Introduction

The cynomolgus macaques (Macaca fascicularis), alias the crab-eating monkeys or long-tailed macaques, are widely dispersed in nature, inhabiting a vast range of Southeast Asia including the Philippines, Indonesia, Vietnam, Malaysia, Thailand, Cambodia, Mauritius, and Brunei (Fuentes et al. 2011). This species along with other macaque species such as the rhesus macaque are often used for biomedical research on infectious and chronic diseases, neurology, immunology, reproduction, regenerative medicine, transplantation, vaccination, and immunotherapy (Gardner and Luciw 2008; Vallender and Miller 2013). Many of these diseases and treatment outcomes are influenced by polymorphisms of the major histocompatibility complex (MHC), a genomic region that encodes the MHC transplantation and immune regulatory molecules (Shiina et al. 2009; Shiina et al. 2004). In this regard, information on the diversity of the MHC alleles and haplotypes may provide an internal genetic control to better define the factors of disease susceptibility and immunity and help to maximize the power of the small study sizes and minimize a priori bias and the effect of heterozygosity in the macaque disease study groups (Arikata et al. 2012; Ericsen et al. 2014; Vallender 2014).

The human classical MHC class I genes, Human Leukocyte Antigen (HLA)-A, HLA-B, and HLA-C, are distinguished by their extraordinary polymorphisms with over 9,600 alleles implicated in disease resistance or susceptibility; whereas the human non-classical MHC class I genes, HLA-E, HLA-F, and HLA-G, are more limited in diversity by their relative monomorphism (IMGT/HLA Database release 3.20.0, http://www.ebi.ac.uk/imgt/hla/) (Robinson et al. 2011). In comparison, the overall structure of the MHC class I regions in the rhesus macaques (Daza-Vamenta et al. 2004) is more complex than the HLA (Shiina et al. 1999) and the common chimpanzee MHC (Patr) (Anzai et al. 2003) because the macaques carry many more duplicated MHC class I genes than humans and chimpanzees. For example, the Mafa-class I region, like the HLA region, is divided into three segments, whereby up to nine duplicated Mafa-B and Mafa-I (Mafa-B/I) genes correspond to HLA-B/C in the beta block, Mafa-E corresponds to HLA-E in the kappa block, and numerous duplicated Mafa-A, Mafa-AG, and Mafa-F genes correspond to HLA-A/G/F in the alpha block (Kulski et al. 2002; Watanabe et al. 2007) and Fig. 1. Overall, the homologous duplications appear to be fewer in number in the Mafa-class I region (Campbell et al. 2009; Otting et al. 2007; Pendley et al. 2008; Uda et al. 2004; Watanabe et al. 2007) than in the Mamu-class I region (Shiina et al. 2006).

Fig. 1
figure 1

The Mafa-class I duplicated genes within the Mafa-class I genomic structure and the representative Mafa-class I haplotypes that have high frequencies in the Filipino cynomolgus macaque population. The number and distribution of the Mafa-class I genes A, B/I, E, and F within each of the shaded segmental block structures were taken from previous publications (de Groot et al. 2012; Kulski et al. 2002; Urvater et al. 2000; Watanabe et al. 2007). Names and numbers in the boxes show Mafa-F allele or A-Hp, B-Hp and E-Hp names and their frequencies in the 127 Filipino cynomolgus macaques. Numbers between boxes show haplotype frequencies of F/A-Hp, A/E-Hp, and E/B-Hp. Hp.Freq. means haplotype frequency

Full size image

The detection and breeding of MHC homozygous primate species are considered necessary for the development of effective vaccine and immunosuppressive drug protocols, evaluation, and validation of organ and regenerated cells originating from induced pluripotent stem (iPS) cells and/or embryonic stem (ES) cells in transplantation medicine (Klimanskaya et al. 2006; Takahashi et al. 2007). In this regard, the cynomolgus macaques from Mauritius and the Philippines are considered the most suitable populations for use in biomedical research because they are thought to be genetically less diverse than those from other geographic locations (Kawamoto et al. 2008; Tosi and Coke 2007; Wiseman et al. 2007). The Mauritian and Filipino macaque populations are believed to have originated from a colonization of Indonesian/continental animals followed by a bottleneck approximately 110,000 years BP (Blancher et al. 2008). In fact, the Mauritian and Filipino macaques show a relatively high genetic diversity, but a smaller number of MHC class I and class II alleles than the Indonesian and Vietnamese populations (Blancher et al. 2012; Blancher et al. 2014; Blancher et al. 2006; Kita et al. 2009; Krebs et al. 2005; Leuchte et al. 2004; Sano et al. 2006). Although the seven most frequent Mafa-class I haplotypes were characterized from 67 distinct Mafa-class I transcripts in the Mauritian population (Budde et al. 2010), detailed polymorphic information for the Mafa-class I genes is still lacking in comparison to the Mafa-class II genes (Blancher et al. 2014) in the Filipino population. Hence, in order to effectively use the Filipino population for medical research based on discovery of MHC homozygous animals, it is necessary to better understand the allele and haplotype diversity of the Mafa-class I genes.

Massively parallel pyrosequencing is one of the next generation sequencing (NGS) techniques and more effective methodologies for high-throughput genotyping of MHC genes and the detection of low-level-expressed MHC alleles (Babik et al. 2009; Wegner 2009; Wiseman et al. 2009). Pyrosequencing has been used successfully to discover and accumulate a large number of MHC alleles in some macaque species (Budde et al. 2010; Karl et al. 2009; O'Leary et al. 2009; Wiseman et al. 2009). This NGS method is precise and speedy for genotyping the highly duplicated macaque MHC class I genes from tissue RNA than the conventional MHC genotyping methods such as microsatellite, sub-cloning, and Sanger sequencing after reverse transcriptase-polymerase chain reaction (RT-PCR).

In this study, to better elucidate the degree and types of allele and haplotype diversity of the Mafa-class I genes in the Filipino population, we identified 56 novel alleles from 112 alleles obtained for the Mafa-A, Mafa-B/I, Mafa-E, and Mafa-F genes in 127 unrelated cynomolgus macaques by using both the conventional and the NGS genotyping methods. We analyzed the allele sequences by performing both population genetic and phylogenetic analyses on the coding regions (exons 1 to 7) of the complementary DNA (cDNA) nucleotide sequences that we obtained from the peripheral white blood cells of the macaques.

Material and methods

Animals

In this study, we used the RNA from the white blood cells of 127 unrelated cynomolgus macaques from the Philippines archipelago (112 from INA Research Philippines INC. and 15 from Sicombrec Co.). The Mafa-DR—Mafa-DQ—Mafa-DP haplotypes of these animals have been previously characterized (Blancher et al. 2012; Blancher et al. 2014). We also used the genomic DNA from the white blood cells of 20 unrelated cynomolgus macaques from Vietnam and Indonesia (ten from Nafovanny and ten from CV Universal Fauna Breeder and Exporter of Non-human Primates for Laboratories) for genotyping of a novel MHC class I gene, Mafa-A8*01:01. The blood collection and animal studies were conducted in accordance with the guidelines for animal experiments specific to each location.

RNA extraction and reverse-transcriptase (RT) reaction

Total RNA was extracted directly from the peripheral white blood cell samples of each of the 127 animals using the TRIzol reagent (Invitrogen/Life Technologies/Thermo Fisher Scientific, Carlsbad, CA). cDNA was synthesized by oligo d(T) primer using the ReverTraAce for reverse transcriptase reaction (TOYOBO, Osaka, Japan) after treatment of the isolated RNA with DNase I (Invitrogen/Life Technologies/Thermo Fisher Scientific, Carlsbad, CA).

Sub-cloning and Sanger sequencing of Mafa-A, Mafa-B, and Mafa-I genes

The Sanger sequencing method genotyped 127 animals using the Mafa-A, Mafa-B, and Mafa-I cDNA sequences, ranging from exon 1 to exon 7 (PCR product size: 1,037~1,061 bp). The sequencing primers used for the initial genotyping procedure were one common sense primer (Mafa-A/B/I_F) and the two locus-specific anti-sense primers, Mafa-A_R for Mafa-A (Kita et al. 2009) and Mafa-B/I_R for Mafa-B and Mafa-I (Table 1). The Sanger sequencing method also was used to confirm 37 novel cDNA sequences of Mafa-A, Mafa-B, and Mafa-I in 26 animals that were identified by the pyrosequencing method. In this case, confirmatory genotyping by the Sanger method was performed with six allele-specific primers (underlined in Table 1) that were specifically based on the newly identified sequences. Mafa-G and Mafa-AG were not examined in this study because Mafa-G is not expressed by white blood cells and the Mafa-G copies are unexpressed pseudogenes. In brief, the 20 μL amplification-reaction-volume contained 10 ng of cDNA, 0.4 units of KOD FX polymerase (TOYOBO), 2× PCR buffer, 2 mM of each dNTP and 0.5 μM of each primer. The cycling parameters were as follows: an initial denaturation of 94°C/2 min. followed by 30 cycles of 98°C/10 s, 60°C/30 s, and 68°C/1 min. PCR reactions were performed using the thermal cycler GeneAmp PCR system 9700 (Applied Biosystems/Life Technologies/Thermo Fisher Scientific, Foster City, CA). After PCR amplification, PCR products were cloned into the PGEM-T Easy vector with the pGEM®-T Easy Vector System according to the protocol provided by the manufacturer (Promega, Madison, WI) and sequenced by using the ABI3130 genetic analyzer (Applied Biosystems/Life Technologies/Thermo Fisher Scientific) in accordance with the protocol of Big Dye terminator method. To avoid PCR and sequencing artifacts generated by polymerase errors, 48 clones per animal were sequenced.

Table 1 PCR primers used for this study
Full size table

Genotyping of Mafa-class I genes by pyrosequencing

A single Mafa-class I-specific primer set (Class_I_F and Class_I_R) was designed based on generic sequences in exon 2 and exon 4 (PCR product size: 514 or 517 bp) that could amplify all known Mafa-class I alleles for massively parallel pyrosequencing (Table 1) of 127 animals. The 20 μL RT-PCR amplification-reaction-volume contained 10 ng of cDNA, 0.4 units of high-fidelity KOD FX polymerase (TOYOBO), 2× PCR buffer, 2 mM of each dNTP and 0.5 μM of each primer. The cycling parameter was as follows: 25 cycles of 98°C/10 s, 58°C/30 s, and 68°C/30 s. The PCR products were purified by the QIAquick PCR Purification Kit (QIAGEN, Hiden, Germany) and quantified by the Picogreen assay (Invitrogen) with a Fluoroskan Ascent micro-plate fluorometer (Thermo Fisher Scientific, Waltham, MA).

Titanium rapid libraries of fragmented DNA linked to AMPure beads (Beckman Coulter Genomics, Danvers, MA) were prepared for the Roche Genome Sequencer 454 FLX system by nebulization, fragment end repair, and multiple identifier (MID)-labeled adaptor ligation, emulsion PCR (emPCR), and emulsion breaking were performed according to the manufacturer’s protocol (Roche, Basel, Switzerland) (Kita et al. 2012). After the emulsion breaking step, the beads carrying the single-stranded DNA templates were enriched, counted, and deposited into a PicoTiterPlate to obtain sequence reads (Kita et al. 2012). After the sequencing run, image processing, signal correction, and base-calling were performed by the GS Run Processor Ver. 3.0 (Roche) with full processing for shotgun or paired-end filter analysis. Quality-filter sequence reads that were passed by the assembler software (single sff file) were binned on the basis of the MID labels into each separate sequence sff file using the sff file software (Roche). These files were further quality trimmed to remove poor sequence at the end of the reads with quality values (QVs) of less than 20. The Mafa-class I alleles were mainly assigned by matching the trimmed and MID labeled sequence reads as 99 and 100 % matching, 200 of minimum overlap length and 10 alignment identity score parameters with all the known Mafa-class I allele sequences released in the IMGT/MHC-NHP database (Robinson et al. 2013) using the GS Reference Mapper Ver. 3.0. On the other hand, to discover novel Mafa-class I sequences, the trimmed and MID-binned sequences were de novo assembled as >85 % matched parameters after the outputs were converted to ace files for the Sequencher Ver. 5.01 DNA sequence assembly software (Gene Code Co., Ann Arbor, MI). A defined consensus sequence obtained from the de novo assembly was used as a reference sequence to identify and map the correct allele sequence in this study.

Nomenclature of novel Mafa-class I alleles

All newly determined Mafa-class I sequences were deposited in the Genbank database (accession references are given in Table 2), and the novel Mafa-A, Mafa-B, and Mafa-I sequences also were submitted to the IMGT/MHC-NHP database (Robinson et al. 2013) for allele nomenclature. The Mafa-class I allele names were assigned by the IPD-MHC NHP database following classical rules (de Groot et al. 2012). An example of the allele nomenclature is Mafa-A1*001:02:01 where Mafa-A1 is the MHC allele of the cynomolgus macaque (M. fascicularis, Mafa) encoded by the class I locus A1. The first three digits after the asterisk define the lineage 001, whereas the two digits after the first colon define the allele number 02. These allele numbers are arbitrary and are numbered in the order in which they were identified. The two digits 01 after the second colon describe a synonymous base pair difference between two sequences. Whereas the Mafa-A alleles may be grouped into at least four lineages or loci, no locus number designation has been introduced for most of the Mafa-B loci because the macaque B genes largely differ in number between haplotypes (de Groot et al. 2012). An exception to the rule is the oligomorphic locus B3, which was previously named as the locus Mamu-I in the rhesus macaque (Urvater et al. 2000), and is found on each haplotype (de Groot et al. 2012). In this study, we refer to Mafa-B3 locus as Mafa-I.

Table 2 112 distinct Mafa-class I alleles identified in 127 Filipino population samples
Full size table

Haplotype estimation

The Mafa-A, Mafa-B, and Mafa-E haplotypes were characterized by manually sorting of Mafa-A, Mafa-B, Mafa-I, and Mafa-E alleles based on Mafa-A, Mafa-B/I, and/or Mafa-E homozygous animals. In contrast, we estimated Mafa-A, Mafa-B/I, and Mafa-E haplotypes by comparison between the homozygous animals and heterozygous animals that have the identical MHC-class I alleles with the homozygous animals as shown in Supplementary figure 1. Estimation of the Mafa-class I haplotypes and haplotype frequencies were performed by the PHASE 2.1.1 program (Stephens et al. 2001) using Mafa-F allele and Mafa-A, Mafa-B, and Mafa-E haplotype data.

Characterization of a novel Mafa-class I locus Mafa-A8*01:01

The full coding sequence of Mafa-A8*01:01 was determined by the conventional method as described above using two pairs of Mafa-A8*01:01 gene-specific primers (MHCI-5.1 and 339-R, and 339-F, and MHCI-3.1, Table 1).

One Mafa-A8*01:01-specific primer set (339genome_F and 339genome_R2) was used for detection of the gene in Vietnamese and Indonesian animals by PCR analysis and for detection of the copy numbers of Mafa-A8*01:01 in allele-positive Filipino animals by quantitative PCR (qPCR) method (Table 1). The primers were designed to amplify the allele sequence in exons 3 and 4 (PCR product size: 857 bp). Twenty genomic DNAs from ten Vietnamese and ten Indonesian animals were extracted from peripheral blood cells using the QIA amp Blood Kit (QIAGEN). The 20-μL amplification-reaction-volume contained 10 ng of genomic DNA, 0.4 units of KOD FX polymerase (TOYOBO), 2× PCR buffer, 2 mM of each dNTP and 0.5 μM of each primer. The cycling parameters were as follows: an initial denaturation of 94°C/2 min followed by 30 cycles of 98 °C/10 s, 65 °C/30 s, and 68 °C/1 min.

Copy numbers of the gene was detected by qPCR method using the StepOnePlus™ Real-Time PCR System (Applied Biosystems/Life Technologies/Thermo Fisher Scientific) with KOD SYBR® qPCR Mix (TOYOBO). Melting curve analysis showed that there was no primer dimer formation. The relative quantitative values were calculated by the comparative C(T) method also referred to as the 2 (−DeltaDeltaC(T)) method (Schmittgen and Livak 2008).

Phylogenetic analysis

Multiple sequence alignment was created using the ClustalW Sequence Alignment program of the Molecular Evolution Genetics Analysis software 5 (MEGA5) (Tamura et al. 2011). Phylogenetic trees of the MHC-class I genes were constructed by the neighbor-joining (NJ) method in MEGA5 (Saitou and Nei 1987) using exons 2 to 4 (alignment length: 822 bp excluding gap sites). A NJ tree was constructed by the Maximum Composite Likelihood model and assessed using 10,000 bootstrap replicates. We used the following MHC-I sequences (DNA accession numbers) for phylogenetic analyses: HLA-A (Accession Number: NM_002116), HLA-B (NM_005514), HLA-C (NM_002117), HLA-E (NM_005516), and HLA-F (NM_018950) and HLA-G (NM_002127), Mafa-A1 (LC043310), Mafa-A2 (LC053847), Mafa-A3 (LC043318), Mafa-A4 (LC043319), Mafa-B (LC043350), Mafa-I (LC043377), Mafa-E (U02976), Mafa-F (DQ367725), Mafa-AG (HQ992797), and mouse H2-K1 (NM_001001892).

Results

Sequence read information from 127 Filipino cynomolgus macaques

Mafa-class I cDNA amplified by using the specific primer set Class_I_F and Class_I_R (PCR product size: 514 bp or 517 bp) were sequenced by parallel pyrosequencing. Supplementary Table 1 shows the number of draft sequences reads (sequences passing quality control (QC) criteria) after base calling, generated by the pyrosequencing method using the manufacturer specifications for 127 Filipino population samples. Draft read numbers in total were 3,328,273 reads with a range of reads from 3,872 in the animal ID189 to 103,942 reads in the animal ID248 that were high-quality reads with more than 20 quality values (QVs) and an average QV of 33.6 ± 3.0 in the high-quality sequence reads. The draft read bases in total were 1,217 Mb with a range between 1.4 Mb in ID189 and 36.2 Mb in ID248 (9.6 ± 6.7 Mb on average), with an overall average read length of 365.7 ± 24 bases and an overall median read length of 410 ± 32 bases (ESM Table 1). Therefore, the sequence reads had sufficient high-quality and sequence volume for further genotyping analysis.

Mafa-class I cDNA alleles

From our initial characterization of the Mafa-A, Mafa-B, and Mafa-I allele sequences using 127 animals (exon 1 to exon 7) by the sub-cloning and Sanger sequencing methods, we detected in total 58 Mafa-A, Mafa-B, and Mafa-I alleles (39 known alleles and 19 novel alleles). After genotyping of Mafa-class I genes using 127 animals by pyrosequencing, we detected an additional 54 Mafa-class I alleles (17 known alleles and 37 novel alleles). Of the 37 novel alleles, 19 were Mafa-A/B/I alleles and the other 18 were Mafa-E or Mafa-F alleles. Finally, we characterized the 19 novel Mafa-A/B/I allele sequences using 26 animals (exon 1 to exon 7) by the sub-cloning and Sanger sequencing methods. Therefore, to summarize this procedure, we detected in total 112 alleles (56 known alleles and 56 novel alleles) using 127 animals, and of the 56 novel alleles we characterized 38 Mafa-A, Mafa-B, and Mafa- I allele sequences by the sub-cloning and Sanger sequencing methods. Of the 38 alleles, 19 were detected by the Sanger method before pyrosequencing and the other 19 were detected by the Sanger method after pyrosequencing.

Table 2 shows the number and frequencies of 112 distinct Mafa-class I alleles (28 Mafa-A, 54 Mafa-B, 12 Mafa-I, 11 Mafa-E, and 7 Mafa-F) that were identified in 127 Filipino cynomolgus macaques by mapping the sequence reads as 99 and 100 % matching parameters with all the known Mafa-class I allele sequences released in the IMGT/MHC-NHP database using the GS Reference Mapper Ver. 3.0. All of the Mafa-A, Mafa-B, and Mafa-I allele sequences were supported by conventional analytical methods, 56 were previously reported in IMGT/MHC-NHP database (Campbell et al. 2009; Kita et al. 2009; Krebs et al. 2005; Lawrence et al. 2012; Otting et al. 2007; Pendley et al. 2008; Uda et al. 2004; Urvater et al. 2000; Wiseman et al. 2007; Wu et al. 2008) and the other 56 (eight Mafa-A, 24 Mafa-B, six Mafa-I, 11 Mafa-E, and seven Mafa-F) were newly identified in this study (Table 2). All of the newly identified Mafa-E and Mafa-F alleles (tentatively named as Mafa-E-like1~Mafa-E-like11 and Mafa-F-like1~Mafa-F-like7) have 94.7 to 99.6 % and 98.7 to 99.6 % nucleotide similarities among the 473-bp peptide-binding region, respectively (ESM Table 2). Overall, the number of Mafa-class I allele sequences per animal ranged from 8 to 27 in total with an average 18.9, and a range of alleles of 1 to 7 for Mafa-A, 1 to 14 for Mafa-B, 1 to 3 for Mafa-I, 1 to 4 for Mafa-E, and 1 to 2 for Mafa-F. The six most frequent alleles in the 127 animals were Mafa-I*01:10/11/14/15 (48.4 %), Mafa-A3*13:03:01 (39.8 %), Mafa-F-like4 (38.2 %), Mafa-E-like3 (35.8 %) and Mafa-E-like10 (32.3 %) (Table 2).

Of the 112 alleles, 13 alleles were perfectly matched with previously reported alleles of the rhesus macaque (Macaca mulatta, Mamu) (Boyson et al. 1995; Boyson et al. 1996; Campbell et al. 2009; Karl et al. 2008; Muhl et al. 2002; Otting et al. 2007; Otting et al. 2005), the southern pig-tailed macaque (Macaca nemestrina, Mane) (Lafont et al. 2003; Lafont et al. 2004) and/or stump-tailed macaque (Macaca arctoides, Maar) (Urvater et al. 2000). These trans-species polymorphisms (Table 3) were probably already generated before speciation of cynomolgus macaques 2.4~4.2 million years ago (Hedges et al. 2006).

Table 3 Mafa-class I allele supporting the trans-species polymorphism
Full size table

Haplotypes

Table 4 shows the numbers and frequencies of the Mafa-A, -B/I, and -E haplotypes estimated for 127 cynomolgus macaques at their different MHC loci. We sorted 17 Mafa-A haplotypes to four Mafa-A loci, 23 Mafa-B/I haplotypes to nine loci including locus Mafa-I, and 12 Mafa-E haplotypes to two loci. In regard to homozygous haplotypes, eight Mafa-A haplotypes (A-Hp1, A-Hp2.1, A-Hp2.2, A-Hp4, A-Hp6, A-Hp7.1, A-Hp7.2, and A-Hp8.1) were supported by 23 Mafa-A homozygous animals, seven Mafa-B/I haplotypes (B/I-Hp1, B/I-Hp2, B/I-Hp3.1, B/I-Hp4.1, B/I-Hp4.2, B/I-Hp5, and B/I-Hp7) were supported by 20 Mafa-B/I homozygous animals, and eight Mafa-E haplotypes (E-Hp1.1, E-Hp1.2, E-Hp2, E-Hp3, E-Hp4, E-Hp5.1, E-Hp6, and E-Hp7.1) were supported by 29 Mafa-E homozygous animals.

Table 4 Mafa-A, Mafa-B/I, and Mafa-E haplotypes estimated using 127 Filipino population samples
Full size table

The three most frequent Mafa-A haplotypes were A-Hp7.2 (26.8 %), A-Hp8.1 (16.1 %), and A-Hp4 (9.4 %), while the least frequent haplotypes were A-Hp5.3 and A-Hp8.2 (0.4 %), haplotypes that are likely variants of A-Hp5.1 or A-Hp5.2, and A-Hp8.1, respectively (Table 4 (A)). In comparison, the four most frequent Mafa-B/I haplotypes were B/I-Hp1 (26.0 %), B/I-Hp2 (16.5 %), and B/I-Hp3.1 (8.3 %) and B/I-Hp4.1 (8.3 %), while the least frequent were B/I-Hp3.2, B/I-Hp6.2, B/I-Hp8.2, B/I-Hp9.2, B/I-Hp13.2, B/I-Hp14, and B/I-Hp15 (0.4 %). Of the least frequent, at least B/I-Hp3.2, B/I-Hp6.2, and B/I-Hp8.2 are possible variants of B/I-Hp3.1, B/I-Hp6.1, and B/I-Hp8.1, respectively, and B-Hp14 and B-Hp15 are rare haplotypes (Table 4 (B)). The two kinds of Mafa-I sequences observed in B/I-Hp6 were not observed in B/I-Hp9 and B/I-Hp13. The six Mafa-B/I haplotypes B-Hp1, B-Hp2, B-Hp4.1, B-Hp4.2, B-Hp6.3, and B-Hp8.1 were identified as consensus Mafa-B/I haplotypes, although some alleles such as Mafa-B*072:01, Mafa-B*079:02:02, Mafa-B*089:01:02, Mafa-B*098:08, Mafa-B*098:10, and Mafa-I*01:12:01 were not observed in these six haplotypes in some animals. The absence of these alleles in some animals might be due to extremely low gene expression or the real absence of the actual allele. Of B/I-Hp2, a part of the Mafa-B/I alleles of the B/I-Hp2 composition were detected such as “B*104:03 - B*144:03N - B*057:04 - B*060:02 - B*46:01:02” in three animals and “B*050:08 - B*114:02 - B*072:01” in two animals. In addition, three recombinants between B/I-Hp3 and B/I-Hp4 and between B/I-Hp2 and B/I-Hp6 were observed in three animals. These structural variants may have been generated by either a genetic recombination or a gene duplication event in the Filipino population in relatively recent times.

The three most frequent Mafa-E haplotypes were E-Hp1.1 (31.9 %), E-Hp2 (17.3 %), and E-Hp3 (14.6 %), while the least frequent haplotype was E-Hp5.2 (0.4 %), a haplotype that is a possible variant of E-Hp5.1 (Table 4c). The Mafa-E genotyping identified one to four distinct Mafa-E alleles per animal. Thus, one or two Mafa-E loci are involved in the gene expression. Although there was no previous information for the Mafa-E haplotype (E-Hp) classification in cynomolgus macaques, we estimated 12 Mafa-E haplotypes by manually sorting of the observed alleles based on the Mafa-A, Mafa-B/I, and/or Mafa-E haplotype homozygous animals.

The number and frequency of the Mafa paired haplotypes estimated by the PHASE 2.1.1 program were 26 Mafa-F—Mafa-A haplotypes (F/A-Hp), 30 Mafa-A—Mafa-E haplotypes (A/E-Hp) and 49 Mafa-E—Mafa-B haplotypes (E/B-Hp), are shown in ESM Table 3. The six-haplotype pairs over 10 % frequencies were F/A-Hp1 (24.8 %), A/E-Hp1 (24.8 %), F/A-Hp2 (15.7 %), A/E-Hp2 (15.7 %), E/B-Hp1 (13.4 %), and E/B-Hp2 (13.0 %). Also, 84 Mafa-F—Mafa-A—Mafa-E—Mafa-B haplotypes (F/A/E/B-Hp) were estimated using Mafa-F allele and Mafa-A, Mafa-B, and Mafa-E haplotype data (Supplementary Table 3). The probability of estimated Mafa-class I haplotypes in each animal ranged from 0.35 to 1.00 with average of 0.91. Figure 1 shows the five most frequent F/A/E/B haplotypes, F/A/E/B-Hp1 (10.6 %), F/A/E/B-Hp2 (10.2 %), F/A/E/B-Hp3 (5.5 %), F/A/E/B-Hp4 (3.9 %), and F/A/E/B-Hp5 (3.9 %) (Supplementary Table 4).

Characterization of a novel Mafa-class I locus, Mafa-A8*01:01

We first identified the novel Mafa-class I sequence Mafa-A8*01:01, which we categorized as a minor allele, during the de novo assembly of the sequence reads. The entire coding sequence was composed of 1,089 bp and 362 deduced amino acids. Although the amino acid sequence contains the structurally essential N-glycosylation sites (amino acid positions 86–88), it lacks one cysteine residue involved in disulfide bonding (amino acid position 101) while the three other cysteine residues are conserved.

We later confirmed the presence of this sequence in 37 of the 127 cynomolgus macaques by PCR based genotyping using a Mafa-A8*01:01-specific primer set (data not shown). Polymorphism of the gene was not detected in the 473-bp peptide-binding region (exons 2–4) of the 37 positive animals or in the 857 bp PCR region (exons 3–4) of nine Vietnamese and Indonesian cynomolgus macaques that were also found to have this sequence (Fig. 2a). Ten Vietnamese and ten Indonesian DNA samples were genotyped with the Mafa-A8*01:01-specific primer set, and PCR products corresponding to the gene were observed in three and five animals, respectively, suggesting that the gene has diffused among some different cynomolgus macaque species (Fig. 2a). The highest nucleotide and amino acid similarities of the coding sequences of Mafa-A8*01:01 with known nonhuman primate MHC class I genes were 85.5 % with Mamu-A1*011:03 (EF580154) and 76.9 % with Mamu-A (D5KRD5), respectively. In comparison, there was 83 to 84 % nucleotide identity with each of the HLA class I genes. However, although we tried the same primer set and PCR conditions in a PCR experiment using human DNA samples, no PCR products were obtained (data not shown). In addition, a phylogenetic tree supports the divergence of the gene from the other Mafa and HLA class I genes (Fig. 3).

Fig. 2
figure 2

Detection of Mafa-A8*01:01 locus in Vietnamese and Indonesian macaques (a) and in the Mafa-class I haplotype structures A-Hp6 (b) and A-Hp10 (c). The electrophoresis images in (a) show the PCR products from ten Vietnamese and ten Indonesian genomic DNA samples using Mafa-A8*01:01-specific primers. Numbers on upper side indicate genomic DNA samples. M and P indicate bands of the 100 bp DNA size marker ladder and a Filipino population DNA sample (ID152) used as a positive control, respectively. The dark background in (b) and (c) indicates the Mafa-A alleles that are composed of A-Hp6 and A-Hp10, respectively. Light gray background indicates Mafa-F alleles, Mafa-E, and Mafa-B haplotypes that link to A-Hp6 or A-Hp10. The gene order of the Mafa-A alleles and the exact location of Mafa-A8*01:01 are estimated by haplotype sorting. ND means “not detected”.

Full size image
Fig. 3
figure 3

Nucleotide sequence based phylogenetic tree using representative Mafa-class I alleles constructed by the neighbor-joining method. Numbers at branches indicate bootstrap values.

Full size image

In an attempt to determine the copy number of the Mafa-A8*01:01 gene by qPCR analysis of 37 allele-positive cynomolgus macaques, two animals (ID165 and ID189) were detected to have two copies in contrast to only one copy in the other 35 animals. From the detailed linkage analysis of the copy numbers with Mafa-F alleles and all their haplotypes, Mafa-A8*01:01 was linked to a genomic region between Mafa-F and Mafa-A (Fig. 2b, c). The gene was detected in 14 animals that have A-Hp6, and in 14 of 15 animals that have A-Hp10. In fact, A-Hp6 and A-Hp10 were observed in animal ID189, but only A-Hp6 was present in ID165 (Fig. 2b, c). However, Mafa-A8*01:01 and Mafa-A4*14:14 were missing from one animal (ID181) with A-Hp10 (Fig. 2c). Therefore, Mafa-A8*01:01 and Mafa-A4*14:14 may have been segmentally deleted from A-Hp10 in relatively recent times.

Discussion

One major advantage of genotyping the macaque MHC by the NGS pyrosequencing method over the conventional Sanger sequencing method was the use of MID-tagged adaptors or PCR primers to identify individual amplicons that can be sequenced simultaneously with an arbitrarily chosen coverage. However, despite its increase in speed and sensitivity, this new sequencing technology is error-prone and poses considerable challenges because it can be more difficult to discriminate between sequencing errors and true rare alleles due to the complex nature of the generated artifacts and errors that require an efficient and accurate quality control (Babik et al. 2009). In fact, 42.0 % of the 3,328,276 draft sequence reads were unavailable for analysis due to various reasons such as short read lengths and incomplete extension, mixed reads, artificial substitutions, insertion, and deletion (indel) errors, including homopolymer errors (ESM Table 1). Nevertheless, the current pyrosequencing method produced 15,205 mapped sequence reads per animal (range 2,683 to 65,382) that were sufficient for effective genotyping and enabled the detection of all of the Mafa-class I alleles in 127 animals including the alleles expressed at extremely low levels. However, all of the known and novel Mafa-class I alleles detected in this study depended on using two different matching parameters (99 and 100 %) as part of an efficient mapping protocol. Both of these matching parameters were necessary to complete the allele assignment and consequently for genotyping and haplotyping of the Mafa-class I alleles with precision.

In this study, 1 to 3 Mafa-A and 2 to 7 Mafa-B classical class I alleles were detected per haplotype. These haplotypic allele numbers are consistent with a previous report that 5 of 13 classical class I alleles in homozygous Mauritian macaque were expressed at relatively high levels (Aarnink et al. 2011). In addition, we found that the Mafa-E region is organized and expressed in the peripheral white blood cells by at least three major Mafa-E alleles and eight minor Mafa-E alleles that have different amino acid sequences and variable Mafa-E haplotype structures (Tables 2 and 4). Although HLA-E has very limited polymorphism, it serves as the ligand for the inhibitory NKG2A receptor expressed by NK cells (Lee et al. 1998). Therefore, Mafa-E and Mamu-E, because of their increased polymorphism and haplotype diversity, may have a different function to HLA-E. On the other hand, the seven Mafa-F alleles were well conserved in each other. This characterization is similar to HLA-F, which was recently identified to be one of the ligands for NK cell Ig-like receptors (Goodridge et al. 2013). These between species comparisons suggest that more comprehensive MHC genotyping and functional studies are still required for cynomolgus macaques to provide better insight into the diversity and the role of the MHC genes in this species.

As part of this study, we discovered a novel Mafa-class I lineage, Mafa-A8*01:01, and linked it to the Mafa-A locus by comparative analysis of the different Mafa-class I haplotypes. Mafa-A8*01:01 was linked most strongly with A-Hp6 and A-Hp10, but it was also detected in 5.7 % of A-Hp7 and 9.5 % of A-Hp8 (Fig. 2b and c). This allele was also observed in 30 and 50 % of Vietnamese and Indonesian cynomolgus macaques, respectively, that we examined (Fig. 2a). The MHC-A region was probably formed by birth and death evolution (Piontkivska and Nei 2003) such as by repeated gene duplication and deletion events in human and non-human primates (Kono et al. 2014; Shiina et al. 2006) as shown in the detailed segmental and phylogenetic sequence analysis of the rhesus macaque MHC alpha block (Kulski et al. 2004). In the HLA-A region, an approximately 70 kb insertion and 50 kb deletion near the HLA-A gene were detected in some HLA-A allele lineages (Watanabe et al. 1997). The genomic rearrangements seen in our study are the likely segmental deletion of Mafa-A8*01:01 and Mafa-A4*14:14 in A-Hp10 (Fig. 2), and the presence of Mafa-A8*01:01 in A-Hp7 and A-Hp8 possibly could have been generated by recombinational crossing over and gene conversion. Phylogenetic analysis suggests that the Mafa-A8*01:01 has diverged from the Mafa-A1-A4 lineage (Fig. 3), and that, in comparison to the Mafa-A1-A4 allele lineage, one of the structurally essential cysteine residues of Mafa-A8*01:01 was replaced by a serine at amino acid position 101. In addition, polymorphism analysis of Mafa-A8*01:01 suggests that this allele is well conserved in various haplotypes among different macaque populations, and that it might have a function different to those of other classical class I genes. However, because the Mafa-A8*01:01 allele does not express a classical α2 domain it might be considered to be a new type of non-classical MHC gene whose function has yet to be determined. Therefore, this allele warrants functional analysis and consideration when matching MHC alleles in macaque transplantation studies.

Because the results of biomedical experiments strongly depend on the immunogenetic background of animals conditioned by various environmental selective factors such as pathogens, MHC homozygous macaques are preferred for use in biomedical research (Vallender 2014). In recruiting macaques for biomedical studies, the origin of the animals and their genetic polymorphisms need to be considered carefully at the population level. In this regard, one of our Filipino animals (ID152) was strictly a “near-homozygote” that has the F/A/E/B-Hp1 in the Mafa-class I region and the #7 haplotype in the Mafa-class II region (Blancher et al. 2014) on both chromosomes (tentatively named “HT1”) (Fig. 4). The HT1 potentially has only five classical class I alleles in the Filipino Mafa-A and Mafa-B genes which is half the number of alleles previously observed in the highly frequent classical class I haplotypes of Mauritian macaques (Aarnink et al. 2011; Budde et al. 2010). However, since only one copy of Mafa-A8*01:01 was detected by qPCR analysis (data not shown) it is likely that this allele is located on only one of the two HT1 chromosomes. Nevertheless, this suggests that it is possible to detect MHC homozygotes in the Filipino macaque population as well as the Mauritius population, which is the preferred population group for recruitment of animals into biomedical studies. In the class II region, the #7 and #11 are highly frequent haplotypes in Filipino macaques with 30.6 and 13.9 %, respectively (Blancher et al. 2014). This suggests that MHC homozygous animals might be easily detected by large scale screening by focusing on F/A/E/B-Hp1, F/A/E/B-Hp2, #7 and #11 in the Filipino macaque population. In this regard, it can be expected that the Mafa-A8*01:01 negative F/A/E/B-Hp1 and F/A/E/B-Hp2 will be preferentially selected for the simple reason that 90~95 % of these haplotypes are Mafa-A8*01:01 negative.

Fig. 4
figure 4

Gene organization of the MHC homozygous-like animal ID152. This animal is a MHC homozygous-like animal because it has both of F/A/E/B-Hp1 and #7 haplotype (HT1) in one chromosome and F/A/E/B-Hp1, #7 and Mafa-A8*01:01 haplotype (HT1+) in the other chromosome. White and gray boxes show common alleles between HT1 and HT1+, and Mafa-A8*01:01, respectively.

Full size image