Intra-epidemic genome variation in highly pathogenic African swine fever virus (ASFV) from the country of Georgia

Resolution of correct homopolymer tract allele states is warranted due to their ability to generate frameshift mutations, alter amino acid repeat number and protein structure, produce gene fusions, and alter putative promoter and termination signals. Substitution mutations and homopolymer tract variations among the Georgian ASFV genomes were elevated in the terminal regions of the genome, particularly in the LVR (Fig. 1). It is unclear whether the homopolymer tract diversity present in the Georgia 2007/1 isolate genome reflects the original wild-type sequence, sequencing error, or mutations arising in cell culture prior to harvesting. Homopolymer tract sequences in the Georgia 2007/1 genome were reported to have been validated using Sanger sequencing [19]. The G-2008/1 and G-2008/2 genomes were sequenced directly from porcine blood samples. In contrast, the Georgia 2007/1 isolate was obtained from infected porcine organ tissue, cultured, and harvested from primary porcine bone marrow cells [19].

Under a scenario where the 2007/1 genome represents the ancestral sequence, the subsequent emergence and maintenance of homopolymeric A/T variations fixed in the G-2008/1 and G-2008/2 genomes would suggest a rapid and stable adaptation to the local host population. Previously published whole genome data gathered following extensive in vitro passage of individual strains revealed only minimal poly A/T tract variability. Whole genome comparisons of the Vero cell culture-adapted BA71V virus and the highly virulent BA71 parent strain revealed only five variable homopolymeric A/T loci and no G-tract polymorphisms [24]. Furthermore, no homopolymeric tract variations were reported in Vero-adapted Georgia 2007/1 strains (ASFV-G passages 30, 60, 80, and 110) although these genomes are not publicly available for independent verification [25]. Compared to such in vitro data, the extent of homopolymeric A/T-tract variability observed between the G-2007/1 genome and G-2008/1 and G-2008/2 appears pronounced, although the context of genome evolution may be expected to differ between published in vitro studies and wild-type genome variation reported here. We found that select homopolymeric A/T-tract loci unique to the 2007/1 genome in our analysis also displayed variation in previous studies, such as the terminal poly-T tract the CP204L gene and the homopolymer tract responsible for fusion of MGF 110-13 L homologs in L60, E75, and NHV28. Homopolymer-based variability has also previously been described in 14 additional ASFV proteins, including p54 and E183L [26, 27]. Homopolymeric tracts that displayed notable tract length variability in other strains, such as that within the M1249 L gene of strains L60 and E7528 showed single-allele state in the three Georgian ASFV genomes here.

Cell culture propagation is frequently performed prior to ASFV genome sequencing. The genetic stability of ASFV epi-markers in traditional ASFV in vitro cell culture systems remains understudied. Limiting errors in phylogenetic inference, such as those arising from homoplasious variation and evolutionary selection are a crucial component of molecular assay design strategy. The extent of in vitro passage or natural transmission events needed to mediate homoplasious variation in current ASFV molecular markers remains unknown and warrants further investigation. We find previous data indicate in vitro propagation may alter genetic loci currently used for ASFV subtyping. Our analysis of genome sequence data generated in a previous study [24] describing adaption of strain BA71 to Vero cells revealed that the adapted strain BA71V uniquely possessed a new CVR (pB602L) allele that is identical to that of the Portuguese strain Por 63 (Supp. 1F). Homoplasious variation was also found to occur at the CVR (B602L) marker in select global strains during replication in pig macrophages and Vero cells [28].

In addition to their use in ASFV subtyping, tandem repeat (TR) markers are also used for dissecting molecular variation in other dsDNA viruses including Hepatitis C virus [29], Adenoviruses [30], human Cytomegaloviruses [31], and Herpesviruses [32‐34]. The approach used by Deback et al. is similar to that commonly used in developing microbial variable-number tandem-repeat (VNTR) assays. This approach includes selection of candidate TR loci that are first informed by comparative genome analyses and subsequently tested for genetic stability in vitro [32]. This approach represents a viable strategy for testing existing ASFV marker stability as well as future markers relevant for tracing current intra-epidemic ASFV isolates. Cataloguing genome-wide variation among a longitudinal set of geographically diverse isolates from the current epidemic would facilitate the identification of new more epidemiologically-informative marker loci. The genomes reported here, in addition to the G-2007/1 genome, provide an early baseline for such efforts.

Variability in the genetic architecture of the genomic region spanning the six hypervariable G/C-tract loci in our study has been reported previously [35, 36]. The repeat tracts are positioned at the core of homopolymeric A + T rich regions within and adjacent to select MGF 110 and MGF 300 genes that together form diverse larger internal repeat sequences in the LVR (38). These regions were previously shown to possess architectural similarity to eukaryotic scaffold-associated regions (SARs) and satellite III repeats of Drosophila [36]. Whether the sequence read heterogeneity at these loci observed in our data represent artifact or a natural mutant spectrum is unclear. Lack of sample volume prevented further molecular validation. Each homopolymeric locus uniquely contained ≥10 Gs or Cs and was confined to the same 7 kb region in the LVR and is illustrated in Additional file 1: Figure S1A. All remaining homopolymeric G/C-tracts in the G-2008/1 and G-2008/2 genomes possessed less than 10 residues and displayed no within-sample copy number variation. Homopolymer A/T tract loci in the genome, by comparison, exhibited no detectable intra-sample variation. In contrast to the G-2007/1 genome, the G-2008/1 and G-2008/2 genomes exhibited only two homopolymeric A/T-tracts that differed and displayed no within-sample heterogeneity in their sequence read populations. The basis for the occurrence of within sample homopolymer tract variation confined only to this region of the genome in both isolates remains unknown.

To date, this study is the first analysis of intra-epidemic ASFV genome sequences and provides insight into genome stability during wild-type infection. DNA viruses generally display low mutation rates largely due their encoded high fidelity DNA proofreading repair enzymes. ASFV genomes uniquely encode the only known X-type polymerase (X Pol) and a DNA ligase that each exhibit low fidelity [37, 38]. The in vitro fidelity of the ASFV DNA ligase is currently the lowest known. The genetic and phenotypic basis for the extreme antigenic diversity displayed by ASFV strains has not been determined [39, 40]. Elevated antigenic diversity in field isolates has been speculated to arise from the activity of the X Pol/DNA ligase. Previous authors have hypothesized that the Pol X/ligase system functions as a strategic DNA mutator in the base excision repair (BER) pathway where rapid genetic drift occurs due to elevated DNA repair-based error [37, 41, 42]. Conflicting data for Pol X fidelity have been reported [37, 41, 43]. It has been speculated that the incongruous reports resulted from enzyme redox state differences between the published studies [44], wherein greater fidelity is exhibited by the reduced form of the enzyme [45]. We speculate realization of the DNA mutator model proposed previously for ASFV would be reflected by greater intra-sample genome-wide variation and elevated genomic polymorphisms between the wild-type isolates studied here. Our data suggest homogeneity of within-host and inter-genome sequences. These data are consistent with other studies describing factors supporting DNA fidelity in ASFV genome replication, including the activity of the viral class II apurinic/apyrimidinic (AP) endonuclease (E296R protein) [46] and the recently reported reparative role for Pol X in vivo [47, 48].