Genomic and phylogenetic characterization of ChPV2, a novel goat PV closely related to the Xi-PV1 species infecting bovines

Origin of animal samples [12] and DNA extraction [13] were described before. The host animal was a Damascus goat (Shami goat) in a herd comprising about 60 animals in Hatay province in southern Turkey. Clinically, only one animal in the herd had a papilloma. According to the animals’ owners, the goats had been allowed to graze in a mountainous area, where goats and cattle used the same grazing land. The clinical sample was obtained from a teat papillomatosis case (Additional file 1: Figure S8) and sent to the laboratory at the department of Virology at Ankara University by veterinary Ali Haciömeroglu and DNA was extracted upon arrival of the specimen. No experiments with living animals were performed, nor were animals harmed or killed during any procedure related to this article.

The clinical sample was minced in phosphate-buffered saline (PBS) supplemented with streptomycin and penicillin using a scalpel and stored at − 80 °C until tested. DNA extraction from the papilloma sample was carried out according to Sambrook et al. [13].

Based on the previously published, partial L1-nucleotide sequence (MG523274, HTY-goat-TR2016) [12], we designed primers (5′GACTGCCCTCCTTTACAGCTT3′ and 5′GCTTTCCTGAACTTGGTAGCC3′) directed towards the edges of this fragment. The remaining part of the PV genome was amplified using Phusion DNA Polymerase kit using with standard buffer (New England Biolabs) and 100 ng of the original sample DNA as template. The resulting PCR-product was purified using the Double Pure kit (PEQLAB) according to manufacturers’ instructions and cloned into the Zero Blunt TOPO PCR Cloning plasmid (Invitrogen) according to manufacturer’s instructions. The nucleotide sequence of the cloned genome fragment was determined by primer walking from both directions using conventional Sanger sequencing (Eurofins). A ~ 2.8 kb fragment spanning the 3′ end of E2 and the 5′ end of L2 ORF was amplified using newly designed new primers (5′GCAAATATGCTTCCCTCCATTAG3′ and 5′CTGCATAATTACACTGTCTGCAG 3′). As amplification of this fragment failed when using Phusion DNA Polymerase we amplified the missing part of the PV genome using One-Taq Polymerase 2 × Mastermix (New England Biolabs) with standard buffer and 100 ng of DNA from the original sample as a template according to manufacturer’s instructions. The resulting ~ 2.8 kb PCR fragment was gel purified using the Double Pure kit (PEQLAB) and cloned into the pGEM-T-easy plasmid (Promega) according to manufacturers’ instructions. To rule out biases resulting from the use of non-proofreading polymerase three individual clones were sequenced by primer walking (Eurofins). Resulting sequences were assembled together with the previously available genome parts and conflicts were corrected after manual inspection of sequencing results. We amplified the genomic region covering the original sequence communicated by Dogan and coworkers using the specific primers (5′TAGCTTGCTTCGCAAATTC 3′, and 5′ ATTTCGTGGCTTGCAAAGC 3′) using One-Taq Polymerase 2 × Mastermix (New England Biolabs) with standard buffer and 100 ng of DNA from the original sample as a template according to manufacturer’s instructions. The resulting PCR product was gel purified using the Double Pure kit (PEQLAB) and cloned into the pGEM-T-easy plasmid (Promega) according to manufacturers’ instructions. Three individual clones were sequenced (Eurofins) and assembled with the previously available genome parts. Conflicts with the previously published sequence from Dogan and coworkers were corrected after comparative alignment and manual inspection. Finally we performed rolling circle amplification (RCA) using the TempliPhi amplification kit (GE healthcare). Resulting RCA products were subjected to MfeI restriction digest to the linearize PV genomes into single genome copies. The resulting restriction fragments of ~ 7 kb was gel-purified and cloned into pShV plasmid. A positive clone was sequenced by primer walking and the previously assembled sequence was confirmed.

ORF analysis was performed using the ORF Finder tool implemented in SnapGene (https://​www.​snapgene.​com/​) and reviewed manually by comparative alignment to closely related PV genotypes. Potential splicing patterns for the E1^E4 ORF were predicted using the Softberry Fsplice program (https://​www.​softberry.​com/​berry.​phtml?​topic=​fsplice&​group=​programs&​subgroup=​gfind) with the Capra hircus genome as a reference. Sequence similarities were calculated based on the single gene alignments. The molecular weight of the putative proteins was calculated using the ExPASy (Expert Protein Analysis System) Compute pI/Mw tool (https://​au.​expasy.​org/​tools/​pi_​tool.​html) [14]. Protein motifs and domains were identified manually or predicted using prosite tool (https://​prosite.​expasy.​org/​) [15, 16]. TATA Box, pA signals as well as E2 and E1 binding sites were identified manually. Potential E2 binding sites were reconfirmed by screening the TRANSFAC Database [17] using the Match tool (https://​gene-regulation.​com/​pub/​programs.​html#match) [18] as described previously [19].

We collected 376 full-length PV genomes from the PaVE database (pave.niaid.nih.gov, accessed 28 May 2019) and the ChPV2 genome was added to this data set (Additional file 1: Table S6). The E1, E2, L2 and L1 genes were extracted from the collected genomes. Genes were aligned individually at the amino acid level using MAFFT v.7.310 [20], corrected manually, and backtranslated to nucleotides using PAL2NAL v.14 [21]. The alignments were filtered using Gblocks v.0.91b [22].

The full 377 PV data set contained recombinant PVs infecting Cetaceans [23‐25], known to have undergone a recombination event between the early and the late gene regions. Therefore initial tree construction was performed on the concatenated E1E2 and L2L1 alignments separately. Maximum Likelihood (ML) phylogenetic inference was done using RAxML v8.2.11 [26], under the GTR + Γ4 model for the nucleotide alignments using six partitions (three for each gene corresponding to each codon position), or under the LG + Γ model for the amino acid alignment using two partitions (one for each gene), and using 1000 bootstrap replicates. The trees were rooted using the SaPV1 sequence.

Subsequently, the recombinant and unresolved taxa were removed from the full data set, leaving us with a reduced data set of 324 PVs. The individual genes were again aligned and filtered as described above. The concatenated E1E2, L2L1, and E1E2L2L1 alignments were used to construct ML trees as described above. For the E1E2L2L1 alignment, twelve partitions were used at the nucleotide level, and four partitions were used at the amino acid level. The trees were rooted using the SaPV1 sequence.

Based on the constructed trees, the close relatives of ChPV2 were extracted from the full data set, leaving us with a reduced data set of 17 PVs. Besides E1, E2, L2, and L1, the E6 and E7 genes were extracted from these selected genomes. The individual genes were aligned without Gblocks filtering. The E6, E7, E1, E2, L2, L1, and the concatenated E1E2L2L1 alignments were used to construct ML trees for this reduced data set as described above. For the individual gene alignments, three partitions were used at the nucleotide level, and no partitions were used at the amino acid level. For the E1E2L2L1 alignment, twelve partitions were used at the nucleotide level, and four partitions were used at the amino acid level. Possible rogue taxa were identified with an algorithm implemented in RAxML. For the individual gene trees, majority rule consensus trees were constructed, and subsequently used for constructing a supernetwork using Splits Tree 4 [27].

To isolate the complete genomic DNA of a PV identified in DNA samples from teat papillomas of a Damascus goat in Turkey [12], we performed long range PCR using a primer set directed towards the boarders of the previously published, partial L1-nucleotide sequence (MG523274, HTY-goat-TR2016). The resulting PCR-product was cloned and sequenced by primer walking. Based on the resulting sequence assembly we finally amplified the genome region spanning the partial L1 sequences published by Dogan and coworkers 2018 from the original DNA sample. The resulting PCR fragment was cloned, sequenced and assembled with the previously available genome parts. Conflicts with the previously published sequence were corrected after manual inspection of a comparative alignment. MfeI restriction digest of rolling circle amplified (RCA) concatenated linearized PV genome copies generated full genome fragments that were cloned respectively. Sequencing confirmed the previously assembled sequence. Plasmids are available from Eric Ehrke-Schulz upon request. The nucleotide sequence of the ChPV2 genome is accessible in GenBank under accession number MN148899.

The genome of ChPV2 spans 7295 bp with an average GC content of 43% and has the typical organization of a PV genome, containing an upstream regulatory region (URR), an early gene region and a late gene region (Additional file 1: Figure S1, Table S1). The early region contains five putative partially overlapping ORFs, E6 (417 bp), E7 (297 bp), E1 (1809 bp), E2 (1293 bp), and E4 nested within E2 (336 bp). Splice site prediction suggested three different potential E1^E4 splice patterns, with nucleotide position 712 as donor and positions 3052, 3088 or 3226 as putative acceptors (Additional file 1: Table S1). Nucleotide and amino acid sequence similarities of ChPV2 genes to their counterparts of closely related PV types were determined based on single gene alignments of ChPV2 and 16 closely related PV types (Additional file 1: Table S2 and Table S3).

Within the translated ORFs of ChPV2 several protein domains/motifs were predicted (Additional file 1: Table S4). ChPV2-E6 contains two zinc-binding motifs (C-X₂-C-X₂₈-C-X₂-C), but it does not seem to contain a standard PDZ-binding motif. ChPV2-E7 contains a casein kinase II phosphorylation site followed by a retinoblastoma protein (pRB) binding motif (LXCXE) and a zinc binding motif (C-X₂-C-X₂₈-C-X₂-C). In ChPV2-E1 a SF3 helicase 1 domain was identified. ChPV2-E2 contains a DNA binding Helix (GCANTLKCFRRRTSHSHPHK). The late proteins ChPV2-L2 and ChPV2-L1 contain lysine and arginine rich nuclear localization signals (RKFKRKTK) (Additional file 1: Table S4).

A number of non-coding regulatory elements were found throughout the ChPV2 genome (Additional file 1: Table S5, Additional file 1: Figure S2). Within the URR a potential TATA box (TATAAA) is located from bp 7273 to 7277, 18 bp upstream of the E6 start codon. Upstream of the TATA Box, three potential E2 binding sites (E2BS), (ACC-N₆-GGT) are located at nucleotide positions 7164–7175, 7242–7253, and 7258–7269. A potential E1 binding site (E1BS), (GTAGTTGTTGTTAACAACAAT) is located between the first and second E2BS. Two imperfect E2BS* (ACT-N₆-GGT, bp 455 to bp 466) and (ACC-N₆-GTG, bp 481 to bp 492) are located within the E7 ORF, 64 bp upstream of a potential TATA box (TATAA) at nucleotide positions 555 to 559, and could constitute the late promoter. Another potential E2BS (ACC-N₆-GGT, positions 6789–6796) is located close to the 3′ end of the L1 ORF. An early polyadenylation signal (pA, AATAAA) is located downstream of the E2 ORF (bp 3749 to bp 3754) and a late pA is located downstream of the L1 ORF (bp 6920 to bp 6925).

In order to assess the phylogenetic relationships of ChPV2, we collected the available 376 full-length PV genomes from the PaVE database (pave.niaid.nih.gov, accessed 28 May 2019) (Additional file 1: Table S6). First, we constructed ML phylogenetic trees of the concatenated early (E1E2) and late (L2L1) gene sequences at the nucleotide (Additional file 1: Figure S2) and amino acid (Additional file 1: Figure S3) levels. Based on these four trees, we observe that ChPV2 clusters within the Beta-XiPV crown group, and is closely related to XiPVs infecting cetartiodactyles. The position of ChPV2 is well-supported (bootstrap support of 97 to 100) and is basal to XiPVs, 1 species (infecting bovines).

After removing the recombinant and unresolved taxa, ML phylogenetic trees were constructed of the concatenated E1E2 and L2L1 alignments (Additional file 1: Figure S4 and Additional file 1: Figure S5), as well as of the concatenated E1E2L2L1 alignments at the nucleotide (Fig. 1) and amino acid (Additional file 1: Figure S6) levels. The position of ChPV2 did not change and remained well supported. Interestingly, ChPV2 does not cluster with the previously described ChPV1, retrieved from the healthy skin of a goat, and classified as PhiPV1 (11). ChPV1 is the sister taxon to the Timor deer RtimPV1, and both are basal to XiPVs. The nucleotide sequence of the ChPV2-L1 gene shows higher similarities to L1 genes of other XiPVs and to RtimPV1 ranging from 58.3 to 71.8%, than to ChPV1 L1, where the nucleotide similarity is 55.3% (Table 1).

Subsequently we constructed ML phylogenetic trees only with the close relatives of ChPV1, including also the E6 and E7 oncogene alignments, as well as the concatenated E1E2L2L1 alignments on the nucleotide (Fig. 2) and amino acid levels (Additional file 1: Figure S7). Not all close relatives of ChPV2 contain the E6 oncogene in their genomes (Fig. 2 and Additional file 1: Figure S7). These PVs lacking the E6 protein encode instead for a small hydrophobic protein (between 40 and 75 amino acids) annotated upstream, named E10 [8]. Interestingly, all PV genomes containing only E6 cluster together, except for BPV12 that clusters close to PVs containing both E6 and E7 in their genomes (Fig. 2 and Additional file 1: Figure S7).

To complement the current view, the individual gene trees (E6, E7, E1, E2, L1, and L1) were combined to construct a phylogenetic supernetwork (Fig. 3). Our results show few conflicting topologies between the trees, and confirmed that ChPV2 is closely related to the Xipapillomavirus 1 species, and in particular to BPV10. Moreover, the positioning of BPV12 does not seem to change much between trees and remains within the clade of PVs containing both E6 and E7.