First isolation and whole genome characterization of porcine deltacoronavirus from pigs in Peru

Abstract Porcine deltacoronavirus is a newly emergent enteric pathogen affecting swine farms worldwide. It has been detected in several countries in Europe, Asia and North America; yet, it has not been reported in South America. In November 2019, an enteric disease outbreak in a pig farm located in San Martin, Peru, was reported along with submission of three intestinal samples from pigs who succumbed to the disease. Samples were processed for molecular detection by qRT‐PCR, viral isolation and further sequencing analysis. A taqman‐based RT‐PCR was performed to differentiate among the most relevant swine enteric coronaviruses described to date. All samples were positive to porcine deltacoronavirus with a cycle threshold (Ct) value between 9 and 14, revealing a high viral load, while testing negative to porcine epidemic diarrhea and transmissible gastroenteritis viruses. Following detection, viral isolation was performed using PK‐15 and Vero cell lines. After 5 days of inoculation, no cytopathic effect was observed. A second blind passage allowed the observation of cytopathic effect on PK‐15 cells, while it remained absent in Vero cells. A fluorescence test using an anti‐N monoclonal antibody confirmed viral replication. One sample was processed for whole genome sequencing (WGS). In short, raw reads were imported into CLC genomics and assembled de novo. Out of 479k reads generated from the sample, 436k assembled into a 25,501 bp contig which was 99.5% identical to a reference porcine deltacoronavirus strain from the USA within the North American phylogroup. Yet, there are relevant differences at the nucleotide and amino acid levels compared with previously described porcine deltacoronavirus strains. Altogether, our findings represent the first report of porcine deltacoronavirus in South America, which provides information of its evolutionary origin. Thus, this study offers new insights into the molecular epidemiology of porcine deltacoronavirus infections in the swine industry.

Interestingly, it appears that the first two groups have originated from bats, whereas the latter two emerged from wild birds (Woo et al., 2012).
Porcine deltacoronavirus (PDCoV) is an emergent virus that causes gastrointestinal disease such as diarrhea, vomiting, dehydration and death in young piglets representing a major threat to swine industry Li et al., 2019;Zhang, 2016;Zhao et al., 2019).
Although PDCoV by itself causes enteric disease, co-infections with other coronaviruses such as porcine epidemic diarrhea virus (PEDV) and transmissible gastroenteritis virus (TGEV) or other viruses are commonly found (Ajayi et al., 2018;Mai et al., 2017;Marthaler et al., 2014a;Song et al., 2015). In this context, PDCoV shows indistinguishable clinical signs from other forms of enteric disease such as PED, TGE and swine acute diarrhea syndrome. Thus, proper differential diagnostic relies on genetic detection-based assays that offer a highly sensitive and specific method.

Samples and RNA extraction
In November 2019, a report of an enteric clinical disease outbreak in a farm located in San Martin, Peru (see Figure 1(c)) was made by a local veterinary professional to the National animal health authorities (SENASA). The report indicated high morbidity in young piglets. Intestinal samples were submitted by the farm professionals to the Laboratory of Virology in the Faculty of Veterinary Medicine at the Universidad Nacional Mayor de San Marcos in Lima, Peru, for testing and detection of PEDV, TGEV and PDCoV by qRT-PCR. Intestinal content (1 ml) was processed accordingly. In short, samples were diluted in 5 ml of phosphate buffer solution and centrifuged at 250 g for 5 min. Following centrifugation, 1 ml of supernatant was processed for RNA extraction using the QIAmp viral RNA mini kit (Qiagen), following manufacturer's specifications.

PCR differentiation and detection
Samples were assayed by qRT-PCR to differentiate three main enteric coronaviruses affecting pigs as PEDV, TGEV and PDCoV. Thus, we used the EZ-PED/TGE/PDCoV MPX 1.1 kit (Tetracore) following manufacturer's specifications.

Viral isolation
For viral isolation, we tested two different cell lines known to support coronavirus replication. We used PK-15 and Vero cells that were plated in 24-well plates at 90% of confluence. PK-15 cells were kindly pro-

Whole genome sequencing
Viral RNA was purified from one sample and processed for nextgeneration sequencing (Illumina, Inc), following manufacturer's (a)

(b) (c)
F I G U R E 1 PDCoV isolation in PK-15 cells and geographical map indicating locations where the outbreak was reported. Following PDCoV detection, intestinal content was filtered, and TPCK-trypsin treated for isolation in cell lines known to be permissible for PDCoV replication. Following 5 days of inoculation, no cytopathic effect (cpe) was evidenced under light microscopy in both cell lines. A second passage evidenced cpe after 24 h, with initial cell rounding and monolayer disruption in PK-15 cells but absent in Vero cells. Following 48 h of second passage, cpe was evidenced in almost 70% of cell monolayer, characterized by cell rounding, cell detaching, pyknosis in PK-15 cell lines but no cytopathic effect was detected in Vero cell lines (a). Moreover, an immunofluorescence test with a PDCoV nucleoprotein monoclonal antibody confirmed viral replication in PK-15 infected cells (b). A representative map of South America (blue) highlighting Peru (orange) is shown in the upper right side. Additionally, a geographical map of San Martin department (yellow) located in the north of Peru (light green) is represented below (c) specifications. Reads were imported into CLC genomics and assembled de novo. Whole genome (n = 44) and ORF 1a/b (n = 23) sequences of PDCoV were obtained from GenBank for phylogenetic analysis.
For analysis based on S gene, we also included those from Mexico (n = 61). General information of nucleotide sequences used is listed in Table 1.

Phylogenetic analysis
The PDCoV whole genome, ORF 1a/b and S nucleotide sequences in this study were aligned using Clustal W from MEGA X software (Kumar et. Et al., 2018). We used PDCoV genome sequences obtained from GenBank isolated in the USA (16), China (13), Japan (5), Haiti (3), South Korea (2) (2) and Laos (1). A general time reversible nucleotide substitution model with gamma distribution among site rate variation was used, with a maximum likelihood estimation model for phylogenetic reconstruction.
Bootstrap analysis was carried out on 1000 data sets. The percentage of nucleotide sequence identity was also calculated.

Molecular detection confirms the presence of PDCoV genome in intestinal samples of pigs suffering from an enteric disease outbreak in Peru
We took advantage of a widely used qRT-PCR assay that allows accurate differentiation of most common enteric coronaviruses in swine such as PEDV, TGEV and PDCoV. qRT-PCR results confirmed the presence of PDCoV RNA in all (n = 3) samples analyzed, which tested negative to PEDV and TGEV. In addition to its qualitative feature, this assay allows relative quantification. Thus, the viral load was quantified through the amount of genetic target of PDCoV amplified during the process. Samples had low Ct values to PDCoV (Ct = 9-14) which indicated a high PDCoV viral load. From these results, we inferred that PDCoV was involved in the enteric disease with high viral titres.

PDCoV replicated in PK-15 while Vero cells did not allow PDCoV propagation
Following PDCoV detection, our objective was to isolate in vitro using PK-15 and Vero cell lines. Despite the high viral load detected by PCR, we did not see evidence of cytopathic effect during the first 5 days following inoculation in any of the cell line tested. Thus, we performed a second blind passage to evidence viral replication. Interestingly, 48 h post inoculation, cytopathic effect was observed in PK-15 cell line, but morphological changes in Vero cell line were absent.
These findings were also confirmed by qRT-PCR. Within the major cel-lular changes observed are: pyknosis, cell rounding, monolayer disruption, cell detachment, which correspond to typical coronavirus cytopathic effect previously described (see Figure 1(a)). Furthermore, the immunofluorescence test in PK-15-infected cells confirmed the presence of PDCoV (see Figure 1(b)). Following isolation, a sample was selected and prepared for genome sequencing. Phylogenetic analysis has typically been performed using key major genes of any organism of interest. However, this analysis tends to limit the analysis to a certain gene or group genes. Conversely, whole genome sequencing offers a more complete and deeper genetic characterization compared with partial approaches. In our study, we took advantage of next generation sequencing of our PDCoV isolate to track its evolutionary origin. Our results indicated that our Peruvian strain belongs to the North American phylogroup and is closely related to a PDCoV strain from the USA isolated in 2015 (99.5% of nucleotide identity    Figures 3(a) and (b)). Similarly, PDCoV protein sequence analysis resembled the topology of the nucleotide analysis.

Phylogenetic analysis of Peruvian PDCoV S gene shows close relationship to the Mexican PDCoV within the North American phylogroup
S gene is one of the most variable genes among coronaviruses. This is due to its function in cell attachment and viral replication. This high polymorphism makes the S gene a powerful tool to estimate the evolutionary relationship among virus strains belonging to the same genetic group. Hence, we performed a phylogenetic analysis using the PDCoV S nucleotide sequences publicly available with our Peruvian S nucleotide sequence to evaluate their evolutionary distance (see Figure 4 Yet, S phylogenetic analysis evidence that Peruvian PDCoV has a certain degree of divergency from the North American phylogroup.

Peruvian PDCoV Spike amino acid sequence reveals unique substitutions compared to other PDCoV strains
As we observed multiple changes at the nucleotide level in the S gene, we were interested in evaluating whether these changes represent modifications at the protein level. Thus, we found multiple changes in the S protein sequence (see Figure 4(b)). Most relevant changes are K96R, G552E, A630V and V1052A that represent unique variations compared with other PDCoV strains. Other amino acid changes have been also found in Chinese strains such as P38L and A137V. We also detected F530L, like the one found in the Vietnamese strains. Furthermore, the Peruvian strains have a Q642K, like those in Vietnamese and Thai strains. These results provide evidence that PDCoV has undergone to unique changes that indicate a degree of genetic diversity in the Peruvian PDCoV strain. Based on previous studies of PDCoV Spike protein characterization (Shang et al., 2018), amino acid substitutions observed here are located randomly across both S protein subunits (S1 and S2) but none of them is located in the RBD region (S1-CTD).

DISCUSSION
PDCoV is one of the most recent and relevant coronaviruses of swine industry. It represents a major threat for swine productivity, and it is responsible for large economic losses worldwide. Yet, PDCoV remains poorly studied despite major efforts made recently. In Peru, multiple cases of enteric disease have occurred; however, these cases are not properly addressed and frequently misdiagnosed. Hence, this study represents the first report of isolation and phylogenetic characterization of a Peruvian PDCoV isolate using the whole genome sequence and its major S protein, revealing unique aspects compared to other PDCoV strains.
Although multiple studies have proven that ST, LLC-PK  and IPI-2I  are the primary option for PDCoV isolation, we showed that PK-15 might be a valid alternative when is needed. This is also demonstrated by Jiang et al. (2019), in the assessment of innate immune activation following PDCoV infection. Nevertheless, it is possible that efficacy in viral replication might vary among different cell lines and PK-15 might not be as permissive as others. This should be considered when viral infectivity is the main objective. In addition, since we detected viral replication in the presence of trypsin, we speculate that PK-15 cells might offer a compared PDCoV permissibility in LLC-PK while differing from that observed in ST. Furthermore, our viral isolation findings contrasted with those detected by qRT-PCR.
This provided evidence that a large proportion of viral particles were unable to replicate into a cell line support. This has also been reported by others indicating that low viral isolation rates might be attributed to sample degradation and viral viability . On the other hand, multiple authors have shown that successful viral isolation are due to other factors such as cell line permissibility and enzyme treatment (Jung et al., 2016;Yang et al., 2020;Zhao et al., 2019). In our study, it remains unclear whether trypsin concentration (10 µg/ml) might have played a role in the lack of viral replication in Vero cells. Additional studies, at different trypsin concentration, will elucidate Vero cell line permissibility to PDCoV. Interestingly, we did not observe cell toxicity in our assays, as that was absent following the first 5 days of inoculation and PDCoV cytopathic effect was evidenced in the second passage.
These results contrast with other claims that cell toxicity is common during PDCoV isolation using intestinal content or foecal samples . Nevertheless, further studies are required to clarify the implications of cell permeability to PDCoV in viral replication and its effects on clinical presentation. those PDCoV isolated in 2017. In our study, phylogenetic analysis using F I G U R E 3 Whole genome and ORF 1a/b phylogenetic analysis of Peruvian PDCoV strain reveals its evolutionary origin from a North American PDCoV strain. For both whole genome and ORF 1a/b, the evolutionary history was inferred by using the maximum likelihood method and general time reversible model. The percentage of trees in which the associated taxa clustered together is shown next to the branches. A discrete gamma distribution was used to model evolutionary rate differences among sites. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. The whole genome analysis involved 45 nucleotide sequences, whereas the ORF 1 a/b involved 24 nucleotide sequences. Evolutionary analyses were conducted in MEGA X

(a) (b)
F I G U R E 4 Phylogenetic analysis of S gene reveals that Peruvian PDCoV strain has a close relationship with a Mexican strain within the North American phylogroup. The evolutionary history was inferred by using the maximum likelihood method and general time reversible model. The percentage of trees in which the associated taxa clustered together is shown next to the branches. A discrete gamma distribution was used to model evolutionary rate differences among. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. This analysis involved 62 nucleotide sequences. Evolutionary analyses were conducted in MEGA X (a). Schematic representation of the porcine deltacoronavirus S protein highlighting the amino acid substitutions found in the present study. Black arrows represent the site of these substitutions. A multiple amino acid sequence alignment of S protein was performed using Clustal W in MEGA X. Eight amino acid modifications were detected within the Peruvian sequence compared to others (b) S gene nucleotide sequences reveals that Peruvian PDCoV grouped closely to the Mexican strain isolated in 2015 within the North American phylogroup. This indicates a close relationship among PDCoV strains from Mexico, the USA and Peru, sharing a common ancestor and evidencing a dissemination route of PDCOV from North America to South America. Interestingly, multiple non-silent mutations were found in the Peruvian PDCoV strain compared with other genomes.
Even though some of these mutations have been described in other PDCoV strains, some are unique to the Peruvian isolate, revealing that this isolate has undergone phenotypical changes after its emergence in North America. Although these substitutions were neither located in critical regions of glycosylation sites nor in the RBD region (S1-CTD), they might have influence in ligand/receptor interaction. Further studies are required to clarify whether these modifications have implications in the pathobiology and development of the clinical disease.
To date, it is unclear how PDCoV was introduced in Peru. However, there is a long history of commerce between Peru and North American countries that has expanded in recent years. The National Service of Animal Health in Peru (SENASA) reported the import of a large number of purebred animals (∼150 tons) during the 2014 and 2018 period. Furthermore, Peru imports feed ingredients for swine farms mainly from the USA (MINAGRI, 2020). Altogether, this might explain the possible routes for PDCoV entrance into the country, similar to that described for other viruses of importance for swine industry (Dee et al., 2018;Ramírez et al., 2019). Interestingly, there is no report of PDCoV in other South American countries so the introduction from those is unlikely.
Nevertheless, further studies are needed to understand the epidemiology of this disease in Peru and its relationship to other countries.
In conclusion, Peruvian PDCoV strain was successfully sequenced, isolated and phylogenetically analyzed demonstrating that this isolate has been derived from a strain identified in the USA. To our knowledge, this is the first report of a PDCoV strain detected in South America and offers new insights about the epidemiology of PDCoV worldwide.