The experimental procedure followed Institutional Animal Care and Use Committee regulations (CRONEX-IACUC 201810005).

We randomly sampled 50 JNP from the Jeju Livestock Promotion Institute (Jeju, Korea), 151 Yorkshire pigs from GGP farms, and 67 Berkshire pigs from Dasan Breeding Farm (Namwon, Korea). The genomic DNAs of the individuals were genotyped using an Illumina Porcine 60 K SNP Beadchip (Illumina, San Diego, CA, USA) following the standard protocol. We merged the three pig breeds’ data using vcftools (vcftools.sourceforge.net) and obtained 62,551 SNPs [7]. For the quality control, we excluded the SNPs with minor allele frequency (<0.05) and Hardy-Weinberg equilibrium (p<0.0001) and genotyping call rate (<0.05).

To understand the characteristics of the JNP, Berkshire and Yorkshire pig populations, structure analysis was performed with the structure program [8]. We set the 5,000 iterations after the burn-in 5,000 iterations and K = 3, 4, and 5. We evaluated the JNP’s purity and structure analysis showed that the JNP was mixed with Berkshire in part. Thus, we eliminated impure JNP individuals for the next analysis.

The correlation coefficient of the linkage disequilibrium (r²) was computed to assess the genetic diversity of the JNP, Berkshire and Yorkshire pigs. Linkage disequilibrium is the nonrandom associations of alleles at two loci, A and B. Linkage disequilibrium is usually used to assess the genetic diversity of a given population, estimation of effective population size and population genetic stability. Using the R package “LDcorSV”, we examined all pairwise r² decays with 100 SNP bins along the physical distance [9]. The formula based on this is as follows:

where DAB2 is the linkage disequilibrium of two alleles A, B and P_X is the frequency of major allele X and P_X is the frequency of minor allele x.

We used the population differentiation statistic (F_st) to find selection signatures of the JNP breeds. The reference pig breeds were Berkshire and Yorkshire. Berkshire and Yorkshire pigs are worldwide commercial pig breeds that we felt provided an adequate comparison to the JNPs. Among various population differentiation statistics, the statistic was the Weir-Cockerham F_st in vcftools [10–12]. The Weir-Cockerham F_st formula is as follows:

Where s² is the sample variance of the allele frequency of A over subpopulations and p̄ is the average of the allele frequency of A in the total populations of the biallelic system [11,13].

Among the 62,551 autosomal SNPs genotyped in this analysis, 46,505 in JNP-Berkshire and 44,306 in JNP-Yorkshire remained after quality control. After filtering and imputation, the number of SNPs per autosome ranged from 1,085 to 6,000 in the JNP-Berkshire data and from 1,030 to 5,864 in the JNP-Yorkshire data, and this value was closely related to the chromosome length and total number of SNPs, as shown in Supplementary Figure S1. The remaining minor allele frequency of SNPs exhibited a uniform distribution, with an average of 0.42±0.22 (standard deviation [SD]). The mean distance between adjacent SNP pairs from this analysis was 50,134±206,871 (SD).

We evaluated the structure distinctness of the JNP, Berkshire and Yorkshire pigs, and the linkage disequilibrium decays were examined to assess the genetic diversity. We used the R package “LDcorSV” to examine the linkage disequilibrium. All pairwise linkage disequilibriums were calculated within adjacent 100 SNPs. Figure 1 shows r² decay with the physical distance in the JNP, Yorkshire and Berkshire pigs. According to Figure 1, the genetic diversity of the JNP were considerably lower than that of the Yorkshire and Berkshire pigs. Additionally, this lower genetic diversity reflects that the number of JNPs has decreased since South Korea has been modernized after the 20th century. We analyzed the population differentiation to examine the population structure of the JNP, Berkshire and Yorkshire pig breeds. Figure 2 shows the result of structure analysis after pre-structure analysis was used to identify and eliminate the impurity of the JNP and other pigs using the case K = 3.

To detect and characterize of the JNP’s selection signatures, we used the population differentiation statistic, F_st in which the selection signature between two compared breeds can be revealed. F_st is the widely used population differentiation statistic in selective sweep analysis. Practically, we used the Weir-Cockerham’s F_st statistic. The mean and standard deviation in F_st analysis of JNP vs Berkshire and JNP vs Yorkshire were 0.27±0.25 and 0.26±0.25, respectively. Figure 3 shows the distribution of F_st of the JNP, Berkshire and Yorkshire pigs. The frequencies of the F_st values decreased with increasing F_st values, and the gray line represents the cutoff F_st (top 5% genes) for gene ontology (GO) analysis. Supplementary file 1 consists of the significant gene’s SNP list and its allele frequencies. Additionally, this file shows the discrepancy of the allele frequency of JNP, Berkshire and JNP, Yorkshire pigs.

The high F_st value (F_st =1) genes in the JNP vs Berkshire were tight junction protein 1 (TJP1), 4-aminobutyrate aminotransferase (ABAT), matrix metallopeptidase 25 (MMP25), methionine sulfoxide reductase B1 (MSRB1), osteoglycin (OGN), plexin C1, and ubiquitin protein ligase E3B (UBE3B). The high F_st value genes in the JNP vs Yorkshire were interferon induced protein 44 and its paralog (IFI44 and IFI44L), DEAD-box helicase 59 (DDX59), cathepsin O, and prostaglandin E receptor 3 (PTGER3).

We performed gene matching with the high F_st SNPs. After gene matching of SNPs using ensemble gene catalogue (www.ensembl.org), the top 5% of genes were used in DAVID GO analysis (The Database for Annotation, Visualization and Integrated Discovery v6.8; https://david.ncifcrf.gov/). The DAVID analysis includes three types of classification (biological process, molecular function, and cellular component). Practically, we used the biological process to query the selective sweep genes’ GO. For the selective sweep genes’ GO, the multiple correction and p-value approach (cutoff p-value 0.01 or 0.05) was not adequate in our analysis because it was very harsh. Table 1 and 2 show the gene ontologies (GO) of the JNP vs Berkshire and JNP vs Yorkshire pigs, respectively. For the JNP vs Berkshire pigs, epithelium development and neurogenesis were enriched and for the JNP vs Yorkshire pigs, cellular response to oxygen-containing compound and neurogenesis were the most enriched GOs.

There are various ways to find selection signatures in two populations such as XP-EHH, XP-CLR, F_st, etc. These methods can be exploited by comparing two populations. XP-EHH and XP-CLR scan the genomic regions and find the selective sweep regions. XP-EHH uses the haplotype homozygosity and XP-CLR uses the population differentiation between the two object and reference population [14,15]. These methods have the advantage not computing the scores of the variants of selective sweep regions but rather comprehensively scanning the genome-wide regions. However, in the current cost restriction, sample size is a primary problem. Therefore, we instead used the F_st statistic which could identify the population differentiation at the direct variant level. In this study, we used SNPs data from a porcine 60K HD chip.

The significant genes in the selective sweep analysis differed considerably from Kim et al [5]. Kim et al [5] used whole genome sequencing data, whereas the SNP chip data were used in our analysis. Due to the long distance between SNPs in our dataset, we utilized another statistic, F_st which assesses the population differentiation at the right position of the SNPs and has the advantage to being able to find the selection signature of the SNPs based on the fixation index between two populations.

The major terms in the GO analysis were “GO:0048729~ tissue morphogenesis”, “GO:0060429~epithelium development”, and “GO:0048589~developmental growth” in the JNP vs Berkshire and “GO:1901701~cellular response to oxygen-containing compound” and “GO:0071396~cellular response to lipid”, and “GO:0022008~neurogenesis” in the JNP vs Yorkshire. The overlapped genes in JNP with Berkshire and Yorkshire pigs were 57. These kinds of genes seem to indicate the JNP’s specific genes with selective sweeps. It is of interest that some genes of the “GO:0022008~neurogenesis”, i.e., zinc finger protein 536 (ZNF536), FRY like transcription coactivator (FRYL), protein tyrosine kinase 7 (PTK7), protein kinase CGMP-dependent 1 (PRKG1), patched 1 (PTCH1), intraflagellar transport 140 (IFT140), was overlapped in JNP vs Berkshire and JNP vs Yorkshire cases. ZNF536 protein is a highly conserved zinc finger protein and is most abundant in the brain, where it negatively regulates neuronal differentiation. The FRYL plays a key role in maintaining the integrity of polarized cell extensions during morphogenesis. The PTK7 gene encodes the proteins of a member of the receptor tyrosine kinase family and is involved in the Wnt signaling pathway. It plays a role in multiple cellular processes including polarity and adhesion. PRKG1 isoforms play as key mediators of the nitric oxide/cGMP signaling pathway. PTCH1 encodes a member of the patched family of proteins and is a component of the hedgehog signaling pathway. Hedgehog signaling is important in embryonic development and tumorigenesis. The encoded protein is the receptor for the secreted hedgehog ligands, which include sonic hedgehog, Indian hedgehog and desert hedgehog. Following binding by one of the hedgehog ligands, the encoded protein is trafficked away from the primary cilium, relieving inhibition of the smoothed G-protein-coupled receptor, which results in activation of downstream signaling. Mutations of this gene have been associated with basal cell nevus syndrome and holoprosencephaly. IFT140 gene encodes one of the subunits of the IFT complex and intraflagellar transport is involved in genesis, resorption and signaling of primary cilia.

The JNP exhibits slower growth, the fat is thicker than other pigs and the litters are smaller. According to the Kyoto encyclopedia of genes and genomes (KEGG), mechanistic target of rapamycin signaling pathway was reported to regulate adiposity [16]. Additionally, the regulation of lipolysis in adipocytes is related to adipose accumulation. Wnt family member 7B (WNT7B; GO:1901701~cellular response to oxygen-containing compound in JNP vs Yorkshire pigs, ssc05205: proteoglycans in cancer in JNP vs Yorkshire pigs) is required for epithelial progenitor growth and is involved in the pancreatic development [17]. Estrogen receptor 1 (ESR1; GO:1901701~ cellular response to oxygen-containing compound in JNP vs Yorkshire pigs, ssc04917: prolactin signaling in KEGG pathway of JNP vs Berkshire pigs) is reported to be associated with litter size in the Chinese-European pig line [18]. Transforming growth factor beta-2 (TGF-beta2: GO:0022008~ neurogenesis in JNP vs Berkshire pigs) enhances connective tissue formation and wound strength in the guinea pig [19]. AKT serine/threonine kinase 3 (AKT3; ubiquitous in KEGG pathway of JNP vs Berkshire pigs) is related to reduced proliferation and facilitated differentiation of myoblasts in skeletal muscle development, whereas phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA) induces multipotency [20] (Tables 3, 4).

TJP1 encodes a member of the membrane-associated guanylate kinase family of proteins, and it acts as a tight junction adaptor protein. Tight junctions play a role in regulating the movement of ions and macromolecules between endothelial and epithelial cells. ABAT is responsible for catabolism of gamma-aminobutyric acid, an inhibitory neurotransmitter in the central nervous system. MMP25 is involved in the breakdown of the extracellular matrix in normal physiological processes, such as embryonic development, tissue remodeling, and reproduction as well as in disease processes, such as arthritis and metastasis. MSRB1 is a member of the MSRB family functions to repair enzymes that protect proteins from oxidative stress by catalyzing the reduction of methionine-R-sulfoxides to methionines. This protein is highly expressed in the liver and kidney and it also has the highest methionine-R-sulfoxide reductase activity. OGN encodes protein that induces ectopic bone formation and may regulate osteoblast differentiation. High expression of the protein may be associated with elevated heart left ventricular mass. Plexins are transmembrane receptors for semaphorins, a large family of proteins that regulate axon guidance, cell motility and migration, and the immune response. Modification of protein with ubiquitin is an important cellular mechanism for targeting abnormal or short-lived proteins for degradation. UBE3B encoded protein are involved in ubiquitination. IFI44 and IFI44L are the interferon induced protein 44 and its paralog. Diseases associated with DDX59 include orofaciodigital syndrome V and orofaciodigital syndrome. The GO annotations related to this gene include nucleic acid binding and helicase activity. Cathepsin O enzyme is involved in cellular protein degradation and turnover. PTGER3 protein is a member of the G-protein coupled receptor family. This protein is one of four receptors identified for prostaglandin E2 and this receptor may have many biological functions, which involve digestion, the nervous system, uterine contraction activities and kidney reabsorption (www.genecards.org).

Kim et al [5] used XP-EHH and XP-CLR to identify the selection signatures of KNBP including JNP. There were overlapped genes such as hydroxysteroid 17-beta dehydrogense 12 (HSD17B12), CUB and sushi multiple domains 3, BTB domain containing 11 (BTBD11), ATP binding cassette subfafmily B member 11, transforming growth factor beta 2 (TGFB2), otoancorin, solute carrier family 4 member 10, thyroid hormone receptor beta, and WNT7B between our study and Kim et al’s study. HSD17B12 gene product converts into estradiol in ovarian tissue and is involved in fatty acid elongation. JNP’s unique meat quality can be implicated in the fatty acid. Specifically, BTBD11 was identified in every four cases of JNP vs Berkshire (with XP-EHH and XP-CLR of Kim et al [5]) and JNP vs Yorkshire (with XP-EHH and XP-CLR of Kim et al [5]).