Large-Scale Polymorphism Analysis of Dog Leukocyte Antigen Class I and Class II Genes (DLA-88, DLA-12/88L and DLA-DRB1) and Comparison of the Haplotype Diversity between Breeds in Japan

Polymorphisms of canine leukocyte antigen (DLA) class I (DLA-88 and DLA-12/88L) and class II (DLA-DRB1) genes are important for disease susceptibility studies, but information on the genetic diversity among dog breeds is still lacking. To better elucidate the polymorphism and genetic diversity between breeds, we genotyped DLA-88, DLA-12/88L, and DLA-DRB1 loci using 829 dogs of 59 breeds in Japan. Genotyping by Sanger sequencing identified 89, 43, and 61 alleles in DLA-88, DLA-12/88L, and DLA-DRB1 loci, respectively, and a total of 131 DLA-88–DLA-12/88L–DLA-DRB1 haplotypes (88-12/88L-DRB1) were detected more than once. Of the 829 dogs, 198 were homozygotes for one of the 52 different 88-12/88L-DRB1 haplotypes (homozygosity rate: 23.8%). Statistical modeling suggests that 90% of the DLA homozygotes or heterozygotes with one or other of the 52 different 88-12/88L-DRB1 haplotypes within somatic stem cell lines would benefit graft outcome after 88-12/88L-DRB1-matched transplantation. As previously reported for DLA class II haplotypes, the diversity of 88-12/88L-DRB1 haplotypes varied remarkably between breeds but was relatively conserved within most breeds. Therefore, the genetic characteristics of high DLA homozygosity rate and poor DLA diversity within a breed are useful for transplantation therapy, but they may affect biological fitness as homozygosity progresses.


Introduction
The major histocompatibility complex (MHC) molecules play important roles in inducing acquired immunity by presenting peptides derived from foreign antigens, such as germs and viruses that T cells recognize as non-self, resulting in the elimination of these antigens. The MHC molecules are classified into class I (MHC-I) and class II (MHC-II), and regulate self-and non-self discrimination in immunity by presenting antigen peptides to CD8 + and CD4 + T cells, respectively [1,2]. The MHC genes encoding MHC-I and MHC-II molecules are composed of multigene families; each of them is extremely polymorphic in many animals. For example, so far, more than 34,000 human leukocyte antigen (HLA) alleles have been identified and reported in the IPD-IMGT database (https: //www.ebi.ac.uk/ipd/imgt/hla/ (accessed on 28 November 2022)). In addition, specific HLA alleles associated with susceptibility or resistance to various diseases [3][4][5] and  Table S1.
In this study, to characterize the intra-and inter-breed DLA diversity, including both DLA-I and DLA-II genes in various breeds, we developed a new genotyping method to separate DLA-12 from DLA-88L accurately and performed polymorphism analysis of the relatively polymorphic DLA-I genes, DLA-88 and DLA-12/88L, and the most polymorphic DLA-II gene, DLA-DRB1, using 829 dogs of 59 breeds that were collected in Japan. We also estimated three-locus DLA-88-DLA-12/88L-DLA-DRB1 haplotypes (88-12/88L-DRB1) from the detected allele information and evaluated the genetic diversity within and between breeds based on the three-locus haplotype frequency. Furthermore, since the DLA- (A) Two different haplotype structures associated with DLA-12 and DLA-88L locus. Interlocus gene conversion events in cis or trans between DLA-88 and DLA-12 are likely to be responsible for generating the DLA-88L locus. (B) Flow chart for locus-specific DLA typing method for DLA-88, DLA-12, and DLA-88L genes with genomic DNA. White and black-filled boxes in the chart represent the schematic location of UTR and CDS regions of a DLA gene. Arrowheads indicate the location and direction of the primers with the primer names whose detailed information is described in Supplementary Table S1.
In this study, to characterize the intra-and inter-breed DLA diversity, including both DLA-I and DLA-II genes in various breeds, we developed a new genotyping method to separate DLA-12 from DLA-88L accurately and performed polymorphism analysis of the relatively polymorphic DLA-I genes, DLA-88 and DLA-12/88L, and the most polymorphic DLA-II gene, DLA-DRB1, using 829 dogs of 59 breeds that were collected in Japan. We also estimated three-locus DLA-88-DLA-12/88L-DLA-DRB1 haplotypes (88-12/88L-DRB1) from the detected allele information and evaluated the genetic diversity within and between breeds based on the three-locus haplotype frequency. Furthermore, since the DLA-88, DLA-12/88L, and DLA-DRB1 genes have the characteristics of classical MHC genes such as HLA-A, HLA-B, and HLA-DRB1, their DLA polymorphisms are thought to play important roles for the allo-recognition mechanism during transplantation. Hence, to evaluate the possibility of 88-12/88L-DRB1-matched transplantation using somatic stem cells in the field of veterinary medicine, we simulated statistically the proportion of recipient dogs that

RNA and DNA Samples
Peripheral blood samples from 829 dogs of 59 breeds were collected from the Animal Medical Center (ANMEC) at Nihon University, Marble Veterinary Medical center, and the Nippon Veterinary and Life Science University in accordance with the guidelines for animal experiments specific to each location when the dog owner approved to use the blood for research. Of these, 403 were genotyped initially using RNA samples extracted in the previous study [18,30], and the remaining 426 were genotyped using newly extracted genomic DNA samples (Table 1). We initially genotyped RNA samples (converted to cDNA for amplification) because transcribed MHC genes and alleles are detected more easily without contamination of amplicons originating from pseudogenes or duplicated genes if primer locations crossover to at least two homologous locations. Limitations of the RNA-based genotyping method [15,18] were corrected by also genotyping genomic DNA samples.

PCR Amplification of DLA-88 and DLA-12/88L Genes
Polymorphism analysis for DLA-88 and DLA-12/88L was performed using the genomic DNA from 426 dogs obtained for this study and 403 dogs that already were genotyped in our previous study [18].

PCR Amplification of DLA-DRB1 Gene
Polymorphism analysis for DLA-DRB1 was performed using RNA samples from the 403 dogs as templates by using DLA-DRB1 specific primer sets (DRB1-F and DRB1-R) (Supplementary Table S1D) [31]. The cDNA samples were synthesized with the oligo-dT primer using the RevaTra Ace reverse transcriptase reaction (TOYOBO, Osaka, Japan) after DNase I treatment using 1 µg of RNA (Invitrogen/Life Technologies/Thermo Fisher Scientific, Carlsbad, CA, USA) according to the manufacturer's protocol. The polymorphism analysis was also performed using DNA samples from the 426 dogs as templates by using DLA-DRB1 specific primer sets (DRB1-g-F and DRB1-g-R) (Supplementary Table S1E) [32]. The composition of the PCR solution consisted of 20 ng of cDNA or genomic DNA, 0.4 units of KOD FX DNA polymerase, 10 uL of 2× PCR buffer, 2 mM of dNTP, and 0.4 uM of each primer in 20 uL. The cycling parameter was as follows: an initial denaturation with 94 • C/1 min followed by 33 cycles of 98 • C/10 s, 60 • C/30 s, and 68 • C/45 s.

Sanger-Sequencing
After purification of PCR products using ExoSAP-IT (GE Healthcare, Piscataway, NJ, USA), the purified PCR products were sequenced directly with Big Dye Terminator Kit Ver. 1.1 or Ver. 3.1 (Life Technologies/Thermo Fisher Scientific, Carlsbad, CA, USA) and ABI3130 genetic analyzer (Life Technologies/Thermo Fisher Scientific, Carlsbad, CA, USA). The nucleotide sequences of the PCR products for DLA-88, DLA-12, and DLA-88L were determined using sequencing primers i1F-T and i3R-T (Supplementary Table S1C). When the DLA-88 allele sequences were difficult to determine due to sequence offsets by nucleotide insertions and deletions in intron 2 and/or exon 3 of the DLA-88 alleles [30,33], additional sequencing was performed using another primer i2F2 ( Figure 1B).

Allele Assignment and Confirmation of Novel DLA Alleles
DLA allelic sequences were assigned using Sequencher Ver. 5.0.1 DNA sequence assembly software (Gene Code Co., Ann Arbor, MI, USA) by comparing them with known DLA-88, DLA-88L, DLA-12, and DLA-DRB1 allele sequences released in the GenBank (https://www.ncbi.nlm.nih.gov/genbank/ (accessed on 26 April 2022)) and the IPD-MHC Canines database (https://www.ebi.ac.uk/ipd/mhc/group/DLA/ (accessed on 26 April 2022)). Allele sequences from the Sanger sequencing data also were assigned using the . New alleles were confirmed by PCR and direct sequencing again. The PCR products were cloned into the pTA2 cloning vector with the TA cloning kit (TOYOBO, Osaka, Japan), and the nucleotide sequence in 4 to 8 clones per DNA sample was analyzed to avoid PCR and sequencing artifacts.

Nomenclature of Novel DLA Alleles
We defined the alleles amplified by using the DLA-88 and DLA-88L specific primer set as the DLA-88L allele and the alleles amplified by using the DLA-12 specific primer set as DLA-12 allele in the 2nd PCR of DLA-12/88L. Since all the DLA-88L alleles reported so far in the IPD-MHC Canines database have been named "DLA-88*", we also followed the rule for the identified DLA-88L novel alleles as well as all published DLA-88 alleles. The official name of the novel allele was assigned according to the DLA nomenclature in the IPD-MHC database. Novel DLA alleles that have not been given an official allele name were named "DLA-88*nov", "DLA-12*nov", or "DLA-DRB1*nov" as tentative allele names. The 88-12/88L-DRB1 haplotypes for each dog, which are 88-12-DRB1 or 88-88L-DRB1, were identified and estimated manually based on genotyping data of 829 dogs as previously described [18,20]. We initially identified the 88-12/88L-DRB1 haplotypes for homozygous dogs and estimated the haplotypes for heterozygous dogs on the basis of those within the homozygous dogs. To confirm our manual haplotype estimation, we also estimated the 88-12/88L-DRB1 haplotypes in each breed by using the maximum likelihood method of the PHASE program [34].

Data Analysis
Calculation of expected heterozygosity (He), Hardy-Weinberg equilibrium (HWE) test, and principal component analysis (PCA) based on the 88-12/88L-DRB1 haplotype frequencies in each breed were performed by GenAlEx Ver. 6.5 [35]. In the PCA, to reduce the 88-12/88L-DRB1 haplotype numbers as explanatory variables, we used the haplotype frequencies composed of the field-1 level alleles. This level reflects differences in immunoresponsiveness due to changes in the amino acid sequences of the peptide-binding region and T-cell recognition region of each DLA allele [33]. The inbreeding coefficient (Fis) and haplotype richness (Hr) were calculated by FSTAT Ver. 2.9.4 (available from https: //www2.unil.ch/popgen/softwares/fstat.htm (accessed on 21 June 2021). In this case, the significant deviation of FIS from zero was also tested by FSTAT Ver. 2.9.4. Pearson productmoment correlation coefficient was calculated with R ver. 3.6.3 (available from https:// www.r-project.org/ (accessed on 17 March 2020)) to evaluate differences in DLA-DRB1 allele diversity in the same dog breeds from different countries, UK and Japan. A phylogenetic tree was constructed by the Neighbor Joining method and assessed using 10,000 bootstrap replicates after aligning the DLA sequences using the MEGA X software (available from https://www.megasoftware.net/ (accessed on 3 March 2020) [36]. A pairwise sequence similarity plot was displayed by a graphical user interface GenomeMatcher [37].  Table 2 and Supplementary Table S2 show detailed information on the types, numbers, and frequencies of DLA-88, DLA-88L, DLA-12, and DLA-DRB1 alleles identified in the 829 dogs. In total, 193 DLA alleles (89 in DLA-88, 18 in DLA-88L, 25 in DLA-12, and 61 in DLA-DRB1) were identified, and 17, 7, 5, and 6 were novel alleles of DLA-88, DLA-88L, DLA-12, and DLA-DRB1, respectively. Figure 2 shows frequencies of the 20 most frequent DLA alleles and the number of animals carrying the alleles in popular breeds in Japan. The highest frequent alleles in each DLA gene were DLA-88*006:01 (allele frequency: 8.9%), DLA-12*001:01:01 (45.1%), and DLA-DRB1*015:01 (14.0%). Although 43 DLA-12/88L alleles were identified, 65.7% (545 dogs) carried DLA-12*001:01:01. In the DLA-DRB1 gene, DLA-DRB1*015 group alleles (DLA-DRB1*015:01, DLA-DRB1*015:02, DLA-DRB1*015:03, and DLA-DRB1*015:04) were the most common, and 37.7% had the alleles. This result showed a similar ratio (23.6%) to the previously published report [20]. The number in the parenthesis of "Novel sequences" represents the number of allele sequences observed from two or more individuals in this study. The detailed information about all alleles detected in this study is described in Supplementary Table S2. a The numbers in the row indicate the number of registered sequences in the IPD-MHC database that were not detected in this study.

Evaluation of DLA-DRB1 Polymorphisms between Same Dog Breeds in Japan and the United Kingdom
The migration or transportation of dog breeds between different geographic locations has been shown to have a detectable effect on breed structures with the generation of genetically differentiated sub-populations [38][39][40]. Therefore, to evaluate the genetic bias of the DLA polymorphisms between the same dog breeds in Japan and another country, we compared our DLA-DRB1 genotyping data with the previously published DLA-DRB1 polymorphism data in the United Kingdom (UK) [21]. We compared the proportion of the dogs with each of the different DLA-DRB1 alleles in 10 breeds that had been analyzed in more than 12 dogs per breed in the present (Japan) and previous studies (UK). Of the 10 breeds in the UK and Japan, moderate to strong correlations with correlation coefficients ranging from 0.453 (Beagle) to 0.977 (Cavalier King Charles Spaniel), and a median value of 0.853 was confirmed in the nine breeds (Beagle, Golden Retriever, Labrador Retriever, Dachshund, Miniature Schnauzer, American Cocker Spaniel, Cavalier King Charles Spaniel, Shih Tzu, and Yorkshire Terrier) ( Figure 4). The Beagles showed a moderate correlation coefficient, but a large difference was observed between the two countries in the proportion of individuals carrying DLA-DRB1*006:01 (74.6% in the UK vs. 10.8% in Japan). However, for Border Collies, whereas 7 out of 10 alleles were commonly observed in both countries, the DLA-DRB1 allele frequency differed markedly, and no positive correlation was observed between the proportions (correlation coefficient r: −0.159) of alleles in each country.

Frequency of the 88-12/88L-DRB1 Haplotypes
To identify the two 88-12/88L-DRB1 haplotypes (88-12-DRB1 or 88-88L-DRB1) within the 829 dogs, we searched homozygous dogs with the three-loci, any two-loci (88-12/88L, 88-DRB1 and 12/88L-DRB1) and one-locus from our genotyping data. Firstly, 52 different sub-haplotypes of the 88-12/88L-DRB1 haplotypes were identified within 198 dogs (23.8%) that were homozygous at the three loci. Then, 54 sub-haplotypes of the 88-12/88L-DRB1 haplotypes were identified within 40 dogs and 163 dogs that were homozygous at twoloci (88-12/88L in 7 dogs, 88-DRB1 in 7 dogs, 12/88L-DRB1 in 16 dogs) and at one-locus (DLA-88 in 3 dogs, DLA-12/88L in 111 dogs, and DLA-DRB1 in 49 dogs), respectively.    Scatter plots based on the proportion of dogs with the DLA-DRB1 alleles in ten breeds are shown. The number of dogs analyzed and the number of observed alleles in the present Japanese and the previous UK study are described on the x-and yaxis in each plot, respectively. Dachshunds are summarized without classification by their size, such as miniature or kaninchen, and hair length, such as short or long, in both studies. An allele name is displayed alongside a marker when the proportion is above 5% in either the present or previous study. A red line in each plot represents a regression line. Also, a correlation coefficient (r) is indicated in each plot.

Frequency of the 88-12/88L-DRB1 Haplotypes
To identify the two 88-12/88L-DRB1 haplotypes (88-12-DRB1 or 88-88L-DRB1) within the 829 dogs, we searched homozygous dogs with the three-loci, any two-loci (88-12/88L, 88-DRB1 and 12/88L-DRB1) and one-locus from our genotyping data. Firstly, 52 different sub-haplotypes of the 88-12/88L-DRB1 haplotypes were identified within 198 dogs (23.8%) that were homozygous at the three loci. Then, 54 sub-haplotypes of the 88-12/88L-DRB1 haplotypes were identified within 40 dogs and 163 dogs that were homozygous at Proportion of dogs with each DRB1 alleles in the present study in Japan (%) Figure 4. Comparisons of the proportion of dogs with each DLA-DRB1 allele in 10 breeds between the present Japanese and the previous UK study. Scatter plots based on the proportion of dogs with the DLA-DRB1 alleles in ten breeds are shown. The number of dogs analyzed and the number of observed alleles in the present Japanese and the previous UK study are described on the xand y-axis in each plot, respectively. Dachshunds are summarized without classification by their size, such as miniature or kaninchen, and hair length, such as short or long, in both studies. An allele name is displayed alongside a marker when the proportion is above 5% in either the present or previous study. A red line in each plot represents a regression line. Also, a correlation coefficient (r) is indicated in each plot.

Comparison of Genetic Diversity between Dog Breeds by Haplotype Numbers and Heterozygosity
We investigated the genetic diversity of the 88-12/88L-DRB1 haplotypes using a total of 725 dogs within 24 different dog breeds (analyzed using ≥ 10 dogs/breed) and mongrels (mixed breeds). The number of different haplotypes in each breed ranged from three Shetland Sheepdogs to 27 Toy Poodles (Table 5), and up to 34 different haplotypes among the mongrels (Figure 2). Six dog breeds (Miniature Schnauzer, Shetland Sheepdog, Shiba, American Cocker Spaniel, Papillon, and Bernese Mountain Dog) had one particular 88-12/88L-DRB1 sub-haplotype at a frequency of greater than 50% ( Figure 5). Only two or three haplotypes represented more than 80% of all the haplotypes in seven breeds (Miniature Schnauzer, Shetland Sheepdog, Shiba, American Cocker Spaniel, Golden Retriever, Miniature Pinscher, and Shih Tzu). In contrast, more than 20 different haplotypes were detected in Chihuahua and Toy Poodle, and each haplotype frequency was distributed similarly ( Figure 5).  Ho: observed heterozygosity, He: expected heterozygosity, Fis: inbreeding coefficient, Hr: haplotype richness. a A number outside and inside the parenthesis indicates the number of dogs whose haplotype could be estimated and whose haplotype could not be estimated in each breed, respectively. b This table is sorted by the ascending order of the number of Hr. "-" in the column of the HWE test indicates no significant differences.
12/88L-DRB1 sub-haplotype at a frequency of greater than 50% ( Figure 5). Only two or three haplotypes represented more than 80% of all the haplotypes in seven breeds (Miniature Schnauzer, Shetland Sheepdog, Shiba, American Cocker Spaniel, Golden Retriever, Miniature Pinscher, and Shih Tzu). In contrast, more than 20 different haplotypes were detected in Chihuahua and Toy Poodle, and each haplotype frequency was distributed similarly ( Figure 5).  Table 5. A numeral following the breed's name indicates the number of different haplotypes detected in each breed. Parenthesis indicates the number of different 88-12/88L-DRB1 haplotypes with a frequency of 5% or higher. The breeds are sorted in ascending order by the percentage frequency of the most frequent haplotype in each breed. For example, the graph of Miniature Schnauzer at the bottom of the figure represents a total of nine sub-haplotypes of 88-12/88L-DRB1 haplotypes with a frequency of 5% or higher for three of them at >50%, 10-29%, and 5-9%, respectively.
We calculated the genetic diversity indices, such as observed heterozygosity (Ho), expected heterozygosity (He), inbreeding coefficient (Fis), and haplotype richness (Hr) to evaluate the 88-12/88L-DRB1 diversity in each breed ( Table 5) Table 5. A numeral following the breed's name indicates the number of different haplotypes detected in each breed. Parenthesis indicates the number of different 88-12/88L-DRB1 haplotypes with a frequency of 5% or higher. The breeds are sorted in ascending order by the percentage frequency of the most frequent haplotype in each breed. For example, the graph of Miniature Schnauzer at the bottom of the figure represents a total of nine sub-haplotypes of 88-12/88L-DRB1 haplotypes with a frequency of 5% or higher for three of them at >50%, 10-29%, and 5-9%, respectively.

Characteristics of Genetic Relationship of the 88-12/88L-DRB1 Haplotypes by Principal Component Analysis
To evaluate genetic relationship of the 88-12/88L-DRB1 haplotypes among different dog breeds, PCA was performed using the 24 breeds listed in Table 5. Of the 24 breeds plotted by PCA, 22 breeds were distributed closely around the centroid of the quadrants as if they were almost one population. The Shetland Sheepdog and Miniature Schnauzer breeds diverged markedly from the other 22 breeds. (Figure 6A and Supplementary Table S3A).
The positions of the Shetland Sheepdog and Miniature Schnauzer breeds within the matrix appear to have reflected the presence of their dominant haplotypes, 88*003-88*017-DRB1*002 (Hp- ID 20) in Shetland Sheepdog (Hp frequency: 67.1%) and 88*013-12*003-DRB1*009 (Hp-ID 23) in Miniature Schnauzer (68.8%). Since the 88*003-88*017-DRB1*002 (Hp-ID 20) also were commonly observed among Welsh Corgi and Border Collie, these two breeds were located slightly outside the large group of the other breeds and closer to Shetland Sheepdog. Removing the Shetland Sheepdog and Miniature Schnauzer outliers from the PCA analysis changed the genetic relationship slightly between some of the 22 breeds on the basis of the 88-12/88L-DRB1 haplotype frequencies ( Figure 6B). For example, the French Bulldog, Bulldog, Border Collie, and Yorkshire Terrier share the 88*028-88*029-DRB1*015 (Hp-IDs 24 and 25) at relatively high frequencies (13.4% to 43.4%), and these four breeds grouped more closely together and at some distance from the other breeds. Similarly, the Golden retriever and Labrador retriever, sharing the 88*508-12*001-DRB1*012 (Hp-ID 21), and American Cocker Spaniel and Cavalier King Charles Spaniel, sharing the 88*003-88*017-DRB1*009 (Hp-ID 31) separated further from each other and at a greater distance from the centroid (0, 0) of the PCA plot ( Figure 6B).

Number of Potential Recipients for 88-12/88L-DRB1-Matched Transplantation, Assuming Homozygous-Derived Somatic Stem Cells as Donors
Assuming that somatic stem cells could be established from the 52 types of homozygotes of the 88-12/88L-DRB1 haplotypes and that these cells could be used as donors for  Table 5 and (B) 22 breeds except for Shetland Sheepdog and Miniature Schnauzer from (A) are represented. The breeds that shared a three-loci haplotype with relatively high frequency (See Section 3.6. in detail) are grouped with a red circle. The contribution ratio of the first (PC1) and second component (PC2) are described on the xand y-axis in parentheses, respectively.

Discussion
In this study, we genotyped the DLA-88, DLA-12/88L, and DLA-DRB1 loci by Sanger sequencing using 829 dogs of 59 breeds and identified 89, 43, and 61 alleles, respectively. We also developed a two-stage PCR method for the polymorphism analysis of the DLA-88, DLA-88L, and DLA-12 genes by separating DLA-88 and DLA-12/88L with the 1st PCR and DLA-12 and DLA-88L with the 2nd PCR ( Figure 1B). This polymorphism analysis by PCR and sequencing clearly distinguished the DLA-88L allele from the DLA-88 allele, which was difficult with conventional RNA-based methods [15,18]. In fact, the previously reported DLA-88*042:02 [17] belongs to DLA-88L rather than DLA-88, and this allele along with DLA-88*008:02 and DLA-DRB1*004:01 constituted the 88-88L-DRB1 haplotype in the Maltese breed (Supplementary Table S2A). Therefore, this simpler and more accurate two-stage PCR method is an important tool to use for a better understanding of the DLA loci and haplotype differences and for evaluating various immune responses in dogs.
Overall, we identified 29 novel DLA-88, DLA-88L, and DLA-DRB1 alleles in this study. Of them, DLA-88*nov65 was newly detected as a DLA-88L allele that showed a different phylogenetic relationship from other DLA-88L alleles, and was highly similar to DLA-12*004:01 of the DLA-12 lineage (Figure 3). Interestingly, our previous study showed that DLA-12*004:01 was generated by a gene conversion event within the exon 2 region between the DLA-12 and DLA-88 alleles [30]. Therefore, the DLA-88*nov65 also might have been generated by gene conversion between the DLA-88 and DLA-12 alleles, similar to DLA-12*004:01. There may be many other unidentified DLA alleles generated by such gene conversions events.
In contrast to the DLA-I genes, the polymorphisms and diversity analyses of the DLA-II genes (DLA-DRB-DLA-DQA-DLA-DQB) have been performed previously in many different dog breeds [21,22,41,42]. Although the Japanese native species of Shiba has not been well analyzed previously, our current analysis showed that DLA-DRB1*056:01 (allele frequency: 55.4%) was the most frequent allele, followed by DLA-DRB1*092:01:1 (18.9%), and DLA-DRB1*011:03 (16.2%) (Supplementary Table S3A). These DLA-DRB1 alleles were detected only in Shiba, and therefore their detection is extremely rare even in past polymorphism analyses of the DLA-II genes, including other dog breeds of Asian origin [21,42]. The DLA polymorphism information on Japanese native breeds is extremely limited [42,43]. In this study, although we analyzed Japanese native species Shiba, Akita, Japanese Spitz, Chin, and Shikoku, the number of animals analyzed was less than 10 animals except for the 37 in the Shiba breed. Therefore, more DLA allele information is necessary for Japanese native species as well as for dog breeds that have not yet been analyzed.
Recent genomic analysis of the remains of extinct Japanese wolves (Canis lupus hodophilax) showed phylogenetically that after the Japanese wolf and modern dog ancestry had diverged from grey wolf lineages, gene flow occurred from the ancestor of Japanese wolves into the ancestor of Japanese dogs, including Shiba, and this flow likely continued and contributed to differentiate between the lineage of Japanese dogs and West Eurasian dogs [44]. Interestingly, DLA-DRB1*056:01 of Shiba was detected in Finnish and Russian wolves with frequencies of 4.0% and 2.9%, respectively [45], and Shiba DLA-DRB1*092:01:1 was detected in Canadian and Croatian wolves with frequencies of 6.0% and 11.0%, respectively [46,47]. These results indicated that Shiba might be a unique breed that shared some of its genomic sequences, including the DLA genomic region, with its ancestor in a different way than those of the European modern dog breeds.
High homozygosity of the DLA haplotypes generally implies a loss of DLA genetic diversity. The Ho values showed significantly lower values than the He values in 5 dog breeds, American Cocker Spaniel, Shiba, Papillon, Shih Tzu, and Beagle (Table 5). This suggests a high level of inbreeding in these five breeds. The high Fis values also observed in Shih Tzu (0.209) and Papillon (0.171) strongly suggest that the DLA diversity in these breeds of our population sample is decreasing by inbreeding (Table 5). Moreover, Shetland Sheepdog showed an extremely low Ho value of 0.314 (Table 5). The loss of genetic diversity due to high homozygosity might increase the deleterious genetic variation in purebreed dogs [48]. Also, high homozygosity of the DLA region due to both inbreeding and genetic bottlenecks by selective artificial breeding was associated with the development of autoimmune diseases in Italian Greyhounds [49]. In contrast, MHC heterogeneity of the Sea lion in wild populations appears advantageous to protect against infectious diseases [50], whereas the pregnancy rate in horses was reported to be decreased by sharing common MHC types between males and females [51]. Therefore, loss of the DLA diversity may affect biological fitness as homozygosity progresses.
The homozygous rate of DLA-DRB1-DLA-DQA1-DLA-DQB1 haplotypes was 35% in a previous study [22]. These three DLA-II genes are located together within 100 kb, while DLA-88 is located far from DLA-DRB1 by over 1.0 Mb [13], resulting in a much stronger linkage disequilibrium (LD) within the DLA-DRB1-DLA-DQA1-DLA-DQB1 haplotype than the 88-12/88L-DRB1 haplotype. Therefore, the lower 88-12/88L-DRB1 homozygous rate (23.8%) in this study than that of DLA-DRB1-DLA-DQA1-DLA-DQB1 haplotype in the previous study [22] might be associated with LD and rates of recombination between different DLA gene loci. The detection of 52 types of homozygotes for the 88-12/88L-DRB1 haplotypes suggests that 90.9% of the dogs analyzed in our study would have a successful 88-12/88L-DRB1-matched transplantation if their somatic stem cells were used in such a procedure. In comparison, if induced pluripotent stem cells (iPSCs) were established from homozygotes of 30 and 50 types of HLA haplotypes (HLA-A-HLA-B-HLA-DRB1) that are frequently observed in Japanese, 82.2% and 90.7% of Japanese would benefit from HLA-matched iPSC transplantation [52]. However, the HLA homozygosity rate for humans is relatively low at 0.5 to 1.5% for the HLA-A-HLA-B-HLA-DRB1 haplotype [53][54][55]. Therefore, the HLA of 15,000 and 24,000 individuals would need to be genotyped to identify these 30 and 50 HLA homozygotes, respectively [52]. In this regard, the MHC homozygotes, preferred donors for somatic stem cell sources, would be much easier to detect in dogs than in humans due to their higher rate of MHC homozygosity. Moreover, our new data on the frequency of DLA haplotypes in various dog breeds could help in the implementation of somatic stem cell transplantation along with a recent development of clinical-grade canine iPSCs derivation [56,57] and assist with the high expectations for regenerative medicine in the veterinary field [58].
In the PCA using the 88-12/88L-DRB1 haplotype frequencies, the values of the first principal component (PC1) and the second principal component (PC2) were extremely low at around 10% in both analyses, but with relatively strong diversity between the DLA haplotypes among the 24 dog breeds, which are popular in Japan ( Figure 6). The Hr values of 88-12/88L-DRB1 were less than five in 11 breeds, suggesting that there were relatively few breeds with many different types of 88-12/88L-DRB1 haplotypes (Table 5). Moreover, haplotype frequency bias was confirmed for six breeds, with some haplotype frequencies at 50% or more within each breed ( Figure 5). These results showed that the DLA diversity is highly conserved within most breeds, but divergent between almost all breeds, as similarly observed in a previous study on the DLA-II haplotype diversity [20,22]. Therefore, the DLA allele and DLA haplotype frequencies appear to change largely depending on the breeds. Of the 829 dogs of 59 breeds analyzed in this study, 68.5% (568 of 829 dogs) were from the top 20 most popular breeds registered in Japan (Table 1). In contrast, only a few dogs were analyzed from breeds that are listed in the top ten most popular breeds in the USA (American Kennel Club; https://www.akc.org/ (accessed on 28 November 2022)), such as German Shepherd and Rottweiler, which were less popular breeds in Japan. The genetic closeness of the DLA haplotypes among different breeds can be evaluated more accurately by enhancing the DLA genotyping data of breeds from which only a small number of dogs were analyzed in our present study. We showed that there are differences in the distribution of DLA-DRB1 alleles within the same breeds, such as Border Collie and Beagle, particularly if they are located in different countries, such as Japan and the UK (Figure 4). Since differences in the DLA diversity within a single breed between different countries have also been reported in some other breeds [49,59,60], geographical location can affect DLA diversity resulting in differences in the susceptibility for various diseases even in a single breed between different countries. From such a discussion, further DLA polymorphisms analysis for various breeds in different countries is warranted to better comprehend the intriguing features of DLA diversity.
A limitation of our study concerning DLA haplotypes containing both DLA-I and DLA-II genes was not to include DLA-DQA1 and DLA-DQB1 polymorphisms that might be linked to the 88-12/88L-DRB1 haplotypes. This was beyond the scope of our present study. The three genes of DLA-DRB1, DLA-DQA1, and DLA-DQB1 often show strong linkage disequilibrium [22], but novel DLA-DRB1-DLA-DQA1-DLA-DQB1 haplotypes might be generated by the recombination between the DLA-DR and DLA-DQ genes. Although the mismatch of HLA-DQ polymorphisms in organ transplantation is associated with the production of de novo donor-specific antibodies (DSA) against the HLA-DQ molecules and contributes to poor graft outcomes [61,62], such studies are lacking in dogs. Therefore, in regard to transplantations in dogs, further studies are necessary to genotype DLA-DQA1 and DLA-DQB1 and consolidate the extended haplotypes that were identified in our study, including all DLA-I and DLA-II genes, to select the most suitable organ donors in future.

Mongrels
Mixtures of different pure dog breeds or interbreeds.