Avian reovirus (ARV) pathogens are found worldwide 1 2 3 and their molecular characterization, using PCR and nucleotide sequence analysis, have been described 4 5 6 . Yet, only a limited number of genomic sequences belonging mainly to the S- class, has been published. These sequences have characterized strains isolated from USA, Taiwan, Germany, Netherlands, Australia, Japan and China 7 8 . Recently, σC protein sequences of some Indian strains were submitted to the Genbank.

Particles of ARV present two layers of capsid and 10 segments of double-stranded RNA 9 . Genomic segments can be separated on polyacrylamide gel electrophoresis into three different classes, based on size, named L (large), M (medium) and S (small). They encode for at least 8 structural and four non-structural proteins 10 11 . Amongst the S-class segments of ARV, the segment S1 contains three open reading frames that are translated into P10, P17 and σC proteins 10 11 . The protein P10 induces cell-cell fusion 12 13 while P17 causes cellular protein translation shutoff and cell cycle arrest 14 15 . The σC protein plays a role for virus attachment 16 and as apoptosis inducer 17 . It has been demonstrated that cell entry of avian reovirus follows a caveolin-1-mediated and dynamin-2-dependent endocytic pathways that require activation of p38 MAPK and Src signaling pathways 18 . It has also been suggested that it represents the target for type-specific neutralizing antibodies. The σB protein, encoded by the S3 gene 10 19 , carries group-specific neutralizing epitopes 20 . Recently, the ARV σA protein, encoded by the S2 gene 16 , has been identified as a double-stranded RNA binding protein that may be involved in interferon resistance 21 . Another viral protein, σNS encoded by the S4 gene, has been reported as having a single-stranded RNA binding activity 22 23 .

This paper reports the analyses of the σC and part of the σB encoding genes along with the nucleotide sequences of 15 ARV isolates, identified over a 12-year period. To the best of our knowledge, this is the first report regarding the genotypic classification of ARV in Tunisia and in Africa. Because ARV are important poultry pathogens, causing arthritis, chronic respiratory disease and malabsorption syndrome, which result in considerable economic losses to the poultry industry 24 25 , better understanding their pathogenicity is economically important.

Being the major laboratory doing diagnostic and research for poultry diseases in Tunisia, samples from different flocks and regions were sent to the laboratory, by private and governmental veterinarians, for the diagnostic of major avian diseases. Fifteen ARV isolates were identified from chickens with viral arthritis, malabsorption syndrome or other suspected symptoms. The flocks were located in governments in the north (Bizerte, Ben Arous and Nabeul), the center (Sousse, Monastir and Mahdia) and the south (Sfax) of the country. ARV were propagated in specific pathogen-free (SPF) embryonnated chicken eggs, followed in cultures of primary chicken embryo fibroblast (CEF) or liver (CEL) cells 7 . The virus was detected when strong cytopathic effects (CPE) were observed after 3 to 5 subcultures; virus stocks were subsequently prepared in 75 cm2 flasks. Once 70-80% CPE were developed, cell cultures were frozen and thawed 3 times, then clarified with low centrifugation at 3000rpm for 20min. Supernatants were then conserved in aliquots at −80°C for later use. The vaccine strain (Nobilis S1133) was propagated in CEF cells and treated as described above. Viral RNA was extracted from supernatants using Trizol (Gibco Brl) as per the procedural modification described by Lee et al. 26 . Briefly, 1ml of Trizol was incubated with 300 μl of clarified supernatant for 5min, at room temperature. RNA was then isolated with chloroform (Sigma-Aldrich, France) and precipitated with isopropan-alcohol (Sigma-Aldrich, France) overnight. Total RNA was used as a template for RT-PCR reactions; non infected-cells and Rnase free water were used as negative controls.

Fifteen Tunisian ARV were isolated (Table 2) in SPF eggs, CEF and/or CEL cell cultures. S1133 vaccine strain, used as a control, was propagated in CEF cultures. Aliquots of clarified freeze-thawed cell supernatants (4 or 5 passages) were used for RNA extraction. Using primer pairs of S1C-S1D, S1E-S1F, S1G-S1H, S1I-S1H for the σC gene and P1-P2 for the σB gene, RNA extracts from all ARV isolates as well as the S1133 strain was transcribed into cDNA and amplified, producing fragments with expected sizes of 738, 342, 720, 500 and 672 bp, respectively. PCR products from S1133 strain and the Tunisian isolates were obtained, indicating that primer regions were conserved in our strains. RNA extracted from non-infected cells and Rnase-free water were used as negative controls and no DNA amplification was observed, indicating that the amplified viral DNA was specific and did not originate from contamination.

ARV pathogenicity is very heterogeneous and ARV strains are associated with disease conditions such as viral arthritis (VA), tenosynovitis and Malabsorption Syndrome (MAS). They are also isolated from chickens without any clinical signs 35 . Although the relationship between reovirus and tenosynovitis has been established, a causative role is less clear in MAS 36 . The characterization and the classification of ARV might be important for the epidemiology of MAS and VA for the choice of the appropriate vaccine in farms with ARV problems. Although ARV infections still cause problems in poultry, the only available ARV sequences are for strains from the USA, Canada, Taiwan, Australia, the Netherlands, Germany, Japan, England and China. Recently, σC encoded gene sequences from Indian isolates became available in Genbank 7 8 (Table 2).

Genotyping of ARV using σC encoded gene revealed at least 6 different genotypes 27 , but correlations between genotypes, pathogenic characteristics and serotype classification have not been well established yet. A phylogenic study using the σC encoded gene indicated that, despite the close relationship with the first USA isolates belonging to clusters I and II, those recently isolated from chickens with Runting-Stuning Syndrome (RSS) showed dispersion and were classified in the 3 different clusters III, IV and VI (Table 2, Figure 2b). They remain quite different from Australian and Dutch strains, classified in the clusters V and VI, respectively. The first strains isolated from Taiwan were classified in cluster I, whereas the more recent ones were dispersed between the 3 clusters II, III, and V. Furthermore, the German strains isolated during the last two years are much more dispersed and classified into 5 clusters. Finally, the trypsin-resistant strain JR1 is classified in the cluster I 8 27 33 .

Molecular characterization of the σC sequences of all Tunisian ARV, isolated from different field conditions over a twelve years period and compared to several published strains, allowed their classification in one cluster. However, no correlation could be established between the σC gene sequences and the disease conditions in the field. ARV were also isolated from apparently healthy chickens, making their classification in a true lineage difficult as stated by Kant et al. 7 .

The σC-encoding gene has evolved into six clusters, while the other S-class genes have diverged into two to four clusters 27 . It displayed features of the highest level of sequence divergence and rapid evolution. Therefore, this gene could be used as a genetic marker for the classification of ARV isolates 27 28 . Such classification in clusters is independent of determination of the 5' end, the 3' end or the whole ORF of the σC gene. Furthermore, sequencing of a representative part of this gene seems to be sufficient 7 .

The major part of σC gene sequences of the Tunisian isolates was closely related to the very well-known American strain S1133. Only nucleotide sequences of the 97.1 and 97.2 strains showed mutations at the positions 354 and 355, respectively. A relationship between the 97.1 and JR1 strains is possible; however, data about the σB gene sequence of the later is not yet published for a possible gene comparison.

Analyses of the whole σB-encoding gene were performed to allow rapid detection of ARV infections with different Tunisian isolates. The constructed Phylogenic trees showed in previous studies that the gene evolves into 2–3 lineages 8 27 . Analyses of the published sequences demonstrated that it evolves into 3 clusters, results obtained when using sequences of the whole gene or the representative sequence used in our work. Recently, we performed a restriction fragment length polymorphism (RFLP) with two restriction enzymes that cut in two sites among those containing mutations on our amplicons (data no shown). PCR/RFLP can be a procedure used for rapid characterization and differentiation of Tunisian ARV isolates.

Mutations do exist in the σB gene sequences of Tunisian strains, allowing their classification into 4 groups: a first group containing the TU430 and TU360 isolates; a second one with the TU429, TU96 and TU119 isolates; a third group containing the TU339, TU87, TU5, TU420, TU105B6, TU1390, TU71B1, TU435 and TU97.1 isolates and the fourth group with only the TU97.2 isolate. The substitution at position 467 was observed for all ARV with respect to the S1133 strain.

Sigma B protein was described as a high variable protein and possesses group-specific neutralizing epitopes 37 . Alignment analysis of amino acids of part of the σB protein showed that only the substitution at position 171 is silent, whereas all other mutations at positions 308, 438, 467, 521 and 656 are non-synonymous and induced substitution of the corresponding translated AA (Figure 1b). These mutations were also observed for the Chinese isolate 99G and indicated a possible relationship between the Tunisian and the Chinese ARVs. Unfortunately, the σC gene sequence of the 99G isolate is not available for comparison. For the Taiwanese strains, the majority of nucleotide substitutions were silent because they occur at the third position of the codons 27 .

The σC protein, located on the surface of the outer layer capsid, induces neutralizing antibody production and is the determinant for ARV serotypes 16 . In previous studies, ARV strains were differentiated by virus neutralization assays 38 . Serological classification of ARV strains has not been successful because of high cross-reactivity of the neutralizing antibodies 33 . Nowadays, genotypic classification of ARV strains is performed using RT-PCR in combination with phylogenic analysis or other molecular techniques, such as RFLP 7 26 . All phylogenic studies have classified ARV isolates into various groups and lineages; however, meta-analytically, there were no identical patterns. Such results suggested that different ARV genome segments may evolve in independent manner 27 . It is puzzling that no correlation was found between genotypes, serotypes and pathotypes, given detectable genetic differences of different ARV strains 7 . Divergences between genotypes, serotypes and pathogenicity suggested the involvement of multiple genes and proteins in serologic and pathogenic determination 33 . It may also result from high mutation rate of viral RNA and the possible reassortment between 2 strains co-infecting the same host 27 39 . Geographic characteristic conditions could facilitate exchanges between segmented genomes within small numbers of viral strains, which would result in less genetic variation within a region 33 .

Vaccination against reovirus infections generally involves a live S1133 vaccine followed by an inactivated vaccine containing strains 1733 and/or 2408, knowing to be related to tenosynovitis and stunting syndrome pathologic conditions and belonging to the same subtype 40 41 . Although chickens were vaccinated, ARV pathogens still persist in Tunisian poultry. Serotyping studies are therefore needed to further elucidate the antigenic characteristics of these isolates and adapt the vaccination program.

In the present study, we characterize genetically avian reovirus Tunisian strains isolated from commercial chickens reared in different geographic areas, during a 12 year period. The classification of ARV could be realized using the σC gene, the most variable known gene but other gene studies may provide additional information. It was not possible, however, to correlate the genotype of isolated strains with neither a specific pathologic conditions nor a geographic or temporal parameter. Although the vaccination against ARV is used in farms, the pathogens still persist in Tunisian poultry flocks. Serotyping studies are underway to characterize the antigenic proprieties of these isolates, which will help adapting the vaccination program.

AA: Amino acid; ARV: Avian reoviruses; SPF: Specific pathogen free; CEF: Chicken embryo fibroblast; CELC: Chicken embryo liver cells; CPE: Cytopathic effect; ARV: Avian reovirus; RT: Reverse transcription; N-PCR: Nested PCR; RSS: Runting-Stuning Syndrome; MAS: Malabsorption Syndrome; RFLP: Restriction fragment length polymorphism.

The authors declare that they have no competing interests.

HKY is a PhD student, who carried out the avian reovirus genes detection by RT-PCR, was involved in sequencing and phylogenic studies and drafted the manuscript. BH and GL helped with experimental procedures and manuscript preparation. SAD reviewed the manuscript critically for important intellectual content. GA conceptualized the study, was involved in design of the trials, supervised all facets of the research and assisted in the writing process. All authors have red and approved the final manuscript.