Genetic Assessment of African Swine Fever Isolates Involved in Outbreaks in the Democratic Republic of Congo between 2005 and 2012 Reveals Co-Circulation of p72 Genotypes I, IX and XIV, Including 19 Variants

African swine fever (ASF) is a devastating disease of domestic pigs. It is a socioeconomically important disease, initially described from Kenya, but subsequently reported in most Sub-Saharan countries. ASF spread to Europe, South America and the Caribbean through multiple introductions which were initially eradicated—except for Sardinia—followed by re‑introduction into Europe in 2007. In this study of ASF within the Democratic Republic of the Congo, 62 domestic pig samples, collected between 2005–2012, were examined for viral DNA and sequencing at multiple loci: C-terminus of the B646L gene (p72 protein), central hypervariable region (CVR) of the B602L gene, and the E183L gene (p54 protein). Phylogenetic analyses identified three circulating genotypes: I (64.5% of samples), IX (32.3%), and XIV (3.2%). This is the first evidence of genotypes IX and XIV within this country. Examination of the CVR revealed high levels of intra-genotypic variation, with 19 identified variants.


Introduction
African swine fever (ASF) is a complex and highly lethal haemorrhagic disease of domestic swine with mortality rates reaching 100%. ASF, which is threatening the world pig industry, is a notifiable disease by the World Organization for Animal Health (OIE) [1]. Since the recognition of ASF in Kenya in the 1920s [2], ASF has expanded to most sub-Saharan countries [3], and was exported outside Africa in 1957 to Portugal [4]. Subsequent exportations to Europe occurred in 1960 and 2007 [5,6]. ASFV DNA was extracted from tissue samples using the QIAGEN blood and tissue extraction kit (Qiagen, Hilden, Germany) according to the manufacturers' protocol. Each sample of extracted DNA was then tested by real-time PCR (qPCR), as described by King et al. [25] to confirm the presence of viral DNA for ASFV using the primers King forward (5 -CTGCTCATGGTATC AATCTTATCGA-3 ), King reverse (5 -GATACCACAAGATCRGCCGT-3 ), and the King probe (5 -Fam-CCACGGGAGGAATACCAACCCAGTG-Tam-3 ).

Molecular Characterization of ASFV
Multiple sequence alignments of both the p54 and p72 genes were generated in MEGA (Version 6.0) [27] using default values of the 'by codon' option with the ClustalW algorithm with additional manual editing as needed. The p72 alignment was 404 bp in length and contained 120 sequences. Of these 120 sequences, 62 were generated for this study and 58 were reference sequences with at least one representing each of the known 23 genotypes. The p54 alignment was 657 bp in length and included 84 sequences; 34 were generated for this study, and 50 were reference sequences for 20 of the 23 known p72 genotypes. Published sequences for genotypes XI, XII, and XVIII were unavailable for examination. The most appropriate model of molecular evolution was determined by the corrected Akaike Information Criterion (AICc) using MEGA [27]. Maximum likelihood (ML) analyses with 1000 bootstrap replicates were performed using the program MEGA with the predetermined model of molecular evolution (GTR+I+G for the p72 dataset and HKY+G for the p54 dataset) using all sites. The intra-genotypic genetic distances for the p72 and p54 nucleotides sequences of DRC samples were calculated using MEGA [27].

Central Hypervariable Region (CVR) of B602L Gene
Grouping of amino acids into tetramers at this locus has been utilized by other researchers, therefore the coding of tetramers followed methods outlined previously [20,26,[28][29][30]. The amino acid tetramer codes are provided in Table 2.

Field Identification and Description of Collecting Localities
Fifty-four out of the 57 sampled pigs exhibited many of the following signs or pathological findings: hemorrhagic edema; enlargement of spleen and some internal lymph nodes; hydropericardium and pericarditis; hydrothorax; ascites; as well as skin cyanosis and petechiae. The three remaining pigs sampled in the markets appeared superficially healthy, but presented with enlarged, congestive or hemorrhagic spleens and/or gastrohepatic lymph nodes. Massive mortality was documented in the 25 sampled localities. Four of the 25 sampling sites, including the cities of Yakoma, Boende, Mahagi and Kipushi, recorded indigenous pigs of local breeds, primarily free ranging to be the most commonly lost. The remaining 21 locations were commercial farms raising improved breeds of pigs in backyards or securely fenced areas, minimizing intermingling of wildlife and domestic swine.

Molecular Characterization of ASFV
Phylogenetic analyses of the p72 gene revealed that the newly sequenced ASFV strains which circulated in the DRC from 2005 to 2012 clustered into three p72 genotypes: I, IX and XIV ( Figure 1). Of the newly analyzed stains, 40 (64.5%) grouped with strains previously identified as belonging to genotype I, including three published strains from the DRC (Katanga63 (Genbank: AF301540) [11], Kat67 (Genbank: FJ174377) [13] and Zaire (Genbank: AY351515) [9]); 20 strains (32.3%) were recognized as genotype IX, and 2 (3.2%) belong to genotype XIV. This is the first report of genotypes IX and XIV circulating in the DRC. Although sequences generated herein grouped with high bootstrap support (>75%) with reference samples of their respective genotypes, support for both inter-and intra-genotypic relationships varied.
Phylogenetic analyses of the p54 gene recovered the same groupings of new sequences with respective genotypes as the p72 analyses with high bootstrap support (>83%) ( Figure 2). The arithmetic means of nucleotide substitutions per site between the DRC ASFVs of each of the three genotypes (within group mean distance) were estimated using MEGA version 6. The within group mean distance for the p72 nucleotide sequences were 0.002 for genotype I and 0.0 for genotype IX and XIV. Likewise, for the p54 nucleotide sequences, the within group mean distance was 0.02 for the genotype I members, and 0.0 for genotype IX members. Only one isolate of genotype XIV was successfully amplified and sequenced for this gene.

CVR of B602L Gene
Sequence analysis of the CVR locus showed distinct variability in nucleotide sequence and recognized 19 unique nucleotide sequences, which were translated into amino acid sequences, and subsequently coded as amino acid tetramers (tet-types). Thirteen tet-types were detected within strains identified as belonging to p72 genotype I, six from strains belonging to genotype IX, and two within genotype XIV strains ( Table 2). All strains grouped within a tet-type contained identical CVR nucleotide sequences; therefore, no resolution was lost by converting nucleotide sequences to tetrameric repeat sequences. Tet-12 and 51 were both detected in p72 genotype I strains, but tet-12 was also found in a p72 genotype IX strain (drc86/10/2), and tet-51 was also found in a p72 genotype

CVR of B602L Gene
Sequence analysis of the CVR locus showed distinct variability in nucleotide sequence and recognized 19 unique nucleotide sequences, which were translated into amino acid sequences, and subsequently coded as amino acid tetramers (tet-types). Thirteen tet-types were detected within strains identified as belonging to p72 genotype I, six from strains belonging to genotype IX, and two within genotype XIV strains ( Table 2). All strains grouped within a tet-type contained identical CVR nucleotide sequences; therefore, no resolution was lost by converting nucleotide sequences to tetrameric repeat sequences. Tet-12 and 51 were both detected in p72 genotype I strains, but tet-12 was also found in a p72 genotype IX strain (drc86/10/2), and tet-51 was also found in a p72 genotype XIV  Figure 4). Regarding the five remaining provinces: Equateur presented four tet-types (within genotype IX); Bas-Congo and Maniema each had a single tet-type (within genotype I); Katanga (within genotype XIV) and Oriental each contained a single tet-type (within genotype IX), as well ( Figure 3). Codes as labeled in previous studies: [20,26,[28][29][30].  [17]; 2 [26]; 3 [16]; 4 [13]; 5 [13]; 6 [14].

Molecular Characterization
This is the first extensive molecular evaluation of circulating ASFV genotypes in the DRC. Previous knowledge was based on data from single samples submitted for ASF

Molecular Characterization
This is the first extensive molecular evaluation of circulating ASFV genotypes in the DRC. Previous knowledge was based on data from single samples submitted for ASF diagnosis [9,11,26,29,31,32], but little was known about the dynamics of circulating ASFV strains in the DRC. Analysis of the p72 gene identified three genotypes (I, IX and XIV) circulating in the DRC. Forty of 62 (64.5%) clustered with DRC historical strains (DRC_Kat63 and DRC_Kat67), as well as other strains from genotype I, confirming that genotype I is the most prevalent in the country. Additionally, DRC strains of this genotype exhibited limited genetic variability, which has been repeatedly documented and hypothesized to be predominantly a result of maintenance through the domestic pig cycle [9]. There were 19 p72 genotype IX strains (30.6%) with no genetic variation detected. Previously undocumented within the DRC, genotype IX is much more geographically restricted, as it is endemic to East and Central Africa (Figure 5), having been reported previously in Republic of Congo, Uganda, and Kenya ( Figure 5) where it is involved in sylvatic and domestic cycles [9,17]. Genotype XIV was also not previously reported in the country prior to the two (3.2%) strains reported herein. Genotype XIV was previously only reported from Zambia ( Figure 5), where it was originally isolated from a tick of the genus Ornithodoros in 1986 [9]. In regards to the distribution of ASFV within the DRC recovered in this study, all genotypes were found near the eastern and western borders, and confirmed localities for all three genotypes occurred in the southern half of the country; however, only genotype IX viruses were collected from localities in the northern DRC (Figures 3 and 4). The presence of all three genotypes within the Kinshasa province is likely a result of pig shipments, as pork sold at markets in Kinshasa has been documented to originate from multiple regions of the country, including the provinces of Equateur, Bandundu, and Bas-Congo [33]. Although the p54 gene was previously determined to be a valuable locus for finer levels of discrimination [19,34,35], no additional resolution was gained for this dataset; however, the p54 locus did corroborate the topology generated by the p72 analysis. Given the sampling scheme of this study (domestic swine only), failure to achieve higher resolution may be due to the low genetic variation consistently detected within the domestic pig cycle, when compared to the higher levels of variation Although the p54 gene was previously determined to be a valuable locus for finer levels of discrimination [19,34,35], no additional resolution was gained for this dataset; however, the p54 locus did corroborate the topology generated by the p72 analysis. Given the sampling scheme of this study (domestic swine only), failure to achieve higher resolution may be due to the low genetic variation consistently detected within the domestic pig cycle, when compared to the higher levels of variation previously detected within the sylvatic cycle [9].
The CVR was capable of providing further resolution than either the p72 or p54 genes. The pattern of genetic variability varied dependent upon the p72 genotype examined. Within p72 genotype I strains, genetic variation was primarily a result of variability in the number of tetramers as previously reported [36], specifically the tetrameric amino acid repeats CAST and CVST. Also of interest within genotype I, is that tet-25, 45, and 51 (also found in one genotype XIV strain) contained a conserved sequence of 11 tetramers (tetrameric repeat sequence number (TRS) ABNABNBT[D/A]BN) not found in other tetrameric sequences within this genotype. This sequence was flanked by a variable number of 'A' coded tetramers. The historical isolate Kat67 genotype I CVR was tet-23, but had a similar sequence BNAxx. Additionally, BTDBN also flanked on each end by variable repeats of 'A' [26]. The new genotype XXIII from Ethiopia had one CVR sequence, ETH/3, which also has a similar motif with xxBNABTDBxx [14]. Nigerian sequences are of genotype I and also share part of this motif in the CVR region xxABNABNxx [16] (Table 2). CVR tet-types of p72 genotype IX consisted primarily of a more complex sequence than genotype I strains, with the exception of tet-12 (common to genotype I strains) being documented in one genotype IX strain. Interestingly, CVR tet-23b which is present in two recent DRC genotype IX strains from 2007 is identical to two genotype IX CVR sequences from Uganda in 2003 and Kenya in 2006 [13] (Table 2). Of the two tet-types detected in genotype XIV strains, the first (tet-51) was identical to a tetrameric repeat sequence found in a genotype I strain, and the second was extremely different than any other tet-types reported herein (Table 2). Several CVR tet-types are discussed below in reference to the detection of multiple strains within outbreaks, and potential links between outbreaks based on this locus.

Disease Ecology, Molecular Epidemiology and Case Investigations
The higher resolution offered by the CVR allowed for the three genotypes to be further broken down into 19 variants. The high level of variability previously noted in the CVR [9,37,38] would suggest that highly similar/identical CVRs are more closely related than more divergent CVRs; however, given the nature of nucleotide repeat regions, such as the CVR, identical sequences can occur due to homoplasy. With these possibilities in mind, a number of putative outbreak connections are proposed and discussed below, as are details of co-circulation of multiple variants.
The detection of genotype I, tet-23 strains during the 2009 outbreaks at Nsele and Kintambo, as well as the 2011 outbreak in Ngafula may suggest that these geographically proximal outbreaks were caused by closely related strains, even though they occurred over a two-year time span which could be facilitated by the asymptomatic infection of domestic pigs [39] in the area. A second potential connection is suggested by analysis of sequences from the Mahagi outbreak in April 2007 (near the Ugandan border), as they were of identical p72 genotype IX and tet-23b to strains from Uganda (Ug03H.1, Genbank: GQ916933), and Kenya (Ken06.B1, Genbank: GQ916935 & Ken06.Bus, Genbank: GQ916940) examined previously [13]. A third connection stems from a strain from a clinically healthy pig sampled at the Kinshasa market (drcKG28040802) in 2008. This strain belonged to the same genotype I and tet-16 as samples from the 2012 outbreak in Mayanda, Bas-Congo province. The geographic proximity of these two provinces, in combination with the high level of commercial traffic between them, could easily result in the spread of ASF from one province to the other [1,20].
Again, either the ability of ASFV to persist in asymptomatic pigs (domestic or feral) or the sylvatic cycle (including ticks) could explain how closely related strains were responsible for outbreaks separated by four years, as Mayanda is a rural locality, where there is a high potential for interaction between domestic and feral swine. Another potential outbreak link was made apparent upon examination of genotype IX strains collected from outbreaks in Yakoma (April 2007) and Boende (March and April 2008). Tet-24c strains were detected in both outbreaks, which severely affected both feral and domestic pigs [40]. The Boende outbreak was estimated to have caused more than 4500 swine fatalities [40]. Both Yakoma and Boende are forested sites, located in North Ubangi and Tshuapa Districts of the Equateur Province, respectively. The forest environment and free ranging animal husbandry practices enable interaction between domestic and wild swine, potentially allowing for crossover between infection cycles and the occurrence of multiple strains.
Another instance of variation occurred in the Ngaliema 2010 outbreak, where tet-51 was found in two strains: drc35/10/1 (genotype I) and drc/35/10/3 (genotype XIV). Co-infection of different tet-types was even documented within the same pig, when two genotype I strains, drc49/05/p1a (tet-5) and drc49/05/p1b (tet-9), were extracted from the spleen and lymph node respectively, of a pig in 2005 from Kinshasa. In similar studies conducted in Mozambique and Nigeria, individual co-infection was not observed [38,41].
Understanding the details of ASFV circulation across the DRC is clearly complex. Co-circulation of genotypes and tet-types and co-infection of multiple tet-types within a single pig suggests a high level of genetic variation and potential support of a previous hypothesis of recombination [42,43]. However, detection of conserved tet-types across large geographic areas, and over multiple years suggests that current markers can provide insight into the movement of ASFV strains. Unfortunately, connecting cases is not straight forward, as anthropogenic factors, such as trade of pigs and animal husbandry practices can play a role. Whole genome examination of these strains may provide a more definitive understanding of relationships.
As the DRC is the largest country in SSA, and borders nine countries, understanding the prevalence and distribution of ASFV genotypes within the DRC is an important step in better understanding large scale patterns of ASFV. Given the high degree of similarity, at all examined loci, between genotype IX strains collected in Mahagi, the DRC (presented herein), and Uganda strains of a previous study [13], it appears that the distribution of this strain spans the border between the DRC and Uganda, suggesting that this strain has been transmitted across boundaries by movement of either feral or domesticated swine. Two more potential cross-boundary transmissions of ASF into/out of the DRC are worth mentioning; however, their molecular evidence is less direct. The first possibility is that genotype XIV may have been transferred, between Zambia (Zam_NYA/12-(Genbank: AY351555), isolated in 1986) [9] and the DRC (drc21/07/22, strain from 2007) through Kipushi, as p72 sequences between these strains share 99.26% nucleotide sequence identity. Unfortunately, no CVR sequence for strain Zam_NYA/12 was available for a more detailed comparison. The next putative trans-boundary migration of ASF could have occurred between Brazzaville, Republic of Congo (Con09/Ni16-(Genbank: HQ645947), isolated 2009) [17] and the DRC (drc86/10/1, strain from in 2010). The p72 sequences were identical, and the CVR locus differed by the insertion of two amino acid tetramers coded as A (AAAAAAAAAF from the Brazzaville strain and AAAAAAAAAAAF in the DRC strain). Rapid mutation rates have been shown in vitro [44], therefore, given the highly variable nature of the CVR, it is difficult to omit the possible link between these outbreaks, especially as the boundary is a narrow aquatic border with high levels of human and animal traffic.
This first, in-depth examination of ASFV in the DRC, has provided evidence of (1) circulation of multiple genotypes previously not reported within the DRC; (2) putative links between both geographically and temporally separated outbreaks; (3) potential movement of ASFV strains across borders between the DRC and Uganda, Zambia, Congo; (4) co-circulation of multiple ASFV genotypes within outbreaks and (5) a pig co-infected with two tet-types. These data, in combination with examination of genotype relationships, will be useful for the optimization of current prevention and control strategies at the regional level given the location and size of the DRC in relation to the rest of the continent and those countries also dealing with ASF.