**3. Discussion**

The main characteristics of the LOM strain are attenuated; however, inoculation into pregnant sows lacking anti-CSF antibodies triggers abortion [7]. In addition, LOM vaccination of immunocompetent pigs leads to an antigen release period of more than 4 weeks, which results in the worsening of lesions if an animal is co-infected with other pathogens [8]. A previous study tested the effect of a live attenuated BVDV vaccine on pig-to-pig transmission and found that the R value was <1 [10]. Here, we found that transmission of the Jeju LOM strain differed significantly depending on the health status of the pig (Group 1 (unhealthy): r = 1.22; and Group 2 (healthy): r = 0). This is because, as for other wasting diseases, the period of virus release from infected pigs is prolonged and the amount of virus shed in the feces increases [8]. A commercial LOM vaccine strain showed 10% of the average transmission rate from inoculated pigs to cohabitating non-inoculated pigs [11]. Although a small number of pigs were used for the transmission experiment in this study, the transmission characteristics of the Jeju LOM strain were revealed to be similar to the commercial LOM vaccine. In a previous study, after oral inoculation of the LOM vaccine strain of 2 <sup>×</sup> <sup>10</sup>4.0TCID50/mL, the LOM vaccine virus was detected in blood and feces in the 9 days post-inoculation [9]. We also detected 11 samples of live Jeju LOM viruses with more than 104.0TCID50/mL in the slaughterhouse experiment, which support the possibility of farm-slaughterhouse-farm transmission.

From 2004 to 2005, LOM-infected pigs in Jeju showed varying symptoms. The clinical symptoms showed both the similarities and differences from those associated with CSF [10]. Similarities to more current Jeju LOM strain infections (2014–2018) included skin ulcers, ulceration of the tonsil, petechial hemorrhage in the kidney, button ulceration of the ileocecal region, infarction of the spleen, and pneumonia. Differences in histologic findings were a weakness in endothelial cell lesions of organs such as the spleen, lymph nodes, and brain. These lesions seem to be more related with PRRS not CSF. In addition, meningitis was observed in the central nervous system, CSF did not show meningitis, and lymphoid organs showed little increase in volume other than lymphocytic atrophy. The Jeju LOM strain that was prevalent from 2004 to 2007 was not related to CSF pathogenicity [10]. IHC (2004–2005) was Jeju LOM virus positive in 52.1% (25/48) of suckling piglets and 28.6% (6/21) of weaned pigs [10]. Our Jeju LOM antigen results (2014–2018) were lower than the results of the Jeju LOM antigen (2004–2005) test (58.3%) from a previous report [10]. Jeju LOM strains (2014–2018) were also suspected to have caused CSF-like lesions in 7 young piglets (average age of 10 days), which was proved by histopathological findings and IHC analysis. It may be deduced that some CSF lesions in suckling piglets, as well as abortion during pregnancy, are associated with LOM strains. However, in this case, it is unlikely that persistent infection will continue to circulate the virus among pigs.

Some experts suggest that Jeju LOM strains detected on Jeju pig farms from 2014 to 2018 have recovered pathogenicity and that it is now different from that of the commercial LOM vaccine strain on the current market. However, gene analysis of Jeju LOM strains revealed no evidence for restoration of pathogenicity. Previous studies have identified differences in pathogenic regions between the ALD and GPE strains as E2 (T830A), NS4B (V2475A, A2563V), and Npro (N136D) [12,13]. Other studies revealed that Shimen's pathogenicity-related gene locus is within E2 (T745I, M979K) [14] and E2 (N850S) [15]. Our analysis of the amino acid sequences of the commercial LOM vaccine strain and the Jeju LOM strains detected in Jeju pigs revealed no changes at these sites. In addition, although there are differences in pathogenicity linked to changes in N-linked glycosylation sites on the E2 protein of CSF, there was no change of N-linked glycosylation sites in the Jeju LOM strains [16]. A study comparing the complete nucleotide and amino acid sequences of the ALD and GPE strains revealed 98.2% and 98.8% identity, respectively [17]. Although the Jeju LOM strain (2014–2018) harbors several amino acid mutations, it is difficult to interpret this as conferring recovery of pathogenicity or additional attenuation. There was no significant difference in the gene mutation rate among Jeju LOM strains, nor was there a significant difference in omega values. Therefore, Jeju LOM strains (2014–2018) from Jeju cannot be considered a pathogenically-reversed LOM strain, which is virulent in pigs but shows a safety profile characteristic of the commercial LOM vaccine strain described in previous study [7]. More precisely, the presence of Jeju LOM strains with recovered pathogenicity should be evaluated in animals (i.e., pregnancy sows and SPF pigs).

In conclusion, the CSF-like histopathogenic lesions of Jeju pigs revealed to be more related to other viral pathogens rather than the Jeju LOM strains (2014–2018), despite the presence of the Jeju LOM strain in organs of piglets. We also confirmed that the pig-to-pig transmission of the Jeju LOM strain and farm-to-farm transmission may have been caused by vehicles visiting the slaughterhouse. Although we found some genetic differences between the Jeju LOM strains (2014–2018) and commercial LOM vaccine strains, more pathogenesis studies may be needed using animals such as pregnant sows and SPF pigs.

#### **4. Materials and Methods**

#### *4.1. Detection of Histopathologic Lesions and Immunohistochemical Straining Analysis*

From 2014 to 2018, tests in this study performed by the diagnostic services of a veterinary medicine college in Jeju national university to detect animal infectious disease identified 122 samples (100 piglets and 22 fetuses) from 25 pig farms that were positive for CSFV by RT-PCR, histopathology, and IHC. The amplification targets for RT-PCR were CSFV [9], porcine circovirus type 2 (PCV2) [18], porcine reproductive and respiratory syndrome virus (PRRSV), swine influenza virus (SIV), cytomegalovirus (PCMV), *Salmonella* spp, *Streptococcus* spp, *Actinobacillus, pleuropneumoniae*, and *Haemophilus parasuis.* Histopathologic lesions specific to CSFV were examined by staining tissues with hematoxylin and eosin. The following were examined: skin (hemorrhage and endothelial damage), lymph nodes (peripheral hemorrhage, lymphoid depletion, and reticular cell hyperplasia), kidney and bladder (hemorrhage, swelling and peripheral hemorrhage), spleen (hemorrhagic infarction, endothelial damage), cecum and colon (necrotic or ulcerative colitis, vascular congestion, and sub-serous hemorrhage), heart (myocardial hemorrhage), and brain and spinal cord (nonsuppurative encephalitis, proliferation of endothelial cells, perivascular cuffing, microgliosis, and focal necrosis). To confirm the presence of CSF antigen in tissues, IHC was performed using the EnVisionTM peroxidase-conjugated polymer reagent (DAKO, Denmark). In brief, tissues were reacted with a primary antibody (mouse anti-CSFV) (WH303, Animal and Plant Health Agency, New Haw, Addlestone, UK), followed by EnVisionTM/HRP, rabbit/mouse (EVN) regent (DAKO, USA), and 3,3'-diamino-benzidine tetrahydrochloride (DAB). The presence of CSF antigen is denoted by a dark brown deposit in the tissue section.

#### *4.2. Detection of Anti-CSF Antibodies*

Seven pig farms with the Jeju LOM infection were selected to estimate how Jeju LOM strains were introduced into the farms and how they spread within the farms. The anti-CSF antibody-positive rates for each pig farm and antibody titer levels for pigs within farms were examined according to age. Blood samples were collected from 20 pigs in each group: sows, suckling pigs (10–20 days old), weaning pigs (40–60 days old), growing pigs (90–120 days old), and finishing pigs (150–180 days old). To detect CSF-specific neutralizing antibodies, a neutralizing peroxidase-linked assay (NPLA) was performed according to the standards manual of the World Organization for Animal Health [19]. For PK-15 cell staining, the monoclonal antibody 3B6 (Median Diagnostics, Chuncheon, Korea) was used to detect the CSF E2 protein.

#### *4.3. Environmental Samples Taken from a Slaughterhouse*

In 2017, 242 samples were obtained from a slaughterhouse in Jeju to investigate CSFV (Jeju LOM strain). These included 71 samples from the driver's foot floor, 71 from vehicle wheels, 74 from the vehicle's pig-holding compartment, and 26 from other sites. Real-time quantitative PCR (qRT-PCR) was performed to detect the CSF antigen copy number in fecal and environmental samples. The VDx® CSFV qRT-PCR (MEDIAN Diagnostic Co. Cat No. NS-CSF-31, Gangwon-do, Korea), which uses TaqMan probes, detects the CSFV 5' UTR with high specificity; it does not detect BVDV or border disease virus, which also belong to the *Pestivirus* genus. Briefly, the qRT-PCR program comprised the following steps: cDNA synthesis (50 ◦C, 30 min) and initial inactivation (95 ◦C, 15 min), followed by a two-step PCR comprising 42 cycles of denaturation (95 ◦C, 10 s) and extension (60 ◦C, 60 s). A peroxidase-linked immunosorbent assay (PLA) was used to confirm viability of Jeju LOM strains. Briefly, PK-15 cells (grown to 80% confluence in 24-well plates) were inoculated for 72 h at 37 ◦C with 10% tissue homogenates including the Jeju LOM strain in a minimum essential medium. The PK-15 cells were fixed in pre-chilled 80% acetone after 72 h and reacted with a 3B6 monoclonal antibody specific for CSFV E2. Subsequently, the PK-15 cells were reacted with biotinylated anti-mouse IgG (H + L) (VECTOR Laboratories, Cat No. BA-9200, Burlingame, CA, USA) and ABC solution (VECTOR Laboratories, Cat No. PK-4000, Burlingame, CA, USA). After staining with DAB

peroxidase substrate (VECTOR Laboratories, Cat No. SK-4100, Burlingame, CA, USA) following the manufacturer's instruction, the PK-15 cells were observed under a microscope.

#### *4.4. Horizontal Transmission between Pigs*

To investigate the transmission probability of the Jeju LOM strain among pigs, 40-day-old CSF antigen- and antibody-negative pigs from two pig farms (n = 24, 12 per farm) were used. Group 1 comprised of pigs with a wasting disease (PRRSV or PCV2) and Group 2 comprised of pigs with specific non-disease. Six pigs from each group were inoculated intramuscularly with 2 mL (104.0TCID50/mL) of the Jeju LOM strain JJ16LOM-YJK08, which is a representative strain circulating in Jeju between 2014 and 2018. After 24 h, the six pigs from each group were housed with non-injected pigs and observed for clinical signs and symptoms. Blood was collected every week for 45 days and tested in the NPLA assay and by qRT-PCR. The estimated reproduction number (R0) and confidence interval (CI) were calculated as described previously [20] using the following formula: R0 = −In((1−AR)/S0)/(AR−(1−S0)); CI <sup>=</sup> AR <sup>±</sup> 1.96 ARx(1 − AR)/n).

#### *4.5. Genomic Analysis of Commercial LOM Vaccine Strains and Jeju LOM Strains*

The complete genome sequences of seven commercial LOM vaccine strains (16LOM-GC00, 16LOM-JY00, 16LOM-KM00 and 16LOM-KR00 (collected from a market in 2016), 02LOM-JY00 (collected from a market in 2002), 88LOM-suri (isolated from LOM-850 strain in 1988), and LOM-850 (original master seed in 1987)) were examined. In addition, five Jeju LOM strains isolated from pigs on Jeju Island between 2004 and 2007, and 16 Jeju LOM strains isolated from pigs on Jeju Island between 2014 and 2018, were analyzed. Positive selection analysis of the complete ORF (Npro, C, Erns, E1, E2, P7, NS2, NS3, NS4A, NS4B, NS5A, and NS5B) was conducted using several models available in the BASEML and CODEML modules of the PAML 4.6 software package [21]. Different values of the non-synonymous/synonymous (*dN*/*dS*) ratio (the omega parameter), were considered in accordance with the user manual. A *dN*/*dS* ratio of <1 indicates a purifying selection, a *dN*/*dS* ratio = 1 suggests an absence of selection (i.e., neutral evolution), and *dN*/*dS* >1 indicates a positive selection. The Bayes empirical Bayes (BEB) calculation of the posterior probabilities of site classes was used to calculate the probability that a site is under positive selection pressure [22]. TempEst (formerly known as "Path-O-gen") is a tool for investigating the temporal signals and "clocklikeness" of molecular phylogenies. The contemporaneous trees (in which all sequences were collected at the same time) and dated-tip trees (in which sequences are collected at different dates) were analyzed using the TempEst v1.5.1 program, which is temporal signal estimator tool [23]. The genome sequences of the Jeju LOM strains (n = 21) and commercial LOM vaccine strains (n = 8) in this study were submitted to the GenBank under accession numbers MN558862–MN558889.
