Evolution of a new nonclassical MHC class I locus in two Old World primate species.

HLA-G is a nonclassical major histocompatibility complex (MHC) class I molecule that is expressed only in the human placenta, suggesting that it plays an important role at the fetal-maternal interface. In rhesus monkeys, which have similar placentation to humans, the HLA-G orthologue is a pseudogene. However, rhesus monkeys express a novel placental MHC class I molecule, Mamu-AG, which has HLA-G-like characteristics. Phylogenetic analysis of AG alleles in two Old World primate species, the baboon and the rhesus macaque, revealed limited diversity characteristic of a nonclassical MHC class I locus. Gene trees constructed using classical and nonclassical primate MHC class I alleles demonstrated that the AG locus was most closely related to the classical A locus. Interestingly, gene tree analyses suggested that the AG alleles were most closely related to a subset of A alleles which are the products of an ancestral interlocus recombination event between the A and B loci. Calculation of the rates of synonymous and nonsynonymous substitution at the AG locus revealed that positive selection was not acting on the codons encoding the peptide binding region. In exon 4, however, the rate of nonsynonymous substitution was significantly lower than the rate of synonymous substitution, suggesting that negative selection was acting on these codons.

Identification of a novel MHC class I gene, Mamu-AG, expressed in the placenta of a primate with an inactivated G locus.

Maternal tolerance of the fetal allograft remains poorly understood. In humans, expression of the highly polymorphic classical HLA-A and HLA-B loci is suppressed, while expression of the nonclassical HLA-G locus is up-regulated at the maternal-fetal interface. Like other nonclassical MHC class I molecules, HLA-G exhibits limited diversity, but certain characteristics of HLA-G distinguish it from other nonclassical MHC class I molecules: it has a truncated cytoplasmic domain, it is the product of alternatively spliced mRNAs, and it is expressed primarily in the placenta. We have examined MHC class I expression in the placenta of the rhesus monkey to determine whether this animal is a suitable model in which to study the function of HLA-G. Although the rhesus monkey possesses orthologs of many MHC class I and II loci found in humans, the HLA-G ortholog is a pseudogene in this nonhuman primate species. In this study, we report the identification of a novel nonclassical MHC class I locus expressed in the placenta of the rhesus monkey, Mamu-AG (Macaca mulatta-AG). Although unrelated to HLA-G, Mamu-AG encodes glycoproteins with all of the characteristics of HLA-G. These Mamu-AG glycoproteins are limited in their diversity, possess truncated cytoplasmic domains, are the products of alternatively spliced mRNAs, and their expression is restricted to the placenta. Taken together, these data suggest that convergent evolution may have resulted in the expression of a unique nonclassical MHC class I molecule in the rhesus monkey placenta, and that the common structural features of Mamu-AG and HLA-G may be functionally significant.

Further diversification of the HLA-B locus in Central American Amerindians: new B*39 and B*51 alleles in the Kuna of Panama.

Several new HLA-B locus alleles have been discovered in South American Amerindians. By contrast, analysis of the MHC class I alleles of North American native populations has revealed few new HLA-B alleles. This suggests that the HLA-B locus is evolving rapidly in South American populations. Here we describe the HLA-B locus alleles present in individuals from a Central American tribe, the Kuna of Panama. Using a sequence-based typing technique that separates alleles by denaturing gradient gel electrophoresis (DGGE) followed by direct sequencing, we determined the HLA-B alleles from eight Kunas. Two of the HLA-B alleles present in the Kuna have been previously described in other South American Amerindian populations; one allele has been characterized in a Mexican-American. We characterized two new HLA-B alleles in the Kuna, HLA-B*3911 and HLA-B*5110. HLA-B*3911 differed from HLA-B*3905 by only a single nucleotide substitution in exon 3. This substitution resulted in an amino acid replacement of leucine by arginine at residue 156 in the alpha 2 domain. Such a change may affect the repertoire of peptides that are bound by this molecule. HLA-B*5110 differed significantly from other HLA-B*51 alleles in that it is the result of an unusually large intra-locus recombination event of minimally 216 nucleotides. This recombination results in an allele that is part HLA-B*51 and part HLA-B*40. Thus, more dramatic recombination events may also play a role in the rapid evolution of the HLA-B locus in Amerindians.

HLA-B typing by allele separation followed by direct sequencing.

Due to the enormous allelic diversity of the HLA-B locus, it has been difficult to design an unambiguous molecular typing method for the alleles at this locus. Here we describe a technique for the direct sequencing of HLA-B alleles. Initially, HLA-B alleles were PCR-amplified after locus-specific reverse transcription of RNA. Alleles were then separated using denaturing gradient gel electrophoresis (DGGE), which separates DNA fragments based on their sequence composition. Amplification products were excised from the gel and eluted DNA was reamplified and directly sequenced. The derived sequences were aligned to a database of published HLA-B sequences, and an initial allele assignment was made. This approach was theoretically sufficient to type 92 of the 118 known HLA-B alleles. The majority of the remaining 26 alleles contain differences at the beginning of exon 2, a region outside the DGGE-separated PCR products. Therefore, we used heterozygous sequencing of this region to identify 19 of these 26 alleles, raising the resolution power to 111 alleles. Using this technique, we analyzed immortalized cell lines and blood samples from several different sources. Nine immortalized cell lines were obtained from the 10th International Histocompatibility Workshop (IHWS) and nine were derived from aboriginal peoples. Additionally, 25 blood samples were acquired from a panel of donors previously shown to be difficult to type using serological techniques. Altogether, using this new method of allele separation by DGGE followed by direct sequencing, we typed 52 different alleles from 57 individuals, covering 40 serological specificities.

Identification of a new HLA-B*08 variant, HLA-B*0804.

The HLA-B locus is the most polymorphic locus known with currently over 100 different alleles described. Many of these alleles encode variants of the serologically-defined tissue transplantation antigens. This high level of diversity makes accurate tissue typing difficult. Here we present the sequence of a new HLA-B*08 variant, HLA-B*0804, found in Caucasian siblings JH and PF serologically typed as HLA-B51/B59 and HLA-B59/B60, respectively. Additionally, DNA-based typing by the polymerase chain reaction using sequence-specific primers (PCR-SSP) identified HLA-B*51 in JH and HLA-B*4001 in PF. However, PCR-SSP failed to identify a second allele in either of these individuals. The unusual finding of a B59 antigen in a Caucasian and the discrepant molecular typing results suggested that these individuals might express novel HLA molecules. Using denaturing gradient gel electrophoresis (DGGE) followed by direct sequencing, we characterized a novel HLA-B*08 variant, HLA-B*0804. The presence of this allele was confirmed by cloning and sequencing. HLA-B*0804 differed from HLA-B*0801 by only one nucleotide substitution resulting in an amino acid replacement of phenylalanine by serine at position 67. Incidentally, this single nucleotide difference was sufficient to prevent amplification by PCR-SSP. This striking difference between both the serologically typed antigen and the PCR-SSP-identified allele compared to the sequenced allele supports the use of sequence-based typing for the analysis of HLA class I locus alleles.

A 25% error rate in serologic typing of HLA-B homozygotes.

The microlymphocytotoxicity technique has been the accepted method for HLA class I typing since the early 1960s. However, it is often difficult to distinguish two related alleles expressed in an individual due to the cross-reactive nature of the alloantibodies used in this technique. This is especially evident at the HLA-B locus, whose more than 180 alleles fall into only 4 major interrelated cross-reactive antigen groups. To estimate the error rate in serologic typing due to the cross-reactive nature of sera, we used polymerase chain reaction with sequence-specific primers (PCR-SSP) amplification to retype 40 individuals who were previously typed as serologic HLA-B locus homozygotes. PCR-SSP revealed that 10 of these 40 individuals (25%) were actually heterozygous at their HLA-B loci. The HLA-B locus alleles of 9 of these 10 discrepant individuals were further analyzed by denaturing gradient gel electrophoresis followed by direct sequencing. The sequence analysis confirmed that all nine individuals were indeed HLA-B locus heterozygotes. This surprisingly high error rate in serologic definition of HLA-B molecules argues for the use of rapid DNA-based techniques in HLA class I typing, even in the setting of solid organ transplantation.

Identification of the rhesus monkey HLA-G ortholog. Mamu-G is a pseudogene.

HLA-G is a nonclassical MHC class I molecule that is primarily expressed in the placenta. To investigate whether rhesus monkeys possess an HLA-G ortholog, we cloned and sequenced MHC class I cDNAs from the rhesus placenta. We identified two rhesus MHC class I cDNAs with sequence similarity to HLA-G. Each cDNA contained premature stop codons and frameshift mutations, suggesting that it was derived from an MHC class I pseudogene. Gene trees constructed using MHC class I alleles from human and nonhuman primates revealed that the rhesus placental pseudogene alleles clustered with HLA-G orthologs from the human, chimpanzee, and gorilla. These data suggested that this rhesus MHC class I pseudogene is an HLA-G ortholog. This locus was, therefore, designated Mamu (Macaca mulatta)-G. PCR amplification of a portion of Mamu-G from the genomic DNA of five rhesus monkeys resulted in the identification of five additional Mamu-G alleles and revealed the presence of four Mamu-G alleles in one rhesus monkey, suggesting that Mamu-G had been duplicated in this individual. Furthermore, the analysis of 81 MHC class I clones isolated from a rhesus placenta cDNA library did not result in the isolation of Mamu-G cDNAs, nor the isolation of any additional HLA-G homologs, suggesting that Mamu-G was transcribed at negligible levels. Given the similarity of rhesus monkey and human placenta structure and function, these data raise interesting questions regarding the role of HLA-G in pregnancy.