Genomic characterization and evolutionary analysis of three bovine viral diarrhea virus 1b strains from Taiwan, highlighting genomic evidence of cytopathic evolution

Abstract

Background

Bovine viral diarrhea virus type 1 (BVDV-1) affects cattle health and productivity worldwide. A nationwide surveillance of BVDV-1 showed persistent circulation in Taiwan in past decade. It exists in two biotypes—noncytopathic (ncp) and cytopathic (cp), which differ in clinical presentation and pathogenic mechanisms. Despite its prevalence in Taiwan, the genomic features that distinguish these biotypes remain poorly characterized.

Result

We recently identified two Taiwanese BVDV-1 in clinical samples, namely BV27 and BV236, from a beef cattle without clinical symptom and a dying dairy cow, respectively, in 2023. Whole genome sequencing using the Illumina NovaSeq X Plus platform revealed genome lengths of 12,207 bp for BV27 and 15,047 bp for BV236. Notably, BV236 derived from cattle exhibiting severe diarrhea prior to death contained a duplicated NS3-NS4A-partial NS4B region and a 240 bp insertion of ubiquitin C gene derived from bovine, a feature commonly associated with causing severe symptom and cytopathic biotype. Moreover, a long-range PCR further revealed the coexistence of two viral genomes within the same sample, one carrying these genomic modifications and one lacking them. Phylogenetic analysis classified both viruses as BVDV-1b, the same subgenotype as 18H17, a previously reported Taiwanese strain. Of note, haplotype network analysis indicated that BV27 and BV236 are genetically distinct from 18H17, which indicated multiple strains of BVDV-1b circulating in Taiwan possible occurred through transboundary transportation.

Conclusion

 This study provides both clinical and genomic evidence on Taiwanese BVDVs. We identified two genetically distinct BVDV-1b strains, and although phenotypic assays were not performed, one strain (BV236) exhibited genomic hallmarks of cytopathogenic evolution that corresponded with severe respiratory symptoms. These findings suggest the coexistence of multiple BVDV-1b lineages in Taiwan and highlight their genomic diversity and potential pathogenic evolution. While based solely on genomic data, this work establishes a foundation for understanding BVDV diversity in Taiwan and underscores the need for continued molecular surveillance and future functional studies to clarify the role of these genomic features in cytopathogenicity.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12917-025-05130-y.

Background

Bovine viral diarrhea virus (BVDV) is a positive-sense single-stranded RNA virus belonging to the Pestivirus genus in the family of Flaviviridae [1]. BVDV is classified into three distinct species: Pestivirus bovis (commonly known as BVDV type 1), Pestivirus tauri (BVDV type 2), and Pestivirus brazilense (BVDV type 3 or Hobi-like pestivirus) based on the relatedness of nucleotide, amino acid sequence, and antigen relatedness, and hosts of origin [2]. BVDV Type 1 (BVDV-1) is a highly contagious pathogen that causes significant economic losses in the bovine industry. Since the first identification of BVDV-1 in North America, this virus has been one of the most widespread cattle pathogens, with an average of over 40% seroprevalence at the animal level and over 50% prevalence at the herd level worldwide [3, 4]. Moreover, the infection of BVDV-1 would highly correlate to severe diseases including bovine respiratory disease complex, diarrhea, mucosal disease, and reproductive abnormalities such as oophoritis and subsequent ovarian dysfunction that cause impaired fertility in female cattle [5–7]. The average direct economic loss per dairy cattle due to BVDV-1 infection worldwide reaches 24.85 US dollars without considering other costs of prevention, medicine, and extra-forage [8].

The genome of BVDV-1 is approximately 12.5 kilobases in length and is composed of a large single open reading frame flanked by an uncapped 5′ untranslated region (5′ UTR) and an unpolyadenylated 3′ untranslated region (3′ UTR) [5]. The 5′ UTR contains an internal ribosome entry sequence that directs ribosomes to initiate translation, while the 3′ UTR contributes to viral replication [9, 10]. The coding region produces a polyprotein that is post-translationally cleaved by host and viral proteases into 11–12 mature proteins, Npro, C, E^rns, E1, E2, p7, NS2, NS3, NS4A, NS4B, NS5A, NS5B [11].

Currently, 23 subgenotypes (1a to 1w) of BVDV-1 were identified based on the partial 5′UTR sequences [12, 13]. Nevertheless, subsequent genetic analysis revealed that the nucleotide sequences of other genomic regions—specifically Npro, E2, and a portion of NS3–NS4A—also displayed relative consistency within the same subgenotypes [14, 15]. BVDV subgenotypes 1a and 1b, the most prevalent subgenotypes, are widely distributed globally [12]. However, subgenotype distribution also exhibits geographic specificity. For instance, the subgenotype 1c is predominant in Australia [16, 17], while the subgenotype 1 d, 1e, and 1f are more commonly found in several European countries [12].

BVDV-1 exists in two biotypes, cytopathic (cp) and non-cytopathic (ncp) biotypes, that could be determined at clinical signs or the occurrence of cytopathic effects of epithelial cells in cell culture [18, 19]. The ncp biotype usually exhibited mild symptoms, but has a strong correlation with persistent infection (PI). PI animals serve as major reservoirs, enabling long-term viral persistence within herds [20, 21]. The PI calves continuously shed the virus throughout their lives via feces, urine, saliva, and nasal discharge, serving as a constant source of infection to other animals [22, 23]. Virus is also secreted in milk, semen, uterine secretions, and aborted fetal membranes to cause vertical virus transmission. If a PI calf survives to adulthood, it can continue to transmit the virus through these routes [24, 25]. Additionally, the immune suppression induced by BVDV-1 makes PI calves more susceptible to illness [26]. On the contrary, the cp. biotype is associated with severe symptoms in cattle, including lethal mucosal disease [5]. The cp. virus typically arises from the ncp virus by point mutations, or through RNA recombination, which either incorporates host cellular transcripts or the viral RNA genome, resulting in genetic changes, for instance, the cleavage of NS3 from NS2 or the expression of NS3 alone [27, 28]. On the other hand, cp virus strains generally incorporate cellular ubiquitin or ubiquitin-like sequences; however, some cp strains also exhibit the incorporation of the DnaJ protein or heat shock protein 40 (Hsp40), referred to as Jiv [29].

In Taiwan, information on BVDV epidemiology and genetic diversity is scarce. A nationwide surveillance study, published recently based on samples collected in 2014, revealed that BVDV is widespread in Taiwan, with a herd-level seroprevalence of 59.2% and an average animal-level seroprevalence of 32.4% [30]. Persistent circulation and regional differences were observed, and BVDV-1b was identified as the predominant subgenotype, along with BVDV-1a and BVDV-2a [30]. Beyond this report, no further epidemiological investigations of BVDV in Taiwan have been conducted, aside from a brief sequence announcement of one local isolate (18H17) [39]. Consequently, the current epidemiological landscape of BVDV in Taiwan remains largely unknown, and the biotypes of circulating strains have not been investigated. To address this gap, the present study conducted whole-genome sequencing of two clinical bovine samples—with and without severe respiratory symptoms—to characterize the genetic features of Taiwanese BVDV-1b strains and explore their potential biotype differentiation.

Methods

Sample description

The complete genome sequences of two bovine samples diagnosed as bovine viral diarrhea virus (BVDV) positive were analyzed. Sample BV27 was obtained from a tracheal swab of a beef cow exhibiting cranioventrial lung lobe consolidation with abscesses. The animal was examined by an experienced veterinarian during a routing slaughterhouse inspection. Sample BV236 was collected from the ileal mucosa of a Holstein heifer showing severe diarrhea with multifocal pinpoint erosive lesions in the nostril, soft palate, and gastrointestinal mucosa. This case was referred to the Animal Disease Diagnostic Center at the College of Veterinary Medicine, National Chung Hsing University, for diagnostic evaluation in 2023. The molecular diagnosis of BVDV was as described [30].

RNA extraction

The RNAs were extracted from the BVDV samples by TRIzol™ (ThermoFisher Scientific). The 200 µL of swab samples were well-mixed with 500 µL of TRIzol™ reagent, followed by the addition of 100 µL of chloroform. For phase separation, the solution was centrifuged at 12,000 ×g for 15 min at 4 °C. The aqueous phase was transferred to a new tube and mixed with the same volume of isopropanol for RNA precipitation at −20 °C for 30 min. Subsequently, the solution was centrifuged using 12,000 ×g for 20 min, followed by a brief wash with 70% EtOH and centrifuged at 8500 xg for 10 min. After EtOH removal, the RNA pellet was air-dried for a few minutes, then dissolved in 30 µL of DEPC-treated distilled water.

NGSs for viral genome sequencing and prepared sequencing data

The two Taiwanese BVDV samples were sequenced by paired-end 150 bp read length kit in 10B flow cell of the Illumina NovaSeq X Plus (Genomics Ltd., New Taipei City, Taiwan). The sequencing files were retained over Q38 score sequencing reads, and all qualified reads were trimmed of sequencing adapters by fastp [31]. The Bos taurus genome dataset, ARS-USD2.0 (GenBank no. GCA_002263795.4) serving as the host reference, was extracted from National Center for Biotechnology Information (NCBI), which was applied to remove host reads in sequencing data by Hisat2 [32].

Viral genome assembly and virus sequence identification

To assemble the BVDV genomes, the Trinity and SPAdes algorithms were applied to conduct viral assembly [33, 34]. The host cleaning read files were uploaded to the Galaxy website (usegalaxy.org) to use the Trinity for de novo assembly without a reference genome [35]. The SPAdes package was used for offline de novo assembly on a personal computer. To identify the virus contig, we made the virus reference dataset that was downloaded from NCBI Virus, and then imported it into the Magic-BLAST v1.6.0 [36].

Gene annotation and multiple alignment

The twelve BVDV-1 gene sequences were extracted from NADL, a reference BVDV-1 strain in GenBank, NCBI. The annotation of the two local strains identified in this study (namely, BV27 and BV236) used higher than 70% similarity to reference viral gene sequences by Geneious Prime Java version 21.0.4 + 7-LTS using Live annotate & Prediction (https://www.geneious.com/). To estimate the genetic distance among Taiwanese BVDV-1, the FFT-NS-i x1000 method of MAFFT was applied to multiple alignment of nucleotide and amino acid sequences. The similarity among strains was calculated using Blosum62 with a threshold of 1.

Verification of unique and host insertion sequences on BV236 genome

To verify the unique and host insertion sequences of BV236, the unique genome region was amplified by PCR, and the identification of the resulting amplicon was confirmed by automated sequencing (Mission Biotechnology, Taipei, Taiwan). Briefly, 1 µg RNA of BV236 was used for reverse transcription via SuperScript™ IV manual (Thermo Fisher Scientific, Inc., Waltham, MA, U.S.A.). The 1 µL of cDNA and 10 µL of KAPA Hifi 2× ReadyMix were mixed with 1 µL of each primer (10 µM), targeting outside of the unique region (Ubi_Check_F: 5′-CCTTCGGTGGTGAGTCGGTATC-3′ and Ubi_Check_R: 5′-CTTGGTGTAGTCCCTCTTGGCAT-3′), and 7 µL of nuclease-free water. The thermal conditions were 95 °C for 3 min and 35 cycles of 98 °C for 20 s, primer annealing of 65 °C; for 20 s; extension at 72 °C for 20 s; and a final extension at 72 °C for 1 min.

Additionally, to confirm whether BV236 represented a co-infection, KAPA HiFi HotStart ReadyMix was applied to amplify the unique viral region containing the viral duplication and host insertion in BV236. Briefly, 10 µL of 2× ReadyMix was combined with 0.75 µL of 10 µM primer (NS2-Out-F: 5′-CGCTAGGGGGCAATTGTTCTT-3′; NS4B-Out-R: 5′-TACGCTCTCCAGTCTAGTAGGGG-3′), 1 µL of cDNA, and nuclease-free water to a final volume of 20 µL. PCR conditions were as follows: initial denaturation at 95 °C for 3 min; 27 cycles of 98 °C for 20 s, annealing at 67 °C for 30 s, and extension at 72 °C for 3 min; followed by a final extension at 72 °C for 6 min in an Applied Biosystems^® 2720 Thermal Cycler (Thermo Fisher Scientific, Inc., Waltham, MA, USA). The amplicons were separated by 0.7% agarose gel electrophoresis. DNA fragments of approximately 5.7 kb and 2.9 kb were excised and bidirectionally sequenced using the same forward and reverse primers by automated sequencing (Mission Biotechnology, Taipei, Taiwan). To extend the sequences, the ~ 5.7 kb DNA fragment was continuously sequenced bidirectionally using primers targeting the host ubiquitin C insertion region (Ubi-F: 5′-GACCACGTGAGCAGAGGTGG-3′; Ubi-R: 5′-GACCAGGTGCAGAGTGGACTCTT-3′). Similarly, the ~ 2.9 kb DNA fragment was extended by primers positioned near the ends of the initial sequencing results to ensure overlapping coverage (NS3-F2: 5′-GGGAAAGAATGAGGAATCTAAACC-3’; NS4A-R2: 5′-GTTACCCCTGACTCTATGGCAT-3′). Sequence reads were mapped to the BV236 genome using the Bowtie 2 algorithm implemented in Geneious Prime Java version 21.0.4 + 7-LTS (https://www.geneious.com).

Construction of phylogenetic trees

The BVDV-1 genomes, including 5′ and 3′ untranslated region, and coding region were downloaded from GenBank, NCBI (Supplementary Table 1). The MAFFT algorithm using FFT-NS-i method in Galaxy Sever version 24.1.3dev was applied to all sequence alignments for phylogenetic tree construction [35]. The phylogenetic tree was constructed using IQ-TREE 1.6.2 with 1000 times bootstraps with the best model, and the visualization of the tree was conducted by FigTree (http://tree.bio.ed.ac.uk/software/figtree/).

Haplotype network analysis of Taiwanese BVDVs

Population analysis with reticulate trees (PopART) version 1.7 was applied to the haplotype network analysis of full E2 and NS2-NS3 sequences of BVDV subgenotype 1b [37]. The shortest network route was calculated by Minimum spanning networks [38]. A total of 41 BVDV-1b strains with whole genomes, including 3 Taiwanese BVDV genomes, were analyzed using haplotype network analysis (Supplementary Table 1).

Result

Histopathologic features of the two cattle

In this study, BVDV sequences were obtained from two Holstein cattle presenting with different clinical backgrounds. BV27 was isolated from a beef cow sent to the slaughterhouse, which showed no overt clinical signs. In contrast, BV236 was derived from a Holstein heifer exhibiting severe diarrhea prior to death. By histopathologic examination, BV236 showed severe lymphoplasmacytic gastroenterocolitis with erosions/ulcers disseminated throughout the mucosa of the soft palate, rumen, reticulum, omasum, and nostril. Within colonic and ileal mucosa, there were marked epithelial necrosis, crypt abscessation and crypt herniated into the submucosa (Fig. 1A and B). Peyer’s patches were severe lymphoid depletion. Mild tracheitis and interstitial pneumonia were also present (Fig. 1C and D). Unlike BV236, BV27 showed marked cranioventral suppurative bronchopneumonia. Multifocal to coalescing, well-demarcated necrosis with bacterial colonies and mild interstitial pneumonia were noted (Fig. 1E and F).

Fig. 1.

Histopathologic features of the two cattle. A dairy cow (BV236) and a beef cow (BV27) were included in this study. Gross lesions (A, C, and E) and corresponding histopathologic changes with H&E staining (B, D, and F) were recorded. Panels A and B show ileal lesions, and panels C and D show lung lesions in BV236. There were reddish plaque and rugal folds like (arrow) thinned ileal wall (A). Low-power photomicrograph of (A) illustrate lymphoid depletion, crypts ectatic and invade into the submucosa and underlying Peyer’s patches of the ileum (B). The majority of the lung parenchyma was mottled red (C). Alveolar septa and perivascular interstitium were infiltrated by small numerous of lymphocytes, plasma cells, and fewer neutrophils (D). Panels E and F illustrate lung lobes in BV27. Approximately 30–40% of the lung parenchyma in the cranioventral regions was dark red, firm and consolidation with multifocal to coalescing, well-demarcated and variably sized yellow-white nodules (E). The bronchioles and surrounding alveoli were obliterated by large numbers of neutrophils, cellular debris, fibrillary fibrin, and admixed with clusters of basophilic bacterial colonies (F). The interstitium of the lung parenchyma was diffusely infiltrated by small numerous of lymphoplasmacytic cells, macrophages, and fewer neutrophils along with proteinaceous fluid

The differing clinical presentations and pathological findings associated with these two cases provide important context for interpreting the genetic and pathogenic features of the corresponding viral strains.

Genome assembly of two biotype Taiwanese BVDVs

High-throughput sequencing using the Illumina NovaSeq X Plus platform generated 26,549,978 and 27,552,891 raw reads from BV27 and BV236 strains, respectively. Among these, a total of 8,994 reads for BV27 and 8,920 reads for BV236 were identified as BVDV-related sequences. The assembled genomes exhibited high coverage depths of 110.5× for BV27 and 88.9× for BV236, providing robust data for downstream genomic analysis.

Genome structure of two Taiwanese BVDVs

The complete genome of the strain BV27 was determined to be 12,207 bp in length, comprising a 371 bp 5′ UTR, an 11,694 bp coding sequence, and a 142 bp 3′ UTR (Fig. 2). In contrast, the strain BV236 exhibited a significantly longer genome of 15,047 bp, including a 372 bp 5′ UTR, a 14,493 bp coding sequence, and a 182 bp 3′ UTR. The complete genome sequences of BV27 and BV236 were submitted to GenBank under accession numbers PV426615 and PV426616, respectively (Supplementary file 1).

Fig. 2.

Whole-genome structure of two Taiwanese BVDV strains. A The BV27 strain has a genome length of 12,207 bp, comprising an 11,694 bp coding region flanked by a 371 bp 5′ untranslated region (UTR) and a 142 bp 3′ UTR. In contrast, the BV236 strain has a genome length of 15,047 bp, including a 14,493 bp coding region, a 372 bp 5′ UTR, and a 182 bp 3′ UTR (B). BV236 harbors a 2,559 bp duplication of the NS3-NS4A-partial NS4B region (highlighted in pink) and a 240 bp insertion corresponding to a partial bovine ubiquitin C gene (ubi^#), highlighted in yellow

Notably, genome annotation revealed unique sequence features in BV236 genome. It contains a 2,559 bp duplication encompassing the NS3-NS4A-partial NS4B coding sequences. In addition, a 240 bp insertion, identified as a partial sequence of the bovine ubiquitin C gene, was located upstream of the duplicated NS3 region (Figs. 2B and 3A). Moreover, there are four non-viral residues (i.e. SRGG encoded by AGCAGAGGTGGG) present between the partial NS4B and Ubiquitin C (Fig. 3A). However, the aforementioned features were not observed in BV27 (Fig. 2A).

Fig. 3.

Verification of host ubiquitin C sequence insertion in the BV236 genome. A Schematic representation of the genetic structure of BV236. A partial ubiquitin C coding sequence (UBC CDS) was aligned to the BV236 polyprotein using Geneious Prime. The UBC CDS is inserted within the duplicated region spanning NS3–NS4A–partial NS4B, specifically located between partial NS4B and NS3. B Validation of the insertion by PCR using primers targeting the flanking regions in NS4B and NS3. A single amplicon of the expected size (379 bp) was obtained, as indicated by the arrow. C Sanger sequencing confirmed the inserted fragment. The chromatogram shows the inserted ubiquitin C sequence, along with the flanking nucleotide regions encoding the amino acid motifs SRGG (upstream) and GPAV (downstream, within the NS3 region) in the BV236 genome

To further confirm the presence and precise location of the host-derived ubiquitin C insertion in the BV236 genome, PCR amplification was performed using primers targeting the regions upstream and downstream of the insertion site, followed by Sanger sequencing. A PCR amplicon of the expected size (379 bp) was obtained (Fig. 3B and Supplementary file 2). As shown in Fig. 3C, sequence alignment confirmed the insertion of the host ubiquitin C sequence within the BV236 genome assembly. Additionally, high-quality chromatograms confirmed the sequences flanking the insertion site. These regions encode the amino acid motifs SRGG (upstream) and GPAV (downstream; on the NS3), located on either side of the ubiquitin C insertion in the BV236 genome (Fig. 3C).

Next, to further examine whether multiple viral strains were present in the BV236 sample, a long-range PCR was performed across the flanking regions of the putative duplication/ubiquitin insertion (Fig. 4A). This assay yielded two distinct amplicons: a larger fragment (~ 5.7 kb) containing both the viral gene duplication and the ubiquitin C insertion, and a smaller fragment (~ 2.9 kb) lacking the insertion (Fig. 4B). These results demonstrate the coexistence of two viral genomes in the same clinical sample, hereafter referred to as BV236 (with modification of duplication and insertion) and BV236-s (without modification). Notably, the duplicated sequences in BV236 were almost identical to the corresponding genomic region in BV236-s, differing by only a few nucleotides that did not alter the encoded amino acids.

Fig. 4.

Validation of BV236 genotypes by long-range PCR. A Schematic of the PCR strategy. To determine whether BV236 represented a co-infection, PCR was performed with primers targeting unique viral regions (NS2 and NS4B; blue arrows), designed to amplify the genomic region containing the viral duplication and host-derived insertion. B PCR results. Two amplicons of approximately 5.7 kb and 2.9 kb were obtained. The corresponding genomic features represented by the two products are illustrated. M: DNA marker; Mock: RNA extracted from uninfected MDBK cells

Genetic analysis of Taiwanese BVDVs

Phylogenetic analyses based on both the 5′ UTR and the complete polyprotein open reading frame confirmed that the two Taiwanese isolates, BV27 and BV236, belong to the BVDV-1b subgenotype (Fig. 5A and B). Notably, both strains clustered closely with BVDV-1b strains previously isolated in the United States. Specifically, in the 5′ UTR-based tree, BV27 and BV236 grouped within the same clade as strain MN188074 (Fig. 5A), while in the polyprotein tree, they were most closely related to strains KP941589, KP941590, and KP941592 (Fig. 5B).

Fig. 5.

Subtype identification of Taiwanese BVDV strains BV236 and BV27 through phylogenetic analysis. A Phylogenetic tree based on the nucleotide sequences of the 5′ untranslated region (UTR), and (B) Tree based on the complete polyprotein coding sequences. A total of 180 BVDV-1 strains were analyzed, including three Taiwanese isolates: BV27 and BV236 (identified in this study) and TW18H17 (previously reported). Phylogenetic trees were constructed using the maximum likelihood method implemented in IQ-TREE v1.6.12. Branch support values were assessed via 1,000 bootstrap replicates. The three Taiwanese strains are highlighted in red

Recently, the genome sequence of another Taiwanese BVDV-1b strain, 18H17, was reported [39]. Although all strains belong to the same subgenotype, phylogenetic analyses revealed that BV27 and BV236 are more closely related to each other and clearly distinct from 18H17, which consistently formed a separate branch in both the 5′ UTR and polyprotein trees. This genetic divergence suggests that BV27 and BV236 did not evolve directly from 18H17, but rather represent a distinct introduction or an independent lineage within the BVDV-1b subgenotype.

Subsequently, the genetic similarity and divergence among the three local BVDV-1b strains were assessed through pairwise distance analysis. Comparisons were made based on the full-length polyprotein sequences as well as the individual amino acid sequences of all twelve viral proteins. Since the insertion of ubiquitin C and the duplication of the NS3–NS4A–partial NS4B region are unique to the BV236 strain, an additional analysis was performed using a modified polyprotein sequence of BV236 with these elements excluded. This approach allowed for a more accurate estimation of sequence divergence attributable to recombination events, rather than overall evolutionary distance (Fig. 6).

Fig. 6.

Pairwise comparison of polyprotein and individual viral protein sequences among Taiwanese BVDV strains. Pairwise amino acid sequence similarity and identity of the 12 individual viral proteins and the full-length polyprotein (FL) were compared between BV27 and BV236 (A), as well as between each of these and the previously reported Taiwanese strain 18H17 (B). Analyses were conducted using the BLOSUM62 substitution matrix with a similarity threshold of 1. Note: “FL-ex” refers to the polyprotein of BV236 excluding the insertion and duplicated regions

Overall, BV27 and BV236 exhibited high sequence similarity across most individual viral proteins, while both showed greater genetic divergence from the 18H17 strain (Fig. 6). Among the 12 individual genes and the both UTRs, the E2 gene displayed the highest level of divergence. The amino acid similarity of the E2 protein between BV27 and BV236 was 98.93% (Fig. 6A), whereas the E2 protein of 18H17 shared only 95.72% similarity with BV27 and 95.45% with BV236 (Fig. 6B).

Interestingly, despite their overall high similarity, the full-length (FL) polyprotein sequence similarity between BV236 and BV27 was only 80.44%, which is notably lower than that between BV27 and 18H17 (98.59%). This discrepancy is primarily due to the unique insertion and duplication events present in BV236, but absent in BV27. Indeed, when these distinct genomic elements—namely the ubiquitin C insertion and the NS3–NS4A–partial NS4B duplication—were excluded from the BV236 sequence, the adjusted sequence similarity with BV27 polyprotein increased markedly to 99.69% (FL-ex, Fig. 6A). Here, ‘FL-ex’ denotes the BV236 polyprotein sequence excluding the insertion and duplicated regions, relative to the full-length sequence. This highlights the significant impact of recombination-associated structural changes on genome-wide similarity estimates.

Distinct origin of the recent BVDV-1b strains (BV27 and BV236) from the 18H17 strain

To elucidate the evolutionary relationships among BVDV-1b strains, we conducted a haplotype network analysis based on the complete nucleotide sequences of the E2 and NS2-3 genes from 41 globally sourced BVDV-1b isolates, including three strains from Taiwan: BV27, BV236, and 18H17 (Fig. 7). The network revealed clear genetic structuring among the strains, with BV27 and BV236 (indicated by orange circles) forming a distinct cluster that is more closely related to BVDV-1b strains isolated in the United States (red circles), rather than to 18H17, which represents a previously circulating Taiwanese strain.

Fig. 7.

Haplotype network analysis of BVDV-1b identified in Taiwan. The genetic relationships and evolutionary pathways of three Taiwanese BVDV-1b strains—18H17 (isolated in 2018) and two recent strains, BV27 and BV236 (identified in 2023)—were inferred using haplotype network analysis. Sequences of the E2 gene (A) and the NS2-3 gene (B) of 41 BVDV-1b strains originating from nine countries were included in the analysis. Haplotype networks were constructed using Minimum spanning networks implemented in PopART v1.7. Circle colors indicate the geographic origin of each haplotype, while circle size represents the number of strains within each haplotype

Discussion

Epidemiological data on BVDV in Taiwan are scarce. Aside from a nationwide survey published recently using samples collected in 2014 [30] and a brief sequence report of one isolate (18H17) [39], no additional studies have addressed the prevalence, genotype distribution, or biotypes of circulating strains. Our study provides the first genomic-level characterization of Taiwanese BVDV-1b since that time, revealing genetic features suggestive of cytopathic evolution. Although limited by sample size and lack of virus isolation, these findings add important genomic and evolutionary insights into the diversity of BVDV in Taiwan.

The comparative genomic analysis of two Taiwanese BVDV-1b strains revealed several notable structural differences, particularly in BV236. Most significantly, BV236 harbored a duplicated region spanning NS3-NS4A and part of NS4B, as well as a host-derived insertion containing sequences from the ubiquitin C gene—both features that have been frequently associated with cytopathogenicity in BVDV-1 strains [29, 40]. Despite being isolated from different cattle species, BV27 and BV236 were linked to respiratory infections of differing severity: BV27 was isolated from a beef cattle without clinical signs, whereas BV236 was obtained from a fatal case in a dairy cow. Given its unique genomic features, BV236 is likely to represent a cytopathic (cp.) biotype, while BV27 appears to be a noncytopathic (ncp) strain.

Cp BVDV strains are known to arise from ncp progenitors through non-homologous RNA recombination events [5]. These events often involve duplication of viral genomic regions or insertions of host- or virus-derived sequences—commonly ubiquitin, heat shock proteins, or fragments of viral genes such as viral Npro protein—into the coding region for NS2-3 or adjacent loci [29, 41–43]. The genome structure of BV236 aligns with this mechanism, featuring both a partial duplication of the NS3–NS4A–NS4B region and an upstream insertion of ubiquitin C sequences preceding the second NS3 gene (Fig. 2).

Although BV27 shares only 80.00% amino acid sequence identity with BV236 across the full length of the polyprotein sequences (FL in Fig. 6A), when the unique ubiquitin C insertion and duplicated viral segments in BV236 are excluded, the similarity between the two strains increases to 99.69% (FL-Ex in Fig. 6A). This high degree of identity and the presence of aforementioned genomic features support the widely accepted model in which cp. strains emerge from closely related ncp progenitors via recombination. Such events are known to result in the independent expression of a cleaved or truncated NS3 protein, bypassing the normal NS2-3 processing [5, 29]. This recombination disrupts the regulated protease-mediated polyprotein processing and leads to constitutive NS3 activity, which enhances cytotoxicity to host cells and is responsible for the cytopathic effects observed in vitro [44].

BV236 was obtained from a fatal dairy cow case with gross and microscopic lesions consistent with mucosal disease. Importantly, in addition to the cp-type genomic configuration identified in BV236, we also detected viral sequences lacking the insertion/duplication events (Fig. 4). The coexistence of both viral forms within the same clinical sample is consistent with the concept that cp viruses can emerge from persistent infections of ncp viruses [45, 46]. Previous studies have demonstrated that genomic insertions, duplications, or incorporation of host-derived sequences in the NS2-3 region are the force to form variant virus in the same animal [5, 29]. Therefore, the detection of both viral forms in BV236 provides genomic evidence supporting the emergence of viral variant formation. This finding also supports the possibility that co-circulating viral forms may contribute to the development of mucosal disease. Nevertheless, the correlation between these genomic features and cytopathogenicity, as well as the mechanisms underlying biotype switching, require further investigation in cell culture models.

Phylogenetically, both BV27 and BV236 strains belong to subgenotype 1b and are closely related to strains identified in the US (Fig. 5). Surprisingly, despite sharing the same subgenotype, BV27 and BV236 exhibited significant divergence from another local BVDV-1b strain, 18H17, isolated in Taiwan in 2018 [39]. Phylogenetic analysis based on both the 5′ UTR and complete polyprotein coding regions placed BV27 and BV236 in a different lineage than 18H17 (Fig. 5). Haplotype network analysis of the E2 and NS2-3 gene regions further supported this finding. As illustrated in Fig. 6, the strains BV236 and BV27 are directly connected within the network, differing by a single mutational step. This close genetic relationship suggests a recent common ancestor or a shared evolutionary origin between the two strains. In contrast, 18H17 is positioned on a separate branch, clearly distant from both BV27 and BV236, supporting the notion that it emerged from an independent lineage within the BVDV-1b subgenotype. These findings indicate that while BV236 and BV27 may be linked by local transmission or a single introduction event, 18H17 likely represents a distinct introduction of BVDV-1b into Taiwan. The genetic proximity of BV27 and BV236 to strains isolated in US raises the possibility of international transmission through trade, importation of livestock, or biological products, and underscores the importance of global surveillance in managing transboundary animal diseases [47–49].

The identification of distinct genetic structures in a potential cp. strain of BVDV-1b circulating in Taiwan underscores the need to refine current surveillance strategies. Routine monitoring often relies on partial genomic regions such as the 5′ UTR or E2 gene for genotyping [12]. However, our findings highlight that key pathogenic features—such as viral recombination and host gene insertions—can only be detected through whole-genome sequencing. This demonstrates the limitations of partial sequencing approaches and emphasizes the necessity of incorporating full-genome sequencing into routine diagnostic and surveillance workflows to enable early detection of virulent or emerging BVDV variants. Furthermore, the close phylogenetic relationship observed between Taiwanese and U.S. strains (Fig. 7) suggests transboundary viral movement, reinforcing the importance of international collaboration and genomic data sharing to track BVDV evolution and dissemination on a global scale.

While this study provides valuable insights into the genomic characteristics and potential origins of Taiwanese BVDV-1b strains, the limited sample size constrains the extent to which our findings can be generalized to the broader viral population. Although genomic analyses suggest hallmarks indicative of cytopathic evolution, confirmation of biotype switching will require future studies involving virus isolation and functional studies, such as cell culture characterization and reverse genetics approaches to directly assess whether these genomic features contribute to cytopathogenicity. The transition from ncp to cp virus likely involves complex host-pathogen interactions, including immune pressure, superinfections, and cellular stress responses, which could not be explored in this genomic study alone [50–53]. Further research should incorporate functional assays to investigate the role of inserted sequences and duplicated regions in cytopathogenicity.

Conclusion

Our comparative genomic analysis of two Taiwanese BVDV-1b strains provides the first genomic evidence of a cytopathic strain in Taiwan. The detection of viral recombination and host sequence insertion events in BV236 highlights the dynamic evolutionary potential of BVDV-1 and underscores the risk of virulence shifts within persistently infected herds. These findings demonstrate the critical importance of genome-level surveillance for the early detection of pathogenic variants. Based on haplotype network analysis, BV27 and BV236 appear to have originated from an independent introduction event, rather than evolving from previously circulating local strains such as 18H17. The continued evolution and spread of BVDV-1b strains reinforce the need for sustained molecular monitoring and timely epidemiological investigations to enable early intervention and effective disease control.

Abbreviations

BVDV

Bovine viral diarrhea virus

ncp

Non-cytopathic

cp

Cytopathic

5′ UTR

5′ untranslated region

3′ UTR

3′ untranslated region

PI

Persistent infection

Hsp40

Heat shock protein 40

Authors’ contributions

Wei-Li Hsu: Conceptualization, Data curation, Validation, Supervision, Funding acquisition, Writing – original draft, Writing – review & editing. Hau-You Tzeng: Data curation, Formal analysis, Investigation, Writing – original draft. Yi-Ying Chiou: Validation, and Supervision. Fong-Yuan Lin: Data curation, Formal analysis. Jia-Jiun Ho, Ching-Yu Tseng, Yumiko Yamada, Ruei-Sheng Tsai and Yuan-Hao Cheng: Data curation. Min-Chuan Lai: Sample Resources, and Experiment design, Hue-Ying Chiou: Data curation, Sample Resources, and original draft, Writing.

Funding

This research was funded by the National of Science and Technology Council, Taiwan, ROC, grant number NSTC-112-2327-B-002-008, as well as partly supported by the framework of the Higher Education Sprout Project from the Ministry of Education (MOE-114-S-0023-A) in Taiwan.

Data availability

Sequence data that support the findings of this study have been deposited in GenBank, National Center for Biotechnology Information with accession number PV426615 and PV426616.

Declarations

Ethics approval and consent to participate

Sequences of BV236 were derived from a fatal dairy cow case submitted to the Animal Disease Diagnostic Center at the College of Veterinary Medicine, National Chung Hsing University, for diagnostic evaluation. The use of the samples for this study was approved, with informed consent obtained from the owner.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.