Detection of atypical porcine pestivirus genome in newborn piglets affected by congenital tremor and high preweaning mortality

Abstract

Recently, piglets from a high-health status farm began exhibiting congenital tremors, high preweaning mortality and incidence of splayed legs. Postmortem histological examination identified a small number of scattered white matter vacuoles in the cerebellum and underlying brainstem of affected piglets. Presence of potential viral sources associated with this neurologic condition was initially infirmed using quantitative PCR for atypical porcine pestivirus (APPV), porcine teschovirus, and porcine sapelovirus. Using metagenomic analysis, APPV was identified as the main microbial species in serum obtained from piglets affected by congenital tremor. These piglets had higher preweaning mortality rates (46.4% vs. 15.3%) and incidence of splayed legs (33.0% vs. 0.8 %) compared to unaffected piglets. Piglets affected by congenital tremor had higher viral titer (P < 0.15) and larger birth weights (P < 0.05) compared to normal litter mates. Whole-genome sequencing and genome assembly of the novel APPV strain (MK728876) was carried out using Oxford Nanopore and related bioinformatics pipelines. Phylogenic analysis demonstrated that this strain along with other completely sequenced APPV strains were grouped into 2 clades, both including strains-inducing congenital tremor. Strains appear to cluster based on region but there were still significant differences within regions. Future research needs to address potential underdiagnosis due to genetic diversity but also to understand mode of transmission, variation in virulence, and the role of host genetics in APPV susceptibility.

INTRODUCTION

Recently, there has been an increase in the number of cases of growing pigs exhibiting a range of neurological conditions such as muscle tremors that can progress to paralysis and death (Swine Health Information Center; https://www.swinehealth.org). A number of viral sources leading to these conditions have been discovered, including atypical porcine pestivirus (APPV), porcine teschovirus, and porcine sapelovirus. Using metagenomic sequencing of swine clinical samples, Hause et al. (2015) were the first to identify and characterize APPV as a new species of Pestivirus. At that time, no symptoms were associated with the presence of this newly characterized virus but later studies were able to associate APPV with the presence of congenital tremor in newborn piglets and with varying degrees of hypomyelination in both brain and spinal cord (Arruda et al., 2016; Schwarz et al., 2017; Mósena et al., 2018). APPV has been detected around the world and characterized by important genetic variations between strains and geographical regions (Hause et al., 2015; Postel et al., 2016; Beer et al., 2017; Zhang et al., 2017; Mósena et al., 2018). This substantial genetic variation could lead to difficulties in detecting the presence of certain strains using traditional diagnostic tests such as quantitative PCR (qPCR). In the recent years, metagenomic analysis was used successfully in identifying novel pathogens and capturing genetic diversity among microbial genomes without any prior requirement of genomic information. In this study, we employed metagenomic sequencing to discover and characterize a novel strain of APPV associated with piglets exhibiting congenital tremor and high preweaning mortality.

MATERIALS AND METHODS

All procedures were approved by the Institutional Animal Care and Use Committee of the University of Nebraska-Lincoln.

Sample and Data Collection

A research farm characterized by high-health status, a comprehensive vaccination program, and naïve for other important pathogens started to observe first cases of congenital tremor in August of 2017. Blood samples were collected from newborn piglets from normal litters and from litters affected by congenital tremor and housed in the same facility. The samples were centrifuged at 2,350 × g at 4 °C for 30 min and the serum was stored at −80 °C. The litters were generated by maternal crossbred dams inseminated with commercial Duroc semen. The source farm has vaccination programs for porcine circovirus 2, porcine parvovirus, Erysipelothrix rhusiopathiae, Clostridium perfringens type C, leptospirosis, and colibacillosis. The farm is naïve for other important viral pathogens such as porcine reproductive and respiratory syndrome virus (PRRSV) or porcine epidemic diarrhea virus. Following initial cases of congenital tremor, a controlled exposure data set was generated (March 2018) by subjecting maternal crossbred gilts (n = 91) to an inoculate obtained from fetal fluids and membranes collected from sows that produced litters affected by congenital tremor. Litter size traits, including total number of piglets born (TNB), number of mummified and stillborn, and presence of splayed legs were recorded at farrowing. The weight of the piglets was measured at birth and at weaning.

Pathological Examination

Cross sections of brain at the level of the cerebellum were fixed by immersion in 10% neutral buffered formalin and then routinely processed overnight on a Leica Peloris II tissue processor. The sections of processed brain were embedded in paraffin blocks and cut to 4 µm thick using a Leica RM2255 microtome, stained with hematoxylin and eosin on a Leica ST5020 H&E stainer, and cover-slipped with a Leica CV5030 coverslip. A veterinary anatomic pathologist who is board certified by the American College of Veterinary Pathologists performed all histological evaluations.

Detection of the Potential Viral Source of Congenital Tremor Using Oxford Nanopore Sequencing

Initial analysis of the potential source of congenital tremor was based on nucleic acids isolated from serum of affected piglets using QIAamp Circulating Nucleic Acid kit (Qiagen). To generate ~1 µg of complementary DNA (cDNA) required for library preparation and Oxford Nanopore sequencing, a preamplification approach was implemented using nanomer-extended primers and experimental conditions as described by Greninger et al. (2015). This approach was tested successfully for sequencing various clinical samples for detection of bacteria, fungi, parasites, DNA, and RNA viruses (reviewed in Greninger et al., 2015). Briefly, the first strand of cDNA was obtained from pooled RNA of 4 affected samples using Superscript III reverse transcriptase (Life Technologies) and random nanomers-extended primers (5′-GTTTCCCACTGGAGGATA-N₉ -3′). The second strand of cDNA was synthesized using Sequenase DNA polymerase (Life Technologies). The cDNA obtained was subjected to preamplification using a primer based on the extended sequence of the random nanomers (5′-GTTTCCCACTGGAGGATA-3′). Metagenomic cDNA library was generated by Oxford Nanopore Ligation Sequencing kit 1D (SQK-LSK108, Oxford Nanopore Technologies), including the following steps: 1) end-repair with NEBNext Ultra II End Repair/dA-Tailing Module (New England BioLabs), 2) AMPure XP beads (Beckman Coulter) based purification, 3) adapter ligation with NEB Blunt/TA Master Mix, 4) AMPure XP beads based purification, and 5) elution in 15-µl buffer (Oxford Nanopore Technologies). The obtained libraries were sequenced on an Oxford Nanopore MinION using R9.4 flow cells loaded with 75 µl of sequencing mix (12-µl cDNA library, 35-µl buffer with fuel mix, 25.5-µl library loading beads, and 2.5-µl ddH₂0) and using Nanopore MinKNOW 1.15.6 version as described in the manufacturer’s protocol.

Genome Sequencing and Assembly of the Novel APPV Strain

Sequencing of the novel APPV strain was based on full-genome preamplification using Primal Scheme (http://primal.zibraproject.org), a web-based tool for primer design (Quick et al., 2017). The primers were designed based on the initial Nanopore sequencing of the MK728876 strain and its genetically closest complete genome sequence of a porcine atypical pestivirus 1 (MF167291.1). The amplicon length was set up to 400 bp with an overlap of 50 bp leading to 39 primer-pair sets. Genome preamplification of the MK728876 strain was based on 22 pooled cDNA samples characterized by high viral titer. First-strand cDNA synthesis of the viral RNA was carried out using Superscript III reverse transcriptase followed by PCR (GoTaq Flexi DNA polymerase; Promega) and agarose gel electrophoresis. Equal amounts of amplicons from each primer set were pooled (390 µl) followed by purification using AMPure XP beads (1× ratio of cDNA to beads). DNA concentration was evaluated using Qubit 1.0 fluorometer (Life Technologies). Sequencing of the amplicons was performed using Ligation Sequencing kit 1D (SQK-LSK108) as describe above. FAST5 reads that contained raw Nanopore signals were base-called using Albacore 2.3.3 (https://nanoporetech.com/). Adapters were removed using Porechop Nanopore application v0.2.3.

Metagenomic classification of the initial Nanopore sequencing data was performed using Centrifuge 1.0.3. package (https://ccb.jhu.edu/software/centrifuge/manual.shtml centrifuge-class option) (Kim et al., 2016). The “centrifuge-build” indexer was used to generate an index on all complete, bacteria, viral, archaea, and host (swine) genomes while “centrifuge-class” option was used to classify the sequencing reads to specific genomes. Pavian application (https://ccb.jhu.edu/software/pavian/) was used in the analysis of the metagenomic classification results (Breitwieser and Salzberg, 2016).

Genome assembly of the new APPV strain was based on several tools initially developed for sequencing the Zika virus (https://github.com/zibraproject/zika-pipeline). Briefly, FAST5 reads that contained raw Nanopore signals were base-called using Albacore 2.3.3 (https://nanoporetech.com/). Adapters were removed using Porechop Nanopore application v0.2.3. The reads were mapped to the reference genome (MF167291.1) using minimap2 (https://github.com/lh3/minimap2) and converted to BAM format by “samtools view.” The script align_trim.py tool was used for trimming the preamplification primers and for normalization of the coverage. The sequencing variants were called using “nanopolish variants” (https://github.com/jts/nanopolish) application, while consensus sequence was generated by script margin_cons.py.

qPCR Testing of APPV

Estimates of the number of APPV copies or viremia in serum was performed using cDNA based on viral RNA isolated by MagMAX Viral RNA Isolation kit (Applied Biosystems) and reverse transcribed by Superscript III reverse transcriptase (Invitrogen). Viral RNA was extracted from blood collected at an average age of 6.6 ± 0.3 d from all available piglets from 25 litters (out of 91 total) affected by congenital tremor (n = 171). The APPV presence was determined based on a qPCR specific to the novel APPV MK728876 strain (Forward: 5’-CGA CTG AAG GCA GGA AAT TTG-3’; Reverse: 5’-GCT ACA AAC ACC AGG CAA TTC-3’; Probe: 5’-TC TCT ATA G/Zen/T GGC AAC CCC AAC CTC T-3’) using TaqMan Master Mix (Life technologies) and CFX384 Real Time PCR (BioRad). The qPCR assay was initially tested on viral cDNA samples collected from piglets exhibiting congenital tremor (n = 82), their unaffected litter mates (n = 2), and their dams (n = 17) but also from piglets from unaffected litters (n = 25). The number of APPV copies/mL of serum (log10) in experimental samples were obtained based on a standard dilution curve of a positive control DNA of known titer.

Statistical Analyses

The relationship between the presence or absence of congenital tremor and occurrence of stillborn, mummies and splayed legs across litters was evaluated by Pearson chi-square contingency test, coding the presence or absence of the targeted phenotypes in each litter. A linear mixed model was used to investigate the association between presence/absence of congenital tremor and birthweight, weaning weight, and average daily gain (ADG) during lactation. The model included litter as random effect, litter size (TNB) as covariate, sex, and occurrence of congenital tremor as fixed effects. A linear mixed model was used to investigate the relationship between APPV viral titer and presence/absence of congenital tremor; the model included birthweight as a covariate, presence/absence of tremor as fixed, and litter as random effects.

Phylogenetic Analyses

To assess the evolutionary relationship of the novel strain with the other APPV strains, the DNA sequence and the viral polyprotein sequence were compared with that of 30 other complete APPV genome and protein sequences, respectively. Genomic data was retrieved from GenBank. DNA and protein alignments were performed using MUSCLE (Edgar, 2004) implemented in MEGA X (Kumar et al., 2018). The phylogenetic tree based on DNA alignment was inferred by using the maximum likelihood method and Tamura–Nei model (Tamura and Nei, 1993) while that based on the protein alignments by using the maximum likelihood method and JTT matrix-based model (Jones et al., 1992).

RESULTS AND DISCUSSIONS

Litters Affected by Congenital Tremor had a Higher Incidence of Preweaning Mortality

A swine research farm characterized by high-health status and a comprehensive vaccination program started to observe first cases of congenital tremor in August of 2017. The affected litters were generated by maternal crossbred dams inseminated with commercial Duroc semen. Potential causes of congenital tremor were investigated histologically and by commercial and literature-based qPCR (Arruda et al., 2016) using multiple tissues. Postmortem histologic examination of affected piglets showed small numbers of scattered white matter vacuoles in the cross sections of the cerebellum and underlying brainstem collected around the level of the pons to myelencephalon. The vacuoles did not have a single distinct region affected but appeared more commonly along the white matter of the cerebellar folia as well as near the junction of the cerebellum and brainstem. The vacuoles varied in size with most ranging from 10 to 30 µm in diameter. Similar findings were reported by other studies in piglets affected by congenital tremor (Schwarz et al., 2017; Mósena et al., 2018). For example, Schwarz et al., (2017) found vacuoles in the white matter of the cerebellum in the affected piglets, while rare or absent vacuoles were found in the unaffected litter mates and no vacuoles were found in the healthy control piglets. Commercial diagnostic qPCR testing for RNA viruses, including APPV, porcine teschovirus, and porcine sapelovirus, known to induce neurological symptoms in pig were all negative in the affected piglets.

A batch of 91 gilts of similar genetics and born in the same farm were subjected to an antigen oral exposure approximately 54 d prior to breeding using fetal fluid from the first group of litters displaying congenital tremor. In the exposed sows, 45.0% of the litters and 30.8% of all piglets were affected by congenital tremor (Table 1). The frequency of piglets with congenital tremor in the affected litters varied from 5.9% to 100%. The time of farrowing (September 4, 2018 to September 27, 2018) did not appear to influence the occurrence of congenital tremor across litters. Overall, the preweaning mortality in piglets affected by congenital tremor was 46.4% compared to 15.3% in unaffected piglets (in both affected and unaffected litters). The preweaning mortality of the unaffected litter mates from affected litters was 24.2% compared to 12.7% of the piglets from unaffected litters. Also, litters affected by congenital tremor had a higher incidence of the average preweaning mortality (17.2%) compared to unaffected litters (6.5%). About half (53.7%) of the mortality in the affected piglets occurred in the first day of life, while three-fourths of the mortality occurred in the first 5 d of life (76%). An impact of congenital tremor on preweaning mortality was previously reported by Schwarz et al. (2017) in 2 farms from Austria in 2015 (mortality up to 25%) and 2016 (up to 30%).

Table 1.

Summary of phenotypic information of a batch of litters affected by congenital tremor (CT) generated by sows (n = 91) exposed to an APPV-based antigen ~54 d prior to breeding.

	Affected litters
Trait/CT phenotype	CT	No CT	Unaffected litters
Number of litters	41		50
Number of piglets born alive	373	186	652
Incidence of stillborn, %	8.92		8.0
Incidence of mummies, %	3.60		3.79
Preweaning mortality, %	46.4	15.3	12.7
Incidence of splayed legs, %	33.0	2.15	0.46
Birth weight (means +/- SE, kg)	1.22 ± 0.02	1.09 ± 0.03	1.3 ± 0.02
Log10 copies of APPV/mL of serum (n = 171)	6.55 ± 0.21	6.25 ± 0.21	—

Litters Affected by Congenital Tremor had a Higher Incidence of Splayed Legs and Smaller Birth Weights

A significant relationship was observed between the frequency of congenital tremor and splayed legs (Pearson χ ² = 40.2, P < 0.0001). Incidence of splayed legs was observed in the majority of the litters affected by congenital tremor (73.2%), while, in the unaffected litters, the incidence was limited (6.0%). The presence of splayed legs in piglets affected by tremor (33.0 %) was larger compared to unaffected piglets (0.8 %) (Table 1). The average frequency of splayed legs in the litters affected by congenital tremor was 22.4%, while, in the unaffected litters, the average frequency was marginal (0.5%). This relationship between congenital tremor and incidence of splayed leg was reported in other studies (Arruda et al., 2016; de Groof et al., 2016). The presence of piglets with congenital tremor across litters was not associated with the presence of stillborn (P > 0.98) or mummies (P > 0.26).

A significant relationship between the occurrence of congenital tremor and birth weight was observed across and within litters (P < 0.0001). Piglets from litters not affected by tremor had greater birth weight (1.30 ± 0.02 kg) compared to piglets exhibiting tremor (1.22 ± 0.02 kg) and their unaffected litter mates (1.09 ± 0.03 kg) from affected litters (P < 0.05). Interestingly, piglets exhibiting congenital tremor had larger birth weights compared to their normal litter mates (P < 0.0001). A similar finding was reported by Harding et al. (2017) using the pregnant gilt model to study host response in PRRSV infection. They hypothesized that large fetuses appear to be more susceptible to PRRSV due to larger placentae or higher nutrient requirements. The presence of congenital tremor across and within litters had no effect on weaning weight or on ADG during lactation (P > 0.12).

APPV, the Predominant Microbial Species in the Serum of Pigs Affected by Congenital Tremor

Oxford Nanopore sequencing was employed for identification of the potential viral source associated with congenital tremor by metagenomic sequencing of cDNA libraries generated from serum samples of piglets affected by congenital tremor. The initial Nanopore sequencing data included 66,898 sequencing reads with an average read length of 568 bases. The sequencing data was classified to 228 reference genomes using Centrifuge 1.0.3 package. APPV (NC_030653.1), with 205 mappable reads obtained, was the most abundant microbial species observed in the samples of the affected pigs (Figure 1; Supplementary Table S1). In addition, other sequencing data classified as part of the virome included 34 reads uniquely mapped to porcine circovirus 3 (PCV3, NC_031753.1). The presence of PCV3 was acknowledged in other swine metagenomic studies (Palinski et al., 2017). The 4 reads originally classified as Shamonda orthobunyavirus (NC_018464.1) were also mapped to host Sus scrofa genome. These sequences are most likely part of the host genome since their average alignment coverage was 74% to S. scrofa genome compared to only 7% to S. orthobunyavirus.

Figure 1.

Alignment of Oxford Nanopore sequencing reads to the reference genomes based on tissues collected from piglets affected by congenital tremor indicated that the most abundant microbial species was APPV.

While metagenomic analysis was used in the discovery efforts of the potential sources of congenital tremor, a qPCR assay, designed based on the new sequencing data, was used to profile the presence of APPV across affected and unaffected piglets. The newly designed qPCR assay provided the required specificity for the diagnostic of the new APPV strain. Specifically, all the piglets exhibiting congenital tremor as well as their litter mates were positive for APPV, while the litters not affected were negative. Profiled at the same time with piglets, all the dams were qPCR negative. The majority of the pigs (69.6%) were still positive for APPV at approximately 30 d of age. The clear signs of tremor disappeared in the vast majority of pigs at 50 d of age (91.6%) and a suggestive decrease in viremia was observed at this point (P < 0.07). Initial Nanopore sequencing data also suggested that the lack of specificity of the reported APPV qPCR assay (Arruda et al., 2016) was the most likely reason for the early negative APPV diagnostic analysis. Specifically, the forward primer of the reported qPCR assay includes 3 nucleotide mismatches compared to the sequences obtained from the newly identified strain.

In a subset of 171 piglets representing 25 litters affected by congenital tremor, a suggestive association was observed between APPV viral titer (P < 0.15) and congenital tremor; piglets exhibiting tremor displayed a larger viral titer in the blood (6.55 ± 0.21 log₁₀ copies of APPV/mL of serum) compared to visually normal litter mates (6.25 ± 0.21), suggesting a role of APPV viral load in the expression of congenital tremor. We hypothesize that the effects are actually larger since most of the preweaning mortality associated with congenital tremor occurred in the first 5 d of life, while the viral titer was measured in all available piglets at an average of 6.6 ± 0.3 d of age.

Genome Sequencing and Assembly of the Novel APPV Strain

Generating whole-genome sequencing data and complete genome assembly directly from clinical samples is challenging due to relatively low viral titer. In this study, we employed a whole-genome preamplification approach used successfully for generating complete genome sequence of Zika and other viral genomes directly from clinical samples (Quick et al., 2017). The initial Nanopore sequencing reads of the novel strain and the complete genome of its closest APPV strain (MF167291.1) led to the design of 39 pairs of primers and overlapping amplicons expected to cover the entire genome. The majority of primer pairs led to successful amplifications (92.3%), while combinations of adjacent primers were used to amplify the remaining regions. After a Nanopore sequencing run, base-calling using Albacore 2.3.3, and adapter removal using Porechop, there were 445,199 reads obtained. The reads were processed using bioinformatic scripts associated with Zika-pipeline (https://github.com/zibraproject/zika-pipeline). Using minimap2, 69.7% of the reads were aligned to the APPV strain (Ger-NRW_L277, MF167291.1). The obtained alignments were subjected to a script (align_trim.py) that removes primer sequences based on their coordinates and normalize the coverage. Following normalization and primer trimming, a subset of 36,166 sequences and the MF167291.1 reference sequence were used for variant detection using “nanopolish variants” application. There were 618 Single Nucleotide Polymorphisms (SNPs) identified with an average of 52.6 SNPs/kb. A script (margin_cons.py) was used to filter out low-coverage and low-quality variants and generate a consensus sequence of the new strain. Compared to MF167291.1 coordinates, the consensus sequence lacked the first 60 nucleotides and had a gap of 217 nucleotides between 4,134 and 4,351 bp. A PCR amplicon covering the gap was sequenced and the new genome was assembled (Acc. No. MK728876).

Phylogenetic Analysis of the APPV Strains

Complete genome sequences available in NCBI Nucleotide/Protein database consisting of 30 APPV strains isolated across the globe were subjected to phylogenetic analysis using the sequences of viral genome and of the viral polyprotein. Both analyses revealed 2 noticeably divergent clades: the large one contained 24 strains, from different geographical locations, while the smaller clade included 7 strains so far reported only in China (Figure 2; Supplementary Figure S1). While some of the strains were isolated from asymptomatic piglets, strains from both clades were associated with occurrence of congenital tremor (Postel et al., 2017; Zhang et al., 2017; Shen et al., 2018). The large clade showed limited clustering based on geographical location; in most cases, important diversity within regions was identified. The DNA sequence of the novel strain (MK728876) showed highest similarity with a cluster of 3 APPV strains isolated from Germany (MF167290.1, MF167291.1, and LT594521.1), forming a strongly supported clade. All these 3 strains were recovered from piglets exhibiting congenital tremor (Postel et al., 2017). Similarities in this clade, at the level of DNA varied from 97% to 98%. Both phylogenies found few affinities between the novel MK728876 strain reported here and the other U.S. strains (Figure 2; Supplementary Figure S1). Clearly, NC_038964 and KR011347 strains are distantly related to MK728876 and KU194229, and the relatedness between the later strains is higher. Consequently, similarities ranging from 94% to 96%, were observed at the DNA level when the new strain was compared to the other 3 strains from the United States. When the predicted polyprotein sequence consisting of 12 peptides was analyzed and compared across predicted viral peptides, the proportion of similarity between the new and a less similar U.S. strain (AKS24977) varied across nonstructural viral proteins ranging from 78.5% (NS5b) to 98.8% (NS4b). Using a short fragment of the nonstructural NS3 protein, Postel et al. (2017) showed similar clustering, emphasizing the widespread and the diversity of APPV.

Figure 2.

Phylogenetic analysis of the DNA sequences of APPV strains across the globe. The phylogenetic tree was generated by using the maximum likelihood method and Tamura–Nei model. The tree with the highest log likelihood is shown. A discrete Gamma distribution was used to model evolutionary rate differences among sites (5 categories [+G, parameter = 0.3303]). Analysis of the 31 sequences involved a total of 10,908 positions.

The genome alignment of the various APPV strains indicates a high overall nucleotide diversity across the strains. The divergence is more evident in the 5’ and 3’ noncoding regions. However, substantial nucleotide polymorphisms are scattered across the viral polyprotein sequence. As a result, qPCR-based identification of all APPV strains could be challenging but employment of degenerate oligonucleotides in qPCR assays or cost-effective metagenomic analyses could circumvent the divergence found among APPV strains.

In this study, we combined 2 approaches in order to detect and sequence the genome of a novel APPV strain. Preamplification of viral genome using random nanomers was employed for the initial sequencing data, while tilling amplicons were obtained and sequenced using partial sequencing data of the novel strain and the most-related and available APPV genome. Compared to the other methods, this approach is characterized by efficient enrichment of the low-abundance genetic material isolated from clinical samples and rapid turnaround in detecting a novel APPV strain. Low viral abundance is associated with a decrease in the likelihood of long RNA fragments being present in samples (Quick et al., 2017), and short-amplicon-based amplification, as employed in this report, alleviated this issue. As demonstrated in this study, qPCR-based detection has limited sensitivity when the targeted strain is highly divergent but, for highly similar family of viruses, the qPCR approach is the cost-effective strategy. While this approach is ideal for infections caused by single-clone viruses, it is less suitable for highly diverse or mixed viruses and strains. As shown here, data generated by amplicon-based sequencing is affected by incomplete and dropouts in coverage and lack of information at both ends not covered by primers. Any approach involving preamplification is prone to contamination based on reagents/tissues used in previous laboratory analyses.

CONCLUSION

Presence of vacuoles in the white matter of the cerebellum, high preweaning mortality and incidence of splayed legs were observed in newborn piglets affected by congenital tremor. Piglets affected by congenital tremor also had lower preweaning survival rates but larger birth weights compared to their unaffected litter mates. Metagenomic analysis showed that APPV was the most abundant microbial species in the blood collected from piglets affected by congenital tremor. A qPCR assay, specific for the new strain, validated the discovery efforts based on metagenomic data, detecting the presence of APPV in all affected piglets and their unaffected litter mates but absent from litters unaffected by congenital tremor. Piglets affected by congenital tremor exhibited higher APPV viral titer compared to normal litter mates. Phylogenetic analysis of APPV strains across the globe revealed 2 highly divergent clades. Evidence of genetic similarity was found based on geographical location; also, important diversity within regions was observed. Variation of the APPV strains in the United States alone could lead to underdiagnosis of pestivirus-associated congenital tremors and emphasize the need for next-generation molecular diagnostic approaches based on robust, cost-effective, and real-time metagenomic analysis. Future studies of this family of swine Pestiviruses will need to provide critical information in areas such as mode of transmission, disease progression, variation in strain virulence, and the role of host genetics in disease susceptibility.

SUPPLEMENTARY DATA

Supplementary data are available at Journal of Animal Science online.

Figure S1. Phylogenetic analysis of the predicted polyprotein sequences of APPV strains across the globe. The phylogenetic tree was generated by using the Maximum Likelihood method and JTT matrix-based model. The tree with the highest log likelihood is shown. A discrete Gamma distribution was used to model evolutionary rate differences among sites (5 categories (+G, parameter = 0.2369)). Analysis of the 31 sequences involved a total of 3,642 positions.

Table S1. Classification results of the sequen- cing reads obtained from piglets affected by congenital tremor aligned to the reference genomes using Centrifuge 1.0.3. package.