**MVA Vectored Vaccines Encoding Rift Valley Fever Virus Glycoproteins Protect Mice against Lethal Challenge in the Absence of Neutralizing Antibody Responses**

## **Elena López-Gil** †**, Sandra Moreno, Javier Ortego, Belén Borrego, Gema Lorenzo and Alejandro Brun \***

Animal Health Research Centre (CISA), National Institute for Agriculture and Food Research and Technology (INIA), Valdeolmos, 28130 Madrid, Spain; melenalopezgil@gmail.com (E.L.-G.); sandramorenofdez@gmail.com (S.M.); ortego@inia.es (J.O.); borrego@inia.es (B.B.); lorenzo.gema@inia.es (G.L.)

**\*** Correspondence: brun@inia.es; Tel.: +34-916-202-300

† Present address: Diagnostic and Therapeutic Applications Laboratory (Diater S. A). Avda. Gregorio Peces Barba, 2. Leganés Technology Park, Leganés, 28918 Madrid, Spain.

Received: 10 January 2020; Accepted: 10 February 2020; Published: 12 February 2020

**Abstract:** In vitro neutralizing antibodies have been often correlated with protection against Rift Valley fever virus (RVFV) infection. We have reported previously that a single inoculation of sucrose-purified modified vaccinia Ankara (MVA) encoding RVFV glycoproteins (rMVAGnGc) was sufficient to induce a protective immune response in mice after a lethal RVFV challenge. Protection was related to the presence of glycoprotein specific CD8+ cells, with a low-level detection of in vitro neutralizing antibodies. In this work we extended those observations aimed to explore the role of humoral responses after MVA vaccination and to study the contribution of each glycoprotein antigen to the protective efficacy. Thus, we tested the efficacy and immune responses in BALB/c mice of recombinant MVA viruses expressing either glycoprotein Gn (rMVAGn) or Gc (rMVAGc). In the absence of serum neutralizing antibodies, our data strongly suggest that protection of vaccinated mice upon the RVFV challenge can be achieved by the activation of cellular responses mainly directed against Gc epitopes. The involvement of cellular immunity was stressed by the fact that protection of mice was strain dependent. Furthermore, our data suggest that the rMVA based single dose vaccination elicits suboptimal humoral immune responses against Gn antigen since disease in mice was exacerbated upon virus challenge in the presence of rMVAGnGc or rMVAGn immune serum. Thus, Gc-specific cellular immunity could be an important component in the protection after the challenge observed in BALB/c mice, contributing to the elimination of infected cells reducing morbidity and mortality and counteracting the deleterious effect of a subneutralizing antibody immune response.

**Keywords:** Rift Valley fever virus (RVFV); modified vaccinia Ankara (MVA); cellular response; neutralizing antibodies; Gn Gc glycoproteins; passive serum:virus transfer

#### **1. Introduction**

Rift Valley fever virus (RVFV), a mosquito-borne bunyavirus, is widely distributed in Sub-Saharan countries, Egypt and the Arabian Peninsula, causing disease in both humans and livestock [1]. RVFV is considered an emerging threat for non-endemic countries due to the movement of infected animals and/or translocation of infected mosquitoes [2,3]. The ample range of RVFV competent mosquito vectors present in many areas of the Mediterranean basin suggests that RVF outbreaks in non-endemic areas could potentially end-up in the establishment of enzootic infection cycles [4]. Should this happen it would cause serious concern for both public and animal health. It is therefore desirable to develop

control tools as well as enhance our knowledge about the immune mechanisms that correlate with the protection elicited by RVFV vaccines. The modified vaccinia Ankara (MVA) virus has been widely used as a carrier of vaccine antigens due to its safety and immunogenicity profile [5]. The MVA vector is a highly attenuated version of a vaccinia virus strain that has lost around 30 kb of sequence upon passage in primary avian cells (CEF) so that many host range and immune-modulatory genes are absent or nonfunctional [6]. This restricts the replication of the virus in most mammalian cells. Besides, it has been demonstrated that MVA itself is very immunogenic (induces humoral responses and is an excellent inducer of T-cell responses). It has been used in human preclinical and clinical trials [7] and more recently it has also been evaluated for different animal diseases, including zoonotic diseases [8]. In the case of zoonotic diseases the data obtained in animal field trials may also help to accelerate the development of corresponding human vaccines.

In order to control RVF it is generally accepted that the induction of neutralizing antibodies is an important correlate of protection [9]. Therefore, the Gn and Gc glycoproteins are the main vaccine antigen targets since both glycoproteins are membrane proteins forming spikes on the surface of the virions that can be accessible to neutralizing antibodies [10–12], precluding either internalization and/or nucleocapsid uncoating. Both glycoproteins are synthesized as a polyprotein precursor that becomes localized in Golgi membranes where they are glycosylated and cotranslationally processed by a yet unidentified cellular protease that releases one end of the membrane attachment sites, allowing both ectodomains to interact and the glycoproteins to acquire their final conformation on the surface of the virion [13,14]. The structure of GnGc architecture has been recently elucidated indicating that Gn shields Gc in the viral particles avoiding to expose the Gc fusion loop to antibodies [15] and offers an hypothesis for the neutralizing mechanisms of protective antibodies [16]. Our previous works using recombinant MVA (rMVA) as vector vaccines expressing both glycoprotein antigens (rMVAGnGc) showed an intriguingly low level of in vitro neutralizing antibody induction upon single dose administration, particularly when compared to other similar rMVA vaccines encoding RNA virus glycoprotein antigens [17]. In our system, the lack of a humoral protective antiglycoprotein response was observed in both mouse models and disease natural hosts [18]. Particularly, for sheep no neutralizing antibody induction was demonstrated either after one or two serial rMVAGnGc vaccine doses, questioning the efficacy of this vaccine in this species [18]. The lack of neutralizing antibody induction could be related to the vector platform used in those experiments (MVA) since the same coding glycoproteins expressed by means of an adenovirus vector induced a potent set of neutralizing antibodies [19,20].

In this work we show that rMVA vaccines expressing independently versions of the glycoprotein Gc or Gn were able to confer substantial protection in mice, albeit inducing no detectable in vitro neutralizing antibody responses. Rather, the protection observed upon challenge was related to a strong T-cell response, which appeared more prominent against Gc epitopes. Intriguingly, the immune response elicited by the rMVA vaccine encoding both glycoproteins or glycoprotein Gn endowed the serum with the capacity to exacerbate disease when a serum:virus mixture was passively transferred to naive mice, as shown in experiments using either mouse or sheep immune sera obtained from rMVA vaccinated animals. It therefore appears that evoking a strong cellular response counteracts the failure in inducing protective humoral responses when an MVA vaccine is used. Conversely, the absence of an effective cell-mediated immune response may lead to disease exacerbation when subneutralizing antibody responses are elicited. These results warrant the optimization of our rMVA vaccines towards the induction of optimal humoral responses.

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

#### *2.1. Generation of Recombinant rMVA Encoding RVFV Gn and Gc Glycoproteins*

The rMVA-GnGc recombinant virus generated by homologous recombination of wild type MVA DNA and a plasmid construct encoding RVFV-MP12 glycoproteins (GB accession: DQ380208.1) into

the TK locus of MVA was described previously [17]. For the generation of rMVA-Gn and rMVA-Gc, the recombination (shuttle) plasmid encoding an RVFV-MP12 GnGc tagged sequence (plasmid #1389 produced at the Viral Vector Core Facility (VVCF), Jenner Institute, Oxford) was used as a template of an inverse PCR reaction using specific 5 phosphorylated primers (Supplementary file, Table S1). After religation of the PCR fragments, two new shuttle vectors were generated, in which the Gn ectodomain (encoding amino acids Met131 to His580 of the translated polyprotein precursor) and a Gc including the C-terminal transmembrane-cytosolic tail (amino acids Cys690 to Ser1197) were placed under the control of the vaccinia p7.5 early/late promoter. The N-terminus of each recombinant polypeptide contained an in-frame fusion of the human tissue plasminogen activator leader sequence (tPA), known to enhance transgene expression and immunogenicity [21]. The C-terminus of each protein contained an H-2K<sup>d</sup> restricted CD8+ T cell epitope from *Plasmodium berghei* circumsporozoite protein (pb9) and an antiV5 monoclonal antibody recognition sequence. The plasmid for MVA construction also includes GFP as a reporter gene under the control of the vaccinia p11 late promoter. Both shuttle vectors were transfected into DF-1 cells (ATCC-CRL-12203) using lipofectamine 2000 (Thermo Fisher Scientific, Waltham, MA, USA), then infected with parental MVA and homologous recombination allowed the insertion of either Gn ectodomain (eGn) or Gc ORFs and the GFP marker gene at the TK locus of the MVA. Three consecutive rounds of green plaque purification were performed in order to obtain a pure preparation of each recombinant virus. The recombinant viruses (named rMVAGn and rMVAGc) were then further expanded in DF-1 cells. Semipurified, concentrated, virus preparations were obtained upon ultracentrifugation of infected cell extracts in a 36% sucrose cushion. The sucrose-purified virus fractions were titrated into DF-1 cells and stored at −80 ◦C until use.

#### *2.2. Western Blot Analysis*

Expression of recombinant RVFV glycoproteins was analyzed by western blots of infected cell lysates using either specific antiGn or Gc antibodies [22] or a monoclonal antibody against V5 peptide tag (Bio-Rad, Hercules, CA, USA)). BHK-21 cells (ATCC CCL-10) were infected with the different recombinant MVA viruses described above, at 5 pfu/cell or were mock infected. At 24 h post infection the cells were harvested, pelleted, washed in PBS-containing protease inhibitor cocktail (Sigma-Aldrich, San Luis, MO, USA), and lysed with cytoplasmic extraction buffer (10 mM HEPES pH 7.9, 10 mM KCl, 0.1 mM EDTA, and 0.3% NP−40). After a centrifugation step to release intact nuclei, extracts were mixed with an equal amount of 2X Laemmli buffer, including DTT as a reducing agent and proteins were resolved in 12% SDS-PAGE and blotted onto nitrocellulose membranes. After a blocking step with 5% low fat dry milk in PBS (blocking buffer), antiRVFV Gn monoclonal antibody 84a (1:3000 dilution), monoclonal antiV5 tag (1:5000), or a rabbit antiGc polyclonal antibody (1:5000) were applied to membranes in blocking buffer with 0.01% Tween-20 and incubated for 1 h at room temperature. Horseradish peroxidase conjugated antimouse or antirabbit antibodies (1:5000) were incubated to the membranes after three washing steps with PBS Tween-20 (PBST). The resulting immunocomplexes were detected by enhanced chemiluminescence (GE Healthcare, Little Chalfont, Buckinghamshire, UK) and X-ray film exposure.

#### *2.3. Indirect Immunofluorescence and Laser Confocal Microscopy*

Cells were grown in either multi-well 96 (MW96) plates or in glass coverslips (CS) and infected with the recombinant MVA viruses at a multiplicity of infection (MOI) of 1. 24 h after infection the cells were fixed and permeabilized with 100% ice-cold methanol (MW96) or fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton-X100 (CS). Fixed cells were blocked with 10% FBS in PBS (10% blocking solution) for 30 min at room temperature (rt). AntiV5tag mAb, glycoprotein specific antibodies or antibodies specific to ER and Golgi proteins calreticulin and human mannosidase II (Bio-Rad's AHP516 and AHP674 antibodies) were incubated for 1 h at rt in 2% blocking solution with 0.01 Tween-20. After three serial washing steps with PBST Alexa 488 conjugated antimouse, or Alexa-Fluor 594-conjugated antirabbit or antigoat mabs (Thermo) were incubated for 30 min at rt. Stained cells in MW96 were visualized using a Zeiss AX10 inverted fluorescence microscope (Zeiss Gmbh, Oberkochen, Germany). Stained CS preparations were mounted onto glass microscopy slides with or without DAPI staining (Thermo Fisher) and were visualized analyzed in a Zeiss LSM880 confocal laser microscope. Images were further processed using the Zen Zeiss software.

#### *2.4. Immunoprecipitation Analysis*

For immunoprecipitation, monolayers of Vero cells were infected with RVFV-MP12 virus in the presence of 300 μCi/mL of [S35]-Methionine-Cysteine solution (Grupo Taper S.L, Alcobendas, Madrid, Spain). After 24 h the cells were lysed in radioimmunoprecipitation assay buffer (150 mM NaCl, 1% NP-40, 0.1% SDS, and 50 mM Tris, pH 7.5) and 20 μl of pooled sera from mice vaccinated with different rMVA viruses were incubated for 1 h at room temperature in a rotary shaker. Paramagnetic protein-G beads (Thermo Fisher) were added and incubated for an additional hour. The immunocomplexes were washed three times with the RIPA buffer and then separated by 12% SDS-PAGE. Fixed gels were subjected to fluorographic enhancement using Amplify solution (GE). After drying the gels were exposed to X-ray film.

#### *2.5. Immunization, Sampling for Immunological Assays and RVFV Challenge*

Groups of 5–10 BALB/c mice eight to ten weeks-old (Envigo RMS, Barcelona, Spain) were immunized intraperitoneally with 107 pfu of sucrose-cushion purified rMVA in phosphate-buffered saline (PBS). One or two weeks postvaccination, blood samples were taken either for neutralization assays (serum) or IFN-γ ELISpot (PBLs), and splenocytes for ex-vivo IFNγ ELISPOT at 7 or 14 dpi (*n* = 4). The remaining mice (*n* = 5), together with additional groups of unvaccinated BALB/c mice and immunized with nonrecombinant MVA (MVA control, expressing only GFP), were all challenged intraperitoneally with 10<sup>3</sup> plaque-forming units (pfu) of the South African RVF virus strain 56/74 [23]. The immunization and challenge studies were also performed in a similar manner using 129SvEv mice (B&K Universal Group Ltd, Hull, UK). To monitor viremia, blood samples were taken at 72 h after RVFV infection, and tested for virus isolation on cell culture as described [24]. Briefly, blood dilutions were incubated with Vero cells cultures and examined for the viral cytophatic effect (cpe). After 96 h the extent of cytophatic effect was recorded and cells were fixed and stained with 2% Crystal Violet in 10% formaldehyde solution. The extent of cpe was quantified to estimate a tissue culture infective titer (TCID50). Serum samples collected at later times postinfection in surviving mice were analyzed for the presence of neutralizing antibodies as described below. Vaccine efficacy estimation was evaluated in terms of morbidity and mortality monitoring daily over three weeks. All surviving mice were culled after 21 days of follow-up. Procedures involving animals received institutional approval (INIA's ethics and biosafety Committee) as well as granted permits from regional veterinary authorities (Comunidad de Madrid PROEX 108/15).

#### *2.6. Assessment of RVFV Serum Neutralizing Antibodies*

Serum neutralizing antibody titers were measured in Vero cell monolayers by serial dilutions of serum mixed with an equal volume of medium containing MP-12 RVF virus strain and incubated for 1 h at 37 ◦C. After 72–96 h the cells were fixed and stained in a solution containing 10% formaldehyde and 2% crystal violet in PBS. The neutralization titer defined as the highest serum dilution at which cell lysis was reduced by 50% relative to cells incubated with RVF virus only. The assays were performed in triplicate and scored by an operator blinded to the vaccination regimen.

#### *2.7. Analysis of T-Cell Responses Against RVFV Glycoproteins*

Viral glycoprotein-specific T cells were measured by ex vivo IFNγ ELISPOT assay on splenocytes and pooled PBLs as described [17]. Gn, Gc-specific and nonspecific class-I restricted peptides were used for restimulation at a final concentration of 5 μg/mL in all assays for 18 h. IL-2, IL-6, IL-4, and IL-5 cytokine capture ELISAs (BD Pharmingen) were also performed using supernatants from peptide restimulated spleen cell cultures. Known concentrations of mouse IL-2, IL-6, IL-4, or IL-5 were used to generate a standard curve to correlate optical densities with cytokine concentration. The sensitivity limit of the assay was estimated in (61.25 pg/mL for IL-6, 250 pg/mL for IL-2, 3.75pg/mL for IL-4, and 37.5 pg/mL for IL-5).

#### *2.8. Passive Transfer of Antibodies*

The sera for passive transfer protection studies were generated by pooling sera from mice immunized with the different rMVA constructs expressing the same antigens. Serum pools were prepared from day 14 post immunization and analyzed by virus neutralization and immunoprecipitation assays. As a positive control, antiRVFV immune mouse serum was used while antihuman adenovirus 5 (AdHu5) pooled mouse serum (collected also at 14 days post immunization) was used as a negative control serum. For passive protection experiments each serum pool was ten-fold diluted in the virus inoculum used to challenge each group of 5 animals. 100 μL of each virus/serum mixture was injected intraperitoneally into adult female BALB/c mice. The virus challenge dose per mouse corresponded to 5 <sup>×</sup> 10<sup>3</sup> pfu. Animals were monitored for clinical signs and mortality during three weeks and were weighed daily to quantify the extent of morbidity after challenge. Additionally, serum from sheep immunized with the rMVAGnGc vaccine, rMVA or from mock vaccinated was also pooled and passively transferred to mice.

#### *2.9. Statistical Analysis*

The log rank (Mantel–Cox) test was used to check for differences in survival analysis following RVFV challenge. Individual ELISPOT values were determined by subtracting background values obtained after stimulation with media only. Statistical significance was calculated by one-way analysis of variance (ANOVA) transforming ELISPOT counts to log10 to limit the range of variation found among individual mice. All analyses were done using the GraphPad 6.0 software (San Diego, CA). Differences were considered significant when *p* value <0.05

#### **3. Results**

#### *3.1. Expression of Recombinant Gn and Gc Glycoproteins in rMVA Infected Cells*

Expression was assessed by western blot analysis (Figure 1A). Both glycoprotein sequences were tagged with the V5 epitope sequence to compare their relative expression levels. It was observed that the infection of cells with both recombinant MVAGn and rMVAGc rendered detectable expression levels for both glycoproteins. This ruled out the possibility of low level expression conditioning the immunity conferred by each vaccine. Expression of glycoproteins was also confirmed using antiGn or Gc specific antibodies [22]. The detecting signal was similar to that of RVFV MP12 infected BHK-21 cells. The size of Gn expressed by rMVAGn was in accordance with its theoretical mass (50.6 kDa) but slightly lower when expressed by rMVAGnGc. Gc expression was also in good agreement with the expected size and similar in size to the one expressed by RVFV infection, although a smaller truncated polypeptide was also evident using the antiGc or the antiV5 tag antibodies. Detection of the expressed antigens was performed also by immunofluorescence assay (IFA)of rMVA infected Vero cell monolayers with an antiV5 tag monoclonal antibody. The subcellular staining pattern of each glycoprotein was in good agreement with intracellular membrane trafficking as it has been described for both glycoproteins (Figure 1B). Gn expression was predominantly cytoplasmatic with no evident association with endoplasmic reticulum (ER). In contrast Gc interaction with ER structures was more obvious as shown by the colocalization with the ER marker calreticulin. No clear association of Gn or Gc was found with Golgi structures at least at the time point assayed (24 hpi), as evidenced by the lack of costaining with an antihuman mannosidase-II mAb.

**Figure 1.** Expression and subcellular localization of recombinant Gn and Gc upon modified vaccinia Ankara (MVA) infection. (**A**). Western blot of different MVA infected BHK-21 cell extracts probed with mAb 84a antiGn or a rabbit polyclonal serum antiGc. The antiV5 tag mAb was used to compare the Gn and Gc expression levels and to confirm the expression of the full-length antigen. As a positive control a RVFV-MP12 infected cell extract was used. Numbers indicate relative molecular mass in kilodaltons. (**B**). Confocal immunofluorescence images of MVA infected Vero cells. Expression of Gn or Gc was detected with anti V5 tag mAb (green). Intracellular membranes were labeled with either antihuman mannosidase-II (Golgi) or anticalreticulin (ER) mAbs (red fluorescence) as indicated. Nuclei were labeled with DAPI stain (blue). All panels correspond to merged fluorescence images. Colocalization of Gc and ER membranes is evidenced by yellow-orange fluorescence.

#### *3.2. E*ffi*cacy Assessment of MVA Vaccines in Mice*

The protective ability of a single dose of our rMVA vaccines was tested in BALB/c mice (Figure 2A). Mice immunized with rMVAGc virus showed an 80% survival after challenge. In this group at 11 dpi one mouse showed signs compatible to a delayed-onset neurological disease, dying at day 14 pi. In contrast, two of the mice vaccinated with rMVAGn showed earlier clinical disease dying at day 4 and 6 post challenge respectively. As expected, in the group of mice vaccinated with the rMVAGnGc construct the survival after challenge was 100%, with only one animal showing mild clinical signs between 3 and 4 dpi. In the mice from both control groups (either nonrecombinant MVA and unvaccinated) the mortality was 80% and 100% respectively with an earlier onset of disease in both groups. Differences in the survival rates observed for each group were statistically significant (χ<sup>2</sup> = 09.503; df = 3; *p* = 0.023) when compared to the control MVA vaccine (Mantel–Cox log-rank test).

**Figure 2.** Survival of MVA vaccinated mice (*n* = 5) upon challenge with virulent Rift Valley fever virus (RVFV). Kaplan–Meier plots of BALB/c mice (haplotype H2d (**A**) or 129EvSv mice (haplotype H2b) (**B**). The mice were vaccinated with a single intraperitoneal dose of 107 pfu of each recombinant virus or were mock-vaccinated (naive). Two weeks after immunization the mice were challenged with 10<sup>3</sup> pfu of RVFV 56/74. The mice were monitored for 3 weeks for the presence of signs of disease.

Since the protective effect of our rMVAGnGc single dose vaccine relies mainly in the induction of specific CD8+T-cell responses [17] we questioned whether its efficacy in a different inbred mouse strain would be compromised. Thus we tested the protective ability of our rMVA vaccines in the context of a different genetic background by using the 129SvEv mouse model (H-2b-haplotype). The sensitivity of this mouse strain to the RVFV challenge is higher than that of the H-2<sup>d</sup> BALB/c strain (our unpublished data). Survival rates upon the RVFV challenge in the 129SvEv mice immunized with rMVAGnGc reached 80% with only one animal dying at 5 dpi. Contrarily, to what was observed in the BALB/c experiments, all of the 129SvEv mice that were vaccinated with rMVAGc or rMVAGn died upon challenge, with a slight delay in mortality in the rMVAGc group with respect to the MVA control group (Figure 2B). Differences in survival times of rMVAGnGc were highly significant (χ<sup>2</sup> = 17.48; df = 3; *p* = 0.0006).

#### *3.3. Analysis of Humoral Responses in rMVA Vaccinated Mice*

We had previously reported the low level of in vitro neutralizing antibody induction induced by a rMVAGnGc vaccine in BALB/c mice and sheep [17,18]. As expected, levels of neutralizing antibodies in the serum from immunized BALB/c mice remained below the established detection threshold (1.3 log10, 1:20 serum dilution; Figure 3A).

**Figure 3.** Humoral responses upon MVA vaccination in mice. (**A**). Virus neutralization titers (VNT) in serum samples taken 14 days after immunization (prech.) of either BALB/c or 129SvEv mice. Neutralization titers were also estimated in mice that survived the challenge (postch.). Bars represent mean plus SD. (**B**). Detection of RVFV-infected cells by IFA with serum from MVA vaccinated mice at 7 or 14 days post immunization (dpi). The figure shows representative images of viral plaques detected on cells. (**C**). Immunoprecipitation of RVFV-infected BHK21 cell extracts with serum from MVA vaccinated BALB/c mice. C+: positive control serum from mice immunized with an adenovirus vector encoding GnGc. Mr: relative mass in kDa.

Only two 129SvEv mice from the rMVAGnGc group showed titers slightly above the threshold limit (1.6 log10, serum dilution 1:40). The rest of mice, either vaccinated with rMVAGn or rMVAGc did not show titers above the detection limit in any of the mouse models used. Moreover, one animal from the MVA control group showed a VNT50 titer of 1:20 indicating that the observed neutralization at this dilution could be unspecific. When a more stringent neutralization determination was applied (i.e., VNT100), no single prechallenge serum showed neutralization in all microtiter wells (not shown). Upon RVFV 56/74 challenge, all surviving animals showed elevated neutralization titers reaching around 3 logs. Although these data could indicate a successful priming of the vaccines, a similar titer observed in a surviving mouse from the control group ruled out this possibility. Interestingly, the mean postchallenge neutralization titer of the rMVAGnGc vaccinated 129EvSv mice was slightly higher, in agreement with the two mice showing prechallenge neutralization titers over the sensitivity limit. In spite of the lack of a clear in vitro neutralization activity, both vaccines induced antibodies able to label RVFV-MP12 infected cells as shown by indirect immunofluorescence (Figure 3B). In the case of the rMVAGc serum, antiGc antibodies developed later, since a clearly positive fluorescent signal on infected cells was only detected in serum collected 14 days post immunization (Figure 3B). However, none of these prechallenge sera was able to inmunoprecipitate metabolically labeled RVFV glycoproteins in infected cell extracts (Figure 3C).

#### *3.4. Analysis of Cellular Immune Responses to Vaccination*

ELISPOT assay using 14 dpi pooled peripheral blood leukocytes (PBLs) from BALB/c mice immunized with rMVAGc showed the highest numbers of IFN-γ secreting cells upon restimulation with two different Gc-specific, MHC-I-restricted, peptides #13 (SYKPMIDQL) and #14 (GGPLKTILL; Figure 4A).

**Figure 4.** Cellular responses upon MVA immunization. (**A**). Interferon gamma ELISPOT assay of pooled (*n* = 5) BALB/c peripheral blood leukocytes (PBLs) taken at day 14 postimmunization with the different MVA vaccines. Each pool was restimulated with either Gn (#4), or Gc (#13 or #14) specific peptides or with peptide pb9. Nonspecific stimulation was induced with phytohemaglutinin (pha). (**B**). Mean ± SD log spot forming cells (SFC) values obtained in spleen cells from BALB/c mice (*n* = 2) at day 7 post MVA immunization. As above, the peptides 4, 13, and 14 were selected on the basis of their ability to stimulate Gn and Gc specific T-cell responses. Cell culture medium with no added peptide (unst) was used to measure the background of the assay. The pb9 peptide was used as a specific positive control for each recombinant MVA (rMVA) vaccinated mice. In all groups asterisks indicate significance levels for each peptide when compared to the unstimulated control (unst) using Dunnett's multiple comparisons test (\*\* *p* < 0.01; \*\*\* *p* < 0.001; \*\*\*\* *p* < 0.0001).

Accordingly, lower numbers of cells were found upon stimulation with the Gn specific peptide (SYAHHRTLL). The rest of the groups showed lower numbers of IFN-γ secreting cells, with the only exception of the rMVAGn group restimulated with the pb9 control peptide. The highest number of IFNγ secreting cells was also found in the MVAGc group in an ELISPOT assay using splenocytes collected at 7 dpi (Figure 4B) or 14 dpi (data not shown). In contrast to the PBL assay, the number of spots was higher for the rMVAGnGc and rMVAGn upon restimulation with a Gn specific peptide. Higher responses were observed upon stimulation with pb9 control peptide in the rMVAGn group in comparison with rMVAGc or rMVAGnGc groups. Of note, the rMVAGn PBLs and spleen cells were also stimulated with Gc peptide 14, although at lower levels. The peptides used for restimulation of BALB/c spleen cells were not able to stimulate IFNγ secretion in 129SvEv spleen cells (not shown), indicative of the restriction imposed by the specific haplotypes. The ELISPOT data correlated with the

higher secretion of IL-2 and IL-6 cytokines, involved in T-cell survival that were detected by ELISA in the supernatants of restimulated cultures (Figure 5), indicative of the induction of a lymphoproliferative environment. Again, the group vaccinated with the rMVAGc vaccine displayed the highest amounts of both cytokines. On the other hand, IL-4 or IL-5, two of the main cytokines involved into B-cell proliferation, class switching, and differentiation to effectors were not detected in the same supernatants (not shown).

**Figure 5.** Detection of IL-6 and IL-2 secretion in ex-vivo restimulated spleen cells with peptides. The levels of each cytokine were estimated, using a capture ELISA, in supernatants collected at different times after stimulation. A standard curve was generated to correlate ELISA absorbance values with cytokine concentrations. The graph represents values after background subtraction from nonstimulated cells. The red dotted line determines the lower range of the standard curve.

#### *3.5. Assessment of E*ffi*cacy of Humoral Responses by Passive Serum Transfer Experiments*

In order to gain insights in the role of the humoral response induced by the rMVA vaccines, a passive serum transfer experiment was designed. In this experiment individual sera from BALB/c mice collected 14 days after immunization with the different rMVA vaccines were pooled. The challenge virus dose (5 <sup>×</sup> 103 pfu/mouse) was preincubated in the presence of each serum pool (final serum dilution 1/14) for 30 min, prior to the inoculation of mice. All mice that received the challenge dose in the presence of RVFV convalescent serum survived with no clinical display nor significant weigh loss (Figure 6A and Supplementary file, Figure S1). In contrast, all mice from the rMVAGn group and most of the mice inoculated with serum pooled from the rMVAGnGc or Ad5 control vaccinated groups died shortly after inoculation (Figure 6A). Interestingly, four out of five mice transferred with virus plus donor serum from rMVAGc or MVA control survived longer than mice from the rMVAGnGc group (χ<sup>2</sup> = 12.11; df = 3; *p* = 0.0070) and eventually recovered from infection (Supplementary file, Figure S1). Accordingly with the survival data, mean viremia titers were more elevated in rMVAGn and rMVAGnGc groups when compared to rMVAGc or MVA control (Figure 6B), and the differences between means were statistically significant (*p* < 0.01, ANOVA test). Of note, the surviving mouse

from the rMVAGnGc group had no conclusive viremia determination but it did not seroconverted (not shown), suggesting that this mouse was not efficiently infected.

**Figure 6.** Effect of rMVA immune serum in BALB/c survival. Mice (*n* = 5) were passively transferred with a mixture of immune serum from MVA vaccinated mice and 5 <sup>×</sup> <sup>10</sup><sup>3</sup> pfu of virulent 56/74 RVFV. For positive and negative control groups *n* = 4 was used. (**A**). Kaplan–Meier plots of survival proportions. The mice were monitored for 3 weeks for the presence of signs of disease. (**B**). Viremia at day three post inoculation tested by tissue culture infection doses in Vero cells. The infectious titer of each sample is defined as the reciprocal of the highest dilution of serum where a 50% of the cytophatic effect (cpe) is observed relative to noninfected controls. Only samples allowing a clear cpe determination are included. Samples inducing a non cpe-like effect were excluded. When no evident cpe was observed an arbitrary value of 10<sup>1</sup> TCID50 below the sensitivity limit (102) was assigned. Black lines represent means. Dotted line represents the sensitivity of the assay. The ANOVA test *p* value for differences among means is indicated. (**C**). Kaplan–Meier plots of BALB/c survival upon transfer with the rMVA ovine immune serum.

In order to confirm these observations a second passive transfer experiment was carried out with a serum pool obtained from a different rMVAGnGc vaccination experiment. Again, the mortality rates were higher and occurred earlier in the rMVAGnGc group than in the control groups (not shown). These data were suggestive of an exacerbating disease effect induced by the serum from animals vaccinated with rMVAGnGc or rMVAGn vaccines. Intriguingly, the extended survival in the rMVA control group was totally unexpected, indicating that the protective effect in the mice was not related to the presence of antiRVFV specific antibodies. In order to check whether the enhanced pathogenic effect of the rMVAGnGc serum was not exclusive of the mouse immune serum, sheep serum pooled from a previous rMVAGnGc vaccination experiment [25] was also used in a similar transfer experiment. The results showed unequivocally an accelerated mortality, with statistical significance (χ<sup>2</sup> = 7.740; df = 2; *p* = 0.0209) in the mice transferred with rMVAGnGc serum with respect to the serum from naive sheep (Figure 6C). Taken together the results observed suggest that the presence of serum anti Gn antibodies may trigger deleterious effects enhancing the infectivity of the virus inoculum.

#### **4. Discussion**

We proposed previously that the protective ability of a recombinant MVA vaccine encoding GnGc antigens relied mostly in T-cell immune responses in the absence of a strong in vitro neutralizing antibody response [17]. Apparently, the lack of neutralizing responses could be due to the type of immunity elicited by the vector itself, since the same coding sequence, either expressed by plasmid DNA, subunit vaccine (Gn) or delivered by means of an adenovirus vector eventually elicited stronger neutralizing antibody responses in mice [17,19,26]. In this previous work several MHC class-I restricted peptides from the glycoprotein sequences were identified for their ability to stimulate the secretion of IFNγ by CD8-T cells [17]. Here, our data confirms that BALB/c mice can be also protected upon the RVFV challenge by rMVAGc and, to a lesser extent, by rMVAGn and that this protection can also be achieved in the absence of neutralizing antibodies. According to the role of a cell-mediated immune response, the protection was restricted to a specific genetic background, as shown by the lack of survival upon challenge of 129SvEv mice immunized with the same vaccines (rMVAGn or rMVAGc). The detection of IL-2 and IL-6 supports the induction of cellular responses since both cytokines play a role in T-cell survival and activation. Particularly, Gc-specific T-cell responses may act as a key component in the protection after challenge observed in the rMVA immunized mice, perhaps contributing to the efficient elimination of RVFV infected cells. Our data also point out that the simultaneous expression of both glycoproteins by the MVA vector is an essential requirement for the induction of a protective response in the 129SvEv mouse strain. At this point it could be interesting to explore further how the genetic background determines the efficacy of the immune response and how differences in susceptibility to RVFV challenge may account for the observed differences in efficacy.

One of the most striking findings in this work is the exacerbating effect of some rMVA immune serum in infectivity. This was somewhat unexpected but may help to explain our previous observations in experiments conducted to evaluate the efficacy of the rMVAGnGc vaccine in sheep [25]. An indirect measure of viral replication in the host is the induction of antibody responses to immunogenic epitopes. For RVFV, the most immunogenic epitopes lie in the viral nucleoprotein N. Therefore, detection of antiN serum antibodies reveals the existence of a productive RVFV infection in the host. Earlier antiN antibody detection was observed in the serum from sheep vaccinated with rMVAGnGc when compared to non- and mock vaccinated controls. In addition, the amount of viral RNA detected in blood was higher at early times upon infection than in controls, indicating faster virus replication. We could reproduce here similar results upon the passive transfer of both mouse and sheep serum. One explanation to these findings would be the induction of subneutralizing antibodies able to enhance rather than block virus replication. Antibody dependent enhancement (ADE) has been described in several viral systems, with more detail in flaviviral infections [27–32]. In the case of RVFV it could be suggested that subneutralizing antiGn antibodies could bind to exposed Gn epitopes on the virus particle. In this scenario internalization of virus-antibody immune-complexes would be augmented in cells bearing complement or Fc receptors, increasing virus uptake and pathogenesis. This could explain the increased mortality of mice that were transferred with rMVAGnGc or rMVAGn serum:virus mixtures compared to that of rMVAGc or the MVA control. However, the protection observed in mice receiving the MVA control immune serum is puzzling. A plausible explanation could be that other nonantibody mediated humoral effectors provide some degree of protection although not sufficient to avoid deleterious effects of subneutralizing antibody responses. Thus, MVA or rMVAGc serum transfer would provide such antiviral effect while the transfer of rMVAGnGc or rMVAGn serum would enhance infectivity through subneutralizing antiGn antibodies. Both the classical and alternative pathways of the complement system can be activated upon viral infections and it has been shown that complement system plays an important role in poxvirus immunity [33]. On the other hand, deposition of complement proteins on the surface of enveloped virions enhances uptake by phagocytosis and potentially interferes with receptor interactions, virus entry, and uncoating [34]. Whether this hypothesis is or not true would deserve further experimentation. Nonetheless, it becomes clear that improving the quality of the antibody response of our MVA vaccines would render them

more efficacious against a lethal RVFV challenge. Current work is underway to test the ability of novel MVA recombinants using different vector source and stronger promoter sequences for increasing antigen expression and enhancing proper processing of antigens.

#### **5. Conclusions**

In conclusion we confirmed the possibility of protecting mice against a lethal RVFV challenge without induction of neutralizing antibody responses, stressing the importance of cell-mediated immune responses in protection. Most importantly, failing in inducing proper neutralizing antibody responses may result in enhanced pathogenesis when the cell mediated immune response is impaired or absent.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-393X/8/1/82/s1, Table S1: primers used for inverse PCR reaction. Figure S1. Weight loss after passive transfer in mice. Figure S2. Amino acid sequences of eGn and Gc encoded by the recombinant MVAs.

**Author Contributions:** Conceptualization, A.B. and G.L.; methodology, E.L.-G., S.M., G.L., J.O., B.B. A.B.; formal analysis, E.L.G., G.L., A.B.; writing—original draft preparation, E.L.-G., G.L., A.B.; writing—review and editing, A.B.; supervision, G.L., A.B.; project administration, A.B.; funding acquisition, A.B., B.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by grants AGL-2011-22485, AGL-2017-83226R from the Spanish Ministry of Science and S2013/ABI-2906, S2018/BAA-4370 from Comunidad de Madrid. ELG was a recipient of a pre-doctoral fellowship program from the Spanish Ministry of Science.

**Acknowledgments:** We thank Nuria de la Losa for her excellent technical assistance and the animal care staff at Centro de Investigación en Sanidad Animal

**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/).

## **Di**ff**erential Immune Transcriptome and Modulated Signalling Pathways in Rainbow Trout Infected with Viral Haemorrhagic Septicaemia Virus (VHSV) and Its Derivative Non-Virion (NV) Gene Deleted**

### **Blanca Chinchilla 1, Paloma Encinas 2, Julio M. Coll <sup>2</sup> and Eduardo Gomez-Casado 2,\***


Received: 20 December 2019; Accepted: 27 January 2020; Published: 30 January 2020

**Abstract:** Viral haemorrhagic septicaemia virus (VHSV) is one of the worst viral threats to fish farming. Non-virion (NV) gene-deleted VHSV (dNV-VHSV) has been postulated as an attenuated virus, because the absence of the *NV* gene leads to lower induced pathogenicity. However, little is known about the immune responses driven by dNV-VHSV and the wild-type (wt)-VHSV in the context of infection. Here, we obtained the immune transcriptome profiling in trout infected with dNV-VHSV and wt-VHSV and the pathways involved in immune responses. As general results, dNV-VHSV upregulated more trout immune genes than wt-VHSV (65.6% vs 45.7%, respectively), whereas wt-VHSV maintained more non-regulated genes than dNV-VHSV (45.7% vs 14.6%, respectively). The modulated pathways analysis (Gene-Set Enrichment Analysis, GSEA) showed that, when compared to wt-VHSV infected trout, the dNV-VHSV infected trout upregulated signalling pathways (*n* = 19) such as RIG-I (retinoic acid-inducible gene-I) like receptor signalling, Toll-like receptor signalling, type II interferon signalling, and nuclear factor kappa B (NF-kappa B) signalling, among others. The results from individual genes and GSEA demonstrated that wt-VHSV impaired 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 promoted the opposite effects. Therefore, these results might support future studies on using dNV-VHSV as a potential live vaccine.

**Keywords:** VHSV; non-virion (NV); transcriptome profiling; rainbow trout; immune pathways

#### **1. Introduction**

Viral haemorrhagic septicaemia virus (VHSV) belongs to the *Novirhabdovirus* genus, together with infectious haematopoietic necrosis virus (IHNV), snakehead rhabdovirus (SHRV), and hirame rhabdovirus (HIRV). They are all enveloped negative-stranded RNA viruses with a single RNA genome of ~11 Kb [1–3], which encodes five virion proteins (N, P, M, G, and L proteins) and the non-virion (NV) protein that gives the name to the *Novirhabdovirus* genus and differentiates it from other fish rhabdoviruses such as spring viremia carp virus (SVCV). VHSV has been isolated from more than 50 fish species from North America, Asia, and Europe, including 15 farmed [4] and free-living marine fish species [5] like trout, salmon, turbot, and eel, among others. Within a farm, the presence of VHSV infection, even if in only one individual fish, has to be notified to the Office International des Epizooties (OIE, Paris, France) and implies the sacrifice of all the farmed fish, thus leading to serious economic losses [6,7]. The *NV* gene was firstly characterised and named from IHNV genome studies [8]. Some

years later, the *NV* gene from VHSV was further characterised by comparative genome studies [9]. Despite the presence of the *NV* gene in the four novirhabdovirus species mentioned above, their NV proteins showed very divergent inter-species sequences [10,11]. Initial studies regarding NV role showed that it was required for the highest efficient replication of IHNV in rainbow trout [12–14] and that of VHSV in olive flounder [13,15] and in *Epithelioma papulosum cyprinid* (EPC) cells [13]. However, NV was not essential for in vitro or in vivo SHRV production in warm-water flatfish [16,17]. Further, in vitro studies using the wild-type (wt) and NV knock-out IHNV or VHSV suggested that NV downregulated the host *ifn1*/*mx* transcriptional levels during in vitro infection in trout (RTG-2, Rainbow Trout Gonad-2) [18] or EPC cells [15], respectively. The higher levels of IFN-induced *mx* transcript in NV knock-out VHSV vs. wt-VHSV injected flounder found in these studies suggested that NV also interferes with IFN defences in vivo to favour VHSV replication [15]. The early anti-apoptotic role of NV during the first stages of VHSV infection has been also demonstrated [19]. Using recombinant NV protein (rNV) and a trout immune-targeted microarray, we have previously determined not only an anti-apoptotic role for NV, but also a plethora of novel expression changes (mainly downregulated) in genes associated with immune innate and adaptive response (i.e., interferons, MX, tumour necrosis factors, antigen presentation, interleukins) [6]. However, the effects driven by the injection of the rNV protein alone will probably differ from those induced by the NV in the course of VHSV infection. Recently, a microarray study in olive flounder liver infected with VHSV described differential gene expression and gene ontology classification of these genes [20], resulting in a global transcriptome profiling where only a few genes have been classified as immune-related. Gene expression has been also characterised with microarrays in olive flounder infected with a VHSV strain that produces high mortality in this species [21]. In addition, the protection of olive flounder against VHSV was previously assessed by immunization with the *NV* gene-knockout recombinant VHSV, which led to good protection against virulent VHSV [22]. However, the underlying mechanisms of olive flounder protection remain unknown. The aim of this work was to characterise the transcriptomic profiling of immune-related genes in trout infected with the wild-type VHSV (expressing NV) and dNV-VHSV (*NV* gene-deleted) in order to find targeted immune genes and signalling pathways implied in the course of a VHSV infection. This study will contribute to better understand how NV modulates gene expression and how the expression pattern changes in response to dNV-VHSV. The results will help tailoring future vaccines against viral haemorrhagic septicaemia virus and other novirhabdoviruses.

#### *1.1. Viral Haemorrhagic Septicaemia Viruses (VHSV)*

Wild type VHSV-23.75 (wt-VHSV) isolated from brown trout [23] (GenBank accession number FN665788) and its derivative *NV* gene-deleted (dNV-VHSV), obtained as previously described [13], were used to infect rainbow trout (*Oncorhynchus mykiss*) by intraperitoneal injection. Both wt- and dNV-VHSV viruses were kindly provided by Dr. Michel Brémont (INRA, France), and further propagated in EPC cells at 14 ◦C and titrated by the plaque-forming assay (pfu) as previously described [14].

#### *1.2. Virus Dosages and Injection of Fingerling Rainbow Trout*

Fingerling rainbow trout (*Oncorhynchus mykiss*) of 6–12 g (approximately 10 cm in length), free of IPNV (Infectious Pancreatic Necrosis Virus) and VHSV antibodies, were obtained from a local fish farm (Los Molinos, Madrid). They were maintained at 14 ◦C in a 200 L aquarium with tap-dechlorinated carbon-filtered water provided with biological filters and fed with a commercial fish diet. After two weeks of acclimation, fish were separated into seven groups of six trout per group. Due to the non-virion (NV) expression by wt-VHSV, a different replication rate of wt-VHSV and dNV-VHSV in the EPC cell line was established [13], and since the maximum NV expression by wt-VHSV is reached at 48 hpi (hours post-infection), we aimed at establishing the equivalent infectious dosage yielding comparable transcriptomic profile for each virus at 48 hpi. For that, we injected trout intraperitoneally with 100 μL of wt-VHSV (104, 105 or 35 <sup>×</sup> 106 pfu), dNV-VHSV (104, 105 or 35 <sup>×</sup> 106 pfu) or phosphate-buffered saline (PBS). Each group of injected trout was then released into a 50-L aquarium and maintained at 14 ◦C. Two days after injection, trout were sacrificed, head kidney and spleen (whole organs) were pooled, and immediately immersed in RNAlater (Ambion, Austin, USA) at 4 ◦C overnight, before being frozen at −70 ◦C until further analysis.

#### *1.3. RNA Extraction and cDNA Synthesis*

The pooled head kidney and spleen (whole organs) from each individual trout were homogenized using the Tissue Lyser Cell Disruptor (Qiagen Iberia, S.L., Madrid, Spain) for 10 min at 50 Hz with 3 mm glass beads in an RTL buffer (Qiagen Iberia, S.L., Madrid, Spain). RNA was then extracted from the homogenates by using the RNAeasyPlus kit (Qiagen Iberia, S.L., Madrid, Spain) and eluted in RNase-free water. RNA concentrations were estimated by Nanodrop and the presence of 18S and 28S rRNA bands was confirmed by denaturing RNA agarose electrophoresis (Sigma-Aldrich Quimica SA, Madrid, Spain). For qPCR experiments of the nucleoprotein (*N*) and non-virion (*NV*) genes, cDNA synthesis was carried out from RNA (1 μg) by using oligo-dT and PrimeScriptTM reverse transcriptase (RR037A TAKARA, Japan) according to the manufacturer's instructions. For microarray experiments, additional RNA quality controls (RNA integrity number, RIN) were performed by NIMGenetics (Madrid, Spain). For each experimental six-trout group, the four trout with best RNA quality (RIN > 7.0) were chosen for microarray hybridisation. cDNA was synthesized by using SuperScript III reverse transcriptase (Invitrogen) and oligo(dT) primer, labelled with Cy3 (GE Healthcare, Spain), and purified with Microcon YM30 (Merck Millipore, Spain).

#### *1.4. Design of Oligo-Microarrays Enriched in Rainbow Trout Immune-Related Genes (Targeted Microarrays)*

Oligo-microarrays were enriched in rainbow trout immune-related genes as previously described (immune-targeted microarrays) [6,24,25]. The final 8 × 15K microarray corresponds to Agilent's ID032303 (Gene Expression Omnibus GEO platform submission number GPL14155) and contains 1474 annotated immune-related probes (60-mer) per duplicate. In order to simplify the analysis of results, annotated probes were classified according to the following gene groups: VIG, VHSV-induced genes (number of probes, *n* = 22); IFN, interferons and their receptors (*n* = 20); MX, interferon-inducible proteins mx (*n* = 3); CO, complement components (*n* = 6); IL, interleukins and their receptors (*n* = 19); APM, antigen-presenting machinery genes (*n* = 4); TNFSF, tumour necrosis factor superfamily (*n* = 16); CD, cluster differentiation antigens (*n* = 15); CK, chemokines and their receptors (*n* = 32); CASP, caspases (*n* = 3); and TF, transcription factors (*n* = 10). The trout microarray used for these experiments was previously validated by real-time quantitative PCR (RTqPCR) [24,25]. The number of biological replicas was four. Four chips of eight samples per chip were used and hybridised simultaneously. This home-made rainbow trout oligo-microarray contains more immune-related genes than any other trout microarray available, since it includes all the immune-related genes from the Agilent's EST-derived rainbow trout oligo-microarray (ID16271) [6,24–26].

#### *1.5. Hybridisation and Gene Expression Changes of Trout Transcripts to the Immune-Targeted Microarrays*

The labelling of 2 μg of RNA (approximately 50 μg/mL) and hybridisation to the microarrays were performed by NIMGenetics (Madrid, Spain) complying with the Minimum Information about a Microarray Experiment (MIAME) standards [24].

Normalisations were performed by correcting the individual fluorescence in each microarray with the sum of all the fluorescent values according to the formula: fluorescence of each probe/sum of all the probe fluorescence signals per microarray. Raw and normalised data were deposited in GEO [27,28]. After normalisation, outlier values (defined by those fluorescence values above or below mean ± standard deviation per probe) were identified and eliminated from the calculations programmed in Origin Pro 8.6 (OriginLab Corporation, Northampton, MA, USA). Fold-change (FC) for each probe was calculated by the following formula: values of wt-VHSV or dNV-VHSV injected trout/mean of PBS injected trout (*n* = 4). Means and standard deviations of individual folds were calculated for

each oligonucleotide probe by the following formula: fluorescent value/mean fluorescent value of the control (*n* = 4). Venn diagrams reflect the percentage of genes which FC value was upregulated, downregulated, and non-regulated for each comparison with these arbitrary criteria previously used [6]: (1) upregulated: FC ≥ 1.5; (2) downregulated: FC ≤ −1.5; and (3) non-regulated (basal) gene expression: −1.5 < FC < 1.5. On the other hand, the heatmap figures reflect the FC for each gene comparison using an arbitrary criteria previously described [6]: (1) non-regulated (basal) gene expression: −1.5 < FC < 1.5 (black); (2) upregulated: 1.5 ≤ FC < 2 (light red box), 2 ≤ FC < 5 (red box), 5 ≤ FC (dark red box); and (3) downregulated: –1.5 ≥ FC > −2 (light green box), −2 ≥ FC > –5 (green box), −5 ≥ FC (dark green box). Differentially expressed gene transcripts were considered significant when FC ≥ 1.5 or FC ≤ −1.5. Negative folds were calculated for those values below 0.66 applying the formula −1/FC. Therefore, FC = 0.66 corresponds to a –1.5 value; FC = 0.5 (more downregulated) corresponds to a −2 value; and FC = 0.2 (even more downregulated) corresponds to a –5 value.

#### *1.6. Quantitative Estimation of Transcripts by Real-Time Quantitative PCR (RTqPCR)*

To estimate the wt- or dNV-VHSV replication in rainbow trout head kidney and spleen, both *N* and *NV* transcript levels were estimated by RTqPCR amplification after intraperitoneal injection of the corresponding VHSV, as described in Section 1.2. RNA extraction and cDNA synthesis were carried out as described above. RTqPCR was carried out by mixing 100 ng of cDNA, 0.9 μM of forward primer, 0.9 μM of reverse primer, and Power SYBR Green PCR Master Mix (Life Technologies, Madrid, Spain). The thermal profile was 10 min at 95 ◦C, followed by 40 cycles of 95 ◦C for 15 s, and 60 ◦C for 1 min. For each experiment, the expression level of the analysed genes was calculated using the 2-ΔΔCt relative quantitation method. The Ct for each viral gene was normalised to β*-actin* gene (ΔCt*gene* = Ct*gene* − Ctβ*-actin*), which was used as an internal control due to its low coefficient of variation (CV) among different virus dosages (CV <sup>≤</sup> 3% for 10<sup>4</sup> pfu/trout and 10<sup>5</sup> pfu/trout of wt- and dNV-VHSV; trout injected with <sup>35</sup> <sup>×</sup> <sup>10</sup><sup>6</sup> pfu of both viruses showed a CV close to 8%). Means and standard deviations were calculated for each experimental infection by intraperitoneal (ip) injection of either the wt- or dNV-VHSV in trout groups (*n*=6) with 104, 105, or 35×106 pfu/trout. Primer sequences used were: <sup>β</sup>*-actin* (accession number AF550583.1) forward 5 CATCACCATCGGCAACGA and reverse 5 GATGTCCACGTCACACTTCAT; nucleoprotein (accession number AJ233396) forward 5 TCTCCGCTCGTCCTCCGTGAG and reverse 5 GTGAGCCCAGAGCCTCTTGTC; and non-virion (*NV*, accession number AJ233396) forward 5 TCAAGGTGACACAGGCAGTCA and reverse 5 CCAGTTCTCTCATGGGCATCAT. Calculated RTqPCR efficiency was 59% for β*-actin*, 45% for *N*, and 43% for *NV* genes. Efficiency was considered to correct the transcript levels obtained by RTqPCR assays.

#### *1.7. Calculations used for Gene Set Enrichment Analysis (GSEA)*

In order to explore the possible biological effects of simultaneous and small changes in several related genes, we screened the transcriptional data with the previously described 51 rainbow trout from the immune-related gene-set (GS) collection [25]. The trout GS collection was manually designed from the KEGG (K) and WIKI (W) trout orthologous human pathways (as accessed in 2013), using the trout genes contained in our home-designed microarray. The trout GS collection was then used for analysis by the Gene-Set Enrichment Analysis (GSEA) program [29–31]. Transcriptional data from the dNV-VHSV and wt-VHSV injected trout were analysed by GSEA to assign a normalised enrichment score (NES) for each GS of the collection in each of the three cases [25].

#### *1.8. Ethics Statement*

All the animal procedures used in this study were approved by the INIA (National Agricultural and Food Research and Technology Institute) ethical and biosecurity committee (authorization CEEA 2011/022) and performed following the National and European Commission guidelines and regulations on laboratory animals' care. Periodic examinations were performed several times a day during infections so as to euthanize fish with abnormal behaviour. To minimize animal suffering, fingerling

rainbow trout were sacrificed by using a lethal dose of tricaine methanesulfonate (MS-222, 50 mg/mL, Sigma, Madrid, Spain).

#### **2. Results and Discussion**

#### *2.1. dNV- and wt-VHSV Dosages used for Microarray Analysis*

In this work, we have firstly defined the appropriate infectious dosage for wt-VHSV and dNV-VHSV in order to establish the comparative transcriptomic profiling between them. At 48 h post-infection, the RNA transcripts from head kidney and spleen were analysed by RTqPCR to estimate the corresponding viral replication loads based on *N* transcript levels (Figure 1). The trout injected with 10<sup>4</sup> pfu, 10<sup>5</sup> pfu, and 35 <sup>×</sup> 10<sup>6</sup> pfu of the wt-VHSV yielded the following *N* (±SD) values: 2.3 <sup>±</sup> 1.1, 9.0 <sup>±</sup> 4.1, and 348.8 <sup>±</sup> 125, respectively (Figure 1). On the other hand, trout injected with 10<sup>4</sup> pfu, 10<sup>5</sup> pfu and 35 <sup>×</sup> 106 pfu of the dNV-VHSV yielded the following *N* (±SD) values: 0.6 <sup>±</sup> 0.2, 1.1 <sup>±</sup> 0.4, and 19.3 ± 7.6, respectively (Figure 1). Wild-type-VHSV/dNV-VHSV ratio for *N* transcripts yielded an approximately 18-fold higher replication rate for the wt-VHSV when both viruses were applied at a dose of 35 <sup>×</sup> 106 pfu/trout dose. However, this proportion was close to 1 when with the dose used was 10<sup>5</sup> pfu for wt-VHSV and 35 <sup>×</sup> 10<sup>6</sup> pfu for dNV-VHSV. For this reason, we considered it best to use the latter for the microarray study. Therefore, we compared the transcriptomic profile of trout injected with 10<sup>5</sup> pfu of wt-VHSV (NV presence) (data deposited on GEO GSE37330) and trout injected with a 350-fold more infectious dose (35 <sup>×</sup> 106 pfu) of dNV-VHSV (NV absence) (data deposited at GEO GSE43285).

**Figure 1.** Nucleoprotein (*N*) and non-virion (*NV*) gene expression values (mean ± SD) obtained by RTqPCR of six trout injected with wild-type (wt)-VHSV (hatched bars) and *NV* gene-deleted (dNV)-VHSV (white bars). Animals were injected with three different dosages (104, 105, and <sup>35</sup> <sup>×</sup> <sup>10</sup><sup>6</sup> pfu/trout) of each virus and analysed 48 hpi. The results showed similar nucleoprotein (boxed N) transcript levels of wt-VHSV (105 pfu/trout) and dNV-VHSV (35 <sup>×</sup> 106 pfu/trout), indicating their similar replication levels. Horizontal dotted line corresponds to negative control. VHSV: viral haemorrhagic septicaemia virus.

#### *2.2. Overview of the Expression Profiles Obtained*

Normalised FC values of microarray datasets from wt-VHSV (105 pfu/trout) and dNV-VHSV (35x10<sup>6</sup> pfu/trout) were classified based on the groups defined for this study (see Section 1.5) and calculated the percentage of upregulated, downregulated and non-regulated genes. Figure 2A shows the upregulated genes between groups (wt-VHSV and dNV-VHSV) with fold changes (FC) ≥ 1.5 in Venn diagrams. The trout injected with dNV-VHSV showed the highest number of upregulated genes (65.6%), followed by those injected with wt-VHSV (45.7%). Probably, the upregulation increase in the dNV-VHSV injected trout was due to the absence of NV. In addition, dNV-VHSV and wt-VHSV shared 25.8% of the upregulated genes, which should be due to other viral proteins rather than NV.

**Figure 2.** Venn diagrams showing the general transcriptomic relationships between trout injected with dNV-VHSV and wt-VHSV. Numbers indicate the percentage of upregulated (fold change, FC ≥ 1.5, (**A**), downregulated (FC ≤ −1.5, (**B**), and non-regulated (−1.5 < FC < 1.5, (**C**) genes, 48 h after injection with dNV-VHSV and wt-VHSV.

Figure 2B displays Venn diagrams with downregulated genes (FC ≤ −1.5), which indicate that fish infected with dNV-VHSV downregulated more genes than fish infected with wt-VHSV (19.2% vs 9.3%, respectively). On the other hand, dNV-VHSV and wt-VHSV do not have any downregulated genes in common.

When the non-regulated gene transcript levels with folds −1.5 < FC < 1.5 (Figure 2C) were analysed by Venn diagrams, the trout infected with wt-VHSV had the highest number of non-regulated genes (45.7%), followed by dNV-VHSV (14.6%). The Venn diagram also showed that 5.3% of non-regulated genes were shared by wt-VHSV and dNV-VHSV.

#### *2.3. dNV- and wt-VHSV Infection E*ff*ects on Trout Immune-Related Genes*

A detailed study of the transcriptomic changes driven by wt-VHSV and dNV-VHSV at 48 hpi was conducted. A heatmap was generated with the FC of the genes grouped in the categories described in methods. In addition, individual genes were also correlated with the pathways (KEGG database) [32] in which they participate (Table 1).



#### 2.3.1. Cytosolic Sensors

Upon infection, viruses are recognised by host receptors and cytosolic sensors that activate mechanisms (signalling molecules) involved in diverse cellular processes such as the antiviral immune response. The cytosolic sensors studied belong to different gene groups and signalling pathways (Figures 3 and 4, and Table 1, RIG-I-like, and NOD-like signalling). Among the genes involved in these pathways, *tnf*, *ifna*, *irf7*, and *dhx58* and *mavs* (Figure 3, IFN and TNFSF) were upregulated by wt-VHSV and slightly more by dNV-VHSV. *Mavs*, another important cytosolic sensor, was upregulated by dNV-VHSV but non-regulated by wt-VHSV. This fact could be due to the NV expression by wt-VHSV and might indicate that *mavs* would be more functional for starting an appropriate immune response against dNV-VSHV as vaccine virus. Other important signalling molecules are *traf2* and *traf3* (Figure 3, TNFSF), and they are also upregulated by dNV-VHSV and non-regulated by wt-VHSV. These findings also support that dNV-VHSV activates better than wt-VHSV at these stages.

#### 2.3.2. IFN System

We studied different genes belonging to the IFN system and IFN-related group of genes that belong to different signalling pathways (Figure 3, IFN pathways, *n* = 14). Regarding the IFN group of genes (IFN, *n* = 20), wt-VHSV induced the upregulation of the lowest number of genes (35%) compared to dNV-VHSV (85%) (Figure 3, IFN). This fact might be due to an inhibitory effect driven by NV after its expression from wt-VHSV. Among the downregulated and non-regulated (basal) genes, we found *iip30*, *ifng1*, *ifng2*, *iip1*, *iip2*, *ifp58*, *ifp35*, *mavs*, *ifn1*, *ifn2*, and *ifn5* (Figure 3, IFN). On the other hand, the expressions of *dhx58*, *hep*, *ifn3*, *ifna*, *irf1*, *irf7*, and *ifn4* were upregulated by wt-VHSV. Thus, some of them (i.e., *dhx58*, *hep*, *irf7*, *ifn3*) were even more upregulated by dNV-VHSV, probably due to the lack of the NV protein. The NV protein expressed by wt-VHSV induced the non-regulation of *mx2* and *mx3*, whereas *mx1* was slightly upregulated. The results also showed that all *mx* genes were upregulated by dNV-VHSV (Figure 3, MX). Mx proteins are implicated in the antiviral interferon-mediated response. In summary, in contrast to wt-VHSV, dNV-VHSV improved the antiviral immune response based on interferons (and related molecules), which would support its use as a potential live vaccine.

#### 2.3.3. TNF Superfamily and Caspases

This gene group is comprised of molecules with diverse functionality that participate in several signalling pathways (*n* = 20, Figure 3 TNFSF, and Table 1). Most of the 16 TNF superfamily genes (*tnfsf*) studied here were upregulated (75%) by dNV-VHSV, whereas wt-VHSV induced a lower upregulation (50%) of these genes (Figure 3, TNFSF). The *tnfsf* genes upregulated by dNV-VHSV but downregulated or non-regulated by wt-VHSV were *balm*, *tnfd* (decoy), *tnfdr* (decoy receptor), *tnfr*, *tnfsf10*, *tnfsf13*, *tnfsf14,* and *tnfsf6* (Figure 3, TNFSF), which participate in several multipaths (Figure 3, TNFSF, Table 1). The *balm* gene is closely related to *tnfsf13b* (BAFF) and *tnfsf13* (APRIL) genes, and seems to be unique to teleost [33]. The *balm* gene has a constitutive expression in adult trout, mainly in the spleen, lymphocytes, posterior kidney, and anterior kidney and, therefore, *balm* has been assigned an immunological role [33]. The *tnfsf6* (FAS ligand, FASL) and *tnfsf13* (APRIL) genes modulate ligand-induced apoptosis [34]. In addition, *tnfsf10* (or TRAIL) is an inductor of apoptosis acting through *casp3* and *casp8*. Among the upregulated genes induced by wt- and dNV-VHSV injection were *tnfsf14* (stimulator of apoptosis), *tnf* (most important inducer of systemic inflammation), and *ltb1*/*ltb2* (involved in proliferation, differentiation, survival, and growth). Another member of TNFSF gene group is *tnfaip3* (A20 protein), which is upregulated in trout infected with wt-VHSV and downregulated in those infected with dNV-VHSV. Previous studies showed that A20 inhibited NF-kappa B and apoptosis [35,36] and our results support that dNV-VHSV would promote NF-kappa B signalling in order to set up a successful immune response.

**Figure 3.** Heatmap showing the transcriptional expression fold changes (FCs) of the selected gene groups VHSV-induced (VIG), myxovirus resistance proteins (MX), caspases (CASP), tumour necrosis factor superfamily (TNFSF), interferon (IFN), and antigen-presenting machinery (APM) induced by dNV-VHSV and wt-VHSV in infected trout. *Gene* denotes names, and the *Pathway* column correlates with Table 1. –, unassigned pathway. Column 1, dNV-VHSV, each box corresponds to the average FC from four trout. Column 2, wt-VHSV, each box corresponds to the average FC from four trout. VIG group: b191 (c-lectin, AF483535), vig-1 (AF076620), vig-2 (AF290477), vig-3 (AF483529), vig-4 (AF483530), vig-5 (clone B17), vig-6 (clone B126), vig-7 (AF483527), vig-8 (clone B68), vig-9 (AF483533), vig-10 (AF483534), b203 (AF483538), b143 (AF483539), b225 (AF483540), b88 (AF483541), b160 (AF483545), b124 (AF483546), b305 (AF483542), cxci7 (VHSV induced protein 7 (vig7)), vi1 (VHSV induced protein 1), gbp (guanylate-binding protein GTPase, b21 (CD9, AF483544)). MX group: mx1 (myxovirus resistance 1), mx2 (myxovirus resistance 2), mx3 (myxovirus resistance 3). CASP group: casp3 (caspase 3), casp6 (caspase 6), casp9 (caspase 9). TNFSF group: balm (BAFF and APRIL-like molecule), ltb1 (lymphotoxin beta 1), ltb2 (lymphotoxin beta 2), tgfb (tumour growth factor beta), tnf (tumour necrosis factor alpha), tnfd (tumour necrosis factor decoy), tnfdr (tumour necrosis factor decoy receptor), tnfr (tumour necrosis factor receptor), tnfsf10 (tumour necrosis factor superfamily 10), tnfsf11 (tumour necrosis factor superfamily 11), tnfsf13 (tumour necrosis factor superfamily 13), tnfsf14 (tumour necrosis factor superfamily 14), tnfsf15 (tumour necrosis factor superfamily 15), tnfsf5 (tumour necrosis factor superfamily 5 (CD40)), tnfsf6 (tumour necrosis factor superfamily 6), tnfaip3 (tumour necrosis factor alpha-induced protein 3). IFN group: dhx58 (RIG-I-like receptor LGP2), iip30 (interferon gamma inducible protein 30), ifna (interferon alpha), ifng1 (interferon gamma 1), ifng2 (interferon gamma 2), iip1 (interferon inducible protein 1), iip2 (interferon inducible protein 2), hep (hepcidin), ifp58 (interferon-induced protein 58), ifp35 (interferon-induced protein 35), irf1 (interferon regulatory factor 1), irf10 (interferon regulatory factor 10), irf2 (interferon regulatory factor 2), irf7 (interferon regulatory factor 7), mavs (mitochondrial antiviral signalling protein), ifn1 (type 1 interferon 1), ifn2 (type 1 interferon 2), ifn3 (type 1 interferon 3), ifn4 (type 1 interferon 4), ifn5 (type 1 interferon 5). APM group: tapbp (tapasin (TAP binding protein)), tap1 (transporter associated with antigen processing 1), psmb9a (proteasome subunit type 9), b2m (beta-2 microglobulin).

**Figure 4.** Heatmap showing the transcriptional expression fold changes (FCs) of the selected gene groups cytokines (CK), interleukins (IL), cluster of differentiation antigens (CD), complement (CO) and transcription factor (TF), induced by dNV-VHSV and wt-VHSV. *Gene* denotes names, and the *Pathway* column correlates with Table 1. –, unassigned pathway. Column 1, dNV-VHSV, each box corresponds to the average FC from four trout. Column2, wt-VHSV, each box corresponds to the average FC from four trout. CK group: ccl13 (cc-chemokine 13), ccl4 (cc-chemokine 4), ck10 (cc-chemokine 10), ck11 (cc-chemokine 11), ck12a (cc-chemokine 12a), ck12b (cc-chemokine 12b), ck1 (cc-chemokine 1), ck3 (cc-chemokine 3), ck4a (cc-chemokine 4a), ck4b (cc-chemokine 4b), ck5a (cc-chemokine 5a), ck5b (cc-chemokine 5b), ck6 (cc-chemokine 6), ck7a (cc-chemokine 7a), ck7b (cc-chemokine 7b), ck8a (cc-chemokine 8a), ck8b (cc-chemokine 8b), ck9 (cc-chemokine 9), cklf7 (chemokine-like factor superfamily member 7), cxc (α-chemokines), cxcd2 (cxc d2 chemokine), il2rg (il2 receptor gamma), nilt4 (novel immunoglobulin-like transcript 4 (FM200774.1)), ckrg (cytokine receptor gamma), crlp1 (chemokine receptor-like protein 1 (AJ620468.1)), socs1 (suppressor of cytokine signalling 1), socs2 (suppressor of cytokine signalling 2), socs3 (suppressor of cytokine signalling 3), socs4 (suppressor of cytokine signalling 4), socs5 (suppressor of cytokine signalling 5), socs6 (suppressor of cytokine signalling 6), socs7 (suppressor of cytokine signalling 7). IL group (r means receptor): il1b, il2, il6, il6m17, il7, il8, il10, il12b, il15, il16, il20, il20ra, il21r, il22, il27 (p28 subunit), il29, Irak4 (interleukin-1 receptor-associated kinase 4), nil1 (novel il-1 cytokine family member). CD group: cd103, cd163, cd2, cd209, cd276, cd28, cd36, cd3e (epsilon), cd4, d79a, cd83, cd8b (beta), cd9. CO group: cr1 (complement receptor type 1), c4 (complement component 4), c6 (complement component 6), c7.1 (complement component 7-1), c9 (complement component 9), prf (perforin). TF group: stat1 (signal transducer and activator of transcription 1), stat5 (signal transducer and activator of transcription 5), sox30 (SRY-related HMG box 30 gene family), sox9 (SRY-related HMG box 9 gene family), sox19a (SRY-related HMG box 19a gene family), sox17 (SRY-related HMG box 17 gene family), sox21 (SRY-related HMG box 21 gene family), sox5 (SRY-related HMG box 5 gene family), tfiia (transcription factor IIA), tbx10 (T-box 10 gene).

TNFs have a role as 'double-edged swords' in cellular proliferation, survival, differentiation or apoptosis. Ligands such as APRIL (*tnfsf13* gene), LIGHT (*tnfsf1*4 gene), RANKL (*tnfsf11* gene), LT-β (*ltb1*, *ltb2* genes), and CD40L (*tnfsf5* gene) bind to receptors with a TRAF-interacting motifs (TIM) domain, leading to the recruitment of TRAF molecules, and the activation of multiple signal transduction pathways such as NF-kappa B, Jun N-terminal kinase (JNK), p38, extracellular signal regulated kinase (ERK), and phosphoinositide-3 kinase (PI3K). On the other hand, ligands such as TNF-α (*tnf* gene), TRAIL (*tnfsf10* gene), FASL (*tnfsf6* gene), and decoy receptor have a dead domain (DD), which ultimately activates apoptosis through caspases. In summary, most of the *tnfsf* genes

were upregulated by dNV-VHSV, giving rise to both activation signalling (NF-Kappa B, JNK, etc) and apoptosis signalling pathways.

Regarding the caspase group (Figure 3, CASP, and Table 1), the effector (*casp3*, *casp6*) and initiator (*casp9*) caspase genes were highly upregulated in the dNV-VHSV infected group while maintained at normal transcription levels in the wt-VHSV infected group. These results suggest that NV expressed by wt-VHSV impairs the upregulation and, consequently, the activation of apoptosis at 48 h post-infection (Figure 3, CASP). Moreover, the upregulation of the *casp* genes by dNV-VHSV promotes the activation of different pathways in which they participate, supporting an appropriate immune response developed by dNV-VHSV.

#### 2.3.4. Antigen Presentation

Among the antigen presenting machinery (APM, *n* = 4) genes studied, TAPASIN (*tapbp*) and proteasome subunit (*psmb9a*) were non-regulated genes, whereas *tap1* and *b2m* were upregulated in wt-VHSV (Figure 3, APM). The APM genes belong to innate and adaptive immune responses. They act within the proteasome for peptide generation (*psmb9a*) and the transport of peptides (*tapbp*, *tap1*) into the endoplasmic reticulum. MHC (Major histocompatibility complex) class I molecules (together with *b2m*) bind these antigenic peptides to present them to CD8+ T-lymphocytes. Other genes belonging to TNFSF and IFN groups are also implicated in the antigen presentation signalling pathway (Figure 3). Overall, the results indicated that dNV-VHSV favoured the antigen processing and presentation in relation to wt-VHSV.

#### 2.3.5. Cluster of Differentiation: B-Cell, T-Cell, and Cell-to-Cell Interactions

Cluster of differentiation genes (CD) conform a functional heterogeneous group of genes that have been involved in cell adhesion, B-cell receptor signalling, T-cell receptor signalling, complement and coagulation cascades, and hematopoietic markers (Figure 4 CD, and Table 1). In this study, 14 CD genes have been analysed. Downregulated and non-regulated CD genes (CD, *n* = 5 genes, Figure 4 CD) were found in the wt-VHSV dataset. The downregulated genes by wt-VHSV were *cd103* (a marker of dendritic cells) [37] and *cd79a* (associated with membrane-bound immunoglobulin in B-cells). On the other hand, the non-regulated genes in wt-VHSV were *cd2* (implicated in the adhesion T cell-APC through the *CD58* protein), *cd276* (participating in the regulation of the T-cell-mediated immune response), *cd83* (involved in the regulation of antigen presentation) [38], and *cd163* (exclusively expressed in monocytes and macrophages in humans), which was downregulated by dNV-VHSV. Some of the CD markers were found upregulated by dNV-VHSV (*cd103*, *cd2*, c*d276*, *cd279*, *cd83*) and others upregulated by wt-VHSV (*cd28*, *cd36*, *cd3e*, *cd83*, *cd11*). This fact might indicate that 48 hpi is a too short time to observe adaptive cellular responses against VHSV.

#### 2.3.6. Cytokines: Chemokines and Interleukins

Among the chemoattractant cytokines or chemokines genes (CK, *n* = *32*), 53% were upregulated by wt-VHSV, whereas dNV-VHSV induced the upregulation by 50%. The chemokines showing downregulated or non-regulated fold changes by wt-VHSV were *ccl13*, *ck11*, *ck12a*, *ck12b*, *ck8b*, *cxc*, *ckrg*, *crlp1*, suppressor of cytokine 1 (*socs1*), *socs2*, *socs3*, and *socs7* (Figure 4, CK). In addition, dNV-VHSV induced the downregulation or non-regulation of the chemokines *ccl13* (basal), *ck1*, *ck4a*, *ck5a*, *ck7a*, *ck7b*, *ck8a*, *ck11*, *nilt4*, *socs4*, and *socs5*. Chemokines have different roles in the coordination of the immune response and may promote the activation or inhibition of different pathways (Figure 4 IL, Table 1), and for the most of them their function is unknown on the basis of viral infections.

Previous studies in rainbow trout have shown that recombinant CK1 has an attractant effect for blood leukocytes) [39]. In addition, CK6 is a chemoattractant for mature macrophages from the RTS11 rainbow trout monocyte-macrophage cell line and may also induce interleukin 8 (IL-8), inducible nitric oxide synthase (iNOS), and the CD-18 integrin in these cells, revealing additional immunomodulatory effects [40]. The capacity of trout recombinant CK12 to attract splenocytes has also been reported, establishing that IgM + B cells were one of the target cells recruited [41]. In the present study, CK1, CK6 and CK12 are upregulated by dNV-VHSV in relation to wt-VHSV. Regarding the interleukin genes group (IL, *n* = 20), the transcriptomic profile obtained after wt-VHSV injection was different from that of dNV-VHSV. Wild-type VHSV downregulated important pleiotropic pro- and anti-inflammatory interleukins such as *il1b, il6,* and its related *il6m17* (Figure 4, IL). On the contrary, interleukins *il1*, *il6*, *il6m7, il7*, *il8*, *il11*, *il21r*, *il27* and *nil1* were upregulated by dNV-VHSV. These interleukins have a key role in immune pathways (Figure 4 IL, Table 1) and their upregulation are required for an effective immune response followed vaccination, suggesting that dNV-VHSV could be an effective attenuated live vaccine.

#### 2.3.7. General Transcription Factors

This group of genes (TF, *n* = 10 selected genes, Figure 4) are implicated in important cellular processes and pathways: Jak–Stat signalling (*stat1*, *stat5*), general transcription factors and regulatory elements (*sox* genes), and multipath genes previously mentioned (*traf2*, *traf3*). In the dNV-VHSV-injected trout, all the TF selected genes were upregulated except for *sox5* (non-regulated) and *tfiia* (downregulated). In wt-VHSV, all the genes studied were upregulated except for *traf2*, *traf3*, *sox30*, *stat5* and *stat1*, which were non-upregulated. These data reflected that Jak–Stat signalling pathway is upregulated by dNV-VHSV, which in turn promoted interferon responses leading to an improved antiviral stage.

#### 2.3.8. Complement and VIG Genes

Among complement genes studied (CO, *n* = 6, Figure 4), the trout injected with dNV-VHSV maintained most of the genes non-regulated and only one was upregulated. It is interesting to note that C9 was upregulated by wt-VHSV, whereas perforin (*prf*) was downregulated by the same virus.

VHSV-induced genes (VIG, *n* = 22) were firstly identified by subtractive hybridisation performed in a previous work [42]. We found that NV from wt-VHSV inhibited the upregulation of 50% of all *vig* genes (*b143*, *b160*, *b191*, *b203*, *b225*, *b88*, *gbp*, *vi1*, *vig-1*, *vig-6*, *vig-8*) while these genes were upregulated in dNV-VHSV, except for *b191* and *gbp* (Figure 3, VIG). Viperin (*vig1*) is expressed by mitochondria and is an IFN-inducible protein that inhibits the replication of a variety of viruses [43]. In addition, *vig-2* [44], *vig-3*, *vig-4*, *vig5*, and probably *vig-6* are also induced by interferon [42]. On the other hand, *vig-7*, *vig-8*, *vig-9* have chemoattractant function, *vig-9* also has an apoptotic function, and *vig-10* is related to apoptosis and transcription repression [42].

The present study is the first one describing the trout transcriptomic profile driven by dNV-VHSV. There are scarce studies regarding whole gene effects upon VHSV infection in trout, the most relevant to our study is the characterization of the RNA microarray profile in olive flounder liver after VHSV infection by immersion [20]. Regarding the role of immune-related genes, we found that both wt-VHSV and dNV-VHSV induced an upregulation of hepcidin, (a regulator of the iron metabolism that is implicated in inflammatory processes) in trout similar to the results observed in VHSV infected olive flounder. However, the *irf2* gene was found non-regulated in the wt-VHSV injected trout, which differs from the upregulation found in olive flounder .The differences found between trout and olive flounder could be due to the expression profile of immune response genes in spleen/head kidney, which could be slightly different from those in liver.

Another recent study determined the miRNA expression profile after VHSV infection. Among the immunity-associated target genes of 63 differentially expressed miRNA after VHSV infection of olive flounder, the authors described IL (*il1b*, *il8*, *il10*), *mx*, interferon regulatory factors (*irf3*, *irf5*, *irf7*), TNF (*tnfsf*), and heat shock proteins (*hsp10*, *hsp60*, *hsp70*, *hsp90*) [45]. For instance, *mx* mRNA was regulated by one miRNA (pol-miR-1388-5p) only, for which the highest expression was at 0 hpi and the lowest at 72 hpi. At 48 hpi, the *mx* expression was not modulated by this miRNA. Our data in trout showed a non-regulation of *mx* at 48h after wt-VSHV injection, being coincident with those of miRNA in olive flounder. Maybe in the future, a correlation between miRNA and mRNA expression levels could be established in infected VHSV fish, as it has been done for some miRNAs in primary human osteoblasts from healthy individuals [46].

#### *2.4. Modulated Pathways in dNV- and wt-VHSV-Injected Trout using GSEA*

To define the impact of dNV-VHSV or wt-VHSV when targeting different pathways or gene sets (GS) in head kidney/spleen, we used the Gene-Set Enrichment Analysis (GSEA) program. Briefly, 51 trout GS collection previously defined [25] were used to obtain normalised enrichment scores (NES). The GS were then classified according to their NES values, as follows: (1) upregulation in dNV-VHSV in relation to wt-VHSV (*n* = 19), and (2) no regulation in dNV-VHSV in relation to wt-VHSV (Figure 5). Among the upregulated GS in dNV-VHSV in relation to wt-VHSV (Figure 5A), higher differences were found in those corresponding to "Protein processing in endoplasmic reticulum", "Regulation of autophagy", and "Autoimmune thyroid disease" (> 2 to 4-fold) (Figure 5A, red lines and symbols). Other GSs which showed lower improvement of upregulations with dNV-VHSV were those implicated in anti-viral interferon networks ("Toll-like receptor signalling pathway", "Toll-like receptor wikipathway", "Type II interferon signalling (IFNG)"), inflammation ("TNFa NF-kappa B signalling", "Cytokine inflammatory response pathway"), recognition of nucleic acids ("Cytosolic DNA sensing pathway", "RIG-I-like receptor signalling"), viral- and bacterial-caused diseases ("Hepatitis C", "Influenza A", "Measles", "Herpes simplex infection", "Epithelial cell *Helicobacter pylori*"), and others ("NF-kappa B signalling pathway", "T cell receptor signalling pathway", "Natural killer cell mediated cytotoxicity-K", "Interleukin 5"). The remaining GSs showed no differences between dNV-VHSV and wt-VHSV (Figure 5B). The "Type II IFN signalling (IFNG)" pathway was one of the top enriched GS found in VHSV survivor zebrafish [47], indicating that both species share similar mechanisms to fight VHSV. Other improved pathways in dNV-VHSV such as "Toll-like receptor signalling", "RIG-I-like receptor signalling", "Natural killer cell-mediated cytotoxicity", "Hepatitis C", and "Influenza A and Measles" were among the most targeted pathways in SVCV zebrafish infections [48]. Finally, "NF-kappa B signalling pathway", "Toll-like receptor wikipathway", "Natural killer cell mediated cytotoxicity", "RIG-I-like receptor signalling", "Autoimmune thyroid disease", "Influenza A", and "Herpes simplex infection" were also modulated in trout when injected with thyroid hormone analogues [25]. All the above-mentioned pathways participated in generating resistance to fish viral infections and underlined the importance that their upregulation by dNV-VHSV might have in case this defective virus is used as a potential attenuated live vaccine.

**Figure 5.** Comparison of significant normalised enrichment scores (NESs) of rainbow trout gene-sets (GSs) by Gene-Set Enrichment Analysis (GSEA). X axis: 1, dNV-VHSV. 2, wt-VHSV. Trout GSs were obtained from the KEGG (-K) and WIKI (-W) pathway databases as described before [25]. For comparative purposes, the NESs calculated by GSEA were represented in a diagram by linking the results obtained with lines. For better clarity, the results were not individually identified by GS. (**A**) GS which show upregulation in dNV-VHSV in relation to wt-VHSV: Toll-like receptor wikipathway-W, Hepatitis C-K, Toll-like receptor signalling pathway-W, RIG-I-like receptor signalling-K, Influenza A-K, Type II interferon signalling (IFNG)-W, Measles-K, NF-kappa B signalling pathway-K, TNFa NF-kappa B signalling-W, Herpes simplex infection-K, Epithelial cell *Helicobacter pylori*-K, T cell receptor signalling pathway-K, Cytosolic DNA sensing pathway-K, Natural killer cell mediated cytotoxicity-K, Interleukin 5-W, Cytokine inflammatory response pathway-W. The most upregulated pathways by dNV-VHSV in relation to wt-VHSV were: Red circle, Autoimmune thyroid disease-K; Red square, Regulation of autophagy-K; and Red triangle, Protein processing in endoplasmic reticulum-K. (**B**) GS which showed no regulation in relation to wt-VHSV: Interferon type I-W, Interferon alpha beta signalling-W, Apoptosis modulation by HSP70-W, B-cell receptor signalling pathway-W, EGFR1 signalling pathway-W, MAPK signalling pathway-K, Interleukin 6-W, Ubiquitin mediated proteolysis-K, TSH signalling pathway-W, Antigen processing and presentation-K, MAPK signalling pathway-W, Interleukin 2-W, HTLV-K, androgen receptor signalling-W, TP53 network signalling-W, AHR pathway-W, Interleukin 3-W, Hematopoietic cell lineages-K, T-cell receptor pathway-W, JAK–STAT signalling pathway-K, PI3K–AKT signalling pathway-K, FGF signalling pathway-W.

#### **3. Conclusions**

The results presented in this study support the hypothesis that dNV-VHSV can be considered an attenuated virus and a potential live vaccine, based on the fact that many critical host gene pathways are activated upon infection. On the contrary, NV expressed at first stages of infection by wt-VHSV modulates the expression levels of interferons, VIG, chemokines, CD, transcription factors, and other immune-related genes, leading to an immune unresponsiveness state that interferes with the early innate immune response. Importantly, this work opens new avenues for the use of NV-deleted novirhabdoviruses as a tool to study the regulation of immune pathways in other teleost fish.

**Author Contributions:** Experimental design, E.G.C. and J.M.C.; Methodology, B.C., P.E., J.M.C., and E.G.C.; Formal analysis, E.G.C.; Original draft preparation, B.C. and E.G.C; Writing—review and editing, B.C., P.E., and E.G.C.; Funding acquisition, J.M.C. and E.G.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partially funded by the Ministry of Science, Innovation and Universities of Spain grants AGL2014-51773-C3-3 and AGL2017-85494-C2-2-R, and the APC was funded by Ministry of Science, Innovation and Universities of Spain grant AGL2017-85494-C2-2-R.

**Acknowledgments:** Authors are in debt to Dr. Rosario Fernández-Godino for her helpful comments.

**Conflicts of Interest:** Authors declare that they have 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/).

## *Review* **Current Progress of Avian Vaccines Against West Nile Virus**

#### **Nereida Jiménez de Oya, Estela Escribano-Romero, Ana-Belén Blázquez, Miguel A. Martín-Acebes and Juan-Carlos Saiz \***

Department of Biotechnology, National Agricultural and Food Research and Technology Institute (INIA), 28040 Madrid, Spain; jdeoya@inia.es (N.J.d.O.); eescribano@inia.es (E.E.-R.); blazquez@inia.es (A.-B.B.); martin.mangel@inia.es (M.A.M.-A.)

**\*** Correspondence: jcsaiz@inia.es; Tel.: +34-9-1347-1497

Received: 31 July 2019; Accepted: 19 September 2019; Published: 23 September 2019

**Abstract:** Birds are the main natural host of West Nile virus (WNV), the worldwide most distributed mosquito-borne flavivirus, but humans and equids can also be sporadic hosts. Many avian species have been reported as susceptible to WNV, particularly corvids. In the case that clinical disease develops in birds, this is due to virus invasion of different organs: liver, spleen, kidney, heart, and mainly the central nervous system, which can lead to death 24–48 h later. Nowadays, vaccines have only been licensed for use in equids; thus, the 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. So far, different commercial and experimental vaccines (inactivated, attenuated, and recombinant viruses, and subunits and DNA-based candidates) have been evaluated, with various regimens, both in domestic and wild avian species. However, there are still disadvantages that must be overcome before avian vaccination can be implemented, such as its cost-effectiveness for domestic birds since in many species the pathogenicity is low or zero, or the viability of being able to achieve collective immunity in wild birds in freedom. Here, a comprehensive review of what has been done until now in the field of avian vaccines against WNV is presented and discussed.

**Keywords:** birds; vaccines; West Nile virus; flavivirus; herd immunity

#### **1. Introduction**

Currently, the ecology of many pathogens is changing because of climate warming that is driving vector colonization of new geographical niches. This fact, together with human behavior and global trade, puts human and animal health at risk. An example is the (re)emergence of West Nile virus (WNV) that nowadays is the most worldwide distributed mosquito-borne flavivirus [1,2]. Since the introduction of a lineage 1 WNV strain in the US in 1999, the virus quickly spread, causing hundreds of deaths in humans and horses and a very high avian mortality [1,2]. More recently, the strains of lineage 2 colonized and spread throughout Europe, leading to outbreaks among wild birds [3] and being responsible for up to 1.875 human cases, including 115 deaths in 2018 [4].

WNV is a small (about 50 nm of diameter), spherical, enveloped flavivirus (*Flaviviridae* family) whose genome consists of a single-stranded RNA molecule of positive polarity that encodes three structural proteins and seven non-structural proteins [1]. Up to nine distinct genetic lineages of WNV have been described, with lineage 1 and 2 being the most distributed worldwide, although only a single serotype is recognized [1,5].

Birds are the main natural host of West Nile virus, though humans and equids can also be sporadically infected [1,2]. Hundreds of avian species have been reported as susceptible to WNV, particularly corvids (Corvidae), which can develop high levels of viremia [6,7], and are notable virus amplifiers [7–9], being, thus, important actors in the epidemiology of the virus [10–14]. Both domestic and wild avian species are susceptible to WNV infection and, in some cases, develop a WNV-associated disease that can lead to high mortality, as occurred during the US outbreak where crow populations declined alarmingly [14–17].

Currently, there is no antiviral therapy against WNV, and the licensed vaccines are only for use in equids [1,18,19]. The availability of avian vaccines would benefit bird populations, both domestic (like farm birds and those grown for restocking and hunting activities) and wild ones (mainly endangered species housed in rehabilitation and wildlife reserves, and birds located at recreational facilities like zoos). Avian vaccination may also help to prevent outbreaks and spread, mainly if herd immunity can be induced. Here, a comprehensive review of our current knowledge, about experimental avian vaccination with different candidates (inactivated, attenuated, and recombinant viruses, and subunit and DNA-based vaccines) in domestic and wild birds, is presented.

#### **2. WNV Biology**

#### *2.1. Genome Organization*

The genome of WNV is composed of a single-stranded positive-sense RNA (ssRNA(+)) of about 11 kb in length (Figure 1) [1]. It contains a 5'-cap structure (m(7)GpppAm) that is methylated at the guanine N-7 and the ribose 2'-OH positions of the first transcribed adenine [20] but lacks a 3' polyA tail. The single open reading frame (ORF) is flanked by two untranslated regions (UTRs) with important functions for viral replication [21]. Remarkably, the 3' UTR is a key determinant of WNV virulence, which makes it attractive for vaccine design [22]. The ORF is translated into a polyprotein that is co-translational and post-translationally cleaved by viral and cellular proteases. The structural capsid (C) protein is involved in the nucleocapsid formation by association with the genomic RNA, the M is produced by cleavage of the prM, and the E is involved in receptor binding, viral entry, and membrane fusion [23]. The non-structural NS2B is the membrane anchor and the co-activator of the NS3 viral serine protease. The NS1 is secreted and has been related to replication, virulence, immunomodulation, and pathogenesis [24]. The NS5 exhibits the methyltransferase activity required for capping of viral RNA and is also the RNA-dependent RNA polymerase in charge of genome replication [21]. Replication of WNV is associated with intracellular membranes of the Endoplasmic Reticulum (ER) [25]. Accordingly, NS2A, NS2B, NS4A, and NS4B are multipass transmembrane proteins. The ER is the place for viral replication and particle biogenesis. The newly assembled immature particles are produced by budding into the lumen of this organelle and traffic across the secretory pathway. Viral particles maturate towards infectious virions by proteolytic processing of the prM to render the M protein. This cleavage takes place inside the *trans*-Golgi network and is catalyzed by the cellular protease furin [23]. Maturation converts the spiky immature particles [26] into smooth mature virions [27] that are released from the cell by exocytosis.

**Figure 1.** Genome organization. Schematic representation of the WNV (West Nile virus) genome. See text for details.

#### *2.2. Molecular Classification and Phylogeny*

WNV is a member of the *Flavivirus* genus, within the *Flaviviridae* family. WNV classification was initially based on cross-neutralization reactions, locating it as a member of the Japanese encephalitis virus (JEV) serocomplex. Later on, the molecular phylogeny analyses supported this antigenic classification and revealed the existence of up to nine distinct genetic lineages of WNV (Figure 2), being lineage 1 and 2 the most worldwide distributed [5]. Lineage 2 was restricted to Africa until recently when it was isolated for the first time in Europe from a goshawk in Hungary in 2004 [28]. Since then, lineage 2 strains have been isolated in mosquitoes, humans, and several domestic and wild birds across the continent [29–31]. In any case, despite this genomic variability, there is only a single WNV serotype described, which could facilitate the development of unique vaccines to protect against all WNV genotypes.

**Figure 2.** Phylogram, showing the relationships between the WNV strains. The tree is based on the complete nucleotide sequence of NS5 (except for HU2925/06 and MH327930). Multiple alignment was performed using MUSCLE [32], and a maximum likelihood tree was built using W-IQ-TREE [33]. The tree was visualized with iTOL [34]. Usutu virus was included as an outgroup for tree rooting. GenBank accession is indicated for each sequence. The country of origin and year of isolation is displayed when available. Circles size denotes the percentage of replicates in the bootstrap analysis (1000 bootstrap analyses). The scale indicates 0.1 substitutions/site. Phylogenetic lineages (Lin.) are indicated according to [5]. Genetic lineage 8 was not included in the tree because only partial sequence, not including NS5, is available (KJ131502).

#### *2.3. Antigenic Structure*

Mature virions are about 50 nm in diameter, and the majority of their surface is occupied by the E glycoprotein (Figure 3A). This external protein shell is composed of 180 copies of E protein arranged as antiparallel homodimers and confers the virions a herringbone T = 3 pseudo-icosahedral symmetry [27]. The E protein is N-glycosylated at Asn 154 in mostWNV isolates (Figure 3B). This surface glycoprotein constitutes the major target for neutralizing antibodies, becoming the base of many vaccine candidates [35]. While the lack of glycosylation influences WNV replication in experimentally infected chickens, it does not compromise the induction of antibodies [36]. Notably, the E protein carries both flavivirus cross-reactive and WNV-specific epitopes. The cross-reactivity between WNV and related flaviviruses is the result of the high degree of structural homology between them and can lead to cross-protection but also to adverse effects due to antibody-dependent enhancement of infection [37–40]. This high cross-reactivity also complicates the precise serological diagnosis of flavivirus infections by immunological techniques, such as ELISA, making necessary the use of confirmatory tests, including related flaviviruses, with the neutralization assay as the gold standard [19]. The E glycoprotein is organized into three domains (DI to DIII), DI is an eight-stranded β-barrel, DII contains the conserved fusion loop (residues 98–110), and DIII adopts an immunoglobulin-like fold form (Figure 3B). Antibody epitopes have been identified in all three domains, with the most prominent neutralizing antibodies targeting DIII, making it an interesting candidate for vaccine development [35]. Antibodies against proteins other than the E have also been identified, so that experimentally infected chickens elicited antibodies against prM and NS1 [41,42]. While antibodies against NS1 have been related to protection in mammals [43], results obtained with red-legged partridges (*Alectoris rufa*) suggest that this could not be always the case [44].

**Figure 3.** Structure of WNV. (**A**) Cryo-electron microscopy reconstruction of a WNV particle (Protein Data Bank accession 3J0B). E monomers are blue, purple, and turquoise. (**B**) Ribbon diagram of the crystal structure of WNV E glycoprotein (Protein Data Bank accession 2HG0). Domain I is red, domain II is yellow, domain III is blue, and the fusion loop is green. The N-linked sugar moiety of Asn 154 is also displayed. Images were produced using Chimera package [45].

#### **3. Avian Susceptibility**

Birds are the main vertebrate hosts for WNV, being commonly infected and frequently developing high levels of viremia [6]. Susceptibility of the avian population to the infection can vary depending on the species, being Corvidae (order Passeriformes) the most susceptible family [46–48], and important virus amplifiers [7–9], that play a key role in the epidemiology of the virus [10–14]. In fact, WNV epidemics in the US were associated with high crow mortality, driving to a significant decrease of native crow species [14–17]. According to the Centers for Disease Control and Prevention (CDC), birds from almost 300 different species have been found dead since 1999 in the US [49]. This avian WNV-associated mortality has been reported around the world in domestic [50–52] and wild birds [11,17,53,54], including endangered species [55,56], as well as in ones adapted to human environments [16].

Differences in pathogenicity, virulence, viremia, the clinical course of the infection, and mortality after experimental infections of birds withWNV strains of either lineage 1 or 2 have been reported [57,58], although no differences have been observed by other authors [8,59].

Main transmission route in birds is by mosquito bites, but other sporadic routes have also been described, such as oral [7,60,61] and bird-to-bird contact [7,13,62–65], suggesting that WNV-infected birds can be a source of contamination in nature [46,64,66].

A great range of viremias has been reported in different species, which may influence viral transmission. Birds that develop viremia greater than 106 pfu/mL are usually considered competent reservoirs to spread the virus [67], although, for some vectors, it has been described in the range of 104–105 pfu/mL [68]. In fact, while in some species (Columbiformes and Galliformes), viremias are quite low, in others (Passeriformes, Charadriiformes, or Strigiformes) are high, making these species more efficient competent hosts for WNV transmission [7]. Viremia can be detected as soon as one day after infection in high susceptible species [7,69–71]. Moreover, WNV has been detected in blood as early as 30–45 min after the bite of infected mosquitoes, suggesting that local replication is not necessary in birds for the primary viremia [72]. Viremia can last up to 7–11 days depending of the avian species [7,59]. Dissemination of the virus to the different tissues has been reported as early as one day after infection in the spleen of crows [73], until 14 days post-infection in kidney and spleen of an American Kestrel (*Falco sparverius*, Falconidae) [7], and even 27 days after infection in the kidney of a horned owl (Strigidae) [74]. WNV can also be detected in oral and fecal swabs from the first day after infection in most of the susceptible species studied with a viral shedding timing that overall reflects that of viremia [8,70,75].

#### **4. Avian Pathology**

No clinical signs are observed in most WNV-infected birds, and, when they show up, the most common are lethargy, reluctance to move, ruffled feathers, and lack of appetite with marked body weight losses (Figure 4) [7,8,63,75]. Dehydration [70], intermittent head twitching [70], convulsions [47,76], profuse oral and nasal discharge [77], or reduced fecal output [78] are less common. When a fatal outcome occurs, it happens within the first 24 h after the onset of clinical signs [7,8].

**Figure 4.** Clinical signs observed in experimentally infected birds. Magpies (upper panel) and partridges (lower panel) experimentally infected with WNV. Birds that die develop high morbidity hours before their death. Evident clinical signs like loss of appetite, ruffled feathers, paralysis, hunchback position, and unresponsiveness are observed in WNV-infected birds from 3 to 8 days post-infection (dpi).

Macroscopic lesions are observed in infected birds between 7 and 10 days after infection, although it can be delayed until 21 days post-infection (dpi) [79], and even become chronic [47]. The most affected organs are the brain, presenting encephalitis with cerebellar involvement, heart, liver, spleen, and kidney [75,80,81]. Lesions, such as diffuse pallor or pale foci in epicardium and myocardium [70], or in the hepatic, splenic, and renal parenchyma [80], as well as hepatomegaly and splenomegaly [58,74,75], have also been reported.

Among the histopathological findings of the affected organs (Figure 5), central nervous system lesions are mainly characterized by hemorrhages in the brain [81–83], mild perivascular cuffs consisting of lymphocytes and plasma cells, scattered individual necrotic neurons, lymphoplasmacytic, and histiocytic meningoencephalitis characterized by gliosis and glial nodules [47,48,59]. The main findings in the heart are lymphoplasmacytic and histiocytic myocarditis with myocardial necrosis, concurrent fibrosis, sometimes with thrombi, hypereosinophilia of cardiomyocytes, myocytolysis, nuclear swelling, pyknosis, loss of striations, myofiber degeneration, and hemorrhages [70,80,84]. Liver lesions include multifocal randomly distributed granulomatous and lymphohistiocytic hepatitis, with mild to moderate coagulative hepatocellular necrosis and deposition of fibrin [75]. The spleen is also affected by WNV infection, where multifocal lymphocytic necrosis occurs characterized by the presence of karyorrhectic nuclear debris [75,81]. Significant histopathological abnormalities present in the kidney are mild multifocal proximal tubular necrosis and mild to moderate lymphoplasmacytic interstitial nephritis that can occasionally be perivascular [70,75,85]. Ocular lesions are also common in WNV-infected birds. These lesions consist of the disarray of the retinal pigmented epithelial cell layer, pectenitis, choroidal or retinal inflammation, cellular necrosis, muscular degeneration in the iris, mild optic neuritis, impaired vision, and even blindness [47,76,86–88]. Other less common described lesions include pancreatitis, pulmonary edema, infiltration of lymphocytes, plasma cells and histiocytes in the intestinal tract, necrotizing mucosal duodenitis, myofiber degeneration with lymphoplasmacytic inflammation, and fibrosis in skeletal muscle [89,90].

**Figure 5.** Histopathological findings in WNV-infected red-legged partridge. (**A**) Moderate gliosis, and lymphoplasmacytic and histiocytic perivascular cuffs observed in the brain. (**B**) Extensive myocardial degeneration and necrosis with inflammatory infiltrate composed of lymphocytes, plasma cells, and histiocytes observed in the heart. (**C**) Extensive liver necrosis with predominantly lymphoplasmacytic infiltrate. Images courtesy of Dr. U. Höfle and Dr. V. Gamino.

WNV can persist in the organs of infected birds up to several months [91], thus playing a possible role in viral overwintering and enabling possible new infections through mosquito bites or bird-to-bird transmission [14,92,93].

#### **5. Vaccines**

Vaccines to protect humans against certain flaviviruses have been available for long time, such as that against yellow fever virus (YFV) in use since more than 80 years, or that against Japanese encephalitis virus (JEV) approved in 2009, and, thus, it is expected that the same principles could be applied to WNV vaccine development. In fact, several commercial formulations are currently available for equid vaccination, and its effectiveness was demonstrated after immunization of horses, which led to a marked decrease of severe WNV disease (WND) in the following years in the US [94,95]. In many cases, experiments with birds have tested some of these commercially available vaccines approved for use in equids [18,19], such as the formalin-inactivated whole-WNV vaccine originally developed by Fort Dodge (Fort Dodge, IA, USA), which has been commercialized under different names (West Nile-Innovator, Duvaxyn® renamed EQUIP WNV®) [96–104], and was licensed in 2003 and subsequent years (Vetera®), a DNA-based vaccine subsequent formulation expressing the prM and E WNV proteins also from Fort Dodge (West Nile-Innovator DNA equine®) [97,105], which was licensed in 2004 in the US but later removed from the market in 2010, and a recombinant live canarypox virus vaccine (ALVAC®) that expresses the prM/E WNV proteins (Recombitek® Equine West Nile virus in the US, Merial, GA, USA; or Proteq WNV® in Europe) [96,105].

Additionally, experimental prototypes have been assayed, like a chimeric virus based on the yellow fever 17D vaccine strain in which the surface proteins were replaced by that of WNV (ChimeriVax-WN) [106], or a vaccine [64] based on WNV recombinant subviral particles (RSPs) produced by a HeLa-3 cell line stably transfected with a plasmid encoding the signal peptide of the C protein and the prM/E proteins [107]. Several other approaches have used DNA vaccines, like the DNA-plasmid vaccine (pCBWN) [108], also encoding the WNV prM/E proteins [98,105,109–111], and a modified version of it [112]. Another DNA vaccine that codes for the prM/M and E proteins of WNV produced by Aldevron [99], and two DNA-plasmid vaccines expressing the ectodomain of the WNV E protein of lineage 1 or 2 in the modified backbone vector pVax1 were also tested [113]. Likewise, a recombinant protein vaccine originally developed for humans, the WN-80E, consisting of a portion of the WNV envelope protein has been assayed too [114]. All these vaccines, commercial and experimental, have been evaluated in different domestic (Table 2) and wild (Table 1) avian species following different routes of administration and vaccination regimens, resulting in varied outcomes.

The availability of vaccines for use in birds, the natural hosts of the virus, will be highly useful, mostly during outbreaks. These vaccines could be used in birds held in captivity in recreational installations and zoos, in wildlife rehabilitation and endangered species breeding centers, and even in birds grown for restocking or hunting activities that are yearly released by the thousands into the environment in many countries. Even more, some of these vaccines could be also administered during surveillance programs [64].

birds.

**Table**

**1.**

Vaccines

tested

in

wild

#### *Vaccines* **2019** , *7*, 126


**Table 1.** *Cont.*

*Vaccines* **2019** , *7*, 126



vaccination and eventually recaptured; Duvaxyn® commercial formulation used; RSP: recombinant subviral particles; WNV: West Nile virus; SC: subcutaneous; IM: intramuscular; OR: oral; IM+EP: intramuscular + electroporation; NC: not commented; NR: not relevant; NA: not applicable; NT: not tested; NAI: not in all individuals; L: low titer; Break of competence: vaccine lowers viremia levels below the threshold of competence (see text).

#### *Vaccines* **2019** , *7*, 126

#### *5.1. Vaccination in Domestic Birds*

The first description of disease and deaths caused by WNV in domestic birds was reported in 1997–1999 in Israel [115], involving hundreds of young geese (*Anser anser*). This species had also been the most affected domestic avian species during virus spread in the US [116]. Symptomatic infections have also been reported in several Psittacine species [90], although experimental infection of birds of these species indicated that they are not very competent [7]. Galliformes, such as chickens (*Gallus gallus domesticus*) and turkeys (*Meleagris gallopavo*), seroconvert but remain asymptomatic. On the other hand, outbreaks among farmed chukar partridges (*Alectoris chukar*) and Impeyan pheasants (*Lophophorus impejanus*) have been reported [52].

After the initial outbreaks in geese in Israel mentioned above [115], both live attenuated and inactivated WNV vaccines have been successfully used there. A live attenuated WNV vaccine was generated by serially passaging a WNV Israeli isolate in a mosquito cell line and selecting an escape mutant using a specific monoclonal antibody [117]. The resulting variant, WN-25A, lost all neuroinvasiveness, while it fully protected geese (20/20) upon a lethal challenge with an Israeli strain isolated from a moribund goose. Later, an attenuated, commercial heterologous flavivirus vaccine derived from Israel turkey meningoencephalitis virus (TMEV) was experimentally tested in laboratory and field settings in geese intra-cranially challenged with WNV two weeks after immunization [115]. The level and duration of protection achieved were quite high and long-lasting (71–93%, 12/17–14/15, protection in laboratory assays, and 39–72%, 9/23–18/25 in the farm ones); however, some goose flocks reacted unfavorably to the vaccination in field trials, showing neurological signs and appreciable mortality. Such undesirable side effects were not observed when they tested a formalin-inactivated WNV strain passaged in suckling mice brains [115]. The same authors reported a 63% (5/8) protection upon intra-cranial challenge when a double dose of this prototype was administered in a single injection and up to 94% (15/16) when a single dose was administered in two injections spaced two weeks apart. Similar studies carried out in farmed goose flocks resulted in 52–80% (13/25–16/20) protection [115]. The efficacy of this vaccine was later evaluated in 829 geese, 298 laboratory-vaccinated, 231 farm-vaccinated, and 300 non-vaccinated, showing 86.58% (258/298), 75.32% (174/231), and 8.33% (25/300) survival rates, respectively, after WNV challenge [118].

Subsequently, an inactivated vaccine was developed using an adapted WNV-Isr98 isolate highly virulent for geese and the PER.C6® cell line platform [119]. When the vaccine was administered with mineral oil as an adjuvant to geese, 91.4% (53/58) survived to the infection, while only 5% (1/20) of the control PER.C6 sham-vaccinated group did. However, the PER.C6-ISR98 candidate did not seem to be sterilizing since, after the challenge, a boost of neutralizing antibodies was detected. In addition, the importance of the use of adjuvants was noted since the administration of the vaccine without adjuvant resulted in 53.3% (32/60) protection [119].

By 2011, the inactivated West Nile-Innovator vaccine was tested for its capability to induce antibodies in chicks and adult thick-billed parrots (*Rhynchopsitta pachyrhyncha*) that received five and three doses with annual boosts along 3 and 7 years, respectively [104]. None of the birds seroconverted after the initial injections, but 2/4 and 3/4 of the chicks developed antibodies 1 and 2 years later, respectively, while only 1/12 and 2/8 of the adults had them 1 and 3 years later, being 6/6 positive after 7 years of annual vaccination. However, as the birds were likely naturally exposed to WNV during the experiment, the interpretation of the results is complicated.

Chimeric vaccines have also been evaluated in domestic birds. So that, an attenuated chimeric vaccine constructed by inserting the prM/E of WNV in dengue virus serotype 4 backbone (WNV/DENV4), and a similar one with a 30-nucleotide deletion in the 3 non-coding region of DENV4 (WN/DEN4-3 Δ30), which were previously shown to prevent viremia in challenged mice and rhesus macaques [120], were tested in young domestic geese [121]. None of these chimeric vaccines stimulated protective immunity against WNV challenge, and high morbidity rates (3/4 in each group), and a high level of viremia were recorded among vaccinated goose, similar to that in non-vaccinated animals [121].

A different approach was used by testing, in domestic geese, aWNV subunit vaccine that comprised 80% of the E protein (WN-80E) combined with adjuvant and administered twice 4 weeks apart [122]. Using viremia as the clinical endpoint, no virus was detected in the serum of groups of six birds immunized with medium or a high-dose (5 or 10 μg) of the vaccine up to 14 dpi. However, the virus was detected in oral swabs 3–6 dpi in some of the birds, and an increase in antibody titers was observed at 14 dpi, indicating that the vaccine did not induce sterile immunity. Likewise, using a recombinant WNV-E as immunogen to orally (20 μg or 100 μg/dose), or intramuscularly (20 μg/dose), vaccinate Leghorn chickens (*G. gallus domesticus*) three times with a 2-week interval [123], it was shown that, in the birds immunized intramuscularly, the levels of viremia were lower and the total production of WNV E protein-specific IgY was significantly higher than in the animals immunized by the oral route. In this line, a recombinant WNV envelope E (rE) protein produced in insects [124], highly protective in mice [125], was assayed in red-legged partridges. Birds were intramuscularly vaccinated twice at the two-week interval with 10 μg/animal of the rE protein administered with adjuvant, and a control group was similarly sham-immunized. Partridges from both groups were subcutaneously challenged with the NY99 WNV strain [63]. All the rE vaccinated birds (22/22) survived to WNV infection, while 33.3% (6/18) of the sham-immunized partridges succumbed between 3 and 8 dpi, being the mortality rate higher among younger (9 weeks of age at the time of challenge) than among older (13 weeks of age) animals (45.5% vs. 14.3%, respectively). An age-dependent susceptibility had been previously reported in a related species, the chukar partridges, in which 25% mortality was observed in juvenile birds, while no mortality was reported in 14 week-old partridges housed nearby during a WNV outbreak in the US [52], and also in domestic geese [51,126]. Analyses of the humoral immune response elicited by rE vaccinated partridges showed that all animals were ELISA positive after two immunizations [63], similar to what had previously been described in geese and American crows (*Corvus brachyrhynchos*) [98,119]. Again, the immunity elicited by rE vaccinated partridges was not sterilizing, as viremia was detected in 4/22 vaccinated birds 3 dpi, and antibody titers significantly increased after viral challenge to levels similar to those found in non-vaccinated animals. Similar results had been observed after vaccination of geese, red-tailed hawks, and Western scrub-jays (*Aphelocoma californica*) [105,110,119].

Another study tested five different vaccine candidates administered intramuscularly in 47 geese [116]. The prototypes tested were an inactivated whole virus prepared with mineral oil as an adjuvant, three recombinant viruses containing the WNV prM/E (the canarypox viruses ALVAC vCP2017 and vCP2018; and the fowlpox virus vFP2000), and an exogenously produced WNV E protein. Birds were challenged 2 weeks after the booster immunization, except in the last case that was done after 1 week. Since no geese died in the challenged sham-immunized groups and only one developed clinical signs, protection was measured through the following five parameters: clinical pathogenicity index (CPI), plasma virus-positive geese on days 1–4 post-inoculation, plasma virus titers, brain histological lesion rates, and severity scores. The best protection was achieved with the vFP2000 fowlpox virus, which was the one that best scored in the five parameters, followed by the vCP2018 canarypox virus that did it in four, the vCP2017 in three, the E protein in one, and the oil-emulsion WNV in none.


#### **2.** Vaccines tested in domestic birds.

**Table** 

*Vaccines* **2019**, *7*, 126

NA: not applicable; NT: not tested; NAI: not in all individuals; L: low titer; Break of competence:

 vaccine lowers viremia levels below the threshold of competence (see text).

#### *5.2. Vaccination in Wild Birds*

Vaccination of wild species presents associated problems, such as the limited access to individuals, aggravated by the high number of susceptible species described, and environmental safety concerns, especially with attenuated or genetically engineered live virus-based vaccines. However, since, in many of them, WNV-related pathogenesis is not clinically relevant and/or they have a limited, if any, role in virus ecology, most of them do not seem to represent a target for vaccine campaigns implementation. Indeed, most efforts in experimental vaccine development have selected members of the Corvidae family as model, although raptors (Accipitridae and Falconidae), nocturnal bird preys (Strigidae and Tytonidae), and members of other families (Phoenicopteridae, Spheniscidae, Gruidae, Turdidae, Cathartidae, Phasianidae, and Anatidae) have also been used (Table 1). For most of them, WNV-associated mortality has been described [8,74,128], and some have been related to virus spreading and re-introduction in different geographical areas due to their migratory behavior.

The main aim of a vaccine is conferring protection. A single intramuscular dose of the pCBWN DNA vaccine administered to fish crows resulted in 100% (8/8) survival rate in comparison with the 50% recorded in non-vaccinated (5/10), or orally immunized (4/8) birds [111]. However, the same vaccine intramuscularly administered to American crows resulted in 44% (4/9) survival, while none (0/10) of the sham-inoculated birds survived [98]. The same authors reported up to 60% (6/10) increased survival rate when intramuscular immunization was performed with an adjuvant, a lack of protection with the adjuvant alone (0/8), or when the oral route was used (0/10), and a low one (11%, 1/9) when the West Nile-Innovator vaccine was intramuscularly administered [98]. Another study reported that a single intramuscular vaccination of Western scrub-jays with the pCBWN vaccine protected 100% (3/3) of the birds, and that 80% (4/5) of the corvids intramuscularly vaccinated with a single dose of the Fort Dodge West Nile-Innovator DNA equine® vaccine or the Recombitek® Equine West Nile virus formulation also survived to the infection compared to the 40% (2/5) of survival reached by the control group [105]. Dispensation to falcons of the Duvaxyn® inactivated vaccine resulted in 80% (4/5) and 100% (4/4) protection when administered twice or thrice, respectively, and 100% protection with two (5/5) or three doses (4/4) was achieved with the Recombitek® Equine West Nile virus formulation, while only 50% (4/8) survival was recorded in non-vaccinated animals [96]. Two DNA vaccines that express the ectodomain of the E protein of WNV of lineage 1 or 2 also tested in large falcons conferred protection against a WNV challenge and, based on their results with different protocols, the authors indicated that protection was dependent on the lineage, regimen, and way of administration used. Birds intramuscularly immunized with the plasmid, encoding the protein of lineage 2, reached 100% (5/5) survival in comparison to the 60% (3/5) reached by those immunized with that of lineage 1, or the 50% (4/8) showed by non-vaccinated birds [113]. Finally, a single intramuscular dose of an RSP-based vaccine protected magpies (*Pica pica*) as 71.4% (5/7) of the vaccinated birds survived to viral challenge, compared to the 22.2% (2/9) survival rate observed among sham-immunized magpies [64].

Remarkably, in one study [106], vaccination of fish crows with the chimeric ChimeriVax-WN resulted in a high mortality rate upon WNV challenge when compared with non-vaccinated birds (25%, 2/8 and 0%, 0/4, respectively) that was accused to a possible antibody-dependent viral enhancement effect, although such effect has not been observed in nature.

Reduced morbidity and pathogenicity were also observed upon experimental vaccination of wild birds. Thus, a reduced pathology was noted in sandhill cranes (*Grus Canadensis*) vaccinated with the Fort Dodge inactivated vaccine [103], clinical signs alleviations were also reported after administration of three doses of Duvaxyn® or Recombitek® Equine West Nile virus vaccines in large falcons [96], and reduced clinical scores and antigen deposition in their organs upon DNA vaccination were also documented [113]. Likewise, a less pronounced and shorter loss of weight and a lack of clinical signs were described in RSP-vaccinated magpies that survived to viral challenge [64].

#### *5.3. Sterilizing Immunity*

As commented above, most of the tested vaccines conferred protection when animals were challenged, but, in none of them, induction of sterilizing immunity was observed, as viral replication could be detected in vaccinated birds, although with the exception of one study performed in vaccinated Western scrub-jays in which no reduction of viral titers was reported [105], those were usually lower in vaccinated than in sham-immunized animals [64,96,98,103,109–111,113]. For instance, viremia was detected in 4/22 rE vaccinated partridges 3 dpi, and antibody titers increased significantly after viral challenge to levels similar to those found in non-vaccinated animals [63].

On the other hand, in many cases [64,96,103,109,113], viremia reached levels below what is considered necessary to be a competent reservoir [7], although, for some vaccines, two [113] or three [96,103] doses were required to achieve it. In this respect, it was reported that for *Culex pipiens* and *Culex quinquefasciatus* [129], two vectors considered key to virus maintenance, these levels must be above 105 pfu/mL, although lower viremia levels (around 104 pfu/mL) also seemed to be relevant for other vectors, such as *Culex univittatus* or *Culex perexiguus* [68]. Additionally, a boost of antibodies in vaccinated birds was usually observed after viral challenge [64,103,105,110].

#### *5.4. Immunogenicity*

Induction of antibodies prior to challenge has not always been detected [101,103,109] and, if so, they are present at low levels [96,100,105,110,112,113], and not in all vaccinated individuals [64,97–100,102,106,110,111,113]. As commented above, humoral immune response priming that induces an enhanced and prominent/lasting antibody production has been described after viral challenge in vaccinated birds [64,103,105,110]. Although an association between antibody induction and protection has been suggested [64,98], sometimes, as no challenge was conducted, this hypothesis could not be confirmed [97,99–102,112,114]. For instance, a non-replicating recombinant adenoviruses, expressing either the WNV envelope (rAdE) or the NS3 (rAdNS3) proteins, was assayed in Japanese quails (*Coturnix japonica*) [127], but, since no challenge was performed, the efficacy of the vaccines was measured in terms of WNV-specific antibodies levels and T cells specific activation, which were both increased in vaccinated birds compared to unvaccinated controls. This antibody response was higher and more robust with the rAdE candidate than with the rAdNS3, probably due to the expression of the entire E protein on the cell surface, thus allowing B cells to bind to any available epitope on the molecule. Even more, vaccination with rAdE triggered the activation of more WNV-specific CD4<sup>+</sup> T cells, which would be required to fully activate the WNV primed B cells to produce antibodies. In line with this, vaccines based on DNA and live vectors favor the availability of cytotoxic T lymphocytes (CTLs) epitopes and therefore, should improve protection after viral challenge if T-cell responses are important. One work performed in large falcons compared the efficacy of the inactivated Duvaxyn® and the live vector-based Recombitek® Equine West Nile virus vaccines, showing a slightly better protection of the later one, as mentioned above [96], and, thus, pointing to a protective role of the cellular immune response. However, another study conducted in American crows with the pCBWN and the inactivated West Nile-Innovator vaccines showed the opposite results, as survival rates were 44% (4/9) and 11% (1/9), respectively [98]. Moreover, many vaccines tested included adjuvant, which could favor antigen presentation to T cells and prolong the presence of viral antigens for B cell recognition. So that, the pCBWN vaccine provided microencapsulated in sodium alginate [111], or with aluminum [98,110], increased the survival rates of immunized and challenged American crows when compared with those which received the vaccine alone (44%, 4/9 and 60%, 6/10, respectively) [98]. Aluminum has also been used with DNA [99] and RSP-based vaccines with good results [64].

#### *5.5. Herd Immunity*

Horizontal transmission of WNV in experimentally infected birds was early described [7]. This can be due to direct contact or by fecal-oral route since the virus can be detected in cloacal and oral

swabs [7,8,61,128] and in feathers of infected birds [8]. Moreover, viremia levels reached in the absence of mosquito-borne transmission can be as high as those obtained by mosquito natural exposure [7]. Several vaccines have demonstrated to be effective in diminishing the risk of this type of transmission as they were able to either shorten [103,110] or reduce viral shedding [96,110,113] and virus presence in feathers [8]. Even more, it has been reported that RSPs vaccination completely broke horizontal transmission, as none (0/4) of the contact birds housed with challenged-vaccinated magpies got infected in contrast to 50% (3/6) that did it when were housed with challenged, unvaccinated cage-mates [64]. These data point to the induction of herd immunity through bird vaccination.

On the other hand, transmission in the absence of mosquito-borne infection has also been reported by the consumption of birds infected with WNV by scavenger species [60,128,130]. Therefore, and even in the absence of sterilizing immunity, reduction of viral load in organs after vaccination is desirable. In fact, reduction of viral load in the organs of challenged falcons vaccinated with commercially available WNV vaccines, such as Duvaxyn®, Recombitek® Equine West Nile virus [96], or with DNA-plasmid vaccines encoding the ectodomain of the E protein, has been reported [113].

#### *5.6. Side E*ff*ects and Feasibility*

An additional important point that must be taken into account for vaccine implementation is the lack of undesirable side effects. Even though local inflammation at the site of administration has been observed, probably due to hypersensitivity reactions to the vaccines or the natural effect of the adjuvants employed in some DNA-based and RSP-based vaccines [64,113], in most cases, no such side effects have been described. Two commercial Fort Dodge vaccines designed for equids (virus inactivated and DNA) showed no adverse side effects in corvids, cranes, or falcons [96,103,105]. The inactivated formulation has also been assayed in flamingos, hawks, eagles, vultures, owls, penguins, and wild chickens [99–102], showing good safety profiles, although, since no viral challenge was done, the immunopathological effects related to the vaccine during the infection were not evaluated. Moreover, vaccines based on live vectors can provoke adverse effects, such as the unexpected increase in mortality observed in corvids vaccinated with the ChimeriVax-WN [106], the development of necrotic lesions also in corvids [105], the massive local inflammation reported in falcons immunized with the Recombitek® Equine West Nile virus formulation [96], or the neurological signs and appreciable mortality observed in geese vaccinated with the heterologous TMEV-based vaccine candidate [115]. Even more, vaccines should avoid any environmental effects. In this regard, shedding by the fecal-oral route of vaccines based on virus or plasmid was not found in falcons immunized with Recombitek® Equine West Nile virus [96], or with DNA-plasmids vaccines expressing WNV proteins of lineage 1 or 2 [113].

As commented above, the biggest drawback for implementation feasibility of wild bird vaccines is access to the target host. This could be bypassed if herd immunity can be established, preferably by oral vaccination in, for example, feeding stations, which has already been useful for controlling other zoonotic diseases, such as rabies [131]. However, so far, experimental vaccination of birds by the oral route has failed in conferring protection [98,111], and it has not even able to stimulate the production of antibodies. In any case, avian vaccination can be a realistic option in specific situations, such as in birds grown for restocking activities, endangered species in captive breeding projects, wildlife reserves, recreation installations, or during epidemiological surveillance programs.

#### **6. Conclusions**

The objective of any vaccine is the induction of protection that, preferably, should be long-lasting and sterilizing, and induce herd immunity. Experimental vaccination with different formulations (attenuated, inactivated, recombinant viruses, and subunits and DNA-based candidates) has been assayed in domestic and wild birds from different species and ages following different routes of administration and regimens, which has resulted in varied outcomes. Even though, due to logistical and ethical concerns, among others, the number of birds included in the studies has generally been rather low, mainly when wild birds were used, the reported data indicate that, overall, vaccination induces humoral and, more probably, cellular responses, and reduces WNV-associated disease, lesions, viremia, viral shedding, and, more significantly, mortality. However, no sterilizing immunity has been observed, induction of antibodies has not always been recorded, and, if detected, it was not always in every bird. Remarkably, when evaluated, no horizontal transmission from challenged-vaccinated birds has been observed, pointing to the induction of herd immunity that would prevent virus maintenance in the environment and, thus, its spread. Nevertheless, the implementation of bird vaccines faces several drawbacks, such as the difficult feasibility of access to the target host, mainly for wild species, as well as the administration route, as oral, the most feasible one, has failed to confer protection. In any case, the availability of effective avian vaccines against WNV would be very helpful, mainly during outbreaks, and therefore, research should go on.

**Author Contributions:** Conceptualization and resources J.-C.S.; writing—original draft preparation, N.J.d.O., E.E.-R., A.-B.B., M.A.M.-A., J.-C.S.; writing—review and editing, J.-C.S.; funding acquisition, J.-C.S.

**Funding:** This research was partially funded by Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria, (INIA) grants RTA-2015-00009-00-00 and E-RTA-2017-00003-C02-01, and Comunidad Autónoma de Madrid grant S2018/BAA-4370 (PLATESA2-CM).

**Acknowledgments:** Authors are in debt to Dr. Ursula Höfle and Dr. Virginia Gamino for kindly providing some of the images.

**Conflicts of Interest:** The authors declare no conflict 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/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Vaccines* Editorial Office E-mail: vaccines@mdpi.com www.mdpi.com/journal/vaccines

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18