Prevalence of fecal viruses and bacteriophage in Canadian farmed mink (Neovison vison)

Abstract Recent viral metagenomic studies have demonstrated the diversity of eukaryotic viruses and bacteriophage shed in the feces of domestic species. Although enteric disease is a major concern in the commercial mink farming industry, few etiologic agents have been well characterized. This study aimed to identify viruses shed in the fecal matter of clinically healthy commercial mink from 40 southern Ontario farms. Viral RNA was extracted from 67 pooled fecal samples (30 adult female mink and 37 kit) and amplified for Illumina sequencing on the NextSeq platform, and the resulting contigs were trimmed and assembled using Trimmomatic 0.36.0 and Spades 3.8.0 in iVirus (CyVerse, AZ, USA) and SeqMan NGen 12 (DNAStar, WI, USA). Identification of assembled sequences >100 bp (Geneious 10.1.3) showed an abundance of bacteriophage sequences, mainly from families Siphoviridae (53%), Podoviridae (22%), Myoviridae (20%), Inoviridae (1%), Leviviridae (0.04%), Tectiviridae (0.01%), and Microviridae (0.01%). A diverse range of vertebrate viruses were detected, of which posavirus 3, mink bocavirus, gyroviruses, and avian‐associated viruses were most abundant. Additionally, sequences from nonvertebrate viruses with water and soil‐associated amebal and algal hosts were also highly prevalent. The results of this study show that viruses shed in the fecal matter of healthy commercial mink are highly diverse and could be closely associated with diet, and that more research is necessary to determine how the detected viruses may impact mink health.

As mink mortality is a production concern, identifying viruses that may play a role in mink health and disease would further the understanding of agents involved in mink enteritis and lead to the development of improved monitoring and treatment strategies.
Additionally, assessment of prevalent bacteriophages may provide insight into the bacterial populations that can cause disease in mink, and help to understand the relationship between phage and bacterial populations. The objective of this study is to identify the prevalent mammalian, environmental, and phage viruses shed in the feces from clinically healthy commercial adult female mink and mink kits from 40 Ontario farms.

| Sample collection, dilution, and filtration
Sixty-seven pooled fecal samples were collected between July and October of 2014 from 40 Ontario mink farms. Thirty-seven pooled kit fecal samples and 30 pooled adult female fecal samples were collected from under three pens, representing up to three adult female mink per sample or up to 15 mink kits per sample. Information on farm location, recent history of antimicrobial use, and mink coat color was collected for each farm. Samples were collected in plastic bags and stored at −80°C until processed. To prepare a 10% fecal sample dilution, the samples were thawed and mixed thoroughly in the bag, and then 1 g of fecal matter was added to 9 ml of phosphate-buffered saline. The sample was then centrifuged at 10,000× g for 15 min at 4°C to remove large particulates and bacteria. The supernatant was removed, filtered (Millipore syringe 0.45 μm filters), and stored at −20°C.

| Purification and extraction of viral nucleic acids
To reduce nonviral nucleic acids, 200 μl of filtered supernatant was treated with a nuclease mixture of 7 μl TURBO DNase (Ambion, Life Technologies, Grand Island, NY, USA), 3 μl Baseline-ZERO DNase (Epicentre, Chicago, IL, USA), and 1 μl of diluted RNase T1 (Fermentas Canada Inc., Burlington, ON) in 7 μl 1× DNase buffer (Ambion). This mixture was incubated at 37°C for 90 min (Victoria et al., 2009;Zhang et al., 2014). DNase and Baseline-Zero were inactivated by incubating for 20 min at 70°C. RNase T1 was inactivated during the first step of nucleic acid extraction. Viral nucleic acids were extracted from 200 μl of the DNase-and RNase-treated product (Invitrogen Viral RNA/DNA Extraction kit; ThermoFisher Scientific, Mississauga, ON, Canada). In the purification procedure, 20 μl of RNase-free water was used to elute nucleic acids.

| Viral cDNA synthesis and preamplification enrichment of viral cDNA and DNA
Ten microliter of extracted viral nucleic acids was incubated with 100 pmol of a primer consisting of a fixed 18 bp sequence with a random nonamer at the 3′ end (GCCGACTAATGCGTAGTCNNNNNNNNN) for 2 min at 85°C. cDNA synthesis was performed using reverse transcriptase from the QuantiTect Reverse Transcription kit (Qiagen, Mississauga, ON, Canada) according to manufacturer's instructions.
For pre-PCR amplification enrichment of viral cDNA and DNA, 10 μl of the cDNA synthesis product was incubated with 50 pmol of the previously described random primer at 92°C for 2 min, 4°C for 2 min, then with 5 U of Klenow fragment with 1× Klenow Buffer (New England Biolabs, Ipswich, MA, USA) at 37°C for 1 h (Li et al., 2015). A subset of randomly selected samples (16/67) were used to test for bacterial contamination using 16S real-time PCR using methods described by Kobayashi et al. (2006).

| PCR amplification and product purification
Klenow products were PCR amplified using KAPA 2G HotStart ReadyMix (Kapa Biosystems, Boston, MA, USA). Five microliter of the Klenow product was mixed with 1 μl of 2.5 mM a primer containing only the 18 bp fixed portion (GCCGACTAATGCGTAGTC) of the previously described primer. An additional 1 μl of 25 mM of MgCl 2 was added to the KAPA master mix. Temperature cycling was performed as follows: 1 cycle of 95°C for 5 min, 33 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 90 s.
Samples were kept at 72°C for an additional 10 min of extension and held at 4°C at the end of the run. PCR products were purified once using the Agencourt AMPure XP beads (Beckman Coulter, Brea, CA, USA) with a 0.8:1 ratio of beads to sample.
Eighty percent ethanol was used for the ethanol wash and 32 μl of elution buffer was used to extract purified DNA fragments from the beads.

| NGS library preparation and sequence data analyses
Sixty-seven samples (weaned kit n = 37, adult female n = 30) were prepared for NGS (next generation sequencing) using Nextera XT DNA Sample Preparation Kit (Illumina, San Diego, CA, USA).
Samples were sequenced using Illumina NextSeq500 V2 chemistry on a 2 × 125 cycle (Donnelly Centre, Toronto, ON, Canada), and reads were demultiplexed by Donnelly software. Low quality reads were filtered using Trimmomatic 0.36.0 in iVirus (CyVerse, AZ, USA) using default parameters. Trimmomatic output was used for de novo assembly in Spades 3.8.0 (CyVerse) using kmer size 65, and SeqMan NGen 12 (DNAStar, Madison, WI, USA) (Zhang et al., 2014). Assembled contigs >700 bp were aligned to the NCBI viral reference database (viral1.1.genomic.fna.gz) using BLASTn in Geneious 10.1.3 (Biomatters Ltd, Auckland, New Zealand) with an E value cut-off 10 −4 . The resulting reads that aligned over at least 100 bp with a reference viral sequence were compiled and used for further analysis. Top phage and nonphage viral families were identified for all sample libraries, and the sequences from specific viruses which had the highest prevalence were compared between for adult females and kits, and between farms grouped based on five geographical regions using JMP Software (SAS Institute, Cary, NC, USA) ( Figure S1). The most prevalent vertebrate virus sequences were further assessed based on identity, sequence length, and prevalence across samples. Viral sequences with lower levels of similarity in amino acid identity (average identity <90%) were then compared to these reference viral sequences (GenBank) to identify the level of identity of protein-encoding genes. All detected sequences for each virus with low average identity were used for de novo assembly in Geneious 10.1.3, followed by phylogenetic analysis in phylogeny.fr with their closest related viral sequences (BLASTn hits with the highest identity) (Dereeper et al., 2008).

| Statistical analysis
JMP (SAS Institute) was used to conduct one-way nonparametric Wilcoxon tests to compare the relative abundances of top phage and mammalian viral sequences between adult female mink and mink kits. For all statistical tests conducted, a p-value ≤ .05 is considered significant. Information collected on mink coat color was not used for statistical analysis due to inconsistent sampling. Seven viral families were identified in the top 12 most prevalent bacteriophage groups (76,558 sequences), including Siphoviridae

| Prevalent phage sequences
(0.04%), Tectiviridae (0.01%), and Microviridae (0.01%). An additional 4.8% of detected bacteriophage sequences were unclassified, with the majority belonging to the order Caudovirales. Pseudomonas phage sequences were found to be significantly higher in adult female mink samples (p = .02), but no other significant differences were found in other detected phage sequences between age groups.  Table 2). The remaining 15 isolates were not found to be resistant to any of the tested antimicrobials. The samples (16/67 sequenced samples) randomly selected for 16S rt-PCR were negative for bacterial contamination.

| Analysis of vertebrate viral sequences with low average identity
This study identified sequences from seven prevalent viruses that had low average identity (<90%) to the reference sequences of vertebrate viruses. The average identities of detected sequences, their prevalence in samples, as well as query-encoded proteins are listed in Table 4. Figure 1a shows the phylogenetic relationship be-   Zhang et al., 2014). The 12 most prevalent phages detected in this study represent 70% (76,558/109,612) (Cao et al., 2015;Gu et al., 2016), further investigation is required to understand the natural role that the associated bacteriophage species play in bacterial populations. Producers were asked to voluntarily report the use of antimicrobials on farms, but due to only partial completion of the survey, the information collected on antimicrobial use from the 2014 sample cohort may not be fully representative. Therefore, any relationship between antimicrobial use and the relative abundance of targeted bacterial species could not be analyzed.  (Fehér et al., 2014;Ng et al., 2014;Smits et al., 2013;Zhang et al., 2014). High numbers of sequences with 84%-96% identity to posavirus 3 strain 958-4 were identified, which has been previously detected in fecal samples collected from commercial swine in high animal density farms (Hause, Hesse, & Anderson, 2015). Hause et al. (2015) suggest that this strain of posavirus is derived from nematodes parasitizing commercial swine. The detected posavirus sequences may be the result of contamination from the soil at the time of fecal sample collection, but also could be attributed to the mink diet, which often consists of pork and poultry products, or nematode infections in the mink gut (Krog, Breum, Jenson, & Larsen, 2013 (Bodewes et al., 2014;Fehér et al., 2014;Krog et al., 2013;Smits et al., 2013). Further research is required to determine the correlation between diet and the fecal virome of mink. This is also the first report of mink bocavirus sequences in commercial mink fecal samples in Canada, with 98%-100% identity to the strain identified in 2016 in China . This strain was most closely related to feline bocavirus (JQ692585). Yang et al. (2016) found no correlation between mink bocavirus and diarrhea, but stated that these results may not be fully representative due to the small sample size.
Viruses with low average identity were used in de novo as- GyV3 were isolated from human fecal samples (Lamberto, Gunst, Muller, Hausen, & de Villiers, 2014;Phan et al., 2012Phan et al., , 2014. Six of the 15 prevalent vertebrate viruses described in this study are of avian origin. Although virus shedding does not represent active infections, some of the viruses identified in this study may have the potential to be transmitted to the humans, commercial and wild animals in close proximity to mink farms due to poor biosecurity (Compo et al., 2017).
In conclusion, this viral metagenomic study provides a preliminary overview of the commercial mink fecal virome, showing a diverse range of bacteriophage and eukaryotic virus sequences, including a potentially novel chapparvovirus. It is not known whether the detected bacteriophage and eukaryotic virus sequences represent commensal species, or if these viruses are capable of influencing bacterial populations and causing disease in mink. Further research is required to clarify the phylogeny of low-identity sequences identified in this study and to determine the role of these prevalent viruses in mink health.

ACK N OWLED G M ENTS
We thank the PHAC for support of AMR work, the Ontario Fur Breeders Association for their support as well as the producers who participated in this study. We would also like to thank Diego Gomez-Nieto, Jutta Hammermueller, Nicol Janecko (PHAC), and Rachel MacDonald for technical support.

AUTH O R CO NTR I B UTI O N
P.V.T., B.T., and J.S.W. conceived of the work and prepared the grant; X.T.X., B.T., A.K., and J.S.W. conducted the work and analyzed the data; X.T.X. and P.V.T. co-wrote the paper, and all authors contributed to manuscript review.

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest.