Serotype Specific Primers and Gel-Based RT-PCR Assays for ‘Typing’ African Horse Sickness Virus: Identification of Strains from Africa

African horse sickness is a devastating, transboundary animal disease, that is ‘listed’ by the Office International des Epizooties (OIE). Although attenuated, inactivated and subunit vaccines have been developed for African horse sickness virus (AHSV), these are serotype-specific and their effective deployment therefore relies on rapid and reliable identification of virus type. AHSV serotype is controlled by the specificity of interactions between neutralising antibodies, and components of the outer-capsid, particularly protein VP2 (encoded by AHSV genome segment 2 (Seg-2)). We report the development and evaluation of novel gel based reverse transcription-PCR (RT–PCR) assays targeting AHSV Seg-2, which can be used to very significantly increase the speed and reliability of detection and identification (compared to virus neutralisation tests) of the nine serotypes of AHSV. Primer sets were designed targeting regions of Seg-2 that are conserved between strains within each of the AHSV serotype (types 1 to 9). These assays were evaluated using multiple AHSV strains from the orbivirus reference collection at IAH (www.reoviridae.org/dsRNA_virus_proteins/ReoID/AHSV-isolates.htm). In each case the Seg-2 primers showed a high level of specificity and failed to cross-amplify the most closely related heterologous AHSV types, or other related orbiviruses (such as bluetongue virus (BTV), or equine encephalosis virus (EEV)). The assays are rapid and sensitive, and can be used to detect and type viral RNA in blood, tissue samples, or cultivated viral suspensions within 24 h. They were used to identify AHSV strains from recent outbreaks in sub-Saharan African countries. These methods also generate cDNAs suitable for sequencing and phylogenetic analyses of Seg-2, identifying distinct virus lineages within each virus-type and helping to identify strain movements/origins. The RT-PCR methods described here provide a robust and versatile tool for rapid and specific detection and identification of AHSV serotypes 1 to 9.


Introduction
African horse sickness virus is a distinct virus species within the genus Orbivirus (of which Bluetongue virus is the 'type species'), within the family Reoviridae. African horsesickness (AHS) is an important transboundary disease that is 'listed' by the Office International des Epizooties (OIE). It is an acute or sub-acute, non-contagious, arthropod-borne viral disease, characterized by severe pyrexia, widespread haemorrhages and oedematous exudations [1]. AHS has case fatality levels as high as 95% in horses and 80% in mules, while donkeys and zebras show few if any clinical signs and are thought to act as 'reservoir' hosts [2,3]. The African horse sickness virus (AHSV) can also infect large carnivores via an oral route and may be a factor in the viability of endangered wildlife populations [4,5].
AHS is endemic in sub-Saharan Africa [6] but single AHSV serotypes have occasionally expanded outside these boundaries, persisting in newly affected areas for a number of years [7]. In 1959-61 an outbreak caused by AHSV serotype 9 (AHSV-9) occurred in the Middle East, spreading as far as Pakistan and India, causing heavy mortality in equids [8]. In 1965, AHSV-9 caused an outbreak that started in North Africa but spread as far as southern Spain [7,9]. In the late 1980s importation of a zebra infected with AHSV-4 from Namibia, caused an outbreak in Spain and Portugal [7,10,11,12]. These incursions demonstrate that the geographic area suitable for successful transmission of AHSV is much greater than its current endemic zone and suggest that if AHS become fully established in Europe, it could inflict massive losses on the equine industry. Control measures have therefore been put in place by the European Community to regulate movements of equids [13] and combat this disease [14].
Bluetongue virus (BTV) and AHSV are transmitted by similar Culicoides vector species [16]. Changes in the global distribution of BTV, with massive outbreaks caused by multiple BTV serotypes throughout Europe (since 1998), incursions into the south-eastern USA by previously exotic virus types, and detection of BTV-2 and 7 in Australia [17,18] have been linked to a combination of increased international trade, recruitment of novel vector species, climate change and its effects on vector distribution [19,20]. There have also been changes in the distribution of Epizootic haemorrhagic disease virus (EHDV), AHSV, and equine encephalosis virus (EEV) [21], suggesting that they could also emerge to pose important threats to European/global livestock and wildlife populations.
Based on field observations, these different AHSV serotypes have previously been linked to differences in immunogenicity, production of different disease patterns and virulence [35,36]. However, the multi-segmented nature of the AHSV genome and the potential for field strains to efficiently exchange/reassort genome segments, suggest that these characteristics may not be solely due those genome segments or proteins that control serotype [37]. Recent experiences with the control of BTV-1 and 8 in northern Europe indicate that vaccination of horses against the AHSV serotype responsible for an outbreak would provide an effective control measure. However, rapid deployment of an appropriate vaccine strain depends on identification of the serotype involved, as quickly and accurately as possible.
Conventional serotyping methods rely on AHSV isolation from clinical specimens, in suckling mouse brain and adaptation to tissue culture (BHK-21, Vero or KC (Culicoides soronensis) cells), followed by serological assays (VNT), taking a minimum of 3-4 weeks. Alternatively it may be possible to identify the virus serotype by the reaction of antisera from an infected animal with reference strains of each AHSV serotype. However, these methods require access to stocks of well characterised reference antisera or reference virus strains for each serotype and cannot distinguish individual virus strains within a single orbivirus serotype. Consequently they are unsuitable for high precision 'molecular' epidemiological studies to differentiate closely related virus lineages, or topotypes. They can also be complicated by cocirculation of different viruses, leading to the generation of a broadly cross-reactive neutralising-antibody responses after sequential infection with more than one serotype [38]. The resulting delays in vaccine deployment could allow the early spread and consequently the duration of an epizootic.
Serotype-specific RT-PCR assays have been developed for the 26 serotypes of BTV [39,40,41]. Attempts to design similar primers and typing assays targeting Seg-2 of AHSV [42] have previously been hampered by a lack of full length Seg-2 sequence data for all AHSV serotypes. Generation of sequence data for Seg-2 from multiple reference and field strains of the nine serotypes of AHSV [31,43, Bankowska et al -manuscript in preparation] has clarified the level and extent of sequence variation within and between different serotypes.
We describe the design and in silico evaluation of Seg-2 based oligonucleotide primers and RT-PCR assays for each AHSV serotype. Assay specificity was evaluated with a range of AHSV isolates from dsRNA virus reference collection at IAH [44]. They were also used to identify AHSV strains circulating in East and West Africa, using RNA samples extracted directly from diagnostic blood or tissue samples. These techniques remove the need for virus isolation and 'reference antisera', increasing the speed and availability of virus identification methods, and reducing the use of experimental animals.

AHSV reference and field isolates
Different AHSV isolates (listed in Table 1) from the Institute for Animal Health (IAH) 'Orbivirus Reference Collection' (ORC) [44] were grown in BHK-21 clone 13 (European Collection of Animal cell Cultures (ECACC -84100501))/Vero (ECACC -84113001) cell monolayers, until 40-100% cytopathic effect (CPE) was observed and in C. sonorensis (KC) cells (originally provided by colleagues at the USDA lab in Laramie, Wyoming) for 7 days. The field samples used in these studies were taken from naturally infected animals in the field, by qualified veterinarians, as part of normal veterinary care and diagnostic testing procedures in the respective countries. No samples were taken from animals specifically for these studies, so we believe that ethical approval is not required.

Extraction of Viral dsRNA
RNA was extracted from BTV infected cell pellets using TrizolH (Invitrogen) [45], or from cell-free tissue culture supernatant using the QIAamp Viral RNA Mini Kit (Qiagen), or a Universal BioRobot (Qiagen), according to manufacturer's instructions.
Group specific RT-PCR assay RNA samples extracted from clinical samples or cell culture material, were tested in duplicate by conventional RT-PCR assays using AHSV genome segment 7 (Seg-7) specific primers [46]. Samples that generated a strong cDNA band of the expected size (1179 Kb), were considered to contain AHSV RNA and were used for further processing/testing.

Serotype specific RT-PCR assays
Multiple primer-pairs were designed based on sequence data for Seg-2, targeting each of the nine AHSV serotypes (Table S1). cDNA copies were reverse transcribed from denatured dsRNA templates, then amplified by One-Step RT-PCR kit (Qiagen) as described by Mertens et al. [41], or by a one-step RT-PCR system (Invitrogen) and high fidelity platinumH Taq as described by Maan et al [47]. In some cases, full length cDNA copies of AHSV Seg-2 were also synthesised and sequenced, as described by Maan et al [48]. RT-PCR typing results were also confirmed by sequencing of amplified cDNA products from Seg-2 and sequence data analysed using MAFFT [49] and/or Clustal W programs [50]. Phylogenetic comparisons were made using Neighbour-Joining (NJ) algorithms and the trees were constructed in Mega 4.1 [51] using nucleotide sequences in a pair-wise deletion, p-distance algorithm, and bootstrapped using 500 replicates.

Development of RT-PCR assays and their evaluation for type specificity
Comparisons of full-length Seg-2 sequences for reference strains of the nine AHSV serotypes and other field isolates [31,43,52,53, Bankowska et al -in preparation], were used to identify regions of conservation within individual AHSV serotypes and variability between different types. These regions were used to design a minimum of two serotype-specific primer-pairs for use in RT-PCR assays to detect each AHSV serotype (Table S1).
The RT-PCR primers were tested in reactions containing RNA extracted from BHK-21 cells infected with the different AHSV serotypes ( Table 1). The cDNAs generated, were analysed by 1% agarose gel electrophoresis (AGE), to check both their size and relative level of synthesis. In each case cDNAs of the expected size were generated from RNA of the homologous serotype ( Figure 1  panels A-D). None of the primer-pairs amplified cDNAs from RNA of even the most closely related heterologous AHSV types, or from members of other related Orbivirus species (data not shown). These initial evaluations confirm in silico analyses, indicating that the primers listed in Table S1 are indeed 'type-specific'. Identification of AHSV-2 isolates from Senegal 2007. RNA was extracted from equine samples, or from cell culture supernatants infected with these viruses (SEN2007/01, SEN2007/02, SEN2007/ 03, SEN2007/04 and SEN2007/05). In each case a strong +ve band of the expected size (1179 Kb) was generated in Seg-7 group-specific RT-PCR assays (Figure 2 -panel A). The RNA samples were also tested by conventional RT-PCR, using Seg-2 specific primers listed in Table S1. Strong +ve bands of the expected sizes were only generated with primer-pairs '2A1' and '2A2', demonstrating the presence of AHSV-2 RNA. None of the other AHSV types were detected, including AHSV-9 (which had recently been used to vaccinate the animals) ( Figure 3). The AHSV-2 Seg-2 amplicons were sequenced, generating data for nucleotides (nt) 500-1648, which showed 97% nt identity with Seg-2 of the South African reference strain of AHSV-2 (Ac. No. AY163332) (Figure 4).
Identification of AHSV-7 from Senegal 2007. AHSV isolate (SEN2007/06) derived from a horse that died in 2007, was also tested in type-specific RT-PCR assays for each of the nine AHSV serotypes. Assays containing only one of the type 7 specific primers ('7A1') (Table S1) generated a band of the expected size (1426 bp) indicating that this sample contains AHSV-7 ( Figure 3 panel C). Although the other type 7 specific primers ('7A2') worked well with the reference strain RSArah7/03 and Ethiopian strain (ETH2010/09), they failed to amplify sequences from SEN2007/06, suggesting a significant variation in the footprint sequence for this primer-pair compared to the reference strain ( Table S1). The '7A1' Seg-2 amplicon (nt 550-1800) from SEN2007/06, was also sequenced, showing 92% nt identity overall to Seg-2 of the South African AHSV-7 reference strain (Ac. No. AY163330) confirming the isolate as AHSV-7 ( Figure 4).

Discussion
AHSV Seg-2 encodes VP2 (the main serotype-specific antigen and most variable of the virus proteins) and shows 47.6% to 71.4% nt identity between types, which can therefore be identified by phylogenetic comparisons [31,43,. Cross-hybridization methods [54] or hybridization and realtime RT-PCR methods targeting AHSV Seg-2 have previously been described [55,56]. Sailleau, et al [42] have also reported typing of AHSV using a serogroup-specific forward-primer and a serotype-specific reverse-primer, based on a limited Seg-2 dataset. These primer-pairs generated a small amplicon for each of nine serotypes of AHSV and were used to 'type' a limited number of reference and field strains. The RT-PCR assay results obtained were also compared with VNT results but were not confirmed by sequencing of the resulting amplicons.
Recent full length sequencing of Seg-2 from reference strains and multiple field strains of AHSV (Bankowska et al -in preparation), have supported the design of additional AHSV 'typespecific' primer-pairs. We have successfully evaluated these primers, RT-PCR assays, sequencing and phylogenetic analyses, for differentiation of individual AHSV strains even within a single serotype. These techniques can determine AHSV serotype with greater accuracy, sensitivity and significantly more quickly than is possible by current serological techniques, facilitating surveillance and molecular epidemiology studies of the virus.
The relatively small number of AHSV isolates (N = 26 (Table 1); N = 107 in silico (Table S1)) that are available for sequence analyses and testing, has limited wider validation of the assays developed reported here. The specificity of the primers therefore primarily reflects their failure to recognise and amplify Seg-2 of other AHSV serotypes, rather than an ability to detect all possible isolates of the homologous type. If in future studies, any of these primers fail to detect Seg-2 of a novel isolate, it will be sequenced, providing a basis for further 'evolution' of primer sequences and assays. The 'fine-scale' differentiation and identification of individual AHSV strains, depends on the availability and characterisation of well documented reference isolates from different locations and belonging to different virus lineages. These need to be stored and maintained in a suitable reference collection (like the ORC) with links to relevant sequence data [44].
Recent disease outbreaks caused by AHSV-4 in Kenya (in 2007); AHSV-2, AHSV-7 and AHSV-9 in Senegal (in 2007), were identified using the 'type-specific' assays (as described here). Serotype 9 was also identified in the Gambia (in 2009) along with multiple serotypes in Ethiopia (in 2010) using the same techniques [15], Bankowska et al, in preparation]. These novel laboratory . Neighbour-Joining tree, showing relationships in Seg-2 between reference and field strains of AHSV. The tree was constructed using distance matrices, generated using the p-distance determination algorithm in MEGA 4.1 (500 bootstrap replicates) [51]. AHSV split into distinct groups based on Seg-2 sequences, reflecting serological relationships between virus strains [31]. Reference strains of AHSV are shown in black and field isolates are shown in blue font. Isolate designations: IAH 'dsRNA virus reference collection number composed of 'country code, year, and the number of the isolate in that year from that country [44]. Seg-2 sequences of two Senegalese AHSV-2 field 2007 strains showed 99.8% nucleotide (nt) identity to each other and showed 97% nt identity to the AHSV-2 reference strain (500-1648 nt). AHSV-4/KEN2007/01 showed 96% nucleotide identity to the AHSV-4 published Seq-2 sequence in the upstream 1170 base pairs (480-1650 nt). Seg-2 (550-1800 nt) of Senegalese AHSV-7 strain from 2007 showed 92% identity to the AHSV-7 South African reference strain. Seg-2 (520-1684 nt) of Senegalese vaccine strain of type 9 showed 98.8% identity to the AHSV-9 South African reference strain. Scale represents number of substitutions per site. Values at major branching points represent NJ bootstraps. doi:10.1371/journal.pone.0025686.g004 tools and the standardised protocol are easy to follow and do not require reference antisera or virus isolates that may have limited availability and which themselves represent a potential disease security risk. These methods are robust and versatile, reliably identifying AHSV type within 24 h of sample receipt, with no evidence of cross-amplification of other related types or other orbiviruses. Although slower than real-time RT-PCR assays, these conventional type-specific primers and assays can be used to generate cDNA amplicons for sequence analyses, providing definitive proof of relationships between Seg-2 of individual virus strains.
These phylogenetic comparisons will help to determine the prevalence, distribution and movements of different AHSV serotypes and topotypes, improving our understanding of the epidemiology and transmission of the virus, providing a basis for the design and rapid implementation of control strategies, including vaccination.

Supporting Information
Table S1 Primers for specific amplification of Seg-2 from various AHSV serotypes in RT-PCR assays. (DOC)