Genotyping of African swine fever virus (ASFV) isolates associated with disease outbreaks in Uganda in 2007

,


INTRODUCTION
African swine fever (ASF) is caused by a virus, ancestrally associated with argasid ticks and wild suids that can infect domestic pigs, frequently resulting in a rapidly lethal disease.However, depending on the combination of virus genotype and pig breed a range of outcomes varying from acute to chronic disease and apparently asymptomatic animals that are carriers of the virus can be observed.Virulent genotypes induce acute haemorrhagic disease, with symptoms including high fever, heamorrhages in the skin and internal organs and death within three to ten days.
ASFV is classified within the Asfarviridae family, genus Asfivirus (Dixon et al., 2000(Dixon et al., , 2005)).It is highly contagious and can spread very rapidly in pig populations by direct or indirect contact.The disease has been reported in *Corresponding author.E-mail: gallardo@inia.es more than twenty African countries, where the virus is maintained either through a sylvatic cycle involving warthogs (Phacochoerus aethiopicus) and soft ticks in the genus Ornithodoros or in a domestic cycle that involves pigs of local breeds, with or without tick involvement (Haresnape and Wilkinson, 1989;Plowright and Pierce, 1969;Sanchez-Botija, 1963;Thomson, 1985).ASF is endemic in most of sub-Saharan Africa, including the island of Madagascar; the highest incidence of disease being recorded from the equator to the northern Transvaal in southern Africa.Disease outbreaks have also occurred in Europe, South America and the Caribbean.In 2007, it was introduced into Georgia, most probably through infected pig-meat that was unloaded from a ship, recycled and fed to local pigs and has since spread throughout the Caucasus and into southern Russia (Rowlands et al., 2008).
Uganda, situated in the eastern-central part of Africa, is divided into 80 districts and shares borders with Kenya, the Sudan, DRC, Rwanda and Tanzania.Currently, African swine fever is officially described as endemic in the whole country and outbreaks are reported sporadically, but without quantitative data on prevalence.Although, no new ASF outbreaks have been officially notified since 2004 (World Animal Health Organization-World Organisation for Animal Health (WHO-OIE), occurrence of hemorrhagic disease has been reported regularly to the Ugandan veterinary authorities by pig farmers.A more recent ASF outbreak occurred in Uganda-Kenyan border (Busia district) in 2007, placing a significant proportion of the Ugandan pig population at risk.In October 2007, an outbreak of virulent hemorrhagic disease was notified in three districts of Central Uganda, Mukono, Wakiso and Nakasongola.Following these outbreaks, blood, serum and tissue samples from pigs with clinical signs of hemorrhagic disease were collected for African swine fever diagnosis and molecular characterization.
We report here the diagnosis and genetic characterization of ASF viruses collected during outbreaks in Uganda in 2007.The genotyping involved sequencing the 3' end of the gene encoding the p72 protein (Bastos et al., 2003;Boshoff et al., 2007;Lubisi et al., 2005) and the full length p54-gene (Gallardo et al., 2009) to place isolates into major subgroups, followed by higher resolution sub-typing through analysis of tandem repeat sequences (TRS) in the central variable region (CVR) of the ASFV B602L gene (Irusta et al., 1996;Lubisi et al., 2007;Nix et al., 2006;Phologane et al., 2005).The very close genetic similarity of isolates associated with recent disease outbreaks in both Uganda and Kenya emphasizes the value of molecular epidemiology for tracing the source and dynamics of ASF infections.

Sampling
During October 2007, the presence of virulent hemorrhagic disease suspected as being caused by ASFV was described by Uganda authorities in Central Uganda districts of Mukono, Wakiso and Nakasongola.Eighty four samples were collected by Ministry of Agriculture, Animal Industry and Fisheries (MAAIF) staff from pigs with clinical symptoms and submitted to the Animal Health Research Centre (CISA-INIA), Valdeolmos, Spain (EU ASF Reference Laboratory).Samples collected comprised sera from 39 pigs (Wakiso n = 13; Nakasangola n = 11; Mukono n = 15), EDTA blood samples from 39 pigs (Wakiso n = 11; Nakasangola n = 9; Mukono n = 19) and six tissue samples from one pig (Wakiso n = 6).They were received at CISA-INIA on 23 rd October 2007.

ASF antibody detection
39 field sera were tested using OIE (2008) prescribed tests (conventional ELISA and Immunoblotting assays) using as antigen a lysate of monkey stable (MS) cells infected with ASF Spanish isolate E70MS48 and protein-A/HRPO as the reporter (OIE 2008).Gallardo et al. 3489 Nucleic acid extraction and genomic DNA amplification DNA was extracted directly from 35 sera, 39 blood samples and 10% suspensions of 6 ground tissues (from one animal) using a nucleic acid extraction kit (Nucleospin/ Machery-Nagel-Cultek) following the manufacturer's procedures.A polymerase chain reaction amplification (PCR) assay using the ASF diagnostic primers PPA1/PPA2, which generates an amplicon of 257 bp within the p72 protein (Aguero et al., 2003), was used to confirm the presence of ASFV DNA.The PCR products were analyzed by electrophoresis through 2% agarose gels visualized under UV light.

Restriction enzyme analysis
The NdeI restriction endonuclease (New England Biolabs) was used to analyze the DNA amplified by PCR.Briefly, 10 µl of the PCR sample was digested for 1 h at 37°C with 8 to 20 U of restriction enzyme in a total reaction volume of 20 µl in the appropriate restriction enzyme buffer.Aliquots of 10 µl of each reaction mixture were size fractionated on a 3% agarose gel and the restriction patterns were visualized under UV light.

Virus isolation
Primary leukocyte cultures were used for the isolation of samples recovered from naïve domestic pigs as previously described (Malmquist and Hay, 1960).Briefly, cells were seeded into 96-well tissue culture grade microtitre plates (200 µl; 300,000 cells per well) in homologous swine serum and incubated in a humidified atmosphere containing 5% CO2 at 37°C.Three day cultures were infected at 1:10 dilution with serum or 10% suspensions of ground tissues supplemented with 5 µg/ml gentamycine sulphate (BioWhittaker) and incubated for 24 h at 37°C.After inoculation, a preparation of 1% homologous red blood cells (RBC) in buffered saline was added to each well.The plates were examined for haemadsorption over a 6 day period.The samples were blind passaged three times.

Genomic amplification and nucleotide sequencing
For molecular epidemiological analysis, DNA was extracted from cell cultures and three separate polymerase chain reactions (PCRs) were set up; for p72 genotyping classification amplifying the Cterminal region of p72 protein using the primers p72-U/D as previously described (Bastos et al., 2003) for p54 genotyping amplifying the complete gene encoding the p54 protein using the primers PPA722/PPA89 (Gallardo et al., 2009); for CVR sub-typing using the primer pairs ORF9L-F/9L-R to amplify the CVR located in the B602L gene (Nix et al., 2006).Amplicons of the expected size were excised, purified by Quiaex gel extraction (QIAGEN) and cloned into a pGEMT-Easy vector (Promega) according the manufacturer's instructions.The nucleotide sequence of the purified products was determined using SP6/T7 primers specific for the pGEMT vector using an automated 3730 genetic analyzer (Applied biosystems).

ASF diagnosis
Among the 39 porcine serum samples received at CISA from the Central Ugandan Districts, ASFV antibody detection performed using the OIE-prescribed ELISA and immunoblotting (IB) assays revealed that, none of them were antibody-positive for ASF.
For ASFV detection, PCR was performed on 35 serum samples, 39 blood samples and also 6 individual tissues from one of the sampled animals.Following PCR amplification, a single major amplicon of approximately 260 bp was generated from 9 of the 35 sera, 17 of the 39 blood samples and all tissue samples examined.Virus isolation was performed using homogenenised pooled tissues and also sera and blood from each PCR positive animal.Seven ASFV strains with a haemadsorption pattern typical of virulent ASF viruses were isolated from the Ugandan samples after three passages in leukocytes (Table 1).In four cases these virus isolates were made directly from serum samples that were negative according to the OIE ELISA and IB (Table 1).DNA was extracted from the cell cultures and the presence of the ASF virus was confirmed using diagnostic PCR.The sequencing and compared with sequences available in the GenBank and with additional sequences obtained in a separate study from ASF isolates recovered during ASF outbreaks that occurred in 2003 and 2006 to 2007, in Uganda and Kenya, respectively (previously unpublished data held at CISA-INIA).Nucleotide sequence analysis of the genome region delimited by the PPA-1/2 primers revealed an NdeI restriction endonuclease site, which which could generate two fragments of 204 and 53 bp in length.NdeI digestion of the amplicons from the Ugandan ASFV isolates resulted in generation of the two expected specific restriction fragments (Figure 1).This site is conserved only in ASFV isolates obtained from eastern African countries, while in sequences from European, Caribbean, West and South African countries a mutation is present at bp 54 within the NdeI restriction site (Table 2).

ASF molecular characterization
Analysis of the p72-gene partial sequences from each of the seven Uganda ASFV isolates showed that, they were identical at the nucleotide level (results not shown).Comparison of these sequences to the 22 (I-XXII) currently known major p72 genotypes (Boshoff et al., 2007) identified the Uganda 2007 sequences as falling within p72 genotype IX (Figure 2).Recent studies have demonstrated the value of p54-gene sequencing as an additional, intermediate-reso-lution, molecular epidemiological tool for typing of ASFV viruses (Gallardo et al., 2009).Amplification of the fragment containing the complete p54gene from all the Ugandan isolates produced PCR products of 550 bp, which were identical in sequence.The sequences of the 7 Uganda isolates were compared with 155 p54 ASFV sequences available in GenBank.As was the case with the p72genotyping, the Uganda isolates were clustered together with Ugandan and Kenyan viruses obtained from domestic pigs (Figure 3).
In an attempt to discriminate genotypes at higher resoamplified products were characterized by nucleotide lution, the CVR sequences of viruses from Uganda were     Uganda isolates (UGA 95/1; Ug03H1-3, Ug03P4-6) and Kenya isolates (Ken06B1-5, Ken06.Bus, Ken06.Kis, Ken07Eld1-2, Ken07.Nak, Ken07.Kia) within CVR subgroup XXIV.Differences were identified mainly in the number of tetrameric amino acid repeats.When the tetrameric repeats within the B602L gene from the Uganda isolates were compared with viruses included within the CVR sub-group XXIV (Nix et al., 2006)  those obtained from the second outbreak that occurred in western Kenya in 2006 (Ken06.Bus) and to all isolates from subsequent outbreaks in the country (Figure 4).

DISCUSSION
Outbreaks of African swine fever have been reported in Uganda sporadically, but regularly in specific areas since 1996 and the pig industry has been very significantly affected.The last major outbreak, according to the OIE reports was in 2004 when an estimated 7,500 pigs were destroyed or slaughtered in twelve districts across the country.Since then, outbreaks of ASF have occurred on a regular basis in Uganda, causing serious losses to both smallholder pig keepers and commercial pig farms.At present, ASF is officially described as "present in the whole country but without quantitative data on the prevalence of the virus".Interestingly, no positive antibody response was detected using OIE-prescribed serological methods in any of the serum samples collected from these outbreaks.This data reconfirms previous studies that describe a low incidence of detectable serological response to ASFV infection in pigs from east African countries using both recombinant antigen and OIE-prescribed African swine fever serological methods (Perez-Filgueira et al., 2006).Since the OIE-prescribed methods are based on the use of total ASF virus as antigen that includes many viral proteins, a possible explanation for the unexpectedly low seropositive response with east African sera may reside in the immunogenetics of the indigenous pig populations in the region and is unlikely to be due to polymorphisms in immunodominant viral antigens.The area of host-virus interaction, including possible differences in both innate and acquired immune responses therefore, represents an important area for future research on African swine fever (ASF).To characterize the viruses responsible for the out-break of ASF in Uganda in 2007 and identify genetic relationships with recent ASFV isolates from neighboring countries, the sequence of three regions of the ASFV genome was determined.Sequence analysis of the Cterminal end of p72-gene and complete p54-gene placed the Uganda isolates within genotype IX together with isolates from Kenya and Uganda.This is a domestic pigassociated genotype that contains 8 isolates from Western and Eastern Uganda (UGA 1/95; Ug03H1-3, Ug03P4-6, UGA2003/1), isolated from outbreaks in 1995 and 2003 (Bastos et al., 2003;Gallardo et al., 2009) and 11 isolates from Western ands Central Kenya (Ken06B1-5, Ken06.Bus, Ken06.Kis, Ken07Eld1-2, Ken07.Nak, Ken07.Kia) obtained from outbreaks in 2006 to 2007 (Gallardo et al., 2009).One additional genetic type of ASFV was also identified in this study as having occurred in Uganda; this is classified within p72 genotype X that comprises viruses recovered from both domestic and wild pigs and the arthropod tick vector.
The data generated using three genes indicated that, African swine fever outbreaks occurring in the three viruses whilst the 1995 outbreak was caused by two diffe-rent viruses belonging to p72 genotypes IX and X (UGA/1/95 and UGA/3/95) (Boshoff et al., 2007;Lubisi et al., 2005).
Sequence analysis of the central variable region (CVR) from the Ugandan isolates identified 22 different tetrameric amino acid units.The sequence of the CVR from the Ugandan isolates was identical to the sequence of 6 isolates responsible for the second wave of infections that occurred in Western and Central Kenya from October 2006 to January 2007, but was distinguishable from Kenyan isolates recovered from an earlier outbreak in May 2006 and from Ugandan viruses recovered from the outbreaks that occurred in Western and Eastern Uganda in 2003.A deletion of one tetrameric repeat that was observed in viruses recovered from the 2007 Ugandan outbreak was also present in viruses responsible for an outbreak that occurred in the Uganda-Kenya border district, subsequently spreading to central Kenya, in 2006Kenya, in to 2007.This is consistent with a single origin for the Kenyan 2006 to 2007 and Ugandan 2007 outbreaks, if there is rapid cross-border movement of the virus.The data from the CVR analysis are consistent with the hypothesis that, ASFV virus exchange between the two countries has occurred on more than one occasion, although, an alternative explanation would be that identical mutations have occurred independently in the central variable region of two closely related virus lineages.Such information should be proved useful to veterinary authorities for better disease control and highlight the requirement for improved regulation and monitoring of domestic pig movement and the pork product trade.

Figure 2 .
Figure 2. Minimum evolution (ME) tree representing the 22 p72 genotypes that distinguish African swine fever viruses (labelled I-XXII) illustrating the phylogenetic position of the Ugandan isolates from 2007.The optimal tree with the sum of branch lengths = 0.23816201 is shown.The percentage of replicate trees in which the associated taxa clustered together according to bootstrap analysis (1000 replicates) is marked adjacent to the nodes.The ME tree was searched using the Close-Neighbor-Interchange (CNI) algorithm at a search level of 1.A neighbor-joining algorithm was used to generate the initial tree.The seven Ugandan viruses characterized in this study are highlighted in grey ( ) within genotype IX (marked in grey).

Figure 3 .
Figure 3. Minimum Evolution phylogenetic tree based on full length p54 gene sequences illustrating the position of Uganda 2007 isolates relative to other ASFV viruses.An unrooted minimum evolution tree was derived from alignments of complete p54-gene nucleotide sequences.The ME tree was searched using the Close-Neighbor-Interchange (CNI) algorithm at a search level of 1.A Neighbor-joining algorithm was used to generate the initial tree.The seven Ugandan viruses characterized in this study are highlighted in black ( ).
, the isolates from Uganda obtained in October 2007, contained a minor change due to the absence of a single internally located tetrameric repeat (CADT) present in the viruses recovered from earlier Uganda outbreaks that occurred in 2003 and in the Kenyan ASF Visolates obtained in May 2006 (Ken06.B1-5).However, they were identical to

Table 1 .
ASFV Uganda isolates obtained from 2007 outbreak characterized in this study.
(Boshoff et al., 2007)erformed with Chromas (www.technelysium.com.au),BioEdit(www.mbio.ncsu.edu/BioEdit/BioEdit.html) and ClustalX version 1.83 (www.clustal.org).For the TRS, analyses including that of the CVR sequences and deduced amino acid sequences were manually aligned with gaps being inserted to optimize the alignment.Two datasets were generated for phylogenetic analyses conducted using MEGA version 4.0(Kumar et al., 2001) (i) A p72-gene dataset comprising 73 taxa (404 characters) in which p72 nucleotide sequences generated in this study from the Ugandan pig viruses were analyzed together with homologous sequences from at least two viruses representative of each of the 22 p72 genotypes identified in a previous study(Boshoff et al., 2007), (ii) A p54-gene dataset comprising 155 taxa in which p54 sequences generated in this study were compared with homologous sequences available in GenBank.Neighbor joining (NJ) and minimum evolution (ME) p72 and p54 trees were constructed employing the p-distance nucleotide substitution model as implemented in the MEGA v4.0 program.To determine the degree of statistical support for each node in the resulting p72 and p54 trees, data were re-sampled 1000 times using the bootstrap method.
determined.Amplification of the CVR fragment generated PCR products of approximately 350 b from all Ugandan isolates.As in the case of the other genes, the Uganda isolates characterized in this study clustered with the In October 2007, a suspected African swine fever outbreak was reported by farmers in three central Uganda districts.Sampling in these areas in collaboration with the Uganda MAAIF resulted in confirmation of gene based PCR and virus isolation at the CISA-INIA EU reference laboratory.Subsequent nucleotide sequence determi-nation of the PCR product allowed identification of a new NdeI restriction site, specific for a particular group of viruses belonging to east African countries.