Logo of immunologyLink to Publisher's site
Immunology. Sep 2009; 128(1 Pt 2): e418–e431.
PMCID: PMC2753908

Evolution of the opossum major histocompatibility complex: evidence for diverse alternative splice patterns and low polymorphism among class I genes


The opossum major histocompatibility complex (MHC) shares a similar organization with that of non-mammals while containing a diverse set of class I genes more like that of eutherian (placental) mammals. There are 11 class I loci in the opossum MHC region, seven of which are known to be transcribed. The previously described Monodelphis domestica (Modo)-UA1 and Modo-UG display characteristics consistent with their being classical and non-classical class I genes, respectively. Here we describe the characteristics of the remaining five transcribed class I loci (Modo-UE, -UK, -UI, -UJ and -UM). All five genes have peptide-binding grooves with low or no polymorphism, contain unpaired cysteines with the potential to produce homodimer formation and display genomic organizational features that would be unusual for classical class I loci. In addition, Modo-UJ and -UM were expressed in alternatively spliced mRNA forms, including a potentially soluble isoform of Modo-UJ. Thus, the MHC region of the opossum contains a single class I gene that is clearly classical and six other class I genes each with its own unique characteristics that probably perform roles other than or in addition to antigen presentation.

Keywords: comparative immunology/evolution, marsupial, major histocompatibility complex


One of the goals of comparative genomics is to understand the functional roles genome organization plays and how it influences evolution. The major histocompatibility complex (MHC), which is among the most gene-dense and polymorphic regions in mammalian genomes, is an important example of a situation in which organization probably plays a significant role in the evolution and maintenance of genes. This may be particularly true for those genes involved in the MHC class I antigen processing and presentation pathway. These include the genes encoding the class I alpha chains that are receptors for antigenic peptides, the proteosome subunit beta (PSMB) involved in generating peptides, and transporter associated with antigen processing (TAP) that translocate peptides to the lumen of the endoplasmic reticulum.

Class I molecules evolve and diversify in species-specific ways with the result that they either perform the classical role of presenting peptides or have other ‘non-classical’ roles not associated with antigen presentation. Non-classical class I molecules such as human leucocyte antigen (HLA)-G have roles in immune regulation at sites of immune privilege such as the fetal–maternal interface.1 Examples of non-classical class I molecules that are involved in non-immune functions have also been reported. For example, non-classical class I molecules have recently been shown to contribute to brain development and plasticity in mice.2,3 In the MHC of non-mammalian species such as teleost fish and the frog Xenopus tropicalis, class I genes, TAP, and PSMB are located close together and appear to be maintained as linked, co-evolving genes that have given rise to haplotypic lineages. These linked class I genes are generally believed to perform the classical roles whereas the non-classical molecules in these species are encoded by unlinked genes outside the MHC. In eutherian (placental) mammals, in contrast, PSMB and TAP are separated from the class I genes. PSMB and TAP are located in the class II region which is separated from the class I genes by the class III region. Overall, the class II region has been relatively stable in eutherians and it is thought that this arrangement has allowed the eutherian class I genes to diversify by being separated from the PSMB and TAP genes, while remaining in the MHC.

In non-eutherian mammals, including marsupials such as opossums and kangaroos, and the monotremes such as the echidna and platypus, a more complicated view of MHC organization and evolution has emerged. In the platypus and echidna the MHC is not contiguous and is located within pseudoautosomal regions of two pairs of sex chromosomes.9 Platypus class I genes are located in close proximity to TAP, providing further evidence that proximity of the class I and antigen-processing genes is the ancestral organization. In the tammar wallaby (Macropus eugenii) the class I genes are not located within the MHC, but rather are scattered to 10 locations on six other autosomes.4 Oddly enough, these scattered wallaby class I genes share a high degree of sequence similarity with each other and resemble classical class I genes.4 So, in the case of the wallaby, moving out of the MHC or becoming unlinked to the TAP and PSMB genes does not appear to have resulted in significant diversification of the class I genes. In contrast, in another marsupial, the grey short-tailed opossum (Monodelphis domestica), there are 11 class I genes located in the MHC and these are in proximity to TAP/PSMB and are highly divergent. Of these 11, seven are known to be transcribed, two appear to be functional at the genomic level but transcripts have yet to be identified, and two appear to be pseudogenes. Interestingly, as in the frog, there appears to be only a single opossum class I gene, Monodelphis domestica (Modo)-UA1, with the characteristics of a classical class I gene.58 However, unlike the frog there are multiple other class I genes also linked to the MHC region that are highly divergent but whose roles are unknown. Only one of the other transcribed opossum class I genes, Modo-UG, has been characterized to any great extent and it was found to have characteristics more like those of a non-classical class I in being non-polymorphic and giving rise to multiple isoforms through alternative mRNA splicing.8 The goal of this study was to characterize the remaining class I loci in the opossum MHC, focusing on those class I genes that we knew were transcribed. The picture emerging from this study is that the opossum MHC region contains only one class I gene with clearly classical characteristics. The remaining six genes, which we previously found to be diverse in their nucleotide sequence,5 each encode a class I alpha chain with a unique set of characteristics.

Materials and methods

Animals and RNA extraction

The M. domestica opossums used in this study were derived from an admixture of two of the partially inbred opossum populations available, Populations 1 and 2, which are described elsewhere.10 The thymus from a 9-week-old male opossum was collected in RNA later® (Ambion, Austin, TX). This age was chosen as being a time-point at which the thymus is fully mature. Total RNA was extracted using Trizol following the manufacturer’s recommended protocols (Invitrogen, Carlsbad, CA) and total RNA was treated with TURBO DNA-free to remove contaminating DNA (Ambion). All experiments were approved by the Institutional Animal Care and Use Committee at the University of New Mexico (protocol number 07UNM005).

Rapid amplification of cDNA end (RACE)

To determine the full-length coding sequences for Modo-UE, -UI, -UJ, -UK and -UM, 5′ and 3′ RACE polymerase chain reactions (PCRs) were performed, using the GeneRacer Kit (Invitrogen) with Advantage™-HF 2 high-fidelity Taq polymerase (Clontech, Mountain View, CA) and the manufacturer’s recommended PCR conditions, on total thymus RNA extracted from a 9-week-old male individual of M. domestica. 5′ RACE PCR was performed in a single step for Modo-UK and Modo-UI, using the reverse 3′ RACE specific primer listed in Table 1, or as a nested PCR for Modo-UE, -UJ and -UM, using the reverse specific primers listed in Table 1 nested with reverse specific primers in exon 3 for Modo-UE, -UJ and -UM. 5′ RACE PCR was also performed as a single step for Modo-UK, using the forward 5′ RACE specific primer listed in Table 1, or as nested PCR using the forward specific primers listed in Table 1 nested with forward specific primers in exon 4 for each locus. Accession numbers for full-length mRNAs encoding the complete class I α chains of Modo-UE, -UI, -UJ, -UK and -UM including alternative mRNA forms are EU886706-EU886714. The primers described in Table 1 were based on sequences for each locus identified in the MonDom5 opossum genome assembly.5 The cDNA nucleotide sequences were aligned against the whole opossum genome assembly using the BLAST algorithm to identify the exons present in each clone.

Table 1
List of gene-specific primer sets used for amplification of class I cDNAs


Genomic DNA was isolated from M. domestica liver or spleen tissue provided by the Southwest Foundation for Biomedical Research (SFBR; San Antonio, TX) and the Division of Genomic Resources, Museum of Southwestern Biology (MSB), University of New Mexico (see Gouin et al.).8 DNA was extracted using a standard phenol/chloroform extraction protocol. The SFBR tissues were from captive-bred animals of Populations 1 and 2. MSB tissues were from wild individuals of M. domestica collected from four different locales in Bolivia (Brecha Tres, Porvenir, Santiago de Chiquitos and Tita). PCR fragments containing sequences of exon 2, intron 2 and exon 3 of the Modo-UE, -UI, -UJ, -UK and -UM genes were amplified using forward primers in intron 1 paired with reverse primers in intron 3. All PCR reactions were performed using High Fidelity Titanium Taq (BD Biosciences Clontech, Palo Alto, CA) and the manufacturer’s recommended PCR conditions with an annealing temperature of 62–64°. Six to 10 cloned PCR products from at least two independent PCR reactions per individual were sequenced. At least 10 independent clones were sequenced from apparent homozygous individuals to confirm homozygosity. Accession numbers are as follows: for Modo-UE alleles, EU886659-EU886668; for Modo-UI alleles, EU886643-EU886658; for Modo-UJ alleles, EU886694-EU886705; for Modo-UK alleles, EU886683-EU886693 and for Modo-UM alleles, EU886669-EU886682.


PCR, reverse transcriptase (RT)-PCR and RACE-PCR products were cloned into the pCR®4-TOPO vector using the TOPO TA Cloning® Kit for sequencing (Invitrogen). Primers for the T3 and T7 promoters were employed for sequencing using BigDye Terminator Cycle Sequencing Kit v3 (Applied Biosystems, Foster City, CA) with non-isotopic dye terminators in 10-μl reactions, according to the manufacturer’s instructions, and analysed on an ABI Prism 3100 DNA automated sequencer (Perkin Elmer, Waltham, MA). Chromatograms were edited manually using the sequencher 4·6 software (Gene Codes Corporation, Ann Arbor, MI) and were compared with sequences in the GenBank database using the BLAST algorithm.11 All sequences were aligned using the clustalx program.12 Nucleotide sequences were aligned and gapped manually using the bioedit program, based on the protein alignment to retain codon positions.


Transmembrane region prediction

The web-based software tmpred(http://www.ch.embnet.org/software/TMPRED_form.html) was used to predict the presence of transmembrane regions in cDNAs corresponding to opossum class I genes.

3D structure modelling

Homology models for Modo-UE, -UM and -UG were generated using the Swiss-Model homology modelling resource.1315 Because amino acid sequences of the antigenic peptides for the complexes have not been determined, models were generated without antigenic peptides localized within the peptide-binding groove.

The homology model for Modo-UG was subjected to nanosecond timescale solvated molecular dynamic simulations to ascertain the extent of motions present for regions of the Modo-UG protein thought to be involved in homodimer formation. For simplicity, only the α1, α2 and α3 domains (residues G21 to D204) of the Modo-UG homology model were simulated. Simulations were conducted using the AMBER8 suite of molecular dynamic simulation algorithms,16 a canonical ensemble and the ff03 force field.17 The conserved disulphide bond present in this domain was created between cysteines 121 and 185.

The model was initially energy-minimized by performing a steep-descent in vacuo minimization (100 steps, conjugate gradient) to relieve bad steric interactions. The model was then charge-neutralized and solvated using a TIP3P water box, and a 4-nanosecond solvated molecular dynamic simulation performed by first completing a restrained minimization (keeping the protein fixed) to relieve bad contacts in the surrounding solvent, and then carrying out an unrestrained minimization to relieve bad contacts in the entire system. This was followed by a 20-picosecond constant-pressure position-restrained dynamic simulation (weak restraints on the protein, raising the temperature from 0 to 300 K) to relax the position of the solvent molecules. The simulation was completed by performing a 4-nanosecond constant-pressure (1 atm) constant-temperature (300 K) molecular dynamic simulation with isotropic position scaling, Langevin dynamics with a collision frequency of 1·0/picosecond, and the SHAKE algorithm to constrain hydrogen bond lengths. The resultant refined Modo-UG model structure was analysed and validated using procheck.18


Gene organization and alternative splice variants

When the opossum genomic region was annotated, 11 class I loci were identified by homology-based searches using known marsupial and eutherian class I sequences.5 For the majority of these class I loci only the exons encoding the α1, α2 and α3 domains (exons 2, 3 and 4) could be identified because of a lack of RNA or cDNA information at the time. Using existing gene prediction software we were unable to reliably identify either the 5′ exon 1 encoding the leader peptide, or the 3′-most exons encoding transmembrane and cytoplasmic domains (MLB and RDM, unpublished observation).5 However, using the sequence information available for exons 2, 3 and 4 we were able to confirm that seven of the 11 class I genes were transcribed in the opossum thymus using RT-PCR.5,8 The complete cDNA sequence and therefore complete gene structure information were already available for two of these loci, Modo-UA1 and Modo-UG.7,8 To determine the complete gene structure of the remaining five transcribed loci, 5′ and 3′ RACE PCR was performed using thymus whole RNA as the template. Using the sequences of the clones amplified we were able to determine the complete exon and intron organization of each locus by aligning the complete cDNA sequences with the whole genome sequence of M. domestica (Fig. 1). The start location of each of the transcribed exons of Modo-UE, -UI, -UJ, -UK and -UM in the M. domestica whole genome assembly MonDom5 is provided in Table 2.

Table 2
Exon start locations for Monodelphis domestica (Modo)-UE, -UI, -UJ, -UK and -UM on chromosome 2 of MonDom5
Figure 1
Gene organization and exon composition of Monodelphis domestica (Modo)-UE, -UI, -UJ, -UK and -UM mRNAs. Modo-UJ and -UM are transcribed into alternatively spliced isoforms which differ in exon composition. The Modo-UJ locus consists of eight exons of ...

Single, full-length mRNAs encoding complete class I alpha chains were isolated for all five of the loci, allowing us to determine the complete gene boundaries. The Modo-UE, -UI and -UK genes have similar lengths (3·2–6 kb) and contain eight exons, typical of mammalian MHC class I genes. Modo-UK is unusual in that exon 8 encodes a lengthy cytoplasmic domain (65 amino acids) that contains three unpaired cysteines (Figs 1 and and2).2). For Modo-UJ and -UM additional alternative mRNA forms were isolated. Modo-UJ, the largest of the transcribed loci at approximately 17 kb, contains seven exons that are used to generate at least three alternative transcripts that would encode proteins varying in their transmembrane and cytoplasmic regions. The exons have been designated 1 to 5 and 6a and 6b, the latter two being used exclusively of each other. Transcripts that include sequence from exon 6a encode a transmembrane protein lacking a cytoplasmic tail, whereas those using the downstream exon 6b encode a cytoplasmic domain typical in length but lacking phosphorylation sites conserved in most MHC class I proteins. The third mRNA form, which we have termed UJsec, appears to be generated from a primary transcript that terminates in intron 5. Furthermore, this transcript utilizes an embedded canonical mRNA splice acceptor that is 3′ of the sequence encoding the transmembrane region site in exon 5. In UJsec, exon 4 is spliced to this internal acceptor site, and would encode a protein predicted to lack a transmembrane region and which could possibly be secreted, hence the designation UJsec. The open reading frame created includes 11 bp of intron 5 before encountering a stop codon followed by an additional 147 bp of intron 5 that is the 3′ untranslated region (UTR).

Figure 2
Cartoon of the predicted proteins encoded by Monodelphis domestica (Modo)-UE, -UI, -UK, -UJ and -UM, including alternative splice forms of Modo-UJ and Modo-UM. P indicates potential phosphorylation sites and SH designates unpaired cysteine residues.

The Modo-UM locus contains nine exons including two 3′ UTRs, these two UTRs being encoded by two different mRNA forms (Fig. 1). At least three alternative mRNA forms are encoded by the Modo-UM locus, including the full-length mRNA which has a typical eight-exon structure. A second mRNA form, Modo-UMΔ3, lacks the α3 domain through splicing out of exon 4. The third isoform, designated Modo-UM5b, is generated from a primary transcript that terminates in intron 5. Modo-UM5b skips the acceptor splice site at the end of exon 5 and continues to transcribe an open reading frame of 152 bp at the 5′ end of intron 5. This isoform is terminated by a stop codon in intron 5 followed by an additional 7 bp of intron 5 that is the 3′ UTR. Modo-UM5b has a shortened cytoplasmic domain as a result of the absence of exons 5, 6 and 7.

Deduced protein sequences and 3D structure modelling

An alignment of the full-length deduced protein sequences of Modo-UE, -UI, -UJ, -UK and -UM contains many of the features conserved in eutherian class I alpha chains,19 including (i) the cysteines involved in intrachain disulphide bonds in the α2 and α3 domains, (ii) the residues that interact with the CD8 co-receptor and (iii) a glycosylation site in the α1 domain (Fig. 3). Each of the M. domestica class I genes are also conserved in the region of the α1 domain implicated in interaction with natural killer (NK) receptors (Fig. 3).20 The cytoplasmic tails of Modo-UE, -UI and -UM contain the tyrosine and serine residues that are conserved phosphorylation sites in eutherian and marsupial class I molecules.7,19,21 Each of the Modo-UJ isoforms has an LI motif in a position corresponding to potential phosphorylation sites of other class I molecules. An LI motif is present in HLA-C and has been shown to serve a similar function in endoplasmic reticulum recycling to the tyrosine and serine residues of other class I genes.22

Figure 3
An alignment of the deduced protein translation of the seven transcribed opossum class I genes, including alternative isoforms of Monodelphis domestica (Modo)-UM and Modo-UJ and the previously described Modo-UA1 and Modo-UG. Human leucocyte antigen (HLA)-A ...

An unusual feature of all five predicted alpha chains is the presence of unpaired cysteines (Fig. 3). In Modo-UE there is an unpaired cysteine in the putative NK receptor binding motif. Modo-UM contains an unpaired cysteine in its α1 domain and another in its cytoplasmic region. Modo-UI and -UJ contain one and two unpaired cysteines, respectively, in their transmembrane regions and Modo-UK has three unpaired cysteines in its extended cytoplasmic domain (Figs 2 and and3).3). Unpaired cysteines are rare in classical class I molecules; however, their presence particularly in the α1 domain is associated with the formation of homodimers or trimers by non-classical class I molecules such as HLA-G and by HLA-B27, encoding an allele of the classical class I gene HLA-B.23,24 Unpaired cysteines were present at positions 74 and 77, respectively, of the peptide-binding grooves of all alleles of Modo-UE and -UM. When Modo-UE and -UM were superimposed on the crystal structure of HLA-B27 the unpaired cysteines in the opossum class I genes were oriented in a similar manner to those in HLA-B27, consistent with the possibility that they may also be involved in the formation of homodimers (Fig. 4a). An unpaired cysteine residue in the previously described opossum class I gene Modo-UG is in a similar position to that in human HLA-G.8 When Modo-UG was superimposed on the crystal structure of HLA-G, its unpaired cysteine was found to point away from the peptide-binding region (PBR) in an orientation where it could be available for homodimer formation, similar to that of HLA-G (Fig. 4b). Thus, it appears likely that the opossum class I genes Modo-UG, -UE and -UM have the potential for homodimer formation. What roles, if any, class I homodimers play in immune regulation and/or in the pathogenesis of disease in the opossum remains to be determined.

Figure 4
Predicted models of the α1 and α2 domains of (a) Monodelphis domestica (Modo)-UE (pink) and -UM (blue) superimposed on human leucocyte antigen (HLA)-B27 (yellow). Unpaired cysteines are shown in pink (Modo-UE), blue (Modo-UM) and yellow ...

Unpaired cysteines are also present in the transmembrane and cytoplasmic regions of several of the opossum class I genes. The unpaired cysteines of Modo-UJ and -UI are located in the transmembrane region where they are unlikely to be involved in homodimer formation. Modo-UK and -UM have unpaired cysteines in their cytoplasmic tails. These cysteines are located in positions where they could potentially be involved in homodimer formation, as described for several mouse class I genes.25


To examine the level of polymorphism in the putative PBRs of the five opossum class I genes, genomic DNA fragments containing exon 2, intron 2 and exon 3 were amplified, cloned and sequenced from captive-bred Brazilian and wild-caught Bolivian M. domestica. The presence of nucleotide substitutions and insertions/deletions in exons 2 and 3 and intron 2 was used to identify unique alleles at each locus (supplementary Fig. S1). The numbers of nucleotide alleles found for each of the loci were similar, ranging from 10 to 16, and not significantly different from that found previously for Modo-UG (Table 3).8 The majority of alleles for each locus are unique to one or the other of the two populations, Brazilian or Bolivian. Not surprisingly, those alleles in common between the two regions were usually also the most frequently isolated (Table 3).

Table 3
Alleles of Monodelphis domestica (Modo)-UE, -UI, -UK, -UJ and -UM present for captive-bred Brazilian and wild-caught Bolivian M. domestica genotyped

To determine if the polymorphism present at any of the loci would influence peptide binding, the deduced protein translations were aligned to HLA-A2 to establish corresponding peptide-binding residues. The majority of alleles appeared to be functional in that they had open reading frames. However, attempts to translate all of the alleles isolated revealed that one allele each for Modo-UK and -UJ contained inframe stop codons, probably resulting in null alleles (Table 3).

For Modo-UM the deduced amino acid translations from each allele revealed eight sites with non-synonymous substitutions in exons 2 and 3 combined, resulting in eight different protein sequences or alleles (supplementary Fig. S1). Only two non-synonymous substitutions occurred in the putative PBR, one of which, at position 67, was at a site that is typically polymorphic. However, the other, at position 159, was unusual in that there was either a leucine or histidine substituted for what is a conserved tyrosine in both classical and non-classical class I molecules (Table 4). For Modo-UE the deduced amino acid translations from each allele revealed 23 sites with non-synonymous substitutions or deletions/insertions in exons 2 and 3 combined, resulting in eight different protein sequences or alleles. A three-codon deletion in the α1 domain was present in six Modo-UE alleles (UE*01, 05, 06, 07, 09 and 11), including alleles from both captive and wild-caught individuals (supplementary Fig. S1). Two of the three codons deleted in several Modo-UE alleles corresponded to the locations of putative peptide-binding pockets and there were eight sites with non-synonymous substitutions, corresponding to those sites associated with the peptide-binding pockets in human class I molecules (Table 4). Although the most polymorphic of the loci analysed, Modo-UE is still substantially less polymorphic than the classical Modo-UA1. The deduced amino acid translation from each Modo-UJ allele revealed 10 non-synonymous substitutions in exons 2 and 3 combined, resulting in 10 different protein sequences. Four of the 10 non-synonymous substitutions occurred in putative PBR sites; however, two, at positions 7 and 24, are conservative substitutions at conserved sites.

Table 4
Comparison of the variation in amino acid residues in the peptide-binding pockets of Modo-UE, -UI, -UJ, -UK and -UM alleles in comparison to human leucocyte antigen (HLA)-A, HLA-G, Modo-UA1 and Modo-UG

The deduced amino acid translation from each Modo-UK allele revealed eight non-synonymous substitutions in exons 2 and 3 combined, resulting in a total of eight different protein sequences. One Modo-UK allele was identified only in a cDNA generated by RACE PCR (allele UK*12; supplementary Fig. S1). Only one of the eight non-synonymous substitutions present in Modo-UK was found in the putative peptide-binding pocket at a site that is generally polymorphic, position 63 (Table 4). Lastly, there were 12 sites with non-synonymous substitutions over exons 2 and 3, resulting in 12 different Modo-UI protein sequences. Two of these substitutions were in the putative PBR (Table 4). All five loci revealed a general absence of significant variation in residues thought to affect peptide binding and substantially less variation than in Modo-UA1. In no case did any of the non-synonymous PBR substitutions result in more than two different residues (Table 4).


The opossum MHC region is similar to the eutherian MHC in both size and gene complexity. However, it contains organizational features akin to those found in non-mammals, including the presence of the antigen transporter and processing genes TAP and PSMB in association with the class I genes.5 Foremost among the characteristics of classical class I genes is the presence of a highly polymorphic peptide-binding groove. The number of non-synonymous substitutions and/or deletions/insertions in the peptide-binding groove of class I genes is indicative of the diversity of peptides capable of being bound. The presence of no or very few non-synonymous substitutions is generally associated with non-classical class I molecules that have roles other than antigen presentation. The five opossum class I genes analysed here had between one and eight non-synonymous substitutions at sites corresponding to their putative peptide-binding pockets. The low polymorphism of these five opossum class I genes contrasted with the high levels of polymorphism observed among classical class I genes, including opossum Modo-UA1 and human HLA-A.8

The five opossum class I genes described here also showed numbers of alleles that were comparable to or only slightly higher than those of non-classical class I genes from humans.26Modo-UE and -UK have 11 and 12 nucleotide alleles, respectively, which encode eight different proteins, Modo-UI has 16 nucleotide alleles which produce 12 different proteins, Modo-UJ has 13 nucleotide alleles which encode 10 unique proteins and Modo-UM has 14 nucleotide alleles and eight proteins. By comparison, HLA-E has eight nucleotide alleles which produce only three different proteins; HLA-F has 20 nucleotide alleles and four unique proteins, and HLA-G has 23 nucleotide alleles which encode seven different proteins plus additional isoforms resulting from alternate splicing.26 In addition, several alleles of Modo-UE have a deletion corresponding to two of the putative peptide-binding sites. Such a deletion would result in the availability of fewer sites at which peptide binding could take place, thus decreasing the diversity of peptides that this molecule could potentially bind or indicating that this molecule is unable to bind peptide at all.

Although most classical class I loci are highly polymorphic, examples of classical class I loci with low polymorphism also exist. Human HLA-C has only 361 nucleotide alleles compared with the 673 alleles known for HLA-A and 1077 known for HLA-B (http://www.ebi.ac.uk/imgt/hla/allele.html). It is possible that the opossum class I genes described here are classical class I genes. However, in addition to low polymorphism, our results demonstrate that the opossum class I genes display other unusual features including atypical genomic organization, alternative mRNA splicing and the presence of unpaired cysteine residues with the potential for homodimer formation. These characteristics are more consistent with Modo-UE, -UI, -UJ, -UK and -UM playing roles other than antigen presentation.

Although the captive-bred opossums used in this study were partially inbred, the same populations of animals were also used for the analysis of Modo-UA1 which was previously found to have a highly polymorphic peptide-binding groove.18 Therefore it is unlikely that the low polymorphism of the five class I genes described here is an artifact of the partial inbreeding of the captive-bred populations. Despite close proximity to the antigen-processing genes, the single classical class I gene Modo-UA1 has evolved a highly polymorphic peptide-binding groove while the other six transcribed class I loci in the same region are less polymorphic yet display considerable sequence diversity between loci.

Polymorphism of MHC class I genes has been examined in only one other marsupial, the Tasmanian devil (Sarcophilus harrisii). Although the MHC region of this species has not yet been mapped, seven class I loci have been identified by PCR, five of which appear to be classical class I genes. A further two have tissue-restricted transcription patterns and low polymorphism, characteristics associated with non-classical class I genes.27,28 Thus, in contrast to the opossum, the devil has multiple class I loci with the potential to be involved in classical antigen presentation. The polymorphism of the devil class I loci appears to be low compared with that of the opossum classical class I gene Modo-UA1, probably as a result of the low overall genetic diversity of current devil populations.27

Modo-UJ and -UM and the previously described Modo-UG share with non-classical class I genes from other species the characteristic of being spliced into alternative mRNA isoforms.8 Phylogenetic analysis has previously revealed that Modo-UG, -UJ and -UM are not closely related to each other and that these loci form separate clades in a phylogenetic tree.5 This result indicates that these three loci evolved independently of one another rather than having arisen from a single gene that underwent alternative splicing. Three mRNA forms each of Modo-UJ and -UM were identified in thymus mRNA from the opossum. Included among the alternative isoforms of Modo-UJ was a potentially soluble isoform, Modo-UJsec, representing the first potentially soluble class I identified in a marsupial. The splice variants of Modo-UJ and -UM are similar to those present in other species, including humans, mice and non-human primates, consistent with the opossum isoforms not being artifacts but representing real transcripts. Similar to Modo-UJsec, soluble isoforms of human HLA-G have open reading frames in intron 4 (HLA-G5 and -G6) or in intron 2 (HLA-G7) which terminate with a stop codon, deleting the fifth and sixth exons.1 In the rhesus monkey placenta, an isoform of Mamu-AG (Mamu-AG5) that retains intron 4 as previously noted in HLA-G5 has been identified. A second alternative splice variant, Mamu-AGv5, retains 18 nucleotides derived from the 3′ end of the fifth intron.29 Modo-UJsec is structurally most similar to soluble human HLA-G5 and rhesus monkey Mamu-AG5. Structural similarities between other alternate splice variants from the opossum and those from other species also exist. Modo-UMΔ3 splices out exon 3 to encode a protein similar in structure to human HLA-G2 and mouse H2.Bl.2.30,31 Modo-UJ6a and Modo-UM5b each have shortened cytoplasmic domains and encode proteins similar in structure to full-length human HLA-G and rhesus monkey Mamu-AG.1,32 Thus, the splice variants identified in the opossum are similar to those found in eutherian mammals, consistent with their roles having evolved prior to the separation of eutherian and metatherian mammals. What role secretory class I genes have remains to be determined; however, they have been found to be critical for the establishment of pregnancy in eutherian mammals.33 Also, whether or not these different alternative splice variants demonstrate tissue-specific expression remains to be determined.

Within the opossum MHC region there is a single classical class I gene and six other class I genes with a diverse set of characteristics more typical of non-classical class I loci. This organization is remarkable when compared with that of eutherian mammals and non-mammalian vertebrates. In the opossum, the antigen processing and transporter genes (PSMB8, PSMB9, TAP1 and TAP2) are located within the MHC region in close proximity to the diverse set of class I genes. In most eutherian mammals, the antigen-processing machinery is located in the class II region, while the classical and non-classical class I genes are encoded in another region of the MHC. In non-mammalian vertebrates and in the rat, the classical class I and antigen-processing genes are in close proximity to each other and appear to be coevolving, whereas the non-classical class I loci are located in a different region of the MHC.3436 Thus, the opossum MHC region displays a unique organization with a diverse set of class I loci, the majority of which display characteristics unusual for classical class I genes, all evolving in close proximity to the antigen-processing machinery.

Close proximity of antigen processing and presenting genes is believed to result in the evolution of class I molecules that can specifically bind peptides loaded by the adjacent TAP and is believed to be the basis for the single dominantly expressed class I molecule in chickens and other non-mammalian vertebrates.37 In the opossum MHC region, Modo-UA1, the opossum classical class I gene, is adjacent to TAP2A, making it possible that these two genes are coevolving. An association between disease resistance and coevolution of TAP and classical class I genes has been observed in chickens and rats. In chickens, coevolution of TAP and classical class I genes appears to have led to the evolution of class I alleles with particular peptide-binding specificities.37 A similar situation exists in rats, where the proximity of TAP and classical class I genes appears to have led to their coevolution, resulting in class I molecules that can bind a specific set of peptides.38 This type of coevolution allows antigen presentation to be efficiently carried out between certain combinations of TAP and class I alleles and may be selecting to preserve linkage of these genes. Such tight linkage can also limit functional class I genes to those that can accept peptides from the adjacent transporter. Thus, many vertebrates may be resistant or susceptible to certain pathogens ultimately because of the genetic organization of their MHC.39

Class I molecules require peptide to be bound to the peptide-binding groove for cell surface expression and stabilization. Non-classical class I molecules, including human HLA-G and mouse Qa-2, acquire peptide in a TAP-dependent manner, similar to the peptide acquisition of classical class I molecules.30,40,41 Thus, close proximity of antigen-processing and non-classical class I genes has the potential to affect the ability of these molecules to bind particular sets of peptides in a similar manner to classical class I molecules from other species.37,38 In the opossum genome, the antigen-processing genes TAP1, PSMB9 and PSMB8 are flanked by Modo-UG and -UI, two loci with features typical of non-classical class I loci but with strikingly different characteristics from each other. Although both Modo-UG and -UI have monomorphic peptide-binding grooves, few residues in the PBR are conserved between these two loci, consistent with them binding different sets of peptides and/or playing different roles in immune regulation. Therefore it is unlikely that both Modo-UG and -UI are coevolving with the antigen-processing machinery to which they are adjacent. Evidence, at least from these two genes, indicates that linkage to the antigen-processing machinery does not appear to have constrained the evolution of class I genes in the opossum.

The organization of the opossum MHC, which is likely to be the ancestral organization, does not appear to be common to all marsupials.4,5 Unlike the opossum, in which all but two class I genes are located within the MHC, in the tammar wallaby, class I genes are located outside the MHC in 10 locations scattered on six different autosomes and TAP2 has been localized to the class II/III region. Similar to the tammar wallaby, the MHC of monotremes is not contiguous and locates within pseudoautosomal regions of two pairs of sex chromosomes. Platypus class I genes are located in close proximity to TAP, providing further evidence that proximity of the class I and antigen-processing genes is the ancestral organization.9 As linkage of TAP and class I genes appears to represent the ancestral organization of the MHC region, the tammar wallaby class I genes probably moved out of the MHC after the divergence of Australian and American marsupials approximately 65 Ma.42,43 The tammar class I genes also share a high degree of sequence similarity with each other, most sharing > 80% nucleotide identity.4 Similarly, two opossum class I genes, Modo-UB and -UC, which are located outside the MHC region, show little evidence of diversification and are similar in sequence to Modo-UA1.5,21 Thus, evidence from the opossum and tammar wallaby demonstrates that diversification of class I genes does not appear to be restricted by close proximity to the antigen-processing machinery.

The opossum provides an important model with which to examine the evolution of the MHC region. At least in the opossum, class I genes are diversifying while being tightly linked to the antigen-processing machinery genes PSMB and TAP. The differences in the organization of the MHC between the opossum and tammar wallaby and between closely related eutherians probably reflect the rapidly evolving nature of this region and highlight the need for comparative studies to provide insight into the evolution of the vertebrate immune system. The opossum provides an important model for understanding how organization affects the evolution of genes. Whether or not Modo-UE, -UG, -UI, -UJ, -UK and -UM are indeed non-classical class I molecules remains to be determined. Another characteristic associated with non-classical class I genes is tissue-specific expression patterns. Indeed, we have found that only Modo-UA1, the apparent classical class I gene, and Modo-UK have clear ubiquitous transcription; all others showed some degree of tissue-restricted transcription patterns (MLB, SDM and RDM, unpublished observation). Further analysis of the functional roles and cell surface expression of each of the opossum class I loci will provide additional insights into how genomic organization influences the evolution of class I genes and the roles of antigen-processing genes in shaping their evolution.


This publication was made possible by support from a National Institutes of Health grant (no. IP20RR18754) from the Institutional Development Award (IDeA) program of the National Centre for Research Resources and a National Science Foundation award (IOB-0641382).



Monodelphis domestica
peptide binding region
proteasome subunit beta
transporter associated with antigen processing
untranslated region

Supporting information

Additional Supporting Information may be found in the online version of this article:

Figure S1. Nucleotide alignment of exon 2 to exon 3 of the Modo-UE, -UI, -UJ, -UK and -UM alleles. The deduced amino acid sequence corresponding to exons 2 and 3 is shown above and places where there are nonsynonymous substitutions are indicated below the alignment. Sites where stop codons are generated either due to frame-shift mutations or substitutions are indicated by a *. Dashes indicate identity. Gaps are inserted to maintain reading frame in coding regions and are indicated by dots. 1. Modo-UK allele UK*12 was identified only in a cDNA isolated by RACE PCR and therefore contains no intron sequence.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than about missing material) should be directed to the corresponding author for the article.


1. Hunt JS, Petroff MG, McIntire RH, Ober C. HLA-G and immune tolerance in pregnancy. FASEB J. 2005;19:681–93. [PubMed]
2. Huh GS, Boulanger LM, Du H, Riquelme PA, Brotz TM, Shatz CJ. Functional requirement for class I MHC in CNS development and plasticity. Science. 2000;290:2155–9. [PMC free article] [PubMed]
3. Syken J, Grandpre T, Kanold PO, Shatz CJ. PirB restricts oculardominance plasticity in visual cortex. Science. 2006;313:1795–800. [PubMed]
4. Deakin JE, Siddle HV, Cross JGR, Belov K, Graves JAM. Class I genes have split from the MHC in the tammar wallaby. Cytogenet Genome Res. 2007;116:205–11. [PubMed]
5. Belov K, Deakin JE, Papenfuss AT, et al. Reconstructing an ancestral mammalian immune supercomplex from a marsupial MHC. PLoS Biol. 2006;4:e46. [PMC free article] [PubMed]
6. Flajnik MF, Ohta Y, Greenberg AS, Salter-Cid L, Carrizosa A, Du Pasquier L, Kasahara M. Two ancient allelic lineages at the single classical class I locus in the Xenopus MHC. J Immunol. 1999;163:3826–33. [PubMed]
7. Miska KB, Miller RD. Marsupial MHC class I: classical sequences from the opossum, Monodelphis domestica. Immunogenetics. 1999;50:89–93. [PubMed]
8. Gouin N, Wright AM, Miska KB, Parra ZE, Samollow PB, Baker ML, Miller RD. Modo-UG, a marsupial nonclassical MHC class I locus. Immunogenetics. 2006;58:396–406. [PubMed]
9. Dohm JC, Tsend-Ayush E, Reinhardt R, Grutzner F, Himmelbauer H. Disruption and pseudoautosomal localisation of the major histocompatibility complex in monotremes. Genome Biol. 2007;8:R175. [PMC free article] [PubMed]
10. VandeBerg JL, Robinson ES. The laboratory opossum (Monodelphis domestica) in biomedical research. In: Saunders NR, Hinds LA, editors. Marsupial Biology: Recent Research, New Perspectives. Sydney: University of New South Wales Press; 1997. pp. 238–53.
11. Altschul SF, Gish W, Miller W, Myers EM, Lipman DJ. Basic local alignment search tool. J Mol Evol. 1990;215:403–10. [PubMed]
12. Thompson JD, Higgins DG, Gibson TJ. CLUSTAL W: improving the sensitivity of the progressive multiple sequence alignment through sequence weighting, position specific gap penalties and weight matrix choice. Nucleic Acids Res. 1994;22:4676–80. [PMC free article] [PubMed]
13. Peitsch MC. Protein modeling by E-mail. Biotechnology. 1995;13:658–60.
14. Guex N, Peitsch MC. SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling. Electrophoresis. 1997;18:2714–23. [PubMed]
15. Schwede T, Kopp J, Guex N, Peitsch MC. SWISS-MODEL: an automated protein homology-modeling server. Nucleic Acids Res. 2003;31:3381–5. [PMC free article] [PubMed]
16. Case DA, Darden TA, Cheatham TE, III, et al. AMBER 8. San Francisco, CA: University of California; 2004.
17. Duan Y, Wu C, Chowdhury S, et al. A point-charge force field for molecular mechanics simulations of proteins. J Comput Chem. 2003;24:1999–2012. [PubMed]
18. Laskowski RA, Rullmannn JA, MacArthur MW, Kaptein R, Thornton JM. AQUA and PROCHECK-NMR: programs for checking the quality of protein structures solved by NMR. J Biomol NMR. 1996;8:477–86. [PubMed]
19. Kaufman J, Salomonsen J, Flajnik M. Evolutionary conservation of MHC class I and class II molecules – different yet the same. Semin Immunol. 1994;6:411–24. [PubMed]
20. Mandelboim O, Reyburn HT, ValesGomez M, Pazmany L, Colonna M, Borsellino G, Strominger JL. Protection from lysis by natural killer cells of group 1 and 2 specificity is mediated by residue 80 in human histocompatibility leukocyte antigen C alleles and also occurs with empty major histocompatibility complex molecules. J Exp Med. 1996;184:913–22. [PMC free article] [PubMed]
21. Miska KB, Wright AM, Lundgren R, Sasaka-McClees R, Osterman A, Gale JM, Miller RD. Analysis of a marsupial MHC region containing two recently duplicated class I loci. Mamm Genome. 2004;15:851–64. [PubMed]
22. Park B, Lee S, Kim E, Chang S, Jin M, Ahn K. The truncated cytoplasmic tail of HLA-G serves as a quality-control function in post-ER compartments. Immunity. 2001;15:213–24. [PubMed]
23. Allen RL, O’Callaghan CA, McMichael AJ, Bowness P. HLA-B27 can form a novel β2-microglobulin free heavy chain homodimer structure. J Immunol. 1999;162:5045–8. [PubMed]
24. Apps R, Gardner L, Sharkey AM, Holmes N, Moffett A. A homodimeric complex of HLA-G on normal trophoblast cells modulates antigen-presenting cells via LILRB1. Eur J Immunol. 2007;37:1924–37. [PMC free article] [PubMed]
25. Capps GG, Robinson BE, Lewis KD, Zuniga MC. In vivo dimeric association of class I MHC heavy chains. J Immunol. 1993;151:159–69. [PubMed]
26. Sullivan LC, Hoare HL, McCluskey J, Rossjohn J, Brooks AG. A structural perspective on MHC class Ib molecules in adaptive immunity. Trends Immunol. 2006;27:413–20. [PubMed]
27. Siddle HV, Kreiss A, Eldridge MDB, Noonan E, Clarke CJ, Pyecroft S, Woods GM, Belov K. Transmission of a fatal clonal tumor by biting occurs due to depleted MHC diversity in a threatened carnivorous marsupial. Proc Natl Acad Sci USA. 2007a;104:16221–6. [PMC free article] [PubMed]
28. Siddle HV, Sanderson C, Belov K. Characterization of major histocompatibility complex class I and class II genes from the Tasmanian devil (Sarcophilus harrisii) Immunogenetics. 2007b;59:753–60. [PubMed]
29. Ryan AF, Grendell RL, Geraghty DE, Golos TG. A soluble isoform of the rhesus monkey nonclassical MHC class I molecule Mamu-AG is expressed in the placenta and testis. J Immunol. 2002;169:673–83. [PubMed]
30. LeMaoult J, Le Discorde M, Rouas-Freiss N, Moreau P, Menier C, McCluskey J, Carosella ED. Biology and functions of human leukocyte antigen-G in health and sickness. Tissue Antigens. 2003;62:273–84. [PubMed]
31. Guidry PA, Stroynowski I. The murine family of gut restricted class Ib MHC includes alternatively spliced isoforms of the proposed HLA-G homolog, “blastocyst MHC” J Immunol. 2005;175:5248–59. [PubMed]
32. Boyson JE, Iwanga KK, Golos TG, Watkins DI. Identification of a novel MHC class I gene, Mamu-AU, expressed in the placenta of a primate with an inactivated G locus. J Immunol. 1997;159:3311–21. [PubMed]
33. Fuzzi B, Rizzo R, Criscuoli L, et al. HLA-G expression in early embryos is a fundamental prerequisite for the obtainment of pregnancy. Eur J Immunol. 2002;32:311–5. [PubMed]
34. Hurt P, Walter L, Sudbrak R, et al. The genomic sequence and comparative analysis of the rat major histocompatibility complex. Genome Res. 2004;14:631–9. [PMC free article] [PubMed]
35. Moon DA, Veniamin SM, Parks-Dely JA, Magor K. The MHC of the duck (Anas platyrhynchos) contains five differentially expressed class I genes. J Immunol. 2005;175:6702–12. [PubMed]
36. Flajnik MF, Kasahara M. Comparative genomics of the MHC: glimpses into the evolution of the adaptive immune system. Immunity. 2001;15:351–62. [PubMed]
37. Kaufman J. Co-evolving genes in MHC haplotypes: the “rule” for non-mammalian vertebrates? Immunogenetics. 1999;50:228–36. [PubMed]
38. Joly E, Le Rolle A-F, Gonzalez AL, et al. Co-evolution of rat TAP transporters and MHC class I RT1-A molecules. Curr Biol. 1998;8:169–72. [PubMed]
39. Kaufman J. The simple chicken major histocompatibility complex: life and death in the face of pathogens and vaccines. Phil Trans R Soc Lond B Biol Sci. 2000;355:1077–84. [PMC free article] [PubMed]
40. Diehl M, Munz C, Keilholz W, Stevanovic S, Holmes N, Loke YW, Rammansee H. Nonclassical HLA-G molecules are classical peptide presenters. Curr Biol. 1996;6:305. [PubMed]
41. Joyce S, Tabaczewski P, Angeletti RH, Nathenson SG, Stroynowski I. A nonpolymorphic major histocompatibility complex class Ib molecule binds a large array of diverse self-peptides. J Exp Med. 1994;179:579–88. [PMC free article] [PubMed]
42. Kirsch JAW, Lapointe F, Springer MS. DNA-hybridisation studies of marsupials and their implications for metatherian classification. Aust J Zool. 1997;45:211–80.
43. Nilsson MA, Arnason U, Spencer PBS, Janke A. Marsupial relationships and a timeline for marsupial radiation in South Gondwana. Gene. 2004;340:189–96. [PubMed]
44. Marsh SGE, Parham P, Barber LD. The HLA Factsbook. San Diego, CA: Academic Press; 2000.

Articles from Immunology are provided here courtesy of British Society for Immunology
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...