Skip to main content

West Nile Integrated Shield Project: Epidemiology, Diagnosis and Prevention of West Nile Virus in Europe

Final Report Summary - WINGS (West Nile Integrated Shield Project: Epidemiology, Diagnosis and Prevention of West Nile Virus in Europe)

Executive Summary:
In recent years, outbreaks of West Nile Virus (WNV), a mosquito-transmitted pathogen, are increasing in number and geographic distribution in Europe. The virus is now considered to be endemic to several European countries (including Italy, Greece and Hungary). In addition, novel WNV strains with higher virulence are emerging. Although WNV causes no or only mild symptoms in most of the cases, it can also lead to severe and even fatal neurological disease, including encephalitis and meningitis. Mostly older or immunocompromised persons are at risk. Although veterinary vaccines for WNV exist, there is no vaccine to protect humans from disease. In addition, WNV infections are difficult to specifically detect, because the virus is present in the blood only for a short time, and WNV-antibodies display a high degree of cross-reactivity to similar viruses, e.g. dengue or tick-borne encephalitis viruses. Especially in areas where different such viruses co-circulate (as is the case for many areas in Europe), this is of concern. Therefore, techniques for the prevention and the improved, specific diagnosis of WNV infections, such as new vaccines and diagnostic test systems, have to be developed and were the major objectives of the WINGS project.
During the three years of funding, the consortium successfully developed several candidate vaccines consisting of either recombinant proteins, or DNA, or a mixture of both. The most promising vaccine candidate showed long-lasting complete protection in mice and non-human primates. The vaccine was well tolerated, protective against different circulating European WNV-strains and displayed higher im-munogenicity and protective capacity than previous developments in this field. In addition, as it is based on a recombinant protein expressed in bacteria, it can be produced in high yields.
Several tools for the specific diagnosis of WNV infections have been developed. A novel method enables the detection and isolation of WNV from urine samples. This is a significant advantage to existing proto-cols because of its higher sensitivity than testing of blood samples. The protocol was distributed to hos-pitals and surveillance institutions all over Europe and will greatly help to facilitate WNV diagnostics and epidemiological studies. In addition, test systems for the detection of WNV-antibodies were devel-oped. These are highly sensitive and specific, as they avoid cross-reactivity to other viral infections.
By using these developments as well as state-of-the art diagnostic methods, novel circulating WNV strains in Europe were identified and their genomic sequences determined. Pathogenicity studies and biochemical experiments revealed important insights into the mechanisms which novel WNV strains use to efficiently spread into new areas.
In sum, the WINGS project delivered several novel technologies and tools to specifically detect and effi-ciently counteract WNV and thereby to enhance Europe´s preparedness for this emerging infection.

Project Context and Objectives:
WNV is a zoonotic neuropathic virus, infecting mainly birds but also mammals, including humans, where infection with WNV can lead to neurological symptoms such as encephalitis. WNV is transmitted via mosquitoes. In humans, first signs of WN-disease are observed 2–6 days post infection, but the incu-bation period may extend to 2 weeks. 20% of the infected persons develop flu-like symptoms with high fever. However, in about 1% of the cases, a WNV infection can lead to a severe neuro-invasive disease. Within this group the mortality rate can reach 20%. In particular older and immune-suppressed persons are at risk.
The virus belongs to the Flaviviridae family, which also includes several other clinically relevant viruses including dengue virus (DENV), Yellow Fever virus (YFV), tick born encephalitis virus (TBEV) and Japa-nese encephalitis virus. Among mosquito-borne flaviviruses, WNV has the broadest vector and host range. Flaviviruses have an icosahedral structure of ~50 nm diameter. The virion’s envelope comprises virus-specific glycosylated proteins, surrounding the capsid which contains the viral RNA genome.
Phylogenetic analyses have revealed that most WNV isolates worldwide can be assigned to two major lineages. Lineage 1 consists of isolates from North Africa, Europe, India, Israel, USA and Australia (Kun-jin-virus), whereas lineage 2 comprises isolates from West, Central and East Africa including Madagas-car. The two lineages share 76–77% identity on the RNA level and 93–94% identity on the amino acid level.
WNV is a zoonotic virus circulating in birds. The role of birds in the enzootic cycle of the virus is well documented. Many avian species are susceptible and develop high viral serum titres during the acute phase of infection. During a time window of 4–5 days these WNV titres are high enough to transmit the virus to blood sucking mosquitoes and infect them. Such serum concentrations have been observed in birds only, hence the mammalian hosts such as humans and horses provide a dead end for WNV. How-ever, in rare cases, WNV can be transmitted among humans via blood transfusion or organ transplanta-tion. Transplacental transmission has been documented as well.
WNV has come into the focus of public interest following its introduction into the USA in 1999. Starting from New York City, the epidemic spread over the entire continent and reached countries in Latin America. In the United States, between 1999 and 2010, more than 30,000 human cases were confirmed that required clinical attention. These were associated with more than 1,200 deaths. Based on sero-prevalence studies several million humans have been infected but not diagnosed. Horses were affected as well, and experienced even higher mortality rates than humans. In addition, WNV has led to a dra-matic decrease in the populations of many American bird species.

WNV in Europe
Although the European situation is currently less dramatic compared to the USA, WNV is being detected more frequently over the last five years, especially in southern Europe, where both major WNV lineages seem to be endemic today. Epidemiological data demonstrate the spreading of WNV and imply the oc-currence of novel WNV variants over the last 15 years with increased virulence and epidemic potential both for birds and for mammals.
Most documented cases of WNV infection derive from the veterinary sector, and outbreaks very often affect birds and horses. Several studies demonstrated that birds in Europe, even in Germany and the UK, had WNV specific antibodies or detectable viral RNA in their blood. Recent outbreaks among horses were seen in 1998 in the Tuscany region / Italy and, since 2008, in Northeastern Italy and in the Ca-margue / France (2002 and 2008). Human WNV cases in Europe have up to now clustered in Southern Europe. WNV of the lineage 2 seems to be endemic in Hungary, where between 2003 and 2007 on aver-age six cases of WNV neuroinvasive infection were recorded per year. In 2008, this number increased to 14 by October, and the virus affected the whole country. In the same year, a WNV outbreak among hu-mans was observed in the Emilia Romagna and Veneto regions of Northeastern Italy with 9 confirmed cases of WNV diseases. In 2009, 16 human cases were reported from Italy, showing a larger geographical distribution than 2008. In 2010, major outbreaks occurred in Romania and Greece, caused by pathogenic lineage 2 WNV strains. More than 200 people suffered from neurological symptoms, more than 30 cases were fatal. Also in 2010, WNV cases were reported from Russia, Italy, Turkey, Israel and Morocco. In the following years, case numbers remained high and outbreaks occurred in a wider area, including countries such as Serbia. WNV from lineage 2 was also found in Northern Italy, where previously only lineage 1 strains were circulating. Consequently, co-circulation of both lineages in the same geographical area was demonstrated. In 2012, the highest case numbers since 2003 were reported from the USA, bringing the problem of WNV back to public attention. Reasons for the spread and diversification of WNV still remain to be fully understood.

Only veterinary WNV vaccines are currently on the market. Equine WNV vaccines in the USA led to a significant decrease in the number of severe clinical signs associated with WNV infections among horses, and the first veterinary vaccine was also licensed in Europe. There is no human vaccine available to date, neither is a specific treatment to counteract WNV caused disease. Hence, at the moment the only way to prevent an infection with WNV is to avoid mosquito bites. There is also a high demand for novel diagnostic tools that minimize cross-reactivity with other flaviviral infections.
Therefore, the development of safe and efficient vaccine candidates for the protection of humans and of innovative diagnostic systems are major goals of the West Nile Integrated Shield Project (WINGS).

Vaccine development
To date, there is no vaccine to protect humans from an infection with WNV. Scientists of the WINGS consortium have used several different innovative techniques of antigen design, vaccine delivery and adjuvants to develop vaccine candidates for the efficient and long lasting protection of humans against WNV. Major focus was on protein- and DNA-vaccines, as well as on combinations of the both methods. The immunization strategies are rapidly adaptable to newly emerging WNV strains.

Development of a specific diagnostic test
The protein components of WNV are very similar to other flaviviruses, such as JEV, TBEV or YFV. There-fore, antibodies produced upon an infection with WNV show a substantial degree of cross-reactivity to antigens from related viruses, which complicates a specific serologic diagnosis. The WINGS project was successfully looking for antigens in the proteome of WNV that allow a specific diagnosis of the virus.

Epidemiology of WNV in Europe
The epidemiologic situation of WNV in Europe is changing quickly. Novel strains (of both major WNV lineages) are detected now every year, with numerous cases of severe neurological symptoms among humans. Research on these novel WNV isolates as well as close collaboration of the WINGS consortium with national and international surveillance agencies helped to understand the spreading of WNV throughout Europe.

Project Results:
Work package 1: Antigen production

• Clone and express WNV-genes and fragments thereof suitable for immunization and diagnostic experiments
• Clone and produce DNA-plasmids carrying WNV sequences or reporter genes for DNA-vaccine experiments
• Supply the consortium partners with these substances
• Integrate protocols existing at the partners´ laboratories to generate SOPs for generalized viro-logical and immunological assays
In this work package, the antigens for the diagnostic and vaccination parts of the WINGS project were generated and distributed to the different consortium partners in charge of the corresponding work packages.

WNV proteome in a library
One of the major objectives in WP1 was the production of antigens for the identification of epitopes suitable for specific diagnosis and vaccination. Therefore, the whole WNV-proteome, separated in 30 amino acid short sequences, was expressed in bacteria. These recombinant fusion proteins are then being contacted with sera from WNV-infected people (European and American outbreaks) to identify regions which are recognized by antibodies. By also using sera from people infected with other fla-viviruses, the specificity of binding was analyzed.
Bacterial expression plasmids coding for GST-fusion proteins were generated (subcontract with ATG:Biosynthetics, Germany). The whole proteome of the WNV-strain “Italy09” (Lineage1, Acc#. GU011992) and the E-protein of the WNV-strain “Hungary04” (Lineage 2, Acc#. AAZ91684) were divid-ed into 30 amino acid long sequences, with an overlap of 10 amino acids to each side.
The 172 plasmids (plus 26 plasmids for the lineage 2 E-protein) code for the peptides fused N-terminally to the GST-protein to facilitate expression, purification and accessibility for the antibodies. Protein expression was done following standard procedures. Only a few clones did not express the proteins in high amounts, hence the bacterial culture had to be enlarged for these bacteria. To obtain proteins of sufficient purity, the purification protocol was optimized by establishing a two-step method, consisting of first a GST-affinity-column followed by gel filtration chromatography.
After quality control the proteins were used for the development of specific diagnostic tests (see WP2).

Recombinant WNV proteins
In addition to the proteome-spanning peptides, also larger fragments of WNV-proteins were expressed, purified and analyzed in vaccination and diagnosis. The two proteins under investigation were E and NS1.

Different versions of the E-ectodomain (i.e. the protein lacking the C-terminal transmembrane region) were cloned into bacterial expression vectors without affinity tags, expressed and purified using oxida-tive refolding techniques. Thereby it was possible to obtain correctly folded E-ectodomain, which was previously only possible by expressing the protein in insect cells. Correct conformation of the proteins was analyzed using antibodies recognizing structural epitopes. All of these proteins were distributed to WP3 and WP4 for immunization experiments.
Besides the wild-type protein (WNV-strain NY-99, lineage 1), two mutants were generated (Equad and EW101R). The aim was to analyze whether the antibody response could be skewed towards domain DIII, as in humans most antibodies are non-neutralizing and against the fusion loop. Therefore, mutant E-proteins might induce immune responses which are more protective than wild type E proteins.
In addition, the mutant and wild type E proteins were used for the development of a specific serological WNV test.
Bacterially expressed and purified NS1 or NS1-delta (amino acids 152 – 352) were mixed with Freund’s adjuvant and immunized into mice. After several rounds of boosting in incomplete Freund’s adjuvant, mice were pre-bled, and then challenged with WNV-NY99. Notably, despite achieving a high-titer of antibodies, there was no significant improvement in mortality after infection. This contrasts with prior published data in which anti-NS1 monoclonal antibodies were transferred passively. Thus, either the polyclonal anti-NS1 response did not target the same epitopes, or competition of binding of protective epitopes occurred.
To develop an additional platform for immunization, the WINGS consortium has collaborated with Dr. Mark Slifka (Oregon Health and Science University, USA). He has developed a methodology for produc-tion of H2O2 inactivated WNV-Kunjin virus, which is an inactivated vaccine that retains antigenicity bet-ter than classical formalin inactivation. The Diamond laboratory (WU) has extensively characterized this candidate vaccine – one of the main advantages is the immunization of a virus particle. The human anti-WNV neutralizing antibody response appears to be skewed towards quaternary epitopes on the virion surface. Thus, immunization with a virus (albeit, inactivated) may induce strongly these types of protective antibodies. WU, ISC and the Slifka lab tested this in greater detail. Indeed, the first set of stud-ies show that H2O2 inactivated WNV-Kunjin virus is highly immunogenic (induces B and T cell respons-es) and stimulates protective (neutralizing) antibody responses.

WNV-DNA plasmids
At the beginning of the project, the plasmid T-WNV-E was finalized at FHG. This plasmid was used to initiate the DNA-vaccine work in the project.
The plasmid codes for the ectodomain of the E-protein (NY-2000 strain), fused to a murine TPA signal sequence on the N-terminus. In addition, the coding sequence for the E protein in pT-WNV-E was changed for the sequence of a pathogenic Lin2 strain, recently occurring in Europe.
The second plasmid produced was prM/E-pVax1, which expresses the prM/E proteins fused to a JEV signal sequence in the same backbone vector as pT-WNV-E. Both plasmids were analyzed in cell culture and were capable to mediate significant expression of the encoded proteins.
At WU, the prM/E expression cassette was inserted into an alphavirus expression vector and used to stably transfect mammalian cell lines. Cells producing high amounts of virus-like particles were gener-ated that way.

Reporter plasmids
A plasmid was generated which contains the gene for the firefly luciferase and allows in vivo tracking of the encoded antigen (Bio-Imaging): the enzyme luciferase, which is produced immediately upon suc-cessful transfer of the plasmid into the animal cells, is metabolizing luciferin (which is injected into the animal). Thereby, a light-signal is emitted which can be detected and quantified by an appropriate in-strument (Bioluminescence Imaging System). Because the same plasmid backbone is used as with the plasmids for WNV immunization, the signals detected gave valuable information about the intensity of antigen expression that can be achieved by a formulation/delivery technique developed within the WINGS-project.
Similarly, the plasmid pVax1-EGFP-SV40 was distributed to the consortium members involved in DNA immunization. Instead of luciferase, it encodes the gene for the enhanced green fluorescent protein, which served as a marker for antigen expression in the DNA-vaccine experiments of the WINGS project (WP4).
SOPs and distribution of research material
At the beginning of the project, protocols for relevant assays were collected and it was decided on the optimal procedures among the consortium members. The corresponding protocols were standardized and distributed among the project partners. The protocols were also part of a number of publications, hence disseminated to the scientific community.
The following SOPs were generated:
•high-throughput focus reduction neutralization test (FRNT)
•plaque reduction neutralization testing (PRNT)
•WNV-ELISA (using recombinant WNV proteins, e.g. DIII or E ecto)

Work package 2: Diagnosis

• Development of a serologic test system for the specific diagnosis of WNV
• Screening of 285 human sera from flavivirus endemic countries
• Development of a memory B-cell assay
As mentioned above, the protein components of WNV are very similar to other flaviviruses, such as JEV, TBEV or DENV. Therefore, antibodies produced upon an infection with WNV show a substantial degree of cross-reactivity to antigens from related viruses, which complicates a specific serologic diagnosis. The WINGS project is looking for antigens in the proteome of WNV that allow a specific diagnosis of the virus.

Study of the human antibody response to WNV
At the beginning of the project, the human antibody response to a WNV infection had not been system-atically studied, besides the identification of epitopes from monoclonal antibodies. Therefore, we aimed to first gain insight into the profile of epitopes recognized by sera from a variety of infected individuals. The proteome library generated in WP1 was used to identify immunodominant linear B-cell epitopes. Serum samples from several outbreaks (Italy, Greece 2010, USA) were used.
It became clear, that the human humoral response to WNV is heterogeneous, as not all patients displayed antibody binding to the same epitopes to similar extends.
This is also illustrated when analyzing the ectodomain of the E protein, which in contrast to the prote-ome library, also contains structural epitopes (Figure 1): some sera showed only low binding.
The heterogeneity presented in Figure 1 became even more evident when the linear protein fragments of the WNV proteome library were used. First, only the structural proteins were studied (named after the amino acid position in the proteins C, prM and E). Although some immunodominant peptides were clearly identified, these were not all recognized by all the sera to the same extend (Figure 2). Hence, to correctly identify all sera as being infected, a mixture of several peptides has to be generated.
However, the results delivered important insight into the human antibody response:
- Many areas recognized by antibodies are in the region of the capsid and the prM protein
- In the E-protein, an immunodominant peptide was identified near the conserved fusion loop (E71-100), corresponding to previous results obtained with structural epitopes suggesting the fusion loop as immunodominant in human infections.
- Other immunodominant regions in the E protein are located in domain II and III
- No significant differences between infections with WNV from lineage 1 and lineage 2 are ob-served
- Antibody responses during European and American outbreaks are comparable
Based on these results, specific peptide mixtures were generated (including parts from the non-structural proteins) for the detection of ENV infections. These experiments are currently being opti-mized and prepared for publication and exploitation as a diagnostic assay.

Mutant E proteins
A form of the E protein was generated with loss-of-function mutations in the fusion-loop domain (Equad mutant). This domain is responsible for the major cross-reactivity observed between antibodies against related flaviviruses (see also above). Hence, by deleting it, we aimed for a specific binding of only WNV-antibodies to the protein, which could then be the basis for a serological test system.
The proteins were generated, and an ELISA-based test platform was established. A similar serum panel as described above was analysed.
As can be seen in Figure 3, under the conditions established there was no significant difference observ-able in binding of WNV-positive sera to either the wild type or the mutant. In contrast, sera positive for DENV or TBEV showed a significantly diminished binding towards the mutant protein (Figure 4). This confirms our assumption that infections with heterologous flaviviruses lead to cross-reactive antibodies against the fusion loop, which are not detected with the loss-of-function E protein. Therefore, this sys-tem can be used for the development of a specific diagnostic test, which would make confirmation tests using live viruses obsolete.
In summary, this part of the project led to the following results, which were published and topic of an IP application:
- A bacterially expressed recombinant E protein from WNV is suitable for serological diagnosis
- A mutant E protein displaying a loss-of-function mutation can be used to reliably discriminate heterologous flavivirus infections from a WNV infection
A second approach was based on computer-algorithm assisted epitope identification:
Due to the high protein sequence conservation among flaviviruses it was assumed that WNV specific antigens are required to develop a WNV specific serologic test system. Therefore, potential B-cell epitopes were identified using readily available bioinformatics programs ( and and predicted B-cell epitope sequences were analysed for sequence conservation among WNV-isolates but sequence differences to other flaviviruses. By this approach WNV specific epitopes were identified located within the structural proteins E, C and prM and non-structural protein NS5. Peptides expressed in P. pastoris did not induce a WNV specific immune response in Balb/c mice. Thus, the E. coli expression system was used alternatively. The recombinant proteins were expressed with N-terminal MBP and C-terminal his-tidine tag and purified by amylose affinity chromatography. Cleavage of MBP by factor Xa from the tar-get protein and subsequent purification by Ni-affinity chromatography worked for protein Cnat only. For ELISA studies the other recombinant antigens, i.e. Cme, E2, NS5.1 NS5.2 were only available as fusion proteins. It turned out that the use of MBP fusion proteins resulted in high background despite several strategies to optimize the purification process and ELISA plates. With synthetic peptide Ep1 only one of 36 serum samples was detected WNV positive. On plates coated with Cnat two sera were detected positive. But there was a lack of reproducibility.
As the peptides were not entirely suitable antigens for ELISA an alternative approach for diagnosis of WNV infections was considered, i.e. an Antigen-Capture-ELISA for NS1 protein. The detection of NS1 in serum samples is used for diagnosis of DENV infections and WNV NS1 also has been described as a useful antigen for diagnostics and as marker of infection. Within the NS1 protein WNV specific sequences containing potential B-cell epitopes were identified by bioinformatic programs and have been synthe-sized as short peptides. Of those peptides NS1.7 has induced high titres of peptide specific antibodies and reaction with native NS1 protein was verified by immunofluorescence assay on WNV infected Vero cells. As expected, serum antibodies cross reacted with JEV but not with TBEV or YFV. Additionally, NS1 was verified by Western blot in supernatant and in cell lysate of WNV infected cells using polyclonal mouse serum as primary antibody. So far generated monoclonal antibodies did not detect native NS1 protein but features of the anti NS1.7 serum were promising and the approach will be further pursued.
Detection of WNV-positive reaction in human sera from endemic countries (maybe in WP7)
To obtain an insight into the prevalence of flavivirus infections, 285 human sera from Ethiopia were screened for IgG antibody reactivity against WNV, TBEV, DENV and YFV by commercial as well as in-house ELISA. For the latter H2O2 inactivated virus was used as antigen. In addition, virus neutralisation test (VNT) was employed to confirm the specificity of the obtained ELISA results. When commercial ELISA was used, 35 out of 285 sera (12.2%) were tested positive for one or more flaviviruses, while with the in-house ELISA 27 sera (9.5%) were analysed as flavivirus positive. Cross-reactive antibodies were found responsible for reactivity with more than one flavivirus in 14 serum samples. By VNT seven sera were verified as WNV antibody positive. DENV-VNT is still in progress as all 4 serotypes have to be analysed. One serum tested positive for YFV by in house ELISA and was confirmed positive by neutrali-zation test. No TBEV-neutralizing antibodies were found. Although the results for DENV neutralisation tests are still incomplete, WNV and DENV seem to cause the majority of flavivirus infections in Gondar. Regarding VNT results the prevalence of WNV infections is 4.3%.

Development of a memory B-cell assay
The memory B-cell ELISpot represents a technique with which duration of vaccination success, i.e. the establishment of memory B-cells, can be surveyed appropriately. It is generally believed that ELISpot gives a better insight into immune reaction than mere screening for antibody. After tests to determine if enriched B-cells or PBMCs were to be used for ELISpot, the decision was made for PBMC-ELISpot using the human IgG ELISpot Plus Kit from MABTECH (Nacka Strand, Sweden). For the assay 9 ml heparinised blood was sufficient, while the work with B-cells requires 30 ml blood. To set up the test and show its functionality memory B-cell frequency was analysed in individuals vaccinated for tick-borne encephali-tis. Included were also sera of non-vaccinated people and people with DENV infection in the past. Re-combinant ED3 protein of TBEV and WNV as well as Influenza nucleoprotein expressed as MBP-fusion protein were used as specific antigens, MBP as control antigen and IgG-coated wells were used to verify activation of memory B-cells. Non-activated cells were plated on IgG-coated wells to screen for antibody secreting cells (ASCs) within the PBMC preparation. Figure 5 shows a characteristic ELISpot result of TBE vaccinated and non-vaccinated blood donors.
The results showed that the median frequency of TBEV-rED3 specific IgG ASCs with 0.05% per total IgG producing cells differed not significantly from the median frequency of NP specific IgG ASCs with 0.104%. Additionally, it turned out that period of time passed since vaccination does not influence the frequency of memory B-cells, whereas serum antibody titres decline significantly. This underlines that there is no correlation between memory B-cell frequency and antibody titre. Interestingly, memory B-cell frequency is significantly increased by booster immunisations.
As already shown by ELISA, rED3-TBEV detects no cross-reactive antibodies from WNV or DENV do-nors and no cross-reactivity was detected by ELISpot. Nonetheless, the sensitivity of the rED3, especial-ly for rED3-WNV is arguable. As shown by ELISA using rED3-TBEV as antigen low positive sera might be not detected. Sensitivity of rED3-WNV as ELISA antigen was even less as only 7 out of 15 WNV positive sera were detected. Thus for WNV-specific memory B-cells the mutated ectodomain of the E-protein (Equad) should be considered as a more suitable antigen.

Work package 3: Protein vaccine

• Development of potent WNV vaccine candidates for human use based on recombinant WNV proteins and novel adjuvants such as Matrix-M™
• Formulation of proteins and adjuvants
• Immunological and safety evaluations
• Coadministration of DNA and adjuvants

Formulation of Proteins and Adjuvants, Immunological Evaluation and Safety Evaluations

Recombinant proteins from WP1 were formulated with the Matrix-M™ adjuvant, tested in mice and immunologically evaluated as possible WNV protein vaccine candidates. The following parameters for a future WNV vaccine were considered: adjuvant composition, antigen/adjuvant-formulation, dosing, and vaccination protocol. The safety of the experimental vaccine formulations was evaluated continuously throughout the project.
The adjuvant Matrix-M™ is a saponin-based adjuvant composed of two individually formed 40 nm large particles, i.e. Matrix-A™ and Matrix-C™. The particles are made from selected quillaja saponin fractions of complementary properties; Matrix-C™ saponins are considered to be highly adjuvant active com-pared to the weaker, but well-toler¬ated, saponins of Matrix-A™. The mixture of the individually formu-lated particles significantly improves both, adjuvant activity and safety profile of the respective saponin fraction.
Different proportions of Matrix-A™ and Matrix-C™ were investigated to see if the most often used Ma-trix-M1™ could be improved by increasing the amount of Matrix-C™. The effect on the antigen-specific immune response of a Matrix-M formulation, made of a higher proportion Matrix-C™, here called Matrix-M3, was compared to antigen adjuvanted with Matrix-M1™ or only Matrix-C™ at different doses. Ovalbumin (OVA) was used as model antigen to economize the use of recombinant WNV E antigens. Despite large variation within the test groups, results showed that Matrix-M1™ adjuvant induced higher antibody titers compared to Matrix-M3 and Matrix-C™, hence the Matrix-M3 formulation with higher amount of Matrix-C™ did not improve the humoral immune response after one injection (Figure 6A, B). However, a strong boost effect was seen two weeks after the second immunization for all adjuvanted groups with high titers of antigen-specific IgG1 and IgG2a (Figure 6C, D). The cellular response was studied by antigen-specific restimulation of splenocytes in vitro and by analyzing the culture superna-tant for its IL-2, IFN-γ, IL-4 and IL-5 content. Splenocytes from mice immunized with OVA and Matrix-M1™ responded with the highest level of IL-2, IL-4 and IL-5 produced, indicating that this is a potent adjuvant formulation also for cellular immune responses. To conclude, an increase in the proportion of Matrix-C™ in the Matrix-M™ formulation did not improve the humoral or cellular response. Thus, the use of Matrix-M1™ was preferred over Matrix-M3 for the rest of the project.

Vaccine antigens
Different recombinant protein antigens based on the WNV E protein from WP1 has been evaluated as potential WNV vaccine candidates in combination with adjuvant Matrix-M1™. The recombinant proteins used are: Ewt, Equad and EW101R, and domain III of the E protein (DIII). The antigens were injected subcu-taneously (s.c) or intramuscularly (i.m.) to mice at antigen doses of 0.5 up to 10 µg and formulated with, or without, 5 or 10 μg of Matrix-M1™ adjuvant. In some cases the antigen was also formulated with the adjuvant Alum for comparative reasons. Additionally, the vaccine candidates were compared to a com-mercial horse WNV vaccine, Duvaxyn® WNV. The efficacy of the different vaccine candidates was evaluated by analyzing antigen-specific antibody titers, virus neutralization antibody titers and antigen-specific cellular responses.
Initially, the different proteins were compared to each other in regard to humoral and cellular immune responses elicited in order to select and proceed with fewer vaccine candidates. First, mice were im-munized with 3 or 10 µg of the different antigens in combination with 10 µg Matrix-M1™. The DIII anti-gen was only tested at 10 µg due to its small size, 12 kD, which is in the smaller size range of the Matrix-M1™ capacity. The mice were immunized twice, on day 0 and day 28. All recombinant proteins elicited an antibody response which was significantly enhanced when adjuvanted with Matrix-M1™ (Figure 7A-D). Serum from all mice was sent to UPA for plaque reduction neutralization testing (PRNT) to investi-gate the functionality of the raised antibodies. Immunization with the different E antigens generated WNV neutralizing antibody titers, which Matrix-M1™ adjuvant clearly enhanced, thus confirming that the formulation and vaccination was successful (Figure 7E). Matrix-M1™ adjuvanted Ewt gave the high-est PRNT titers compared to the other antigens, while the DIII antigen did not generate any neutralizing antibodies.
In addition to improving the antibody response, Matrix-M1™ clearly enhanced the antigen-specific cel-lular immune response for all vaccine antigens tested (Figure 8). This was tested in vitro by restimulat-ing splenocytes from immunized mice with the respective antigens, and then analyzing the cytokines produced. Of the different E antigens EW101R gave the highest IFN- response while the cytokine re-sponses for the other E antigens were of similar magnitude.
Recombinant Ewt as the main vaccine candidate
On the basis of these initial experiments Ewt was selected as the most promising candidate as both, high neutralizing antibody titers and good cellular immunity were recorded. To continue, an anti-gen/adjuvant dosing and durability study was done. Mice immunized with 0.5 1 or 3 g Ewt adjuvanted with 5 or 10 g Matrix-M1™ all developed high IgG1-and IgG2a-antibody titers after the booster im-munization (Figure 9). Even the lowest Ewt dose (0.5 μg) adjuvanted with Matrix-M1™ gave higher IgG1- and IgG2a-antibody titers compared to 3 μg Ewt alone. Hence, Matrix-M1™ clearly displays a dose sparing effect for Ewt in mice.
Antigen-specific splenocyte restimulation two weeks after booster immunization showed that the low-est Ewt dose adjuvanted with 10 µg Matrix-M™ induced as high cytokine levels as the higher adjuvanted antigen doses and compared to the antigen control (Figure 10).
To study the durability of the immune response four mice per group were kept until day 214 after the primary immunization. At this time point the IgG1- and IgG2a-titers were still high for the Matrix-M1™ adjuvanted vaccine groups and a significant antigen-specific T-cell response was detected when spleno-cytes from these mice were restimulated in vitro (data not shown).
Matrix-M1™ adjuvanted Ewt compared to Alum adjuvanted Ewt and Duvaxyn® WNV
To compare Matrix-M1™ adjuvanted Ewt with other adjuvant or vaccine preparations, mice were im-munized with Ewt alone, Ewt adjuvanted with Matrix-M1 or Alum, or with the commercially available veterinary vaccine Duvaxyn® WNV. Mice vaccinated with Matrix-M1 adjuvanted Ewt responded with significantly higher IgG1 titers compared to mice immunized with only Ewt or Duvaxyn® two weeks after booster (Figure 11A). Furthermore, Matrix-M1 adjuvanted Ewt also induced significantly higher IgG2a levels compared to Alum adjuvanted Ewt and Ewt alone. Only Matrix-M1™ adjuvanted Ewt elicited increased levels of neutralizing antibodies compared to antigen alone two weeks after boosting (Figure 11B). Antigen-specific splenocyte restimulation revealed that mice vaccinated with Matrix-M1 adju-vanted Ewt had greater secretion of IL-2, IFN-, IL-4 and IL-5 compared to only Ewt, Alum adjuvanted Ewt and/or Duvaxyn® WNV (Figure 11C).

WNV challenge
To ultimately test the protective efficacy of Matrix-M1™ adjuvanted Ewt, the vaccine candidate was test-ed in two separate WNV challenge experiments performed by UZH. The first challenge was done with a WNV strain homologous to the Ewt protein, ITA-09, and the second challenge with a heterologous WNV strain, AUT-08. Both challenges were successful and conferred complete protection against WNV disease in all vaccinated groups (Figure 12).
Coadministration of DNA and adjuvants
Together with UGENT the optimal conditions for combining DNA vaccines with Matrix-M1™ were investigated and are primarily reported by UGENT in work package 4. As injected DNA has to be translated into protein before an immune response can be elicited, the time point to inject Matrix-M1™ to ensure the best adjuvant effect for the antigen was evaluated. To guarantee that Matrix-M1™ was assessed together with an expressed protein we injected Matrix-M1™ and Ewt (protein) i.m in mice at different time points and in separate syringes to investigate the timing of antigen and adjuvant when draining to the same lymph node. The results suggest that it is possible to acquire an adjuvant effect even though the adjuvant and antigen are injected separately and at different time points (data not shown).

In WP3 we have shown that by combining a recombinant WNV E protein with adjuvant Matrix-M1™ we have developed a vaccine candidate that is highly immunogenic. Matrix-M1™ increased the humoral and cellular immune response towards WNV Ewt compared to unadjuvanted Ewt. Further, the addition of Matrix-M1™ to the vaccine reduced the antigen dose that was needed and increased the durability of the antigen-specific immune responses. Moreover, the vaccine candidate induced a balanced and stronger Th1 and Th2 response compared to Alum adjuvanted Ewt. Importantly, Ewt protein adjuvanted with Matrix-M1™ protected mice from lethal WNV infection using both, lineage 1 and 2 WNV, and one dose was enough for protection.

Work package 4: DNA vaccine

• Development of potent formulations for a WNV DNA vaccine
• Evaluation of DNA vaccine functionality
• Determination of the most optimal delivery route of these formulated DNA vaccines
• Evaluation of DNA vaccination in combination with Matrix-M1™

Nanoparticle formulations of the DNA vaccine
The WNV DNA vaccine that was used in this project encoded for the ectodomain of the E-protein (NY-2000 strain) fused to a murine TPA signal sequence on the N-terminus. We called this DNA vaccine “WNV-E-DNA vaccine” and each batch of produced WNV-E-DNA vaccine underwent a throughout quali-ty control before use. We formulated the WNV-E-DNA vaccine with four different carriers: DOTAP-liposomes, SECosomes, linear polyethylene imine (lPEI) and lPEI-mannose. The DOTAP-liposomes con-tain the cationic lipid DOTAP and the helper lipid DOPE in a 1:1 molar ratio. The SECosomes are highly deformable liposomes containing DOTAP, cholesterol and sodium cholate in a 6:1:1 molar ratio. In the past we have shown that siRNA/SECosome complexes (SECoplexes) can penetrate much deeper into the human skin than other siRNA/liposomes complexes. We used linear PEI with a molecular weight of 22 kDa. The lPEI-mannose was prepared by covalently coupling of mannose to the N-groups. The LPEI-mannose contained 3% mannose (calculated on the nitrogen content of the polymer). The formulation of DNA vaccines with lPEI-mannose has been extensively optimized in the past for topical delivery of DNA vaccines by GIM. Therefore, further optimization studies with DermaVir will not be considered in this project. WNV-E-DNA vaccine containing nanoparticles were prepared by mixing the carriers with the DNA vaccine at different +/- or N/P (nitrogen/phosphate) ratios using standardized protocols. The quality of the produced nanoparticles was monitored by measuring their physicochemical properties.
Selection of the most optimal WNV-E-DNA vaccine formulation for each carrier
The gene transfer efficiency of DNA formulated with a carrier into nanoparticle depends on the DNA/carrier ratio. It was our original goal to select the optimal ratio based on the in vitro gene transfer efficacy. However, preliminary in vivo experiments demonstrated that in vitro optimized ratio cannot always be translated to in vivo. Therefore, in vitro ratio optimization experiments were stopped early in the project. To find the optimal carrier/DNA vaccine ratios for each carrier we performed vaccination experiments. The optimal ratio for the DOTAP-liposomes, the SECosomes and the lineair polyethylene imine (lPEI) polymer was respectively 8/1 (+/- ratio), 4/1 (+/- ratio) and 12/1 (nitrogen to phosphate ratio; N/P ratio). However, the efficacy of the lPEI based nanoparticles at a ratio 8/1 was only slightly below the efficacy of the nanoparticles prepared at 12/1. The ratio of the lPEI-mannose based nanopar-ticles was 4/1 (N/P ratio).
Selection of the most efficient carrier for formulating the WNV-E-DNA vaccine
After having optimized the ratios we selected the most efficient carriers for formulating the WNV-E-DNA vaccine. The selection was done based on a DNA prime/DNA boost schedule. Form these experi-ments we retained DOTAP-liposomes, lPEI and lPEI-mannose formulated with the WNV-E-DNA vaccine at 8/1 (+/- ratio), 8/1 (N/P ratio) and 4/1 (N/P ratio). Nanoparticles based on lPEI formulated at 12/1 (N/P ratio) were also selected for further experiments.
Evaluating of the effect of the WNV-E-DNA vaccine dose on the immune response
Increasing doses (0.1 1, 10 and 20 µg) of WNV-E-DNA vaccine containing nanoparticles were given to mice in DNA prime/DNA boost schedule. From the cellular and humoral immune response measured 14 days after the boost we could not detect a clear dose response effect. However, the lowest dose (0.1 µg) seemed to inferior to the other ones. It is important to mention that the humoral as well as the cellular immune response after homologous vaccination (i.e. a DNA prime /DNA boost vaccination schedule) with the nanoparticle formulated DNA vaccine is very low. Therefore, we decided to replace the DNA boost by a protein boost. As protein boost we used domain III of the E-protein or the E-protein. Both proteins were used in combination with Matrix–M1TM adjuvant. This heterologous approach was clearly much more efficient than the homologous vaccination schedule (Figure 13). The data in figure 13 also demonstrate that priming with 1 µg of formulated WNV-E-DNA vaccine is enough.
In vivo toxicity study of the candidate DNA vaccine formulations
The in vivo toxicity of the different candidate formulations was investigated by measuring basic cyto-toxicity markers, i.e. total serum bilirubin and serum alanine transaminase activity. The only formula-tion that did not result in increased levels of these markers was the one based on lPEI-mannose. Espe-cially the toxicity of the lPEI based nanoparticles was the highest when they were formulated at a 12/1 ratio. The toxicity also increased with the dose. Based on these efficacy and toxicity data we decided to focus on the lPEI-mannose based particles for a throughout characterization of the immune response after intradermal, topical and intramuscular administration.
Selection of the most optimal delivery route of the WNV-E-DNA vaccine formulated with lPEI-mannose
The WNV-E-DNA vaccine formulated with lPEI-mannose into nanoparticles (which we called WNV-DermaVir nanoparticles) was administered intradermal, topical or intramuscular in balb/c mice. For topical application, the hair of the dorsum was removed by shaving and the skin was exfoliated with a sponge. Finally, tape stripping was performed and 80 µl WNV-DermaVir nanoparticles (containing 1 µg WNV-E-DNA vaccine) was applied on the entire tape stripped area. After air-drying, mice were returned to the corresponding cages. The same dose of WNV-DermaVir nanoparticles was also given intrader-mally and intramuscularly. Mice were boosted four weeks after the prime either via an intramuscular injection of the same dose of WNV-DermaVir nanoparticles or via subcutaneous injection of recombi-nant E-protein formulated with Matrix-M1TM.
The levels of E-protein specific IgG1 and IgG2a antibodies were measured 14 days after the boost. The endpoint titer of anti-E-protein antibodies in mice primed with WNV-DermaVir nanoparticles was very low (i.e. < 100). After boosting these mice with WNV-DermaVir nanoparticles, the endpoint titers of the anti-E-protein antibodies were still below 100 (Figure 14). In contrast, mice primed with WNV-DermaVir nanoparticles and boosted with E-protein/Matrix-M1TM developed measurable IgG1 (Figure 14A) and IgG2a (Figure 14B) titers against the WNV-E protein. Mice primed with WNVDermaVir nano-particles and boosted with the E-protein also showed a T-helper (Th)-2 skewed response. Indeed, the IgG1/IgG2a ratios for intradermal, intramuscular and topical administration were respectively 3, 1.7 and 2.1. The highest antibody titers were obtained in the groups primed intramuscularly or topically with the nanoparticles. When a single vaccination with E-protein (without priming) was given, the IgG1 (Figure 14A) and IgG2a (Figure 14B) titers were lower and a Th1 shifted response was induced. The titers of the unprimed mice were statistical significantly lower compared to the titers obtained for the intramuscularly primed mice.
Next, virus neutralizing titers, which are very important for protection against infection, were deter-mined after vaccination via the best administration routes of the WNV-DermaVir nanoparticles (i.e.intramuscular and topical). All the mice primed with WNV-DermaVir nanoparticles and boosted with E-protein had detectable virus neutralizing antibodies while none (topical priming) or three out of five mice (intramuscular priming) raised virus neutralizing antibodies after boosting with WNV-DermaVir nanoparticles (Figure 15). The latter was surprising as intramuscularly priming and boosting with WNV-DermaVir nanoparticles resulted in antibody titers below 100 (Figure 15).Subsequently, the cellular immune response was measured two weeks after boosting. WNV-E protein specific IL-4 and IFN-gamma responses were measured via ELISA and intracellular cytokine staining, respectively. Mice that were topically or intramuscularly primed with WNV-DermaVir nanoparticles had a higher percentage of IL-4 producing splenocytes than mice that received the WNV-DermaVir na-noparticles prime via an intradermal injection. Mice receiving a single injection with E protein (without priming) did not produce IL-4 positive spots (Figure 16). Next, the amount of IFN-gamma producing CD4+and CD8+ splenocytes was determined. The percentage of CD4+ cells that produced IFN-gamma was 2 to 2.5-fold higher in the mice boosted with WNV-DermaVir nanoparticles compared to the mice boosted with WNV-E protein. In comparison, in the CD8+ population mice boosted with WNV-E protein showed the highest amount of IFN-gamma positive cells, and their relative numbers were higher than their respective relative numbers of IFN-gamma positive CD4+ cells. Altogether, these data demonstrate that a cellular immune response can be obtained by combining a WNV-DermaVir nanoparticle prime with a E-protein/Matrix-M1TM boost.
Finally, protection against a lethal WNV infection was studied with the two most potent vaccination regimes, i.e. intramuscular or topical priming with WNV-DermaVir nanoparticles followed by subcuta-neous boosting with WNV-E protein. All mice primed by either route survived the infection. The surviv-al rates of these groups were significantly higher than that of the control group vaccinated with egfp-DermaVir nanoparticles, in which all of the mice succumbed to the infection.
Electroporation mediated delivery of WNV-E DNA vaccine as a benchmark
The optimal parameters for electroporation mediated intradermal and intramuscular DNA vaccination with the WNV-E-DNA vaccine were established. Electroporation mediated delivery of the WNV-E-DNA vaccines resulted in a much higher humoral response than vaccination with the WNV-E-DNA vaccine containing nanoparticles. Additionally, in contrast to vaccination with the WNV-E-DNA vaccine contain-ing nanoparticles, the antibody titers also increased as a function of the WNV-E-DNA vaccine dose. Cel-lular immune response was rather low. This may be related to the fact the encoded E-protein contains a secretion signal.
Effect of Matrix-M1TM on the immune response after vaccination with WNV-E-DNA vaccine
Two sets of experiments were performed. In a first experiment we immunized mice by intramuscular injection of the WNV-E-DNA vaccine (10 µg) formulated with DOTAP-liposomes in the presence or ab-sence of Matrix-M1TM. Two different doses of Matrix-M1TM, i.e. 5 μg or 10 μg, were injected intramuscu-larly near the injection spot of the formulated WNV-E-DNA vaccine. The adjuvant was injected either directly or one day after administration of the DNA vaccine containing nanoparticles. The mice were immunized twice, i.e. on day 0 and day 28. These experiments did not detect any improvement in the cellular or humoral response when Matrix-M1TM was combined with the DNA vaccine containing nano-particles. In a second experiment 10 µg Matrix-M1TM was given two days after intramuscular electro-poration of 10 µg of a DNA vaccine encoding the prM/E proteins fused to a JEV signal sequence. Again, we did not detect a significant improvement of the immune responses when the DNA vaccine was com-bined with Matrix-M1TM.

Work package 5: Challenge Studies in mice

• Infectivity of West Nile Virus (WNV) isolates – Disease Model
• Performance of Different Vaccination Schemes
• Protection Studies

The aim of WP5 is to develop a vaccine that would be effective within a wide spectrum of WNV isolates. The age component was reflected in the first deliverable study design as the vaccine is supposed to work for all the patients. Additionally, to ensure a higher transferability to the primate model, we chose to work with 2 age groups and 2 different mouse strains (NMRI and BALB/c). Including an outbred mouse strain like NMRI mice (instead of e.g. the inbred C57BL/6J) covers a larger genetic background and improves the likelihood of a good transferability. To test whether the vaccines work on different virus strains, we used three different virus strains for an infection model: Two on a linage 1 background (IS-98-ST1 and ITA09) and one on a linage 2 background (HU-08). Once the infection model was set up for both the lineages, different vaccination schemes were tested and the challenge study was performed to compare the efficacy of the vaccine in protecting the mice from lineage 1 and lineage 2 infection.
Infectivity of WNV isolates – Disease Model
In order to keep the animal numbers low, the study was developed in a progressive design. In the “Dose Finding” part, the viral strain in question was applied i.p. to heterogeneous groups of four mice with increasing concentrations. If one dosage of virus caused that all animals of one group died, reached their endpoint or fell ill, this dosage was validated in a second study with more animals. Each group contained mice of two different strains and age groups: One young NMRI mouse, one old NMRI mouse, one young BALB/c mouse and one old BALB/c mouse. Before and after the infection the animals were scored according to a scoring scheme as described in Annex I. As soon as the animals reached a certain score they were considered as terminally ill and were euthanized.
WNV strains ITA09 at a dose of 104 plaque forming units (pfu) and HU-08 at a dose of 105 pfu showed 100% mortality (Figure 17) and were then validated with 12 control animals (3 NMRI young, 3 NMRI old, 3 BALB/c young, 3 BALB/c old) running in parallel for both studies (n total (control) = 24). With ITA09 (104 pfu) 15 of 16 animals were classified as infected due to reaching the endpoint, death of showing signs of disease (cf. textbox below). One died (on day 9 p.i.) 11 reached the endpoint (between day 7 and 10 p.i.) and three showed signs of a disease, but did not reach the endpoint. With HU-08 (105 pfu) 16 of 16 animals were classified as infected. One animal died on day 8 p.i. and 15 animals reached the endpoint and were euthanized between day 6 and 11 p.i.. None of the animals in the control groups were classified as infected.
Therefore, both virus strains are suitable as an infection model and can be used in the future protection studies. This allows testing the vaccines against a linage 1 and a linage 2 WNV strain and contributes to a higher reliability for a broad-spectrum efficacy.

Potency of the vaccine candidates to protect mice from WNV infection (lineage 1)
Based on the earlier report from WP3 and WP4 on various vaccine candidates, the best vaccination schemes using different potent delivery methods were chosen for the challenge studies in mice. The first challenge experiment was carried out using the ITA-09 WNV strain model established above. In order to test the potency of the vaccines, both DNA and protein vaccines were administered with protein booster vaccination for the first challenge study (Table 1).
Control groups showed infection and were euthanized between day 8 and day 10. However, 10-20% of the animals survived, which corresponds with the data of the previous studies, when the infection mod-el was established. ITA-09 generally did not produce 100% mortality even in higher doses than 104 pfu. Overall, all the vaccines provided protection against the ITA-09 strain (Figure 19). The losses in group ISC-2 were not infection related (see below), while one animal in group ISC-1 clearly showed signs of disease and reached its endpoint. As expected, the total scores for the control groups were 6 or more and showed all signs of infection. Also a very small elevation in scoring points can be observed around day 3 post-infection. This can be because of either viraemia or the blood sampling procedure on day 2 after the infection.
Efficacy of the vaccine candidates to lineage 2 WNV infection
Data on the potency of the vaccines from the first challenge study on ITA-09 strain previously reported provided the groundwork for planning the second challenge experiment. In addition to DNA & protein vaccines with protein booster tested in the first one, two other schemes were added to vaccination pro-tocol; DNA vaccine with DNA booster and protein vaccine without a booster (Table 2).
Protective effect was observed against HU-08 strain with the protein + protein boost and DNA + protein boost vaccines (Figure 20) similar to ITA-09 strain (Figure 19). ISC-3 and G3 were new groups added to the second study. Interestingly, ISC-3 group, where no booster vaccination was given, showed “0” scor-ing from p.i. day 3 for AUT-08 strain, unlike the other vaccinated groups, suggesting that the protection in this group was slightly more effective than in the others. The new group “G3” where only DNA vac-cine was used showed symptoms of an infection (Figure 20). The DNA vaccine + DNA boost did not pro-vide any protection to the animals against the AUT-08 strain. The effect of G3 and ISC-3 vaccines on ITA-09 strain has not been tested and therefore a comparative study between two strains regarding these vaccinations is not possible. Regarding the bodyweight between the two strains, we observed a higher weight gain in animals infected with lineage 1 strain in comparison to the lineage 2 one. This might in-dicate a quicker recovery in animals from lineage 1 challenge with the different vaccinations adminis-tered.

Significant Results
- A broad spectrum murine WNV infection model is established and the scoring system for the in-fection is refined and confirmed
- Both protein and DNA-based vaccines administered with the protein boost protect animals against the two strains
- Protein vaccine without the booster provides protection against AUT-08 strain

Work package 6: Evaluation of the vaccine in macaques

• Evaluation of WNV vaccine candidates intended for future use in humans in rhesus macaques (Macaca mulatta)
• Experimental infection study of 4 rhesus macaques and 4 common marmosets using the same challenge virus
• Evaluation of infection route (intradermal), viral dose, course of infection in both species

Infection study
Macaques were intradermally inoculated with 2 x 105 TCID50 of WNV strain Italy2009, while marmosets were infected with a lower dose of virus (1 x 105 TCID50). All animals became infected within 2 days after infection. Two macaques and two marmosets were followed for 14-days (Figure 21A). The other 4 monkeys were monitored for 3 days (Figure 21B). During infection the animals were bled regularly, and body temperature was measured 24/7 (in the 14-days study) by using a transponder. At the end of the studies, the animals were euthanized, and necropsy was performed. Tissues were collected to determine the bio-distribution of the virus, and to allow the evaluation of pathological changes during acute infection. Virus was detectable at day 1 post-infection, and peak viremia (dark blue area) developed at day 3 in rhesus macaques (R01034 and R05066), and in marmosets at day 2-3 (M08082 and M08009). Peak viremia in marmosets was 1 log higher than in the macaques (106 RNA genome copies/ml vs. 105 RNA genome copies/ml)(Figure 21C). In the marmosets viremia was detectable until day 8 and 12, while in rhesus macaques viremia tended to be shorter (virus detectable until day 5 and 6). Due to technical problems with the temperature transponders only 1 out of 4 functioned (R01034). In this an-imal no changes in body temperature (yellow line) were measured that could be related to WNV infec-tion.
No significant changes in behavior, but also in hematological parameters and blood biochemistry were detected in both species. Based on these findings it was decided to euthanise the animals in study B at day 3 post-infection. Again, virus was detectable in both species at day 1, and viremia quickly developed. At the time-point of euthanasia virus titers in the rhesus macaques were already slightly. The total virus production in both studies was highest in the marmosets.
Sequential plasma samples from the rhesus macaques R01034 and R05066 (14-days study) were ana-lyzed for the presence of an anti-WNV E-protein antibody response at partner FHG (Figure 22). Anti-WNV IgM was first detected 9 days post-infection with WNV-Ita09, and titers further increased at days 12 and 14 p.i. Anti-WNV IgG was detectable at 10 days p.i. and similar to IgM, IgG levels increased over time. Limited blood sampling from marmosets allowed the determination of antibody levels only at the start of the study and at euthanasia at day 14. At the start of the study the marmosets were negative for E-antigen-specific IgM and IgG, but both isotypes were clearly detectable at the time point of euthanasia.
Tissue samples collected at necropsy were analyzed for the presence of WNV RNA by using PCR assays. Three days after infection more tissues were marmosets than in the macaques. Seven tissues were WNV-positive in common marmoset M09134, while in marmoset M09135 nine tissues tested positive. Viral RNA was found in the axillary and mesenteric lymph nodes, the lung, and the adrenal gland of both animals. Samples from the tonsil, gall bladder, trachea, thymus, ileum, stomach, kidney, and urinary bladder were tested positive in at least one of the marmosets. In sharp contrast, only one rhesus ma-caque (R04097) tested positive in the spleen, while in the other macaque (R04093) no virus was de-tected in any of the solid tissue samples. Fourteen days post-infection the number of WNV-positive tis-sues was comparable in both monkey species. WNV RNA was primarily detected in spleen and lymph node samples from both species. In addition, viral RNA was found in the urinary bladder of macaque R01034, and in brain tissues (cerebellum, hippocampus) of macaque R05066. Marmoset M08009 also tested positive in the hippocampus, and M08082 was positive in one of the kidneys.

General conclusions
- Both monkey species can be intradermally infected with WNV strain Ita09
- Peak-viremia was reached around day 3 post-infection
- In marmosets the viremic period was longer, and peak-viremia was higher
- None of the animals showed any signs of disease or change in behavior
- Broader tissue distribution of WNV was found in the common marmosets

Vaccine evaluation study
A vaccine evaluation study in 18 rhesus macaques was performed. Two different vaccination strategies had been selected on basis of studies in mice that had been performed by partners UZH, UGENT, GIM and ISC.
The first strategy consisted of 3 immunizations with recombinant WNV E-protein (provided by partner UPA) mixed with the Matrix-M1TM adjuvant (partner ISC), while the second protocol consisted of a priming immunization with a DNA vector expressing the E-protein mixed with LPEIm, followed by 2 booster immunizations with WNV E-protein adjuvanted with Matrix-M1TM. Protein immunizations were given intramuscularly in the upper arm with 20 μg E-protein in 25 μg Matrix-M1TM. This mixture was diluted in PBS and brought to a volume of 0.25 ml. The total volume of the DNA vaccine was 800 µl, and this was given as 8 doses of 100 µl, injected intradermally on the back of the animal. The immuniza-tions were given with 3-weeks intervals. Nine weeks after the third immunization, the animals were intradermally challenged with 2 x 105 TCID50 of WNV strain Italy2009 (provided by partner UPA).
During the immunization phase and during the 2-weeks post-challenge follow-up period the animals were regularly bled for immunogenicity and efficacy testing. As read-out parameters for immunogenici-ty we used the induction of anti-WNV antibodies (tests at partner FHG), induction of WNV-neutralizing antibodies, and the IFNγ response to purified E-protein. Vaccine efficacy was evaluated by the detection of WNV in blood by real-time PCR, and the PCR analysis of tissue samples taken at necropsy 14-days post-challenge.

Immunogenicity testing of WNV vaccines
The antibody responses to WNV were measured from the start of immunization period (week 0) until 4 week pre-challenge (week 11) (Figure 23).
In figure 24 we have shown the IgG responses targeted to the E-protein ecto domain (top panel), to inactivated-whole virus (middle), and the WNV neutralizing antibodies (bottom). Antibodies in group 1 were detectable one week after the 2nd immunization. The responses reached a plateau at one week after the 3rd immunization, and then titers slightly decreased towards challenge.
In group 2, IgG reponses were first detectable after the 2nd immunization, but then quickly rose to the same levels as measured in group 1, indicative of a priming effect by the DNA vaccination.
Virus neutralizing antibodies (vna) were measured for both vaccine groups at start of the study and 4 weeks pre-challenge (week 11). At the latter time-point vna were clearly detectable. Evaluation of vna using another WNV genotype 2 virus revealed that both vaccine strategies, that were based on the genotype 1 Italy09 strain, induced cross-reactive vna.
In figure 25, we have shown the number of IFNγ-producing cells in the blood of the different groups. Similar to the antibody responses, the number of IFNγ-producing cells in the blood reached a plateau after the third immunization. These cells were already detectable after one immunization, but more cells were responsive after the protein priming immunization than after the DNA prime.
It can be concluded that both vaccine strategies were able to induce anti-WNV antibodies, and that both strategies led to the induction of cross-neutralizing antibodies to both WNV genotypes.

Post-challenge follow-up
After the intradermal inoculation with 2x105 TCID50 of WNV strain Italy2009, the animals were fol-lowed-up for 14 days. The first 10 days the animals were bled daily, then the animals were bled at day 12 and at day 14 (day of euthanasia). At all time-points of the post-challenge, we performed real-time PCR on plasma to determine the WNV load. The results are depicted in the table below (RNA copies/ml plasma):
No virus was found in plasma of the vaccinated animals, while in 5 out of 6 control animals virus was detectable. We also tested tissue samples for WNV RNA using real-time PCR and nested diagnostic PCR. In all control animals (including R08058 that was plasma-negative) WNV was detectable in one or more tissues. In contrast, residual WNV was tested in one tissue sample of individual in each group.
Conclusions and suggestions for further research
Both vaccine strategies were equally capable of fully preventing plasma viremia in rhesus macaques after experimental challenge with the European West Nile virus strain Italy2009. In only 2 out of 12 animals residual WNV was found in solid tissue samples, in contrast to 6 out of 6 controls. The positive outcome of this study warrants further studies in nonhuman primates to improve vaccination strategies (less immunizations), and to evaluate in vivo cross-protection against other WNV genotypes. In addition, in order to develop vaccine strategies for the human target populations, i.e. individuals with a less-competent immune system, like the elderly or otherwise immunocompromised individuals, it may be essential to evaluate vaccine protection in artificially immunocompromised nonhuman primates.

Work package 7: Epidemiology
• Investigation of occurrence and spread of WNV in Europe
• Investigation of WNV prevalence and collection of information on WNV cases in humans and an-imals to establish a Europe-wide epidemiologic map
• Maintenance of contacts with authorities in charge of surveillance (ECDC, public health agencies, veterinary institutes, etc.)
• Sample collection from possible and probable WNV cases for laboratory testing and confirma-tion of diagnosis and for viral isolation
• Performing genetic analysis and characterization of WNV strains based on whole genome se-quencing data; ultra-deep sequencing of WNV to identify intra-host variants in patients with in-fection; phylogenetic analysis of new WNV strains by comparison with other European isolates as well as with WNV isolates from all over the world and used to follow WNV spreading; charac-terization of WNV mutations potentially relevant for virulence and viral transmission by bioin-formatics modelling and in vitro and in vivo experiments
Work achievements
WNV Prevalence
WNV prevalence was investigated in northern Italy by UPA within the surveillance activity as regional reference centre for WNV diagnosis in humans. Results were timely reported to public health authorities in charge to communicate WNV cases to the Italian Ministry of Health and to ECDC. In addition, UPA collaborated with other European Reference Centres for WNV surveillance and diagnosis to provide advice and technical support and maintained contacts with international agencies in charge of surveil-lance (ECDC, ENIVD, EpiSouth) and National and Regional institutions (Italian Ministry of Health, Re-gional Public Health Systems, and Veterinary Reference Centres) to communicate surveillance data and get information on WNV epidemiology data in Italy, Europe, and worldwide. Relevant information for public health was reported with press releases, communications at scientific meetings, and publications in international scientific journals. Updates on the epidemiology situation of WNV infection, circulation, and strains were provided to the WINGS Consortium. Epidemiological data on the spread of WNV in Europe during the time of funding (i.e. 2011-2013) was collected and analyzed in a final report, which included surveillance results obtained within the WINGS project and results on human, entomological, and veterinary surveillance that were available from the scientific literature and public databases. As described in detail in the report, the 2011-2013 seasons were characterized by an increase in the num-ber of human cases of WNV infection reported and in the number of EU and neighbouring countries affected by WNV circulation (Figures 26 and 27). Besides the production of epidemiological data, UPA provided a significant improvement to the diagnosis of WNV infection with the set-up of methods for WNV detection and isolation from urine samples of patients with acute infection. In addition, protocols for direct whole genome sequencing and ultra-deep sequencing of WNV genome directly from human samples were set-up.
Sample Collection
Serum samples and WNV lineage 1 and 2 isolates were obtained from patients with WNV infection by UPA and made available for the consortium. Serum samples were collected also from subjects with an-tibodies against other flaviviruses, as well as from control patients without antibodies against fla-viviruses. These samples and viral strains were used for the development of WNV tests (WP2) and for the setup of mouse and primate challenge models for testing vaccines (WP5 and WP6).

Genetic Analysis and Characterization of WNV Strains
An important limitation to the understanding of WNV circulation and spread in Europe is the lack of genome sequence information on WNV strains from Europe. The WINGS project provided a great con-tribution to the knowledge on the molecular epidemiology of WNV in Europe. In fact, most of full ge-nome sequences of WNV strains published in GenBank in the 2011-2013 period (17 out of 20 full WNV genome sequences) were obtained within the activities of the WINGS project (Table 3).
These genome sequences included WNV lineage 1 strains from Italy and WNV lineage 2 strains from Greece, Italy, and Austria. Phylogenetic analysis of all available full and partial WNV sequences from Europe reported in the recent years was done in order to follow the spread and evolution of WNV. In addition, in silico and experimental studies were performed to characterize WNV genome mutations potentially associated with pathogenicity and transmissibility. The results of these studies were pub-lished in international scientific journals and in a scientific report. Briefly, phylogenetic and genetic analyses demonstrated that the 2011-2013 seasons were characterized by the spread of virulent WNV lineage 2 strains belonging to the Hungarian cluster to several Central and Southern European countries, including Hungary, Austria, Greece, Serbia, Albania, and Italy, where they caused human outbreaks of neuroinvasive disease. Another WNV lineage 2 strain caused large outbreaks in the Russian Federation and in Romania. In the period from 2011 to 2013, circulation of WNV lineage 1a in EU countries was documented only in Italy and, among neighbouring countries, in Turkey. Also WNV lineage 1a strains detected in Europe in the recent years belonged to two distinct clades, i.e. the Western European and Mediterranean clade (WNV strains isolated in Italy and in Spain) and the Middle-East and Sub Saharan clade (strains detected in Turkey) (Figure 28). These different subgroups of WNV lineage 1 and lineage 2 strains have been probably introduced independently from Africa through migratory birds, as suggested by their geographic clustering with bird migration routes from Africa to Europe. After introduction in Europe, these WNV strains have overwintered and established endemic transmission cycles.
Phylogenetic trees of full genome sequences including of WNV lineage 1 and lineage 2 strains isolated in EU countries during the 2011-2013 seasons are reported in Figure 29. The close genetic relatedness of these strains indicates that the different WNV strains probably evolved locally in Europe, while only a few introductions occurred from Africa. It is conceivable that WNV strains with increased virulence and spread capacity have emerged as a consequence of adaptation to local environmental conditions and have been responsible for outbreaks. Sequencing of the full genome of different WNV isolates during outbreaks allowed determining the inter-host variability of the virus. In addition, inter-host variability of WNV was characterized by direct deep-sequencing of the viral genome from human samples. These analyses demonstrated the high genetic stability of WNV.
Mutations in viral genes potentially associated with increased pathogenicity were identified in WNV strains responsible for outbreaks in Europe. These mutations included the presence of a proline in the 249 position of NS3 that has been associated with lethality in the American crow. This mutation was present in the WNV lineage 1 WNV Ita09 and Livenza strains that caused large outbreaks in Italy and in the WNV lineage 2 strain circulating in Greece. An in silico prediction analysis indicated that presence of a proline in position 249 could increase NS3 protein stability at high temperature, i.e. 40-42 °C typical of avian hosts. In vitro experiments demonstrated that recombinant proteins with either proline (P), alanine (A), or threonine (T) at position 249 had similar ATPase activity at 28°C (mosquito tempera-ture) and 37°C (mammalian temperature), while at 42°C NS3-249P had greater activity than NS3-249A and NS3-249T, in agreement with the prediction model and with data on bird lethality. The WNV lineage 2 strains of the Hungarian cluster that were responsible for human outbreaks in Greece, Serbia, and Italy in the recent years, were characterized by a high similarity of the polyprotein amino acid sequence and by the presence of some amino acid changes that characterized some strain, such as the V119I mutation in NS2B, H249P in NS3, and S14G, T49A, and V113M in NS4A which were present in Greek strain and the I159T change in the E protein, S192C in NS2A, and A298T in NS5 that distinguished the Italian WNV lineage 2 genomes from other genomes of the Hungarian cluster. In particular, the E I159T mutation was predicted in silico, but not confirmed experimentally, to introduce a new glycosylation site. We cannot however exclude that this mutation, like the E-A159V mutation which characterized the North American WN02 strain, is associated with increased transmissibility and virulence. Finally, in vivo ex-periments were also performed, which demonstrated the high pathogenicity in mice of WNV lineage 1 Ita09 and WNV lineage 2 Austria/08 strains, as detail in the WP5 report.

Potential Impact:

The project was composed in response to the call HEALTH.2010.2.3.3-3 (Integrated disease-specific research on West Nile Virus infections, Chikungunya and/or Crimean Congo Haemorrhagic Fever) which was published in 2009 and had the following content:
The aim is to integrate research on one or more of the above diseases – or other relevant newly emerging human infections in Europe. Research areas such as basic virology, transmission, epidemiology, diagnostic approaches and/or treatment and prevention strategies should contribute to a comprehensive approach towards these diseases. Proposals are expected to be output-oriented and to build on existing basic research results to the extent possible in order to advance to late term pre-clinical studies and deliver tools countering the threat of these infections.
Expected impact: Projects in this field are expected to create an integrated European research capacity on diseases, which are among the prime threats of vector-borne emerging diseases in Europe. In case of a significant new unforeseen threat before the time of the call closure, research on this should be included. Significant synergy is expected from uniting researchers from different fields around the common goal of generating new knowledge on these diseases, and delivering improved ways to monitor its spread, diagnose, prevent, control and treat it. The ultimate impact is to enhance Europe's preparedness for these new diseases.
The problem of WNV is a rather recent one for Europe, as regular outbreaks involving humans are oc-curring only since a couple of years. However, since the time the call for projects was launched from the EU, the situation got dramatically worse:
- Case numbers among humans increased from 10-20 per year (in 2009) to several hundreds of severe infections (since 2010).
- Number of affected countries increased from Italy and Hungary to almost all Southern and South-Eastern European countries.
- Circulating strains became more diverse with WNV from the two major genetic lineages co-circulating in the same geographical area.
- WNV seems to overwinter in many affected areas.
- Many countries ban persons returning from countries with WNV circulation from blood dona-tion. E.g. in Germany this currently applies to 13 European and Mediterranean countries (
But not only in Europe, WNV is an emerging pathogen of increasing importance also in other parts of the world:
- Case numbers in the USA of 2012 were the highest since the peak of the epidemic in 2003, de-spite major efforts in vector control activities and public information campaigns.
- Outbreaks of severe WNV disease are being reported from Australia, South America and several countries in Asia where WNV was never observed before.
Hence, the demand for tools to counteract the threat of WNV on the population (and on the veterinary sector) is becoming even higher as compared to the baseline in 2009. Therefore, it was a central objec-tive of WINGS to deliver tools in order to address this demand, and the focus in this respect was three-fold:
• Development of vaccine candidates for the protection of humans
• Establishment of novel diagnostic test systems for the secure and specific diagnosis of infections
• Investigation of the spread of WNV in affected areas by analyzing the circulating strains

The project delivered two vaccine candidates (based on protein and DNA+protein) that meet all crucial parameters for a successful WNV vaccine in Europe and worldwide:
• 100% protective in mice and non-human primates
• Well tolerated in mice and non-human primates
• No side effects
• Cross-protective against WNV from lineage 1 and lineage 2
• Long-lived protective immune responses
• More immunogenic than existing technologies
• Easy production of vaccine antigen
Therefore, the requirements set out by the call in 2009 (i.e. tools advanced to late-term pre-clinical studies) were fully met by the project. These candidates have clear advantages over previous develop-ments, mostly in the USA (which were already tested clinically), because the antigen is a bacterial pro-tein which can be produced in a cheaper way and in higher quantities than the insect-derived antigens tested by others. In addition, due to the adjuvant, immunogenicity is clearly superior.
A vaccine will be the most effective control measure of WNV infections, particularly in the at-risk popu-lation (the elderly and immunocompromised individuals). Vaccines to similar flaviviruses, such as TBEV, have helped to efficiently control these infections or even diminish case numbers to zero in some areas in Europe. Due to the excellent features of the WINGS vaccine candidate in immunogenicity and safety, it can immediately enter the last tests before the application in humans. Especially as the vaccine developed in WINGS protects from lineage 1 and 2 infections, it will be functional worldwide. However, in light of the high costs associated with a clinical vaccine development, the sporadic nature of out-breaks and the resulting difficulties in efficacy testing, the further implementation of this technology needs partnership with the pharmaceutical industry, at best in an open discussion with European au-thorities.

Several diagnostics test systems are available for WNV, which rely on the detection of either viral RNA or antibodies produced against the virus. As the virus is only shortly detectable in serum and has often already disappeared when disease symptoms start, measurement of its RNA is challenging. On the other hand, detection of WNV-antibodies is complicated by the structural similarity of many flaviviruses (e.g. TBEV or DENV), which leads to cross-reactivities in serological diagnosis. Therefore, the WINGS consor-tium focused on alternative techniques to detect the virus in patients and to measure WNV-antibodies for a rapid and specific diagnosis.
A system to detect/isolate WNV in urine was developed and the protocol distributed to diagnostic la-boratories and networks, also towards the European Centre for Disease Prevention and Control (ECDC). The protocol allows for a detection of viral RNA and even infectious viruses several days after onset of symptoms. This system will have substantial impact on the routine diagnostic of WNV infections, by increasing the rate of cases confirmed by molecular tests and avoiding confirmation by the time con-suming neutralization tests. In addition, urine sample can be collected abundantly and non-invasively from patients. WNV-detection in urine, based on the protocol developed in WINGS, is now included into the case definition criteria by the ECDC.
In addition, a serological diagnostic platform was established based on mutant viral E-proteins. This is currently transformed into an ELISA-based test-format for the detection of IgG and IgM antibodies in human sera. The test system meets the following criteria:
• Sensitive and specific detection of WNV infections in Europe and America
• Detection of WNV lineages 1 and 2
• Discrimination between different flavivirus infections (DENV, TBEV, JEV)
The major problem with the existing serological tests is the cross-reactivity displayed by WNV-antibodies with other, related flaviviruses, such as TBEV or Dengue. Especially in areas where several flaviviruses co-circulate (e.g. WNV and TBEV in Northern Italy) or where people are vaccinated with existing flavivirus-vaccines (such as against YFV or TBEV) this is of significant concern. All current tests cannot rule out cross-reactions and have therefore to be interpreted with caution. As a consequence, positive results using the existing tests have to be confirmed by virus neutralization tests. These use live viruses and have therefore to be performed in high-security laboratories (biosafety-level 3, BSL-3), which is time consuming and expensive.
The method developed in WINGS allows for the secure differentiation of WNV-infections from other flaviviruses and makes virological confirmation assays obsolete. After implementation through a part-nership with the diagnostics industry (bilateral discussions are in progress) it will make diagnosis sig-nificantly faster and cheaper. Importantly, tests can also be performed in less well-structured areas where no BSL-3 laboratory is present.
In addition, the first systematic study of the human antibody response to WNV infections was undertak-en during the WINGS project. The results demonstrate a significant degree of heterogeneity in the hu-moral immune response to the virus, which was not different in samples from Europe and the USA or Canada. These results have important implications for the further development of vaccines and diag-nostic tests: despite the fact that some antigenic structures are clearly dominant, it has to be taken into account that not all humans develop antibodies against the same parts of WNV.

Using state-of-the-art technologies as well as developments of WINGS (e.g. WNV isolation from urine samples) emerging strains in Italy were investigated. The varying distribution of lineage 1 and lineage 2 viruses was demonstrated as well as co-circulation of different strains within the same area. Moreover, the protein sequences of these viruses were determined and several mutations were found. It is critical to study these novel proteins for different biochemical features, as these might make the virus more virulent, for instance via an increased temperature stability. Biochemical tests have indeed shown that some of the mutations identified could render WNV more pathogenic. These results are important to understand how a WNV strain which previously was considered of low pathogenicity suddenly causes an outbreak.
Taken together, the results of WINGS will contribute significantly to the enhancement of Europe´ pre-paredness for WNV as well as to the current countermeasures taken by the regulatory authorities worldwide.

Exploitation of results
The developments of WINGS were IP protected and published. In collaboration with the European Commission and stakeholders of the vaccine industry ways are being designed to start the clinical de-velopment (i.e. the testing in humans) in order to have a usable product for protection of humans from WNV as soon as possible.

The methodology to detect WNV RNA and isolate the virus from urine was published and has been dis-tributed to diagnostic authorities and laboratories in Europe. It is already being used in a variety of countries and greatly facilitates WNV diagnosis.
The serological test system for measuring WNV-antibodies was also IP protected and published. It is being developed into a marketable product (ELISA based test-kit) in collaboration with the diagnostics industry.
In addition, the library spanning the WNV proteome was published and is open for the scientific com-munity for collaborations on the further study of the humoral immune response to WNV. Potential fu-ture projects using this system might address the avian hosts or other WNV susceptible animals such as horses.
The results obtained during the project (on vaccines, diagnostics and epidemiology) were all dissemi-nated in peer-reviewed publications and presented to the scientific community on various workshops, conferences etc. In addition, press releases, radio interviews etc. were used to also inform the public on results of WINGS.
The project´s website will stay online and will be updated on a regular basis for new publications and results derived from work done during the WINGS project.

List of Websites:
Dr. Sebastian Ulbert
Fraunhofer Institute for Cell Therapy and Immunology
Perlickstraße 1
04103 Leipzig, Germany
Tel: +49 341 35536-2106
Fax: +49 341 35536-9921