High prevalence, genetic diversity and a potentially novel genotype of Sapelovirus A (Picornaviridae) in enteric and respiratory samples in Hungarian swine farms

Abstract

All of the known porcine sapeloviruses (PSVs) currently belong to a single genotype in the genus Sapelovirus (family Picornaviridae). Here, the complete genome of a second, possibly recombinant, genotype of PSV strain SZ1M-F/PSV/HUN2013 (MN807752) from a faecal sample of a paraplegic pig in Hungary was characterized using viral metagenomics and RT-PCR. This sapelovirus strain showed only 64 % nucleotide identity in the VP1 region to its closest PSV-1 relative. Complete VP1 sequence-based epidemiological investigations of PSVs circulating in Hungary showed the presence of diverse strains found in high prevalence in enteric and respiratory samples collected from both asymptomatic and paraplegic pigs from 12 swine farms. Virus isolation attempts using PK-15 cell cultures were successful in 3/8 cases for the classic but not the novel PSV genotype. Sequence comparisons of faeces and isolate strains derived VP1 showed that cultured PSV strains not always represent the dominant PSVs found in vivo.

Introduction

The genus Sapelovirus of family Picornaviridae currently consists of two species, Sapelovirus A and B. The species Sapelovirus A contains only a single known sero/genotype, porcine sapelovirus (PSV)−1, which could infect only swine hosts [1]. The first strains of PSVs (formerly called PEV-8) as members of the historic group of porcine ‘enteroviruses’ were isolated more than 50 years ago from faecal samples of swine using cell cultures with various swine-derived cell lines [2–4]. Since then porcine kidney cell lines such as LLC-PK or PK-15 are widely used for isolation and typing of PSVs [5–7].

Porcine sapeloviruses have been associated with a wide variety of diseases including diarrhea, pneumonia and reproductive disorders like SMEDI (stillbirth, mummification, embryonic death and infertility) syndrome [4, 8, 9] of pigs, although due to the relatively high prevalence of PSVs detected in asymptomatic animals the exact role of natural PSV infections in such diseases are not yet definitely established [7, 10]. Nevertheless certain neurotropic PSV strains are increasingly associated with various CNS-related symptoms of swine including paralysis and paraplegia of various severity [11, 12]. PSVs are generally identified from faecal specimens although studies of their pathogenicity indicate that certain strains can result in viraemia and disseminated infections affecting extra-intestinal organs including the respiratory system [5, 6, 9]. Furthermore the presence of neurotropic PSVs infecting various parts of the brain and the spinal cord had been also recently reported [11, 12].

The single-stranded RNA genome of sapeloviruses contains a single viral polyprotein-encoding ORF flanked by a 5′ and a 3′ untranslated regions (UTRs) and a poly(A)-tail [13, 14]. Sapleoviruses have a type-IV internal ribosomal entry site (IRES) in the 5′ UTR and a complex secondary RNA structure with three stem-loops and a pseudoknot in the 3′UTR [15, 16]. The cis-acting replication elements (cre), which generally consist of a single stem-loop with conserved AAACA motif in the loop are usually found in the 2C genome regions of PSVs [16, 17]. The viral polyprotein of sapeloviruses is build up from 12 mature peptides located in consecutive order as l-P1(VP4-VP2-VP3-VP1)-P2(2A-2B-2C)-P3(3A-3B-3C-3D) (L=leader peptide) [13, 14]. The N-terminal cleavage sites of the P1 capsid proteins of PSVs are experimentally determined [18]. The VP1 capsid protein, which contains major antigenic epitopes, is generally used for sero/genotyping of sapelovirus strains [10, 18–20]. Although PSVs are known to be a genetically highly diverse group of viruses, with recombination potential, only a single genotype of PSV has been identified based largely on VP1 sequence analyses [10, 16, 20]. Beside the recently published large-scale epidemiological studies on PSV conducted mostly in Far East countries such as China and Japan [7, 10, 20] little information is available about the prevalence and genetic diversity of PSVs circulating in Europe particularly in Hungary. Mostly single PSV strains or partial PSV sequences have been described from European countries including Spain, France and Italy [21–23].

Here, this study focused on (i) the complete genome characterization of a novel, possible recombinant genotype of porcine PSV identified from a faecal sample of a paraplegic pig using viral metagenomics and RT-PCR methods; (ii) epidemiological investigation and VP1-based typing of PSVs in enteric and respiratory samples collected from asymptomatic and paraplegic pigs in 12 different herds in Hungary and (iii) isolation attempts of randomly selected PSV strains in PK-15 cell cultures including the possible novel genotype-strain and comparisons of VP1 of inoculum and isolated strains.

Methods

Background of samples and sources

A total of 111 faeces, sera, rectal, or nasal swab samples were analysed (Tables 1 and S2, available in the online version of this article). Samples include a total of 33 rectal−nasal swab pairs from 21- to 35-day-old asymptomatic pigs from five swine farms (farms A, B, C, D, GD) and ten faeces, and nasal−serum pairs from 3-month-old apparently healthy swine (farm EF) all collected in 2016 (Fig. 1a, Tables 1 and S2). Faecal samples from asymptomatic, first-parity sows (TKo-1–3; SZ1K-4K), which gave birth to paraplegic piglets were also collected from two farms (farms TM and SZ) (Tables 1 and S2). Beside the above mentioned samples from asymptomatic pigs, a total of 21 faecal samples from paraplegic animals of different ages (ranging between 21 days and 12 months) from seven different swine farms (farms GD, TM, NSZ, KEV, ZS, BUV and SZ) collected in 2013 were also analysed (Fig. 1a, Tables 1 and S2). In addition to faecal samples additional nasal swabs were available from five paraplegic pigs from farm GD (Tables 1 and S2). No other types of samples including any types of tissue or organ specimens were collected either from asymptomatic or from paraplegic pigs. Faecal samples, nasal and rectal swabs were collected using individual containers and separate polyester-tipped swabs. Swabs were then rinsed in 500 µl sterile 0.1M PBS in a sterile tube. All the collected samples were stored at −80 °C until use.

Table 1.

Origin and characteristics of biological samples used in this study. List of swine farms, locations (farm ID column), animal IDs and numbers, collection dates, age and health status (AS: asymptomatic, PP: paraplegic). The percentages of PSV positivity based on the results of PSV 3D^RdRp RT-PCR screening reactions was indicated in the sample type (nasal swab, rectal swab/faeces and serum) columns. For an extended version see Table S2

Farm ID	Animal IDs	No. of animals	Collection date	Age	Health status	Nasal swab	Rectal swab/faeces	Serum
Farm GD (Orosháza)	GD01-05	5	09/15/2016	≈21–25 days	PP	80 %	100 %	n/a
	GD06-09	4	09/15/2016	≈21–25 days	AS	75 %	100 %	n/a
Farm A (Városföld)	A1-5	5	08/12/2016	≈35 days	AS	100 %	80 %	n/a
Farm B (Katymár)	B1-5	5	08/18/2016	30–35 days	AS	0 %	80 %	n/a
Farm C (Bácsalmás)	C1-5	5	08/18/2016	≈35 days	AS	60 %	100 %	n/a
Farm D (Nyíribrony)	D1-5	5	08/10/2016	30–35 days	AS	20 %	0 %	n/a
Farm TM (Tázlár)	TKo1-3	3	01/2013	≈ 11–12 mo.	AS	n/a	0 %	n/a
	TM-1121, 1130, 1205	3	01/2013	7–21 days	PP	n/a	33 %	n/a
Farm NSZ (Nagyszokoly)	NSzK1	1	01/23/2013	≈ 11–12 mo.	PP	n/a	0 %	n/a
Farm KEV (Kevermes)	KEV1-3	3	09/05/2013	≈ 4.5 mo.	PP	n/a	0 %	n/a
Farm ZS (Zsana)	ZS1-3	3	04/14/2013	≈70 days	PP	n/a	100 %	n/a
Farm BUV (Balmazújváros)	BUV1-2	2	04/24/2013	21–35 days	PP	n/a	100 %	n/a
Farm SZ (Székelyszabar)	SZ1K-4K	4	01/2013	≈ 11–12 mo.	AS	n/a	25 %	n/a
	SZ1M-4M	4	01/2013	7–21 days	PP	n/a	50 %	n/a
Farm EF (Egyházasfalu)	EF1-10	10	08/17/2016	≈ 3 mo.	AS	0 %	90 %	10 %

n/a: no sample available or not applied

Fig. 1.

(a) Localizations, farm IDs and numbers of collected samples (N=111) of Hungarian swine farms used for investigation of PSVs in this study. Empty markers: asymptomatic swine, black markers: swine herds with paraplegic cases. (b) The virus family-based classification of individual metagenomic sequence reads identified from a single faecal specimen of a 3-week-old paraplegic pig (SZ1M). The classification of reads is based on the best blastx-scores (E-value ≤10⁻⁵). Numbers in brackets represent the numbers of reads belong to the certain taxon.

The severity, duration and variety of clinical signs of CNS-related symptoms with unknown etiology (e.g. movement disorders, pitching, hind limb weakness/paraplegia, later paralysis of both legs, paresis and serious spastic paralysis) showed great variations between the affected animals of different farms with the consistent presence of hind limb weakness/paraplegia. The diseases with CNS-related symptoms were mostly mild, self-limited and always present as local endemics usually affecting approximately 1–3 % of animals (Tables 1 and S2). Additional signs including mechanical damage, gastroenteric or respiratory symptoms could not be observed among the affected animals. Previous laboratory tests were not conducted in these cases. Serum samples were originally collected for regular testing of PCV-2 vaccination efficiency.

Next-generation sequencing and bioinformatics pipeline

For diagnostic purposes a single faecal specimen of a 3-week-old pig (animal ID: SZ1M) with signs of mild paraplegia with unknown aetiology as a part of an ongoing local epidemic in January, 2013 in Székelyszabar, Hungary (Fig. 1a, Tables 1 and S2) was subjected to viral metagenomics and next-generation sequencing (NGS) as previously described [24]. Briefly, 200 µl of the 35–40 v/v% (diluted in sterile 0.1M PBS) faecal suspension was centrifuged (15 000 g for 10 min) and filtered through a 0.45 µm filter (Millipore). The filtered sample was then treated with a mixture of DNases (Turbo DNase from Ambion, Baseline-ZERO from Epicentre, and Benzonase from Novagen) and RNase (Fermentas) to digest unprotected nucleic acids. Viral cDNA library was constructed by ScriptSeqTM v2 RNA-Seq Library Preparation Kit (Epicentre) and sequenced using the MiSeq Illumina platform. The generated sequence reads were assembled to contigs and sequences (single reads and contigs) were compared to the protein database of GenBank using blastx. The expected value (E) cut-off was set up to 10⁻⁵ for the categorization of the reads. Only sequences with the lowest E-value to known viruses were used for further analyses.

RT-PCR-based genome acquisition, screening and typing reactions

Total RNA was extracted from 150 µl of faecal suspensions (35–40 v/v% diluted in sterile PBS) as well as 150 µl re-suspended rectal, and nasal swabs, undiluted serum and cell-culture supernatant samples using TRI Reagent (MRC, USA) according to the manufacturer’s instructions. The amount of extracted total RNAs were measured using NanoDrop 2000 spectrophotometer (Thermo Fisher, Waltham, MA, USA). For complete genome acquisition of the study PSV strain SZ1M-F/PSV/HUN/2013 (MN807752), and complete coding sequence (CDS) determination of an additional three picornaviruses found in the sample of SZ1M, picornavirus-related reads and a contigs (Table 2, Fig. 1b) were used for virus-specific oligonucleotide primer design. Multiple RT-PCR reactions and 5′/3′ RACE (rapid amplification of cDNA ends) methods [25, 26] as well as virus-specific primer sets were used to amplify missing genomic regions uncovered by metagenomic sequences. The primer sets used for the genome determinations are available on request.

Table 2.

List and sequence relationships of various picornaviruses of four genera identified by NGS and blast analyses from a single faecal specimen of a 3-week-old paraplegic pig (SZ1M). Where it was necessary the missing genomic regions from the complete coding sequences (CDS) or complete genomes were determined by RT-PCR and Sanger sequencing methods using contig-specific primer pairs (data not shown). The lengths of the determined viral sequences are found in the ‘Final sequence length’ column

Genus	No. of reads	No. of contigs	Total length of the contig(s)	Final sequence length	Genome completeness	Accession No.	Closest relative by blast	nt identity	aa identity
Enterovirus	3205	4	5331 nt	7228 nt	complete CDS	MN807749	Porcine enterovirus 3 h [HQ702854]	78 %	88 %
Kobuvirus	895	1	7951 nt	7951 nt	complete CDS	MN807751	Porcine kobuvirus 1 [GQ249161]	89 %	96 %
Teschovirus	674	2	6593 nt	6981 nt	complete CDS	MN807750	Porcine teschovirus 3 [MF170908]	87 %	96 %
Sapelovirus	111	1	779 nt	7534 nt	complete genome	MN807752	Porcine sapelovirus 1 [LC425415]	78 %	84 %

nt, nucleotide; aa, amino acid.

For PSV screening reactions, a generic oligonucleotide primer pair (PSV-3Dscreen-6595-R / PSV-3Dscreen-6209-F, Table S1) was designed for the conserved 3D RNA-dependent RNA polymerase (3D^RdRp) genomic regions of all available PSV sequences including SZ1M-F/PSV/HUN/2013. For complete VP1 sequence-based PSV typing multiple sets of PSV-specific primers were designed to the VP3, VP1 and 2A regions of PSVs including SZ1M-F/PSV/HUN/2013 (Table S1).

The reaction conditions and reagents used for the RT and PCR reactions of screening, typing and genome acquisitions were the same as described previously with some modifications [25]. Briefly, in the RT reactions, 1 µl of total RNA sample was used in the final reaction volume of 25 µl. The 25 µl of cDNA was then used for PCR in a final volume of 50 µl. The generated PCR products were directly sequenced in both directions using BigDye Terminator v1.1 Cycle Sequencing Kit (Thermo Fisher) and run on an automated sequencer ABI 3500 Genetic Analyzer (Applied Biosystems, Hitachi, Tokyo, Japan).

RT-qPCR-based absolute quantification of viral copy numbers of PSV

For the absolute quantification of PSVs in faecal samples, 1 µl of total RNA samples were transcribed using random hexamer (500 ng/reaction), oligo dT (250 ng/reaction) primers and M-MLV-RT enzyme (Promega, USA) in a final volume of 10 µl. For quantitative PCR (qPCR), 1 µl of undiluted cDNA products and a PSV-specific qPCR primer-pair targeting the 3D^RdRp regions of PSVs was amplified in a SYBR Green-based real-time PCR assays (Maxima SYBR Green qPCR Master Mix, Thermo Scientific, USA) in a final volume of 20 µl (Table S1). All of the quantification steps of the qPCR reactions were followed by DNA duplex dissociation assays run on a Rotor-Gene Q MDx 5plex thermal cycler (Qiagen, Hilden, Germany). For the generation of standard curve tenfold dilution series of purified and spectrophotometrically quantified (NanoDrop 2000) 3D^RdRp screening RT-PCR product of SZ1M-F/PSV/HUN/2013 was used (Table S1). The qPCR assays contained three technical repeats of all the samples and standards. The slope of the standard curve was −3.72765. The calculated PCR-efficiency was 99.829 %. Note that due to the random amount of nasal and rectal samples collectable by polyester-tipped swabs therefore absolute quantification of PSVs by RT-qPCR was not conducted in swab samples.

Sequence and phylogenetic analyses

The donut charts of metagenomic sequences were generated by the GraphPad Prism ver 6.01 software. Sapelovirus sequences were downloaded from GenBank database and aligned with the study strains using Multiple Sequence Comparison by Log-Expectation (muscle) algorithm (https://www.ebi.ac.uk/Tools/msa/muscle/). The pairwise nucleotide (nt) and amino acid (aa) identity calculations were performed by the GeneDoc v. 2.7 software [27]. Phylogenetic trees were constructed using mega software ver. 7.026 [28]. The neighbour-joining model with the Jukes–Cantor method was used to construct all nt trees, while the Jones–Taylor–Thornton model was used to create the aa phylogenetic trees of 3D^RdRp, complete P1 and complete VP1. Bootstrap (BS) values were set to 1000 replicates and only BS ≥50 % are indicated in phylogenetic trees. Predictions of potential proteolytic cleavage sites of SZ1M-F/PSV/HUN/2013 were based on aa alignments with strains of PSV-1. The RNA folds of the 5′ and 3′ UTR sequences were predicted using the Mfold program [29] and the generated models were visualized and annotated using VARNA software and the CorelDraw Graphics Suite v. 12 [30]. Similarity plot was calculated with RDP software ver. 4.16 using the Kimura (two-parameter) model with a window size of 200 nt and a step size of 20 nt [31]. SimPlot ver. 3.5.1 and RDP ver. 4.16 software were used for bootscanning analyses (recombination detection) [31, 32]. The complete genomic sequence of PSV strain SZ1M-F/PSV/HUN/2013 (MN807752), the complete CDS sequences of further three picornaviruses (MN807749-MN807751) and the 25 complete (MN807753-MN807777) and two partial VP1 sequences (MN807778, MN807779) of the Hungarian PSV strains identified in this study are available in GenBank.

Cell culture and virus inoculation

PK-15 cell line was cultured at 37 °C in a humidified 5 % CO₂ atmosphere using Dulbecco’s Modified Eagle’s Medium (DMEM, Lonza, Walkersville, USA) supplemented with 5 % FBS (Biowest, Nuaillé, FR) and antibiotic-antimycotic solution (Sigma-Aldrich, St. Louis, USA). Faecal suspensions, re-suspended rectal and nasal swabs were centrifuged (10.000 g, 10 min) then the supernatants were passed through first a 0.45 µm membrane filters (Millipore, Bedford, MA). The filtered suspensions were inoculated on 90 % confluent PK-15 cell line grown in sterile 12-well cell-culture plates (Greiner Bio-One, Kremsmünster, A) either undiluted or after 1:100 dilution with culture medium. In total, 500 µl inoculum was incubated with the cells for 60 min at 37 °C followed by washing twice with PBS and addition of fresh DMEM. After 4 days' incubation the cell cultures were assessed for cytopathic effects (CPE). Following repeated freeze–thawing and centrifugation the culture supernatants were transferred to fresh PK-15 cell cultures. After two passages the supernatants were collected for subsequent analysis.

Results

Metagenomic overview

For diagnostic purposes a single faecal specimen (sample/animal ID: SZ1M) of a 3-week-old pig with a sign of mild paraplegia as a part of an ongoing local epidemic in January 2013 in Székelyszabar, Hungary was subjected to viral metagenomics (Fig. 1a, Tables 1 and S2). A total of 18 844 viral reads were identified by bioinformatics analyses of which 10 480 reads originated from viruses infecting bacteria (Microviridae: N=7247 and Siphoviridae: N=48…), from various unassigned viruses (N=3119) and from viruses with unresolved hosts (Picobirnaviridae: N=69) (Fig. 1b). The remaining 8360 reads showed similarity to multiple viruses of eukaryotic organisms including invertebrates (Iflaviridae: N=35) and vertebrates (Picornaviridae: N=4885; Circoviridae: N=2481; Astroviridae: N=830; Caliciviridae: N=69 and Parvoviridae: N=45) (Fig. 1b).

Initial blast analyses of the most abundant picornavirus-related sequences indicated the presence of at least four different viruses related to the members of genera Enterovirus, Kobuvirus, Teschovirus and Sapelovirus (Table 2). The complete coding sequences (CDS) of SZ1M-F/PEV/HUN/2013 (MN807749) of genus Enterovirus, SZ1M-F/PKV/HUN/2013 (MN807751) of genus Kobuvirus and SZ1M-F/PTV/HUN/2013 (MN807750) of Teschovirus and in the complete genomic (CG) sequence of SZ1M-F/PSV/HUN/2013 (MN807752) of genus Sapelovirus was determined using different RT-PCR methods and Sanger-sequencing. The closest genetic relatives of the four different picornaviruses are shown in Table 2. According to the sequence comparisons the most divergent picornavirus found in sample SZ1M was porcine sapelovirus with an overall 84 % aa identity to the closest relative, Japanese porcine sapelovirus 1 strain PSV/Porcine/JPN/MoI3/2016 (LC425415) identified by blastp search (Table 2), therefore further analyses were focused to this PSV strain.

Genome analysis of porcine sapelovirus strain SZ1M-F/PSV/HUN/2013

The complete genome of the PSV strain SZ1M-F/PSV/HUN/2013 (MN807752) is 7534 nt [excluding the poly(A)-tail], which is predicted to contain only a single, 6987-nt-long ORF flanked by a 464-nt-long 5′ UTR and 80-nt-long 3′ UTR (Fig. 2a). The 5′UTR of the study virus and other PSVs shows 88–90 % nt pairwise identities. The 5′UTR of SZ1M-F/PSV/HUN/2013 is predicted to contain a type IV-like IRES (data not shown). The short 3′UTR of the study virus shows 86–98 % identity to the 3′UTR sequences of other known PSVs found in GenBank and is predicted to form three hairpins with a pseudoknot (Fig. 2b). The transcription initialization site of the single ORF was found in an optimal Kozak context (cCCA ₄₆₅ UGG, start codon is underlined, conserved nts are capitalized). The single ORF encodes a 2328-aa-long viral polyprotein, which shares the same organization of L-4-3-4 and similar predicted cleavage sites as previously found in other sapeloviruses [4, 14] (Fig. 2a). The potential cre with the conserved AAACA motif in the loop is presumably located at the 2C region between nt position 4463 and 4505 (Fig. 2c). Amino acid sequence comparisons of the study strain and the closest relative of PSV-1 strain PSV/Porcine/JPN/MoI3/2016 (LC425415) show that while the identity values are relatively high between the main non-structural P2 (85%) and P3 (91%) regions the level of identity is considerably lower between the P1 capsid-encoding (75%) regions (Fig. 2d). The lowest identity between the study virus and PSV/Porcine/JPN/MoI3/2016 (64 % nt/62 % aa) was measured in the capsid (VP1) and the most diverse regions are also located in the P1 (Fig. 2d). The highest identity was found in the non-structural 3D^RdRp region (88 % nt/96 % aa) (Fig. 2a and d).

Fig. 2.

(a) Genome maps with presumed cleavage sites (P5/P2′) of SZ1M-F/PSV/HUN/2013 (top) and its closest relative (bottom) of PSV-1 strain PSV/Porcine/JPN/MoI3/2016. The nucleotide (upper numbers) and amino acid (lower numbers in brackets) lengths of the corresponding genomic regions of the study virus are shown in each gene box. The pairwise nucleotide (upper numbers) and amino acid identity values (lower numbers in brackets) (%) of each genomic region are found between the genome maps. (b) Predicted secondary RNA structure of the 3′ end of the genome of SZ1M-F/PSV/HUN/2013. PK: pseudoknot. Nucleotides highlighted with grey represent the presumed stop codon. (c) Predicted secondary RNA structure of the cre. Conserved nucleotides of the structure are highlighted with grey. (d) Similarity plot of the complete genome of SZ1M-F/PSV/HUN/2013 with pairwise identity (ID) values (%) compared with the closest relative sequence of PSV-1 strain PSV/Porcine/JPN/MoI3/2016. Black arrows indicate the most diverse sites in the capsid.

The SZ1M-F/PSV/HUN/2013 clustered together with the other PSVs in both of the nt and aa phylogenetic trees of 3D^RdRp (Fig. 3) however, it was separated from the other known PSV strains and formed a novel cluster with a relatively high bootstrap support in both of the nt and aa phylogenetic trees of P1 (Fig. 4). Based on the sequence divergence of P1 and phylogenetic position in the P1 trees SZ1M-F/PSV/HUN/2013 is the most divergent porcine sapelovirus strain relative to those currently known.

Fig. 3.

Phylogenetic trees of 3D^RdRp [(a) nucleotide, (b) amino acid] and full-length P1 capsid [(c) nucleotide, (d) amino acid] of SZ1M-F/PSV/HUN/2013 (bold with black arrows) and most closely related porcine sapeloviruses (Sapelovirus A) as well as representative sequences of species Sapelovirus B of genus Sapelovirus. Human poliovirus 1 of genus Enterovirus and the unassigned Bat sapelovirus 1 were used as outgroups in the 3D and P1 trees. The PSV strains related closely to SZ1M-F/PSV/HUN/2013 were chosen from the list of closest hits of blast analyses.

Fig. 4.

Complete nucleotide (a) and amino acid (b) phylogenetic trees of VP1 capsid of study strains and closest relatives of PSV with countries of origins and collection dates as well as representative sequences of species Sapelovirus B of genus Sapelovirus. The unassigned Bat sapelovirus 1 was used as the outgroup. Sequences from the same swine farm are highlighted with the same colour. Sequences from paraplegic pigs are marked with black circles. Double-headed arrows indicate the VP1 sequences identified from different samples (i.e. nasal and rectal swabs) of the same animal. Black arrows: phylogenetic positions of SZ1M-F/PSV/HUN/2013 and its closest relative. Box with a dotted line: a main branch which contains the majority of the Hungarian VP1 sequences. The study strains are named as follows: FarmID-sample number-sample source (-F: faecal, -RS: rectal swab, -NS: nasal swab, -VC: viral culture)/PSV/country (HUN=Hungary)/year of sample collection. Similar trees with all of the currently known PSV strains with completely determined VP1 are shown in Fig. S1.

Before virus isolation attempts the average viral genome copy number of the sapelovirus SZ1M-F/PSV/HUN/2013 in the faecal sample used for the inoculation (2.90E+06/ml faeces) was determined by RT-qPCR. Although clearly visible CPE was observed after multiple passages in the SZ1M faecal sample-inoculated swine-derived PK-15 cell culture (culture ID: VC2-F, Fig. S1) neither entero-, kobu- and teschoviruses nor porcine sapelovirus was detectable by different RT-(q)PCR methods from the cell-culture supernatant (Table S1).

Epidemiological investigation of PSV in Hungarian swine herds

In order to investigate the prevalence of PSVs as well as to detect divergent PSV strains similar to SZ1M-F/PSV/HUN/2013 in Hungarian swine herds, a novel, in-house designed universal PSV screening primer pair was used (Table S1). Faecal, serum, rectal and nasal swab samples (N=111 including N=33 rectal–nasal swab pairs) from asymptomatic (N=41) and paraplegic (N=21) swine of different ages collected from 12 geographically distant swine farms were screened (Tables 1 and 3 S2, Fig. 1a). Altogether, 57 out of the 111 samples collected from 44 of 62 animals were RT-PCR-positive for PSV (Table 3), the overall PSV RNA positivity was 51.4 %. The 387-nt/129-aa long partial 3D polymerase sequences acquired from the screening RT-PCR reactions showed 85–97 % nt and 93–100 % aa identity between each other (data not shown).

Table 3.

Prevalence of porcine sapleoviruses in different age groups of pigs based on the results of 3D^RdRp RT-PCR screening reactions

Age group	No. of animals	Nasal swabs	Rectal swabs/faeces	Serum
Suckling pigs (0–35 days)	31/38 (81.2 %)	16/29 (55.2 %)	27/38 (71.1 %)	n/a
Nursery pigs (36–77 days)	9/10 (90.0 %)	0/10 (0 %)	9/10 (90.0 %)	1/10 (10 %)
Fattening pigs (78–140 days)	3/6 (50.0 %)	n/a	3/6 (50.0 %)	n/a
First parity sows (>140 days)	1/8 (12.5%)	n/a	1/8 (12.5 %)	n/a
					∑	44/62 (71.0 %)	16/39 (41.0 %)	40/62 (64.5 %)	1/10 (10.0 %)

n/a, sample not available; No: number of

Majority of the enteric samples (64.5 %) were PSV-positive followed by positive nasal swabs (41 %). Only one (10 %) of the ten investigated serum samples contained detectable PSV RNA (Table 3). The majority of the infected animals were nursery pigs (90 % positivity) followed by suckling pigs (81.2 %), fattening pigs (50 %) and sows (12.5 %) (Table 3).

The prevalence of PSV-positive animals in different farms (where sample number was ≥10) ranged between 20 % (farm D) and 100 % (farm GD, farm C) (Tables 1 and S2). Interestingly, there are farms (e.g. farm A and farm D) where PSV was more often detectable from nasal swab samples (20 and 100 %) than enteric samples (0 and 80% positivity), while in other farms like farm EG or farm B, PSV was detectable only in enteric (90 and 80% positivity) but not in nasal samples (Tables 1 and S2). The age groups (suckling pigs) of the sampled animals and sampling times (August, 2016) of these farms are similar with the exception of farm EF where nursery pigs were sampled (Tables 1 and S2).

Fourteen nasal-rectal swab pairs collected from suckling pigs from three different farms (farms A, C and GD) were PSV-positive by RT-PCR (Table S2). Asymptomatic (AS) and paraplegic (PP) animals were both sampled in only one farm (farm GD). No significant difference was found in the PSV-positivity between the nasal (PP: 4/5, 80 %; AS: ¾, 75 %) and rectal swab samples (both 100 %) of asymptomatic and paraplegic animals (Tables 1 and S2).

For PSV isolation beside SZ1M two additional nasal swabs (NS), one rectal swabs (RS) and four faecal samples (F) were randomly selected (culture IDs: VC-1, VC-3–8) to inoculate swine-derived PK-15 cell cultures (Table S2). The average viral copy numbers of PSVs of the faecal samples used for the inoculations were ranged between 2.90E+06 and 4.24E+08 copies/ml faeces. Although clearly visible CPE was observed after multiple passages in all of the inoculated cultures only three culture supernatants (VC4-F, VC6-RS and VC7-F) were positive to PSV by RT-PCR (Table S2, Fig. S1). The initial viral copy numbers of faecal samples used for successful cultivations were 5.01E+06 (VC7-F, BUV1-F/PSV/HUN/2014) and 4.24E+08 (VC4-F, EF2-F/PSV/HUN/2016). None of the PK-15 cell cultures inoculated by nasal swab samples were RT-PCR-positive to PSV. There were no entero-, tescho- or kobuviruses identifiable from the culture supernatants with the exception of VC4-RS where, beside PSV, a porcine teschovirus was also detectable by RT-PCR (data not shown).

VP1 capsid sequence-based typing of field strains of PSV

From the total of 57 PSV 3D^RdRp RT-PCR-positive samples only 25 complete PSV VP1 sequences could be determined by RT-PCR sequencing using multiple primer sets targeting the VP3, VP1 and 2A genomic regions of different PSVs including SZ1M-F/PSV/HUN/2013 (Table S1). Twenty-one VP1 sequences were identified from enteric samples from nine swine farms while only four VP1 sequences were originated from nasal swabs of four swine farms. The VP1 sequences show 61–99 % nt and 62–100 % aa pairwise sequence identities to each other. The study strains of PSVs are named as follows: FarmID-sample number-sample source (-F: faecal, -RS: rectal swab, -NS: nasal swab)/PSV/country (HUN=Hungary)/year of sample collection, i.e. B4-RS/PSV/HUN/2016.

Phylogenetically, the majority (19/25) of VP1 sequences from 9/12 swine farms clustered together and formed a main branch with PSV sequences originating from Italy, Spain and China (Figs 4 and S2). Interestingly, there were PSV strains such as B4-RS/PSV/HUN/2016 (farm B) and EF8-F/PSV/HUN/2016 (farm EF) or SZ1M-F/PSV/HUN/2013 (farm SS) and EF9-F/PSV/HUN/2016 (farm EF) that belonged to the same lineages but originated from geographically distant farms (Katymár, Egyházasfalu and Székelyszabar) (Figs 1a and 4 and S2). Farm C, farm BUV and farm ZS yielded phylogenetically closely related strains while in farm B, farm EF and farm SS multiple, phylogenetically distant PSV strains were co-circulating (Figs 4 and S2). Interestingly, three closely related strains (ZS1-F/PSV/HUN/2013, ZS2-F/PSV/HUN/2013, ZS3-F/PSV/HUN/2013) from faecal samples of paraplegic pigs from the same farm (farm ZS) clustered together with a historic PSV isolates of 26-T-XII (AY392544) and 16-S-X (AY392543) from Hungary, 1963 (Figs 4 and S2).

The identified PSV strains of paraplegic pigs mostly formed multiple lineages distant from each other and from the three currently known ‘neurotropic’ PSV strains identified from CNS samples (Figs 4 and S2). Some of the Hungarian strains like TM1121-F/PSV/HUN/2013 (MN807769) or SZ4M-F/PSV/HUN/2013 (MN807753) being closely related to strains identified from asymptomatic pigs (Figs 4 and S2).

There are a total of three PSV-positive nasal–rectal swab pairs collected from suckling pigs from three different farms (farms A, C and GD) where complete VP1 sequences could be determined from each sample type. The VP1 sequences from both nasal and rectal swabs were identical in two cases (B1 and GD01) while the VP1 sequence from nasal swab of the third animal (A3-NS/PSV/HUN/2016, MN807755) from farm A shows only 84 % nt/92 % aa identity to the VP1 sequence from the same animal’s rectal sample (A3-RS/PSV/HUN/2016, MN807756). These two VP1 sequences also separated phylogenetically from each other in both the VP1 nt and aa phylogenetic trees (Figs 4 and S2).

Only a single VP1 sequence (EF9-F/PSV/HUN/2016, MN807773) detected in a faecal sample of a nursery pig from farm EF showed relatively high (87 % nt/92 % aa) pairwise sequence identity with the divergent PSV strain SZ1M-F/PSV/HUN/2013. These two strains form a single cluster separated from the other known PSV strains with relatively high bootstrap support in both of the nt and aa phylogenetic trees of VP1 (Figs 4 and S2). The 3D^RdRp sequences of these two viruses share 86 % nt and 95 % aa identity (data not shown).

VP1 capsid sequence-based typing of PSV isolates

The partial, 561-nt-long VP1 sequences of strain VC7-BUV1-F/PSV/HUN/2016 and VC6-GD01-RS/PSV/HUN/2016 and complete, 879-nt-long VP1 of VC4-EF2-F/PSV/HUN/2016 were also determined from the PSV-positive cell-culture supernatants. Two of these sequences (VC4-EF2-F/PSV/HUN/2016 and VC6-GD01-RS/PSV/HUN/2016) were nearly identical (99 % nt identity with one and three synonymous mutations) to the corresponding genome parts of PSV strains (EF2-F/PSV/HUN/2016 and GD01-RS/PSV/HUN/2016) identified from the enteric samples used for inoculation of the cultures. However the third, partial VP1 sequence of VC7-BUV1-F/PSV/HUN/2016 shows only 88 % nt and 93 % aa identities to the corresponding genome part of inoculum strain of BUV1-F/PSV/HUN/2016 (data not shown). Note that because the 3′ ends of VP1 sequences of VC7-BUV1-F/PSV/HUN/2016 and VC6-GD01-RS/PSV/HUN/2016 (ca 318 nt was missing) were unable to amplify using RT-PCR and various capsid primers (Table S1) these strains are not included in the phylogenetic analyses of complete VP1.

Discussion

In this study, the complete genomic sequence of a novel, divergent porcine sapelovirus strain SZ1M-F/PSV/HUN/2013 (MN807752) was determined using NGS, RT-PCR and Sanger sequencing. The study strain of PSV was identified from a faecal sample (SZ1M) from a paraplegic pig parallel with multiple other co-infecting viruses including a porcine entero-, tescho-, and kobuvirus of which complete CDS sequences were also determined. Beside PSVs certain strains of porcine teschoviruses (PTVs) were also reported to cause CNS infections associated with mild or severe neurological disorders including paraplegia [33, 34]. Although in our case without available tissue samples from the CNS the pathogenic role of any picornaviruses (including PTV and PSV) found in the faeces are not determinable.

The identified PSV strains have identical IRES (type-IV), same localization of cre in the 2C and similar organization of the single viral polypeptide (L-4-3-4) and structurally similar 3′UTR as reported in other sapeloviruses [10, 13, 14, 16]. The strain SZ1M-F/PSV/HUN/2013 shows various levels of sequence similarity in structural P1 (75 % aa identity) and non-structural P2-P3 (85–91 %) genomic regions compared to the closest relative of PSV-1 strain PSV/Porcine/JPN/MoI3/2016 (LC425415), which could suggest the recombinant nature of the virus. The results of pairwise sequence comparisons between the study virus and its closest PSV-1 relative of all 12 viral peptide-encoding genome regions and different positions of SZ1M-F/PSV/HUN/2013 in the P1 and 3D phylogenetic trees may also reflect the independent origin of the P1 and P2-P3 genome regions. Recombination is relatively frequent among PSVs and a putative recombinant hotspot was identified near the 3′ end of the P1 region of PSVs [10, 20], which could be the potential recombination breakpoint of SZ1M-F/PSV/HUN/2013, therefore the recombinant nature of the study PSV strain is plausible. Unfortunately, due to the lack of potential parental sequences of P1 none of the recombination detection programs (RDP4 or SimPlot) showed clear in silico evidence supporting this hypothesis (data not shown). However, beside recombination, selective pressure of the capsid-encoding genome region of the SZ1M-F/PSV/HUN/2013 could also explain the separation of the virus from the other known PSVs in the P1 phylogenetic trees.

The species Sapelovirus B of genus Sapelovirus currently consists of at least three different simian sapelovirus genotypes while until now all of the currently known PSV strains belong to a single geno/serotype of species Sapelovirus A [1, 10, 19]. Serotyping of historic porcine ‘enteroviruses’ including sapelovirus strains has traditionally been largely based on neutralization or indirect immunofluorescence assays using multiple sets of serotype-specific monoclonal antibodies applied on cultured viruses [35, 36]. Although most PSVs are able to replicate in cell lines of swine-origin including PK-15 [5–7] all the cultivation attempts of PSV strain SZ1M-F/PSV/HUN/2013 in PK-15 cell cultures failed despite the relatively high initial viral copy number (2.90E+06 copies/ml) of the virus in the sample used for inoculation. It is currently not known whether SZ1M-F/PSV/HUN/2013 is unable to grow in PK-15 cell line or the sample did not contain replication competent PSV viruses, but the classical serotyping of SZ1M-F/PSV/HUN/2013 using antibody-based serological assay could therefore not be conducted.

To our knowledge there are no widely accepted sequence-based genotype demarcation criteria in species Sapelovirus A but in Sapelovirus B viruses belong to different genotypes are ≤73 % identical in VP1 nucleotide sequence (share ≤85 % amino acid identity), and the phylogenetic relationships of candidate strains in the VP1 phylogenetic trees generally mirror those revealed by pairwise sequence comparisons, with high bootstrap support [19]. Using the above mentioned criteria, the low (64 % nt/63 % aa) sequence homology in the VP1 region between the study strain of SZ1M-F/PSV/HUN/2013 and its closest relative of PSV-1 and the phylogenetic separation from the other known PSV strains in both of the nt and aa phylogenetic trees of P1 and VP1 with high bootstrap support could suggest that SZ1M-F/PSV/HUN/2013 could be the first member of a novel, second genotype of PSV. VP1 sequence identities between the currently known PSV strains are measured to be ≥70.8 % nt and ≥75.1 % aa and none of the previously known PSVs formed a distinct lineage in the VP1 phylogenetic trees [10].

In order to investigate the prevalence and VP1-based genetic diversity of PSVs circulating in Hungarian swine farms as well as to detect further divergent PSV strains similar to SZ1M-F/PSV/HUN/2013 3D^RdRp, and VP1-based RT-PCR screening reactions were conducted on enteric (faecal, rectal swabs), respiratory (nasal swabs) and serum samples of asymptomatic and paraplegic pigs from 12 different swine farms. PSVs were detectable with variable prevalence in all of the investigated swine farms located in different regions of Hungary (Fig. 1a) suggesting the endemic presence and widespread of this virus in both local industrial and backyard herds similarly as found in Japan, China and the USA [10, 20, 37].

According to our results PSVs were most prevalent among nursery (90 % positivity) and suckling pigs (81.2 %), found in considerably lower prevalence in fattening pigs (50 %) and only a single sow was PSV-positive (Table 3). The overall sample positivity (51.4 %) and the high prevalence of PSVs in younger animals (suckling and nursery pigs) are fairly similar to those found in China and Japan [10, 20] although highest prevalence (69.4 %) of PSVs were found among fattening pigs in China. This incongruence was most likely due to the low number of available samples from Hungarian fattening pigs (n=6, Table 3). PSV was detectable from a single faecal sample of a first-breeding sow, which further supports the hypothesis of sows could serve as initial sources of PSV infection of piglets [8, 9].

PSVs were mostly detectable from enteric samples followed by nasal swabs and rarely in serum (Table 3). Although PSVs have been suspected as porcine pathogens in both enteric as well as respiratory diseases [5, 38, 39], and PSV is detectable in lung tissue samples [5, 6] this is to our knowledge the first report of PSV RNA in nasal swabs. Despite the same age of animals and time of sampling the prevalence of PSVs in enteric samples and nasal swabs showed great variations. In some populations PSVs were more prevalent in nasal samples while in other enteric samples were more frequently PSV positive. There are a total of 14 nasal–rectal swab pairs collected from suckling pigs from three different farms where both sample types were PSV positive. The presence of PSVs in the nose could indicate either disseminated, or respiratory infections or even passive, non-replicating presence of PSV breathed into nasal passages from the environment. As pigs are kept in crowded barns, contamination of snouts after contact with faeces or the rear end of litter mates could also explain virus detection in nasal swabs. The ability of some isolated PSV strains to cause viraemia and disseminated infections, which affect the respiratory system in experimentally infected pigs [5, 6] suggest that – at least in certain cases – PSV found in the nasal cavities could be originated from dissemination. According to VP1 sequence analyses, the same PSV strains were identifiable from both of the enteric and nasal swab samples collected from two suckling pigs held in distant farms while different PSVs were identifiable in the nasal and rectal swabs collected from a third animal. Furthermore, previous reports showed that PSVs could be associated with variety of diseases including pneumonia [8, 9]. Beside host-related factors such as genetic background, age, present state of the immune system etc. the diverse receptor spectra of different PSVs could be responsible for the various clinical outcomes of the infections [40, 41]. Whether PSV exists with different tissue tropism will require further experimental studies and large-scale epidemiological investigations.

From the total of seven, randomly selected PSV-positive enteric (n=5) and respiratory samples (n=2) used for virus cultivation in PK-15 cell line only three PSV strains, all from enteric samples could be successfully cultivated. Only a minority of the analysed Hungarian field PSV strains can therefore be cultivated in swine kidney cells (PK-15) similar to results found in another study [20]. The VP1 sequences of the three cultured (cell-line-adapted) PSVs and in their original feacal samples were also compared. Beside nearly identical sequences (only a few synonymous substitutions were found) considerably different (88 % nt and 93 % aa identity) VP1 sequences of the cell culture adopted and the inoculum-strains could also be detected. Cultured PSV strains may therefore not always reflect the dominant PSV strains detectable in the original faecal sample used for inoculation.

Beside the use of multiple sets of general primer-pairs targeting different parts of the capsid (Table S1) full-length VP1 sequences could only be determined from less than half of the PSV 3D^RdRp RT-PCR positive samples (mostly from enteric origin) probably due to primer mismatch, which could reflect a high sequence divergence of field strains of PSVs as found in other countries [10, 20, 39].

Most of the VP1 sequences create farm-specific, closely related lineages, although there were farms where phylogenetically distant PSV strains were present suggesting the local endemic circulation of multiple PSV strains. We detected a few VP1 clusters with phylogenetically closely related PSV originating from geographically distantly farms. Strain-exchange of PSVs between different farms through unknown routes may account for these results.

During our search for divergent PSVs related to SZ1M-F/PSV/HUN/2013, only a single VP1 sequence (EF9-F/PSV/HUN/2016, MN807773) could be identified from a faecal sample of an asymptomatic nursery pig from farm EF, which shares relatively high sequence homology (87 % nt/92 % aa). The EF9-F/PSV/HUN/2016 fell into the same phylogenetical VP1 cluster as SZ1M-F/PSV/HUN/2013 (Figs 4 and S2), and could therefore belong to the same novel second porcine sapelovirus-A genotype. These two sapeloviruses were detected in geographically distant swine farms approximately 500 km apart (Fig. 1a). Further sampling will be required to measure the prevalence of these related sapeloviruses in Europe, which based on our initial results is significantly lower in Hungary than that of the ‘classic’ PSVs.

Only enteric, and in cases of animals from farm GD additional nasal swab samples were collected from paraplegic pigs. No tissue specimens were therefore available for further investigations. Beside the recently recognized pathogenic potential of PSVs in certain cases of CNS infections [11, 12] any role of classic and novel PSV strains in the development of any CNS-related symptoms among investigated paraplegic cases in this study remains unknown. In only one farm (farm GD) were samples (rectal–nasal swab pairs) collected from both asymptomatic and paraplegic pigs. No significant difference was found in the PSV positivity between the nasal and rectal swab samples of asymptomatic and paraplegic animals (Tables 1 and S2). Furthermore, the PSV strains of paraplegic pigs formed multiple lineages in the VP1 phylogenetic trees, distant from the three currently known ‘neurotropic’ PSV strains, some of them related closely to strains identified from asymptomatic pigs (Figs 4 and S2).

The discovery of a proposed novel genotype of PSV, the high prevalence of PSVs found in nasal swabs, and the high genetic diversity of VP1 sequences previously reported and also seen here in Hungary indicates that the genetic and possibly phenotypic diversity of PSVs are still not completely characterized. It should be necessary to explore the full genetic diversity of circulating PSVs by more sensitive methods in swine, which is one of the most important food animals for humans.