Significant differences were calculated by Student's *t* test (\* *p* < 0.05). (**C**) The viral copy numbers in PAMs at necropsy upon challenge with VR2332. (**D**) The viral copy numbers in PAMs at necropsy upon challenge with MN184C. Each dot represents the mean value of each animal. The bars represent the standard error of the mean. Three independent experiments were performed for each. Student's *t* test was used to calculate significant differences (\* *p* < 0.05, \*\* *p* < 0.01).

#### *3.4. Lower Lung Lesion Scores Detected in GP5-Mosaic-Vaccinated Animals*

The lung lesion scores after challenge with MN184C were significantly lower in GP5-Mosaic-vaccinated animals than those in both GP5-WT and vector-control animals when evaluated using nine sections of lung (Figure 5 right panel, *p* < 0.05). In contrast, after VR2332 challenge, lung lesion scores were lower on average in both the GP5-Mosaic and GP5-WT groups than those in vector-control animals. There were no significant differences between the lung lesion scores of GP5-WT and GP5-Mosaic-vaccinated groups after challenge with VR2332 (Figure 5).

**Figure 5.** Lung lesion scores. The lung lesion scores of GP5-Mosaic-vaccinated animals were significantly lower (*p* < 0.05) than those of GP%-WT or vector-control animals after challenge with MN184C. Each figure represents the mean value of each individual. The bars represent the standard error of the mean. Three independent experiments were performed for each. Student's *t* test was utilized to calculate significant differences (days post-challenge (DPC)).

#### **4. Discussion**

There is extraordinary diversity among PRRSV strains. The diversity among PRRSV strains is as high as, or may even surpass that of, HIV. Recently, several vaccine candidates, including a synthetic consensus PRRSV strain [27], a chimeric PRRSV strain containing multiple ORFs DNA shuffled [46–48] and an intranasal live virus vaccine with adjuvant [49] induced various levels of heterologous protection in pigs. We have developed GP5-Mosaic vaccines that incorporate sequences derived from naturally circulating viruses. The data obtained from our GP5-Mosaic vaccinated pigs supports the Mosaic vaccine approach as an effective strategy to address PRRSV diversity. It has also been demonstrated that Mosaic T-cell vaccines have great potential for other viruses with extraordinary diversity, such as HIV [33–38]. In an earlier study, we demonstrated the immunogenicity of GP5-Mosaic vaccines in swine and their ability to induce cross-reactivity, as shown by their broad recall responses ex-vivo, and conferred the partial protection they provide in pigs [29,30]. The ability of the GP5-Mosaic vaccines to induce cross-protection in pigs against heterologous PRRSV strains was confirmed in the present study.

A sequence analysis of two GP5-Mosaic vaccines and the two Genotype 2 PRRSV strains utilized in the study, VR2332 and MN184C, that belong to lineage 5 and to lineage 1, respectively, revealed a 15% difference in the GP5 aa sequences. The aa sequences were used to generate a phylogenetic tree using the neighbor-joining method [24]. The GP5-Mosaic 1 was closer to VR2332 (93% aa identity), while GP5-Mosaic 2 was closer to MN184C (90% aa identity), which suggested that the two GP5-Mosaic sequences broaden the coverage to strains belonging to different lineages. Based on this analysis, a broader protection could be expected with GP5-Mosaic vaccines. The data of the present study supports clearly the capacity of the GP5-Mosaic vaccine to cross-protect pigs against VR2332 and MN184C that share only 85% aa identity.

VACV provides a very good platform for the development and testing of vaccines, including animal vaccines, human vaccines and cancer immunotherapy [50]. In our study, a GP5-Mosaic DNA vaccine prime/GP5-Mosaic rVaccinia (VACV) boost regime was used to immunize pigs. This type of prime-boost regime has previously been demonstrated to increase vaccination efficacy substantially [51,52]. In addition, Quil-A®, used as adjuvant for the priming of the GP5-Mosaic DNA vaccine, was previously shown to further increase the immune responses [53–55]. Therefore, the prime-boost vaccination regime used here for testing GP5-Mosaic vaccine efficacy, in terms of broad protection, proved appropriate for testing the hypothesis.

PBMCs collected on the challenge day from GP5-Mosaic-vaccinated pigs responded to a broader range of virus strains, as measured by IFN-γ mRNA expression, than those from GP5-WT vaccinates. Thus, PBMC from GP5-Mosaic-vaccinated pigs expressed higher levels of IFN-γ mRNA in response to stimulation with VR2332 or MN184C strains, respectively, compared to those of the vector-control pigs. In contrast, PBMC from GP5-WT vaccinated pigs expressed higher levels of IFN-γ mRNA only when stimulated with VR2332. These data were consistent with those of our previous study where the GP5-Mosaic vaccine induced broad recall cellular responses, as measured by significantly higher levels of IFN-γ mRNA expression in response to the four diverse Genotype 2 PRRSV strains tested, including VR2332, NADC9, NADC30 and SDSU73 [30]. The significant induction of IFN-γ is an important asset of a PRRSV vaccine, as IFN-γ reportedly plays a critical role in the control of and protection from PRRSV infection [56]. In addition, both the GP5-Mosaic and the GP5-WT vaccines induced relatively high levels of neutralizing antibodies in VR2332-challenged pigs. These data indicate that the GP5-Mosaic vaccine preserves virus-neutralizing epitopes that appear to contribute to the overall efficacy of this vaccine.

In terms of vaccine-induced protection, both GP5-Mosaic and GP5-WT-vaccinated pigs had significantly lower serum viral loads at 7, 10 and 14 DPC than the vector-control group after the VR2332 challenge. In contrast, only the GP5-Mosaic-vaccinated group showed significantly lower viral loads in sera at 7 and 14 DPC than the vector-control group after the MN184C challenge. These data demonstrate that the GP5-Mosaic vaccine is capable of cross-protecting pigs against two divergent PRRSV strains (15% difference) while the GP5-WT vaccine only provided protection in pigs against the homologous strain. Furthermore, a similar pattern was found with respect to viral loads in tissues, where the GP5-Mosaic-vaccinated pigs had significantly lower viral loads in their lungs, TBLN, spleen, ILN and tonsils compared to the vector-control group after challenge with MN184C, while the GP5-WT group only showed slightly lower (numerically, but not significant) viral loads. In the VR2332 challenge experiment, both the GP5-Mosaic and GP5-WT vaccinates had significantly lower viral loads in tissues generally—the viral loads in the GP5-Mosaic vaccinates were significantly lower in their lungs, TBLN and tonsils, while viral loads in the GP5-WT group were significantly lower in their lungs, ILN and spleen, compared to the vector-control group. Similar protective effects were also demonstrated in the GP5-Mosaic vaccine vaccinates, where lower viral loads were detected in the BAL fluids or PAMs after MN184C challenge, further confirming the cross-protective capability of the GP5-Mosaic vaccine. Indeed, this is finding may be particularly relevant since PAMs are the primary target of PRRSV infection. Lung lesion scoring is an essential and crucial method to evaluate vaccine efficacy, as infection by PRRSV tends to increase susceptibility to coinfection with other virus or bacterial pathogens [57–59] which may result in more severe lesions and mortality. Pigs vaccinated with the GP5-Mosaic vaccine had significantly lower lung lesion scores than both the GP5-WT and vector-control groups after challenge with MN184C, while both the GP5-Mosaic and GP5-WT groups had lower lung lesion scores compared to the vector-control group in the VR2332 challenge experiment.

#### **5. Conclusions**

Taken altogether, our results clearly demonstrate that the GP5-Mosaic vaccine provided cross-protection against two heterologous PRRSV strains, VR2332, the prototype Genotype 2 strain, and MN184C, a highly virulent strain, while the GP5-WT vaccine provided only homologous protection against VR2332. In summary, our results demonstrate that our GP5-Mosaic vaccine is able to induce broader protection against PRRS and warrants further research for field applications.

**Author Contributions:** Conceptualization, P.H.V. and A.E.G.; Methodology, J.C., C.M.O., J.A.S., P.H.V. and A.E.G.; Formal analysis, J.C. and A.E.G.; Investigation, J.C., C.M.O., C.H., K.S., J.A.S., P.H.V., H.J.V.K., and A.E.G.; Data curation, J.C. and A.E.G.; Writing—original draft preparation, J.C.; Writing—review and editing, J.C., C.M.O., J.A.S., P.H.V. and A.E.G.; Supervision, A.E.G.; Project administration, A.E.G.; Funding acquisition, J.A.S., P.H.V. and A.E.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was funded by USDA/NIFA Grant Number 2011 67015-30176 and NC229.

**Acknowledgments:** The authors thank Kay Faaberg and Kelly Lager kindly providing the PRRSV strains.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **Significant Interference with Porcine Epidemic Diarrhea Virus Pandemic and Classical Strain Replication in Small-Intestine Epithelial Cells Using an shRNA Expression Vector**

### **Da Shi** †**, Xiaobo Wang** †**, Hongyan Shi** †**, Jiyu Zhang, Yuru Han, Jianfei Chen, Xin Zhang, Jianbo Liu, Jialin Zhang, Zhaoyang Ji, Zhaoyang Jing and Li Feng \***

State Key Laboratory of Veterinary Biotechnology, Harbin Veterinary Research Institute, Chinese Academy of Agricultural Sciences, Xiangfang District, Haping Road 678, Harbin 150069, China; dashi198566@163.com (D.S.); wxb1901@163.com (X.W.); shy2005y@163.com (H.S.); 18841618894@163.com (J.Z.); hyr1968274346@163.com (Y.H.); chenjianfei@126.com (J.C.); zhangxin2410@163.com (X.Z.);

liujianbo@caas.cn (J.L.); zhangjialin0106@gmail.com (J.Z.); Zy\_ji2010@163.com (Z.J.); 15204604415@163.com (Z.J.)

**\*** Correspondence: fengli@caas.cn; Tel.: +86-0451-5105-1720

† These authors contributed equally to this work.

Received: 27 August 2019; Accepted: 16 October 2019; Published: 2 November 2019

**Abstract:** Porcine epidemic diarrhea (PED) re-emerged in China in 2010 and is now widespread. Evidence indicates that highly virulent porcine epidemic diarrhea virus (PEDV) strains belonging to genotype G2 caused a large-scale outbreak of diarrhea. Currently, vaccines derived from PEDV classical strains do not effectively prevent infection by virulent PEDV strains, and no specific drug is available to treat the disease. RNA interference (RNAi) is a novel and effective way to cure a wide range of viruses. We constructed three short hairpin RNA (shRNA)-expressing plasmids (shR-N307, shR-N463, and shR-N1071) directed against nucleocapsid (N) and determined their antiviral activities in intestine epithelial cells infected with a classical CV777 strain and LNCT2. We verified that shR-N307, shR-N463, and shR-N1071 effectively inhibited the expression of the transfected *N* gene in vitro, comparable to the control shRNA. We further demonstrated the shRNAs markedly reduced PEDV CV777 and LNCT2 replication upon downregulation of N production. Therefore, this study provides a new strategy for the design of antiviral methods against coronaviruses by targeting their processivity factors.

**Keywords:** porcine epidemic diarrhea virus; RNA interference; processivity factor; intestine epithelial cells; *N* gene

#### **1. Introduction**

Porcine epidemic diarrhea (PED), caused by porcine epidemic diarrhea virus (PEDV), is a highly contagious intestinal infectious disease. PED is an important disease in swine-producing countries. PED causes the death of newborn piglets and weight loss in pigs of all ages from PEDV-induced severe symptoms, such as serious diarrhea, vomiting, and dehydration, which seriously damage the swine industry [1]. Following reports in 1978 [1], PED had an outbreak in swine-farming countries in Asia, North America, South America, and Europe [2–5]. Starting from the end of 2010, highly virulent PEDV variants that differed from the classic European strain CV777 were widespread in China, resulting in high mortality of newborn piglets and huge economic losses [6–9]. PEDV can be divided into genotypes G1 and G2 based on phylogenetic analysis of full-length *S* gene sequences [10]. PEDV strains detected in China since 2010 mostly belonged to genotype G2, which differed genetically from the CV777 vaccine strain belonging to subtype G1 [11]. However, many studies demonstrated that

commercially available PEDV vaccines derived from classical strains of PEDV do not provide effective protection against highly virulent PEDV variant infections in China [3]. Thus, PEDV infection remains a major veterinary problem. Understanding it is essential to developing novel antiviral drugs that will specifically inhibit PEDV propagation.

RNA interference (RNAi) is a short, double-strand RNA-induced process that targets and degrades the messenger RNA (mRNA) of specific sequences [12,13]. Post-transcriptional gene silencing can be mediated by exogenous small interfering RNAs (siRNAs), endogenous microRNAs (miRNAs), and short hairpin RNAs (shRNAs). Effective gene knockdown is achieved by shRNAs by inducing the endogenous RNAi process [14,15]. Transcribed shRNAs are exported from the nucleus by Exportin-5 and processed by the RNase III Dicer into small double-stranded RNA (dsRNA) molecules of 19 to 23 bp called siRNAs. The complementary guide strand of processed siRNA is incorporated into the RNA-induced silencing complex to mediate the cleavage of target mRNAs [16,17]. RNAi evolves in the host defense system directed at infectious viruses and transposable elements. Effective silencing of transgene expression and endogenous genes in vivo was shown by a number of groups [6,7,18]. These findings raised the possibility that RNAi could be another therapeutic approach for inhibiting virus infection. ShRNAs were employed for therapy against human viral diseases, as well as cancer and neurogenerative diseases [8]. PEDV primarily infects villous epithelial cells throughout the small intestine and causes serious injury of intestine epithelial cells (IECs), including superficial villous enterocyte swelling and severe diffuse atrophic enteritis [19]. Infection of PEDV variant strains in the field complicates the development of prophylactic and therapeutic strategies to protect suckling pigs from diarrhea. As an RNA virus, PEDV could be an ideal target for studying its biology and therapeutics using RNAi.

The PEDV nucleocapsid (N) protein, which is abundantly expressed in infected cells, has multiple functions. It is one of the structural proteins that forms complexes with genomic RNA, enhancing viral transcription and assembly. Therefore, the N protein may be a potential drug target for antiviral therapy against PEDV infection. In this study, we compared and analyzed *N* gene sequences from 25 different PEDV G1 and G2 isolates from different countries. Three novel shRNAs targeting conserved and unexploited regions in the *N* gene were tested for inhibition of PEDV CV777 and LNCT2 replication. Cell viability, viral titer, and protein expression were examined as indicators of the efficacy of targeted gene silencing by the shRNAs. All three shRNAs effectively inhibited PEDV replication and *N* expression of the G1 and G2 subtypes.

#### **2. Materials and Methods**

#### *2.1. Viral Propagation and Titer Assays*

Swine intestinal epithelial cells (IECs) were donated by Dr. Yanming Zhang (Northwest A&F University, China), and derived from the mid-jejunum of neonatal, unsuckled, one-day-old piglets. Primary intestinal epithelial cells were isolated by the tissue explant adherent method and purified by trypsin digestion with citric acid, forming an undifferentiated porcine intestinal epithelial cell line that was immortalized. Cells were maintained in a Dulbecco's minimum essential medium (DMEM; Gibco, Thermo Fisher Scientific, Waltham, MA, USA)/nutrient mixture F-12 (Ham) (1:1) containing 10% heat-inactivated fetal bovine serum (FBS; Gibco), 5 ng/mL epidermal growth factor (Life Technologies, Carlsbad, CA, USA), 5 mg/mL insulin-transferring selenium supplements (Life Technologies), and 1% penicillin–streptomycin (Life Technologies). Cell culture media were changed every two days, and cells were passaged every 3–4 days by trypsinization with 0.25% trypsin–ethylenediaminetetraacetic acid (EDTA) (Life Technologies). Vero E6 (African green monkey kidney cells, American Type Culture Collection (ATCC)) were grown and maintained in DMEM supplemented with 10% heat-inactivated FBS and penicillin–streptomycin, and incubated at 37 ◦C with 5% CO2.

The PEDV G1 CV777 vaccine strain (GenBank accession: AF353511.1) was preserved at Harbin Veterinary Research Institute (Harbin, China). PEDV G2 strain LNCT2 (GenBank accession: KT323980.1) was isolated in Vero E6 cells in our laboratory. Vero E6 cells were cultured and used to amplify PEDV as previously described [9]. After 70% of virus-infected cells showed cytopathic effects (CPEs), cultures were collected for three freeze–thaw cycles. Viral titration used 96-well microplates with Vero E6 cells. Viral cultures were 10-fold serially diluted with virus replication medium containing trypsin (10 μg/mL). Confluent Vero E6 cells from microplates were washed three times with phosphate-buffered saline (PBS) and inoculated at 0.1 mL per well into eight wells. Following adsorption for 1 h at 37 ◦C, the inocula were removed, and cells were washed three times with PBS. Subsequently, 0.1 mL of fresh virus replication medium was transferred into each well, and cells were incubated 4–5 days at 37 ◦C. The 50% tissue culture infective dose (TCID50) was expressed as the reciprocal of the highest dilution showing CPE by the Reed and Muench method. Assays were performed in triplicate in three independent experiments.

#### *2.2. Plasmid Construction*

Plasmids expressing GFP-tagged N and myc-Tagged N were described previously [20]. The design of shRNAs targeting the PEDV strain LNCT2 genome *N* gene (Figure 1A) used methods from the literature [21] and the web-based Block-iTTM RNAi Designer program [22]. A Basic Local Alignment Search Tool (BLAST) search [23] was performed to exclude possible homologous sequences. Three individual targeting sites were selected and chemically synthesized (Sangon Biotech, Shanghai, China) (Table 1). Control shRNA was designed at the same time to have no homology with PEDV or the IEC cell genome. All sequences were arranged as *Bbs*I + sense + loop + antisense + termination signal + *BamH*I and cloned into the pGPU6-Hygro vector to make the shRNA-expressing plasmids shR-N307, shR-N463, shR-N1071, and shR-NC (Figure 1B,C). Expression of siRNAs was driven by the U6 promoter.

**Figure 1.** Schematic description of the target viral genome, and small interfering RNA (siRNA)-expressing cassette. (**A**) Genomic structure of porcine epidemic diarrhea virus (PEDV) and position of target short hairpin RNA (shRNA) at *N* gene. ShR-N307, shR-N463, and shR-N1071 indicate initial shRNA target sites; (**B**) sequences and design map for shRNA constructs; (**C**) structure of predicted shRNA.



#### *2.3. Plasmid DNA Preparation*

Plasmid DNA was transformed into electrocompetent DH5α *Escherichia coli* and purified with EndoFree Plasmid Maxi Kits (QIAGEN, Hilden, Germany).

#### *2.4. Cell Transfection and Antiviral Activity In Vitro*

We validated the inhibitory effects of the shRNAs against N by target gene expression in an in vitro transfection system. One day before transfection, IEC cells were seeded into 12-well plates at 5 <sup>×</sup> 104 cells per well without antibiotics and grown at 37 ◦C overnight with 5% CO2. The 60–80% confluent cells were transiently transfected with the indicated plasmid using Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA), according to the manufacturer's instructions. At 48 h after transfection, *N* expression was analyzed by Western blot or inverted fluorescence microscope. To examine the inhibitory effects of shRNAs against N on target gene expression during PEDV replication, IEC cells were transfected with or without 1 μg, 2 μg, or 4 of μg shR-N307, shR-N463, and shR-N1071 or 4 μg of shR-NC for 24 h and infected with 100 TCID50/mL PEDV strain CV777 or LNCT2. At 48 h post-infection (hpi), antigen slides were prepared to detect PEDV N protein expression by Western blot. In parallel experiments, cells from individual wells were collected and frozen and thawed twice, followed by centrifugation at low speed (1000× *g*), and supernatants were serially diluted, inoculated with Vero E6 cells, and titrated by the Reed and Muench method.

#### *2.5. Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis and Western Blots*

IEC cells grown in 12-well plates were transfected and infected as described. Cells were harvested at the indicated time points after transfection or virus infection, washed once with cold PBS, and lysed in radioimmunoprecipitation assay (RIPA) buffer (Sigma-Aldrich, St. Louis, MO, USA) to determine protein concentrations. Equal amounts of protein were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) followed by blotting onto nitrocellulose membranes. Membranes were washed twice in Tris-buffered saline (TBS) and incubated 2 h in Superblock blocking buffer (ThermoFisher, Waltham, MA, USA). Membranes were incubated with anti-N monoclonal antibody (mAb) 3G2 (prepared by our laboratory, diluted 1:1000), anti-myc monoclonal (Sigma, diluted 1:1000) or anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH) monoclonal antibody (Sigma, diluted 1:10,000). Proteins were revealed by IRDye 800CW goat anti-mouse immunoglobulin G (IgG) (H + L) (1:10,000) (LiCor BioSciences, Lincoln, NE, USA), and blots were visualized using an Odyssey infrared imaging system (LiCor BioSciences).

#### *2.6. Cell Viability Assays*

Cell viability assays were performed using cell counting kit-8 (CCK-8) (CK04; Dojindo, Shanghai, China), according to the manufacturer's protocol. In brief, IEC cells were seeded in 96-well plates at 10,000 per well and incubated at 37 ◦C for 24 h. Cells were transfected or not with shRNAs, and plates were incubated for 48 h before 10 μL of CCK-8 was added to wells for incubation for 2 h. Optical density at 450 nm was measured. Viability of treated cells was expressed as a percentage relative to untreated cells.

#### *2.7. Immunofluorescence Assays*

IEC cells were seeded in 12-well plates, and confluent cell monolayers were transfected with shR-N307, shR-N463 and shR-N1071, or shR-NC (4 μg) with Lipofectamine 3000 (Invitrogen) before PEDV infection. Cells were infected with PEDV strain CV777 or LNCT2 at 100 TCID50/mL. PEDV infection was analyzed using immunofluorescence assays (IFAs) at 48 hpi. Cells were fixed with 4% paraformaldehyde at 4 ◦C for 30 min and washed with PBS. Fixed cells were permeabilized with 0.2% Triton X-100 for 15 min at room temperature and blocked with blocking buffer (PBS with 5% bovine serum albumin (BSA)) for 2 h. Preparations were labeled with the mouse anti-PEDV N mAb (1:100 dilution) at 37 ◦C for 2 h followed by labeling with Alexa Fluor 488 goat anti-mouse IgG antibody (1:200 dilution) (ThermoFisher) for 1 h at 37 ◦C. Cell nuclei were stained with 4 ,6-diamidino-2-phenylindole (DAPI) (0.05 μg/mL) (D9542; Sigma) for 15 min and analyzed using an AMG EVOS F1 florescence microscope.

#### *2.8. Statistical Analysis*

All statistical data are expressed as means ± standard deviation (SD) of three independent experiments and analyzed using the Student's *t*-test. A *p*-value <0.05 was considered statistically significant.

#### **3. Results**

#### *3.1. IEC Cells and PEDV Infection*

Phylogenetic analysis of CV777 and LNCT2 was based on complete genomic sequences (Figure 2A). CV777 clustered with PEDV G1 genotype (classic strains), whereas LNCT2 clustered with PEDV G2 genotype (epidemic strains), with >98% nucleotide identity to these strains. To determine PEDV propagation in IEC cells, CPEs and IFAs were used to monitor after infection (Figure 2B). At 48 hpi, CV777 and LNCT2 caused similar CPEs in IEC cells, characterized by rounding, aggregation, and rupturing. The morphology of negative control cells remained unchanged and did not exhibit any signs of CPEs. Inoculation of CV777 and LNCT2 with IEC cells resulted in positive immunofluorescent staining with mAb that recognized the virus N protein. Controls were negative for immunostaining.

**Figure 2.** Phylogenetic analysis and virus-induced cytopathic effect (CPE) in intestine epithelial cells (IECs). (**A**) Phylogenetic trees of PEDV based on complete genomic DNA. Black squares, strains in United States of America (USA); black triangles, strains in South Korea; black rhombuses, strains in China; red circle, G1 CV777 strain from this study; red rhombus, G2 LNCT2 strain from this study. (**B**) Production and growth properties of PEDV CV777 and LNCT2 in IEC cells. CPE 48 h post-infection (hpi) (upper panels, bar: 400 μm) and immunofluorescence assay (IFA) 48 hpi (lower panels, bar: 200 μm) of CV777 and LNCT2 in IEC cells. IEC cells were infected with PEDV CV777 or LNCT2 at 100 50% tissue culture infective dose (TCID50)/mL. CPE and IFA were examined at 48 hpi, and cell images were captured.

#### *3.2. Selection of Targeted Sites for N Gene-Specific shRNAs*

The prevalent strains that caused the outbreak of PEDV in China in 2010 and in North America in recent years belonged to the G2 genotype. PEDV is a positive-stranded RNA virus with higher mutation rates than DNA viruses [24]. Specific shRNAs were designed to target the conserved gene. We used the *N* gene of LNCT2 as the target sequence. Using nucleotide substitutions in the *N* gene and previous reports [25], we chose three unused regions that were well conserved in *N* genes among 25 isolates of PEDV G1 and G2 genogroups. The sites were at +307 to +327 nt, +463 to +483 nt, and +1071 to +1091 nt relative to the 5 ATG initiation codon (Table 1).

#### *3.3. Cell Viability Is Not A*ff*ected by Plasmids with shRNAs*

ShRNAs induce large amounts of cell death when transfected at large volume, resulting in interference with experimental results. To detect if the three shRNAs we used led to cytotoxicity at 4 μg, IEC cells were seeded in 96-well microplates and transfected with shR-N307, shR-N463, shR-N1071, or shR-NC (4 μg) using Lipofectamine 3000 (Invitrogen), or transfected with transfection reagent alone as mock. After transfecting for 48 h, CCK-8 solution (10 μL) was added to wells, and plates were incubated at 37 ◦C for 2 h. Absorbance was measured at 450 nm using a microtiter plate reader (Bio-Rad, Hercules, CA, USA). Viable cells that were mock transfected with transfection reagent alone were the reference of 100% cell viability. ShR-N307, shR-N463, and shR-N1071 exhibited no obvious cytotoxicity in transfected Vero E6 and IEC cells at 4 μg (Figure 3).

**Figure 3.** ShRNA treatment did not affect cell viability. Cell viability was detected by cell counting kit-8 (CCK-8) assays after transfection of IEC cells with shRNAs for 48 h. Absorption at 450 nm was recorded and expressed as a percentage of relative cell viability. Values are means ± SD (*n* = 3). Significant differences were assessed by Student's *t*-test; ns, no significant difference compared to control; *p* > 0.05.

#### *3.4. E*ffi*cient Inhibition of PEDV Myc-N or AcGFP-N Expression by shRNA Expression Cassettes*

We designed three shRNAs specifically targeting regions shown in Figure 1, named shR-N307, shR-N463, and shR-N1071. Western blots determined the inhibitory effects of the shRNAs on *N* gene expression. PMyc-N was co-transfected into IEC cells with increasing doses of shR-N307, shR-N463, or shR-N1071 using Lipofectamine 3000. At 48 h after transfection, N protein expression was analyzed by Western blots using mouse anti-Myc mAb. Western blots demonstrated that shR-N307, shR-N463, and shR-N1071 inhibited Myc-tagged N protein expression in a dose-dependent manner (Figure 4A).

**Figure 4.** Inhibitory effects of shR-N307, shR-N463, and shR-N1071 on the PEDV N protein. (**A**) Western blots for effect of shR-N307, shR-N463, and shR-N1071 on *N* gene expression. PMyc-N was co-transfected with increasing doses of shR-N307, shR-N463, or shR-N1071 into IEC cells. After 48 h, transfected cells were lysed. Equal amounts of cell lysates were resolved by 12.5% SDS-PAGE. Reaction products were probed with anti-Myc or anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH). Densitometric data for N/GAPDH from three independent experiments are shown as means ± SD. \* *p* < 0.05. The *p* value was calculated using Student's *t*-test; (**B**) shR-N307, shR-N463, and shR-N1071 influence on pAcGFP-N expression in cultured IEC cells. PAcGFP-N was co-transfected with increasing doses of shR-N307, shR-N463, or shR-N1071 into IEC cells. PAcGFP-N expression plasmid was the unrelated control. Images show enhanced GFP (EGFP) expression at 48 h post-transfection.

The *N* gene was fused with the *AcGFP* gene to make pAcGFP-N. The effect of shRNAs on *N* gene expression was monitored by *AcGFP* expression. Plasmid pAcGFP-N was transfected into IEC cells alone or with the indicated concentrations of shR-N307, shR-N463, or shR-N1071 expression cassettes. At 48 h post-transfection, the effects of shRNA on enhanced GFP (EGFP) expression were monitored by fluorescence microscopy. The shR-N307, shR-N463, and shR-N1071 expression cassettes targeting *N* gene inhibited the expression of *AcGFP* to some extent compared with cells transfected with plasmid pAcGFP-N alone. The shR-N307, shR-N463, and shR-N1071 expression cassettes targeting the *N* sequence were similar, with weak fluorescence observed from cells transfected with 4 μg of shRNAs (Figure 4B).

#### *3.5. ShRNA-Mediated Inhibition of N Expression and PEDV Production in Infected IEC Cells*

To examine the inhibitory effects of shRNAs against N on target gene expression during CV777 and LNCT2 replication, IEC cells were transfected with indicated plasmids for 24 h and then infected with PEDV CV777 orLNCT2 strain (100 TCID50/mL). At 48 hpi, PEDV-infected cells were lysed with RIPA buffer containing 1 mM phenylmethylsulfonyl fluoride (PMSF), and Western blots were used to analyze *N* expression using mAb 3G2. Treatment with shR-N307, shR-N463, and shR-N1071 significantly reduced *N* expression compared to shR-NC- or mock-transfected IEC cells. The inhibitory effects of the three shRNAs on *N* expression were elevated with increased shRNA concentration (Figures 5A and 6A). These results show that *N*-targeted shRNA efficiently inhibited target gene expression during PEDV replication.

**Figure 5.** Dose-dependent inhibitory effects of shR-N307, shR-N463, and shR-N1071 against N on target gene expression and PEDV CV777 replication in infected IEC cells. IEC cells with shRNA (1, 2, 4 μg) for 24 h before infection with PEDV CV777 for 48 h. (**A**) PEDV N protein by Western blot with anti-N protein monoclonal antibody (mAb). Densitometric data for N/GAPDH from three independent experiments, shown as means ± SD. \* *p* < 0.05. The *p* value was calculated using Student's *t*-test; (**B**) PEDV CV777 titers in shRNA-transfected IEC cells. ShRNA transfection and viral infection were as in panel A. Viral titers in supernatants collected at 48 hpi were determined using the Reed–Muench method. Error bars represent standard errors of the mean from three independent experiments. \* *p* < 0.05. The *p* value was calculated using Student's *t*-test.

**Figure 6.** Dose-dependent inhibitory effects of shR-N307, shR-N463, and shR-N1071 directed against N on target gene expression and PEDV LNCT2 replication in infected IEC cells. IEC cells with shRNA (1, 2, 4 μg) for 24 h were infected with PEDV LNCT2 for 48 h. (**A**) PEDV N protein was detected by Western blot with anti-N protein mAb. Densitometric data for N/GAPDH from three independent experiments are shown as means ± SD. \* *p* < 0.05. The *p* value was calculated using Student's *t*-test; (**B**) PEDV LNCT2 titers in shRNA-transfected IEC cells. ShRNA transfection and viral infection were as in panel A. Viral titers in supernatants collected at 48 hpi were determined using the Reed–Muench method. Error bars represent standard errors of the mean from three independent experiments. \* *p* < 0.05. The *p* value was calculated using Student's *t*-test.

To determine if downregulation of *N* expression by shRNAs decreased PEDV CV777 and LNCT2 strain replication, IEC cells were transfected with indicated plasmids for 24 h and infected with CV777 or LNCT2 (100 TCID50/mL). Cell cultures were collected at 48 hpi, and viral titers were determined. Transfection with shR-N307, shR-N463, or shR-N1071 significantly reduced PEDV CV777 and LNCT2 replication compared to shR-NC- or mock-transfected IEC cells (*p* < 0.05) (Figures 5B and 6B). Although differences in inhibition of PEDV CV777 and LNCT2 strain replication induced by 1 μg, 2 μg, or 4 μg of shR-N307, shR-N463, and shR-N1071 were seen, PEDV replication gradually reduced with increased shRNA concentrations.

#### **4. Discussion**

In this study, we demonstrated that shRNAs against PEDV N protein broadly inhibited PEDV G1 and G2 strains. This report found that shRNAs inhibit swine coronavirus replication in epithelia. Most coronaviruses such as severe acute respiratory syndrome coronavirus (SARS-CoV), Middle East respiratory syndrome coronavirus (MERS-CoV), and PEDV infect epithelial cells in the respiratory and/or enteric tracts. The control of coronavirus diseases is important for human public health security. Further developing RNAi as a potential therapeutic agent against coronavirus infection is worthwhile.

The PEDV N protein is predominantly produced in susceptible cells, which makes it a major target for early and accurate diagnosis [26]. The N protein forms complexes with coronavirus genomic RNA and enhances the viral transcription and assembly. Because of the importance of N protein in viral replication and the coronavirus *N* gene long used as a major target for shRNA design, we focused on the inhibition of PEDV infection using RNAi targeting the *N* gene of PEDV and systematically evaluated suppression efficiency. Alterations of viral genomes such as nucleotide substitution, insertion, and deletion have the potential to decrease the inhibitory effect of shRNA. We analyzed 25 *N* gene sequences derived from PEDV genotypes G1 and G2 and only found single-nucleotide substitutions. Considering the nucleotide substitutions in the *N* gene, we chose three conserved and unexploited regions for the design of shRNAs. PEDV is primarily transmitted through the fecal–oral route and infects intestinal villous epithelial cells in vivo [27]. Current in vitro cell cultures of PEDV include Vero E6 cells, MARC-145 cells (another monkey kidney cell line), and HEK293 cells [28]. Most are non-porcine intestinal epithelial cells and, thus, not ideal in vitro cellular models for studying the interaction between PEDV infection and the host response due to interspecific variation. The IEC cell line represented a better model of normal porcine intestinal epithelium than transformed cell lines, and provided a unique opportunity to explore host–pathogen interactions in an in vitro system [29]. Using this cell model, we showed that shRNA expression did not affect the viability of IEC cells. Co-transfection of recombinant plasmid pMyc-N and different concentrations of shRNAs highlighted the success of the gene knockdown at the protein level. GFP was an important reporter for *N* gene expression. For example, if the *N* gene was silenced by the shRNAs, the translation of GFP protein in frame with the *N* gene in the recombinant plasmid would also be inhibited. This hypothesis was consistent with our findings. Our shRNAs shR-N307, shR-N463, and shR-N1071 strongly reduced GFP expression, and the inhibition was dose-dependent. In addition, the model system was exploited to further study the pathological functions of PEDV genes. We assessed the capacity of shR-N307, shR-N463, and shR-N1071 for inhibiting the gene and protein expressions of PEDV CV777 and LNCT2 strains in vitro. We observed that shR-N307, shR-N463, and shR-N1071 were able to knock down target gene expression of PEDV CV777 and LNCT2 strains at 48 hpi. We also showed that shR-N307, shR-N463, and shR-N1071 against the *N* gene suppressed PEDV CV777 and LNCT2 strain replication.

Currently, siRNA import into cells requires vectors such as plasmids and recombinant lentiviruses that express shRNAs efficiently and stably. As recombinant lentiviruses have a higher adaptability and replication ability in host cells, and as their genes can be integrated into cellular genomes, the silencing effect of lentiviruses is more efficient than plasmids [30]. However, this kind of gene integration may induce host genome mutations and cause cellular injury. In addition, excessive expression of shRNAs competitively inhibits cell endogenous miRNAs and causes cytotoxicity [31]. Therefore, RNAi mediated by a plasmid is currently safer for high gene silencing efficiency. However, off-target effects from using plasmid-based siRNA are still a shortfall [32]. To solve this problem, designing improved siRNA sequences, as well as developing novel and progressive vectors, is necessary.

In conclusion, plasmids expressing three shRNAs targeting different sites of the *N* gene of PEDV were constructed and transfected into IEC cells. Infection of PEDV CV777 (G1) or LNCT2 (G2) occurred post transfection. Detection by Western blot and viral titer assays was used to measure levels of viral replication in cells. The results demonstrated that PEDV G1 and G2 strains were susceptible to RNAi pathways targeting the *N* gene. All shRNA tests led to the silencing of *N* gene expression and suppressed proliferation of PEDV G1 and G2 strains in IEC cells. In sum, our data showed the potential for the shRNA expression vectors to precisely and effectively interfere with the replication of PEDV G1 and G2 strains in vitro. To our knowledge, this is the first report of the inhibition of PEDV G1 and G2 strain infection with shRNAs in IEC cells. Therefore, this report enriches the antiviral spectrum of RNAi treatment. Although IEC cells are a non-transformed porcine intestinal epithelial cell line, these cells lack the complexity of the cell types found in the architecture of the intestinal epithelium and, thus, do not satisfactorily mimic the natural infection process. This method merits further investigation in animal studies to define its therapeutic potential. Determining if the technology could be used in vivo for anti-PEDV therapy is still under investigation.

**Author Contributions:** Conceptualization, L.F.; data curation, D.S., X.W., and H.S.; funding acquisition, L.F.; METHODOLOGY, D.S., X.W., and H.S.; project administration, L.F.; resources, J.C.; software, X.Z., J.L., J.Z. (Jialin Zhang), Z.J. (Zhaoyang Ji), and Z.J. (Zhaoyang Jing); supervision, L.F.; writing—original draft, D.S., X.W., H.S., and L.F.; writing—review and editing, J.Z. (Jiyu Zhang) and Y.H.

**Funding:** This work was supported by grants from the National Natural Science Foundation of China (31602072 and 31572541), the China Postdoctoral Science Foundation (2017M610136), and the Natural Science Foundation of Heilongjiang Province of China (C2017079 and C2018066).

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **A Single Dose of Dendrimer B2T Peptide Vaccine Partially Protects Pigs against Foot-and-Mouth Disease Virus Infection**

**Rodrigo Cañas-Arranz 1, Mar Forner 2, Sira Defaus 2, Patricia de León 1, María J. Bustos 1, Elisa Torres 1, Francisco Sobrino 1,\*, David Andreu 2,\* and Esther Blanco 3,\***


Received: 16 December 2019; Accepted: 8 January 2020; Published: 10 January 2020

**Abstract:** Foot-and-mouth disease virus (FMDV) causes a highly contagious disease of cloven-hoofed animals whose control relies on efficient vaccination. We have reported that dendrimer peptide B2T, with two copies of FMDV B-cell epitope VP1 (136–154) linked through maleimide units to T-cell epitope 3A (21–35)], elicits potent B- and T-cell specific responses and confers solid protection in pigs to type-O FMDV challenge after two doses of peptide. Herein we now show that B2T evokes specific protective immune responses after administration of a single dose of either 2 or 0.5 mg of peptide. High titers of ELISA and neutralizing antibodies against FMDV were detectable at day 15 post-immunization. Likewise, activated T cells and induced IFN-γ response to in vitro recall with FMDV peptides were also detected by the same day. Further, in 70% of B2T-vaccinated pigs, full protection—no clinical signs of disease—was observed upon virus challenge at day 25 post-immunization. These results strengthen the potential of B2T as a safe, cost-effective candidate vaccine conferring adequate protection against FMDV with a single dose. The finding is particularly relevant to emergency scenarios permitting only a single shot immunization.

**Keywords:** FMDV; peptide vaccine; single dose; amount; pig

#### **1. Introduction**

Vaccination remains the most effective approach to prevent human and animal diseases [1]. In animal health, development of safe, cost-effective and marker vaccines, capable of telling infected from vaccinated animals (DIVA), remains a challenge for many diseases, particularly viral ones [2,3]. Conventional vaccines based on inactivated or attenuated viruses entail risks such as accidental escape or incomplete inactivation of infectious viruses, as well as possible reversion of attenuated into virulent forms. Subunit or epitopic vaccines represent an alternative that solves most such problems by excluding the infectious agent and allowing targeting to well characterized viral epitopes relevant for protection [4–6].

Foot-and-mouth disease virus (FMDV) is the prototype member of the *Aphthovirus* genus within the *Picornaviridae* family [7] and the etiological agent of FMD, a highly transmissible infection of pigs and other cloven-hoofed animals, with huge economic impact worldwide [8,9]. FMD underscores paradigmatically the challenge of finding alternative strategies to the classic vaccines still used to prevent this highly contagious disease [10]. The massive amplification and shedding of FMDV in

infected pigs turns this species to a key epidemiological factor for the spread of the virus over the course of outbreaks in many regions of the world [11]. In addition, the growing numbers of domestic pigs worldwide, particularly in Asian countries, make the development of pig-suited FMD vaccines a strategic task.

Conventional FMDV vaccines based on chemically inactivated virus have allowed FMD control and eradication in some countries, although their manufacturing process—not upgraded over recent decades—poses significant biosafety concerns that have been related to occasional escape episodes of diverse consequence [12–14]. This risk plus other limitations, such as the need for a strict cold chain to preserve stability, and the use of updated vaccine strains, because of the high potential antigenic diversity of the virus, underlie the adoption of non-vaccination policies in FMDV-free countries, a controversial and by no means risk-free practice, as borne out by not infrequent outbreaks in those locations. In crisis scenarios of this kind [15], vaccines incorporating outbreak-relevant epitopes, eliciting protective responses and generated as a quick response to the epidemic, can become an invaluable emergency resource for FMD containment [16]. Among such emergency vaccines, those based on synthetic peptides [6] are particularly appealing because of their (i) total lack of biological hazard; (ii) possibility of displaying various epitopes on a single platform; (iii) DIVA compliance; (iv) efficient synthetic production and characterization as pharmaceuticals, and (v) no cold-chain required; easy transport and storage [17].

The main B-cell antigenic site in FMDV, located at the GH loop of capsid protein VP1 (residues ca. 140–160), is structurally continuous [18,19]. Linear peptides reproducing this loop, either alone or in combination with T-cell FMDV epitopes, have been shown to confer limited protection in natural hosts [20–23].

A substantial enhancement in immunogenicity can be achieved by multiple display of B- and/or T-cell epitopes on a single molecular scaffold [17] inspired on the multiple antigenic peptide (MAP) platform of Tam [24]. In an initial realization in this regard, a peptide spanning residues 21–35 of FMDV protein 3A [thereafter T3A], which delimit an immunodominant T-cell epitope in domestic pigs [25], was N-terminally elongated into a Lys tree to which four copies of a B-cell epitope (residues 140–158 of VP1; containing the RGD motif that mediates binding to integrins, the cell receptors) were covalently linked in a dendrimeric (branched) fashion. The sequence of the B-cell epitope corresponded to that of the epidemiologically relevant O/UKG/11/01 isolate, belonging to serotype O the most prevalent worldwide [26]. This multivalent construct (named B4T) elicited high titers of FMDV-neutralizing antibodies, activated specific T cells, and fully protected pigs against FMDV challenge [27]. Interestingly, a simpler version (i.e., two B-cell epitope branches) of the peptide vaccine candidate, termed B2T, also elicited potent specific responses and conferred solid protection in pigs to challenge [28] (Table 1). In both B4T and B2T trials, animals were immunized with two 2-mg doses of peptide (3 weeks apart from each other) before challenge.


As mentioned above, emergency FMD vaccines, eliciting protective responses upon a single shot, are particularly valuable in containing uncontrolled FMDV outbreaks [14]. The modular nature of B2T affords considerable versatility in the sequences that can be integrated into the constructions, either as part of a single molecule or as mixtures of different molecules. Thus, incorporation of VP1 GH-loop sequences from different FMDV isolates can modulate/enhance the protective spectrum of the candidates. This is particularly relevant because of the high FMDV antigenic diversity reflected in seven serotypes and many variants within each of them, which makes matching of vaccine strains with circulating virus a critical issue for vaccine efficacy [11,15]. On the other hand, the T3A sequence mentioned above is highly conserved among FMDV serotypes and therefore can evoke heterologous responses, again contributing to broaden the protective response conferred by B2T. In addition to versatile, readily adaptable responses to new virus threats, emergency vaccines require fast and cost-effective manufacturing programs, which are ideally met by the expediency and flexibility inherent to chemical synthesis production.

Herein we have explored the possibility of: (i) Eliciting protective responses in pigs upon administration of a single B2T dose, and (ii) reducing the amount of antigen required to elicit protective responses. Thus, we report results with both the 2 mg dose of previous experiments [28] and with a reduced 0.5 mg inoculum. Remarkably, the latter dose elicits a rapid immune response involving high titers of FMDV neutralizing antibodies and specific IFN-γ secreting T cells at 15 days post-immunization (dpi). In addition, solid protection is observed in 80% of pigs vaccinated with 0.5 mg B2T, with no clinical signs upon viral challenge at day 25 dpi. Taken together, our results highlight the value of B2T as candidate FMD vaccine for pigs in emergency scenarios.

#### **2. Results**

#### *2.1. A Single B2T Dose Elicits Rapid Humoral Specific Responses Including FMDV Neutralizing Antibodies*

Domestic pigs, in two different groups of five animals each, were immunized once with 2 or 0.5 mg of B2T (pigs 1 to 5 and 6 to 10, respectively), and two additional non-immunized animals were kept as controls (pigs 11 and 12 inoculated with PBS). Total FMDV-specific IgG antibodies were determined by ELISA at 0, 15, and 21 dpi. Both 2 and 0.5 mg doses elicited consistent and comparable IgG titers (log10) at 15 (4.3 ± 0.4 vs. 3.7 ± 0.3) and 21 (4.2 ± 0.2 vs. 4.3 ± 0.3) dpi (Figure 1A). Upon FMDV challenge (day 25) these titers were not boosted up, remaining similar in both B2T-immunized groups (4.3 ± 0.1 vs. 4.6 ± 0.5). Non-immunized control pig 11 that survived 10 days post-challenge (dpc) showed anti-FMDV titers >2log10 units lower than those of the immunized and challenged groups (Figure 1A).

Regarding induction of FMDV neutralizing antibodies, as observed with the ELISA results, no major differences were noticed between virus neutralization titers (VNT) in the groups immunized with either 2 or 0.5 mg B2T at day 15 (1.5 ± 0.2 vs. 1.4 ± 0.4) or day 21 pi (1.9 ± 0.4 vs. 2 ± 0.5) (Figure 1B). On the other hand, and in contrast to ELISA antibody titers, post-challenge VNT increased at 10 dpc (35 dpi) in both groups (3.1 ± 0.3 vs. 3.2 ± 0.4). As expected, no neutralizing antibodies were detected in the two control animals before challenge. The non-vaccinated animal that survived the challenge (pig 11), showed VNT (3.1) similar to those in immunized groups (Figure 1B), showing that neutralizing antibody levels in B2T-immunized/challenged pigs—including those found to be protected—were as high as those in infected and recovered animals.

**Figure 1.** B- and T-cell responses in pigs immunized with a single dose and different amounts of B2T. Time course of the specific antibody responses in sera collected on the indicated days pi. (**A**) Total anti peptide B IgG titers analyzed by ELISA. Each point depicts mean antibody titers (calculated as described in Methods) ± SD for each group of pigs (*n* = 5). (**B**) Virus neutralization titers, VNT, expressed as the reciprocal log10 of the last serum dilution that neutralized 100 TCID50 of FMDV isolate O/UKG 11/01. Each symbol represents the value for an individual pig (numbering is included). Horizontal lines indicate the geometric mean for each animal group. In no case individual spontaneous reactivity was observed in the titers determined at day 0. Dotted lines denotes the assay detection limit. In (**A**,**B**) arrows point FMDV challenge (day 25 pi). (**C**) Specific T-cell responses measured by an ex vivo IFN-γ ELISPOT at days 15 pi. The frequency of FMDV-specific IFN-γ secreting cells was determined as detailed in Methods. Horizontal bars represent the mean frequencies of IFN-γ release spots of triplicates of peripheral blood mononuclear cells (PBMCs) from pigs stimulated in vitro with B2T (circles) or T (T3A) (squares) peptides (pig numbering is included).

#### *2.2. B2T Elicits Early FMDV-Specific IFN-*γ *Responses*

Specific T cell responses elicited by B2T at 15 dpi were determined by ELISPOT analysis of IFNγ-expressing PBMCs. High frequencies of spot-forming cells were found at day 15 in pigs immunized with either dose of B2T in response to in vitro recall with homologous peptide (284.8 ± 321.1 vs. 645.6 ± 467.9). On average, pigs in the 0.5 mg group showed higher frequencies of IFNγ-expressing

PBMCs than those in the 2 mg group (Figure 1C). All responses were specific, as no peptide-driven IFN-γ-producing cells were detected in the two non-immunized pigs. PBMC stimulation with T3A peptide paralleled those observed with B2T (241.3 ± 365.8 vs. 614.4 ± 403.1) (Figure 1C), supporting the recognition of T3A as a T-cell epitope.

#### *2.3. A Single Dose of B2T Peptide Confers Clinical Protection against FMDV Challenge*

At 25 dpi, pigs in all three groups were challenged with FMDV. Animals were examined daily for clinical signs (see Methods) and considered protected when lesions were not observed or appeared only at the inoculation site [29]. As expected, PBS-inoculated control pigs 11 and 12 showed full FMD signs upon challenge, developing vesicular lesions on all four feet at 4 dpc and on the snout at 7 and 5 dpc, respectively (Table 2). The acute FMDV infection caused myocarditis, leading to heart failure and sudden death, of pig 12 at 5 dpc. In contrast, only 3 out of 10 peptide-immunized animals developed lesions outside the inoculation site: two pigs (3 and 5) in the 2 mg group and one (pig 10) in the 0.5 mg group. The remaining immunized animals did not develop any clinical signs, the cumulative lesion score of immunized groups being significantly lower than that of the non-immunized one (Table 2). In general, a correlation between lower body temperature and protection was observed, being the protected animals immunized with 0.5 mg B2T those that did not develop fever.


**Table 2.** Evidences for protection in immunized pigs.

<sup>a</sup> VNT and IFNγ spots/106 PBMCs determined at day 21 and 15 post-immunization, respectively. <sup>b</sup> Temperature ( ◦C) and (in parenthesis) day pi when maximum temperature registered. No fever: ≤39.0–39.5 ◦C. <sup>c</sup> Animals were monitored up to 10 days pc for lesions. Lesion score (maximum value of 7): 1 point/vesicle in foot (up to 4 points); 1 point/mouth, tongue or snout lesion; 1 point/>2 lesions of diameter ≥10 mm. In parenthesis, day pc when lesion(s) was first observed. <sup>d</sup> Animals were considered fully protected (++) if lesion score ≤ 1; partially protected (mild/delayed disease) (+) if lesion score ≤ 2, or non-protected (−) if lesion score <sup>&</sup>gt; 2. <sup>e</sup> Detection of FMDV RNA in serum samples. ND: RNA not detected. The amount of RNA in positive animals are expressed as viral RNA copies (VRC)/mL serum. Detection limit: 5 × 103 VRC/mL serum. The values from RNA positive animals are presented as VRC/mL serum. The day(s) post-challenge when RNA was detected is shown in brackets. <sup>f</sup> Animal died on day 5 pc.

Detection of FMDV RNA in serum samples from challenged pigs by RT-qPCR showed the presence of viral RNA in non-immunized control pigs 11 (days 3 and 5 pc) and 12 (day 3 pc), as well as in the two immunized but non-protected pigs 10 (days 3 and 5 pc) and 3 (only at day 5 pc), which showed the higher lesion score at day 7 pc (Table 2). Hence, virus detection in sera associated with the severity of the lesions of challenged pigs. Further work is necessary to assess the potential contribution of viruses circulating in non-protected animals, and to a lesser extent in protected animals, to the spread of the disease under field conditions.

In conclusion, immunization with a single 2 mg dose and, remarkably, even more with a 0.5 mg dose of B2T dendrimer afforded substantial protection against conventional FMDV challenge.

#### **3. Discussion**

Protection after a single inoculation is a must for effective FMD vaccines, as it reduces both the cost of the vaccine as well as the logistics and labor expenses associated with the double immunization schedules. The application of a single dose vaccination program to the pig population can favor disease eradication, considering that pigs serve as amplifiers of FMDV and therefore a rapid control of the infection in swine is essential for disease control [11]. This is particularly relevant in settings such as diseases outbreaks occurring in areas where FMD is not enzootic and livestock remains unvaccinated [30]. To our knowledge, with the exception of adenovirus-vectored vaccines expressing FMDV empty capsids [31], few studies exist on the capability of FMD subunit vaccines alternative to classical ones to confer protection in relevant hosts after a single administration [32]. In this context, our optimization of the minimal amount of B2T dendrimer still conferring protection is pertinent, both for understanding the response mechanisms to B2T and for scaling down production costs, a relevant issue for FMD vaccine development.

We had previously shown that two 2 mg doses of B2T conferred solid protection to FMDV challenge in swine, and that protection correlated with the induction of strong B- and T-cell specific responses [28]. Here, we explore the feasibility of B2T as a protective, single dose subunit vaccine alternative to conventional FMD formulations. To this end, the original 2 mg dose and a four-fold lower one (0.5 mg) have been compared.

The challenge protocol followed in our experiment was as recommended by the OIE manual, for evaluation of protection against podal generalization (PPG) test, except that virus was inoculated 3 days earlier than indicated [33]. Under such conditions, full protection, considered as the absence of lesions at points other than the inoculation site, was observed in 70% of pigs vaccinated with B2T; 3 out of 5 in the 2 mg dose group, and 4 out of 5 in the 0.5 mg group. This level of protection correlated with an average increase in virus neutralizing (VN) antibodies that was not observed for B-specific IgG antibodies detected by ELISA (Figure 1), suggesting the recall upon viral infection of a subset of FMDV-specific memory B cells, as well as of memory T cells capable to promote B-cell maturation. This increment of VN antibodies in protected animals, matching that of a non-immunized pig, might reflect virus multiplication to a limited extent not resulting in detectable viremia (Table 2). Similar levels of VN antibodies boost upon virus challenge have been reported in pigs and cattle immunized with live-vectored vaccines or FMDV-like particles [34,35].

Non-protected (pigs 3 and 10) and partially protected (pig 5) animals showed lower levels of VN antibodies at the day of challenge, albeit animals with lower VNT yet protected were also observed (Figure 1B); the lack of correlation in a fraction of protected animals immunized with different FMDV vaccines has been described [21,36]. Experiments aimed at assessing the amount of antibodies that may target the FMDV receptor [37] as well as at examining cross-neutralization among other type O FMDVs will be foreseen to further characterize the antibody response evoked by the dendrimer peptide vaccines.

In addition, and supporting T3A recognition as a T-cell epitope, specific IFN-γ releasing activated T cells were detected rather early, at 15 dpi, in 7 of the 10 immunized pigs. The average frequencies of IFN-γ releasing cells in PBMCs from pigs in the 0.5 mg dose group were higher than those in the 2 mg group (Figure 1C and Table 2), possibly reflecting differences in the in vitro dose-effect of peptide stimulation and/or on the MHC allele composition of the animals analyzed. High doses of antigen might also facilitate its capture by APC/DC subsets with suboptimal costimulatory capacity that favor the development of regulatory T cells, which could limit the activation of B cells and subsequent antibody production as well as the development of IFN-γ producing cells [38]. In any case, additional studies of the effector mechanisms triggered by the dendrimers will be necessary to understand the effect on the protective responses of different peptide vaccine doses. One animal that was low responder at 15 dpi (pig 7), turned out to be protected upon challenge and pig 5, also low responder, was partially protected. Among responder animals, no clear correlation between IFN-γ releasing cell frequency at 15 dpi and protection to FMDV challenge was observed, i.e., non-protected pigs 3 and 10

showed high responses. Thus, as previously observed [21], VNT and T-cell stimulation do not fully correlate with the protection conferred by FMD peptide vaccines.

A potential problem of epitopic vaccines such us the dendrimeric peptides used in this study is the selection in non-protected animals of escape viruses capable to avoid antibody neutralization [21] and/or T cell recognition. In our experiment, we could not rule out that possibility since the limited number and size of the vesicles developed by non-protected animals did not allow virus recovery for sequencing analyses.

Our evidence of the level of full protection conferred by a single B2T dose upon challenge at 25 dpi is particularly remarkable with regard to the 0.5 mg dose group (80% of the pigs), as it opens the way to significant savings in manufacturing costs. Incidentally, on a related note, we have found that long-lasting (up to 19/20 weeks post-boost) reduced susceptibility to FMDV infection can be attained with two B2T doses and, remarkably, a similarly lasting protective response with a single dose of B2T (Cañas-Arranz et al., submitted).

Attempts to use B2T and B4T dendrimer peptides as vaccines in cattle showed a trend toward a reduced capacity to confer protection relative to swine, including the need for a third immunization to elicit protective levels on neutralizing antibodies [39,40]. Failure in conferring protection in cattle has been described for linear FMDV peptide vaccines containing a heterologous T cell epitope [41]. Further work on the requirements of the interactions between neutralizing epitopes and B cells, as well as the identification of new, effective T helper epitopes frequently recognized by individuals will contribute to the improvement of the effectiveness of this kind of vaccines for cattle and, eventually, other host species.

In summary, our findings portray the B2T peptide as a safe, potentially cost-effective candidate to be included in FMD vaccine formulations conferring single-shot protection in pigs. Formulation of dendrimers with alternative adjuvants as well as the incorporation of peptides corresponding to other T cell epitopes previously identified in pigs [42,43] can contribute to improve their immunogenicity in swine.

#### **4. Methods**

#### *4.1. Peptides*

Peptides identified as B- and T-cell epitopes of FMDV O/UKG/11/01 (Table 1) were assembled by Fmoc-solid phase synthesis, purified by reverse-phase liquid chromatography, and characterized by mass spectrometry. Dendrimeric B2T construct was conjugated in solution using two previously synthesized precursors: (i) B epitope (VP1, residues 136–154) plus a C-terminal Cys residue (free thiol form) and (ii) T epitope (3A, residues 21–35) elongated at the N-terminal with two Lys residues followed by an additional Lys branching derivatized as two maleimide groups. The B2T peptide was efficiently obtained after the thiol-maleimide ligation at pH = 6, RP-HPLC purification and MS characterization [27,28,44].

#### *4.2. Virus*

A virus stock derived from FMDV isolate O/UKG/11/01 [45] by two amplifications in swine kidney cells (IB-RS-2 cells) was used. The resulting virus maintained the consensus sequences at the capsid region [46].

#### *4.3. Animals and Experimental Design*

The immune response to B2T dendrimer peptide was assessed in ten 9-weeks-old white cross-bred Landrace female pigs (Agropardal SA breed), free of antibodies to FMDV. The study was approved (CBS2014/015 and CEEA2014/018) by the INIA Committees on Ethics of Animal Experiments and Biosafety, and by the National Committee on Ethics and Animal Welfare (PROEX 218/14). Pigs were randomly assigned to two groups of five animals each and immunized once (days 0) by intramuscular injection with 2 mL of Montanide ISA 50V2 emulsion (Seppic, Paris, France) containing 2 or 0.5 mg of B2T peptide. Two additional non-vaccinated pigs were kept as infection controls (#11 and 12). Animals were housed in separate units of the high-containment facility and challenged at day 25 with 1.6 <sup>×</sup> <sup>10</sup><sup>4</sup> plaque forming units (pfu) of FMDV O/UKG/11/01, by inoculation at two sites of both main claws of the left-hindfoot pad (0.1 mL/site). Animals were monitored for clinical signs of disease during the 10 days (Table 2), and then euthanized at day 35.

#### *4.4. Viral RNA Detection after Challenge*

Serum samples were examined for the presence of viral RNA by real time RT-qPCR. Briefly, the cDNA obtained in a RT reaction using primer A (5 -CACACGGCGTTCACCCA(A/T)CGC-3 ) [47] was amplified by qPCR using the "Light Cycler RNA Master SYBR Green I" kit (Roche, Basel, Switzerland) and LightCycler equipment following the instructions of the manufacturer. The amplicon synthesized spanned a conserved region of 290 pb length in the 3D protein coding sequence, and was amplified using primers A (5 -CACACGGCGTTCACCCA(A/T)CGC-3 ) and B (5 -GACAAAGGTTTTGTTCTTGGTC-3 ). The values for the quantification of the samples were obtained from a standard curve from a RNA transcript derived from the infectious clone pMT-28 codifying the genomic RNA from FMDV C-S8c1 [48].

#### *4.5. Virus Neutralization Test (VNT)*

Serial two-fold dilutions of sera were incubated with 100 infection units—50% tissue culture infective doses (TCID50) of FMDV O/UKG 11/01, for 1 h at 37 ◦C. End-point titers were calculated as the reciprocal of the serum dilution that neutralized FMDV infection in 50% of the wells [28].

#### *4.6. Detection of Specific Anti-FMDV Antibodies by ELISA*

Total anti-FMDV antibodies were determined by ELISA. Briefly, 96-well plates (Nunc, Rochester, NY, USA) were coated with peptide B (1 μg) overnight at 4 ◦C. Duplicate three-fold dilution series of each serum sample were prepared in 50 μL, starting at 1/100. Pre-immune sera from peptide-immunized pigs and sera from non-immunized animals were used as negative controls. Specific antibodies were detected with HRP-conjugated protein A (Thermo Fisher, Rochester, NY, USA), diluted 1/4000. Color development was obtained after addition of 100 μL/well of TMB (Sigma Aldrich, St. Louis, MO, USA) and stopped by an equal volume of 1M H2SO4. Plates were read at 450 nm, titers expressed as the reciprocal of the last dilution giving the absorbance recorded in the control wells (serum at day 0) plus 2 SD.

#### *4.7. PBMC Isolation and IFN-*γ *Detection by ELISPOT*

Porcine peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation using Histopaque-1077 (Sigma, St. Louis, MO, USA) and cryopreserved prior to assay. Vials were defrosted and resuspended in complete RPMI 1640 and incubated overnight at 37 ◦C, 5% CO2. Cell counting and viability were tested by Trypan blue staining. For the IFN-γ ELISPOT assay 2.5 <sup>×</sup> <sup>10</sup><sup>5</sup> PBMCs were shed in triplicate wells of Immobilon-P plates (Merck Millipore, Burlington, MA, USA) coated with 5 μg/mL of anti-pig IFN-γ antibody (clone P2G10, Becton Dickinson, Franklin Lakes, NJ, USA). For in vitro antigen recall, PBMCs were stimulated with 25 μg/mL of the peptide used for pig immunization [49]. As positive control, PBMCs were incubated with 5 μg/mL of phytohaemagglutinin (Sigma) using cells incubated without antigen as negative control. After 48 h at 37 ◦C, 5% CO2, plates were washed and incubated with 2 μg/mL of biotinylated anti-mouse IFN-γ antibody (clone P2C11, BD) and HRP-streptavidin (BD). Antibody was visualized with 3-amino-9-ethyl carbazole (BD). The frequency of peptide-specific T cells in was expressed as the mean number of spot-forming cells/106 PBMCs, with background values (number of spots in negative control wells) subtracted from the respective counts of stimulated cells.

#### *4.8. Statistical Analyses*

Differences among the peptide-immunized groups in FMDV-antibody titers and number of IFN-γ producing cells were analyzed by two-way ANOVA, followed by Tukey's post-hoc comparisons tests. Values are cited in the text as means ± SD. All *p* values are two sided, and *p* values < 0.05 were considered significant. Statistical analyses were conducted using GraphPad Prism Software 5.0 (Graphpad Software, San Diego, CA, USA).

**Author Contributions:** Conception, design and funding of the work (D.A., E.B. and F.S.). Data Acquisition and analysis (R.C.-A., M.F., S.D., P.d.L., M.J.B. and E.T.). Interpretation of data for the work (R.C.-A., M.F., S.D., P.d.L., D.A., F.S. and E.B.). Drafting the work (R.C.-A., M.F., S.D., P.d.L., F.S. and E.B.). Revising the data critically for important intellectual content (D.A., E.B. and F.S.). All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Spanish Ministry of Science, Innovation and Universities (grant AGL2017-89097-C2 to FS and DA; AGL2016-76445-R to EB), Comunidad de Madrid co-financed with ECFEDER funds (S2013/ABI-2906-PLATESA and P2018/BAA-4370 to FS and EB) and Generalitat de Catalunya (2009SGR492 to DA). Work at Centro de Biología Molecular "Severo Ochoa" and at UPF was supported by Fundación Ramón Areces and by the Maria de Maeztu Program of the Spanish Ministry of Science, Innovation and Universities, respectively. R.C.-A. and M.F. were holders of a PhD fellowship from the Spanish Ministry of Science, Innovation and University (FPI programme). Article processing free of charge.

**Acknowledgments:** We thank P. Pallarés for her advice and support.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
