Classical genes within the Major Histocompatibility Complex (MHC) are responsible for peptide presentation to T cells, thus playing a central role in immune defense against pathogens. These genes are subject to strong selective pressures including both balancing and directional selection, resulting in exceptional genetic diversity—thousands of alleles per gene in humans. Moreover, some allelic lineages appear to be shared between primate species, a phenomenon known as trans-species polymorphism (TSP) or incomplete lineage sorting, which is rare in the genome overall. However, despite the clinical and evolutionary importance of MHC diversity, we currently lack a full picture of primate MHC evolution. In particular, we do not know to what extent genes and allelic lineages are retained across speciation events. To start addressing this gap, we explore variation across genes and species in our companion paper (Fortier and Pritchard, 2025), and here we explore variation within individual genes. We used Bayesian phylogenetic methods to determine the extent of TSP at 17 MHC genes, including classical and non-classical Class I and Class II genes. We find strong support for ancient TSP in 7 of 10 classical genes, including—remarkably—between humans and old-world monkeys in MHC-DQB1. In addition to the long-term persistence of ancient lineages, we additionally observe rapid evolution at nucleotides encoding the proteins’ peptide-binding domains. The most rapidly-evolving amino acid positions are extremely enriched for autoimmune and infectious disease associations. Together, these results suggest complex selective forces—arising from differential peptide binding—that drive short-term allelic turnover within lineages while also maintaining deeply divergent lineages for at least 31 million years in some cases.
primatesMajor Histocompatibility Complextrans-species polymorphismhuman leukocyte antigenallelic lineageResearch organismHumanPrimatehttp://dx.doi.org/10.13039/100000002National Institutes of HealthR01 HG011432FortierAlyssa LynPritchardJonathan Khttp://dx.doi.org/10.13039/100000002National Institutes of HealthR01 HG008140FortierAlyssa LynPritchardJonathan Khttp://dx.doi.org/10.13039/100000001National Science FoundationDGE-1656518FortierAlyssa LynThe funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.Author impact statementTrans-species polymorphism at primate Major Histocompatibility Complex genes is old, especially in the 'classical' genes, while rapidly-evolving regions of each gene correspond with proteins' functional domains.publishing-routeprcIntroduction
The Major Histocompatibility Complex (MHC) is a large locus containing many immune genes that is shared among the jawed vertebrates (Radwan et al., 2020). In humans, the MHC is also known as the HLA (Human Leukocyte Antigen) region; it spans about 5 megabases (Mb) on chromosome 6 and contains 412 genes (Figure 1A; Genome Reference Consortium, 2022; O’Leary et al., 2016). Many of these are part of the MHC gene family, a large group of evolutionarily related genes with varying functions. The ‘classical’ MHC genes are responsible for presenting protein fragments for inspection by T cells. MHC peptide presentation allows T cells to monitor the body for the presence of foreign peptides, which might indicate infection or cancer; this is crucial for vertebrate immune surveillance (Neefjes et al., 2011). ‘Non-classical’ MHC genes are essential to the innate immune system, where they perform a variety of niche roles. See the appendices of our companion paper (Fortier and Pritchard, 2025) for more detail.
The MHC region in humans (HLA).
(A) Each point at top represents the location of a gene. The different types of HLA genes are distinguished by different colors, shown in the key at left. The 19 functional HLA genes are labeled with their name (omitting their ‘HLA’ prefix due to space constraints). Gray points represent non-HLA genes and pseudogenes in the region. The black line shows nucleotide diversity (Nei and Li’s π) across the region, while the pink horizontal line shows the genome-wide average nucleotide diversity (π≈0.001\begin{document}$ \pi\approx 0.001$\end{document}) (Sachidanandam et al., 2001). (B) Nucleotide diversity around classical Class I gene HLA-A, with exon structure shown. (C) Nucleotide diversity around classical Class II gene HLA-DRB1, with exon structure shown. (D) Species tree showing the phylogenetic relationships among selected primates from this study (Kuderna et al., 2023). The colors of the icons are consistent with colors used throughout the paper to distinguish species. The pink vertical dashed lines indicate split times of the new-world monkeys (NWM) from the apes/old-world monkeys (OWM) (39 MYA), OWM from the apes (31 MYA), and the lesser apes (gibbons) from the great apes (23 MYA).
Nucleotide diversity in the human HLA region.
Nucleotide diversity (Nei and Li’s π) around each HLA gene is shown in black, while the pink horizontal line shows the genome-wide average nucleotide diversity (π≈0.001\begin{document}$ \pi\approx 0.001$\end{document}) (Sachidanandam et al., 2001). The genes are shown in order along the genome (from top left to bottom right), but the x-axis is repeatedly broken in order to zoom in on detail around each gene. The genes are colored according to their type, with key shown at bottom right. For the functional genes (and some pseudogenes), the exon structure is shown by boxes; other pseudogenes are not annotated with this level of detail.
Species key.
On the left-hand side is the species tree relating the species used in this study (Kuderna et al., 2023; Foley et al., 2023). Each species/tip is labeled with a unique color and 4-letter abbreviation, which is composed of the first two letters of the genus name and first two letters of the species name. The common name and Latin name for each species is shown on the right-hand side. Plecturocebus moloch is listed in the IPD-MHC database under its old name, Callicebus moloch, and uses a different abbreviation (Camo) in that resource. Similarly, Leontocebus fuscicolis was formerly known as Saguinus fuscicollis (Safu) in the IPD-MHC database. There appears to be some debate as to whether the pygmy marmoset should be placed in the Callithrix or Cebuella genus, but we have used the name Cebuella pygmaea (Cepy) in accordance with a recent primate study (Kuderna et al., 2023). This species is known as Callithrix pygmaea (Capy) in IPD-MHC. OWM, Old-World Monkeys; NWM, New World Monkeys; Str., Strepsirrhini; Gli., Glires; Lau., Laurasiatheria; Atl., Atlantogenata.
The MHC locus is extraordinarily polymorphic. Haplotypes can vary widely in gene content, and thousands of distinct alleles are observed at the classical genes in humans and other primates (Maccari et al., 2017; Maccari et al., 2020; Robinson et al., 2019). Different alleles are functionally diverse, with distinct peptide-binding affinities and, consequently, allelic differences in pathogen detection (Neefjes et al., 2011; Adams and Luoma, 2013). Given this huge diversity of functionally distinct alleles, the MHC is by far the most important locus in the genome for inter-individual variation in both infectious and autoimmune disease risk, with thousands of GWAS hits (Buniello et al., 2019; Smith et al., 2024). In our companion paper (Fortier and Pritchard, 2025), we built large multi-gene trees to explore the relationships between the different classical and non-classical genes. Here, we look within 17 specific genes—representing classical, non-classical, Class I, and Class II —to characterize trans-species polymorphism, a phenomenon characteristic of long-term balancing selection.
Historically, the MHC provided some of the first clear examples of positive selection in early studies of molecular evolution. By the 1980s and 1990s, researchers had noted an excess of missense variants (i.e. dN/dS>1\begin{document}${dN}/{dS} \gt 1$\end{document}) in the peptide-binding regions of classical MHC genes (Hughes and Nei, 1988; Hughes and Nei, 1989), alleles shared across species (Arden and Klein, 1982; Mayer et al., 1988), and high nucleotide diversity across the region (Wakeland et al., 1987; Nei and Hughes, 1991) in rodents and primates. Indeed, modern data show that nucleotide diversity in the human MHC (HLA region) exceeds 70 times the genome-wide average near the classical genes, suggesting ancient balancing selection (Figure 1A–C). Meanwhile, the MHC also features prominently in genome-wide scans for short-term directional selection (Mathieson et al., 2015; Field et al., 2016; Allentoft et al., 2022; Cong et al., 2022; Okada et al., 2018; Yasumizu et al., 2020).
In the present paper, we explore a particularly striking feature of the selection signals at MHC, namely the evidence for extremely deep coalescence structure. Some alleles (haplotypes) are more closely related to corresponding alleles from another species than they are to distinct alleles from their own species. This phenomenon is referred to as trans-species polymorphism (TSP).
TSP is rare overall in humans. Across most of the genome, human alleles coalesce to a common ancestor well within the human lineage, typically around 2 million years (MY) ago (Mallick et al., 2016). Indeed, only ∼100 loci genome-wide show compelling evidence for sharing of ancestral alleles between humans and our closest relatives, chimpanzees (Leffler et al., 2013). TSP among humans and more distantly related species is even rarer; besides the MHC, the only other clear example of deep TSP is at the ABO locus (which influences blood type; Azevedo et al., 2015). At this locus, both the A and B alleles are shared by descent throughout the apes, implying that the A and B lineages date back to at least the divergence point of humans and gibbons 23 MY ago (Ségurel et al., 2012; Kuderna et al., 2023). Such deep coalescence is extraordinarily unlikely under a neutral model, and instead points to some form of balancing selection.
Meanwhile, TSP is evident at multiple MHC genes and in many different phylogenetic clades. TSP at this locus was first proposed in the 1980s on the basis of unusual sequence similarity between mice and rats (Klein, 1980; Arden and Klein, 1982; Klein, 1987; Figueroa et al., 1988; McConnell et al., 1988; Wakeland et al., 1987), and between humans and chimpanzees (Lawlor et al., 1988; Mayer et al., 1988). Later work has reported likely TSP between humans and apes (Slierendregt et al., 1995; Boyson et al., 1996; McKenzie et al., 1999; Wroblewski et al., 2017) and humans and old world monkeys (Mayer et al., 1992; Brändle et al., 1992; Kupfermann et al., 1992; Slierendregt et al., 1992; Geluk et al., 1993; Satta et al., 1996; Otting et al., 2000; Kriener et al., 2000; Gyllensten et al., 1990; Kriener et al., 2001; Otting et al., 1992; Otting and Bontrop, 1995; Otting et al., 1992; Otting et al., 2002); deep TSP is also consistent with the high levels of genetic diversity within the MHC. Such ancient TSP would make the MHC unique compared to any other locus in the genome. However, most previous work has not fully accounted for the inherent uncertainty in phylogenetic inference, especially given the potential for convergent evolution at functional sites. Although there is clear evidence for TSP, its exact age at each gene is still uncertain.
To address these questions, we used data from the IPD-MHC/HLA database—a large repository for MHC allele sequences from humans, non-human primates, and other vertebrates—along with supplementary sequences from NCBI RefSeq (Maccari et al., 2017; Maccari et al., 2020; Robinson et al., 2019). This represents the most complete sampling of primate MHC genes to date, spanning the entire primate tree (Figure 1D; Tables 2–4). We account for the uncertainty in phylogenetic inference using a Bayesian MCMC approach (BEAST2), which is well-suited to handle highly variable and rapidly-evolving sequences. In our companion paper (Fortier and Pritchard, 2025), we built trees to compare genes across dozens of species. When paired with previous literature, these trees helped us infer orthology and assign sequences to genes in some cases. That process helped inform this work, where we assess support for TSP within individual genes.
We find support for TSP among the African apes for genes MHC-C, -DPA1, and -DRB3, among the great apes for MHC-DPB1, and among all apes for MHC-B. We also find conclusive evidence for TSP at least back to the ancestor of humans and OWM in MHC-DQB1, implying—remarkably—that allelic lineages have been maintained by balancing selection for at least 31 MY. Rapidly-evolving sites are mainly located in the critical peptide-binding regions of the classical genes, but are spread throughout the coding region of the non-classical genes. Moreover, the most rapidly-evolving sites are also frequently associated with immune phenotypes and diseases in the literature, connecting our evolutionary findings with their functional consequences. These results highlight the contrasting roles of ancient balancing selection and short-term directional selection within the peptide-binding regions of the classical genes and motivate further evolutionary and functional studies to better understand this unique system.
ResultsData
We collected MHC nucleotide sequences for all genes from the IPD-MHC/HLA database, a large repository for MHC alleles from humans, non-human primates, and other vertebrates (Maccari et al., 2017; Maccari et al., 2020; Robinson et al., 2024). Although extensive, this database includes few or no sequences from important primates such as the gibbon, tarsier, and lemur. Thus, we supplemented our set of alleles using sequences from NCBI RefSeq (O’Leary et al., 2016). Because the MHC genes make up an evolutionarily related family, they can all be aligned (Kaufman, 2022; Adams and Luoma, 2013). In our companion paper (Fortier and Pritchard, 2025), we utilized these large multi-gene alignments for Class I, Class IIA, and Class IIB to compare genes. Here, we analyze subsets of those alignments, each focusing on a single gene or group of closely related genes.
We considered 16 gene groups spanning MHC classes and functions. These include the classical Class I genes (MHC-A-related, MHC-B-related, MHC-C-related), non-classical Class I genes (MHC-E-related, MHC-F-related, MHC-G-related), classical Class IIA genes (MHC-DRA-related, MHC-DQA-related, MHC-DPA-related), classical Class IIB genes (MHC-DRB-related, MHC-DQB-related, MHC-DPB-related), non-classical Class IIA genes (MHC-DMA-related, MHC-DOA-related), and non-classical Class IIB genes (MHC-DMB-related, MHC-DOB-related). See Tables 2–5 for a breakdown of the sequences from each species included in each group. We studied two or three different genic regions for each group: exon 2 alone, exon 3 alone, and (for Class I) exon 4 alone. Exons 2 and 3 encode the peptide-binding region (PBR) for the Class I proteins, and exon 2 alone encodes the PBR for the Class II proteins. For the Class I genes, we also considered exon 4 alone because it is comparable in size to exons 2 and 3 and provides a good contrast to the PBR-encoding exons. Because few intron sequences were available for non-human species, we did not include them in our analyses.
Trans-species polymorphism is widespread
For each gene group and genic region, we used the Bayesian phylogenetics software BEAST2 (Bouckaert et al., 2014; Bouckaert et al., 2019) with package SubstBMA (Wu et al., 2013) to infer phylogenies. One major advantage of BEAST2 over less tunable methods is that it can allow evolutionary rates to vary across sites, which is important for genes such as these which experience rapid evolution in functional regions (Wu et al., 2013). We also considered each exon separately to minimize the impact of recombination as well as to compare and contrast the binding-site-encoding exons with non-binding-site-encoding exons.
We can visualize each set of phylogenies as a single summary tree, which maximizes the product of posterior clade probabilities (BEA, 2024). Three of these summary trees are shown in Figure 2, constructed from the second exons of classical Class I gene MHC-C, classical Class II gene MHC-DQB, and non-classical Class II gene MHC-DOA, respectively (see Figure 2—figure supplements 1–16, Figure 3—figure supplements 1–12, Figure 4—figure supplements 1–10 for the other exons and genes).
<italic>BEAST2 allele summary trees using sequences from exon 2</italic>.
(A) MHC-C, (B) MHC-DQB1, and (C) MHC-DOA. Each tip represents an allele, with color and four-letter abbreviation representing the species (see Figure 1—figure supplement 2 for full species key). The species label is followed by the allele name (see Appendix 1 for more details on nomenclature) or RefSeq accession number. For simplicity, monophyletic groups of similar alleles are collapsed with a triangle and labeled with their one-field allele name. The color/abbreviation key (center) also depicts the species tree (Kuderna et al., 2023). Human alleles (HLA; red) are bolded for emphasis. Dashed outgroup branches are scaled by a factor of 110\begin{document}$ \frac{1}{10}$\end{document} to clarify tree structure within the clade of interest. The smaller inset tree in panel B highlights the relationships between two human allele groups (red) and two OWM allele groups (green). The indicated human and OWM lineages coalesce more recently between groups than within each group. Pri., primate backbone sequences; Mam., mammal outgroup sequences.
MHC-A-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; > 100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-B-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1Appendix 1for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; > 100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-C-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-E-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-F-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-G-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DRA-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DQA-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DPA-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DMA-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DOA-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DRB-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DQB-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DPB-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DMB-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DOB-related group <italic>BEAST2</italic> tree for exon 2 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-C is a classical Class I gene that duplicated from MHC-B in the ancestor of the great apes (Piontkivska, 2003; Fukami-Kobayashi et al., 2005; Abi-Rached et al., 2010; Adams and Parham, 2001; Lugo and Cadavid, 2015). Its protein product participates in classical antigen presentation and also serves as the dominant Class I molecule for interacting with killer cell immunoglobulin-like receptors (KIRs) in innate immunity (Adams and Parham, 2001; Guethlein et al., 2015; Vollmers et al., 2021). MHC-DQB is a classical Class II gene which pairs with MHC-DQA. Apes have two MHC-DQ copies, MHC-DQA1/MHC-DQB1 and MHC-DQA2/MHC-DQB2, while the second copy was deleted in OWM. NWM can have two or three sets of MHC-DQ genes, depending on species, but it has been unclear whether any of them are 1:1 orthologous with the ape or OWM genes. Lastly, MHC-DOA is a non-classical Class II gene whose protein product modulates MHC-DM activity, indirectly affecting Class II peptide presentation (Heijmans et al., 2020; Neefjes et al., 2011). The genes’ differing roles result in different patterns in the phylogenetic trees.
Critically, we observe that, at classical genes MHC-C (Figure 2A) and MHC-DQB (Figure 2B), the alleles fail to cluster together according to species, as indicated by the mixed-color clades throughout the trees. In MHC-C, human (HLA; red rectangles), chimpanzee (Patr; dark pink), bonobo (Papa; light pink), and even gorilla (Gogo; orange) alleles can be found throughout the tree, indicating that variation in this gene is almost as old as the gene itself.
Figure 2B displays the BEAST2 tree consisting of MHC-DQB1, -DQB2, and outgroup -DQB alleles all together. It shows many mixed-color clades throughout, consisting of ape (red/orange/yellow rectangles) and OWM (green) alleles grouping together. Alleles often group by first-field name instead of by species, indicating that some allelic lineages have been maintained since before the split of humans and OWM—at least 31 MY. An example of this is shown in the inset to the left of this tree, ‘Example of Human-OWM TSP’. Here, human alleles coalesce with OWM alleles before they coalesce with each other. Near the bottom of the tree is a clade consisting of ape and NWM MHC-DQB2 sequences, suggesting that they are orthologous. However, NWM species have expanded their MHC-DQ regions, so these genes may not actually be 1:1 orthologous (see our companion paper, Fortier and Pritchard, 2025). Additionally, Strepsirrhini sequences do not group with either the MHC-DQB1 or -DQB2 clade, showing that the duplications of the MHC-DQB genes must have happened in or after the Simiiformes ancestor.
Figure 2C shows the BEAST2 summary tree for non-classical MHC-DOA. In this tree, alleles group exclusively by species (clades are collapsed for clarity) and the branching order of the species deviates only slightly from the species tree. This shows that not all MHC genes are affected by long-term balancing selection, despite the complicated linkage disequilibrium across classical and non-classical genes in the region (Smith et al., 2024). This also suggests that antigen presentation specifically, as opposed to a general role in immune function, is the driving force behind this long-term balancing selection.
While the BEAST2 summary trees in Figure 2 are suggestive of deep TSP, they do not directly quantify the statistical confidence in the TSP model. Moreover, standard approaches to quantifying uncertainty in trees, such as bootstrap support or posterior probabilities for specific clades, do not relate directly to hypothesis testing for TSP. We therefore implemented an alternative approach using BEAST2 output, as follows (see the Materials and methods; Bayes factors for details).
We performed formal model testing for TSP within quartets of alleles, where two alleles are taken from a species (or taxon) A, and two alleles are taken from a different species (or taxon) B. If the alleles from A group together (and the alleles from B group together) in the unrooted tree, this quartet supports monophyly of A (and of B). In a neutral genealogy, monophyly of each species’ sequences is expected. But if alleles from A group more closely with alleles from B in the unrooted tree, then this comparison supports TSP. Since BEAST2 samples from the posterior distribution of trees, we counted the number of trees that support TSP versus the number that support monophyly as an estimate of the posterior support for each model. We then summarized the relative support for each model by converting these to Bayes factors (see Materials and methods; Bayes factors for more detail). The precise interpretation of Bayes factors depends on one’s prior expectation; however, following standard guidelines (Jeffreys, 1998), we suggest that Bayes factors >100 should be considered as strong support in favor of TSP. Bayes factors <1 are evidence against TSP. For each comparison of two taxa, we report the maximum Bayes factor across the possible quartets, as we are interested in whether any quartet shows compelling evidence for TSP.
Gene conversion, the unidirectional transfer of short tracts of DNA from a donor to an acceptor sequence, can affect the inferred trees. In particular, acceptor sequences may group more strongly with donor sequences than with sequences that share DNA by descent. This can make it difficult to distinguish trees influenced by trans-species polymorphism from those influenced by gene conversion. Thus, we inferred gene conversion tracts using GENECONV (Sawyer, 1999) and excluded significant gene-converted acceptor alleles from the Bayes factor calculations. While GENECONV cannot possibly infer all past events, this procedure should ameliorate any biasing effects. Additionally, we consider each exon separately; analyzing short tracts reduces the effect of recombination on the tree (see our companion paper for more specifics; Fortier and Pritchard, 2025). Note that the number of sequences available for comparison also affects the detectability of TSP. For example, if the only sequences available are from the same allelic lineage, they will coalesce more recently in the past than they would with alleles from a different lineage and would not show evidence for TSP. This means our method is well-suited to detect TSP when a diverse set of allele sequences is available, but it is conservative when there are few alleles to test. There were few available alleles for some non-classical genes, such as MHC-F, and some species, such as gibbon. This uneven sampling of taxa means that some TSPs cannot be detected at this time.
Bayes factors are shown in Figures 3 and 4. See Figure 2—figure supplements 1–16, Figure 3—figure supplements 1–12, and Figure 4—figure supplements 1–10 for examples of high-Bayes-factor quartets for each comparison.
Strong support for TSP at Class I genes MHC-B and -C.
Bayes factors computed over the set of BEAST trees indicate deep TSP. Different species comparisons are listed on the y-axis, and different gene regions are listed on the x-axis. Each table entry is colored and labeled with the maximum Bayes factor among all tested quartets of alleles belonging to that category. High Bayes factors (orange) indicate support for TSP among the given species for that gene region, while low Bayes factors (teal) indicate that alleles assort according to the species tree, as expected. Bayes factors above 100 are considered decisive. Tan values show poor support for either hypothesis, while white boxes indicate that there are not enough alleles in that category with which to calculate Bayes factors. MHC-A is not present in the NWMs, and MHC-C was not present before the human-orangutan ancestor, so it is not possible to calculate Bayes factors for these species comparisons.
MHC-A-related group <italic>BEAST2</italic> tree for exon 3 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-A-related group <italic>BEAST2</italic> tree for exon 4 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-B-related group <italic>BEAST2</italic> tree for exon 3 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-B-related group <italic>BEAST2</italic> tree for exon 4 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-C-related group <italic>BEAST2</italic> tree for exon 3 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-C-related group <italic>BEAST2</italic> tree for exon 4 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-E-related group <italic>BEAST2</italic> tree for exon 3 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-E-related group <italic>BEAST2</italic> tree for exon 4 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-F-related group <italic>BEAST2</italic> tree for exon 3 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-F-related group <italic>BEAST2</italic> tree for exon 4 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-G-related group <italic>BEAST2</italic> tree for exon 3 (PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-G-related group <italic>BEAST2</italic> tree for exon 4 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
TSP among old-world monkey groups for the Class I genes.
Different species comparisons are listed on the y-axis, and different gene regions are listed on the x-axis. Each table entry is colored and labeled with the maximum Bayes factor among all tested quartets of alleles belonging to that category. High Bayes factors (orange) indicate support for TSP among the given species for that gene region, while low Bayes factors (teal) indicate that alleles assort according to the species tree, as expected. Bayes factors above 100 are considered decisive. Tan values show poor support for either hypothesis, while white boxes indicate that there are not enough alleles in that category with which to calculate Bayes factors. MHC-C was not present before the human-orangutan ancestor, so it is not possible to calculate Bayes factors for MHC-C for these species comparisons.
TSP among new-world monkey groups for the Class I genes.
Different species comparisons are listed on the y-axis, and different gene regions are listed on the x-axis. Each table entry is colored and labeled with the maximum Bayes factor among all tested quartets of alleles belonging to that category. High Bayes factors (orange) indicate support for TSP among the given species for that gene region, while low Bayes factors (teal) indicate that alleles assort according to the species tree, as expected. Bayes factors above 100 are considered decisive. Tan values show poor support for either hypothesis, while white boxes indicate that there are not enough alleles in that category with which to calculate Bayes factors. MHC-C was not present before the human-orangutan ancestor, and MHC-A is not present in the NWM, so it is not possible to calculate Bayes factors for these genes.
Strong support for TSP at the classical Class II genes.
Bayes factors computed over the set of BEAST trees indicate deep TSP. Different species comparisons are listed on the y-axis, and different gene regions are listed on the x-axis. Each table entry is colored and labeled with the maximum Bayes factor among all tested quartets of alleles belonging to that category. High Bayes factors (orange) indicate support for TSP among the given species for that gene region, while low Bayes factors (teal) indicate that alleles assort according to the species tree, as expected. Bayes factors above 100 are considered decisive. Tan values show poor support for either hypothesis, while white boxes indicate that there are not enough alleles in that category with which to calculate Bayes factors.
MHC-DRA-related group <italic>BEAST2</italic> tree for exon 3 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DQA-related group <italic>BEAST2</italic> tree for exon 3 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DPA-related group <italic>BEAST2</italic> tree for exon 3 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DMA-related group <italic>BEAST2</italic> tree for exon 3 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DOA-related group <italic>BEAST2</italic> tree for exon 3 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DRB-related group <italic>BEAST2</italic> tree for exon 3 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DQB-related group <italic>BEAST2</italic> tree for exon 3 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DPB-related group <italic>BEAST2</italic> tree for exon 3 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DMB-related group <italic>BEAST2</italic> tree for exon 3 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
MHC-DOB-related group <italic>BEAST2</italic> tree for exon 3 (non-PBR-encoding).
In the tree, each tip represents a sequence (see Appendix 1 for more details on nomenclature), with the colored rectangle and four-letter abbreviation indicating the species (see Figure 1—figure supplement 2 for full species key). Following the rectangle, tips are labeled with the sequence name; sequences which have been assigned to loci are colored according to the gene group, while unassigned sequences are written in gray. Dashed branches are shortened to 10% of their length to expand detail in the rest of the tree. The left-hand side shows example Bayes factors we calculated from the set of posterior trees; the tree tips correspond to the rows of the grid. Each panel represents a type of comparison (labeled at top) and each column represents one of the top 5 highest-Bayes-factor comparisons (exact value at bottom; >100 indicates strong evidence for TSP). The four colored blocks in each column include two red blocks (human sequences) that were tested against two sequences from other species (see Materials and methods; Bayes factors). Grayed-out rows correspond to sequences that were not considered for Bayes factors, either because they belong to a non-orthologous gene or backbone sequence (and thus not relevant to compare with the human gene in question), or because they may have been involved in a gene conversion event in this exon (according to GENECONV or from the literature).
TSP among old-world monkey groups for the Class II genes.
Different species comparisons are listed on the y-axis, and different gene regions are listed on the x-axis. Each table entry is colored and labeled with the maximum Bayes factor among all tested quartets of alleles belonging to that category. High Bayes factors (orange) indicate support for TSP among the given species for that gene region, while low Bayes factors (teal) indicate that alleles assort according to the species tree, as expected. Bayes factors above 100 are considered decisive. Tan values show poor support for either hypothesis, while white boxes indicate that there are not enough alleles in that category with which to calculate Bayes factors.
TSP among new-world monkey groups for the Class II genes.
Different species comparisons are listed on the y-axis, and different gene regions are listed on the x-axis. Each table entry is colored and labeled with the maximum Bayes factor among all tested quartets of alleles belonging to that category. High Bayes factors (orange) indicate support for TSP among the given species for that gene region, while low Bayes factors (teal) indicate that alleles assort according to the species tree, as expected. Bayes factors above 100 are considered decisive. Tan values show poor support for either hypothesis, while white boxes indicate that there are not enough alleles in that category with which to calculate Bayes factors. There is not enough data in each category to compute Bayes factors for these groups for MHC-DPA1, -DMA, -DMB, -DOA, and -DOB.
At the Class I genes (Figure 3), MHC-C shows strong support for TSP within the African apes: human, chimpanzee, and gorilla. Having arisen fairly recently in the ancestor of human and orangutan, MHC-C has thus maintained some allelic lineages for most of its history. TSP has not previously been reported for this gene.
For MHC-A, Bayes factors vary considerably depending on exon and species pair. Past work suggests that this gene has had a long history of gene conversion affecting different exons, resulting in different evolutionary histories for different parts of the gene (Hans et al., 2017; Gleimer et al., 2011; Adams and Parham, 2001). Indeed, we excluded many MHC-A sequences from our Bayes factor calculations because they were identified as gene-converted in our GENECONV analysis or were previously suggested to be recombinants. As shown in Figure 3, the lack of concordance in Bayes factors across the different exons for MHC-A is evidence for gene conversion, rather than balancing selection, being the most important factor in this gene’s evolution. In contrast, the other gene groups generally show concordance in Bayes factors across exons. We interpret this as evidence in favor of TSP being the primary driver of the observed deep coalescence structure for MHC-B and -C (rather than recombination or gene conversion).
The non-classical Class I genes MHC-E, -F, and -G (bottom row of Figure 3) are interspersed with the classical Class I genes in the MHC region (see Figure 1), but their products have niche functions in innate immunity. Their indirect involvement in adaptive immunity means they experience different selective pressures. They exhibit lower polymorphism and dN/dS<1\begin{document}$dN/dS \lt 1$\end{document}, reflecting the fact that they have not been subject to the same pathogen-mediated balancing selection. The Bayes factors for all three of these genes show strong evidence against TSP, as expected. However, since there are fewer alleles available for the non-classical genes, we note that our method may be conservative here. Interestingly, despite its non-classical role, MHC-E has a known balanced polymorphism in humans; the two main alleles are at similar frequencies worldwide but may have different expression levels and peptide preferences (Paganini et al., 2019; Grant et al., 2020). Our approach—meant to detect ancient TSP—does not reveal balancing selection in MHC-E, showing that this balanced polymorphism is young. For MHC-G, there were not enough sequences available to perform many of the tests (at least two from each species group are required). While we do not expect to see evidence of TSP in this gene, sequencing more alleles is necessary to address this.
Each Class II MHC molecule has an α and β component which are encoded by an A and B gene, respectively. Bayes factors for the Class IIA genes are shown in the top row of Figure 4, while those for their Class IIB partners are shown in the bottom row. The non-classical MHC-DM and -DO molecules assist the classical Class II genes with peptide loading and are not believed to be shaped by balancing selection (Heijmans et al., 2020; Neefjes et al., 2011). As expected, we see strong evidence against TSP between humans and all other primate species for these genes (first two columns of Figure 4).
In contrast, we find evidence for deep TSP within the classical Class II genes. MHC-DPA1 shows TSP between human, chimpanzee, and gorilla in exon 2, but not in exon 3. We find that this TSP in MHC-DPA1 is less deep than has been previously suggested with non-Bayesian methods (Otting and Bontrop, 1995), underscoring the importance of this methodology for handling the MHC. Meanwhile, its partner MHC-DPB1 shows strong evidence for TSP between human and orangutan in exon 3 and suggestive evidence in exon 2; our work provides the first evidence of TSP between humans and other apes for this gene (Slierendregt et al., 1995).
The MHC-DR genes behave somewhat differently than the other classical Class II molecules. While the α and β components of all the other molecules engage in exclusive binding, there are many different MHC-DRβ molecules which all bind to the same MHC-DRα. The MHC-DRA gene is conserved across species with little polymorphism, while the MHC-DRB region is highly variable both in gene content and allelic diversity. Consistent with this, Bayes factors for the MHC-DRA gene reveal strong evidence against TSP for all species pairs, while MHC-DRB1 shows strong evidence in favor of TSP between human, chimpanzee, gorilla, OWM, and even NWM in exon 2. However, because the Bayes factors only support TSP between humans and OWM/NWM in MHC-DRB1 in exon 2, but not in exon 3, this could mean alleles are not actually that ancient. We show in our companion paper that individual MHC-DRB genes are short-lived, and only three are truly orthologous between apes and OWM (Fortier and Pritchard, 2025). These pieces of evidence suggest previous work may have overstated the extent of TSP at this locus.
Remarkably, the MHC-DQB1 gene shows definitive evidence for TSP back to at least the ancestor of humans and OWM. While this result has been presented previously, we confirm it with decisive evidence (Bayes factor >100) for allelic lineages being maintained for over 31 MY (Otting et al., 1992; Otting et al., 2002; Loisel et al., 2006; Simons et al., 2017).
Many of our BEAST trees also showed intermingling of sequences from different species within the OWM or NWM (e.g. Figure 2—figure supplement 4), even if they did not form trans-species clades with ape sequences. This could indicate trans-species polymorphism within the OWM or NWM that is still ancient, but not old enough to be shared with the apes. Therefore, we also calculated Bayes factors across different clades of OWM and NWM (Figure 3—figure supplement 13, Figure 3—figure supplement 14, Figure 4—figure supplement 11, and Figure 4—figure supplement 12). We see very strong evidence for TSP between all groups of OWM for MHC-DPA1, -DPB1, -DQA1, and -DQB1, indicating that allelic lineages at these genes have been maintained within the OWM for at least 19 MY. Unexpectedly, we also see evidence for TSP in some non-classical genes. MHC-E, a gene that is non-classical in humans and is presumed non-classical in OWM (yet is duplicated in some species), shows evidence for 15-MY-old TSP within the OWM. In non-classical MHC-DMB, we also observe TSP within the OWM as old as 11 MY. This could indicate differing roles for these genes in the OWM lineage, and functional experiments are needed to explore this. Due to the uncertainty of locus assignments for alleles of the OWM MHC-A, -B, and -DRB genes and of the NWM genes, we cannot make definitive conclusions about TSP within these clades for these other genes.
In summary, the phylogenetic analyses point to ancient TSP in classical genes MHC-B, -C, -DPA1, -DPB1, -DQB1, -DRB1, and -DRB3. Bayes factors for the non-classical genes MHC-E, -F, -G, -DMA, -DMB, -DOA, and -DOB do not indicate TSP involving apes at these loci—as expected for non-classical genes. However, we detected possible TSP at some of these genes within other clades, such as the OWM, hinting at possible functional differences. Overall, TSP is more ancient among the Class II genes than the Class I genes, consistent with the genes’ older age.
From evolution to function
Alongside the evidence for ancient TSP, the MHC region is also notable for its high rate of missense substitutions (dN/dS>1\begin{document}$ dN/dS \gt 1$\end{document}) (Hughes and Nei, 1988; Hughes and Nei, 1989) and its large number of GWAS hits for autoimmune and infectious diseases (Buniello et al., 2019; Kennedy et al., 2017). We next aimed to understand how these observations relate to signals of TSP and known features of the MHC proteins.
To explore these questions, we first estimated the per-site evolutionary rates within each gene. As in our TSP analysis, we used the BEAST2 package SubstBMA, which estimates evolutionary rates at every site concurrently with a tree. We averaged these rates over all states in the chain to get per-site evolutionary rates, then calculated their fold change relative to the average rate among mostly-gap sites in the alignment (‘baseline’; see Materials and methods; Rapidly-evolving sites).
Figure 5A shows the substitution rate fold change for each nucleotide along the concatenated coding sequence of Class I genes MHC-B, -C, and -E (see Figure 5—figure supplement 1 for the other genes). In classical genes MHC-B and -C, nearly all the rapidly-evolving sites lie within exons 2 and 3, which encode the protein’s peptide-binding domain. While exons 2 and 3 make up only ∼50% of the gene’s length, they contain 94% and 90% of the sites evolving at more than four times the baseline evolutionary rate for the classical genes MHC-B and MHC-C, respectively. In MHC-B, exons 2 and 3 each show significantly higher proportions of rapidly-evolving sites compared to the ‘other’ exons (exons in the gene excluding 2, 3, or 4), while the difference is not significant for MHC-C (Figure 5—figure supplement 2). This result could reflect the relatively young age of MHC-C or its additional role as the dominant Class I molecule for interacting with KIRs (Adams and Parham, 2001; Guethlein et al., 2015; Vollmers et al., 2021; Piontkivska, 2003).
Rapidly-evolving sites in the Class I genes.
(A) Rapidly-evolving sites are primarily located in exons 2 and 3. Here, the exons are concatenated such that the cumulative position along the coding region is on the x-axis. The dashed orange lines denote exon boundaries. The three genes are aligned such that the same vertical position indicates an evolutionarily equivalent site. The y-axis shows the substitution rate at each site, expressed as a fold change (the base-2 logarithm of each site’s evolutionary rate divided by the mean rate among mostly-gap sites in each alignment; see Materials and methods). (B) Rapidly-evolving sites are located in each protein’s peptide-binding pocket. Structures are Protein Data Bank (Berman et al., 2000) 4BCE (Teze et al., 2014) for HLA-B, 4NT6 (Choo et al., 2014) for HLA-C, and 7P4B (Walters et al., 2022) for HLA-E, with images created in PyMOL (Schrödinger, LLC, 2021). Substitution rates for each amino acid are computed as the mean substitution rate of the three sites composing the codon. Orange indicates rapidly-evolving amino acids, while teal indicates conserved amino acids. (C) Rapidly-evolving amino acids are significantly closer to the peptide than conserved amino acids. The y-axis shows the BEAST2 substitution rate and the x axis shows the minimum distance to the bound peptide, measured in PyMOL (Schrödinger, LLC, 2021). Each point is an amino acid, and distances are averaged over several structures (see Table 5). The orange line is a linear regression of substitution rate on minimum distance, with slope and p-value annotated on each panel.
Rapidly-evolving sites in the Class I genes.
Rapidly-evolving sites are primarily located in exons 2 and 3. Here, the exons are concatenated such that the cumulative position along the coding region is on the x-axis. The dashed orange lines denote exon boundaries and the exon numbers are labeled in the top panel. The genes are aligned such that the same vertical position indicates an evolutionarily equivalent site. The y-axis shows the substitution rate at each site, expressed as a fold change (the base-2 logarithm of each site’s evolutionary rate divided by the mean rate among mostly-gap sites in each alignment; see Materials and methods).
Proportions of rapidly-evolving sites for Class I.
Sites were binned into slowly-evolving (≤-1), rapidly-evolving (>1), or baseline (>-1 but ≤1) categories. We then calculated proportions of these categories for exon 2, exon 3, exon 4, and the ’other’ exons (not exons 2, 3, or 4). These bins are all approximately the same size, ∼270 bp. For each gene and exon, we tested the difference in the proportion of rapidly-evolving sites between that exon and the ‘other’ exons group (two-sample z-test for equality of proportions with continuity correction). Significant tests (Bonferroni corrected; p<0.05/3) are marked with an asterisk.
Rapidly-evolving sites on Class I protein structures.
Structures are Protein Data Bank (Berman et al., 2000) 6 J1W (Zhu et al., 2019) for HLA-A, 3BVN (Kumar et al., 2009) for HLA-B, 4NT6 (Choo et al., 2014) for HLA-C, 7P4B (Walters et al., 2022) for HLA-E, 5IUE (Dulberger et al., 2017) for HLA-F, and 3KYN (Walpole et al., 2010) for HLA-G, with images created in PyMOL (Schrödinger, LLC, 2021). Substitution rates for each amino acid are computed as the mean substitution rate of the three sites composing the codon. Orange indicates rapidly-evolving amino acids, while teal indicates conserved amino acids.
Evolutionary rate is related to the distance to peptide.
The y-axis shows the BEAST2 substitution rate, expressed as a fold change (the base-2 logarithm of each site’s evolutionary rate divided by the mean rate among mostly-gap sites in each alignment; see Materials and methods). The x-axis shows the minimum distance to the bound peptide, measured in PyMOL (Schrödinger, LLC, 2021). Each point is an amino acid, and distances are averaged over several structures (see Table 5). The orange line is a linear regression of substitution rate on minimum distance, with slope and p-value annotated on each panel. Amino acids with a fold change greater than 1.5 are labeled.
Class I rapidly-evolving sites by binned distance to peptide.
Nucleotide sites were divided into those whose corresponding amino acids contact the peptide (<4Å) versus do not contact the peptide (≥4Å), shown on the x-axis. This distance has been used previously to define plausible peptide-contacting residues (Nielsen et al., 2007). The y-axis shows the substitution rate at each site, expressed as a fold change (the base-2 logarithm of each site’s evolutionary rate divided by the mean rate among mostly-gap sites in each alignment; see Materials and methods). For each gene, the groups are compared using a Wilcoxon test, with p-value displayed at the top of each panel.
In contrast to these classical genes, non-classical MHC-E primarily presents self-peptides for recognition by NK cell receptors, and its peptide-binding groove is tailored to accommodate a very specific set of self-peptides—leader peptides cleaved from other Class I MHC proteins during processing (Miller et al., 2003). As shown in Figure 5A, this gene has fewer rapidly-evolving sites than the classical genes. These sites are also relatively evenly distributed across the gene, with exons 2 and 3 (which cover ∼50% of the gene’s length) containing 45% of the sites evolving at over four times the baseline evolutionary rate. Interestingly, exon 4—a non-peptide-binding exon of equal size—displays a significantly lower proportion of rapidly-evolving sites compared with the ‘other’ exons (Figure 5—figure supplement 2). These results support that MHC-E has been remarkably conserved across the primates and that its evolution may not be driven by differential peptide binding (Heijmans et al., 2020).
We then examined where the rapidly-evolving sites lie within the physical protein structures. To do this, we averaged the per-site rates within each codon to get per-amino-acid rates, then mapped these onto the known human protein structures. Unfortunately, there are few non-human primate protein structures in the Protein Data Bank (Berman et al., 2000), but the macaque structures we found were nearly identical to those of human. Figure 5B shows structures for human HLA-B, -C, and -E; this view features the peptide (black) sitting in the peptide-binding groove (flanked on top and bottom by helices) (see Figure 5—figure supplement 3 for the rest of the Class I proteins). In MHC-B and -C, rapidly-evolving amino acids (orange) tend to be located within the peptide-binding groove. To quantify this, we measured the minimum distance between each amino acid and the bound peptide. We averaged these distances over several structures, which are listed in the Materials and methods (Table 5). For all three proteins, amino acids closer to the peptide have significantly higher evolutionary rates than amino acids further from the peptide, as shown in Figure 5C (see also Figure 5—figure supplements 4 and 5). The effect is much less pronounced in non-classical MHC-E, where even the amino acids closest to the peptide do not exhibit high evolutionary rates. These results are consistent with the expectation that rapid evolution and diversity at the classical MHC genes would be mediated by selective pressures for changes in peptide binding.
The rapidly-evolving sites for the Class II genes are shown in Figure 6. Panel A shows the substitution rate fold change for each nucleotide along the concatenated coding sequence of the Class II MHC-DRA, -DQA, -DRB, and -DQB gene groups (see Figure 6—figure supplements 1 and 3 for the other genes). Both MHC-DQA and -DQB are extraordinarily polymorphic, but MHC-DRA is conserved compared to its multiple, highly-variable MHC-DRB partners. Indeed, Figure 6A shows that rapidly-evolving sites are concentrated in binding-site-encoding exon 2 for MHC-DRB, -DQB, and to a lesser extent MHC-DQA. Exon 2, which makes up ∼30% of the coding region, contains 32% of sites evolving at more than twice the baseline rate in MHC-DRA, but 57% of such sites in MHC-DQA, 61% in MHC-DQB, and 73% in MHC-DRB. Comparing across exons, exon 2 contains a significantly higher proportion of rapidly-evolving sites compared to the "other" exons in classical MHC-DQA, -DQB, -DRB, and -DPB, but also—curiously—in non-classical MHC-DMA and -DMB (Figure 6—figure supplements 2 and 4). It is interesting that MHC-DM appears to be evolving rapidly in its binding-site-encoding exons, despite the fact that it is not thought to bind peptides. Instead, it is responsible for assisting with peptide loading onto the classical genes. Co-evolution with MHC-DR in particular seems possible; the interaction between the MHC-DM and -DR molecules depends on the affinity of the peptide trying to bind with MHC-DR. MHC-DM thus shapes the repertoire of peptides presented by MHC-DR, favoring high-affinity peptides (Dijkstra and Yamaguchi, 2019; Schulze and Wucherpfennig, 2012). It is plausible that the host-pathogen evolution shaping the MHC-DRB genes has resulted in co-evolution of MHC-DMA and -DMB to maintain this regulatory interaction.
Rapidly-evolving sites in the Class II genes.
(A) Rapidly-evolving sites are primarily located in exon 2. Here, the exons are concatenated such that the cumulative position along the coding region is on the x-axis. The dashed orange lines denote exon boundaries. The α genes (top two plots) are aligned such that the same vertical position indicates an evolutionarily equivalent site; the same is true for the β genes (bottom two plots). The y-axis shows the substitution rate at each site, expressed as a fold change (the base-2 logarithm of each site’s evolutionary rate divided by the mean rate among mostly-gap sites in each alignment; see Materials and methods). (B) Rapidly-evolving sites are located in each protein’s peptide-binding pocket. Structures are Protein Data Bank (Berman et al., 2000) 5JLZ (Gerstner et al., 2016) for HLA-DR and 2NNA (Henderson et al., 2007) for HLA-DQ, with images created in PyMOL (Schrödinger, LLC, 2021). Substitution rates for each amino acid are computed as the mean substitution rate of the three sites composing the codon. Orange indicates rapidly-evolving amino acids, while teal indicates conserved amino acids. (C) Rapidly-evolving amino acids are significantly closer to the peptide than conserved amino acids. The y-axis shows the BEAST2 substitution rate and the x axis shows the minimum distance to the bound peptide, measured in PyMOL (Schrödinger, LLC, 2021). Each point is an amino acid, and distances are averaged over several structures (see Table 5). The orange line is a linear regression of substitution rate on minimum distance, with slope and p-value annotated on each panel.
Rapidly-evolving sites in the Class IIA genes.
Rapidly-evolving sites are primarily located in exon 2. Here, the exons are concatenated such that the cumulative position along the coding region is on the x-axis. The dashed orange lines denote exon boundaries and the exon numbers are labeled in the top panel. The genes are aligned such that the same vertical position indicates an evolutionarily equivalent site. The y-axis shows the substitution rate at each site, expressed as a fold change (the base-2 logarithm of each site’s evolutionary rate divided by the mean rate among mostly-gap sites in each alignment; see Materials and methods).
Proportions of rapidly-evolving sites for Class IIA.
Sites were binned into slowly-evolving (≤-1), rapidly-evolving (>1), or baseline (>-1 but ≤1) categories. We then calculated proportions of these categories for exon 2, exon 3, and the ‘other’ exons (not exons 2 or 3). These bins are all approximately the same size, ∼270bp. For each gene and exon, we tested the difference in the proportion of rapidly-evolving sites between that exon and the ‘other’ exons group (two-sample z-test for equality of proportions with continuity correction). Significant tests (Bonferroni corrected; p<0.05/2) are marked with an asterisk.
Rapidly-evolving sites in the Class IIB genes.
Rapidly-evolving sites are primarily located in exon 2. Here, the exons are concatenated such that the cumulative position along the coding region is on the x-axis. The dashed orange lines denote exon boundaries and the exon numbers are labeled in the top panel. The genes are aligned such that the same vertical position indicates an evolutionarily equivalent site. The y-axis shows the substitution rate at each site, expressed as a fold change (the base-2 logarithm of each site’s evolutionary rate divided by the mean rate among mostly-gap sites in each alignment; see Materials and methods).
Proportions of rapidly-evolving sites for Class IIB.
Sites were binned into slowly-evolving (≤-1), rapidly-evolving (>1), or baseline (>-1 but ≤1) categories. We then calculated proportions of these categories for exon 2, exon 3, and the ‘other’ exons (not exons 2 or 3). These bins are all approximately the same size, ∼270 bp. For each gene and exon, we tested the difference in the proportion of rapidly-evolving sites between that exon and the ‘other’ exons group (two-sample z-test for equality of proportions with continuity correction). Significant tests (Bonferroni corrected; p<0.05/2) are marked with an asterisk.
Rapidly-evolving sites on Class II protein structures.
Structures are Protein Data Bank (Berman et al., 2000) 5JLZ (Gerstner et al., 2016) for HLA-DR, 2NNA (Henderson et al., 2007) for HLA-DQ, 7 T2A (Ciacchi et al., 2023) for HLA-DP, 2BC4 (Nicholson et al., 2006) for HLA-DM, and 4I0P (Guce et al., 2013) for HLA-DO, with images created in PyMOL (Schrödinger, LLC, 2021). Substitution rates for each amino acid are computed as the mean substitution rate of the three sites composing the codon. Orange indicates rapidly-evolving amino acids, while teal indicates conserved amino acids.
Class II rapidly-evolving sites by binned distance to peptide.
Nucleotide sites were divided into those whose corresponding amino acids contact the peptide (<4Å) versus do not contact the peptide (≥4Å), shown on the x-axis. This distance has been used previously to define plausible peptide-contacting residues (Nielsen et al., 2007). The y-axis shows the substitution rate at each site, expressed as a fold change (the base-2 logarithm of each site’s evolutionary rate divided by the mean rate among mostly-gap sites in each alignment; see Materials and methods). For each gene, the groups are compared using a Wilcoxon test, with p-value displayed at the top of each panel. MHC-DM and -DO are not shown because they do not bind peptides.
Number of associations per amino acid as a function of evolutionary rate.
The x-axis shows the substitution rate at each site, expressed as a fold change (the base-2 logarithm of each site’s evolutionary rate divided by the mean rate among mostly-gap sites in each alignment; see Materials and methods). The y-axis shows the number of unique associations for each amino acid, including diseases, TCR phenotypes, and protein expression levels. Only genes with associations are shown; at the time of publishing, there were no amino acid associations meeting our criteria for MHC-E, -F, -G, -DRA, -DMA, -DMB, -DOA, or -DOB. For each gene, a regression line is shown in orange, with slope and p-value displayed at the top of each panel. The amino acids with the greatest number of associations within each gene are also labeled.
We again mapped the evolutionary rates onto human protein structures, shown in Figure 6B. In each molecule, the α chain is positioned at the top and encompasses the upper helix forming the binding site, while the β chain is oriented toward the bottom and encompasses the lower helix. The peptide is shown in black. Rapidly-evolving sites are concentrated in each protein’s binding site, although in MHC-DR this is more prominent in the bottom helix (MHC-DRB) (see Figure 6—figure supplement 5 for the other proteins).
We then measured the distance between each amino acid and the bound peptide, shown in Figure 6C. MHC-DRA did not show a significant relationship between evolutionary rate and distance, as expected by its relatively uniform distribution of evolutionary rates across the sequence (Figure 6A). For the other three proteins, amino acids closer to the peptide had significantly higher evolutionary rates. This held true for the classical MHC-DPA and -DPB genes as well (Figure 5—figure supplement 4 and Figure 6—figure supplement 6). Again, this is consistent with differential peptide binding and TCR responsiveness driving the diversity and long-term balancing selection at the classical genes. We could not measure peptide distances for non-classical MHC-DOA, -DOB, -DMA, and -DMB genes because they do not engage in peptide presentation.
Lastly, since the rapidly-evolving sites are likely involved in peptide binding, they also influence the response to pathogens and self-antigens, presumably affecting risk for infectious and autoimmune diseases. To bridge the gap between evolution and complex traits, we collected HLA fine-mapping studies for infectious, autoimmune, and other diseases, as well as for biomarkers and TCR phenotypes. These studies report associations between a disease or trait and classical HLA alleles, SNPs, and amino acid variants, often with multiple independent hits per gene. They demonstrate that HLA variation affects disease in complex ways—sometimes, a single variant is strongly associated with a condition, while other times, a combination of amino acids or even an entire allele (haplotype) is the strongest indicator of disease susceptibility or protection.
We were interested in whether our rapidly-evolving amino acids from the BEAST2 analysis corresponded with disease-associated amino acids from the literature. Table 1 lists disease, trait, and TCR-phenotype associations for the most rapidly-evolving amino acids (fold change ≥ 1) of the MHC-B group (see Table 1—source data 1 for the other genes). The majority of rapidly-evolving positions in MHC-B have at least one association. Furthermore, all three classical Class I genes (MHC-A, -B, and -C) show a significant positive relationship between per-amino-acid evolutionary rate and the number of amino acid associations (Figure 6—figure supplement 7). Interestingly, this relationship is not significant for the Class II genes, possibly because they evolve more slowly overall.
Rapidly-evolving amino acids in MHC-B and their trait and disease associations.
Shown here are all amino acid positions in the MHC-B group evolving at more than twice the baseline rate (fold change ≥ 1). Many corresponding positions in human HLA-B have associations with autoimmune or infectious diseases, biomarkers, or TCR phenotypes. Disease associations were collected from a literature search of HLA fine-mapping studies with over 1000 cases (see Materials and methods).
Rapidly-evolving amino acids and their trait and disease associations for all studied genes.
For each gene group, the rapidly-evolving amino acid positions (substitution rate 𝑙𝑜𝑔2 fold change >1) are listed alongside their disease and trait associations. group: gene group; position: amino acid position; evol_rate: substitution rate 𝑙𝑜𝑔2 fold change; min_dist: minimum distance to peptide (Å); disease_assoc: list of disease, trait, and other associations, with citations.
Amino Acid Position
Evol. Rate Fold Change
Distance to Peptide (Å)
Associations
156
3.42
3.55
Chronic Hepatitis C (Hirata et al., 2019), HIV Set Point Viral Load (Luo et al., 2021), Asthma (Sakaue et al., 2021), Eosinophil Count (Sakaue et al., 2021), Hypothyroidism (Sakaue et al., 2021), Pediatric Asthma (Sakaue et al., 2021), Systolic Blood Pressure (Sakaue et al., 2021), Total Protein (Sakaue et al., 2021), TCR β Interaction Probability >50% (Sharon et al., 2016), Plasma Protein Levels of ADAM8, AGER, ASPSCR1, B2M, CCL16, CCL28, CCL4, CD200R1, CD5L, CDSN, CX3CL1, FCRL5, IGF2R, IL12A, IL12B, IL5RA, MICB, NUCB2, PDCD1, RARRES2, SFTPD, SIGLEC6, SNX2, TIMD4, TNFRSF4, TNR, TYRP1 (Krishna et al., 2024)
95
3.10
3.82
KLRF1 Plasma Protein Level (Krishna et al., 2024)
114
3.08
4.80
Rheumatoid Arthritis (Sakaue et al., 2021), Plasma Protein Levels of AIF1, CD1C, DDR1, IL15, LILRB2, MICB (Krishna et al., 2024)
116
2.84
3.44
Eosinophil Count (Hirata et al., 2019), HIV Control (McLaren et al., 2012), Angina (Sakaue et al., 2021), Allergic Rhinitis (Waage et al., 2018), Psoriasis (Zhou et al., 2016), Plasma Protein Levels of ADAM15, APOM, BTN2A1, CD1C, CFB, CXCL11, CXCL9, FLT4, GNLY, KLRF1, LILRB1, MICB, PLXDC2, TNF, TNFRSF13C, TNXB (Krishna et al., 2024)
70
2.64
3.71
Platelet (Sakaue et al., 2021), Plasma Protein Levels of CD8A, GZMA, MICB, NRP2 (Krishna et al., 2024)
97
2.37
4.29
Ankylosing Spondylitis (Butler-Laporte et al., 2023), HIV Set Point Viral Load (Luo et al., 2021), HIV Control (McLaren et al., 2012), Adult Height (Sakaue et al., 2021), Alkaline Phosphatase (Sakaue et al., 2021), Body Weight (Sakaue et al., 2021), C-reactive Protein (Sakaue et al., 2021), IgA Nephritis (Sakaue et al., 2021), Mean Arterial Pressure (Sakaue et al., 2021), Mean Corpuscular Hemoglobin (Sakaue et al., 2021), Pneumonia (Tian et al., 2017), Tonsillectomy (Tian et al., 2017), Plasma Protein Levels of ADGRE2, BTN3A2, CCL21, CCL3, CD1C, CD8A, CDSN, CPVL, DXO, EBI3, EFCAB14, HBEGF, HCG22, IL12A, IL12B, LILRB2, LRP1, LRPAP1, LTB, LY75, MANF, MANSC1, MICB, OSCAR, PLA2G10, PRTN3, SIGLEC10, STAB2, TEK, TNFSF13, ZNRD2 (Krishna et al., 2024)
24
2.35
4.68
Hepatic Cancer (Sakaue et al., 2021), Plasma Protein Levels of ADAM15, CXCL10, FCRL6, GZMB, MICB, TNFSF8 (Krishna et al., 2024)
163
2.23
4.13
Lung Cancer (Squamous Cell Carcinoma) (Ferreiro-Iglesias et al., 2018), Alanine Aminotransferase (Sakaue et al., 2021), TCR Expression (TRAV38-1) (Sharon et al., 2016), TCR α\begin{document}$ \alpha$\end{document} Interaction Probability >50% (Sharon et al., 2016), Plasma Protein Levels of DDR1, GZMA, LILRB1, MICB, NPTX1, SEPTIN3, TEK, WFDC2 (Krishna et al., 2024)
67
1.96
3.72
Graves’ Disease (Hirata et al., 2019), HIV Set Point Viral Load (Luo et al., 2021), Asthma (Sakaue et al., 2021), Psoriasis (Stuart et al., 2022), Plasma Protein Levels of AMBP, C2, CD160, CD28, CD48, CFB, FCRL1, FCRL6, FRZB, GP1BB, LILRB1, LTA, LY96, NID1, SIGLEC9, SORT1, THBD, TNFRSF4, TNFSF13B, TNXB, TP53BP1, VCAM1 (Krishna et al., 2024)
152
1.96
3.53
JIA (Oligoarthritis/RF-negative Polyarthritis) (Hinks et al., 2017), Plasma Protein Levels of LTBR, MICB, PLXNA4, RARRES2 (Krishna et al., 2024)
63
1.71
2.94
HIV Control (McLaren et al., 2012), Skin Cancer (Sakaue et al., 2021), B2M Plasma Protein Level (Krishna et al., 2024)
99
1.70
3.00
Plasma Protein Levels of APOM, CRTAM, DXO, IL15, MICB, OSCAR (Krishna et al., 2024)
66
1.58
3.41
TNFSF11 Plasma Protein Level (Krishna et al., 2024)
69
1.56
4.69
Parkinson’s Disease (Naito et al., 2021)
74
1.31
4.30
Chronic Sinusitis (Sakaue et al., 2021)
62
1.29
3.89
PGLYRP1 Plasma Protein Level (Krishna et al., 2024)
138
1.26
9.18
9
1.20
3.36
Primary Biliary Cholangitis (Darlay et al., 2018), Systemic Lupus Erythematosus (Molineros et al., 2019), Rheumatoid Arthritis (Raychaudhuri et al., 2012), Hyperthyroidism (Sakaue et al., 2021), Monocyte Count (Sakaue et al., 2021), Serum Creatinine (Sakaue et al., 2021), Psoriasis (Zhou et al., 2016), Plasma Protein Levels of CD1C, CX3CL1, IL12B, LIPF, LTA, MICB, PDCD1, RGMA, SGSH, SLAMF7, TNFRSF8 (Krishna et al., 2024)
81
1.03
4.04
Thus, in summary, we find that rapid evolution has primarily targeted amino acids within the peptide-binding region of each gene, and that these specific positions are likely the primary drivers of phenotypic associations at the MHC locus.
Discussion
The MHC region contains the clearest signals of balancing and directional selection in mammalian genomes, including extreme diversity, ancient trans-species polymorphism, and high rates of nonsynonymous evolution between allelic lineages. In humans, MHC/HLA variation is associated with risk for infectious and autoimmune diseases and many other traits, and HLA matching is critical for successful tissue transplantation (Kennedy et al., 2017; Smith et al., 2024; Lee et al., 2007).
Despite its evolutionary and clinical importance, the extreme diversity of the MHC makes it challenging to study, and basic questions about its evolutionary history remain unresolved. While past work has suggested ultra-deep TSP at this locus, in this study, we re-examined the region with modern, comprehensive data and a unified analysis framework. Using Bayesian evolutionary analysis, we report conclusive evidence for long-term TSP in seven classical genes, including between humans and OWM at MHC-DQB1. Thus, remarkably, allelic lineages at this gene have been maintained for at least 31 MY.
Our evidence for TSP at MHC-DQB1 spanning at least 31 MY places it among the most ancient examples of balancing selection known in any species, and almost certainly the oldest in primates. Aside from MHC, the deepest example within primates is at the ABO locus controlling blood type; it exhibits trans-species polymorphism between humans and gibbons, an age of 23 MY (Ségurel et al., 2012). In various chimpanzee species, OAS1, which helps inhibit viral replication, contains alleles up to 13 MY old (Ferguson et al., 2012). TSP between chimpanzee and human includes LAD1, a protein that maintains cell cohesion (6 MY; Teixeira et al., 2015), retroviral transcription factor TRIM5α (4-7 MY in apes and >8 MY in OWM; Cagliani et al., 2010; Newman et al., 2006), and ZC3HAV1, an antiviral protein leading to viral RNA degradation (6 MY; Cagliani et al., 2012), among others (Leffler et al., 2013).
Looking more broadly across the tree of life, ancient trans-species polymorphism occurs widely, albeit rarely. Several of the oldest examples are found in the MHC locus: MHC polymorphisms have been maintained for 35 MY in cetaceans (Xu et al., 2009), 40 MY in herons (Li et al., 2011), 48 MY in mole rats (Kundu and Faulkes, 2007), 70 MY in tree frogs (Zhao et al., 2013), and over 105 MY in salmonid fishes (Kiryu et al., 2005; Grimholt et al., 2015). There are examples in non-MHC loci as well; in cyanobacteria, polymorphism at the HEP island controlling heterocyst function has been maintained for 74 MY (Sano et al., 2018), in plants, S-genes determining self-incompatibility exhibit TSP spanning 36 MY (Ioerger et al., 1990; Igic et al., 2006; Fujii et al., 2016), and in Formica ants, alleles at a supergene underlying colony queen number have been maintained for over 30 MY (Purcell et al., 2021).
Paradoxically, given the extremely long-lived balancing selection acting in these lineages, many authors have also reported strong directional selection at the MHC (Brandt et al., 2018; Nunes et al., 2021; Bhatia et al., 2011). Indeed, within the phylogeny, we find that the most rapidly-evolving codons are substituted at around two- to fourfold the baseline rate, generating ample mutations upon which selection may act. For the classical genes (except MHC-DRA), these rapidly-evolving sites lie within the peptide binding regions of the corresponding proteins, usually very close to the peptide-contact surfaces. This does not hold true for the non-classical genes, supporting the fact that selection at the MHC is mainly driven by the peptide presentation pathway. Non-classical MHC-DMA and -DMB are a surprising exception, showing significantly elevated proportions of rapidly-evolving sites in exon 2 similarly to the classical genes (even though the MHC-DM molecule does not bind peptides). This pattern may be caused by its co-evolution with the classical Class II genes, and more research is needed to address this.
The primary role of classical MHC proteins is to present peptides for T cell recognition; we found that the same rapidly-evolving amino acids are associated with shaping T cell receptor (TCR) repertoires. Moreover, these amino acids are frequently associated with autoimmune and infectious diseases in HLA fine-mapping studies, particularly for Class I.
Taken together, we begin to see a comprehensive picture of the nature of primate MHC evolution. In response to rapidly-changing pathogen pressures, the PBRs of classical MHC proteins evolve to bind changing pathogen antigens and present them to TCRs. Broad lineages of MHC alleles are maintained over tens of millions of years by strong balancing selection, providing defense against a wide variety of different pathogens. Yet within these lineages, alleles turn over quickly in response to new specific threats. This reconciles evidence for TSP, the presence of thousands of alleles, and the existence of rapidly-evolving sites.
MHC molecules must evolve to detect pathogens with both specificity and sensitivity, and distinguishing self from non-self peptides is a challenging task. As MHC proteins evolve, there is an unavoidable flux between infection defense and autoimmune susceptibility. Additionally, many MHC proteins have roles in both innate and adaptive immunity. As a result, rapidly-evolving amino acids are associated with both infections and autoimmune conditions. In the future, disease studies within other primate species could provide insight into the trajectory of MHC evolution and might reveal evolutionary trade-offs. Perhaps balancing selection has kept the same amino acids disease-relevant across the entire primate evolutionary tree, or maybe the rapid turnover of MHC variation means different primate clades will have different disease associations.
One limitation of the data we used is that the vast majority of nonhuman MHC sequences were obtained via Sanger sequencing or next-generation sequencing methods. While highly accurate sequence-wise, these methods are limited by PCR-related technical artifacts such as heteroduplexes and chimeras. Allelic dropout is also a problem, because similar genes may not be amplified uniformly, resulting in alleles or entire genes being missed when the entire region is amplified simultaneously (Cheng et al., 2022). Additionally, MHC allele assignments require sequences from multiple exons, so the phasing of distant variants can make it difficult to assign alleles confidently (Liu, 2021). Many nonhuman MHC regions are also more complex than their human counterparts, containing unknown numbers of recently duplicated paralogs, copy number variants, and structural variants (Cheng et al., 2022; Heijmans et al., 2020). This makes it difficult to assemble MHC sequences spanning the entire region and assign sequences to genes, which can make the inference of TSP challenging. For example, if sequences from two entirely different genes are thought to be from the same gene, one may falsely conclude that they are highly-diverged alleles and suspect TSP. We relied mostly on common alleles from the IPD-MHC/HLA database, which have been confirmed in multiple individuals often from different research groups. This helps reduce the issue of chimeric or heteroduplex alleles being wrongly considered to be highly-diverged alleles. We also did not assess TSP when orthology was too ambiguous, instead only calculating Bayes factors when our trees and other work strongly supported sequences coming from the same gene. Nevertheless, convergent evolution and shared history can still result in ambiguous gene assignments (Dilthey, 2021). Luckily, long-read sequencing of the MHC region has the potential to solve many of these issues. Oxford Nanopore and PacBio HiFi sequencing have already been used to obtain high-quality MHC sequences in humans (Wenger et al., 2019; Jain et al., 2018; Liu, 2021; Bruijnesteijn, 2023), and researchers are beginning to explore their potential in non-model organisms (Cheng et al., 2022). These methods will be instrumental in increasing the number of alleles detected at MHC loci, resolving entire MHC haplotypes (thus facilitating detection of copy number and structural variation), and even detecting epigenetic modifications (Bruijnesteijn, 2023; Cheng et al., 2022; Karl et al., 2023; Viļuma et al., 2017; Fuselli et al., 2018; Maibach et al., 2017).
Although the primate MHC has been of interest to evolutionary biologists for more than 30 years, there is still much to be done to more fully document the evolution of the MHC genes within and between species. Moreover, we still have a limited understanding of how sequence changes map to functional differences among alleles, and how these relate to allele-specific profiles of pathogen protection (and autoimmunity risk). However, functional and computational advances will provide key opportunities for progress on these problems (Radwan et al., 2020; Vizcaíno et al., 2020).
Materials and methodsData
We downloaded MHC allele nucleotide sequences for all human and nonhuman genes from the IPD Database (updated January 2023) (Barker et al., 2023; Maccari et al., 2017; Maccari et al., 2020). To supplement the alleles available in the database, we also collected nucleotide sequences from NCBI using the Entrez E-utilities with query ‘histocompatibility AND txidX AND alive[prop]’, where X is a taxon of interest.
We wanted to provide ‘zoomed-in’ versions of various subtrees within the multi-gene trees presented in our companion paper (Fortier and Pritchard, 2025) Thus, we included more species and more alleles per species than in the original trees. In each tree, we also included a ‘backbone’ of sequences from the overall multi-gene tree to provide context for each expanded clade (lists of alleles provided as Supplementary file 1).
For Class I, we expanded the following clades: (1) MHC-A-related genes (MHC-A group), (2) MHC-B-related genes (MHC-B group), (3) MHC-C-related genes (MHC-C group), (4) MHC-E-related genes (MHC-E group), (5) MHC-F-related genes (MHC-F group), and (6) MHC-G-related genes (MHC-G group). For Class IIA, we expanded: (1) MHC-DMA-related genes (MHC-DMA group), (2) MHC-DOA-related genes (MHC-DOA group), (3) MHC-DRA-related genes (MHC-DRA group), (4) MHC-DPA-related genes (MHC-DPA group), and (5) MHC-DQA-related genes (MHC-DQA group). For Class IIB, we expanded: (1) MHC-DMB-related genes (MHC-DMB group), (2) MHC-DOB-related genes (MHC-DOB group), (3) MHC-DRB-related genes (MHC-DRB group), (4) MHC-DPB-related genes (MHC-DPB group), and (5) MHC-DQB-related genes (MHC-DQB group). These sets were inclusive of all orthologs and paralogs of a given human gene across all species we included (see our companion paper for more information; Fortier and Pritchard, 2025). For example, the MHC-A group includes human HLA-A and its 1:1 orthologs in the apes, the expanded MHC-A and -AG paralogs of the OWM, chimpanzee-specific Patr-AL, gorilla-specific Gogo/Gobe-OKO, orangutan-specific Poab/Popy-Ap, and pseudogenes MHC-H and -Y. Tables 2–4 provide an overview of which genes from which species were included in each of these named groups. See Supplementary file 1 for lists of all alleles included within each group.
Data summary for Class I.
Each row represents a species, and each column represents a gene group. Each cell lists the number of alleles included for each gene represented by that gene group. Bolded entries are ‘backbone’ sequences that are included in every group.
Each row represents a species, and each column represents a gene group. Each cell lists the number of alleles included for each gene represented by that gene group. Bolded entries are ‘backbone’ sequences that are included in every group.
Clade
Species Group
Species
Latin Name
Pref.
MHC-DPA Group
MHC-DQA Group
MHC-DRA Group
MHC-DMA Group
MHC-DOA Group
Ape
Human
Human
Homo sapiens
Hosa
22 –DPA, 2 –DPA, 2 –DQA, 1 –DRA, 1 –DMA, 1 –DOA
2 –DPA, 22 –DQA, 2 –DQA, 1 –DRA, 1 –DMA, 1 –DOA
2 –DPA, 2 –DQA, 4–DRA, 1 –DRA, 1 –DMA, 1 –DOA
2 –DPA, 2 –DQA, 1 –DRA, 8–DMA, 1 –DMA, 1 –DOA
2 –DPA, 2 –DQA, 1 –DRA, 1 –DMA, 14–DOA, 1–DOA
Chimpanzee
Bonobo
Pan paniscus
Papa
1 –DPA
2 –DQA
1 –DRA
1 –DMA
1 –DOA
Chimpanzee
Pan troglodytes
Patr
5 –DPA
6 –DQA
3 –DRA
1 –DMA
1 –DOA
Gorilla
Western gorilla
Gorilla gorilla
Gogo
3 –DPA
10 –DQA
1 –DRA
1 –DMA
1 –DOA
Orangutan
Sumatran orangutan
Pongo abelii
Poab
4 –DPA
6 –DQA
4 –DRA
1 –DMA
1 –DOA
Bornean orangutan
Pongo pygmaeus
Popy
4 –DPA
4 –DQA
2 –DRA
Gibbon
Silvery gibbon
Hylobates moloch
Hymo
1 –DPA
3 –DQA
1 –DRA
1 –DMA
1 –DOA
Northern white-cheeked gibbon
Nomascus leucogenys
Nole
1 –DPA
2 –DQA
1 –DRA
1 –DMA
1 –DOA
Lar gibbon
Hylobates lar
Hyla
6 –DQA
OWM
Baboon
Olive baboon
Papio anubis
Paan
13 –DPA
8 –DQA
3 –DRA
1 –DMA
1 –DOA
Hamadryas baboon
Papio hamadryas
Paha
1 –DPA
3 –DQA
Yellow baboon
Papio cynocephalus
Pacy
7 –DQA
Guinea baboon
Papio papio
Papp
4 –DQA
Gelada
Gelada
Theropithecus gelada
Thge
2 –DPA
3 –DQA
1 –DRA
1 –DMA
1 –DOA
Mangabey
Sooty mangabey
Cercocebus atys
Ceat
2 –DPA
2 –DQA
1 –DRA
1 –DMA
1 –DOA
Black crested mangabey
Lophocebus aterrimus
Loat
1 –DQA
Drill
Drill
Mandrillus leucophaeus
Male
1 –DPA
2 –DQA
1 –DRA
1 –DMA
1 –DOA
Macaque
Crab-eating macaque
Macaca fascicularis
Mafa
30 –DPA, 1 –DMA, 1 –DOA
11 –DQA, 1 –DMA, 1 –DOA
16 –DRA, 1 –DMA, 1 –DOA
7 –DMA, 1 –DMA, 1 –DOA
1 –DMA, 6 –DOA, 1 –DOA
Northern pig-tailed macaque
Macaca leonina
Malo
6 –DPA
8 –DQA
5 –DRA
2 –DMA
7 –DOA
Rhesus macaque
Macaca mulatta
Mamu
22 –DPA, 1 –DPA, 2 –DQA, 1 –DRA
1 –DPA, 9 –DQA, 2 –DQA, 1 –DRA
1 –DPA, 2 –DQA, 12 –DRA, 1 –DRA
1 –DPA, 2 –DQA, 1 –DRA, 4 –DMA
1 –DPA, 2 –DQA, 1 –DRA, 1 –DOA
Southern pig-tailed macaque
Macaca nemestrina
Mane
14 –DPA
10 –DQA
11 –DRA
1 –DMA
1 –DOA
Tibetan macaque
Macaca thibetana
Math
7 –DPA
1 –DQA
1 –DRA
10 –DMA
Stump-tailed macaque
Macaca arctoides
Maar
1 –DPA
2 –DQA
Grivet
Grivet
Chlorocebus aethiops
Chae
6 –DQA
Green Monkey
Green monkey
Chlorocebus sabaeus
Chsa
5 –DPA
2 –DQA
1 –DRA
1 –DOA
Guenon
Blue monkey
Cercopithecus mitis
Cemi
5 –DQA
De Brazza’s monkey
Cercopithecus neglectus
Cene
2 –DQA
Colobus
Angola colobus
Colobus angolensis
Coan
2 –DQA
1 –DRA
1 –DMA
1 –DOA
Ugandan red colobus
Piliocolobus tephrosceles
Pite
1 –DPA
1 –DQA
1 –DRA
1 –DMA
1 –DOA
Mantled guereza
Colobus guereza
Cogu
1 –DQA
Langur
Francois’ langur
Trachypithecus francoisi
Trfr
1 –DPA
1 –DQA
1 –DRA
1 –DMA
1 –DOA
Snub-Nosed Monkey
Black-and-white snub-nosed monkey
Rhinopithecus bieti
Rhbi
1 –DRA
1 –DMA
1 –DOA
Golden snub-nosed monkey
Rhinopithecus roxellana
Rhro
1 –DPA
1 –DQA
1 –DRA
1 –DMA
1 –DOA
NWM
Tamarin
Cotton-top tamarin
Saguinus oedipus
Saoe
4 –DQA
Marmoset
Common marmoset
Callithrix jacchus
Caja
1 –DPA, 2 –DQA, 1 –DMA, 1 –DOA
1 –DPA, 6 –DQA, 2 –DQA, 1 –DMA, 1 –DOA
1 –DPA, 2 –DQA, 1 –DRA, 1 –DMA, 1 –DOA
1 –DPA, 2 –DQA, 1 –DMA, 1 –DOA
1 –DPA, 2 –DQA, 1 –DMA, 1 –DOA
Night Monkey
Nancy Ma’s night monkey
Aotus nancymaae
Aona
1 –DPA, 1 –DRA
6 –DQA, 1 –DRA
2 –DRA, 1 –DRA
1 –DRA, 1 –DMA
1 –DRA, 1 –DOA
Gray-bellied night monkey
Aotus lemurinus
Aole
3 –DQA
Spix’s night monkey
Aotus vociferans
Aovo
3 –DRA
Capuchin
Panamanian white-faced capuchin
Cebus imitator
Ceim
1 –DPA
3 –DQA
1 –DRA
1 –DMA
1 –DOA
Tufted capuchin
Sapajus apella
Saap
1 –DPA
3 –DQA
1 –DRA
1 –DMA
1 –DOA
Squirrel Monkey
Black-capped squirrel monkey
Saimiri boliviensis
Sabo
2 –DQA
1 –DRA
1 –DMA
1 –DOA
Common squirrel monkey
Saimiri sciureus
Sasc
3 –DPA
Tarsier
Tarsier
Philippine tarsier
Carlito syrichta
Casy
3 –DPA
2 –DQA
1 –DRA
1 –DMA
1 –DOA
Strepsirrhini
Lemur
Ring-tailed lemur
Lemur catta
Leca
1 –DPA, 1 –DQA, 1 –DRA, 1 –DMA, 1 –DOA
1 –DPA, 1 –DQA, 1 –DRA, 1 –DMA, 1 –DOA
1 –DPA, 1 –DQA, 1 –DRA, 1 –DMA, 1 –DOA
1 –DPA, 1 –DQA, 1 –DRA, 1 –DMA, 1 –DOA
1 –DPA, 1 –DQA, 1 –DRA, 1 –DMA, 1 –DOA
Gray mouse lemur
Microcebus murinus
Mimu
1 –DPA
1 –DQA
1 –DRA
1 –DMA
1 –DOA
Loris
Sunda slow loris
Nycticebus coucang
Nyco
1 –DPA
2 –DRA
1 –DMA
1 –DOA
Galago
Northern greater galago
Otolemur garnettii
Otga
1 –DPA
1 –DQA
1 –DRA
1 –DMA
1 –DOA
Sifaka
Coquerel’s sifaka
Propithecus coquereli
Prco
2 –DPA
1 –DQA
1 –DRA
1 –DMA
1 –DOA
Flying Lemur
Flying Lemur
Sunda flying lemur
Galeopterus variegatus
Gava
2 –DPA
1 –DRA
1 –DMA
1 –DOA
Tree Shrew
Tree Shrew
Chinese tree shrew
Tupaia chinensis
Tuch
4 –DPA
1 –DQA
1 –DRA
1 –DMA
1 –DOA
Glires
Rodent
Groundhog
Marmota monax
Mamo
1 –DPA, 1 –DPA
1 –DPA
1 –DPA, 1 –DRA
1 –DPA, 1 –DMA
1 –DPA, 1 –DOA
Brown rat
Rattus norvegicus
Rano
1 –DQA, 1 –DRA, 1 –DMA, 1 –DOA
2 –DQA, 1 –DQA, 1 –DRA, 1 –DMA, 1 –DOA
1 –DQA, 2 –DRA, 1 –DRA, 1 –DMA, 1 –DOA
1 –DQA, 1 –DRA, 1 –DMA, 1 –DOA
1 –DQA, 1 –DRA, 1 –DMA, 1–DOA, 1–DOA
Pika
Plateau pika
Ochotona curzoniae
Occu
2 –DPA
2 –DQA
1 –DRA
1 –DMA
1 –DOA
American pika
Ochotona princeps
Ocpr
2 –DPA
1 –DQA
1 –DRA
1 –DMA
1 –DOA
Laurasiatheria
Artiodactyla
Bactrian camel
Camelus bactrianus
Caba
1 –DPA
1 –DQA
1 –DMA
1 –DOA
Wild boar
Sus scrofa
SLA
1 –DQA, 1 –DRA, 1 –DMA, 1 –DOA
1 –DQA, 1 –DQA, 1 –DRA, 1 –DMA, 1 –DOA
1 –DQA, 3 –DRA, 1 –DRA, 1 –DMA, 1 –DOA
1 –DQA, 1 –DRA, 4 –DMA, 1 –DMA, 1 –DOA
1 –DQA, 1 –DRA, 1 –DMA, 1 –DOA, 1 –DOA
Even-toed ungulates
Bos sp.
BoLA
2 –DQA
Domestic yak
Bos grunniens
Bogr
2 –DQA
Water buffalo
Bubalus bubalis
Bubu
2 –DQA
Sheep
Ovis aries
Ovar
4 –DQA
3 –DRA
Dromedary camel
Camelus dromedarius
Cadr
1 –DRA
Bighorn sheep
Ovis canadensis
Ovca
3 –DRA
Ferungulata
Sea otter
Enhydra lutris
Enlu
1 –DQA
1 –DRA
1 –DMA
1 –DOA
Cat
Felis catus
Feca
1 –DRA
1 –DMA
1 –DOA
Sunda pangolin
Manis javanica
Maja
1 –DQA
1 –DRA
1 –DMA
1 –DOA
Cougar
Puma concolor
Puco
1 –DPA
1 –DMA
1 –DOA
Jaguarundi
Puma yagouaroundi
Puya
1 –DPA
1 –DRA
1 –DMA
1 –DOA
Steller sea lion
Eumetopias jubatus
Euju
1 –DPA
1 –DQA
Horse
Equus caballus
Eqca
5 –DQA
3 –DRA
Bat
Big brown bat
Eptesicus fuscus
Epfu
1 –DPA
2 –DQA
1 –DRA
1 –DMA
Kuhl’s pipistrelle
Pipistrellus kuhlii
Piku
1 –DPA
1 –DQA
1 –DRA
1 –DMA
Large flying fox
Pteropus vampyrus
Ptva
2 –DQA
1 –DRA
1 –DOA
Mole
Star-nosed mole
Condylura cristata
Cocr
3 –DQA
1 –DOA
Atlantogenata
Xenarthra
Linnaeus’s two-toed sloth
Choloepus didactylus
Chdi
1 –DPA
1 –DQA
1 –DRA
1 –DMA
1 –DOA
Nine-banded armadillo
Dasypus novemcinctus
Dano
1 –DPA
2 –DQA
1 –DRA
1 –DMA
1 –DOA
Afrotheria
Cape golden mole
Chrysochloris asiatica
Chas
1 –DPA
1 –DRA
1 –DMA
1 –DOA
Cape elephant shrew
Elephantulus edwardii
Eled
1 –DPA
1 –DPA, 1 –DQA
1 –DPA, 1 –DRA
1 –DPA, 1 –DMA
1 –DPA, 1 –DOA
Aardvark
Orycteropus afer
Oraf
1 –DPA
1 –DQA
1 –DMA
1 –DOA
West Indian manatee
Trichechus manatus
Trma
2 –DPA
1 –DQA
1 –DRA
1 –DMA
Lesser hedgehog tenrec
Echinops telfairi
Ecte
1 –DRA
1 –DOA
Data summary for Class IIB.
Each row represents a species, and each column represents a gene group. Each cell lists the number of alleles included for each gene represented by that gene group. Bolded entries are ‘backbone’ sequences that are included in every group.
Clade
Species Group
Species
Latin Name
Pref.
MHC-DPB Group
MHC-DQB Group
MHC-DRB Group
MHC-DMB Group
MHC-DOB Group
Ape
Human
Human
Homo sapiens
Hosa
74 –DPB, 2 –DPB, 2 –DQB, 9 –DRB, 1 –DMB, 1 –DOB
2 –DPB, 24 –DQB, 2 –DQB, 9 –DRB, 1 –DMB, 1 –DOB
2 –DPB, 2 –DQB, 46 –DRB, 9 –DRB, 1 –DMB, 1 –DOB
2 –DPB, 2 –DQB, 9 –DRB, 6 –DMB, 1 –DMB, 1 –DOB
2 –DPB, 2 –DQB, 9 –DRB, 1 –DMB, 14 –DOB, 1 –DOB
Chimpanzee
Bonobo
Pan paniscus
Papa
8 –DPB
2 –DQB
5 –DRB
2 –DMB
1 –DOB
Chimpanzee
Pan troglodytes
Patr
6 –DPB
9 –DQB
17 –DRB
2 –DMB
2 –DOB
Gorilla
Western gorilla
Gorilla gorilla
Gogo
5 –DPB
10 –DQB
7 –DRB
2 –DMB
1 –DOB
Orangutan
Sumatran orangutan
Pongo abelii
Poab
5 –DPB
6 –DQB
7 –DRB
1 –DMB
1 –DOB
Bornean orangutan
Pongo pygmaeus
Popy
5 –DPB
3 –DQB
7 –DRB
3 –DMB
Gibbon
Silvery gibbon
Hylobates moloch
Hymo
1 –DPB
2 –DQB
5 –DRB
1 –DMB
1 –DOB
Northern white-cheeked gibbon
Nomascus leucogenys
Nole
1 –DPB
2 –DQB
1 –DMB
1 –DOB
Lar gibbon
Hylobates lar
Hyla
4 –DQB
OWM
Baboon
Olive baboon
Papio anubis
Paan
5 –DPB
5 –DQB
11 –DRB
1 –DMB
Hamadryas baboon
Papio hamadryas
Paha
3 –DQB
1 –DRB
Chacma baboon
Papio ursinus
Paur
8 –DRB
Gelada
Gelada
Theropithecus gelada
Thge
2 –DPB
1 –DQB
3 –DRB
1 –DMB
1 –DOB
Mangabey
Sooty mangabey
Cercocebus atys
Ceat
3 –DPB
2 –DQB
1 –DRB
1 –DMB
1 –DOB
Drill
Drill
Mandrillus leucophaeus
Male
1 –DPB
1 –DRB
1 –DMB
1 –DOB
Mandrill
Mandrill
Mandrillus sphinx
Masp
10 –DRB
Macaque
Crab-eating macaque
Macaca fascicularis
Mafa
7 –DPB, 2 –DRB, 1 –DMB, 1 –DOB
5 –DQB, 2 –DRB, 1 –DMB, 1 –DOB
6 –DRB, 2 –DRB, 1 –DMB, 1 –DOB
2 –DRB, 4 –DMB, 1 –DMB, 1 –DOB
2 –DRB, 1 –DMB, 6 –DOB, 1 –DOB
Northern pig-tailed macaque
Macaca leonina
Malo
5 –DPB
5 –DQB
3 –DRB
2 –DMB
3 –DOB
Rhesus macaque
Macaca mulatta
Mamu
5 –DPB, 2 –DPB, 1 –DQB, 4 –DRB
2 –DPB, 4 –DQB, 1 –DQB, 4 –DRB
2 –DPB, 1 –DQB, 10 –DRB, 4 –DRB
2 –DPB, 1 –DQB, 4 –DRB, 5 –DMB
2 –DPB, 1 –DQB, 4 –DRB, 1 –DOB
Southern pig-tailed macaque
Macaca nemestrina
Mane
5 –DPB
5 –DQB
6 –DRB
1 –DMB
1 –DOB
Tibetan macaque
Macaca thibetana
Math
13 –DPB
6 –DQB
1 –DRB
1 –DMB
1 –DOB
Stump-tailed macaque
Macaca arctoides
Maar
5 –DQB
2 –DRB
Japanese macaque
Macaca fuscata
Mafu
3 –DRB
Lion-tailed macaque
Macaca silenus
Masi
3 –DRB
Grivet
Grivet
Chlorocebus aethiops
Chae
3 –DQB
6 –DRB
Green Monkey
Green monkey
Chlorocebus sabaeus
Chsa
3 –DPB
4 –DQB
7 –DRB
1 –DMB
1 –DOB
Colobus
Angola colobus
Colobus angolensis
Coan
2 –DPB
2 –DQB
1 –DMB
1 –DOB
Ugandan red colobus
Piliocolobus tephrosceles
Pite
1 –DPB
1 –DQB
3 –DRB
1 –DMB
1 –DOB
Langur
Francois’ langur
Trachypithecus francoisi
Trfr
2 –DPB
1 –DQB
3 –DRB
1 –DMB
1 –DOB
Gray langur
Semnopithecus entellus
Seen
1 –DPB
Snub-Nosed Monkey
Black-and-white snub-nosed monkey
Rhinopithecus bieti
Rhbi
1 –DMB
1 –DOB
Golden snub-nosed monkey
Rhinopithecus roxellana
Rhro
1 –DPB
1 –DQB
2 –DRB
1 –DMB
1 –DOB
NWM
Tamarin
Cotton-top tamarin
Saguinus oedipus
Saoe
3 –DPB
4 –DQB
8 –DRB
White-lipped tamarin
Saguinus labiatus
Sala
3 –DRB
Marmoset
Common marmoset
Callithrix jacchus
Caja
1 –DPB, 1 –DPB, 2 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
1 –DPB, 3 –DQB, 2 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
1 –DPB, 2 –DQB, 3 –DRB, 1 –DRB, 1 –DMB, 1 –DOB
1 –DPB, 2 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
1 –DPB, 2 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
Night Monkey
Nancy Ma’s night monkey
Aotus nancymaae
Aona
4 –DPB
3 –DQB
6 –DRB
1 –DMB
1 –DOB
Gray-bellied night monkey
Aotus lemurinus
Aole
3 –DPB
1 –DQB
Azara’s night monkey
Aotus azarae
Aoaz
2 –DRB
Black-headed night monkey
Aotus nigriceps
Aoni
3 –DRB
Three-striped night monkey
Aotus trivirgatus
Aotr
3 –DRB
Spix’s night monkey
Aotus vociferans
Aovo
3 –DRB
Capuchin
Panamanian white-faced capuchin
Cebus imitator
Ceim
1 –DPB
3 –DQB
1 –DRB
1 –DMB
1 –DOB
Tufted capuchin
Sapajus apella
Saap
1 –DPB
4 –DQB
3 –DRB
1 –DMB
Squirrel Monkey
Black-capped squirrel monkey
Saimiri boliviensis
Sabo
1 –DPB
3 –DQB
1 –DMB
1 –DOB
Common squirrel monkey
Saimiri sciureus
Sasc
2 –DRB
Spider Monkey
White-bellied spider monkey
Ateles belzebuth
Atbe
2 –DRB
Howler Monkey
Guatemalan black howler
Alouatta pitta
Alpi
2 –DRB
Saki
White-faced saki
Pithecia pithecia
Pipi
3 –DRB
Tarsier
Tarsier
Philippine tarsier
Carlito syrichta
Casy
2 –DPB
1 –DQB
2 –DRB
1 –DMB
1 –DOB
Strepsirrhini
Lemur
Ring-tailed lemur
Lemur catta
Leca
1 –DPB, 1 –DPB, 1 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
1 –DPB, 1 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
1 –DPB, 1 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
1 –DPB, 1 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
1 –DPB, 1 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
Gray mouse lemur
Microcebus murinus
Mimu
1 –DPB
3 –DQB
2 –DRB
1 –DMB
1 –DOB
Loris
Sunda slow loris
Nycticebus coucang
Nyco
1 –DPB
Galago
Northern greater galago
Otolemur garnettii
Otga
1 –DPB
1 –DQB
1 –DRB
1 –DMB
1 –DOB
Sifaka
Coquerel’s sifaka
Propithecus coquereli
Prco
1 –DPB
1 –DQB
1 –DMB
1 –DOB
Flying Lemur
Flying Lemur
Sunda flying lemur
Galeopterus variegatus
Gava
2 –DPB
2 –DRB
1 –DMB
1 –DOB
Tree Shrew
Tree Shrew
Chinese tree shrew
Tupaia chinensis
Tuch
1 –DPB
1 –DMB
1 –DOB
Glires
Rodent
Groundhog
Marmota monax
Mamo
1 –DPB, 1 –DPB
1 –DPB
1 –DPB
1 –DPB, 1–DMB
1 –DPB
Brown rat
Rattus norvegicus
Rano
1 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
2 –DQB, 1 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
1 –DQB, 3 –DRB, 1 –DRB, 1 –DMB, 1 –DOB
1 –DQB, 1 –DRB, 1 –DMB, 1 –DMB, 1 –DOB
1 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
Pika
American pika
Ochotona princeps
Ocpr
1 –DQB
1 –DRB
1 –DMB
Plateau pika
Ochotona curzoniae
Occu
1 –DPB
2 –DQB
1 –DOB
Laurasiatheria
Artiodactyla
Bactrian camel
Camelus bactrianus
Caba
1 –DMB
1 –DOB
Dromedary camel
Camelus dromedarius
Cadr
1 –DPB
1 –DPB
1 –DPB
1 –DPB
1 –DPB
Wild boar
Sus scrofa
SLA
1 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
2 –DQB, 1 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
1 –DQB, 4 –DRB, 1 –DRB, 1 –DMB, 1 –DOB
1 –DQB, 1 –DRB, 1 –DMB, 1 –DOB
1 –DQB, 1 –DRB, 1 –DMB, 2 –DOB, 1 –DOB
Even-toed ungulates
Bos sp.
BoLA
1 –DQB
2 –DRB
Water buffalo
Bubalus bubalis
Bubu
1 –DQB
Wild Bactrian camel
Camelus ferus
Cafe
1 –DQB
Sheep
Ovis aries
Ovar
6 –DQB
1 –DRB
Goat
Capra hircus
Cahi
2 –DRB
Bighorn sheep
Ovis canadensis
Ovca
1 –DRB
Ferungulata
Horse
Equus caballus
Eqca
6 –DQB
3 –DRB
5 –DMB
3 –DOB
Sunda pangolin
Manis javanica
Maja
1 –DMB
1 –DOB
Cougar
Puma concolor
Puco
1 –DPB
1 –DMB
Jaguarundi
Puma yagouaroundi
Puya
1 –DMB
Northern elephant seal
Mirounga angustirostris
Mian
1 –DRB
1 –DOB
Sea otter
Enhydra lutris
Enlu
1 –DPB
1 –DQB
Steller sea lion
Eumetopias jubatus
Euju
1 –DQB
Bat
Big brown bat
Eptesicus fuscus
Epfu
2 –DQB
1 –DRB
1 –DMB
Large flying fox
Pteropus vampyrus
Ptva
1 –DQB
1 –DOB
Kuhl’s pipistrelle
Pipistrellus kuhlii
Piku
1 –DPB
1 –DQB
Mole
Star-nosed mole
Condylura cristata
Cocr
2 –DQB
1 –DRB
1 –DMB
1 –DOB
Atlantogenata
Xenarthra
Linnaeus’s two-toed sloth
Choloepus didactylus
Chdi
2 –DPB
1 –DQB
1 –DMB
1 –DOB
Nine-banded armadillo
Dasypus novemcinctus
Dano
1 –DPB
Afrotheria
Aardvark
Orycteropus afer
Oraf
1 –DQB
1 –DOB
Cape elephant shrew
Elephantulus edwardii
Eled
1 –DPB
West Indian manatee
Trichechus manatus
Trma
1 –DPB
Lesser hedgehog tenrec
Echinops telfairi
Ecte
1 –DRB
We aligned each group separately using MUSCLE (Edgar, 2004) with default settings and manually adjusted, following the alignments we already produced for the multi-gene trees in our companion paper (Fortier and Pritchard, 2025).
Nucleotide diversity
The classical MHC region is defined as chr6:28,510,120–33,480,577 (GRCh38) (Genome Reference Consortium, 2022). Nucleotide diversity (π) was calculated on modern human data from the 1000 Genomes Project (Auton et al., 2015) using VCFtools (0.1.15) (Danecek et al., 2011). For the entire MHC region (Figure 1A), π was calculated in 5000 bp sliding windows with a step size of 1000 bp. For each gene separately (Figure 1B and C, Figure 1—figure supplement 1), π was calculated in 50 bp sliding windows with a step size of 10 bp.
Bayesian phylogenetic analysis
We constructed phylogenetic trees using BEAST2 (Bouckaert et al., 2014; Bouckaert et al., 2019) with package SubstBMA (Wu et al., 2013). SubstBMA implements a spike-and-slab mixture model that simultaneously estimates the phylogenetic tree, the number of site partitions, the assignment of sites to partitions, the nucleotide substitution model, and a rate multiplier for each partition. Since we were chiefly interested in the partitions and their rate multipliers, we used the RDPM model as described by Wu et al., 2013. In the RDPM model, the number of nucleotide substitution model categories is fixed to 1, so that all sites, regardless of rate partition, share the same estimated nucleotide substitution model. This reduces the number of parameters to be estimated and ensures that only evolutionary rates vary across site partitions, reducing overall model complexity. We used an uncorrelated lognormal relaxed molecular clock because, in reality, evolutionary rates vary among branches (Bergeron et al., 2023).
Priors
For the Dirichlet process priors, we used the informative priors constructed by Wu et al., 2013 for their mammal dataset. This is appropriate because they include several of the same species and their mammals span approximately the same evolutionary time that we consider in our study. We also use their same priors on tree height, base rate distribution, and a Yule process coalescent prior. We did not specify a calibration point—a time-based prior on a node—because we did not expect our sequences to group according to the species tree.
Running <italic>BEAST2</italic>
We aligned sequences across genes and species and ran BEAST2 on various subsets of the alignment. For the Class I gene groups (MHC-A group, MHC-B group, MHC-C group, MHC-E group, MHC-F group, and MHC-G group), we repeated the analysis for (1) exon 2 only (PBR), (2) exon 3 only (PBR), (3) exon 4 only (non-PBR), and (4) exons 1, 5, 6, 7, and 8 together (non-PBR; ‘other’ exons). For the Class IIA gene groups (MHC-DMA group, MHC-DOA group, MHC-DRA group, MHC-DPA group, and MHC-DQA group), we used (1) exon 2 only (PBR), (2) exon 3 only (non-PBR), and (3) exons 1, 3, 4, and 5 together (non-PBR; ‘other’ exons). For Class IIB gene groups (MHC-DMB group, MHC-DOB group, MHC-DRB group, MHC-DPB group, and MHC-DQB group), we analyzed (1) exon 2 only (PBR), (2) exon 3 only (non-PBR), and (3) exons 1, 3, 4, and 5 together (non-PBR; ‘other’ exons). In the following, each ‘analysis’ is a collection of BEAST2 runs using one of these sets of exons of a particular gene group.
The XML files we used to run BEAST2 were based closely on those used for the mammal dataset with the RDPM model and uncorrelated relaxed clock in Wu et al., 2013 (https://github.com/jessiewu/substBMA/blob/master/examples/mammal/mammal_rdpm_uc.xml; Vaughan et al., 2018). Running a model with per-site evolutionary rate categories and a relaxed clock means there are many parameters to estimate. Along with the large number of parameters, the highly polymorphic and often highly diverged sequences in our alignments make it difficult for BEAST2 to explore the state space. Thus, we undertook considerable effort to ensure good mixing and convergence of the chains. First, we employed coupled MCMC for all analyses. Coupled MCMC is essentially the same as the regular MCMC used in BEAST2, except that it uses additional ‘heated’ chains with increased acceptance probabilities that can traverse unfavorable intermediate states and allow the main chain to move away from an inferior local optimum (Müller and Bouckaert, 2020). Using coupled MCMC both speeds up BEAST2 runs and improves mixing and convergence. We used four heated chains for each run with a delta temperature of 0.025. Second, we ran each BEAST2 run for 40,000,000 states, discarding the first 4,000,000 states as burn-in and sampling every 10,000 states. Third, we ran at least 8 independent replicates of each analysis. The replicates use the exact same alignment, but explore state space independently and thus are useful for improving the effective sample size of tricky parameters. As recommended by BEAST2, we examined all replicates in Tracer version 1.7.2 (Rambaut et al., 2018) to ensure that they were sampling from the same parameter distributions and had reached convergence. We excluded replicates for which this was not true, as these chains were probably stuck in suboptimal state space. Additionally, Tracer provides estimates of the effective sample size (ESS) for the combined set of states from all chosen replicates, and we required that the ESS be larger than 100 for all parameters. If there were fewer than 4 acceptable replicates or if the ESS was below 100 for any parameter, we re-ran more independent replicates of the analysis until these requirements were satisfied. We obtained between 5 and 18 acceptable replicates per analysis (median 8).
For some analyses, computational limitations prevented BEAST2 from being able to reach 40,000,000 states. In these situations, more replicates (of fewer states) were usually required to achieve good mixing and convergence. The first 4,000,000 states from each run were still discarded as burn-in even though this represented more than 10% of states in these cases.
This stringent procedure ensured that all of the replicates were exploring the same parameter space and were converging upon the same global optimum, allowing the ≥4 independent runs to be justifiably combined. We combined the acceptable replicates using LogCombiner version 2.6.7 (Drummond and Rambaut, 2007), which aggregates the results across all states. We then used the combined results to perform downstream analyses.
The XML files required to run BEAST2 are provided as Source code 1.
Phylogenetic trees
After combining acceptable replicates, we obtained 12,382–64,818 phylogenies per group/gene region (mean 34,499). These trees are provided in https://doi.org/10.5061/dryad.zcrjdfnrz. We used TreeAnnotator version 2.6.3 (Drummond and Rambaut, 2007) to summarize each set of possible trees as a maximum clade credibility tree, which is the tree that maximizes the product of posterior clade probabilities. Since BEAST2 samples trees from the posterior, one could in principle perform model testing directly from the posterior samples; the complete set of trees can typically be reduced to a smaller 95% credible set of trees representing the ‘true’ tree (BEA, 2024). However, given the high complexity of the model space, all our posterior trees were unique, meaning this was not possible in practice. (Since the prior over tree topologies is unstructured, this effectively puts minuscule prior weight on trees with monophyly. Thus, sampling directly from the posterior provides an unacceptably high-variance estimator.).
Gene conversion
We calculated gene conversion fragments using GENECONV version 1.81a (Sawyer, 1999) on each alignment. It is generally advisable to use only synonymous sites when running the program on a protein-coding alignment, since silent sites within the same codon position are likely to be correlated. However, the extreme polymorphism in these MHC genes meant there were too few silent sites to use in the analysis. Thus, we considered all sites but caution that this could slightly overestimate the lengths of our inferred conversion tracts. However, we were mainly concerned with the presence of a conversion tract rather than its precise length. For each alignment, we ran GENECONV with options ListPairs, Allouter, Numsims = 10000, and Startseed = 3 10. We collected all inferred ‘Global Inner’ (GI) fragments with sim_pval<0.05\begin{document}$ sim\_pval \lt 0.05$\end{document} (this is pre-corrected for multiple comparisons by the program). GI fragments represent a possible gene conversion event between two sequences in the alignment.
For each GI fragment, we made an educated guess on which sequence was the donor sequence and which was the acceptor sequence by comparing how sequences clustered using sites within the fragment bounds to how sequences clustered using sites outside of the fragment bounds (but within the same exon). Sequences that were determined to be acceptor sequences were excluded from the Bayes factor analyses for the relevant exon because their non-tree-like behavior has the potential to bias results. For sequence pairs where the direction could not be determined, both sequences were excluded from subsequent analyses.
Bayes factors
Because we could not perform model testing directly on the full phylogenies, we used an alternative approach—computing Bayes factors for TSP within manageable subsets of the data, i.e. quartets of alleles. Let D\begin{document}$ D$\end{document} be a sample of phylogenies from BEAST2, sampled from the posterior with uniform prior. For a chosen species, we have a null hypothesis H\begin{document}$H$\end{document}, that human alleles form a monophyletic group, and an alternative hypothesis, Hc\begin{document}$ H^{c}$\end{document}, that is also the complement of H\begin{document}$ H$\end{document}—that the human alleles do not form a monophyletic group. The Bayes factor, K\begin{document}$ K$\end{document}, is a ratio quantifying support for the alternative hypothesis:K=Pr(D|Hc)Pr(D|H)=Pr(Hc|D)Pr(H|D)⋅Pr(H)Pr(Hc)\begin{document}$$\displaystyle K=\frac{{\rm Pr}(D|H^c)}{\Pr(D|H)}=\frac{{\rm Pr}(H^c|D)}{{\rm Pr}(H|D)}\cdot\frac{{\rm Pr}(H)}{{\rm Pr}(H^c)}$$\end{document}
where the first term on the right-hand side is the posterior odds in favor of the alternative hypothesis and the second term is the prior odds in favor of the null hypothesis. Bayes factors above 100 are considered decisive support for the alternative hypothesis (Jeffreys, 1998).
Because it is difficult to evaluate monophyly using a large number of alleles, we evaluate Bayes factors considering four alleles at a time: two alleles of a single species and two alleles of different species. For example, to assess support for TSP between humans and chimpanzees, we could use two human alleles and two bonobo alleles. Or, to assess support for TSP between humans and OWM, we could use two human alleles, one baboon, and one macaque allele. Because there are many possible sets of four alleles for each comparison, we tested a large number of quartets. We reported the maximum Bayes factor among all tested allele sets to represent evidence for TSP for that species comparison, because our aim was to find any evidence of TSP among any set of four alleles.
Next, we calculated the prior odds of the null hypothesis (that the chosen species, i.e. humans, form a monophyletic group). The prior odds Pr(H)Pr(Hc)=12\begin{document}$\frac{{\rm Pr}(H)}{{\rm Pr}(H^{c})}=\frac{1}{2}$\end{document}, because if the trees were assembled at random, there is one possible unrooted tree where the two human alleles would form a monophyletic group and two possible unrooted trees where the two human alleles would not form a monophyletic group, as shown in Figure 7.
Possible unrooted trees of 4 alleles.
There is one tree where the human alleles are monophyletic, and two trees where they are non-monophyletic.
The data, D\begin{document}$ D$\end{document}, is the set of BEAST2 trees, so the posterior odds Pr(Hc|D)Pr(H|D)\begin{document}$\frac{{\rm Pr}(H^{c}|D)}{{\rm Pr}(H|D)}$\end{document} is the fraction of BEAST2 trees where the two human alleles do not form a monophyletic group divided by the fraction of BEAST2 trees where the two human alleles do form a monophyletic group. If either fraction is 0, we set its probability to p=1n+1\begin{document}$ p=\frac{1}{n+1}$\end{document}, where n\begin{document}$ n$\end{document} is the number of BEAST2 trees for that gene/sequence subset, and set the complement’s probability to 1−p\begin{document}$1-p$\end{document}. This is the reason that some labels in Figures 3 and 4 contain a > sign (e.g. if no trees in a set of 14,000 were monophyletic, then the Bayes factor must be at minimum 7,000).
Bayes factors K\begin{document}$ K$\end{document} were then computed as follows and interpreted according to the scale given by Jeffreys, 1998.K=Pr(Hc|D)Pr(H|D)⋅12\begin{document}$$\displaystyle K=\frac{{\rm Pr}(H^c|D)}{{\rm Pr}(H|D)}\cdot\frac{1}{2}$$\end{document}
For each gene and genic region, we tested for TSP between human and chimpanzee, gorilla, orangutan, gibbon, OWM, and NWM. For the Class II genes, which have orthologs beyond the primates, we also tested for TSP between human and tarsier, Strepsirrhini, Flying Lemur, Treeshrew, Glires, Laurasiatheria, and Atlantogenata. For the Class I genes, we considered outgroup species to be the tarsier and Strepsirrhini.
Rapidly-evolving sites
BEAST2 places sites into partitions and estimates evolutionary rates for each partition. We averaged these rates over all sampled states, resulting in an overall (relative) evolutionary rate for each nucleotide position. We then normalized these rates. In designing the gene groups, we included common ‘backbone sequences’ in every set, although we expanded one particular clade for each focused tree. The inclusion of backbone genes spanning the whole MHC family caused the alignments to contain many gaps. To normalize the rates, we took advantage of the fact that every alignment had many mostly-gap sites. We defined ‘gappy’ sites as those in which the alignment had a gap in all of the following human backbone alleles: HLA-A*01:01:01:01, HLA-B*07:02:01:01, HLA-C*01:02:01:01, HLA-E*01:01:01:01, HLA-F*01:01:01:01, HLA-G*01:01:01:01, and HLA-J*01:01:01:01 for Class I, HLA-DRA*01:01:01:01, HLA-DQA1*01:01:01:01, HLA-DPA1*01:03:01:01, HLA-DMA*01:01:01:01, and HLA-DOA*01:01:01:01 for Class IIA, and HLA-DRB1*01:01:01:01, HLA-DQB1*02:01:01:01, HLA-DPB1*01:01:01:01, HLA-DMB*01:01:01:01, and HLA-DOB*01:01:01:01 for Class IIB. Because mostly-gap sites are not expected to affect the BEAST2 run very much and were common to all focused gene group alignments, we considered these sites’ BEAST2 evolutionary rates as a baseline. Since the rates obtained from SubstBMA are relative anyway, we simply needed a set of sites that behaved similarly across each BEAST2 run so that we could normalize the rates in a consistent manner. For each group and gene region, we calculated the mean rate among all of these baseline gappy sites. Then, we expressed normalized per-site rates as fold changes by taking the base-2 logarithm of the ratio of each site’s evolutionary rate to the baseline mean.
Protein structures
To map the rapidly-evolving sites onto protein structures, we first translated our nucleotide alignments into protein sequences using Expasy translate (Gasteiger et al., 2003). We then aligned our translated sequences with amino acid sequences from selected Protein Data Bank (PDB) (Berman et al., 2000) (https://www.rcsb.org/) structures (Table 5) using MUSCLE (Edgar, 2004) with default settings.
Structures used to calculate distances to peptide.
This table lists the Protein Data Bank (Berman et al., 2000) structure codes and references for all structures used to calculate peptide distances.
Gene
Struct.
Reference
MHC-A
1ZVS
Chu et al., 2007
3JTT
Dai et al., 2010
3OX8
Liu et al., 2011
3OXR
Liu et al., 2011
3OXS
Liu et al., 2011
3RL2
Zhang et al., 2011
4HX1
Niu et al., 2013
6J1V
Zhu et al., 2019
6J1W
Zhu et al., 2019
6MPP
Flores-Solis et al., 2019
6PBH
van de Sandt et al., 2019
7SR0
Finton et al., 2023
7SRK
Finton et al., 2023
7WT5
Asa et al., 2022
8I5C
Lu et al., 2023
MHC-B
1JGD
Hillig et al., 2004
3BVN
Kumar et al., 2009
3KPL
Macdonald et al., 2009
3KPN
Macdonald et al., 2009
3LN4
Bade-Doding et al., 2011
3LN5
Bade-Doding et al., 2011
3RWJ
Wu et al., 2011
3W39
Yagita et al., 2013
3X13
Saunders et al., 2015
4JQV
Rist et al., 2013
4JRY
Liu et al., 2013
4MJI
Motozono et al., 2014
4O2E
Sun et al., 2014
4PRA
Liu et al., 2014
4PRB
Liu et al., 2014
5EO0
Du et al., 2016
5IEK
Alpizar et al., 2016
5VUD
Illing et al., 2018
5VVP
Illing et al., 2018
5VWF
Illing et al., 2018
6IWG
Yamamoto et al., 2019
6MTM
Grant et al., 2018
6PYL
Lim Kam Sian et al., 2019
6PYV
Lim Kam Sian et al., 2019
6UZP
Schutte et al., 2020
6VIU
Schutte et al., 2020
6Y27
Loll et al., 2020
7R7V
Li et al., 2023
7T0L
Vivian and Rossjohn, 2022
7TUC
Jiang et al., 2022a
7X1 C
Huan et al., 2023
7YG3
Jiang et al., 2022b
MHC-C
4NT6
Choo et al., 2014
5VGD
Kaur et al., 2017
5VGE
Kaur et al., 2017
5W67
Mobbs et al., 2017
6PAG
Moradi et al., 2021
7WJ3
Asa et al., 2022
MHC-E
2ESV
Hoare et al., 2006
3CDG
Petrie et al., 2008
5W1V
Sullivan et al., 2017
7P49
Walters et al., 2022
7P4B
Walters et al., 2022
MHC-F
5IUE
Dulberger et al., 2017
MHC-G
1YDP
Clements et al., 2005
2DYP
Shiroishi et al., 2006
3KYN
Walpole et al., 2010
MHC-DM
1HDM
Mosyak et al., 1998
2BC4
Nicholson et al., 2006
4FQX
Pos et al., 2012
4GBX
Pos et al., 2012
4I0P
Guce et al., 2013
MHC-DO
4I0P
Guce et al., 2013
MHC-DP
3LQZ
Dai et al., 2010
3WEX
Kusano et al., 2014
7T2A
Ciacchi et al., 2023
7T6I
Klobuch et al., 2022
7ZAK
Racle et al., 2023
MHC-DQ
2NNA
Henderson et al., 2007
4D8P
Tollefsen et al., 2012
5KSA
Petersen et al., 2016
5KSU
Nguyen et al., 2017
6DIG
Jiang et al., 2019
6PX6
Ting et al., 2020
MHC-DR
1BX2
Smith et al., 1998
1FV1
Li et al., 2000
1H15
Lang et al., 2002
1T5X
Zavala-Ruiz et al., 2004
2Q6W
Parry et al., 2007
3C5J
Dai et al., 2008
4FQX
Pos et al., 2012
4H1L
Yin et al., 2012
5JLZ
Gerstner et al., 2016
5V4M
Ooi et al., 2017
6ATF
Scally et al., 2017
8EUQ
Kassardjian et al., 2023
The translated alignments and the PDB structures’ sequences allowed us to determine which nucleotides in the alignment corresponded to which amino acids in each structure. We then calculated per-amino-acid evolutionary rates by averaging the per-site evolutionary rates among the sites composing each codon. We caution that the vast majority of PDB structures we considered were from human (and were functional), and thus they cannot capture all of the indels and structural changes that define different primate proteins. Additionally, some of our alignments included pseudogenes and diverse sets of related genes, for which it may seem reductionist to map to a single human protein structure. However, we note that despite significant sequence variation, the structures of different MHC proteins (from different genes within the same class) are extremely similar, and thus we do not expect such complications to alter the overall conclusions.
We used PyMOL version 2.4.2 (Schrödinger, LLC, 2021) to visualize the per-amino acid evolutionary rates on each gene’s protein structure. We used model 4BCE (Teze et al., 2014) for HLA-B, 4NT6 (Choo et al., 2014) for HLA-C, 7P4B (Walters et al., 2022) for HLA-E, 5JLZ (Gerstner et al., 2016) for HLA-DR and 2NNA (Henderson et al., 2007) for HLA-DQ from Protein Data Bank (Berman et al., 2000) to prepare the main figures (https://www.rcsb.org/).
We calculated the distances between all atoms of all amino acids of the HLA molecule and all atoms of all amino acids of the peptide in PyMOL, then took the minimum distance to represent each amino acid’s distance to the peptide. Where possible, we averaged the minimum distances over multiple alternative structures for each protein, to avoid relying too heavily on a particular structure. The structures chosen (Table 5) represented different alleles, different contexts (e.g. bound to a receptor or not, bound to a self or non-self peptide), and in a few cases different species.
Disease and trait literature
We conducted a literature search for papers that used HLA fine-mapping to discover disease and trait associations, limiting our selection to those including at least 1000 cases and which identified putatively independent signals via conditional analysis. We included all independent amino acid signals identified as significant by the original authors. If there was more than one study for the same disease/trait, but in different populations, we included all unique independent hits. We also collected associations between amino acids and TCR phenotypes.
For Figure 6—figure supplement 7, we counted the number of significant, unique trait associations for each amino acid and plotted them against each amino acid’s evolutionary rate. We drew simple linear regression lines to evaluate the association for each gene. We caution that this simple analysis does not take into account the significance level of each hit nor the ranking (e.g. top hit, second independent hit) of the associations, and that each study’s authors may have performed conditional analyses differently. References are listed in Table 1 and in Table 1—source data 1.
Additional informationCompeting interests
No competing interests declared
Author contributions
Data curation, Software, Formal analysis, Investigation, Visualization, Methodology, Writing – original draft
This zip file contains all xml files we used to run <italic>BEAST2</italic> with <italic>SubstBMA</italic> on each gene group/exon alignment.
This zip file contains lists of IPD database allele names or RefSeq accession numbers for the sequences we analyzed in each gene group.
Data availability
The current manuscript is a computational study, and all data used is publicly available. Supplementary file 1 and Source code 1 contains lists of alleles used in this study and xml files for running BEAST2, respectively. Sets of posterior trees from BEAST2 for each gene group and gene region are available at https://doi.org/10.5061/dryad.zcrjdfnrz.
The following dataset was generated:
FortierALPritchardJK2025The primate Major Histocompatibility Complex: Sets of posterior trees from BEAST2 for each gene group and regionDryad Digital Repository10.5061/dryad.zcrjdfnrz
Acknowledgements
We acknowledge support from NIH grants R01 HG011432 and R01 HG008140. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant No. DGE-1656518. We appreciate helpful comments from Jeffrey Spence, the Pritchard lab, and the reviewers of the previous version of this work.
ReferencesAbi-RachedLKuhlHRoosCten HallersBZhuBCarboneLde JongPJMootnickARKnaustFReinhardtRParhamPWalterL2010A small, variable, and irregular killer cell Ig-like receptor locus accompanies the absence of MHC-C and MHC-G in gibbonsJournal of Immunology1841379139110.4049/jimmunol.090301620026738AdamsEJParhamP2001Species‐specific evolution of MHC class I genes in the higher primatesImmunological Reviews183416410.1034/j.1600-065x.2001.1830104.xAdamsEJLuomaAM2013The adaptable major histocompatibility complex (MHC) fold: structure and function of nonclassical and MHC class I-like moleculesAnnual Review of Immunology3152956110.1146/annurev-immunol-032712-09591223298204AllentoftMESikoraMRefoyo-martínezAIrving-peaseEK2022Population genomics of stone age eurasiabioRxiv10.1101/2022.05.04.490594AlpizarAMarcillaMSantiagoC2016Structure of HLA-B*40:02 in complex with the endogenous peptide REFSKEPELWorldwide Protein Data Bank10.2210/pdb5IEK/pdbArdenBKleinJ1982Biochemical comparison of major histocompatibility complex molecules from different subspecies of Mus musculus: evidence for trans-specific evolution of allelesPNAS792342234610.1073/pnas.79.7.23426954545AsaMMoritaDKurohaJMizutaniTMoriNMikamiBSugitaM2022Crystal structures of N-myristoylated lipopeptide-bound HLA class I complexes indicate reorganization of B-pocket architecture upon ligand bindingThe Journal of Biological Chemistry29810210010.1016/j.jbc.2022.10210035667438AutonABrooksLDDurbinRMGarrisonEPKangHMKorbelJOMarchiniJLMcCarthySMcVeanGAAbecasisGR1000 Genomes Project Consortium2015A global reference for human genetic variationNature526687410.1038/nature1539326432245AzevedoLSerranoCAmorimACooperDN2015Trans-species polymorphism in humans and the great apes is generally maintained by balancing selection that modulates the host immune responseHuman Genomics92110.1186/s40246-015-0043-126337052Bade-DodingCTheodossisAGrasSKjer-NielsenLEiz-VesperBSeltsamAHuytonTRossjohnJMcCluskeyJBlasczykR2011The impact of human leukocyte antigen (HLA) micropolymorphism on ligand specificity within the HLA-B*41 allotypic familyHaematologica9611011810.3324/haematol.2010.030924BarkerDJMaccariGGeorgiouXCooperMAFlicekPRobinsonJMarshSGE2023The IPD-IMGT/HLA DatabaseNucleic Acids Research51D1053D106010.1093/nar/gkac101136350643BEA2024Summarizing posterior treesv.2.7.8Centre for Computational Evolutionhttps://www.beast2.org/summarizing-posterior-trees/BergeronLABesenbacherSZhengJLiPBertelsenMFQuintardBHoffmanJILiZSt LegerJShaoCStillerJGilbertMTPSchierupMHZhangG2023Evolution of the germline mutation rate across vertebratesNature61528529110.1038/s41586-023-05752-y36859541BermanHMWestbrookJFengZGillilandGBhatTNWeissigHShindyalovINBournePE2000The protein data bankNucleic Acids Research2823524210.1093/nar/28.1.23510592235BhatiaGPattersonNPasaniucBZaitlenNGenoveseGPollackSMallickSMyersSTandonASpencerCPalmerCDAdeyemoAAAkylbekovaELCupplesLADiversJFornageMKaoWHLLangeLLiMMusaniSMychaleckyjJCOgunniyiAPapanicolaouGRotimiCNRotterJIRuczinskiISalakoBSiscovickDSTayoBOYangQMcCarrollSSabetiPLettreGDe JagerPHirschhornJZhuXCooperRReichDWilsonJGPriceAL2011Genome-wide comparison of African-ancestry populations from CARe and other cohorts reveals signals of natural selectionAmerican Journal of Human Genetics8936838110.1016/j.ajhg.2011.07.02521907010BouckaertRHeledJKühnertDVaughanTWuC-HXieDSuchardMARambautADrummondAJ2014BEAST 2: a software platform for Bayesian evolutionary analysisPLOS Computational Biology10e100353710.1371/journal.pcbi.100353724722319BouckaertRVaughanTGBarido-SottaniJDuchêneSFourmentMGavryushkinaAHeledJJonesGKühnertDDe MaioNMatschinerMMendesFKMüllerNFOgilvieHAdu PlessisLPopingaARambautARasmussenDSiveroniISuchardMAWuC-HXieDZhangCStadlerTDrummondAJ2019BEAST 2.5: An advanced software platform for Bayesian evolutionary analysisPLOS Computational Biology15e100665010.1371/journal.pcbi.100665030958812BoysonJEShufflebothamCCadavidLFUrvaterJAKnappLAHughesALWatkinsDI1996The MHC class I genes of the rhesus monkey: Different evolutionary histories of MHC class I and II genes in primatesThe Journal of Immunology1564656466510.4049/jimmunol.156.12.4656BrändleUOnoHVincekVKleinDGolubicMGrahovacBKleinJ1992Trans-species evolution of Mhc-DRB haplotype polymorphism in primates: organization of DRB genes in the chimpanzeeImmunogenetics36394810.1007/BF002092911587553BrandtDYCCésarJGoudetJMeyerD2018The effect of balancing selection on population differentiation: a study with HLA GenesG3: Genes, Genomes, Genetics82805281510.1534/g3.118.200367BruijnesteijnJ2023HLA/MHC and KIR characterization in humans and non-human primates using Oxford nanopore technologies and pacific biosciences sequencing platformsHLA10120522110.1111/tan.1495736583332BunielloAMacArthurJALCerezoMHarrisLWHayhurstJMalangoneCMcMahonAMoralesJMountjoyESollisESuvegesDVrousgouOWhetzelPLAmodeRGuillenJARiatHSTrevanionSJHallPJunkinsHFlicekPBurdettTHindorffLACunninghamFParkinsonH2019The NHGRI-EBI GWAS Catalog of published genome-wide association studies, targeted arrays and summary statistics 2019Nucleic Acids Research47D1005D101210.1093/nar/gky112030445434Butler-LaporteGFarjounJNakanishiTLuTAbnerEChenYHultströmMMetspaluAMilaniLMägiRNelisMHudjashovGYoshijiSIlboudoYLiangKYHSuC-YWilletJDSEskoTZhouSForgettaVTaliunDRichardsJBEstonian Biobank Research Team2023HLA allele-calling using multi-ancestry whole-exome sequencing from the UK Biobank identifies 129 novel associations in 11 autoimmune diseasesCommunications Biology611710.1038/s42003-023-05496-5CaglianiRFumagalliMBiasinMPiacentiniLRivaSPozzoliUBonagliaMCBresolinNClericiMSironiM2010Long-term balancing selection maintains trans-specific polymorphisms in the human TRIM5 geneHuman Genetics12857758810.1007/s00439-010-0884-620811909CaglianiRGueriniFRFumagalliMRivaSAgliardiCGalimbertiDPozzoliUGorisADuboisBFenoglioCForniDSannaSZaraIPitzalisMZoledziewskaMCuccaFMariniFComiGPScarpiniEBresolinNClericiMSironiM2012A trans-specific polymorphism in ZC3HAV1 is maintained by long-standing balancing selection and may confer susceptibility to multiple sclerosisMolecular Biology and Evolution291599161310.1093/molbev/mss00222319148ChengYGrueberCHoggCJBelovK2022Improved high-throughput MHC typing for non-model species using long-read sequencingMolecular Ecology Resources2286287610.1111/1755-0998.1351134551192ChooJALLiuJTohXGrotenbregGMRenEC2014The immunodominant influenza A virus M158-66 cytotoxic T lymphocyte epitope exhibits degenerate class I major histocompatibility complex restriction in humansJournal of Virology88106131062310.1128/JVI.00855-1424990997ChuFLouZChenYWLiuYGaoBZongLKhanAHBellJIRaoZGaoGF2007First glimpse of the peptide presentation by rhesus macaque MHC class I: crystal structures of Mamu-A*01 complexed with two immunogenic SIV epitopes and insights into CTL escapeJournal of Immunology17894495210.4049/jimmunol.178.2.94417202356CiacchiLvan de GardeMDBLadellKFarencCPoelenMCMMinersKLLlerenaCReidHHPetersenJPriceDARossjohnJvan ElsC2023CD4+ T cell-mediated recognition of a conserved cholesterol-dependent cytolysin epitope generates broad antibacterial immunityImmunity561082109710.1016/j.immuni.2023.03.02037100059ClementsCSKjer-NielsenLKostenkoLHoareHLDunstoneMAMosesEFreedKBrooksAGRossjohnJMcCluskeyJ2005Crystal structure of HLA-G: a nonclassical MHC class I molecule expressed at the fetal-maternal interfacePNAS1023360336510.1073/pnas.040967610215718280CongP-KBaiW-YLiJ-CYangM-YKhederzadehSGaiS-RLiNLiuY-HYuS-HZhaoW-WLiuJ-QSunYZhuX-WZhaoP-PXiaJ-WGuanP-LQianYTaoJ-GXuLTianGWangP-YXieS-YQiuM-CLiuK-QTangB-SZhengH-F2022Genomic analyses of 10,376 individuals in the Westlake BioBank for Chinese (WBBC) pilot projectNature Communications13293910.1038/s41467-022-30526-x35618720DaiSCrawfordFMarrackPKapplerJW2008The structure of HLA-DR52c: comparison to other HLA-DRB3 allelesPNAS105118931189710.1073/pnas.080581010518697946DaiSMurphyGACrawfordFMackDGFaltaMTMarrackPKapplerJWFontenotAP2010Crystal structure of HLA-DP2 and implications for chronic beryllium diseasePNAS1077425743010.1073/pnas.100177210720356827DanecekPAutonAAbecasisGAlbersCABanksEDePristoMAHandsakerRELunterGMarthGTSherrySTMcVeanGDurbinR1000 Genomes Project Analysis Group2011The variant call format and VCFtoolsBioinformatics272156215810.1093/bioinformatics/btr33021653522DarlayRAyersKLMellsGFHallLSLiuJZAlmarriMAAlexanderGJJonesDESandfordRNAndersonCACordellHJ2018Amino acid residues in five separate HLA genes can explain most of the known associations between the MHC and primary biliary cholangitisPLOS Genetics14e100783310.1371/journal.pgen.100783330507971de GrootNGOttingNRobinsonJBlancherALafontBAPMarshSGEO’ConnorDHShiinaTWalterLWatkinsDIBontropRE2012Nomenclature report on the major histocompatibility complex genes and alleles of Great Ape, Old and New World monkey speciesImmunogenetics6461563110.1007/s00251-012-0617-122526602de GrootNGOttingNMaccariGRobinsonJHammondJABlancherALafontBAPGuethleinLAWroblewskiEEMarshSGEShiinaTWalterLVigilantLParhamPO’ConnorDHBontropRE2020Nomenclature report 2019: major histocompatibility complex genes and alleles of great and small Ape and old and new world monkey speciesImmunogenetics72253610.1007/s00251-019-01132-x31624862DijkstraJMYamaguchiT2019Ancient features of the MHC class II presentation pathway, and a model for the possible origin of MHC moleculesImmunogenetics7123324910.1007/s00251-018-1090-2DiltheyAT2021State-of-the-art genome inference in the human MHCThe International Journal of Biochemistry & Cell Biology13110588210.1016/j.biocel.2020.10588233189874DouillardVCastelliECMackSJHollenbachJAGourraudP-AVinceNLimouS2021Approaching genetics through the MHC Lens: Tools and methods for HLA researchFrontiers in Genetics1277491610.3389/fgene.2021.77491634925459DrummondAJRambautA2007BEAST: Bayesian evolutionary analysis by sampling treesBMC Evolutionary Biology721410.1186/1471-2148-7-21417996036DuVYBansalACarlsonJSalazar-GonzalezJFSalazarMGLadellKGrasSJosephsTMHeathSLPriceDARossjohnJHunterEGoepfertPA2016HIV-1-Specific CD8 T cells exhibit limited cross-reactivity during acute infectionJournal of Immunology1963276328610.4049/jimmunol.150241126983786DulbergerCLMcMurtreyCPHölzemerANeuKELiuVSteinbachAMGarcia-BeltranWFSulakMJabriBLynchVJAltfeldMHildebrandWHAdamsEJ2017Human leukocyte antigen F presents peptides and regulates immunity through interactions with NK cell receptorsImmunity461018102910.1016/j.immuni.2017.06.00228636952EdgarRC2004MUSCLE: multiple sequence alignment with high accuracy and high throughputNucleic Acids Research321792179710.1093/nar/gkh34015034147FergusonWDvoraSFikesRWStoneACBoissinotS2012Long-term balancing selection at the antiviral gene OAS1 in Central African chimpanzeesMolecular Biology and Evolution291093110310.1093/molbev/msr24722104212Ferreiro-IglesiasALesseurCMcKayJHungRJHanYZongXChristianiDJohanssonMXiaoXLiYQianDCJiXLiuGCaporasoNSceloGZaridzeDMukeriyaAKonticMOgnjanovicSLissowskaJSzołkowskaMSwiatkowskaBJanoutVHolcatovaIBolcaCSavicMOgnjanovicMBojesenSEWuXAlbanesDAldrichMCTardonAFernandez-SomoanoAFernandez-TardonGLe MarchandLRennertGChenCDohertyJGoodmanGBickeböllerHWichmannH-ERischARosenbergerAShenHDaiJFieldJKDaviesMWollPTeareMDKiemeneyLAvan der HeijdenEHFMYuanJ-MHongY-CHaugenAZienolddinySLamSTsaoM-SJohanssonMGrankvistKSchabathMBAndrewADuellEMelanderOBrunnströmHLazarusPArnoldSSloneSByunJKamalAZhuDLandiMTAmosCIBrennanP2018Fine mapping of MHC region in lung cancer highlights independent susceptibility loci by ethnicityNature Communications9392710.1038/s41467-018-05890-230254314FieldYBoyleEATelisNGaoZGaultonKJGolanDYengoLRocheleauGFroguelPMcCarthyMIPritchardJK2016Detection of human adaptation during the past 2000 yearsScience35476076410.1126/science.aag077627738015FigueroaFGüntherEKleinJ1988MHC polymorphism pre-dating speciationNature33526526710.1038/335265a03137477FintonKAKRupertPBFriendDJDincaALovelaceESBuergerMRusnacDVFoote-McNabbUChourWHeathJRCampbellJSPierceRHStrongRK2023Effects of HLA single chain trimer design on peptide presentation and stabilityFrontiers in Immunology14117046210.3389/fimmu.2023.117046237207206Flores-SolisDMcShanASgourakisN2019HLA-a*01:01 complex with NRAS Q61K peptide by NMRWorldwide Protein Data Bank10.2210/pdb6MPP/pdbFoleyNMMasonVCHarrisAJBredemeyerKRDamasJLewinHAEizirikEGatesyJKarlssonEKLindblad-TohKSpringerMSMurphyWJZoonomia Consortium‡2023A genomic timescale for placental mammal evolutionScience380eabl818910.1126/science.abl818937104581FortierALPritchardJK2025The primate major histocompatibility complex: an illustrative example of gene family evolutioneLife14RP10354510.7554/eLife.103545FujiiSKuboKITakayamaS2016Non-self- and self-recognition models in plant self-incompatibilityNature Plants21613010.1038/nplants.2016.13027595657Fukami-KobayashiKShiinaTAnzaiTSanoKYamazakiMInokoHTatenoY2005Genomic evolution of MHC class I region in primatesPNAS1029230923410.1073/pnas.0500770102FuselliSBaptistaRPPanzieraAMagiAGuglielmiSToninRBenazzoABauzerLGMazzoniCJBertorelleG2018A new hybrid approach for MHC genotyping: high-throughput NGS and long read MinION nanopore sequencing, with application to the non-model vertebrate Alpine chamois (Rupicapra rupicapra)Heredity12129330310.1038/s41437-018-0070-529572469GasteigerEGattikerAHooglandCIvanyiIAppelRDBairochA2003ExPASy: The proteomics server for in-depth protein knowledge and analysisNucleic Acids Research313784378810.1093/nar/gkg56312824418GelukAElferinkDGSlierendregtBLvan MeijgaardenKEde VriesRROttenhoffTHBontropRE1993Evolutionary conservation of major histocompatibility complex-DR/peptide/T cell interactions in primatesThe Journal of Experimental Medicine17797998710.1084/jem.177.4.9798459225Genome Reference Consortium2022Human Genome Region MHChttps://www.ncbi.nlm.nih.gov/grc/human/regions/MHCAugust 12, 2025GerstnerCDubnovitskyASandinCKozhukhGUchtenhagenHJamesEARönnelidJYtterbergAJPieperJReedETandreKRieckMZubarevRARönnblomLSandalovaTBucknerJHAchourAMalmströmV2016Functional and structural characterization of a novel HLA-DRB1*04:01-Restricted α-Enolase T cell epitope in rheumatoid arthritisFrontiers in Immunology749410.3389/fimmu.2016.0049427895642GleimerMWahlARHickmanHDAbi-RachedLNormanPJGuethleinLAHammondJADraghiMAdamsEJJuoSJaliliRGharizadehBRonaghiMGarciaKCHildebrandWHParhamP2011Although divergent in residues of the peptide binding site, conserved chimpanzee Patr-AL and polymorphic human HLA-A*02 have overlapping peptide-binding repertoiresJournal of Immunology1861575158810.4049/jimmunol.100299021209280GrantEJJosephsTMLohLClemensEBSantSBharadwajMChenWRossjohnJGrasSKedzierskaK2018Broad CD8+ T cell cross-recognition of distinct influenza A strains in humansNature Communications9542710.1038/s41467-018-07815-530575715GrantEJNguyenATLobosCASzetoCChatzileontiadouDSMGrasS2020The unconventional role of HLA-E: The road less traveledMolecular Immunology12010111210.1016/j.molimm.2020.02.01132113130GrimholtUTsukamotoKAzumaTLeongJKoopBFDijkstraJM2015A comprehensive analysis of teleost MHC class I sequencesBMC Evolutionary Biology153210.1186/s12862-015-0309-125888517GuceAIMortimerSEYoonTPainterCAJiangWMellinsEDSternLJ2013HLA-DO acts as a substrate mimic to inhibit HLA-DM by a competitive mechanismNature Structural & Molecular Biology20909810.1038/nsmb.2460GuethleinLANormanPJHiltonHGParhamP2015Co-evolution of MHC class I and variable NK cell receptors in placental mammalsImmunological Reviews26725928210.1111/imr.1232626284483GyllenstenUBLashkariDErlichHA1990Allelic diversification at the class II DQB locus of the mammalian major histocompatibility complexPNAS871835183910.1073/pnas.87.5.18352308943HansJBBerglRAVigilantL2017Gorilla MHC class I gene and sequence variation in a comparative contextImmunogenetics6930332310.1007/s00251-017-0974-x28332079HeijmansCMCde GrootNGBontropRE2020Comparative genetics of the major histocompatibility complex in humans and nonhuman primatesInternational Journal of Immunogenetics4724326010.1111/iji.1249032358905HendersonKNTye-DinJAReidHHChenZBorgNABeissbarthTTathamAManneringSIPurcellAWDudekNLvan HeelDAMcCluskeyJRossjohnJAndersonRP2007A Structural and immunological basis for the role of human leukocyte antigen DQ8 in celiac diseaseImmunity27233410.1016/j.immuni.2007.05.015HilligRCHülsmeyerMSaengerWWelfleKMisselwitzRWelfleHKozerskiCVolzAUchanska-ZieglerBZieglerA2004Thermodynamic and structural analysis of peptide- and allele-dependent properties of two HLA-B27 subtypes exhibiting differential disease associationThe Journal of Biological Chemistry27965266310.1074/jbc.M30745720014555655HinksABowesJCobbJAinsworthHCMarionMCComeauMESudmanMHanBBeckerMLBohnsackJFde BakkerPIWHaasJPHazenMLovellDJNigrovicPANordalEPunnaroMRosenbergAMRyggMSmithSLWiseCAVidemVWedderburnLRYarwoodAYeungRSMPrahaladSLangefeldCDRaychaudhuriSThompsonSDThomsonWJuvenile Arthritis Consortium for Immunochip2017Fine-mapping the MHC locus in juvenile idiopathic arthritis (JIA) reveals genetic heterogeneity corresponding to distinct adult inflammatory arthritic diseasesAnnals of the Rheumatic Diseases7676577210.1136/annrheumdis-2016-21002527998952HirataJHosomichiKSakaueSKanaiMNakaokaHIshigakiKSuzukiKAkiyamaMKishikawaTOgawaKMasudaTYamamotoKHirataMMatsudaKMomozawaYInoueIKuboMKamataniYOkadaY2019Genetic and phenotypic landscape of the major histocompatibilty complex region in the Japanese populationNature Genetics5147048010.1038/s41588-018-0336-030692682HoareHLSullivanLCPietraGClementsCSLeeEJElyLKBeddoeTFalcoMKjer-NielsenLReidHHMcCluskeyJMorettaLRossjohnJBrooksAG2006Structural basis for a major histocompatibility complex class Ib-restricted T cell responseNature Immunology725626410.1038/ni131216474394HuanXZhuoNLeeHYRenEC2023Allopurinol non-covalently facilitates binding of unconventional peptides to HLA-B*58:01Scientific Reports13937310.1038/s41598-023-36293-z37296297HughesALNeiM1988Pattern of nucleotide substitution at major histocompatibility complex class I loci reveals overdominant selectionNature33516717010.1038/335167a03412472HughesALNeiM1989Nucleotide substitution at major histocompatibility complex class II loci: evidence for overdominant selectionPNAS8695896210.1073/pnas.86.3.9582492668HurleyCK2021Naming HLA diversity: A review of HLA nomenclatureHuman Immunology8245746510.1016/j.humimm.2020.03.005IgicBBohsLKohnJR2006Ancient polymorphism reveals unidirectional breeding system shiftsPNAS1031359136310.1073/pnas.050628310316428289IllingPTPymmPCroftNPHiltonHGJojicVHanASMendozaJLMifsudNADudekNLMcCluskeyJParhamPRossjohnJVivianJPPurcellAW2018HLA-B57 micropolymorphism defines the sequence and conformational breadth of the immunopeptidomeNature Communications9469310.1038/s41467-018-07109-w30410026IoergerTRClarkAGKaoTH1990Polymorphism at the self-incompatibility locus in Solanaceae predates speciationPNAS879732973510.1073/pnas.87.24.97322263623JainMKorenSMigaKHQuickJRandACSasaniTATysonJRBeggsADDiltheyATFiddesITMallaSMarriottHNietoTO’GradyJOlsenHEPedersenBSRhieARichardsonHQuinlanARSnutchTPTeeLPatenBPhillippyAMSimpsonJTLomanNJLooseM2018Nanopore sequencing and assembly of a human genome with ultra-long readsNature Biotechnology3633834510.1038/nbt.406029431738JeffreysH1998The Theory of ProbabilityOxford University PressJiangWBirtleyJRHungS-CWangWChiouS-HMacaubasCKornumBTianLHuangHAdlerLWeaverGLuLIlstad-MinnihanASomasundaramSAyyangarSDavisMMSternLJMellinsED2019In vivo clonal expansion and phenotypes of hypocretin-specific CD4+ T cells in narcolepsy patients and controlsNature Communications10524710.1038/s41467-019-13234-x31748512JiangJTaylorDKKimEJBoydLFAhmadJMageMGTruongHVWoodwardCHSgourakisNGCresswellPMarguliesDHNatarajanK2022aStructural mechanism of tapasin-mediated MHC-I peptide loading in antigen presentationNature Communications1311310.1038/s41467-022-33153-8JiangHWangC-WWangZDaiYZhuYLeeY-SCaoYChungW-HOuyangSWangH2022bFunctional and structural characteristics of HLA-B*13:01-mediated specific T cells reaction in dapsone-induced drug hypersensitivityJournal of Biomedical Science2912110.1186/s12929-022-00845-8KarlJAPrallTMBussanHEVargheseJMPalAWisemanRWO’ConnorDH2023Complete sequencing of a cynomolgus macaque major histocompatibility complex haplotypeGenome Research3344846210.1101/gr.277429.12236854669KassardjianASunESookhooJMuthuramanKBoliganKFKucharskaIRujasEJethaABranchDRBabiukSBarberBJulienJP2023Modular adjuvant-free pan-HLA-DR-immunotargeting subunit vaccine against SARS-CoV-2 elicits broad sarbecovirus-neutralizing antibody responsesCell Reports4211239110.1016/j.celrep.2023.11239137053069KaufmanJ2022The new W family reconstructs the evolution of MHC genesPNAS11911912110.1073/pnas.2122079119KaurGGrasSMobbsJIVivianJPCortesABarberTKuttikkatteSBJensenLTAttfieldKEDendrouCACarringtonMMcVeanGPurcellAWRossjohnJFuggerL2017Structural and regulatory diversity shape HLA-C protein expression levelsNature Communications81592410.1038/ncomms1592428649982KennedyAEOzbekUDorakMT2017What has GWAS done for HLA and disease associations?International Journal of Immunogenetics4419521110.1111/iji.1233228877428KiryuIDijkstraJMSarderRIFujiwaraAYoshiuraYOtotakeM2005New MHC class Ia domain lineages in rainbow trout (Oncorhynchus mykiss) which are shared with other fish speciesFish & Shellfish Immunology1824325410.1016/j.fsi.2004.07.007KleinJ1980Generation of diversity at MHC loci: implications for t-cell receptor repertoiresFougereauMDaussetJImmunologyAcademic Press2392536968289KleinJ1987Origin of major histocompatibility complex polymorphism: the trans-species hypothesisHuman Immunology1915516210.1016/0198-8859(87)90066-83305436KlobuchSLimJJvan BalenPKesterMGDde KlerkWde RuAHPothastCRJedemaIDrijfhoutJWRossjohnJReidHHvan VeelenPAFalkenburgJHFHeemskerkMHM2022Human T cells recognize HLA-DP–bound peptides in two orientationsPNAS119e221433111910.1073/pnas.2214331119KrienerKO’hUiginCTichyHKleinJ2000Convergent evolution of major histocompatibility complex molecules in humans and New World monkeysImmunogenetics5116917810.1007/s002510050028KrienerKO’hUiginCKleinJ2001Independent origin of functional MHC class II genes in humans and new world monkeysHuman Immunology6211410.1016/S0198-8859(00)00233-0KrishnaCChiouJSakaueSKangJBChristensenSMLeeIAksitMAKimHIvon SchackDRaychaudhuriSZiemekDHuX2024The influence of HLA genetic variation on plasma protein expressionNature Communications15646910.1038/s41467-024-50583-839085222KudernaLFKGaoHJaniakMCKuhlwilmMOrkinJDBataillonTManuSValenzuelaABergmanJRousselleMSilvaFEAguedaLBlancJGutMde VriesDGoodheadIHarrisRARaveendranMJensenAChumaISHorvathJEHvilsomCJuanDFrandsenPSchraiberJGde MeloFRBertuolFByrneHSampaioIFariasIValsecchiJMessiasMda SilvaMNFTrivediMRossiRHrbekTAndriaholinirinaNRabarivolaCJZaramodyAJollyCJPhillips-ConroyJWilkersonGAbeeCSimmonsJHFernandez-DuqueEKanthaswamySShiferawFWuDZhouLShaoYZhangGKeyyuJDKnaufSLeMDLizanoEMerkerSNavarroANadlerTKhorCCLeeJTanPLimWKKitchenerACZinnerDGutIMelinADGuschanskiKSchierupMHBeckRMDUmapathyGRoosCBoubliJPRogersJFarhKK-HMarques BonetT2023A global catalog of whole-genome diversity from 233 primate speciesScience38090691310.1126/science.abn782937262161KumarPVahedi-FaridiASaengerWMerinoELópez de CastroJAUchanska-ZieglerBZieglerA2009Structural basis for T cell alloreactivity among three HLA-B14 and HLA-B27 antigensThe Journal of Biological Chemistry284297842979710.1074/jbc.M109.03849719617632KunduSFaulkesCG2007A tangled history: patterns of major histocompatibility complex evolution in the African mole-rats (Family: Bathyergidae)Biological Journal of the Linnean Society9149350310.1111/j.1095-8312.2007.00814.xKupfermannHMayerWEO’hUiginCKleinDKleinJ1992Shared polymorphism between gorilla and human major histocompatibility complex DRB lociHuman Immunology3426727810.1016/0198-8859(92)90026-j1464555KusanoSKukimoto-NiinoMSattaYOhsawaNUchikubo-KamoTWakiyamaMIkedaMTeradaTYamamotoKNishimuraYShirouzuMSasazukiTYokoyamaS2014Structural basis for the specific recognition of the major antigenic peptide from the Japanese cedar pollen allergen Cry j 1 by HLA-DP5Journal of Molecular Biology4263016302710.1016/j.jmb.2014.06.02025020231LangHLEJacobsenHIkemizuSAnderssonCHarlosKMadsenLHjorthPSondergaardLSvejgaardAWucherpfennigKStuartDIBellJIJonesEYFuggerL2002A functional and structural basis for TCR cross-reactivity in multiple sclerosisNature Immunology394094310.1038/ni83512244309LawlorDAWardFEEnnisPDJacksonAPParhamP1988HLA-A and B polymorphisms predate the divergence of humans and chimpanzeesNature33526827110.1038/335268a03412487LeeSJKleinJHaagensonMBaxter-LoweLAConferDLEapenMFernandez-VinaMFlomenbergNHorowitzMHurleyCKNoreenHOudshoornMPetersdorfESetterholmMSpellmanSWeisdorfDWilliamsTMAnasettiC2007High-resolution donor-recipient HLA matching contributes to the success of unrelated donor marrow transplantationBlood1104576458310.1182/blood-2007-06-09738617785583LefflerEMGaoZPfeiferSSégurelLAutonAVennOBowdenRBontropRWallJDSellaGDonnellyPMcVeanGPrzeworskiM2013Multiple instances of ancient balancing selection shared between humans and chimpanzeesScience3391578158210.1126/science.123407023413192LiYLiHMartinRMariuzzaRA2000Structural basis for the binding of an immunodominant peptide from myelin basic protein in different registers by two HLA-DR2 proteinsJournal of Molecular Biology30417718810.1006/jmbi.2000.419811080454LiLZhouXChenX2011Characterization and evolution of MHC class II B genes in ardeid birdsJournal of Molecular Evolution7247448310.1007/s00239-011-9446-3LiXSinghNKCollinsDRNgRZhangALamothe-MolinaPAShahinianPXuSTanKPiechocka-TrochaAUrbachJMWeberJKGaihaGDTakou MbahOCHuynhTCheeverSChenJBirnbaumMZhouRWalkerBDWangJ2023Molecular basis of differential HLA class I-restricted T cell recognition of a highly networked HIV peptideNature Communications143857310.1038/s41467-023-38573-8Lim Kam SianTCCIndumathySHalimHGreuleACryleMJBownessPRossjohnJGrasSPurcellAWSchittenhelmRB2019Allelic association with ankylosing spondylitis fails to correlate with human leukocyte antigen B27 homodimer formationThe Journal of Biological Chemistry294201852019510.1074/jbc.RA119.01025731740583LiuJChenKYRenEC2011Structural insights into the binding of hepatitis B virus core peptide to HLA-A2 alleles: towards designing better vaccinesEuropean Journal of Immunology412097210610.1002/eji.20104137021538979LiuYCMilesJJNellerMAGostickEPriceDAPurcellAWMcCluskeyJBurrowsSRRossjohnJGrasS2013Highly divergent T-cell receptor binding modes underlie specific recognition of a bulged viral peptide bound to a human leukocyte antigen class I moleculeThe Journal of Biological Chemistry288154421545410.1074/jbc.M112.44718523569211LiuYCChenZNellerMAMilesJJPurcellAWMcCluskeyJBurrowsSRRossjohnJGrasS2014A molecular basis for the interplay between T cells, viral mutants, and human leukocyte antigen micropolymorphismThe Journal of Biological Chemistry289166881669810.1074/jbc.M114.56350224759101LiuC2021A long road/read to rapid high-resolution HLA typing: The nanopore perspectiveHuman Immunology8248849510.1016/j.humimm.2020.04.00932386782LoiselDARockmanMVWrayGAAltmannJAlbertsSC2006Ancient polymorphism and functional variation in the primate MHC-DQA1 5’ cis-regulatory regionPNAS103163311633610.1073/pnas.060766210317053068LollBRueckertCZieglerB2020Crystal structure of HLA-B2709 complexed with the nona-peptide maWorldwide Protein Data Bank10.2210/pdb6Y27/pdbLuDChenYJiangM2023Crystal structure of A TCR in complex with HLA-A*11:01 bound to KRAS peptide (VVGAVGVGK)Worldwide Protein Data Bank10.2210/pdb8I5C/pdbLugoJSCadavidLF2015Patterns of MHC-G-like and MHC-B diversification in new world monkeysPLOS ONE10e013134310.1371/journal.pone.013134326121030LuoYKanaiMChoiWLiXSakaueSYamamotoKOgawaKGutierrez-ArcelusMGregersenPKStuartPEElderJTForerLSchönherrSFuchsbergerCSmithAVFellayJCarringtonMHaasDWGuoXPalmerNDChenYDIRotterJITaylorKDRichSSCorreaAWilsonJGKathiresanSChoMHMetspaluAEskoTOkadaYHanBMcLarenPJRaychaudhuriSNHLBI Trans-Omics for Precision Medicine (TOPMed) Consortium2021Author Correction: A high-resolution HLA reference panel capturing global population diversity enables multi-ancestry fine-mapping in HIV host responseNature Genetics53172210.1038/s41588-021-00979-934728834MaccariGRobinsonJBallingallKGuethleinLAGrimholtUKaufmanJHoC-Sde GrootNGFlicekPBontropREHammondJAMarshSGE2017IPD-MHC 2.0: an improved inter-species database for the study of the major histocompatibility complexNucleic Acids Research45D860D86410.1093/nar/gkw105027899604MaccariGRobinsonJHammondJAMarshSGE2020The IPD Project: a centralised resource for the study of polymorphism in genes of the immune systemImmunogenetics72495510.1007/s00251-019-01133-wMacdonaldWAChenZGrasSArchboldJKTynanFEClementsCSBharadwajMKjer-NielsenLSaundersPMWilceMCJCrawfordFStadinskyBJacksonDBrooksAGPurcellAWKapplerJWBurrowsSRRossjohnJMcCluskeyJ2009T cell allorecognition via molecular mimicryImmunity3189790810.1016/j.immuni.2009.09.025MaibachVHansJBHvilsomCMarques-BonetTVigilantL2017MHC class I diversity in chimpanzees and bonobosImmunogenetics6966167610.1007/s00251-017-0990-xMallickSLiHLipsonMMathiesonIGymrekMRacimoFZhaoMChennagiriNNordenfeltSTandonASkoglundPLazaridisISankararamanSFuQRohlandNRenaudGErlichYWillemsTGalloCSpenceJPSongYSPolettiGBallouxFvan DriemGde KnijffPRomeroIGJhaARBeharDMBraviCMCapelliCHervigTMoreno-EstradaAPosukhOLBalanovskaEBalanovskyOKarachanak-YankovaSSahakyanHTonchevaDYepiskoposyanLTyler-SmithCXueYAbdullahMSRuiz-LinaresABeallCMDi RienzoAJeongCStarikovskayaEBMetspaluEParikJVillemsRHennBMHodoglugilUMahleyRSajantilaAStamatoyannopoulosGWeeJTSKhusainovaRKhusnutdinovaELitvinovSAyodoGComasDHammerMFKivisildTKlitzWWinklerCALabudaDBamshadMJordeLBTishkoffSAWatkinsWSMetspaluMDryomovSSukernikRSinghLThangarajKPääboSKelsoJPattersonNReichD2016The simons genome diversity project: 300 genomes from 142 diverse populationsNature53820120610.1038/nature1896427654912MarshSGEAlbertEDBodmerWFBontropREDupontBErlichHAFernández-ViñaMGeraghtyDEHoldsworthRHurleyCKLauMLeeKWMachBMaiersMMayrWRMüllerCRParhamPPetersdorfEWSasazukiTStromingerJLSvejgaardATerasakiPITiercyJMTrowsdaleJ2010Nomenclature for factors of the HLA system, 2010Tissue Antigens7529145510.1111/j.1399-0039.2010.01466.x20356336MathiesonILazaridisIRohlandNMallickSPattersonNRoodenbergSAHarneyEStewardsonKFernandesDNovakMSirakKGambaCJonesERLlamasBDryomovSPickrellJArsuagaJLde CastroJMBCarbonellEGerritsenFKhokhlovAKuznetsovPLozanoMMellerHMochalovOMoiseyevVGuerraMARRoodenbergJVergèsJMKrauseJCooperAAltKWBrownDAnthonyDLalueza-FoxCHaakWPinhasiRReichD2015Genome-wide patterns of selection in 230 ancient EurasiansNature52849950310.1038/nature16152MayerWEJonkerMKleinDIvanyiPvan SeventerGKleinJ1988Nucleotide sequences of chimpanzee MHC class I alleles: evidence for trans-species mode of evolutionThe EMBO Journal72765277410.1002/j.1460-2075.1988.tb03131.xMayerWEO’hUiginCZaleska-RutczynskaZKleinJ1992Trans-species origin of Mhc-DRB polymorphism in the chimpanzeeImmunogenetics37122310.1007/BF002235401428008McConnellTJTalbotWSMcIndoeRAWakelandEK1988The origin of MHC class II gene polymorphism within the genus MusNature33265165410.1038/332651a02895893McKenzieLMPecon-SlatteryJCarringtonMO’BrienSJ1999Taxonomic hierarchy of HLA class I allele sequencesGenes & Immunity112012910.1038/sj.gene.636364811197301McLarenPJRipkeSPelakKWeintrobACPatsopoulosNAJiaXErlichRLLennonNJKadieCMHeckermanDGuptaNHaasDWDeeksSGPereyraFWalkerBDde BakkerPIWInternational HIV Controllers Study2012Fine-mapping classical HLA variation associated with durable host control of HIV-1 infection in African AmericansHuman Molecular Genetics214334434710.1093/hmg/dds22622718199MillerJDWeberDAIbegbuCPohlJAltmanJDJensenPE2003Analysis of HLA-E peptide-binding specificity and contact residues in bound peptide required for recognition by CD94/NKG2Journal of Immunology1711369137510.4049/jimmunol.171.3.136912874227MobbsJIIllingPTDudekNLBrooksAGBakerDGPurcellAWRossjohnJVivianJP2017The molecular basis for peptide repertoire selection in the human leukocyte antigen (HLA) C*06:02 moleculeJournal of Biological Chemistry292172031721510.1074/jbc.M117.806976MolinerosJELoogerLLKimKOkadaYTeraoCSunCZhouX-JRajPKochiYSuzukiAAkizukiSNakaboSBangS-YLeeH-SKangYMSuhC-HChungWTParkY-BChoeJ-YShimS-CLeeS-SZuoXYamamotoKLiQ-ZShenNPorterLLHarleyJBChuaKHZhangHWakelandEKTsaoBPBaeS-CNathSK2019Amino acid signatures of HLA Class-I and II molecules are strongly associated with SLE susceptibility and autoantibody production in Eastern AsiansPLOS Genetics15e100809210.1371/journal.pgen.100809231022184MoradiSStankovicSO’ConnorGMPymmPMacLachlanBJFaoroCRetièreCSullivanLCSaundersPMWidjajaJCox-LivingstoneSRossjohnJBrooksAGVivianJP2021Structural plasticity of KIR2DL2 and KIR2DL3 enables altered docking geometries atop HLA-CNature Communications12217310.1038/s41467-021-22359-x33846289MosyakLZallerDMWileyDC1998The structure of HLA-DM, the peptide exchange catalyst that loads antigen onto class II MHC molecules during antigen presentationImmunity937738310.1016/s1074-7613(00)80620-29768757MotozonoCKuseNSunXRizkallahPJFullerAOkaSColeDKSewellAKTakiguchiM2014Molecular basis of a dominant T cell response to an HIV reverse transcriptase 8-mer epitope presented by the protective allele HLA-B*51:01Journal of Immunology1923428343410.4049/jimmunol.130266724600035MüllerNFBouckaertRR2020Adaptive Metropolis-coupled MCMC for BEAST 2PeerJ8e947310.7717/peerj.947332995072NaitoTSuzukiKHirataJKamataniYMatsudaKTodaTOkadaY2021A deep learning method for HLA imputation and trans-ethnic MHC fine-mapping of type 1 diabetesNature Communications12163910.1038/s41467-021-21975-x33712626NeefjesJJongsmaMLMPaulPBakkeO2011Towards a systems understanding of MHC class I and MHC class II antigen presentationNature Reviews. Immunology1182383610.1038/nri308422076556NeiMHughesAL1991Polymorphism and evolution of the major histocompatibility complex loci in mammalsSelanderRClarkAWhittamTEvolution at the Molecular LevelSinauer Associates, Inc222247NewmanRMHallLConnoleMChenG-LSatoSYusteEDiehlWHunterEKaurAMillerGMJohnsonWE2006Balancing selection and the evolution of functional polymorphism in Old World monkey TRIM5alphaPNAS103191341913910.1073/pnas.060583810317142324NguyenTBJayaramanPBergsengEMadhusudhanMSKimCYSollidLM2017Unraveling the structural basis for the unusually rich association of human leukocyte antigen DQ2.5 with class-II-associated invariant chain peptidesThe Journal of Biological Chemistry2929218922810.1074/jbc.M117.78513928364043NicholsonMJMoradiBSethNPXingXCunyGDSteinRLWucherpfennigKW2006Small molecules that enhance the catalytic efficiency of HLA-DMJournal of Immunology1764208422010.4049/jimmunol.176.7.420816547258NielsenMLundegaardCBlicherTLamberthKHarndahlMJustesenSRøderGPetersBSetteALundOBuusS2007NetMHCpan, a method for quantitative predictions of peptide binding to any HLA-A and -B locus protein of known sequencePLOS ONE2e79610.1371/journal.pone.000079617726526NiuLChengHZhangSTanSZhangYQiJLiuJGaoGF2013Structural basis for the differential classification of HLA-A*6802 and HLA-A*6801 into the A2 and A3 supertypesMolecular Immunology5538139210.1016/j.molimm.2013.03.01523566939NunesKMaiaMHTDos SantosEJMDos SantosSEBGuerreiroJFPetzl-ErlerMLBedoyaGGalloCPolettiGLlopETsunetoLBortoliniMCRothhammerFSingleRRuiz-LinaresARochaJMeyerD2021How natural selection shapes genetic differentiation in the MHC region: A case study with Native AmericansHuman Immunology8252353110.1016/j.humimm.2021.03.00533812704OkadaYMomozawaYSakaueSKanaiMIshigakiKAkiyamaMKishikawaTAraiYSasakiTKosakiKSuematsuMMatsudaKYamamotoKKuboMHiroseNKamataniY2018Deep whole-genome sequencing reveals recent selection signatures linked to evolution and disease risk of JapaneseNature Communications9163110.1038/s41467-018-03274-029691385O’LearyNAWrightMWBristerJRCiufoSHaddadDMcVeighRRajputBRobbertseBSmith-WhiteBAko-AdjeiDAstashynABadretdinABaoYBlinkovaOBroverVChetverninVChoiJCoxEErmolaevaOFarrellCMGoldfarbTGuptaTHaftDHatcherEHlavinaWJoardarVSKodaliVKLiWMaglottDMastersonPMcGarveyKMMurphyMRO’NeillKPujarSRangwalaSHRauschDRiddickLDSchochCShkedaAStorzSSSunHThibaud-NissenFTolstoyITullyREVatsanARWallinCWebbDWuWLandrumMJKimchiATatusovaTDiCuccioMKittsPMurphyTDPruittKD2016Reference sequence (RefSeq) database at NCBI: current status, taxonomic expansion, and functional annotationNucleic Acids Research44D733D4510.1093/nar/gkv118926553804OoiJDPetersenJTanYHHuynhMWillettZJRamarathinamSHEggenhuizenPJLohKLWatsonKAGanPYAlikhanMADudekNLHandelAHudsonBGFuggerLPowerDAHoltSGCoatesPTGregersenJWPurcellAWHoldsworthSRLa GrutaNLReidHHRossjohnJKitchingAR2017Dominant protection from HLA-linked autoimmunity by antigen-specific regulatory T cellsNature54524324710.1038/nature2232928467828OttingNKenterMvan WeerenPJonkerMBontropRE1992Mhc-DQB repertoire variation in hominoid and Old World primate speciesJournal of Immunology14946147010.4049/jimmunol.149.2.4611624794OttingNBontropRE1995Evolution of the major histocompatibility complex DPA1 locus in primatesHuman Immunology4218418710.1016/0198-8859(94)00095-8OttingNde GrootNGNoortMCDoxiadisGGBontropRE2000Allelic diversity of Mhc-DRB alleles in rhesus macaquesTissue Antigens56586810.1034/j.1399-0039.2000.560108.x10958357OttingNde GrootNGDoxiadisGGMBontropRE2002Extensive Mhc-DQB variation in humans and non-human primate speciesImmunogenetics5423023910.1007/s00251-002-0461-912136334PaganiniJAbi-RachedLGouretPPontarottiPChiaroniJDi CristofaroJ2019HLAIb worldwide genetic diversity: New HLA-H alleles and haplotype structure descriptionMolecular Immunology112405010.1016/j.molimm.2019.04.01731078115ParryCSGorskiJSternLJ2007Crystallographic structure of the human leukocyte antigen DRA, DRB3*0101: Models of a directional alloimmune response and autoimmunityJournal of Molecular Biology37143544610.1016/j.jmb.2007.05.025PetersenJKooy-WinkelaarYLohKLTranMvan BergenJKoningFRossjohnJReidHH2016Diverse T cell receptor gene usage in HLA-DQ8-associated celiac disease converges into a consensus binding solutionStructure241643165710.1016/j.str.2016.07.010PetrieEJClementsCSLinJSullivanLCJohnsonDHuytonTHerouxAHoareHLBeddoeTReidHHWilceMCJBrooksAGRossjohnJ2008CD94-NKG2A recognition of human leukocyte antigen (HLA)-E bound to an HLA class I leader sequenceThe Journal of Experimental Medicine20572573510.1084/jem.2007252518332182PiontkivskaH2003Birth-and-death evolution in primate MHC class I genes: divergence time estimatesMolecular Biology and Evolution2060160910.1093/molbev/msg064PosWSethiDKCallMJSchulzeMAndersAKPyrdolJWucherpfennigKW2012Crystal structure of the HLA-DM-HLA-DR1 complex defines mechanisms for rapid peptide selectionCell1511557156810.1016/j.cell.2012.11.02523260142PurcellJLagunas-RoblesGRabelingCBorowiecMLBrelsfordA2021The maintenance of polymorphism in an ancient social supergeneMolecular Ecology306246625810.1111/mec.1619634570409RacleJGuillaumePSchmidtJMichauxJLarabiALauKPerezMASCroceGGenoletRCoukosGZoeteVPojerFBassani-SternbergMHarariAGfellerD2023Machine learning predictions of MHC-II specificities reveal alternative binding mode of class II epitopesImmunity561359137510.1016/j.immuni.2023.03.00937023751RadwanJBabikWKaufmanJLenzTLWinternitzJ2020Advances in the evolutionary understanding of MHC polymorphismTrends in Genetics3629831110.1016/j.tig.2020.01.00832044115RambautADrummondAJXieDBaeleGSuchardMA2018Posterior summarization in bayesian phylogenetics using tracer 1.7Systematic Biology6790190410.1093/sysbio/syy03229718447RaychaudhuriSSandorCStahlEAFreudenbergJLeeHSJiaXAlfredssonLPadyukovLKlareskogLWorthingtonJSiminovitchKABaeSCPlengeRMGregersenPKde BakkerPIW2012Five amino acids in three HLA proteins explain most of the association between MHC and seropositive rheumatoid arthritisNature Genetics4429129610.1038/ng.107622286218RistMJTheodossisACroftNPNellerMAWellandAChenZSullivanLCBurrowsJMMilesJJBrennanRMGrasSKhannaRBrooksAGMcCluskeyJPurcellAWRossjohnJBurrowsSR2013HLA peptide length preferences control CD8+ T cell responsesJournal of Immunology19156157110.4049/jimmunol.130029223749632RobinsonJBarkerDJGeorgiouXCooperMAFlicekPMarshSGE2019IPD-IMGT/HLA DatabaseNucleic Acids Research48D948D95510.1093/nar/gkz950RobinsonJBarkerDJMarshSGE202425 years of the IPD-IMGT/HLA DatabaseHLA103e1554910.1111/tan.1554938936817SachidanandamRWeissmanDSchmidtSCKakolJMSteinLDMarthGSherrySMullikinJCMortimoreBJWilleyDLHuntSEColeCGCoggillPCRiceCMNingZRogersJBentleyDRKwokPYMardisERYehRTSchultzBCookLDavenportRDanteMFultonLHillierLWaterstonRHMcPhersonJDGilmanBSchaffnerSVan EttenWJReichDHigginsJDalyMJBlumenstielBBaldwinJStange-ThomannNZodyMCLintonLLanderESAltshulerDInternational SNP Map Working Group2001A map of human genome sequence variation containing 1.42 million single nucleotide polymorphismsNature40992893310.1038/3505714911237013SakaueSKanaiMTanigawaYKarjalainenJKurkiMKoshibaSNaritaAKonumaTYamamotoKAkiyamaMIshigakiKSuzukiASuzukiKObaraWYamajiKTakahashiKAsaiSTakahashiYSuzukiTShinozakiNYamaguchiHMinamiSMurayamaSYoshimoriKNagayamaSObataDHigashiyamaMMasumotoAKoretsuneYItoKTeraoCYamauchiTKomuroIKadowakiTTamiyaGYamamotoMNakamuraYKuboMMurakamiYYamamotoKKamataniYPalotieARivasMADalyMJMatsudaKOkadaYFinnGen2021A cross-population atlas of genetic associations for 220 human phenotypesNature Genetics531415142410.1038/s41588-021-00931-x34594039SanoEBWallCAHutchinsPRMillerSR2018Ancient balancing selection on heterocyst function in a cosmopolitan cyanobacteriumNature Ecology & Evolution251051910.1038/s41559-017-0435-9SattaYMayerWEKleinJ1996Evolutionary relationship ofHLA-DRB genes inferred from intron sequencesJournal of Molecular Evolution4264865710.1007/BF02338798SaundersPMVivianJPBaschukNBeddoeTWidjajaJO’ConnorGMHitchenCPymmPAndrewsDMGrasSMcVicarDWRossjohnJBrooksAG2015The Interaction of KIR3DL1*001 with HLA Class I molecules is dependent upon molecular microarchitecture within the Bw4 epitopeThe Journal of Immunology19478178910.4049/jimmunol.1402542SawyerSA1999GENECONV: A Computer Package for the Statistical Detection of Gene ConversionScienceOpenScallySWLawSCTingYTvan HeemstJSokoloveJDeutschAJBridie ClemensEMoustakasAKPapadopoulosGKvan der WoudeDSmolikIHitchonCARobinsonDBFerucciEDBernsteinCNMengXAnapartiVHuizingaTKedzierskaKReidHHRaychaudhuriSToesRERossjohnJEl-GabalawyHThomasR2017Molecular basis for increased susceptibility of Indigenous North Americans to seropositive rheumatoid arthritisAnnals of the Rheumatic Diseases761915192310.1136/annrheumdis-2017-21130028801345Schrödinger, LLC2021The pymol molecular graphics systemv.2.4.2PyMOLhttp://www.pymol.org/SchulzeM-SEWucherpfennigKW2012The mechanism of HLA-DM induced peptide exchange in the MHC class II antigen presentation pathwayCurrent Opinion in Immunology2410511110.1016/j.coi.2011.11.004SchutteRLiDOstrovD2020HLA-b*15:01 complexed with a synthetic peptideWorldwide Protein Data Bank10.2210/pdb6UZP/pdbSégurelLThompsonEEFlutreTLovstadJVenkatAMargulisSWMoyseJRossSGambleKSellaGOberCPrzeworskiM2012The ABO blood group is a trans-species polymorphism in primatesPNAS109184931849810.1073/pnas.121060310923091028SharonESibenerLVBattleAFraserHBGarciaKCPritchardJK2016Genetic variation in MHC proteins is associated with T cell receptor expression biasesNature Genetics48995100210.1038/ng.3625ShiroishiMKurokiKRasubalaLTsumotoKKumagaiIKurimotoEKatoKKohdaDMaenakaK2006Structural basis for recognition of the nonclassical MHC molecule HLA-G by the leukocyte Ig-like receptor B2 (LILRB2/LIR2/ILT4/CD85d)PNAS103164121641710.1073/pnas.060522810317056715SimonsNDEickGNRuiz-LopezMJOmejaPAChapmanCAGoldbergTLTingNSternerKN2017Cis-regulatory evolution in a wild primate: Infection-associated genetic variation drives differential expression of MHC-DQA1 in vitroMolecular Ecology264523453510.1111/mec.1422128665019SlierendregtBLvan NoortJTBakasRMOttingNJonkerMBontropRE1992Evolutionary stability of transspecies major histocompatibility complex class II DRB lineages in humans and rhesus monkeysHuman Immunology35293910.1016/0198-8859(92)90092-21478892SlierendregtBLOttingNKenterMBontropRE1995Allelic diversity at the Mhc-DP locus in rhesus macaques (Macaca mulatta)Immunogenetics41293710.1007/BF001884297806271SmithKJPyrdolJGauthierLWileyDCWucherpfennigKW1998Crystal structure of HLA-DR2 (DRA*0101, DRB1*1501) complexed with a peptide from human myelin basic proteinThe Journal of Experimental Medicine1881511152010.1084/jem.188.8.15119782128SmithCJStrauszSSpenceJPOllilaHMPritchardJKFinnGen2024Haplotype Analysis Reveals Pleiotropic Disease Associations in the HLA RegionmedRxiv10.1101/2024.07.29.24311183StuartPETsoiLCNairRPGhoshMKabraMShaiqPARajaGKQamarRThelmaBKPatrickMTPariharASinghSKhandpurSKumarUWittigMDegenhardtFTejasviTVoorheesJJWeidingerSFrankeAAbecasisGRSharmaVKElderJT2022Transethnic analysis of psoriasis susceptibility in South Asians and Europeans enhances fine-mapping in the MHC and genomewideHGG Advances310006910.1016/j.xhgg.2021.10006934927100SullivanLCWalpoleNGFarencCPietraGSumMJWClementsCSLeeEJBeddoeTFalcoMMingariMCMorettaLGrasSRossjohnJBrooksAG2017A conserved energetic footprint underpins recognition of human leukocyte antigen-E by two distinct αβ T cell receptorsThe Journal of Biological Chemistry292211492115810.1074/jbc.M117.80771928972140SunMLiuJQiJTefsenBShiYYanJGaoGF2014Nα-terminal acetylation for T cell recognition: molecular basis of MHC class I-restricted nα-acetylpeptide presentationJournal of Immunology1925509551910.4049/jimmunol.140019924829406TeixeiraJCde FilippoCWeihmannAMeneuJRRacimoFDannemannMNickelBFischerAHalbwaxMAndreCAtenciaRMeyerMParraGPääboSAndrésAM2015Long-term balancing selection in LAD1 maintains a missense trans-species polymorphism in humans, Chimpanzees, and BonobosMolecular Biology and Evolution321186119610.1093/molbev/msv00725605789TezeDHendrickxJCzjzekMRopartzDSanejouandYHTranVTellierCDionM2014Semi-rational approach for converting a GH1 β-glycosidase into a β-transglycosidaseProtein Engineering, Design & Selection27131910.1093/protein/gzt05724287187TianCHromatkaBSKieferAKErikssonNNobleSMTungJYHindsDA2017Genome-wide association and HLA region fine-mapping studies identify susceptibility loci for multiple common infectionsNature Communications859910.1038/s41467-017-00257-528928442TingYTDahal-KoiralaSKimHSKQiaoS-WNeumannRSLundinKEAPetersenJReidHHSollidLMRossjohnJ2020A molecular basis for the T cell response in HLA-DQ2.2 mediated celiac diseasePNAS1173063307310.1073/pnas.191430811731974305TollefsenSHottaKChenXSimonsenBSwaminathanKMathewsIISollidLMKimCY2012Structural and functional studies of trans-encoded HLA-DQ2.3 (DQA1*03:01/DQB1*02:01) protein moleculeThe Journal of Biological Chemistry287136111361910.1074/jbc.M111.32037422362761van de SandtCEClemensEBGrantEJRowntreeLCSantSHalimHCroweJChengACKotsimbosTCRichardsMMillerATongSYCRossjohnJNguyenTHOGrasSChenWKedzierskaK2019Challenging immunodominance of influenza-specific CD8+ T cell responses restricted by the risk-associated HLA-A*68:01 allomorphNature Communications10557910.1038/s41467-019-13346-431811120VaughanTXieWWuJ2018SubstBMAdafa621GitHubhttps://github.com/jessiewu/substBMAViļumaAMikkoSHahnDSkowLAnderssonGBergströmTF2017Genomic structure of the horse major histocompatibility complex class II region resolved using PacBio long-read sequencing technologyScientific Reports74551810.1038/srep4551828361880VivianJRossjohnJ2022HLA-b*27:05 in complex with the pan-HLA-ia monoclonal antibody W6/32Worldwide Protein Data Bank10.2210/pdb7t0l/pdbVizcaínoJAKubiniokPKovalchikKAMaQDuquetteJDMongrainIDeutschEWPetersBSetteASiroisICaronE2020The human immunopeptidome project: a roadmap to predict and treat immune diseasesMolecular & Cellular Proteomics19314910.1074/mcp.R119.001743VollmersSLobermeyerAKörnerC2021The new kid on the block: HLA-C, a key regulator of natural killer cells in viral immunityCells10310810.3390/cells1011310834831331WaageJStandlMCurtinJAJessenLEThorsenJTianCSchoettlerNFloresCAbdellaouiAAhluwaliaTSAlvesACAmaralAFSAntóJMArnoldABarreto-LuisABaurechtHvan BeijsterveldtCEMBleeckerERBonàs-GuarchSBoomsmaDIBrixSBunyavanichSBurchardEGChenZCurjuricICustovicAden DekkerHTDharmageSCDmitrievaJDuijtsLEgeMJGaudermanWJGeorgesMGiegerCGillilandFGranellRGuiHHansenTHeinrichJHendersonJHernandez-PachecoNHoltPImbodenMJaddoeVWVJarvelinM-RJarvisDLJensenKKJónsdóttirIKabeschMKaprioJKumarALeeY-ALevinAMLiXLorenzo-DiazFMelénEMercaderJMMeyersDAMyersRNicolaeDLNohrEAPalviainenTPaternosterLPennellCEPershagenGPino-YanesMProbst-HenschNMRüschendorfFSimpsonAStefanssonKSunyerJSveinbjornssonGThieringEThompsonPJTorrentMTorrentsDTungJYWangCAWeidingerSWeissSWillemsenGWilliamsLKOberCHindsDAFerreiraMABisgaardHStrachanDPBønnelykkeK23andMe Research TeamAAGC collaborators2018Genome-wide association and HLA fine-mapping studies identify risk loci and genetic pathways underlying allergic rhinitisNature Genetics501072108010.1038/s41588-018-0157-130013184WakelandEKTarruzzeeRWLuCC1987The evolution of MHC class II genes within the genus musDavidCSH-2 Antigens: Genes, Molecules, FunctionSpringer13915310.1007/978-1-4757-0764-9_14WalpoleNGKjer-NielsenLKostenkoLMcCluskeyJBrooksAGRossjohnJClementsCS2010The structure and stability of the monomorphic HLA-G are influenced by the nature of the bound peptideJournal of Molecular Biology39746748010.1016/j.jmb.2010.01.05220122941WaltersLCRozbeskyDHarlosKQuastelMSunHSpringerSRamboRPMohammedFJonesEYMcMichaelAJGillespieGM2022Primary and secondary functions of HLA-E are determined by stability and conformation of the peptide-bound complexesCell Reports3911095910.1016/j.celrep.2022.11095935705051WengerAMPelusoPRowellWJChangP-CHallRJConcepcionGTEblerJFungtammasanAKolesnikovAOlsonNDTöpferAAlongeMMahmoudMQianYChinC-SPhillippyAMSchatzMCMyersGDePristoMARuanJMarschallTSedlazeckFJZookJMLiHKorenSCarrollARankDRHunkapillerMW2019Accurate circular consensus long-read sequencing improves variant detection and assembly of a human genomeNature Biotechnology371155116210.1038/s41587-019-0217-931406327WroblewskiEEGuethleinLANormanPJLiYShawCMHanASNdjangoJBNAhuka-MundekeSGeorgievAVPeetersMHahnBHParhamP2017Bonobos maintain immune system diversity with three functional types of MHC-BJournal of Immunology1983480349310.4049/jimmunol.160195528348269WuYGaoFLiuJQiJGostickEPriceDAGaoGF2011Structural Basis of diverse peptide accommodation by the rhesus Macaque MHC Class I Molecule Mamu-B*17: Insights into immune protection from simian immunodeficiency virusThe Journal of Immunology1876382639210.4049/jimmunol.1101726WuCHSuchardMADrummondAJ2013Bayesian selection of nucleotide substitution models and their site assignmentsMolecular Biology and Evolution3066968810.1093/molbev/mss25823233462XuSXRenWHLiSZWeiFWZhouKYYangG2009Sequence polymorphism and evolution of three cetacean MHC genesJournal of Molecular Evolution6926027510.1007/s00239-009-9272-z19693422YagitaYKuseNKurokiKGatanagaHCarlsonJMChikataTBrummeZLMurakoshiHAkahoshiTPfeiferNMallalSJohnMOseTMatsubaraHKandaRFukunagaYHondaKKawashimaYAriumiYOkaSMaenakaKTakiguchiM2013Distinct HIV-1 escape patterns selected by cytotoxic T cells with identical epitope specificityJournal of Virology872253226310.1128/JVI.02572-1223236061YamamotoYMoritaDShimaYMidorikawaAMizutaniTSuzukiJMoriNShiinaTInokoHTanakaYMikamiBSugitaM2019Identification and Structure of an MHC Class I-encoded protein with the potential to present N-Myristoylated 4-mer Peptides to T CellsJournal of Immunology2023349335810.4049/jimmunol.190008731043477YasumizuYSakaueSKonumaTSuzukiKMatsudaKMurakamiYKuboMPalamaraPFKamataniYOkadaY2020Genome-wide natural selection signatures are linked to genetic risk of modern phenotypes in the Japanese populationMolecular Biology and Evolution371306131610.1093/molbev/msaa00531957793YinLCrawfordFMarrackPKapplerJWDaiS2012T-cell receptor (TCR) interaction with peptides that mimic nickel offers insight into nickel contact allergyPNAS109185171852210.1073/pnas.121592810923091041Zavala-RuizZStrugIAndersonMWGorskiJSternLJ2004A polymorphic pocket at the P10 position contributes to peptide binding specificity in class II MHC proteinsChemistry & Biology111395140210.1016/j.chembiol.2004.08.007ZhangSLiuJChengHTanSQiJYanJGaoGF2011Structural basis of cross-allele presentation by HLA-A*0301 and HLA-A*1101 revealed by two HIV-derived peptide complexesMolecular Immunology4939540110.1016/j.molimm.2011.08.01521943705ZhaoMWangYShenHLiCChenCLuoZWuH2013Evolution by selection, recombination, and gene duplication in MHC class I genes of two Rhacophoridae speciesBMC Evolutionary Biology1311310.1186/1471-2148-13-11323734729ZhouFCaoHZuoXZhangTZhangXLiuXXuRChenGZhangYZhengXJinXGaoJMeiJShengYLiQLiangBShenJShenCJiangHZhuCFanXXuFYueMYinXYeCZhangCLiuXYuLWuJChenMZhuangXTangLShaoHWuLLiJXuYZhangYZhaoSWangYLiGXuHZengLWangJBaiMChenYChenWKangTWuYXuXZhuZCuiYWangZYangCWangPXiangLChenXZhangAGaoXZhangFXuJZhengMZhengJZhangJYuXLiYYangSYangHWangJLiuJHammarströmLSunLWangJZhangX2016Deep sequencing of the MHC region in the Chinese population contributes to studies of complex diseaseNature Genetics4874074610.1038/ng.357627213287ZhuSLiuKChaiYWuYLuDXiaoWChengHZhaoYDingCLyuJLouYGaoGFLiuWJ2019Divergent peptide presentations of HLA-A*30 alleles revealed by structures with pathogen peptidesFrontiers in Immunology10170910.3389/fimmu.2019.0170931396224Appendix 1MHC Nomenclature
MHC allele nomenclature.
(A) Human HLA alleles are named in a standard fashion, with the gene name followed by four colon-separated fields. The first field indicates a broad-scale allele group which sometimes corresponds to a serological antigen. The second field denotes a specific HLA protein. The third field indicates synonymous changes to the nucleotide sequence in the coding region, while the fourth field is used to distinguish alleles with differences in the noncoding regions. If an allele’s expression has been characterized, an informative suffix is sometimes added (Robinson et al., 2024; Marsh et al., 2010). (B) Researchers have applied the same format to non-human alleles, with some key differences. Instead of ‘HLA’, a prefix which concatenates the first two letters of the genus name with the first two letters of the species name is used, except in certain cases where the species’ MHC system was named long ago. Paralogs can be distinguished using numbers, but sequences unassigned to a particular locus or paralog might incorporate a ‘W’ in the gene name. Use of expression tags varies, with some being added to the end of the gene name instead of the end of the entire allele name. Pseudogenes can be denoted with gene name suffixes, gene names themselves, expression suffixes, or not at all. For both human and non-human alleles, the lack of an expression suffix does not imply normal expression (de Groot et al., 2020). SLA: Swine Leukocyte Antigen; Chsa: Chlorocebus sabaeus—green monkey; Lero: Leontopithecus rosalia—golden lion tamarin; Mamu: Macaca mulatta—rhesus macaque; Aotr: Aotus trivirgatus—three-striped night monkey; Popy: Pongo pygmaeus—Bornean orangutan; Ceat: Cercocebus atys—sooty mangabey; Sala: Saguinus labiatus—white-lipped tamarin; Gogo: Gorilla gorilla—Western gorilla; Rano: Rattus norvegicus—brown rat; Patr: Pan troglodytes—chimpanzee; Papa: Pan paniscus—bonobo.
The large number of genes, some with thousands of alleles, necessitates a consistent naming scheme (Appendix 1—figure 1). Known alleles are given names such as ‘Aole-DQB1*23:01’, and names are maintained and updated by the WHO Nomenclature Committee for Factors of the HLA System (Robinson et al., 2024; Marsh et al., 2010). First, the species of origin is indicated by a four-letter prefix consisting of the first two letters of the genus name and the first two letters of the species name, for example ‘Chsa-’ for Chlorocebus sabaeus, the green monkey. There are some exceptions, usually because these MHC systems were first investigated before the naming scheme was put into place. These include ‘HLA-’ for human, ‘H2-’ for mouse, ‘RT1-’ for rat, and ‘SLA-’ for swine, among others (de Groot et al., 2020; de Groot et al., 2012).
After the hyphen is the locus designation. Some species, such as humans, have a relatively simple landscape of MHC genes, making it easy to identify sequences that belong to a particular gene. However, other species have recent gene expansions and considerable region conformation diversity, making it difficult to assign alleles to genes. In some cases, these are given generic locus designations; for example, rhesus macaques have at least 19 paralogous B loci, but most are given the ambiguous name ‘Mamu-B’, with the exception of a few well-characterized genes such as Mamu-B17. In other cases, unassigned sequences are given a working designation indicated by a ‘W’, such as ‘Popy-DRB*W113:01’. In this example, the allele definitely belongs to a DRB paralog, but it is unclear which one. Some locus names are given a ‘Ps’ suffix to indicate they are pseudogenes, such as ‘Caja-G5Ps’. However, not all pseudogenes are labeled this way, so one should not assume the lack of a ‘Ps’ suffix means a gene is functional (de Groot et al., 2020; de Groot et al., 2012).
After the species and locus name, each MHC allele is designated by up to four fields separated by colons. The first field designates the type or family. Types often, but not always, correspond to the broad serological reactivity of the allele, as many were named before full sequences were known. To facilitate comparison across closely related species, researchers generally try to give related MHC alleles the same first-field designation, for example Gogo-A*02 and HLA-A*02. However, certain genes do not follow this general rule. For example, MHC-DPB1 has undergone considerable gene conversion, resulting in no distinct types; thus, a shared first-field designation between species is meaningless for this gene (de Groot et al., 2012; de Groot et al., 2020). The second field designates the allele subtype, or unique amino acid sequence. For example, ‘Patr-A*08:01’ and ‘Patr-A*08:02’ are part of the same allelic family, but have some nonsynonymous differences. Synonymous changes are specified by the third field. For example, ‘Paan-DPB1*03:01:01’ and ‘Paan-DPB1*03:01:02’ have silent substitutions which ultimately result in the same protein. Lastly, the fourth field is used to describe changes to the noncoding regions—that is, the 5’ and 3’ UTRs and the introns. Of course, this requires that these regions have been sequenced, so not all alleles will have a fourth field. Finally, alleles can also be followed by an optional suffix to describe expression changes, most commonly ‘N’ for a null/nonexpressed allele or ‘L’ for a lowly-expressed allele (Hurley, 2021; Douillard et al., 2021).
In general, caution must be taken in interpreting allele names. First, because not all alleles are resolved at three- and four-field resolution, the names are not all strictly hierarchical; alleles which have all four fields cannot always simply be truncated to obtain the two-field version. Second, because human alleles were named in order of discovery, alleles with very different one- or two-field designations could ultimately have the same nucleotide or amino acid sequence in the peptide-binding groove. When discussing functional consequences, it is relevant to group alleles by their nucleotide or amino acid sequence in the PBR (designated G- and P-groups, respectively) and not necessarily by their one- or two-field name (Hurley, 2021; Douillard et al., 2021). Additionally, the suffixes can be misleading because not every allele has had its expression level characterized—the absence of an ‘L’ does not mean that an allele has normal expression (Hurley, 2021). Despite these small issues, the naming system is generally intuitive and very useful for understanding alleles at a glance. In this work, alleles obtained from the IPD-MHC and IPD-IMGT/HLA databases are named this way, but sequences obtained from RefSeq are labeled by accession number or location in a genome.
10.7554/eLife.103547.3.sa0eLife AssessmentEnardDavidReviewing EditorUniversity of ArizonaUnited StatesCompellingImportant
This important manuscript presents a thorough analysis of trans-specific polymorphism (TSP) in Major Histocompatibility Complex gene families across primates. The analysis makes the most of currently available genomic data and methods to substantially increase the amount and evolutionary time that TSPs can be observed. Both false negative TSPs due to missing genes at the assembly and/or annotation level, as well as false positives due to read mismapping with missing paralogs, are well assessed and discussed. Overall the evidence provided is compelling, and the manuscript clearly delineates the path for future progress on the topic.
In this study, the authors characterized population genetic variation in the MHC locus across primates and looked for signals of long-term balancing selection (specifically trans-species polymorphism, TSP) in this highly polymorphic region. To carry out these tasks, they used Bayesian methods for phylogenetic inference (i.e. BEAST2) and applied a new Bayesian test to quantify evidence supporting monophyly vs. transspecies polymorphism for each exon across different species pairs. Their results, although mostly confirmatory, represent the most comprehensive analyses of primate MHC evolution to date and novel findings or possible discrepancies are clearly pointed out. However, as the authors discuss, the available data are insufficient to fully capture primates' MHC evolution.
Strengths of the paper include: using appropriate methods and statistically rigorous analyses; very clear figures and detailed description of the results methods that make it easy to follow despite the complexity of the region and approach; a clever test for TSP that is then complemented by positive selection tests and the protein structures for a quite comprehensive study.
That said, weaknesses include: lack of information about how many sequences are included and whether uneven sampling across taxa might results in some comparisons without evidence for TSP; frequent reference to the companion paper instead of summarizing (at least some of) the critical relevant information (e.g., how was orthology inferred?); no mention of the quality of sequences in the database and whether there is still potential effects of mismapping or copy number variation affecting the sequence comparison.
Comments on revisions:
The authors have sufficiently addressed the reviewers' comments or provided additional details justifying their work. In particular, expansion of the discussion section on limitations of the analysis and clearer reference to how this relates to their companion paper represent improvements. Remaining suggestions are to still make clearer how much sparsity of sequences in the database may impact the conclusions (e.g., is this more of a problem for some genes or taxa than others? Is it a small problem or a large problem?). The data summary tables are a bit hard to read and seem to contain some information not used in the article - maybe the presentation of these could be improved or the full details, or a shorter table summer in the main paper and full details only in the supplement.
The study uses publicly available sequences of classical and non-classical genes from a number of primate species to assess the extent and depth of TSP across the primate phylogeny. The analyses were carried out in a coherent and, in my opinion, robust inferential framework and provide evidence for ancient (even > 30 million years) TSP at several classical class I and class II genes. The authors also characterise evolutionary rates at individual codons, map these rates onto MHC protein structures, and find that the fastest evolving codons are extremely enriched for autoimmune and infectious disease associations.
Strengths:
The study is comprehensive, relying on a large data set, state-of-the-art phylogenetic analyses and elegant tests of TSP. The results are not entirely novel, but a synthesis and re-analysis of previous findings is extremely valuable and timely.
Weaknesses:
Following the revision by the Authors I see mostly one weakness - Older literature on the subject is duly cited, but the discussion of the findings the context of this literature is limited.
Comments on revisions:
Lines 441-452 - In this section, you discuss an apparent paradox between long-lived balancing selection and strong directional selection, referencing elevated substitution rates. However, this issue is more nuanced and may not be best framed in terms of substitution rates. That terminology is common in phylogenetic analyses, where differences between sequences-or changes along phylogenetic branches-are often interpreted as true substitutions in the population genetic sense. In the case of MHC trees and the rates you're discussing here, the focus is more accurately on the rate at which new mutations become established within particular allelic lineages. So while this still concerns evolutionary rates at specific codons, equating them directly with substitution rates may be misleading. A more precise term or framing might be warranted in this context.
10.7554/eLife.103547.3.sa3Author responseFortierAlyssa LynAuthorStanford UniversityStanfordUnited StatesPritchardJonathan KAuthorStanford UniversityStanfordUnited States
The following is the authors’ response to the original reviews
Public Reviews:
Reviewer #1 (Public review):
Summary:
MHC (Major Histocompatibility Complex) genes have long been mentioned as cases of trans-species polymorphism (TSP), where alleles might have their most recent common ancestor with alleles in a different species, rather than other alleles in the same species (e.g., a human MHC allele might coalesce with a chimp MHC allele, more recently than the two coalesce with other alleles in either species). This paper provides a more complete estimate of the extent and ages of TSP in primate MHC loci. The data clearly support deep TSP linking alleles in humans to (in some cases) old world monkeys, but the amount of TSP varies between loci.
Strengths:
The authors use publicly available datasets to build phylogenetic trees of MHC alleles and loci. From these trees they are able to estimate whether there is compelling support for Trans-species polymorphisms (TSPs) using Bayes Factor tests comparing different alternative hypotheses for tree shape. The phylogenetic methods are state-of-the-art and appropriate to the task.
The authors supplement their analyses of TSP with estimates of selection (e.g., dN/dS ratios) on motifs within the MHC protein. They confirm what one would suspect: classical MHC genes exhibit stronger selection at amino acid residues that are part of the peptide binding region, and non-classical MHC exhibit less evidence of selection. The selected sites are associated with various diseases in GWAS studies.
Weaknesses:
An implication drawn from this paper (and previous literature) is that MHC has atypically high rates of TSP. However, rates of TSP are not estimated for other genes or gene families, so readers have no basis of comparison. No framework to know whether the depth and frequency of TSP is unusual for MHC family genes, relative to other random genes in the genome, or immune genes in particular. I expect (from previous work on the topic), that MHC is indeed exceptional in this regard, but some direct comparison would provide greater confidence in this conclusion.
We agree that context is important! Although we expected to get the most interesting results from studying the classical genes, we did include the non-classical genes specifically for comparison. They are located in the same genomic region, have multiple sequences catalogued in different species (although they are less diverse), and perform critical immune functions. We think this is a more appropriate set to compare with the classical MHC genes than, say, a random set of genes. Interestingly, we did not detect TSP in these non-classical genes. This likely means that the classical MHC genes are truly exceptional, but it could also mean that not enough sequences are available for the non-classical genes to detect TSP.
It would be very interesting to repeat this analysis for another gene family to see whether such deep TSP also occurs in other immune or non-immune gene families. We are lucky that decades of past work and a dedicated database exists for cataloging MHC sequences. When this level of sequence collection is achieved for other highly polymorphic gene families, it will be possible to do a comparable analysis.
Given the companion paper's evidence of genic gain/loss, it seems like there is a real risk that the present study under-estimates TSP, if cases of TSP have been obscured by the loss of the TSP-carrying gene paralog from some lineages needed to detect the TSP. Are the present analyses simply calculating rates of TSP of observed alleles, or are you able to infer TSP rates conditional on rates of gene gain/loss?
We were not able to infer TSP rates conditional on rates of gene gain/loss. We agree that some cases of TSP were likely lost due to the loss of a gene paralog from certain species. Furthermore, the dearth of MHC whole-region and allele sequences available for most primates makes it difficult to detect TSP, even if the gene paralog is still present. Long-read sequencing of more primate genomes should help with this. We agree that it would also be very interesting to study TSPs that were maintained for millions of years but were lost recently.
Figure 5 (and 6) provide regression model fits (red lines in panel C) relating evolutionary rates (y axis not labeled) to site distance from the peptide binding groove, on the protein product. This is a nice result. I wonder, however, whether a linear model (as opposed to non-linear) is the most biologically reasonable choice, and whether non-linear functions have been evaluated. The authors might consider generalized additive models (GAMs) as an alternative that relaxes linearity assumptions.
We agree that a linear model is likely not the most biologically reasonable choice, as protein interactions are complex. However, we made the choice to implement the simplest model because the evolutionary rates we inferred were relative, making parameters relatively meaningless. We were mainly concerned with positive or negative slopes and we leave the rest to the protein interaction experts.
The connection between rapidly evolving sites, and disease associations (lines 382-3) is very interesting. However, this is not being presented as a statistical test of association. The authors note that fast-evolving amino acids all have at least one association: but is this really more disease-association than a random amino acid in the MHC? Or, a randomly chosen polymorphic amino acid in MHC? A statistical test confirming an excess of disease associations would strengthen this claim.
To strengthen this claim, we added Figure 6 - Figure Supplement 7 (NOTE: this needs to be renamed as Table 1 - Figure Supplement 1, which the eLife template does not allow). Here, we plot the number of associations for each amino acid against evolutionary rate, revealing a significant positive slope in Class I. We also added explanatory text for this figure in lines 400-404.
Reviewer #2 (Public review):
Summary
In this study, the authors characterized population genetic variation in the MHC locus across primates and looked for signals of long-term balancing selection (specifically trans-species polymorphism, TSP) in this highly polymorphic region. To carry out these tasks, they used Bayesian methods for phylogenetic inference (i.e. BEAST2) and applied a new Bayesian test to quantify evidence supporting monophyly vs. transspecies polymorphism for each exon across different species pairs. Their results, although mostly confirmatory, represent the most comprehensive analyses of primate MHC evolution to date and novel findings or possible discrepancies are clearly pointed out. However, as the authors discuss, the available data are insufficient to fully capture primates' MHC evolution.
Strengths of the paper include: using appropriate methods and statistically rigorous analyses; very clear figures and detailed description of the results methods that make it easy to follow despite the complexity of the region and approach; a clever test for TSP that is then complemented by positive selection tests and the protein structures for a quite comprehensive study.
That said, weaknesses include: lack of information about how many sequences are included and whether uneven sampling across taxa might results in some comparisons without evidence for TSP; frequent reference to the companion paper instead of summarizing (at least some of) the critical relevant information (e.g., how was orthology inferred?); no mention of the quality of sequences in the database and whether there is still potential effects of mismapping or copy number variation affecting the sequence comparison.
To address these comments, we added Tables 2-4 to allow readers to more readily understand the data we included in each group. We refer to these tables in the introduction (line 95), in the “Data” section of the results (lines 128-129), and the “Data” section of the methods (lines 532-534). We also added text (lines 216-219 and 250-252) to more explicitly point out that our method is conservative when few sequences are available.
We also added a paragraph to the discussion which addresses data quality and mismapping issues (lines 473-499).
We clarified the role of our companion paper (line 49-50) by changing “In our companion paper, we explored the relationships between the different classical and non-classical genes” to “In our companion paper, we built large multi-gene trees to explore the relationships between the different classical and non-classical genes.” We also changed the text in lines 97-99 from “In our companion paper, we compared genes across dozens of species and learned more about the orthologous relationships among them” to “In our companion paper, we built trees to compare genes across dozens of species. When paired with previous literature, these trees helped us infer orthology and assign sequences to genes in some cases.”
Reviewer #3 (Public review):
Summary
The study uses publicly available sequences of classical and non-classical genes from a number of primate species to assess the extent and depth of TSP across the primate phylogeny. The analyses were carried out in a coherent and, in my opinion, robust inferential framework and provided evidence for ancient (even > 30 million years) TSP at several classical class I and class II genes. The authors also characterise evolutionary rates at individual codons, map these rates onto MHC protein structures, and find that the fastest evolving codons are extremely enriched for autoimmune and infectious disease associations.
Strengths
The study is comprehensive, relying on a large data set, state-of-the-art phylogenetic analyses and elegant tests of TSP. The results are not entirely novel, but a synthesis and re-analysis of previous findings is extremely valuable and timely.
Weaknesses
I've identified weaknesses in several areas (details follow in the next section):
- Inadequate description and presentation of the data used
- Large parts of the results read like extended figure captions, which breaks the flow. - Older literature on the subject is duly cited, but the authors don't really discuss their findings in the context of this literature.
- The potential impact of mechanisms other than long-term maintenance of allelic lineages by balancing selection, such as interspecific introgression and incorrect orthology assessment, needs to be discussed.
We address these comments in the more detailed section below.
Recommendations for the authors:
Reviewer #1 (Recommendations for the authors):
The abstract could benefit from being sharpened. A personal pet peeve is a common habit of saying we don't know everything about a topic (line 16 - "lack a full picture of primate MHC evolution"); We never know everything on a topic, so this is hardly a strong rationale to do more work on it. This is followed by "to start addressing this gap" - which is vague because you haven't explicitly stated any gap, you simply said we are not yet omniscent on the topic. Please clearly identify a gap in our knowledge, a question that you will be able to answer with this paper.
That makes sense! We added another sentence to the abstract to make the specific gap clearer. Inserted “In particular, we do not know to what extent genes and alleles are retained across speciation events” in lines 16-17.
Reviewer #2 (Recommendations for the authors):
- Some discussion of alternative explanations when certain comparisons were not found to have TSP - is this consistent with genetic drift sometimes leading to lineage loss, or does it suggest that the proposed tradeoff between autoimmunity and pathogen recognition might differ depending on primates' life history and/or exposure to similar pathogens? Could the trade-off of pathogen to self-recognition not be as costly in some species?
This is consistent with genetic drift, as no lineages are expected to be maintained across these distantly-diverged primates under neutral selection. These ideas are certainly possible, but our Bayes Factor test only reveals evidence (or lack thereof) for deviations from the species tree and cannot provide reasons why or why not.
- It would be interesting to put these results on very long-term balancing selection in the context of what has been reported at the region for shorter term balancing selection. The discussion compares findings of previous genes in the literature but not regarding the time scale.
Indeed, there is some evidence for the idea of “divergent allele advantage”, in which MHC-heterozygous individuals have a greater repertoire of peptides that they can present, leading to greater resistance against pathogens and greater fitness. This heterozygote advantage thus leads to balancing selection (Pierini and Lenz, 2018; Chowell et al., 2019). Our discussion mentions other time scales of balancing selection across the primates at the MHC and other loci, but we choose to focus more on long-term than short-term balancing selection.
- Lines 223-226 - how is the difference in BF across exons in MHC-A to be interpreted? The paragraph is about MHC-A, but then the explanation in the last sentence is for when similar BF are observed which is not the case for MHC-A. Is this interpreted as lack of evidence for TSP? Or something about recombination or gene conversion? Or that one exon may be under balancing selection but not the other?
Thank you for pointing out the confusing logic in this paragraph.
Previous: “For MHC-A, Bayes factors vary considerably depending on exon and species pair. Many sequences had to be excluded from MHC-A comparisons because they were identified as gene-converted in the GENECONV analysis or were previously identified as recombinants (Hans et al., 2017, Gleimer et al., 2011, Adams and Parham, 2001). Importantly, for MHC-A we do not see concordance in Bayes factors across the different exons, whereas we do for the other gene groups. Similar Bayes factors across all exons for a given comparison is thus evidence in favor of TSP being the primary driver of the observed deep coalescence structure (rather than recombination or gene conversion).” Current (lines 228-238):
“For MHC-A, Bayes factors vary considerably depending on exon and species pair. Past work suggests that this gene has had a long history of gene conversion affecting different exons, resulting in different evolutionary histories for different parts of the gene (Hans et al., 2017, Gleimer et al., 2011, Adams and Parham, 2001). Indeed, we excluded many MHC-A sequences from our Bayes factor calculations because they were identified as gene-converted in our GENECONV analysis or were previously suggested to be recombinants. As shown in Figure 3, the lack of concordance in Bayes factors across the different exons for MHC-A is evidence for gene conversion, rather than balancing selection, being the most important factor in this gene's evolution. In contrast, the other gene groups generally show concordance in Bayes factors across exons. We interpret this as evidence in favor of TSP being the primary driver of the observed deep coalescence structure for MHC-B and -C (rather than recombination or gene conversion).”
- In Figures 5C and 6C, the points sometimes show a kind of smile pattern of possibly higher rates further from the peptide. Did authors explore other fits like a polynomial? Or, whether distance only matters in close proximity to the peptide? Out of curiosity, is it possible to map substitution time/branch into the distance to the peptide binding region for each substitution? Is there any pattern with distance to interacting proteins in non-peptide binding MHC proteins like MHC-DOA? Although they don't have a PBR they do interact with other proteins.
Thank you for these ideas! We did not explore other fits, such as a polynomial, because we wanted to implement the simplest model. Our evolutionary rates are relative, making parameters relatively meaningless. We were mainly concerned with positive or negative slopes and we leave the rest to the protein interaction experts.
There is most likely a relationship between evolutionary rate and the distance to interacting proteins in the non-peptide-binding molecules MHC-DM and -DO. However, there are few currently available models and it is difficult to determine which residues in these models are actually interacting. However, researchers with more experience in protein interactions would be able to undertake such an analysis.
- How biased is the database towards human alleles? Could this affect some of the analyses, including the coincidence of rapidly evolving sites with associations? Are there more associations than expected under some null model?
While the database is indeed biased toward human alleles, we included only a small subset of these in order to create a more balanced data set spanning the primates. This is unlikely to affect the coincidence of rapidly-evolving sites with associations; however, we note that there are no such association studies meeting our criteria in other species, meaning the associations are only coming from studies on humans.
- To this reader, it is unnecessary and distracting to describe the figures within the text; there are frequent sentences in the text that belongs in the figure legend instead (e.g., lines 139-143, 208-211, 214-215, 328-330, etc). It would be better to focus on the results from the figures and then cite the figure, where the colors and exactly what is plotted can be in the figure legend.
We appreciate these comments on overall flow. We removed lines 139-143 and lengthened the Figure 2 caption (and associated supplementary figure captions) to contain all necessary detail. We removed lines 208-211 and 214-215 and lengthened the captions for Figure 3, Figure 4, and associated supplementary figures. We removed a sentence from lines 303-304.
- I'm still concerned that the poor mappability of short-read data is contributing in some ways. Were the sequences in the database mostly from long-reads? Was nucleotide diversity calculated directly from the sequences in the database or from another human dataset? Is missing data at some sites accounted for in the denominator?
The sequences in the database are mostly from short reads and come from a wide array of labs. We have added a paragraph to the discussion to explain the limitations of this (lines 473-499). However, the nucleotide diversity calculations shown in Figure 1 do not rely on the MHC database; rather, they are calculated from the human genomes in the 1000 Genomes project. Nucleotide diversity would be calculable for other species, but we did not do so for exactly the reason you mention–too much missing data.
- The Figure 2 and Figure 3 supplements took me a little bit to understand - is it really worth pointing out the top 5 Bayes-factor comparisons when there is no evidence for TSP? A lot of the colored squares are not actually supporting TSP but in the grids you can't see which are and which aren't without looking at the Bayes Factor. I wonder if it would help if only those with BF > 100 were shown? Or if these were marked some other way so that it was easy to see where TSPs are supported.
Thank you for your perspective on these figures! We initially limited them to only show >100 Bayes factors for each gene group and region, but some gene groups have no high Bayes factors. Additionally, the “summary” tree pictured in these figures is necessarily a simplification of the full space of posterior trees. We felt that showing low Bayes factor comparisons could help readers understand this relationship. For example, allele sets that look non-monophyletic on the summary tree may still have a low Bayes factor, showing that they are generally monophyletic throughout the larger (un-visualizable) space of trees.
Reviewer #3 (Recommendations for the authors):
Specific comments
Abstract
I think the abstract would benefit from some editing. For example, one might get the impression that you equate allele sharing, which would normally be understood as sharing identical sequences, with sharing ancestral allelic lineages. This distinction is important because you can have many TSPs without sharing identical allele sequences. In l. 20 you write about "deep TSP", which requires either definition of reformulation. In l. 21-23 you seem to suggest that long-term retention of allelic lineages is surprising in the light of rapid sequence evolution - it may be, depending on the evolutionary scenarios one is willing to accept, but perhaps it's not necessary to float such a suggestion in the abstract where it cannot be properly explained due to space constraints? The last sequence needs a qualifier like "in some cases".
Thank you for catching these! For clarity, we changed several words:
● “alleles” to “allelic lineages” in line 13
● “deep” to “ancient” in line 21
● “Despite” to “in addition to” in line 22
● Added “in some cases” to line 28
Results - Overall, parts of the results read like extended figure captions. I understand that the authors want to make the complex figures accessible to the reader. However, including so much information in the text disrupts the flow and makes it difficult to follow what the main findings and conclusions are.
We appreciate these comments on overall flow. We removed lines 139-143 and lengthened the Figure 2 caption (and associated supplementary figure captions) to contain all necessary detail. We removed lines 208-211 and 214-215 and lengthened the captions for Figure 3, Figure 4, and associated supplementary figures. We removed a sentence from lines 303-304.
l. 37-39 such a short sentence on non-classical MHC is necessarily an oversimplification, I suggest it be expanded or deleted.
There is certainly a lot to say about each of these genes! While we do not have space in this paper’s introduction to get into these genes’ myriad functions, we added a reference to our companion paper in lines 40-41:
“See the appendices of our companion paper (Fortier and Pritchard, 2025) for more detail.”
These appendices are extensive, and readers can find details and references for literature on each specific gene there. In addition, several genes are mentioned in analyses further on in the results, and their specific functions are discussed in more detail when they arise.
l. 47 -49 It would be helpful to briefly outline your criteria for selecting these 17 genes, even if this is repeated later.
Thank you! For greater clarity, we changed the text (lines 50-52) from “Here, we look within 17 specific genes to characterize trans-species polymorphism, a phenomenon characteristic of long-term balancing selection.” to “Here, we look within 17 specific genes---representing classical, non-classical, Class I, and Class II ---to characterize trans-species polymorphism, a phenomenon characteristic of long-term balancing selection.“
l.85-87 I may be completely wrong, but couldn't problems with establishing orthology in some cases lead to false inferences of TSP, even in primates? Or do you think the data are of sufficient quality to ignore such a possibility? (you touch on this in pp. 261-264)
Yes, problems with establishing orthology can lead to false inferences of TSP, and it has happened before. For example, older studies that used only exon 2 (binding-site-encoding) of the MHC-DRB genes inferred trees that grouped NWM sequences with ape and OWM sequences. Thus, they named these NWM genes MHC-DRB3 and -DRB5 to suggest orthology with ape/OWM MHC-DRB3 and -DRB5, and they also suggested possible TSP between the groups. However, later studies that used non-binding-site-encoding exons or introns noticed that these NWM sequences did not group with ape/OWM sequences (which now shared the same name), providing evidence against orthology. This illustrates that establishing orthology is critical before assessing TSP (as is comparing across regions). This is part of the reason we published a companion paper (https://doi.org/10.7554/eLife.103545.1), which clears up questions of orthology and supports the analyses we did in this paper. In cases where orthology was ambiguous, this also helped us to be conservative in our conclusions here. The problems with ambiguous gene assignment are also discussed in lines 488-499.
l. 88-93 is the first place (others are pp. 109-118 and 460-484) where a fuller description of the data used would be welcome. It's clear that the amount of data from different species varies enormously, not only in the number of alleles per locus, but also in the loci for which polymorphism data are available. In such a synthesis study, one would expect at least a tabulation of the data used in the appendices and perhaps a summary table in the main article.
l. 109-118 Again, a more quantitative summary of the data used, with reference to a table, would be useful.
Thank you! To address these comments, we added Tables 2-4 to allow readers to more readily understand the data we included in each group. We refer to these tables in the introduction (line 95), in the “Data” section of the results (lines 128-129), and the “Data” section of the methods (lines 532-534). Supplementary Files listing the exact alleles and sequences used in each group are also included in the resubmission.
l. 123-124 here you say that the definition of the "16 gene groups" is in the methods (probably pp. 471-484), but it would be useful to present an informative summary of your rationale in the introduction or here
Thank you! We agree that it is helpful to outline these groups earlier. We have changed the paragraph in lines 123-135 from:
“We considered 16 gene groups and two or three different genic regions for each group: exon 2 alone, exon 3 alone, and/or exon 4 alone. Exons 2 and 3 encode the peptide-binding region (PBR) for the Class I proteins, and exon 2 alone encodes the PBR for the Class II proteins. For the Class I genes, we also considered exon 4 alone because it is comparable in size to exons 2 and 3 and provides a good contrast to the PBR-encoding exons. See the Methods for more detail on how gene groups were defined. Because few intron sequences were available for non-human species, we did not include them in our analyses.” To:
“We considered 16 gene groups spanning MHC classes and functions. These include the classical Class I genes (MHC-A-related, MHC-B-related, MHC-C-related), non-classical Class I genes (MHC-E-related, MHC-F-related, MHC-G-related), classical Class IIA genes (MHC-DRA-related, MHC-DQA-related, MHC-DPA-related), classical Class IIB genes (MHC-DRB-related, MHC-DQB-related, MHC-DPB-related), non-classical Class IIA genes (MHC-DMA-related, MHC-DOA-related), and non-classical Class IIB genes (MHC-DMB-related, MHC-DOB-related). We studied two or three different genic regions for each group: exon 2 alone, exon 3 alone, and (for Class I) exon 4 alone. Exons 2 and 3 encode the peptide-binding region (PBR) for the Class I proteins, and exon 2 alone encodes the PBR for the Class II proteins. For the Class I genes, we also considered exon 4 alone because it is comparable in size to exons 2 and 3 and provides a good contrast to the PBR-encoding exons. Because few intron sequences were available for non-human species, we did not include them in our analyses.”
l. 100 "alleles" -> "allelic lineages"
Thank you for catching this. We have changed this language in line 104.
l. 227-238 it's important to discuss the possible effect of the number of sequences available on the detectability of TSP - this is particularly important as the properties of MHC genealogies may differ considerably from those expected for neutral genealogies.
This is a good point that may not be obvious to readers. We have added several sentences to clarify this:
Line 193-194: “In a neutral genealogy, monophyly of each species' sequences is expected.”
Line 213-219: “Note that the number of sequences available for comparison also affects the detectability of TSP. For example, if the only sequences available are from the same allelic lineage, they will coalesce more recently in the past than they would with alleles from a different lineage and would not show evidence for TSP. This means our method is well-suited to detect TSP when a diverse set of allele sequences are available, but it is conservative when there are few alleles to test. There were few available alleles for some non-classical genes, such as MHC-F, and some species, such as gibbon.”
Line 244-246: “However, since there are fewer alleles available for the non-classical genes, we note that our method is likely to be conservative here.”
l. 301 and 624-41 it's been difficult for me to understand the rationale behind using rates at mostly gap positions as the baseline and I'd be grateful for a more extensive explanation
Normalizing the rates posed a difficult problem. We couldn’t include every single sequence in the same alignment because BEAST’s computational needs scale with the number of sequences. Therefore, we had to run BEAST separately on smaller alignments focused on a single group of genes at a time. We still wanted to be able to compare evolutionary rates across genes, but because of the way SubstBMA is implemented, evolutionary rates are relative, not absolute. Recall that to help us compare the trees, we included a common set of “backbone” sequences in all of the 16 alignments. This set included some highly-diverged genes. Initially, we planned to use 4-fold degenerate sites as the baseline sites for normalization, but there simply weren’t enough of them once we included the “backbone” set on top of the already highly diverse set of sequences in each alignment. This diversity presented an opportunity. In BEAST, gaps are treated as missing and do not contribute any probability to the relevant branch or site (https://groups.google.com/g/beast-users/c/ixrGUA1p4OM/m/P4R2fCDWMUoJ?pli=1). So, we figured that sites that were “mostly gap” (a gap in all the human backbone sequences but with an insertion in some sequence) were mostly not contributing to the inference of the phylogeny or evolutionary rates. Because the “backbone” sequences are common to all alignments, making the “mostly gap” sites somewhat comparable across sets while not affecting inferred rates, we figured they would be a reasonable choice for the normalization (for lack of a better option).
We added text to lines 680 and 691-693 to clarify this rationale.
l. 380-84 this overview seems rather superficial. Would it be possible to provide a more quantitative summary?
To make this more quantitative, we plotted the number of associations for each amino acid against evolutionary rate, shown in Figure 6 - Figure Supplement 7 (NOTE: this needs to be renamed as Table 1 - Figure Supplement 1, which the template does not allow). This reveals a significant positive slope for the Class I genes, but not for Class II. We also added explanatory text for this figure in lines 400-404.
Discussion - your approach to detecting TSP is elegant but deserves discussion of its limitations and, in particular, a clear explanation of why detecting TSP rather than quantifying its extent is more important in the context of this work. Another important point for discussion is alternative explanations for the patterns of TSP or, more broadly, gene tree - species tree discordance. Although long-term maintenance of allelic lineages due to long-term balancing selection is probably the most convincing explanation for the observed TSP, interspecific introgression and incorrect orthology assessment may also have contributed, and it would be good to see what the authors think about the potential contribution of these two factors.
Overall, our goal was to use modern statistical methods and data to more confidently assess how ancient the TSP is at each gene. We have added several lines of text (as noted elsewhere in this document) to more clearly illustrate the limitations of our approach. We also agree that interspecific introgression and incorrect orthology assessment can cause similar patterns to arise. We attempted to minimize the effect of incorrect orthology assessment by creating multi-gene trees and exploring reference primate genomes, as described in our companion paper (https://doi.org/10.7554/eLife.103545.1), but cannot eliminate it completely. We have added a paragraph to the discussion to address this (lines 488-499). Interspecific introgression could also cause gene tree-species tree discordance, but we are not sure about how systematic this would have to be to cause the overall patterns we observe, nor about how likely it would have been for various clades of primates across the world.
l. 421 -424 A more nuanced discussion distinguishing between positive selection, which facilitates the establishment of a mutation, and directional selection, which leads to its fixation, would be useful here.
We added clarification to this sentence (line 443-445), from “Indeed, within the phylogeny we find that the most rapidly-evolving codons are substituted at around 2--4-fold the baseline rate.” to “Indeed, within the phylogeny we find that the most rapidly-evolving codons are substituted at around 2--4-fold the baseline rate, generating ample mutations upon which selection may act.”
l. 432-434 You write here about the shaping of TCR repertoires, but I couldn't find any such information in the paper, including Table 1.
We did not include a separate column for these, so they can be hard to spot. They take the form of “TCR 𝛽 Interaction Probability >50%”, “TCR Expression (TRAV38-1)”, or “TCR 𝛼 Interaction Probability >50%” and can be found in Table 1.
l. 436-442 Here a more detailed discussion in the context of divergent allelic advantage and even the evolution of new S-type specificities in plants would be valuable.
We added an additional citation to a review article to this sentence (lines 438-439).
l. 443 The use of the word "training" here is confusing, suggesting some kind of "education" during the lifetime of the animal.
We agree that “train” is not an entirely appropriate term, and have changed it to “evolve” (line 465).
489-491 What data were used for these calculations?
Apologies for missing this citation! We used the 1000 genomes project data, and the citation has been updated (line 541-542).