Pathotyping of European West Nile virus (WNV) strains and analysis of the impact of viral molecular diversity on host-pathogen interaction

  • Maha Dridi

Student thesis: Doc typesDoctor of Sciences


Since the re-emergence of West Nile virus (WNV) in the mid 1990’s in Europe and the Mediterranean Basin countries, a wide range of genetic variants has been circulating in those areas. European WNV isolates sometimes cause sporadic or recurrent outbreaks, and those neuro-invasive disease cases mostly affect human and equine hosts. Given the rare WNV-associated bird mortalities so far recorded in Europe, and despite their role as reservoir and amplifying hosts for WNV, European birds were only scarcely targeted for pathogenicity and transmission studies. Conversely, the single WNV genotype that was introduced to North-America (NY99), and the narrow spectrum of genetic variants that has later derived from it, quickly resulted in generalized epidemics affecting not only humans and horses, but also causing massive bird mortalities. In North-America, passerines, especially from the corvid family, were the most susceptible avian species to WNV infection, as evidenced by field observations and numerous experimental infections of Nearctic birds. In the highly sensitive American crows (Corvus brachyrhynchos), this increased pathogenicity was linked to a NS3-Thr249Pro mutation. In Eurasia, this mutation had also been reported in the sole WNV strain that had ever associated with remarkable bird mortalities, namely the Is98 strain that had caused goose mortalities in Israel in 1998. This mutation was also detected in the strains responsible for the most deadly European WNV outbreaks for humans, like those elicited by lineage 1 strains in Romania in 1996 and Russia in 1999, as well as by the lineage 2 strain in Greece in 2010. If lineage 1 WNV strains are widely distributed and were responsible for the majority of recent epidemics/epizootics in Europe, America and the rest of the world, lineage 2 strains had only been described in Africa, until 2004 when a lineage 2 isolate was reported in raptors in Hungary. This lineage 2 virus spread to Austria and had acquired a NS3-His249Pro mutation by the time it reached Greece in 2010. Nevertheless, the NS3-249Pro genotype has arisen recently in countries of the Western Mediterranean region, like Spain and Italy, but did not correlate with any enhanced virulence for birds, horses or humans, what questions the relevance of a NS3-249Pro genotype as a universal determinant of WNV virulence.
In this thesis, avian infection models for WNV infection using SPF chickens and wild-caught Carrion crows were developped for studying the pathogenicity of European WNV strains for local birds. Those same avian models were subsequently used to evaluate the relevance of the NS3-249Pro mutation as a virulence marker of WNV.
In the first part of this thesis (chapter I), the suitability of the SPF chicken for use as a model for WNV infection was investigated, and the best experimental parameters pertaining to this model were defined. For that purpose, four lineage 1 WNV strains were selected: Is98 and Kunjin that were hypothesized to be respectively highly and weakly pathogenic for birds, as well as It2008 and Fr2000 that were posited to be moderately pathogenic for birds. Different ages at inoculation (one day, one week or four weeks), viral doses (101, 102 or 103 TCID50) and routes of inoculation (subcutaneous (SC) or intracranial (IC)) were tested on SPF chickens in order to determine the best experimental setting for pathotyping the four assayed WNV strains. If chickens aged one or four weeks at inoculation did not develop signs of the disease, the mortality rates of chicks inoculated at the age of one day were directly correlated to the observed field pathogenicity of the tested WNV strains. The discrimination between the four WNV strains was sharpest for an IC inoculation of a dose of 103 TCID50 of virus. Those parameters were thus set as those defining the newly-developed SPF chicken model for WNV pathotyping. Subsequently, further viral infection parameters were explored in one-day-old and one-week-old SPF chickens infected by the SC with a dose of 103 TCID50 of Is98 or Fr2000. For that purpose, viral loads in serum and feathers were followed-up by a NS2a-specific rRT-PCR. However, those virological profiles were rather correlated to the age of the inoculated chicks than to the pathotype of the tested strain. Given that occasional spill-overs to dead-end hosts like humans and horses are only allowed when WNV-induced viremia in birds reach a critical level, the SPF chicken model for WNV infection cannot predict the zoonotic potential of a given WNV strain.
In the second part of this thesis (chapter II), the susceptibility for infection with Eurasian strains of WNV of a common Paleartic corvid, namely, the Carrion crow (Corvus corone), was investigated. A WNV infection protocol for wild-caught juvenile (i.e., less than one year of age) Carrion crows was therefore developed. Birds were SC-inoculated with a dose of 103.5 TCID50 of IS98 or Fr2000 and clinical and virological follow-ups were carried-out. Both WNV strains were pathogenic for Carrion crows, especially Is98 that induced a mortality rate of 100%. Conversely to the SPF chicken model, the differential pathogenicities of Is98 and Fr2000 for crows also reflected in the viral RNA load levels elicited in serum, feathers and oral secretions. These findings thus confirmed the predictivity of the SPF chicken model for the pathotype of a given WNV strain in a wild bird. Moreover, those experiments demonstrated that Paleartic corvids, just like their Nearctic counter-parts, are susceptible to WNV infection despite limited reports of corvids die-offs in the field in Europe. However, with an estimated maximum viremia that was limited to 104 TCID50/ml of serum (vs. more than 109 PFU/ml of serum in American crows ), Palearctic Carrion crows are not likely to act as an amplifying host for WNV in the case of a hypothetical natural outbreak.
The two first chapters of this thesis aimed at the development of avian models for WNV infection. Those models were used in a subsequent series of experiments that aimed at characterizing some of the putative molecular determinants of WNV virulence.
In the third part of this thesis (chapter III), the impact of the NS3-249Pro mutation on the pathogenicity of WNV in the SPF chick and Carrion crow models was studied. Firstly, three highly related (99% identity at the nucleotide level) lineage 2 viral isolates from Eastern Europe, namely Hun2004, Aus2008 and Gr2011, that yet differ amongst others as of the residue in position 249 of the NS3 (histidine or proline), were used. Hun2004 was associated with only a few avian deaths and is characterized by a NS3-249His genotype; Aus2008 was associated with significant outbreaks in birds, horses and humans and is also characterized by a NS3-249His genotype. Gr2011 directly derives from the Nea Santa Gr2010 strain that was responsible for the major human outbreak of neuroinvasive disease in Thessaloniki in 2010, and is characterized by a NS3-249Pro genotype. Unexpectedly, clinical and viral RNA loads monitoring in sera and feathers of SC-inoculated day-old SPF chicks showed that Gr2011 was the least pathogenic strain for the SPF chicken model, despite the presence of the NS3-249Pro genotype and its association with heavy human fatalities. On the one hand, this finding suggests that the NS3-249Pro genotype is neither sufficient nor necessary for the pathogenic potential of a given WNV strain, what suggests a polygenic foundation for WNV virulence. On the other hand, this observation confirms the predictability of the SPF chicken model for WNV pathogenicity in birds, as well as points out that the predictability of the SPF chicken model for WNV pathotypes is limited to avian hosts. In order to grasp more so accurately the true weight of the NS3-249Pro substitution as a virulence determinant, two cloned viruses that only differ as of the residue present at position 249 of the NS3 (proline or threonine) were subsequently used. Those cloned viruses had been derived from the Is98 strain that had elicited high rates of avian deaths in 1998 in Israel: (1) a wild type (WT Pro) clone bearing a NS3-249Pro genotype and (2) a mutant clone (mutant Thr) bearing a NS3-249Thr genotype. Both clones were characterized in the SPF chicken model as well as in the Carrion crow model. In chickens as well as in Carrion crows, WT Pro (NS3-249Pro) induced higher mortality rates and viral RNA loads in serum as well as in feathers than mutant Thr (NS3-249Thr). These results suggest that for a given viral genetic backbone, and notably for the Is98 strain, the NS3-249Pro mutation is indeed a determinant of WNV virulence for birds.
In a final study (chapter IV), we focused on the quasispecies structure of WNV. Particularly, we tried to investigate whether Carrion crows would modulate WNV quasispecies population dynamics in a pattern similar to the one observed in the day-old SPF chicken model, the latter being so far the sole avian species used to simulate WNV intra-host genetic evolution. For that purpose, we inoculated day-old SPF chicks and juvenile Carrion crows with a genetically homogeneous cloned virus, i.e. the WT Pro cloned virus. Genetic diversification was assessed by the deep-sequencing technique in the sera of both hosts after 2 and 4 days of infection. At both time-points, WNV was more diverse in Carrion crows than in SPF chicks. Indeed, in the latter host, minor genotypes variability and frequency did not differ much from that of the inoculum, whereas in crows, multiple novel minor variants emerged sporadically. Similarly to mosquitoes, which were previously shown to exert negligible selection pressure on WNV in comparison to SPF chickens, susceptible birds like Carrion crows may promote WNV evolution by allowing extensive genetic expansion. Hence, complementarily to the emergence of some point mutations that may confer enhanced virulence, the dynamics of WNV quasispecies when they circulate between different host species may also constitute a mechanism whereby WNV can undergo a pathogenic evolution.
In the light of the dramatic developments of the WNV issue in North-America, the local circulation of WNV in European countries appears as a latent threat that could broke into an explosive epidemic at any time. As suggested by our results, this feared pathogenic evolution cannot be easily predicted since the same single-point mutations are not always correlated with comparable impacts on virulence, and as passage in natural susceptible avian hosts seems to promote WNV genetic diversification. The research carried out in the frame of this thesis is coherent with the European and particularly Belgian preparedness efforts for dealing with a possible incursion of WNV or with existing outbreaks. We indeed developed an SPF chicken infection model that would allow the pathotyping of unknown WNV isolates, and we proved that the Carrion crow, which is the main species targeted by the Belgian WNV surveillance program, is susceptible to WNV infection.
Date of Award3 Nov 2014
Original languageEnglish
Awarding Institution
  • University of Namur
SupervisorBenoit Muylkens (Supervisor), JEAN-JACQUES LETESSON (President), UTE ZIEGLER (Jury), Stephan ZIENTARA (Jury), Alain Vanderplasschen (Jury), Thierry VAN DEN BERG (Jury) & Benedicte Lambrecht (Co-Supervisor)


  • West Nile virus
  • pathogenicity
  • specific pathogen free chicken
  • carrion crow
  • virulence molecular marker
  • quasispecies

Cite this