The objective of this study was to characterize CPXVs isolated from Fennoscandia and compare their biological and genetic characteristics to that of CPXVs isolated from other geographic regions, as well as to other OPV species. The Fennoscandian CPXVs were characterized on the basis of ATI phenotype, RFLP of atip gene fragment amplicon, sequence and phylogenetic analysis based on full length atip and p4c genes. We have demonstrated that CPXVs isolated from Fennoscandia produced wild type V+ ATI (except for CPXV-No-H2) and encode full length atip and p4c genes. The V+ ATI was produced in both Vero and A549 cells, suggesting that it is a strain specific trait. Functionally intact atip and p4c genes have been shown to be essential but not sufficient[24, 33] for the formation of V+ ATI. In addition to functionally intact atip and p4c genes, it has been demonstrated that VACV Copenhagen A27L homologue is required for the formation of V+ ATI. The production of V+ ATI in cells infected with Fennoscandian CPXVs may be due to the presence of functionally intact atip, p4c and A27L sequences. Although these three genes have been shown to be essential for the formation of V+ ATI, it cannot be excluded that other genes are involved in the formation of this phenotype. We are currently investigating whether the three genes (atip, p4c, A27L homologue) are sufficient for the production of wild type V+ ATI or if additional genes are required. The occlusion of virions within ATI may facilitate host to host transmission by protecting the virions from the harsh environment during transmission between hosts. Thus, Fennoscandian CPXVs with the exception of CPXV-No-H2 may have evolved the V+ATI phenotype to increase virus survival capability outside the host and aid host-to-host transmission under extreme climatic conditions in Fennoscandia, especially the seasonal freeze-thaw cycles.
The RFLP profiles generated by digesting PCR products amplified with ATI-2 primer pairs have enabled correct species assignment of 73 OPV isolates already known to belong to different OPV species. Our application of this method to CPXVs isolated in Fennoscandia yielded RFLP profiles that correlate with the geographic region of the isolates. Although the RFLP profiles 2–4 were unique to Fennoscandian CPXVs, they were more or less related to published profiles of other CPXVs. However the RFLP profile 5 for CPXV-No-H2 was similar to that of ECTV and was completely different from any known CPXV profile. This observation was the first clue of a suspected recombination event. We have reported elsewhere that CPXV-No-H2 is a novel recombinant between CPXV and ECTV. Thus, in addition to robust differentiation and classification of OPV strains, RFLP profiles generated by digesting ATI-2 primer amplicons can serve as a first indicator of atypical or recombinant CPXV.
The phylogenetic tree topology in tandem with genetic and patristic distances has been used for robust molecular taxonomy of OPVs. We employed the same method in classifying Fennoscandian CPXVs and other CPXVs/OPVs, with the modification that we chose to use the TATV-CMLV threshold rather than VARV-TATV that was reported by others. We chose the TATV-CMLV threshold because TATV is closer to CMLV than to VARV, thus the distances between TATV and CMLV represents the lowest distance between distinct OPV species, and CPXVs whose genetic/patristic distances equals or exceeds the TATV-CMLV threshold should not be grouped as the same species. The atip gene phylogeny showed that OPVs were divided into two major monophyletic clades that were further subdivided into six clusters. The CPXV-like clade was exclusively made of five different CPXV clusters (CPXV 1–5) that did not contain any other OPV species, while the VACV-like clade contained VACV-like CPXVs (CPXV 6) in addition to MPXV, VACV, TATV and CMLV. These results are in agreement with a recent phylogenetic analysis based on ten concatenated conserved genes. However, it has to be noted that the bootstrap support for the VACV-like clade is low. One possible explanation for the low VACV-clade support is that vaccine or laboratory strains of VACV were used for the construction of the phylogenetic tree. These vaccine/laboratory strains usually have high number of passages in cell cultures or laboratory animal, and these passages may have introduced mutations into the genome of these strains. Presumably, these mutations might have interfered with the phylogenetic signals. To test this hypothesis; we reconstructed the atip gene phylogeny using sequences of naturally isolated VACV strains. The most improved VACV clade support was obtained when only HSPV-MNR76 and VACV-3737 sequences were used. VACV clade support was improved from less than 50% (Figure 3) to 61% (Additional file6) while the CPXV-clade was improved from 73% (Figure 3) to 82% (Additional file6). These results suggest that mutations introduced in the atip gene of vaccine/laboratory strains of VACV following passages in cell culture or laboratory animals may in part account for the low VACV clade support. Alternatively, low clade support may indicate that members of the clade are not phylogenetically related. However this is unlikely in this case as phylogenetic analysis based on multiple genes as well as the entire conserved central region of OPV genome have yielded tree topology consistent with the atip gene phylogeny reported in this study with the exception that bootstrap support for the VACV-like clade was 95% or more. It has been estimated that CPXV-like and VACV-like clades of OPVs diverged from the common ancestor some ten thousand years ago (TYA), a period that corresponds to the last ice age over Fennoscandia. It has been speculated that the wide abundance of various rodents (natural host to CPXVs and other OPVs) may have fuelled OPV divergent evolution. However, full genome OPV phylogeny grouped CPXVs classified as CPXV-like as sister to CMLV, TATV and VARV. This is in disagreement with the atip gene phylogeny reported in this paper and also with phylogeny based on concatenated conserved genes published elsewhere. The reason(s) for this discrepancy is unclear but it may be that phylogeny based on whole genome provides better resolved trees than those obtained with single genes or concatenated multiple genes. Again, while the atip gene phylogeny reported in this study resolved CPXV_NOR_1994_MAN into the same species cluster as other CPXVs isolated from Norway (CPXV group 1) and distinct from isolates from the United Kingdom (CPXV group 2), the genome based phylogenetic tree grouped CPXV_NOR_1994_MAN into the same cluster as isolates from the United Kingdom. There are two likely reasons for this. First, only one isolate from Norway and no isolate from Sweden were used in the whole genome phylogenetic tree reconstruction. This might have under-represented phylogenetic signals for CPXVs isolated from Norway. Secondly, the higher TATV-VARV threshold was used in the whole genome study as opposed to the lower TATV-CMLV threshold that was used in this study. Indeed if the lower TATV-CMLV threshold was used in the whole genome study, CPXV_NOR_1994_MAN would be distinct from isolates from the United Kingdom. Thus, it can be concluded that CPXV_NOR_1994_MAN clustered with other CPXV isolates from Norway and is distinct from CPXVs isolated in the United Kingdom.
The phylogeny based on the p4c gene showed that CPXVs classified as CPXV-like (CPXV group 1–5) formed a major OPV monophyletic clade that included CMLV, TATV and VARV. Thus, as opposed to the atip gene phylogram, the p4c tree shows that CPXVs classified as CPXV-like (CPXV group 1–5) were sister to TATV, CMLV and VARV. This is in accordance with the results of whole genome phylogeny, but in contrast to the phylogram obtained from concatenating conserved genes located at the central region of the genome. In addition, the p4c genetic and patristic distance measures showed that CPXV group 7 was closer to VACV than to any other OPV. Surprisingly, the p4c gene phylogeny with ECTV as outgroup taxa (Figure 4, Additional file4) does not corroborate the findings obtained from patristic and genetic distance measures. We suspected that the reason for this is that the ECTV sequence is not an uncontroversial outgroup taxa as it is too close to VACV. We tested this hypothesis by reconstructing an unrooted p4c gene phylogenetic tree and our result confirmed that the ECTV p4c sequence was very close to the VACV (data not shown). When ECTV was excluded and an unrooted tree was reconstructed, it clearly showed that CPXV group 7 formed the same monophyletic clade as VACV (Additional file5). Previously, we have shown that the p4c gene of CPXV-No-H2 is diverged from the homologues in all OPV species and probably represent an ancestral sequence. A reconstruction of the p4c gene phylogeny with CPXV-No-H2 sequence as out group taxa demonstrated that CPXV group 7 formed the same clade with VACV (Figure 5). This is in agreement with the whole genome phylogeny. The evidence that CPXV-No-H2 has ECTV atip gene while its p4c gene was shown to be distant from other CPXVs (Figure 5) may raise the suspicion that it is not a CPXV strain. In addition to clinical history we have shown that CPXV-No-H2 is a CPXV strain based (i) presence of two copies of cytokine response modifier B (crmB) gene, (ii) sequence and phylogenetic analysis based on multiple genes including crmB, Chinese hamster ovary host range (CHOhr) gene, and the haemagglutinin (HA) gene[18, 33]. Our previous HA phylogenetic tree construction did not include all the CPXVs used in this study. Therefore we reconstructed the HA phylogenetic tree including all the CPXVs used in this study as well as representatives of both “Old World” and “North America” OPV species. Our result clearly showed that CPXV-No-H2 has phylogenetic affinity with other CPXVs belonging to CPXV-like clade and it is phylogenetically distinct from other “Old World” and “North American” species of OPVs (Additional file7). Unlike the p4c BI phylogenetic tree which has strong posterior probability of 1.0 (100%) in most nodes including the node that grouped CPXV 1–5 with CMLV, TATV and VARV, the p4c ML tree has low bootstrap support in some of nodes. The low clade support in some nodes in the p4c gene ML tree may be due to under-estimation of clade support by the ML algorithm.
The conflicting phylogenetic signals between the atip and p4c genes as shown by both the paired tests and some of the branching and nodes in the tree topologies were rather surprising since both genes are located in the same central part of the genome (the p4c gene is just upstream of the atip) and are required for the same function, that is, the formation of ATI phenotypes. The conflicting phylogenetic signals between atip and p4c genes may indicate that either or both genes are involved in functions other than the formation of ATI, and thus may have different rates of evolution. In spite of this conflict in phylogenetic signals between the atip and p4c, both phylograms have demonstrated that (i) CPXVs are genetically heterogeneous and can be subdivided into six or seven species clusters, (ii) CPXVs isolated from Fennoscandia belong to three of the distinct clusters and that each of the clusters contain isolates from one specific geographic region or country, (iii) CPXV_NOR_1994_MAN belonged to the same cluster as other Norwegian isolates and was distinct from isolates from the United Kingdom, (iv) VACV-like CPXVs (CPXV_GRI_90, CPXV-FIN/T2000, CPXV_FIN2000_MAN, CPXV_AUS1999_867) are closer to VACV than any of the other 14 CPXV isolates (CPXV group 1–5), (v) TATV, CMLV and VARV belong to the same monophyletic cluster, and TATV is closer to CMLV than to VARV, (vi) HSPV is grouped together with VACV isolates, and (vii) MPXVs were resolved into Congo basin and West Africa sub-clusters. Other investigators have demonstrated similar results with respect to OPV phylogeny and evolution[17, 35–37].
The fact that CPXV isolates from Norway and Sweden were classified as CPXV-like while isolates from Finland in conjunction with the isolate from Russia were VACV-like may be an indication that CPXV-like and VACV-like CPXVs have two distinct evolutionary histories in geographically or otherwise separated rodent lineages that re-colonized Fennoscandia after the post glacial retraction 12 – 8 TYA. A recent retrospective study on OPV molecular evolution showed that CPXV-like viruses may have separated from the common ancestor approximately 10 TYA and commenced individual evolution. This event may have triggered the divergent evolution of OPVs that lead to the emergence of VACV-like CPXVs and other OPV species. Intriguingly, the estimated 10 TYA age for OPV divergent evolution corresponds well to the time of post glacial retraction (12 – 8 TYA), thus supporting the post glacial retraction hypothesis for OPV evolution. Moreover, post-glacial re-colonization hypothesis has been used to explain the distribution of different Puumala Hantavirus (PUU) genotypes in Norway and Sweden[38, 39], and it is interesting that both PUU and CPXV have bank voles as reservoir species[38–40]. Detailed studies including a large number of CPXVs isolated from different geographic regions in Fennoscandia as well as their genome sequences will be required in order to determine whether or not (or to what degree) post-glacial re-colonization could explain the genetic diversity among CPXVs.