Species of the genus Pestivirus of the family Flaviviridae [1] are responsible for cosmopolitan disease affecting cattle pig and other ruminants, presenting a wide range of clinical manifestations, with relevant impact on zootechnic production. Different reports described atypical variants within the bovine viral diarrhoea virus type 1 (BVDV-1) (Pestivirus A), characterized by genetic traits different from most common previously reported strains, in particular from Asian countries [2] - [10]. Similarly, Asian atypical sequences have been reported in other Pestivirus species as BVDV-3 (Pestivirus H) and Classical swine fever virus (CSFV) (Pestivirus C) [11] [12] [13]. Taking into account that a certain confusion in the nomenclature of the increasing number of Pestivirus genetic variants was evident and it is still a matter of concern, indicating that the need for harmonization [14] [15] [16], and the utility to have an alternative evaluation method are always interesting in any discipline, the palindromic nucleotide substitutions (PNS) method [17] [18] was proposed for an analytical procedure not limited to primary structure analysis but extended to secondary structure, focusing on strategic genomic sequences corresponding to the internal ribosome entry site (IRES), in the 5’ untranslated region (UTR), responsible for translational, transcriptional and replication events in pestiviruses [19]. In order to evaluate more exhaustively the taxonomical segregation of the genus Pestivirus in Asian countries, the 5’-UTR genomic sequences of the BVDV-1, BVDV-2 (Pestivirus B), BVDV-3, Border Disease virus (BDV) (Pestivirus D) and CSFV species were analyzed and compared applying the PNS genotyping method, encompassing all Pestivirus species circulating in domestic animals in Asia and excluding the pestiviruses from chiropters.

In order to determine genotypic variations in the BVDV-1, BVDV-2, BVDV-3, BDV and CSFV species, the 5’-UTR genomic region of six hundred-fifty-one Pestivirus strains, have been analyzed for numerical taxonomy. The virus nucleotide sequences, with different geographical origin, primarily of Asian origin, from different host species or contaminants of biological products, were obtained from the DDBL/EMBL/GenBank DNA database or provided by authors (Table 1). Strains from Europe and Americas have been considered as reference of genetic variants and complete comparison of strain sequences of BVDV-2, BVDV-3 and CSFV species. The majority of the tested virus sequences were originated from strains isolated from cattle (Bos taurus) from China (n 131). Other bovine strains originated from various other countries from Middle East, Indian subcontinent, Far East and Austral Asia (Turkey, Iran, Bangladesh, India, Thailand, Japan, South Korea, Australia and New Zealand) (n 69), and 89 were from Europe and Americas. Sixteen strains were isolated from buffalo (Bubalus bubalis) in China. Strains isolated from domestic Bactrian camels (Camelus bactrianus) and yaks (Bos grunniens) have also been reported from China (n 60). Seventeen strains were reported from Zebu (Bos indicus) in India, 116 strains were isolated from domestic and wild suids, pigs (Sus scrofa domesticus) and wild boar (Sus scrofa), 63 were from small ruminants, sheep (Ovis aries), goat (Capra hircus) and black goat (Capra aegagrus hircus). Three strains were from Alpine chamois (Rupicapra rupicapra), Pyrenean chamois (Rupicapra pyrenaica) and Reindeer (Rangifer tarandus), and 3 from Japanese serow (Capricornis crispus). A single

Alpine chamois (Rupicapra rupicapra); Bactrian camel (Camelus bactrianus); Black Goat (Capra aegagrus hircus); Buffalo (Bubalus bubalis); Cattle (Bos taurus); Goat (Capra hircus); Human (Homo sapiens); Pig (Sus scrofa domesticus); Pyrenean chamois (Rupicapra pyrenaica); Reindeer (Rangifer tarandus); Serow (Capricornis crispus); Sheep (Ovis aries); Wild boar (Sus scrofa); Yak (Bos grunniens), Zebu (Bos indicus).

sequence was identified from a human pathological sample in Brazil. Seven sequences were CSFV vaccinal strains. Eighty-three strains were contaminants of biological products, mainly detected in fetal bovine serum (FBS) samples for laboratory use, but also cell lines, interferon for human use, vaccines for veterinary and human use. Most of the samples were from Japan (n 37). Other contaminats were reported from China, India, Australia, New Zealand, United Kingdom, Italy, Belgium, Netherland, Tunisia, USA, Canada, Mexico, Brazil and Argentina. Sequences were compared to previously evaluated strain, reaching a total number of 1426 observations obtained with PNS procedure (Table 2). Qualitative and quantitative evaluation of genomic sequence divergence, in terms of palindromic nucleotide base pairings variations, has been applied for taxonomical segregation, through the evaluation of relevant secondary structure regions in the 5’-UTR of the viral RNA, the three variable regions, V1, V2 and V3 genomic sequences (Table 3), according to the genotyping based on the PNS method [17] [18]. As described by Deng and Brock [19] and Harasawa [20], the 5’-UTR is characterized by a series of secondary sequence stem-loop structures which are divided into 4 domains: A, B, C and D. The most significant is Domain D which corresponds to the IRES and it covers two thirds of the 5’-UTR sequence from nucleotides 139 to 361 and it is conserved among NADL, Osloss, SD-1, Alfort and Brescia Pestivirus reference strains [19]. The predicted Domain D complex palindromic stem-loop structures D2, D3 and D4 [19], or as recently renamed as IRES domains IIIb, IIId1 and IIId2 [21], correspond to PNS V1, V2 and V3 variable loci, respectively. Within the genus, the palindromic structures identifiable in linear sequences were generally characterized by the absence of nucleotides in position 22 of V1 and V2 locus composed by a constant number of 23 nucleotides.

The variation of loops in V1 and V3 loci determined difference in size of palindromes. Secondary structures were obtained for the entire 5’-UTR sequence of each strain. Palindromic sequences corresponding to the IRES three variable loci were identified in the predicted secondary structure and considered out of the rest of the nucleotide sequence. Nucleotide sequence secondary structures were predicted according to the algorithm of Zuker and Stiegler [22] using the Genetyx-Mac version 14 program package (Software Development Co., Ltd., Tokyo, Japan). The minimum free energy was calculated by the method of Freier et al. [23]. The PNS software version 2.0 [24], prepared for the application on the genotyping procedures with the keys for Pestivirus identification of genomic sequences, using the C# programming language, was also applied for the construction of secondary structure sequence alignment, in order to compute genetic distance among strains. Segregation of Pestivirus species strains into genotypes and relatedness among genotypes within the species was evaluated according to changes in nucleotide base pairs at the level of the secondary palindromic structure of the three variable loci. Genotypes have been defined applying a nomenclature reflecting the level of divergence within the species, and ranked according to increasing divergence expressed in number of divergent base pair (bp), with reference to most common observed sequence combinations in the genus. Genotype roots, base-pair (bp) combinations at the level of low-variable positions (LVP), also defined determinative LVP [14] [17], were identified to support genetic clustering of strain sequences and characterize species genotypes, varying for each species (LVP V1/14, V2/7 and V3/4 for BVDV-1; V1/9, V1/12, V1/16, V1/17 and V1/18 for BVDV-2 or V1/9, V2/1, V2/5 and V3/8 for BDV). Among genotypes, homology was evaluated in terms of shared base pairs in the three variable loci. Cross comparison between types within the genus has been evaluated by computing the divergence percentage, identifying strains showing multiple relation (sequences sharing base pairings specific to different genomic groups, and scoring low divergence values) or borderlines (sequences showing qualitative similarities with a genomic group, but with high divergence values, candidates for reallocation as separate groups in the genus), and indicating divergence within groups and among groups quantifying the heterogeneity of a genotype and the genetic distance between groups. Phylogenetic trees based on the 5’-UTR were constructed following sequence alignment of representative species strains with Clustal X [25] by using the neighbour-joining method [26] and visualized with Newick tree format option (Mega version 7.0.26) [27].

The observation made on the nucleotide sequences of the three variable loci at the level of the 5’-UTR genomic region of Pestivirus strains, according to the PNS genotyping method, allowed the identification of consensus motifs shared by all the Pestivirus species, genus specific base-pairings, and characteristic species and genotype specific PNS, respectively (Table 4). Ten PNS positioned in the V1 and V2 loci were characteristic for the genus. PNS consensus motifs shared by all species in the genus, as the bulge formed by two cytosine nucleotides in the V1 stem position eleven, are presented in Table 4 and schematically shown in Figure 1. Based on the divergence limit value of 9 bp for genotype determination [17], twenty-three genotypes within the BVDV-1 species have been identified, from 1a to 1.23. BVDV-2 accounted for six genotypes, from 2a to 2f. BDV was clustered into ten genotypes, from a to j. BVDV-3 and CSFV were less heterogeneous with four genotypes.

Secondary sequence construction, efficiently obtained by both available software, Genetyx and PNS [24], revealed a conserved palindromic structure in the genus species (Figure 1). Only in few cases the elaboration of secondary structure resulted slightly problematic. Some difficulties have been encountered due to atypical sequences, showing insertions in the variable loci, as the Chinese BVDV-3 ovine strains HN1559 [KU053497], HN1507 [KU563155], HN1568 [KU053499] and HN1564 [KU053498] (Shi et al., unpublished) with U U, A A, G G or C C bulge insertion in the V1 locus. Similarly, BVDV-3 strains IndMDV18963/12 [KM201318] and IndBHA6604/12 [KM201307] showed deletions in V3 locus positions 5 and 6. The predicted secondary structures of the three variable loci were aligned for comparison of base pairings in the different positions (Table 5). Different base pairing combinations were identified for genotype characterization (BVDV-1 n 416; BVDV-2 n 66; BVDV-3 n 33, BDV n 99 and CSFV n 37), and considered for identification marker definition. Strains showing sequence identity at the level of the three variable loci or sharing non-relevant variations as G*U or G-C (G:Y) were excluded (Table 6). The summary of strains clustering into genotypes of Pestivirus species is presented in Table 7. At the species level, the observed taxonomic status of the examined strains corresponded to the estimation obtained by phylogenetic trees constructed from the alignment with the representative strains from the identified genogroups (Figures 2-4). Different species genotypes, showing peculiar genetic

characteristics, were restricted to specific geographic areas as in Turkey, China and other Asian countries (Figure 5).

The BVDV-1 species was heterogeneous. Some Asian strains showed exception at the level of BVDV-1 species markers. For example, Chinese strains of the genotype BVDV-1.10 showed divergence in V1/15 with A A or C A bulges, instead of species marker U-A pairing. In sub genotype 1.15.2, all strains showed species marker exception in V2/5, with A-U instead of G-C. Similarly, strain S153 [8], genotype 1.23, showed an atypical G A bulge at the level of species marker position in V3/5. The Asian clusters 1.8 and 1.15 shared a root characteristic of genotype BVDV-1a (V1/14 C-G, V2/7 G-C and V3/4 A-U). All the strains belonging to the group 1.8 showed an A C bulge in position 12 in V1 locus, a base pairing not present in any member in the genotype 1a. In this group have been included bovine strains from Australia, China and Japan as Bega (Mackintosh et al., unpublished), Shitara/01/05 [28], Manasi [9], the Australian contaminant strain A1-114/AU [29] and the contaminant strain CRFK reported from Japan [30]. Among Chinese strains of the genotype 1.15, reported as genotype M [2] [7], root A was present in only the strain LZ05 of the sub genotype 15.1. In the sub genotype 15.2, characterized also by an exception at the level of species marker in V2/5 (A-U instead of G-C), root A was present in the majority of the strains. The genotype 1.21 showed a new and atypical U-A/G-C/U-A root associated to the genotype b (sub genotype 2). BVDV-1b, characterized by a variation of base pairing at the level of position 7 in V2 (b.1 V2/7 A-U; b.2 V2/7 G*U or G-C), was the only other BVDV-1 genotype showing U-A pairing in LVP V1/14. The V1/14 U-A was shared exclusively with BVDV-1b and BVDV-1.21

in which was characteristic, and in all the BVDV-2 species strains.

Among BVDV-1 genotypes, only BVDV-1a, 1b, 1.3 and 1.13 were widely distributed. The geographic distribution of genotypes varied considerably, and in certain countries as China, over the half of the 23 genotypes described in the species, 12 could be identified: BVDV-1a, 1b1 and 1b2, 1.3, 1.6, 1.7, 1.8, 1.10, 1.11, 1.15, 1.18, 1.21 and 1.23. BVDV-1a (n = 4) was reported in two bovines [6], two Bactrian camels [31] and one pig [32]. Despite belonging to the same BVDV-1a genotype, strains NH01 (bovine) and SH1060 (pig) were low related, showing 89.2% homology and a divergence of 11 bp in the IRES. Among BVDV-1b genotype, only two BVDV-b1 have been reported in cattle (Zhang et al., unpublished) [2], while the large majority (n = 41) were BVDV-1b2, 13 in cattle, 27 in yaks [2] and one in a Bactrian camel (Camelus bactrianus) [31]. Four bovine strains (Gao et al., unpublished) [6] belonged to genotype BVDV-1.3. Two other bovine strains (S133 and TY05), reported as M [7] [8] resulted BVDV-1.6. One pig strain, reported as genotype O by Deng et al. [32] was clustered in the genotype BVDV-1.7, with four other strains all reported as genotype M, three from cattle, S43, S121 and BJ0924 [8] and one from camel, isolate 9 [31]. BVDV-1.8 (n = 5) was reported in three bovines [9] and two camels [31]. Genotype BVDV-1.10, reported as Q, resulted circulating in cattle (n= 6), pigs (n = 5) and camels (n = 3) [3] [4] [6] [8] [31] [32]. Two of the strains from camel, isolate 6 and isolate 7, have been reported as M by Gao et al. [31]. Only one bovine strain [8] resulted BVDV-1.11. BVDV-1.15.1 (clustered as M) were isolated in 56 bovines, 9 buffaloes, 4 Yaks, [2] [6] [7] [8] [10] 2 pigs [32] and one in a Bactrian camel [31]. BVDV-1.15.2 (28 strains) was reported as M in 13 cattle and 14 yaks [2] [8] and one in a Bactrian camel [31]. Six strains were BVDV-1.18, four cattle isolates reported as P [7] and two from Bactrian camel, isolate 5 and isolate 8, reported as M [31]. Genotype BVDV-1.21 (U) was reported in 14 bovines, 7 buffaloes and in two yaks (Sun et al., unpublished) [2]. The BVDV-1.23 strain S153 [8] was isolated in cattle. In Japan most of the strains isolated from cattle and contaminant of biological products belonged to BVDV-1 genotypes 1a and 1b1. Genotypes 1.3, 1.6, 1.7, 1.8 and 1.13 were also reported.

Some genetic variants appeared to be restricted in certain areas, as certain genotypes circulating only in Turkey or China, suggesting geographic isolation [7]. Most of BVDV-1 genotypes were related to specific geographic distribution, showing two main populations circulating in Asian and European countries, respectively. Apart few exceptions, a very limited number of strains isolated in Italy belonging to 1.6, 1.8 and 1.21, some genotypes were characteristic from Asia. Genotypes 1.10, 1.15, 1.18 and 1.23 were circulating exclusively in China, and also genotype 1.21 has been reported almost exclusively in China. Other genotypes (1.6, 1.7 and 1.8) appear to be restricted to Asian or Austral Asian countries. U-A pairing in V2/2 was present only among certain Asian genotypes (1.7, 1.10 and 1.15.1). Other genotypes were rare and present only in few countries, BVDV-1.16 bovine strains reported by as genotype L [33], and strains TR70,

TR73 and TR75 belonging to the BVDV-1.14, reported by Yesilbag et al. [34] as genotype R, were restricted to Turkey.

In the BVDV-2 species, while BVDV-2a showed cosmopolitan diffusion and 2b and other related genotypes were more frequently distributed in South America, only one group appeared specific to Asia. The genotype b variant 4 (BVDV-2b4) included only Chinese isolates: the bovine strain SD-1301 [5] and the contaminants S143, S172 and S51 [8]. BVDV-3 species showed low heterogeneity with four different genotypes. Out of four, three (BVDV-3.2, BVDV-3.3 and BVDV-3.4) were specific to zebu and bovine isolates from India and Bangladesh, respectively [12] [13]. Strains IndMDV18963/12 and IndBHA6604/12, representative for BVDV-3.2 genotype, were divergent in the species, scoring divergence 96.66% when compared with other BVDV-3 strains. Similarly, with divergence values of 82.14% for BGD/ZS5 and 100% for BGD/ZS3, respectively, bovine strains from Bangladesh were divergent from genotypes 1 and 2, constituting a separate genotype in the BVDV-3 species. In the BVDV-3 genotype 1, subgenotype 1 variants 2 (contaminant strain JS12/01) [35] and 3 (bovine strain Th/04_KhonKaen) [36] were reported from China and Thailand. BVDV-3 genotype 1 subgenotype 2 was specific to small ruminants from China (Shi et al., unpublished), with divergence 27.27%, divergence mean value 6.82 from other BVDV-3.1 genotype strains. BVDV-3 genotype 1 subgenotypes 3 and 4 were specific to zebu and bovine isolates from India and Bangladesh, respectively [12] [13]. Strains IndABI16023/12 and IndABI16020/12, representative for BVDV-3.1.3, showed a divergence of 23.33% from strains of other genotype variants. BVDV-3.1.4 strain BGDZS1 showed a divergence of 40%.

BDV species sequence characteristics of Chinese and Turkish strains were highly divergent from other genogroups, indicating geographic segregation. Chinese strains AH12-01, AH12-02 and AHHX15, reported as genotype BDV-3 (Ghiforn type―PNS BDV-j) [37], have been clustered as genotype BDV-d, sub genotype d1. These strains showed high homology with strain 297, also reported as BDV-3 [38], clustered in the same genotype, but as separate sub genotype BDV-d2. Similarly, other Chinese strains JS12/04, JSLS12-01 and JSYZ15, reported as genotype BDV-3 [37] have been clustered as genotype BDV-h. Both genotypes resulted partially related with BDV-j. However, qualitative non correspondence at root LVP level (BDV-d GC/GU/UA-AA-AG; BDV-h GC/GU-AU/UA-AA-AU; BDV-j GC/AU/UA-UG-GC-UC) (exception AA in V3/8 of strain BDV-j2 CH-BD1, discriminated from BDV-h with additional identification marker V1/3: BDV-j2 G-C; BDV-h A-U) and quantitative evaluation of divergence values justified separate clustering. Strains of sub genotype BDV-d1 were related only with BDV-j1 (India type) (no divergence values exceeding genotype limit value 9; divergence mean value 8.5) and divergent from sub genotypes BDV-j2 (divergence 92.85%, divergence mean value 10.78) and BDV-j3 (divergence 100%, divergence mean value 11.83). Strain 297 (sub genotype BDV-d2) resulted divergent from BDV-j3 (Ghiforn type) (divergence 100%, divergence mean value 11) and related with BDV-j1 and BDV-j2 (divergence mean values 8 and 8.14, respectively), but showing lowest divergence value mean of 6.5 with BDV-d1 strains. BDV-h resulted related only with BDV-j1 (homology 33.33%; divergence mean value 8.67) and divergent from other BDV-j sub genotypes (BDV-j2: divergence 61.9%, divergence mean value 9.67; BDV-j3: divergence 100%, divergence mean value 12.33).

Turkish strains TR-13 and TR-14, reported as distinct group in the BDV species [39], have been clustered as genotype BDV-i. Due to the divergence value of 9, obtained comparing their respective sequences, related to nucleotide variations particularly in the V1 stem, but maintaining clear base pairing homology in the rest of the secondary structure, they have been considered as two separate sub genotypes. BDV-i1 strain TR-13 showed bulges in V1/8 C C and V1/9 G A and BDV-i2 strain TR-14 showed V1/4 G A bulge, V2/12 insertion U and V1/8 A G bulge as characteristic bp in the sequence. Comparison with the other BDV sequences, applying species determination limit value 13, resulted related to the majority of BDV genotypes, showing a divergence percentage of 36.08, with divergence value range from 7 to 19 (mean 12.91). High divergence values have been observed with genotype BDV-f (including Aveyron and Pyrenean Chamois types) (values up to 17 - 19), BDV-g (Tunisian type), sub genotype 2 (values 15 - 16). High divergence values (16-19) have been obtained also with Pestivirus I (Turkish type Aydin―Burdur). However, strains TR-13 and TR-14 shared all BDV species markers. Comparison with the other BDV sequences, applying genotype determination limit value 9, resulted divergent from all other BDV genotypes, despite the partial relation scored with some members of genotype BDV-a, sub genotype 2, BDV-c, sub genotype 1, and BVD-j, which, however, did not correspond to relatedness when considering the entire group to determine overall homology (out of 192 comparisons, only 19 values were 7 - 9 and 13 corresponded to the limit value). Also BDV-j1 included the Indian ovine strain Ind 830-09 [40].

In the CSFV species, secondary structure analysis revealed corresponding main genomic groups, defined by phylogenetic analysis, dividing CSFVs into three major lineages (Brescia, Alfort and Kanagawa/Okinawa types) and their sublineages [41]. Three genetic clusters referred specifically to Asian countries. The CSFV genotype a variant 4 (type Parambi) included only pig and wild boar strains from India (Ravishankar et al., Tomar et al., Chandramohan et al., unpublished) [42]. CSFV pig strains clustered into genotype C (type Okinawa) were reported only from Japan and Taiwan (Liu, unpublished) [41] [43] [44]. The Chinese strain S171 [8], isolated from bovine serum, was clustered as CSFV-d. The strain S171 showed affinity with genotype CSFV-a2, but the difference was marked at the level of the V3 locus nucleotide base pairings. Exception made for a partial relation with CSFV-a2, S171 was divergent from all other CSFV types showing divergence values from 10 to 15. Divergence percentage with CSFV-a2 was 62.5%, with a divergence mean value of 9.87, resulting related, but belonging to a separate cluster. In addition, it was significantly less related to BDV species, a known characteristic of other CSFV strains (Figure 6 & Figure 7 ). Base pairings comparison between strain S171 and strains of the different BDV genotypes scored high divergence values, ranging from 17 to 25, with a mean value of 20.31. These values were very high in confront of the mean divergence value of 14.57 obtained comparing the other CSFV strains with BDV strains. Divergence from BDV-2 tentative species resulted also higher with CSFV-d (divergence values ranging from 21 to 22; mean divergence value 21.33), in confront of scores obtained with other CSFV genotypes CSFV-a, CSFV-b and CSFV-c (divergence values ranging from 15 to 20; mean divergence value 17.06). No relation was observed between CFSV and any other Pestivirus species (Figure 7). Other CSFV bovine strains have been reported from India and China, contaminants or associated to natural infection with clinical signs [8] [11] [45]. Indian strains, reported from the states of Tamil Nadu and Meghalaya, were genetically closely related to CSFV genotype a1 and a5 strains, respectively. However, a G*U pairing in V1/3 was peculiar only in the Indian bovine strains from Tamil Nadu. In other CSFV strains this position is characterized by a conserved A-U pairing, while G-C or G*U is conserved among BVDV-1, BVDV-2 and BVDV-3 strains (only 5 strains show exceptions, out of 1073 considered sequences).

Different genetic characteristics were specific to Asian clusters in different

Pestivirus species, showing great potential in the evolution of the genus. Until 1997, only two BVDV-1 genotypes were known, 1a and 1b. Thanks to the development of virologic procedures and the increased interest in the pathogen, different new genomic types have been described. However, the heterogeneity of the species become even more evident in particular by the contribution of scientists from China and other Asian countries. BVDV-1 is known to have a cosmopolitan diffusion, but genotype geographic distribution may vary considerably. For example, while predominant in UK [46], in China, BVDV-1a genotype is rare, first reported in cattle in 2010 [5] and in pigs in 2012 [32], and shows also heterogeneity, suggesting that the virus have the great variation in evolution between the different host species [5]. BVDV-1 species resulted heterogeneous especially in China. The number of different BVDV-1 genotypes increased progressively. BVDV-1b and 1m (PNS 1.15) are predominant [7]. Currently, BVDV-1 is represented by 12 types. Only in Italy, genetic diversity of BVDV-1 is higher, accounting for 14 different types [15]. Despite, the first description of the species in 1980 was related to 1b infected cattle imported from Europe [10] [47], some of the genetic variants further described appeared to be restricted to China. For example, the phylogenetic reconstructions indicated that the clustering of the Chinese BVDV-1m genotype in the phylogenetic tree is a result of geographic isolation [7]. The Chinese strain ZM-95 for many years was the only representative of a separate cluster in the species, suggesting the circulation of a rare Pestivirus restricted in pigs. The strain was the first BVD virus isolated in China (Inner Mongolia) from pigs showing clinical symptoms and pathological lesions resembling mild classical swine fever [48], and characterized as genotype M [49] (PNS BVDV-1.15). Zhu et al. [10] reported a noncytopathic strain, genetically closely related to the pig strain ZM-95, associated with a disease characterized by high morbidity and mortality, a first BVDV-1m virus originated from cattle. In the present study, many other closely related strains have been considered from recent reports from China. Only the bovine strains BJ11_01 and S73 isolated in China in 2011 and 2013, 16 - 18 years later (Zhang et al., unpublished) [8], shared all the nucleotide characteristics of ZM-95, and all the other newly reported strains clarified genotype identification parameters. The genotype 1.15 resulted the most commonly diffused in China, and not restricted in pigs but circulating in cattle, buffaloes, yaks and Bactrian camels. The yak strain M31182 (Sun et al., unpublished) showed also characteristic nucleotide base pairs indicating radical sequence variations generating new genotype in the BVDV-1 species. Reported as genotype U, PNS 1.21, the genotype was identified also in other Chinese bovine and buffalo sequences.

The heterogeneity revealed in the BDV species was among the highest observed among pestiviruses. Only BVDV-1 accounted for more genotypic variants [14]. Sequence characteristics of Turkish and Chinese strains were highly divergent from other genogroups, indicating geographic segregation. HoBi-like pestiviruses (BVDV-3) was less heterogenous, but characterized by three Asian lineages. Strains from India and Bangladesh [12] [13] showed genetic diversity and molecular characteristics clearly distinct from those previously reported circulating globally, from bovine fetal serum or naturally infected cattle and buffaloes in America, Europe, Thailand and Australia, highlighting the independent evolution of the species in the Indian subcontinent. Similarly, the CSFV species showed low heterogeneity. However, two Asian genotypes were highly divergent. According to primary sequence analysis, the strain Kanagawa/74 (PNS genotype CSFV-c) was divergent in the species [50] [51], and proposed as harmonized outgroup strain for phylogenetic analyses [52]. This suggests that strain S171 (PNS genotype CSFV-d) should be considered as another outgroup of the species with even more enhanced divergence, with atypical characteristics, possibly related to host species adaptation or a result of geographic isolation, occurrences suggested also for BVDV-1 isolates in China [5] [7]. During a survey for the detection and genetic characterization of 5’-UTR and E2 gene of CSFV from bovine population of the northeastern region of Indian state of Meghalaya, out of 134 cattle serum samples tested, all were positive in reverse transcription-polymerase chain reaction (RT-PCR) for 5’-UTR region and 10 samples were positive for CSFV antigen by a commercial antigen capture enzyme linked immunosorbent assay (Ag-ELISA) [45]. Full length E2 region of CSFV were amplified from two positive samples, CS/ML/911/IDP/13 [KY860532] and CS/ML/AF/Umiam/14 [KY860531]. Phylogenetic analysis showed similarity with isolates reported from the neighboring state of Assam, as the pig strain IND/AS/GHY/G4 [KM362426] [53]. According to E2 region primary sequence analysis, these strains were clustered in the CSFV sub-genogroup 2.1 [45] [53]. The E2 regions of the bovine strains CS/ML/911/IDP/13 [KY860532] and CS/ML/AF/Umiam/14 [KY860531] showed 99% nucleotide identity with IND/AS/GHY/G4 (PNS CSFV-a5) and 92% with strain Paderborn [GQ902941] (PNS CSFV-a2). 5’-UTR evaluation by PNS from complete genome of Paderborn and IND/AS/GHY/G4 suggested the appurtenance of both bovine strains CS/ML/911/IDP/13 and CS/ML/AF/Umiam/14 to PNS CSFV-a5. As PNS CSFV-a4 (type Parambi), also this cluster was characteristic to India.

The high level of heterogeneity was reflected also in a certain confusion in the nomenclature of types. For example, the Chinese camel isolate 9, clustered as M [31], diverged of only 2 bp from the strain AQGN96BI5 of genotype 1.7 (O), and 4 bp from the 1.7 reference strain IS25CP/01 [28], thus justifying the reallocation in this BVDV group. Similarly, the Chinese camel isolate 5, reported as M [31] was reclustered in the present study, showing relation with genotype 1.17 (P) reference strain TJ06 [7] (divergence 6 bp). Confusion was also due to the use of same names to define different genetic clusters, as in the case of strains TR-2007-A-2368MS, TR-2007-Gu-175454-4695 [33], TR1, TR16 and TR72 [34], clustered as genotype L, reported from cattle in Turkey, genetically distant from homonymous isolates from continental Europe [54] [55] [56].

It is unlikely that geographic specificity of some types might be due to sampling bias, taking into account that strains have been reported from qualified laboratories, often identifying in the same epidemiological survey typical isolates. In certain cases, observations have been further validated at the level of the International Committee on Taxonomy of Viruses. For example, atypical ovine Turkish isolates BDV/Aydin/04-TR and BDV/Burdur/05-TR [57], previously considered a separate Pestivirus species [58], a new BDV subgroup [57] or BDV borderline candidate for reclustering as new species [59], have been recently classified as new genus Pestivirus member species (Pestivirus I) [1]. Furthermore, it is highly probable that future investigations will reveal the existence of new types in the genus since most of routinely conducted diagnostic work is based on serology and only few isolates are reported from certain regions. Similarly, despite evidence of Pestivirus seropositive animals in a large number of wild animal species [60], the number of reported isolates is very limited and related to few species as deer (Cervus elaphus) and roe deer (Capreolus capreolus) [61] [62], or the Japanese serow (Capricornis crispus) [63], Pyrenean chamois (Rupicapra pyrenaica) [64], Alpine chamois (Rupicapra rupicapra) [65], and captive mousedeer (Tragulus javanicus) [66], wisent (Bison bonasus) and reindeer (Rangifer tarandus) [67].

Also investigations in man, may reveal new aspects of Pestivirus epidemiology. Using mass spectrometry proteomics to analyze protein extracts from three Zika positive brains of deceased babies with severe brain lesions and arthrogryposis, peptides from the polyprotein of a Bovine-like viral diarrhea virus have been detected, suggesting that Zika virus may not be the only etiological agent responsible for microcephaly [68]. During the outbreak in Paraíba, in 2015, BVDV RNA was also found in the amniotic fluid collected from four mothers with babies affected by Zika and microcephaly. Two 5’-UTR RNA genomic sequences, have been provided by Prof Tanuri, Federal University of Rio de Janeiro, for further secondary structure analysis. The strain 4p, considered in the present study, belonged to genotype BVDV-2b, typical genotype circulating in South America, genetically close to Brazilian strains LV56-1013 [69] and LVPatol0209 (Silveira et al., unpublished), but showing unreported variants in the IRES. The second isolate (12p) was a BVDV-1b1, genotype of cosmopolitan diffusion, very similar to bovine strains UEL9-BR/11 (Rodrigues et al., unpublished), isolated in Brazil in 2011, about 10 years later than the three other closely genetically related strains 133/02, 4092/00 and 3310/01 [70] previously identified in Spain in 2002. The human strain 12p showed also homology (98% nucleotide identity) with four contaminant strains isolated in Mexico in 2012, the BVDV-1b1 NGR2, NGR3, NGR11 and NGR12 (Gomez-Romero et al., unpublished). The only strain previously reported in humans belonged to genotype BVDV-1.3 [71], another cluster with cosmopolitan diffusion.

Observed heterogeneity in the Pestivirus species has to be considered for potential implications on diagnostic tests, control and preventive measures, since commonly available tests and vaccines are based on viral antigenic substrate [72]. Due to possible geographical segregation, recognition of the molecular characteristics of field strains present in a population is fundamental for the control or eradication programs design, vaccine development or retracing infection sources in case of outbreaks [46] [73]. Laboratory testing difficulties may be due to serological cross-reactivity or divergence among pestiviruses. Serological surveillance of BVDV by ELISA does not distinguish between BVDV and BDV as source of infection. During routine genetic typing of pestiviruses in India, BDV was detected in sheep by real time RT-PCR [40]. All the samples yielded positive virus isolates in cell culture but were found negative by a BVDV antigen ELISA, suggesting that for diagnosis of BDV infection, the commercial BVDV Ag-ELISA should be used with caution. For adequate differentiation between BVDV and BDV, cross-serum neutralization test procedure has been recently developed [74]. Current BVDV diagnostic tests may fail to detect HoBi-like viruses or to differentiate between BVDV and HoBi-like viruses [75]. Furthermore, available commercial serological tests for BVDV do not reliably detect HoBi-like virus exposure, and cross protection against HoBi-like viruses conferred by current BVDV vaccines is likely limited [75]. Therefore, accurate genetic analysises are necessary for epidemiological studies, traceability and characterization of atypical strains.

The PNS method allowed an exhaustive representation of genomic variants in the genus, based on the comparative evaluation of about 1,400 Pestivirus strain sequences. The particularity of the PNS method is the exclusive consideration of strategic genomic sequences corresponding to the 5’-UTR IRES. Thus, at this level stable nucleotide variations assume high importance in terms of virus evolutionary history. Theoretically, applicable for the evaluation of all positive polarity RNA viruses, the PNS method was used for the determination of genotypes of the human hepatitis C virus [76], considering the unambiguous functional and genomic similitude between the two genera [77]. PNS sequence characterization easily discriminated species as Giraffe, Pronghorn or Bungowannah [78] [79]. Even if E2 glycoprotein gene is preferred in recent taxonomy of the species [80], CSFV was evaluated with 5’-UTR, and otherwise not comparable with 5’-UTR sequences of the other different species. Despite a limiting factor for differentiation of closely related isolates [80], short length target fragments in other regions of the viral genome as 5’-UTR or NS5B have been largely applied for genotyping purposes and molecular epidemiology [50] [52] [81]. In addition, according to the Terrestrial Manual of the World Organization for Animal Health (Chapter 2.8.3 CSFV; Paragraph 1.1.5.) [82], the 5’-UTR of the CSFV genome (150 nucleotides) is also foreseen for genetic typing and molecular epidemiology of CSF. However, in light of the movement towards whole genome sequencing and recent species typing [1], the routine work performed in veterinary laboratories in the future will possibly be based on full-length genome comparison to delineate Pestivirus genogroups and species, despite still problematic due to lack of a sufficient number of sequences representing existing genetic variants. For example, current classification of BVDV isolates is generated using short genome fragments, giving that 97.7% of deposited sequences are restricted to 5’-UTR or Npro, mainly [16].

In addition to accurate virologic investigations, it is important to understand ways of diffusion and in particular prevent spread of atypical genetic clusters. The major concern is represented by the fact that if genetically related types induce effective cross immunity, at the contrary, divergent genetically atypical types might pose problems in terms of diagnostic reliable detection or vaccine conferred cross protection against such viruses. In addition, even non-naïve local animal populations might be more exposed in case of heterogeneous strains due to non-sufficient cross immunity. Trade of live animals is generally considered the main route of pathogen diffusion. For example, in UK, only three BVDV-1 types were detected in 1999. In 2013, phylogenetic analysis demonstrated the existence of six genotypes of BVDV-1 circulating in the country. This indicated that restocking of cattle from continental Europe has increased the genetic diversity [46]. Unless illegal animal movements occur, trade is the relatively easier controllable and preventable among ways of introduction of the virus in free herds or free areas. Contamination of biological products and iatrogenic infection are probably also relevant aspects of the virus epidemiology. This recalls also the risk of spreading exotic strains, requiring not only controls on live animals, animal products, semen, ova and embryos, but also potentially contaminated biological products. Eighty-three strains contaminant of biological products, considered in the present study, were reported mainly from Asian countries. Commercialized batches of BVDV-1 contaminated bovine serum have been reported also in China [8]. The contamination of fetal bovine serum (FBS) as a vehicle for the rapid spread of the pathogen is a known and not really infrequent phenomenon [29] [83]. This might be the source of introduction of BVDV-3 in Italy, first identified in Europe in FBS imported from Brazil [84], and further reported in Italian cattle herds with respiratory distress and reproductive failures [85] [86] [87]. A similar hypothesis could be formulated for the BVDV-1 genotype 1.21. Some characteristic Asian strains of the genotype 1.21 have been shown to circulate also in southern Italy [88], suggesting ways of diffusion other than direct contact among animals. Being unlikely the direct import of live animals from China, other factors were implicated. The strain 441/09, isolated in 2009 in Puglia region, showed high similarity with the originally reported Chinese yak strain. The other two strains, 130/15-4215 and 130/15-5364, have been isolated six years later in 2015 in Sicily. Their sequences were very similar to each other. When compared to other BVDV-1 genotypes, both strains resulted divergent from most of other types, with high divergence values as 20 bp with the European genotype 1.22. However, the lowest divergence values were obtained with strains belonging to genotype 1.21, justifying allocation in the cluster, but in a separate sub genotype 1.21.1. These observed genomic variations were coherent with virus evolutionary adaptation related to spatial temporal factors. Strain 441/09 may represent the evidence of the introduction of an exotic Pestivirus variant in the animal population in Italy, while strains 130/15-4215 and 130/15-5364 may represent the result of the virus adaptational evolution in the new environment, after an adequate period of time. This suggest the introduction of genotype 1.21 from China in 2009 and the potential risk of the endemic status of the virus in the area, supported by evidence of genomic characteristics mutations, possibly related to geographical situation or contact with virus populations circulating in the region. Interestingly, the Italian bovine strain SI/207/12 [89], reported as genotype T, was clustered in the genotype 1.6, scoring low divergence values with most of the strains in the group, all reported as genotype BVDV-1n, essentially of Asian origin, including the Japanese strains so CP/75 [90] and Shitara/02/06 [91] or the South Korean strains 06z71, 06z127 (Yang and Kweon, unpublished) and KB01 (Oem et al., unpublished).

Other ways of diffusion may be revealed by the unusual circulation of CSFV in animal host different from suids. CSFV is generally considered to be restricted to domestic and wild suids. Current knowledge indicates that experimental infections without clinical signs have been reported in cattle, sheep, goats and deer, but there is no evidence that these species become infected in nature [92] [93] [94] [95]. Only in one case, hog cholera virus has been identified in sheep [70]. In Spain, the reported 5440/99 ovine strain showed sequence similarities with CSFV vaccine strains used in pigs, suggesting spillovers from prophylactic campaigns. Another ovine isolate (strain 12 Ovine liver 113nt) [96], reported from United Arab Emirates in the framework of investigations for pestiviruses, was suspected as CSFV. CSFV bovine strains detected in China might find their origin from vaccine prophylaxis performed in domestic animal populations [11]. Anti BVDV killed virus vaccines are available in China. However, for long time there were no commercial BVDV vaccines in the Chinese market. In cattle and yaks, BVDV was prevented by the wide application of triple dose of the live attenuated hog cholera lapinized vaccine (HCLV) strain in different areas of China, and especially in Tibet and Qinghai provinces. Since previous experimental studies conducted by Yuan Qingzhi in 1957 on HCLV, the only one CSFV authorized vaccine in China, efficacy and safety of the prophylactic use for BVDV in lactating and pregnant cows, calves and yaks have been demonstrated [97]. In Tibet, the immunization with HCLV of bovine species resulted beneficial in reducing losses due to BVDV infection. However, long-term use of live attenuated HCLV gave the opportunity for vaccinal strains adaptation in cattle, with further natural diffusion as possible consequence.

In the present study, according to secondary structure analysis, the genus Pestivirus resulted heterogeneous. Sequence characteristics among Asian genomic clusters within the different Pestivirus species suggested geographic segregation and occurrence of micro-evolutive steps in the genus evolutionary history. This aspect was particularly evident in atypical sequences originated from China or Turkey, indicating risk of diffusion by animals and products trade or contamination of biological products as bovine calf serum, with potential diagnostic and control difficulties. Determination of species heterogeneity is important for diagnostic efficiency and prophylactic purposes, taking into account possible wider animal host range by different Pestivirus species, causing overlapping clinical features, potential influence on eradication programs, and eventual regulatory measures.

The authors declare no conflicts of interest regarding the publication of this paper.

Giangaspero, M., Zhang, S.-Q. and Apicella, C. (2019) Heterogeneity of Pestivirus Species in Asia. Advances in Microbiology, 9, 266-342. https://doi.org/10.4236/aim.2019.93019