Review Comparative Genomics of the MHC: Glimpses Into the Evolution Of

Home , 2R hypothesis

Immunity, Vol. 15, 351–362, September, 2001, Copyright 2001 by Cell Press Comparative Genomics of the MHC: Review Glimpses into the Evolution of the Adaptive Immune System

Martin F. Flajnik1 and Masanori Kasahara2,3 especially the section adjacent to the class I region, as 1Department of Microbiology and Immunology the “inflammatory region” (Figure 3) and have suggested University of Maryland at Baltimore that linkage of such genes to each other and to class Room 13-009, 655 West Baltimore Street I/class II genes is somehow advantageous. Yet, there Baltimore, Maryland 21021 is no evidence to support a model that the genes must 2Department of Biosystems Science be linked; many genes in the class III region have no School of Advanced Sciences obvious connections to the adaptive or innate immune The Graduate University for Advanced Studies systems (Klein and Sato, 1998). Hayama 240-0193 One test of whether the linkage of immune-related Japan genes might be beneficial to the organism is to study whether such genes have remained physically linked over long periods of evolutionary time. Here we describe MHCs of nonmammalian vertebrates, emphasizing those Summary features that seem universal as well as those characteristics unique to each group of animals. We also specu- MHC gene organization (size, complexity, gene order) late on MHC origins, concentrating on the role of ancient differs markedly among different species, and yet all en bloc duplications in the genesis of the MHC. nonmammalian vertebrates examined to date have a true “class I region” with tight linkage of genes encod- Origins of the Adaptive Immune System ing the class I presenting and processing molecules. Class I and class II genes have been isolated from all Three paralogous regions of the human genome con- jawed vertebrates (gnathostomes) studied to date (Ka- tain sets of linked genes homologous to various loci sahara et al., 1995; Flajnik et al., 1999a), the cartilaginous in the MHC class I, class II, and/or class III regions, fish (e.g., sharks) being the oldest group (Figure 1). In providing insight into the organization of the “proto fact, all of the genes that define adaptive immunity, MHC” before the emergence of the adaptive immune including T cell receptors (TCR), immunoglobulins (Ig), system in the jawed vertebrates. and MHC (class I/II, LMP, TAP, TAPBP), as well as the recombination-activating genes (RAG1, RAG2) and the Introduction (unknown) mutational machinery acting upon the anti- The human major histocompatibility complex (MHC) is gen receptor genes, are all present in these oldest jawed a large genetic region of almost 4 megabases with well vertebrates (Figure 1, reviewed in Laird et al., 2000). over 100 genes, up to 40% of which are involved to some The two living forms of the older, jawless vertebrates degree in immunity (The MHC Sequencing Consortium, (agnathans), hagfish and lamprey, have yielded no such 1999; Trowsdale, 2001). The genes are historically as- genes, molecules, or mechanisms so far despite exten- sembled into three major regions, class I, class II, and sive scrutiny; furthermore, these animals lack the pri- class III (Figures 2 and 3). The class I region encodes mary and secondary lymphoid organs (thymus and the polymorphic, classical class I (or class Ia) genes, spleen) found in all jawed vertebrates. Thus, the adap- two lineages of nonclassical class I (or class Ib) genes, tive immune system as we know and cherish it arose in and other genes unrelated to the immune system. The its entirety seemingly over a relatively short period of class II region contains the polymorphic class II ␣ and geological time (Bernstein et al., 1996). Some believe ␤ genes, as well as genes encoding the ␥ interferon- that the introduction of a RAG-dependent transposable inducible (immune) proteasome components (LMP2 and element into an existing Ig variable (V)-like gene initiated LMP7) and transporters associated with antigen pro- the adaptive immune system and that a method to gen- cessing (TAP1 and TAP2) that are involved in the produc- erate diversity of antigen-recognition elements became tion and transport of peptides that bind to class Ia mole- manifest (e.g., Agrawal et al., 1998). Because the RAG- cules, respectively. The so-called “extended class II mediated rearrangement break occurs in the part of the region” includes mostly nonimmune-related genes but V gene exon encoding CDR3, and this region is in the does encode the TAP binding protein (TAPBP or ta- center of the TCR that interacts with peptides bound pasin), which is essential for class Ia biosynthesis, and in MHC molecules, it was proposed that ␣/␤ TCR-like RXRB, one transcription factor that regulates class I molecules may have arisen first in evolution, followed expression (Dey et al., 1992). The central class III region by Ig and ␥/␦ TCR that recognize antigen in its free is the most gene-dense and well-conserved (colinear) form (Davis and Bjorkman, 1988). However, it has been section of the MHC in mammals, with the presence of difficult, using phylogenetic analyses, to discern which various genes involved in innate/adaptive immunity antigen receptor is oldest (Richards and Nelson, 2000), such as the complement components C4 and factor B and as described, the oldest vertebrates with an adap- (Bf), the inflammatory cytokine tumor necrosis factor tive immune system have all of the rearranging antigen (TNF), and HSP70 (among others, Figure 3). Gruen and receptor families found in mammals. Weissman (1997) have christened the class III region, Similarly, it is not known whether class I or class II molecules arose first in evolution. Phylogenetic trees 3 Correspondence: [email protected] made by two groups argue for a class II ancestor (re- Immunity 352

Figure 1. Phylogenetic Relationships among Extant Deuterostome Classes and Phyla Species discussed in the text are shown in parentheses. Divergence times are based on molecular data compiled by Kumar and Hedges (1998). The 2R hypothesis assumes two rounds of genome-wide duplication at indicated stages. It is a matter of debate whether hagfish and lampreys are monophyletic. Mya, million years ago. viewed in Hughes and Yeager, 1997; Klein and Sato, Rached et al., 1999). Of over 100 genes located in the 1998), proposed previously on thermodynamic grounds, HLA complex, nearly 40 genes have one, two, or three i.e., a homodimer with each chain having two domains paralogous copies on specific regions of chromosomes arising first, evolving into a class II heterodimer and 1, 9, and 19, namely 1q21-q25/1p11-p32, 9q32-q34, and finally into class I (Kaufman, 1989). Yet, the trees have 19p13.1-p13.3 (for a complete listing of MHC genes with little statistical support, and it is also possible that the such paralogous copies, see Tables S1–S3 [see Supple- peptide binding region (PBR) structure existed in an mental Tables online at http://www.immunity.com/cgi/ ancestor predating both class I and class II and that content/full/15/3/351/DC1]). With the exception of chro- exons encoding a proto-PBR were shuffled onto an exon mosome 1, the paralogous region is basically confined encoding an Ig-like C1-set domain (Flajnik et al., 1991). to either a long or short arm. There is compelling evi- Finally, it is not known whether classical or nonclassical dence that human chromosome 1 underwent a pericen- class I molecules arose first in evolution, or ancestrally tric inversion after the divergence of the human and whether the PBR groove was used to bind antigen, pep- chimpanzee lineages, which is probably responsible for tidic or otherwise. The nonclassical class I molecules the occurrence of the paralogous regions on both arms CD1 and FcRn are just as old as the peptide binding of chromosome 1. class I, and thus the ancestral class I-like molecule may Figure 2 shows that the genes with paralogous copies have had a function outside the immune system (dis- on chromosomes 1, 9, or 19 are distributed from the cussed below). Such problems can only be addressed extended centromeric border of the HLA to its extended by examining the molecular evolution of the adaptive telomeric border containing a class Ib gene, designated immune system. HFE (the gene, the mutation of which causes hemochro- matosis). Therefore, the paralogous region on chromosome 6 encompasses the entire MHC and spans at least megabases. Gene families such as NOTCH and PBX 8ف Genome Paralogy and the MHC Genes within a single species that arose by duplication have copies on all four paralogous regions including the are termed paralogous genes or paralogs. Close inspec- MHC. However, the majority of gene families listed in tion of the vertebrate genome occasionally reveals that Tables S1–S3 have copies only on two or three paralo- closely linked sets of paralogous genes are located on gous regions. This suggests the existence of gene fami- more than two, and typically four, different chromo- lies that share paralogous copies only among chromosomes (Nadeau and Kosowsky, 1991; Lundin, 1993). somes 1, 9, and 19 or between two of them; indeed, a This phenomenon is referred to as genome paralogy, survey of the NCBI database revealed more than 40 such and the chromosomal segments that display genome gene families (Table S4). Thus, when the information in paralogy are called paralogous regions. One of the major Tables S1–S4 is combined, more than 80 gene families developments in the field of MHC genomics is the dis- share at least two paralogous copies among the specific covery that the human genome contains at least three regions of chromosomes 1, 6, 9, and 19. We call these regions paralogous to the MHC (Figure 2, Kasahara et gene families the members of the MHC paralogous al., 1996; Katsanis et al., 1996; Kasahara, 1999; Abi- group. Review 353

Figure 2. Distribution of Paralogous Genes Constituting the MHC Paralogous Group in the Human Genome Genes in the human MHC (top line) are arranged from the centromere to telomere (map not drawn to scale). The size of the extended MHC depicted here is approximately 8 megabases. MHC genes that do not have paralogous copies are indicated below the line; these genes were chosen arbitrarily to give a better orientation of the region. Corresponding paralogous copies on different chromosomes are indicated with the symbols of the same color and shape. Genes that share paralogous copies only among chromosomes 1, 9, and 19 or between two of them are indicated by gene symbols with blue lettering to distinguish them from those with paralogous copies in the MHC, the gene symbols of which are in black lettering. Genes, whose positions are known only cytogenetically, are shown in parentheses for chromosomes 1, 9, and 19. For detailed description of the genes that appear in the figure, see Tables S1–S5. The positions of centromeres (abbreviated as cen) are not drawn to scale except on chromosome 1. The NOTCH2 gene, which has been mapped to the p arm cytogenetically, is on the long arm according to the GenBank map.

Some gene families constituting the MHC paralogous immediate common ancestor based on the observation group have copies outside of the well-established para- that, in several gene families, paralogous copies located logous regions located on chromosomes 1, 6, 9, and on chromosomes 1 and 9 are more closely related to 19. The regions of the human genome that contain more each other than they are to the corresponding copies than two copies of the members of the MHC paralogous on the MHC (Katsanis et al., 1996; Kasahara et al., 1997). group include 5q11-q26, 9p13-p24, 12p11-p13, 15q21- However, if the MHC paralogous group arose by allo- q26, and 21q22.3 (Tables S1–S4). Among them, 12p11- polyploidization (see below), more comprehensive anal- p13 is interesting in that it contains a tapasin-like gene ysis of gene families will be required to deduce the order (Du Pasquier, 2000) and a cluster of NK receptor genes by which the MHC and its three paralogous regions of the C-type lectin family (Yokoyama, 1998). The emerged. chicken MHC contains two NK receptor-like genes of Origin of the MHC Paralogous Group the C-type lectin family (Figure 3, Kaufman et al., 1999). Two explanations have been put forward as to how the Hence, the lectin-like NK receptor gene family might MHC and its three major paralogous regions emerged. have been originally a member of the MHC paralogous The first is that they are descended from a common group (Kasahara, 1999). Another region of interest is ancestral region and emerged as a result of large-scale 15q21-q26 containing the ␤2-microglobulin gene. Many block or chromosomal duplication (Kasahara et al., of these newly discovered paralogous regions of smaller 1996; Kasahara, 1999; Abi-Rached et al., 1999). The sec- size probably emerged as a result of fragmentation and ond is that they represent assemblages of indepen- subsequent translocation of the paralogous segments dently duplicated genes brought into proximity by selec- that now reside on chromosomes 1, 6, 9, and 19. How- tive forces (Hughes, 1998). The major drawback of the ever, their exact nature is not known. latter explanation, which has been proposed mainly on Previously, it was suggested that the paralogous re- the basis of the observation that the age of paralogous gions on chromosomes 1 and 9 probably shared an copies shows considerable variation among gene fami- Immunity 354

Figure 3. Comparative MHC Maps Including the Inferred Ancestral (Ur) MHC Yellow indicates that the gene is found in at least one other species besides human; blue designates a gene with known or inferred immune function (see text); red lettering indicates an old gene either found in all four human MHC paralogous regions or at least in two paralogous Review 355

lies (discussed below), is that it is difficult to envisage Also, the organization of the MHC has undergone dy- a molecular mechanism that enables sorting of a large namic changes during vertebrate evolution (Figure 3). number of paralogous copies to four specific regions These characteristics may explain why the gene order of the human genome. Hence, the “block duplication” is less conserved on chromosomes 1 and 6. hypothesis has been a preferred explanation in general. Genome Duplication The most decisive difference between the block duplica- A gene, which occurs in a single copy in invertebrates, tion and “functional clustering” hypotheses is that only has often up to four paralogous copies in humans (Spring, the former predicts the existence of a preduplicated 1997). Furthermore, in the human genome, duplicated region. Recent studies are providing evidence that inver- copies are often found in the paralogous regions that tebrates indeed have such a region (see below). There- apparently arose by large-scale block duplication (Lun- fore, the MHC paralogous group seems to have emerged din, 1993). These observations have rekindled great in- primarily by large-scale block duplications, with func- terest in the hypothesis that genome-wide duplications tional clustering presumably playing only a minor role. occurred at least once at the stage of fish or amphibians All jawed vertebrates are equipped with the full- (Ohno, 1970). Genome duplication could be a highly fledged MHC region. Furthermore, there is preliminary efficient way of making complex networks of interacting evidence that the zebrafish genome contains paralo- molecules; thus, Ohno suggested that genome dup- gous regions similar to those in humans (Postlethwait lication by polyploidization played a pivotal role in the et al., 1998). Therefore, the duplications that formed the evolution of vertebrates. The most popular form of the four paralogous clusters appear to have taken place genome duplication hypothesis—modified to accom- before the emergence of jawed vertebrates. Recent evi- modate recent molecular data—is the 2R hypothesis dence indicates that the cephalochordate Amphioxus (reviewed in Ohno, 1999; Wolfe, 2001), which assumes (Figure 1) has only a single cluster that qualifies as a two rounds of allopolyploidization (polyploidization be- precursor of the four paralogous regions (P. Pontarotti, tween two different species), the first round in a common personal communication). Hence, the block duplications ancestor of all vertebrates and the second round in a that formed the MHC paralogous group appear to have common ancestor of jawed vertebrates after its separa- taken place after the emergence of cephalochordates tion from jawless fishes (Figure 1). Because the esti- but before the emergence of a common ancestor of mated timing and number of block duplications that jawed vertebrates. To have four paralogous regions, at formed the MHC paralogous group closely match those least two rounds of duplication are required. Currently, of the 2R hypothesis, it is possible that the MHC region it is not known whether the first round of duplication was formed by two rounds of genome duplication. took place before or after the emergence of jawless Whether the 2R hypothesis is correct is the subject fishes. Preliminary analysis of gene families such as PBX of intense controversy (Wolfe, 2001). The recently com- and tenascin suggests that jawless fish are likely to have pleted human draft sequence has revealed that the hu- at least two paralogous clusters (T. Suzuki and M.K., man genome contains more than 1000 paralogous seg- unpublished data). Thus, it is possible that one round ments of various lengths (Venter et al., 2001). In addition, of duplication took place in a common ancestor of all all of the well-characterized paralogous regions, includ- jawed and jawless vertebrates, and another in a com- ing those coding for HOX clusters, have their origin close mon ancestor of jawed vertebrates after its separation to the origin of vertebrates. Genome duplication is from jawless fishes (Figure 1). clearly an attractive mechanism to account for these Near completion of the human genome project enables observations. However, compared to the genome dupli- us to assess the extent to which the gene order is con- cation events proposed to have taken place in baker’s million years ago), Arabidopsis thallana 100ف) served among the four sets of paralogous regions (Figure yeast -million years ago), and Xenopus spe 180ف and 112ف) -In general, the relative order of the corresponding para .(2 logous copies is poorly conserved, suggesting that, after cies (5–30 million years ago), the duplication events pos- the block duplication, each region underwent major tulated to have taken place in vertebrate ancestors structural rearrangements including inversions and trans- (Ͼ500 million years ago) are much more ancient. This locations. This is not surprising if one considers that has made it difficult to obtain definitive evidence for more than 500 million years have passed since the last genome duplication in the human genome. A method duplication event. Nevertheless, there are copies ar- commonly used to test the 2R hypothesis is phyloge- ranged in the same order between chromosomes. For netic (tree) analysis. This type of analysis assumes that example, 9q31-q34 contains the following genes from if the paralogous regions arose by genome duplication, centromere to telomere: CTL1–TNFSF18/15–C5–DNM1– the divergence time between paralogous copies map- BRD3–NOTCH1. The corresponding paralogous copies ping to different paralogous regions should be more or of these genes are arranged in the same order on 19p13, less identical in all gene families. However, this logic is suggesting that this arrangement might represent the not necessarily correct. First, tandem duplication can primordial one. As described above, there is evidence take place prior to genome duplication. If tandemly du- that chromosome 1 underwent a pericentric inversion. plicated copies were lost differentially from the paralo-

regions and one other species besides human; green lettering denotes a class I/II gene; an asterisk indicates a species-specific feature; a large box indicates that there are at least two genes encoded in that region; a double slash indicates linkage on the same chromosome; a space indicates that the genes are on different linkage groups; Rec is a recombinant. The teleost MHC is a composite of the Fugu and zebrafish maps. Refer to the text for details. Refer to Tables S1–S5 for descriptions of each gene. Immunity 356

gous regions, the estimated divergence time of the para- nature of the MHC has prompted Kaufman to propose logous copies should match that of the tandem but not that chickens have a “minimal essential MHC,” in which the genome duplication. Hence, phylogenetic analysis the complex has been stripped clean of all genes dis- will not allow us to know when the genome duplication pensable to a functioning antigen presentation system took place. Second, if genome duplication occurred by (Kaufman et al., 1995, 1999). Furthermore, this miniature allotetraploidization as suggested by Spring (1997), par- complex may allow for coevolution of all of the linked alogous copies are derived from either the same or dif- genes, e.g., the two C-type lectin genes mentioned ferent parent species. Copies derived from the different above that may encode NK receptors coevolving with parent species should have a longer divergence time the particular class I alleles to which they are linked. than those derived from the same parent species. These Birds have microchromosomes and many other genes, examples illustrate the limitations inherent in the phylo- and gene families show such compactness; thus, it is genetic analysis. Presumably, final resolution of the not clear whether there is some advantage to having a genome duplication hypothesis will require studies of minimal essential MHC or if the organization is a conse- genome architecture in several major classes of verte- quence of another selective pressure (Hughes and brates, including jawless and cartilaginous fish. Hughes, 1995). One can easily make a case for an immune function for each gene in the chicken MHC except Comparative Analysis of the MHC the one encoding RING3, a nuclear kinase containing Besides human and several other mammals, genomic a bromodomain whose specific function is not known (physical mapping) information exists for only two verte- (Thorpe et al., 1996); thus, it will be interesting to deter- brate MHCs, birds (chicken) and bony fish (Fugu or puff- mine whether RING3 is involved in transcriptional con- erfish, zebrafish, and medaka fish). In other vertebrates trol of the MHC (Flajnik et al., 1999a). such as amphibians (Xenopus and axolotl), additional Expression levels vary for different chicken class Ia bony fish (many species owing to their great diversity alleles, and these levels correlate with resistance to Mar- and economic importance), and cartilaginous fish, MHC ek’s disease, a leukemia caused by a herpesvirus (and genes have been isolated, and useful genetic analyses to other viruses as well, reviewed in Kaufman et al., 1995, are available (Figure 3). Studies of MHC organization in 1999). Those chickens with MHC haplotypes resulting in nonmammalian vertebrates indicate that the genes are a lower class Ia expression are most resistant to this not organized in the same way in all animals, not surpris- disease, but it is not known which arm of the immune (or other) system (e.g., T cell, NK cell, virus entry) is most ing since the organization can differ even among mam- affected. Coevolution of the very tightly linked TAPs and malian taxa. Nevertheless, there are common features class Ia genes may dictate which peptides are presented that distinguish the MHCs of nonmammalian vertebrates from a pathogen by particular class Ia alleles, perhaps from those of mammals: (1) in chicken, teleost fish, and similar to what has been shown for the biallelic TAP probably amphibians and sharks, the class Ia genes are system in the rat MHC (Joly and Butcher, 1998; Kaufman, closely linked to LMP and TAP genes in a true “class I 1999). Thus, such a gene arrangement would be inher- region,” indicating that these genes may coevolve in a ently structured to reveal Ir gene effects more often concerted fashion; in general, MHC-linked class Ia than is detected in most mammals where class Ia is genes in these species do not undergo extensive dupli- separated from TAPs by the class III region. cations/deletions, and (2) several immune genes in the The chicken Rfpy complex contains a set of class I human class III region are also present in the amphibian and class II ␤ genes that segregates independently from Xenopus but apparently not at all in the teleost fish, and the MHC proper. This complex is nevertheless located most have been deleted in birds. on the same microchromosome as the MHC (Miller et The Avian MHC al., 1996), and the first Rfpy class I gene sequences The entire chicken MHC has been sequenced, and it indicate that they encode class Ib proteins (Afanassieff consists of only 19 genes in less than 100 kb (Figure 3, et al., 2001); such genes may serve auxiliary functions Kaufman et al., 1999). It is extremely compact, both at to the single expressed polymorphic gene, especially the levels of intron size and intergenic distances. In since the chicken MHC is so compact and the linked contrast to most mammalian MHCs having a large num- genes therein may be functionally inter-dependent, or ber of class I genes, the chicken MHC has just two of the Rfpy gene products may have another function en- them, and only one is highly expressed as protein at the tirely. cell surface. These two class Ia genes sandwich the Preliminary genomic studies have also been per- TAP1 and TAP2 genes in what might be considered a formed in other birds. Species from other taxa may not true class I region. The immunoproteasome genes LMP2 have a minimal essential MHC owing to the suggested and LMP7 (and MECL1) are not in this MHC and probably presence of a higher number of expressed class I/II were deleted from the genome (Kandil et al., 1996); the genes, but more studies are required before definitive lack of immunoproteasome elements may explain the statements can be made (reviewed in Kaufman, 1999). unusual C termini of peptides (bearing negative charges) The Amphibian MHC bound to some chicken class Ia allelic forms (Kaufman The MHCs of two amphibian species, Xenopus and axo- et al., 1995). There is also a single polymorphic class II lotl, have been examined, and each species has unique ␤ molecule expressed; class II ␣ is monomorphic and properties. Xenopus, like the chicken, has a single highly encoded on the same chromosome but far from the expressed class Ia locus, the only detectable class I “compact” MHC. Only the C4 gene remains of the class locus in the MHC (Shum et al., 1993). There are two or III region, as most of the other class III genes have been three ␣ and ␤ class II loci, depending on the species exiled to other parts of the genome. The compressed examined. The class III region immune genes C4, Bf, and Review 357

HSP70 have been mapped to MHC, the oldest vertebrate nome-wide duplications (Wolfe, 2001), there is no doubt studied to date with a mammalian-like structure for this that Xenopus can speciate by allopolyploidization, and region (Figure 3; reviewed in Flajnik et al., 1999a). Very several species exist ranging from the true diploid X. little Xenopus genomic work has been done as the genes tropicalis with 20 chromosomes to the tetraploid X. are generally large, and genomic libraries with large in- laevis with 36 chromosomes to the dodecaploid X. ru- serts have yet to be prepared due to technical problems. wensoriensis with 108 chromosomes (see Nonaka et al., Recombinants detected in Xenopus families suggest 2000). Functional studies carried out over 20 years ago that the single class Ia locus is located between the indicated that the MHC is prone to silencing, with most class II and class III genes, probably in close linkage polyploid species reverting to a disomic inheritance of with TAP1, TAP2, and LMP7 (Nonaka et al., 1997). There MHC (Du Pasquier et al., 1977). This evolutionarily rapid are two ancient lineages of TAP1,2/LMP7/class I alleles diploidization of MHC suggests that expression of too that are not apparent for class II or class III (Nonaka et many MHCs is not advantageous, either because of al., 2000; Flajnik et al., 1999b; Y. Ohta, S. Powis, M. deletion of a large part of the T cell repertoire during Nonaka, and M.F.F., unpublished data); LMP7 and thymic development or perhaps for regulation of NK cell TAP1/2 seem truly biallelic, but within the two class recognition or to maintain a certain density of particular Ia lineages there are many highly diverse alleles. Age class Ia molecules at the surface. We have found that estimates for all of these lineages suggest that they select MHC subregions can be silenced on the various split from a common ancestor near the human/mouse constituting chromosomes during diploidization (brack- divergence 60–80 million years ago. Recombinants be- eted in Figure 3), suggesting that mutations of cis-acting tween TAP1,2/LMP7/class Ia genes from different lin- transcriptional elements might repress whole segments eages in wild-caught animals have never been detected, of the MHC (L. Salter-Cid, M.F.F., M. Nonaka, and L. Du suggesting either that there is a block to recombination Pasquier, unpublished data). between genes in these lineages or that there is a selec- The axolotl, a urodelan amphibian, is notorious for tion against creatures that have recombined TAP/LMP7/ its poor humoral and cell-mediated immune responses. class I (Y. Ohta and M.F.F., unpublished data). Similar The class II proteins, first analyzed by isoelectric focus- to what has been described in chickens, these genes ing and now deduced from cDNA sequences, seem to are likely to be in close linkage and probably constitute be truly biallelic, which is evidently not due to a recent a class I region in which the genes have coevolved bottleneck (Kaufman et al., 1995; Laurens et al., 2001). (Nonaka et al., 1997). In contrast to Xenopus there are many class I genes There is a large cluster of Xenopus class Ib genes that are all apparently linked to class II (Sammut et al., that, like the chicken Rfpy complex, does not segregate 1999). It is appealing to call attention to the low class with the functional MHC. Again, like Rfpy, in situ analysis II polymorphism and the unusual expansion of class I has shown these class Ib genes to be on the same genes as either causes or consequences of the weak chromosome as MHC but half a chromosome away at adaptive responses in axolotls; note that TCR repertoire the telomere (Courtet et al., 2001). In contrast to the diversity seems in line with estimates in most other ver- stable MHC-linked class Ia gene, the numbers of class tebrates (Charlemagne et al., 1998). Unfortunately, the Ib genes vary greatly in different species, probably be- very large genome size of this creature makes physical cause the genes in their telomeric location are subjected analysis of its MHC architecture a daunting task. to rapid changes (L. Du Pasquier and M.F.F., unpub- The MHC of Fishes lished data). These rapidly evolving class Ib genes Teleost (bony) fish are the largest group of vertebrates, (again, like Rfpy) may play roles complementary to the comprising about half of the total living vertebrates. class Ia gene that coevolves with other MHC genes. However, most existing bony fish (Euteleosts) evolved Situating class Ib genes far from MHC proper probably from a common ancestor in a large expansion 150 Mya, minimizes gene exchange with the class Ia gene and and these are the species for which we have acquired hence does not affect coevolution of class Ia with TAP/ data. MHC studies have been carried out in many spe- LMP. cies from several subtaxa with the following common The class Ia gene is expressed ubiquitously in adult features: (1) class Ia and class II genes are found in frogs, and particular class Ib genes are expressed either different linkage groups; (2) the immunoproteasome, ubiquitously or in a tissue-specific fashion (Flajnik et al., TAP, and TAPBP genes are closely linked to class Ia, 1987; Salter-Cid et al., 1998). Tadpoles do not express comprising a true class I region as suggested above for class I genes of either type or LMP7 at high levels, with chickens and amphibians; (3) some nonimmune genes induction of high expression at the metamorphic climax. found in the human class III region and in the extended Class II molecules are expressed in larvae with a distri- class II region are also linked to class I, but none of the bution similar to mammals, on B cells and APC; after class III immune genes is linked to class I or class II metamorphosis, class II becomes expressed on T cells (Figure 3). Hard core genomic studies have been done as well. This differential regulation of MHC genes during in the Japanese pufferfish Fugu rubripes (Clark et al., the two immunocompetent lives of the frog likely im- 2001), the zebrafish Danio rerio (Michalova´ et al., 2000; pacts T cell and NK cell repertoires in ontogeny, a specu- Su¨ ltmann et al., 2000), and medaka fish (Naruse et al., lation that has some experimental support (Horton et 2000). The genomes of these species are compact, and al., 1998). The transcriptional regulation of class I/class in Fugu, a contig BAC map of 300 kb encompassing the II in ontogeny has not been studied, but it is attractive entire MHC was analyzed (Clark et al., 2001, Figure 3). to suggest that RXRB may somehow be involved. Like in Xenopus and chicken, at least in the trout it While as described above, it is controversial whether was shown definitively that there is only a single ex- the vertebrate genome underwent two rounds of ge- pressed class Ia gene, which also falls into ancient lin- Immunity 358

eages (Hansen et al., 1996; Shum et al., 2001). It will tions should have a single precursor region (Figure 1). be interesting to determine whether such old class Ia Recent work has provided convincing evidence that the lineages also will be detected in most other fish, but genome of Amphioxus indeed contains such a region now this would not be surprising. In contrast, unlinked (P. Pontarotti, personal communication). For example, trout and salmon class II genes do not form old lineages Amphioxus has a single RXR gene that qualifies as a and change rapidly over evolutionary time more like common ancestor of three vertebrate RXR genes. The mammalian class Ia genes; i.e., the class II alleles exam- cosmid clone containing Amphioxus RXR contains three ined are species specific (Shum et al., 2001). In cichlids, other genes, one resembling NEU1 mapping to the HLA variable numbers of class II genes are found in different class III region, another resembling a human EST se- haplotypes, again suggesting rapid evolution (Malaga- quence mapping to 9q34.11, and the third resembling Trillo et al., 1998). Such results show that the fragmented human genes mapping to 1p31.1. teleost MHC allows class I and class II genes to evolve Studies of other invertebrates indicate that the linkage independently, presumably under different selection of some MHC genes is much more ancient than the pressures. Having unlinked class I and class II genes origin of cephalochordates. For example, the human may be the result of differential diploidization after a MHC encodes C4, Bf, C2, PBX2, NOTCH4, RXRB, and polyploidization event proposed to have occurred spe- TNX (Figures 2 and 3). In the sea urchin Strongylocentro- cifically in a Euteleost ancestor (Amores et al., 1998). tus purpuratus,aC3-like gene (related to the ancestor While most species have relatively few class I genes in of three vertebrate paralogs C3, C4, and C5) and a Bf- the MHC, the cod (Persson et al., 1999) and cichlid like gene (related to the common ancestor of Bf and (Murray et al., 2000) have many class I genes, but it is C2) are located within a distance of 140 kb, and genes not yet known whether these are linked to TAP/LMP or coding for NOTCH and PBX (again, both related to the are in another linkage group as in chicken and Xenopus. ancestral form of the four paralogs) are within 680 kb Perhaps not coincidentally, the cod, like the axolotl that (Rast et al., 2000). Furthermore, genes coding for the also has a large number of MHC-linked class I genes, RXR-, PBX-, NOTCH-, and TN-like molecules are located has a miserable adaptive immune response. within a distance of 5.8 megabases in the nematode In Fugu and zebrafish, at least one class I gene is Caenorhabditis elegans (Trachtulec et al., 1997), sug- physically associated with several immunoproteasome gesting that a primitive form of the class III region exists genes and the TAP2 gene. A second LMP2 gene, inter- even in protostomes. mediate in sequence from LMP2 found in all other verte- The Jawed Vertebrate (Postduplication) Ur MHC brates and the constitutive proteasome element ␦,is Although we are still at the beginning of our analysis of present in all teleosts so far studied (Hansen et al., 1999). genomic structure of MHC in nonmammalian verte- The third ␥ interferon-inducible immunoproteasome brates, at least we can make models for its origins based gene, MECL1, exiled from MHC in mammals, is in the on the comparative studies and the genes found in the proteasome/class Ia cluster (Clark et al., 2000), a result MHC-paralogous groups (Figure 3). MHC organizations strongly supporting the en bloc duplication model de- in representative classes of vertebrates are largely con- scribed above. There are several other fish genes that sistent with the assumption that the preduplicated re- in humans map to the extended class II region and even gion was made up of the precursors of the members of a few nonimmune class III genes (Su¨ ltmann et al., 2000; the MHC paralogous group. For example, RING3, PSMB Figure 3). (LMP), BAT2, PBX, and RXR are members of the MHC Since class Ia and class II genes are not linked in paralogous group. Thus, one can expect that these teleost fish it is possible that they arose on different genes were present in the Ur MHC and that the MHCs chromosomes (the ancient paralogs?) and then were of some jawed vertebrates should contain related genes, “attracted together” to form MHC in a tetrapod ancestor which is indeed the case with the teleost MHC. The (Klein and Sato, 1998). However, studies in the oldest former two genes are also encoded in the Xenopus MHC extant jawed vertebrates, the cartilaginous fish, showed (Flajnik et al., 1999a; Figure 3). These findings indicate linkage of class Ia and class II genes in two divergent that systematic identification of the members of the shark species, suggesting that the situation in teleosts MHC paralogous group is a powerful means of deducing is derived (Ohta et al., 2000). Further linkage studies in the primordial MHC organization. This information in sharks showed that TAP and LMP genes are also MHC turn enables us to deduce the major structural changes linked and that class Ia genes are closely linked to immu- that have taken place in the MHCs during vertebrate noproteasome gene fragments, suggesting that a class evolution (Figure 3). I region also will be found in this group (Y. Ohta, M. In the ancestor of all jawed vertebrates, class II and Criscitiello, E.C. McKinney, and M.F.F., submitted). Like class I presumably arose by cis-duplication from an an- the teleosts, an additional immunoproteasome gene cestor having a C1-set Ig domain and a PBR, but this (LMP7-like) is present in cartilaginous fish (Kandil et al., synteny was broken in the teleosts (which may actually 1996). Thus far, class Ib genes that are related to the have been to their advantage, allowing class II genes MHC-linked class Ia genes are not closely linked to the to evolve at a faster pace when not closely linked to MHC, although there are several divergent class I fami- genes like TAP/LMP that evolve conservatively; Klein lies currently being studied (Bartl, 1998). et al., 1992.) Based upon phylogenetic analyses, it is “MHC” in Creatures without an Adaptive expected that the DM class II genes arose around the Immune System same time that class I and class II diverged from a The idea that the four paralogous regions arose as a common ancestor (Kasahara et al., 1995; Figure 3) and result of block duplication implies that animals derived thus were in the Ur MHC and could be found in all jawed from ancestors that have not experienced such duplica- vertebrates. The discoveries of the three inducible, im- Review 359

munoproteasome genes in the teleost MHC (Clark et al., have been a “primordial immune complex” that gov- 2000), coupled with the presence of the constitutive erned some responses in animals relying solely on non- genes in the Amphioxus proto-MHC, strongly support adaptive immunity (Salter-Cid and Flajnik, 1995). As de- the en bloc duplication model. Immune-related genes in scribed, genes encoding a C3-like molecule, the central the class III region either were present and subsequently player in the complement cascade, and its structurally lost in the bony fish or were drawn to MHC in a functional unrelated partner Bf are closely linked in the predupli- cluster in a tetrapod ancestor; the ancient paralogs cated sea urchin “MHC” (Rast et al., 2000). Evidence would predict that C3/C4/C5 and TNFSF genes indeed from the MHC paralogs strongly suggests that genes were present in the ancestral MHC, and thus the first encoding TNFSF members, a toll receptor, cathepsins possibility is most likely. Class Ia was probably closely (Table S4) and other genes involved in proinflammatory linked to TAP/LMP/TAPBP genes in a true, functional responses and immunoregulation described in the preclass I region in the most primitive ancestor with an ceding section were also present. Thus it is possible adaptive immune system (as well as to class II) but lost that large syntenic blocks of genes encoding structurally its close association to these genes in a mammalian unrelated proteins involved in innate immunity were in- ancestor—it was then capable of rapid evolution and deed functionally clustered (Hughes, 1998) in the proto consequently relinquished its “need” to coevolve with MHC region, perhaps permitting global transcriptional LMP/TAP (Nonaka et al., 1997). Perhaps Xenopus and control over this battery of dangerous immune mole- chicken will be models revealing how a second cluster cules (see Volpi et al., 2000). Inconsistent with this of rapidly evolving class Ib genes can complement the model, vertebrate C3 serum protein is present at high stable classical class I gene. constitutive levels; however, in the sea urchin, C3 is essentially not expressed in quiescent animals but can Role of Block Duplications in the Emergence be induced with a stimulus such as lipopolysaccharide of Adaptive Immunity (Clow et al., 2001). Thus, if such transcriptional regula- As discussed above, current evidence indicates that the tion already had been imposed on the preduplicated prototypes of the four paralogous regions, including the MHC (e.g., through TNF or type I interferons), genes MHC itself, emerged for the first time in a common an- involved in adaptive immunity after their emergence cestor of jawed vertebrates after its separation from could have “piggy-backed” onto a similar regulation. jawless vertebrates. Thus, the genome of jawless verte- Vaux et al. (1994) theorized that CTL killing mecha- brates is unlikely to contain a full-fledged MHC region, nisms evolved because viruses “learned” to block cell- consistent with the observation that the adaptive im- autonomous apoptotic pathways induced by viral infec- mune system characterized by MHC/Ig/TCR has been tion, the primitive manner by which cells protected identified only in jawed vertebrates (Figure 1). The MHC themselves from intracellular pathogens. This is a clever paralogous group encodes several accessory mole- idea made all the more plausible in recent years by cules of the adaptive immune system, such as LMP, studies demonstrating the seemingly endless number VAV1 (involved in lymphocyte development/activation; of mechanisms that viruses have evolved to subvert Cantrell, 1998), NOTCH1 (involved in T cell differentia- immune function (Alcami and Koszinowski, 2000). How- ever, the primeval killer cells need not have been bona tion; Robey, 1997), C4, RXRB, and the OX40 ligand (Fig- fide CTL but instead NK-like cells that monitored the ure 2 and Tables S1–S4). Because all of these molecules cell surface for up and downregulation of self-mole- presumably emerged as a result of the block duplica- cules, much like modern-day NK cells; the NK surveil- tion event, it is likely that this event contributed to the lance system in mammals is astonishingly sophisticated emergence of the adaptive immune system. Of particu- (Trowsdale, 2001), and one can imagine a similar innate lar interest is the fact that the gene coding for a class arsenal before the introduction of adaptive immunity. I-like molecule of innate immunity, CD1, maps to 1q22- Vaux et al.’s theory infers that some self-molecule moni- q23. This raises the interesting possibility that an MHC toring the internal integrity of the cell must have evolved class I molecule, a key player of adaptive immunity, as a signal to alert killer cells. The specialized Ig super- emerged from its common ancestor with CD1 only after family C1 domain is found in the antigen receptors (Ig, the block duplication. It has been suggested (Abi- TCR), their ligands class I and class II, and in the MHC- Rached and Pontarotti, 2000) that “relaxation of func- linked TAPBP and butyrophylin, but in very few other tional constraints” on duplicates of the MHC precursor proteins. Du Pasquier (2000) has speculated that ances- region may have allowed rapid evolution of new func- tors of the rearranging antigen receptor genes were tions by the paralogs leading to the initiation of adaptive once encoded in the MHC as well, perhaps as a hypo- immunity. thetical nonrearranging NK-like receptor having a V-C1 structure. Such receptors may have originally scanned Speculations on MHC Origins for C1-type molecules induced by the stress of infection, It is most fun to reflect on the origins of MHC (the predu- like recognition of the stress-induced human class Ib plicated ancestral MHC) before the establishment of MIC proteins today (Groh et al., 2001). The recognition adaptive immunity. It is possible that genes in the origi- itself may have been similar to interaction of the extant nal “MHC,” before the emergence of Ig/TCR/class I/II, CD8 molecule with the class I C1-type Ig domain. were syntenic purely by chance and that coordination But what do we make of the proteasome connection; of immune functions arose only after the en bloc duplica- i.e., why were constitutive proteasome elements linked tions described above (Kasahara et al., 1997). Alterna- to C3/Bf, TNFSF, and IgSF genes before the emergence tively, the presence of genes encoding essential players of adaptive immunity? Since constitutive proteasomes in innate immunity suggests that the original MHC may are expressed at high levels in all living cells, it is difficult Immunity 360

to envision that their simple upregulation had much to Afanassieff, M., Goto, R.M., Ha, J., Sherman, M.A., Zhong, L., Auf- do with immune responses—however, proteasomes fray, C., Courdert, F., Zoorob, R., and Miller, M.M. (2001). At least also are known to be involved in control of cell cycle/ one class I gene in Restriction fragment pattern-Y (Rfp-Y), the second MHC gene cluster in the chicken, is transcribed, polymorphic, transcription/apoptosis in stressed cells, so the associa- and shows divergent specialization in antigen binding region. J. tion of genes encoding the proteolytic proteasome sub- Immunol. 166, 3324–3333. units with genes involved in innate immunity may have Agrawal, A., Eastman, Q.M., and Schatz, D.G. (1998). Transposition been advantageous in ways that are not obvious. Exons mediated by RAG1 and RAG2 and its implications for the evolution encoding PBR of a “chaperone” may have been trans- of the immune system. Nature 394, 744–751. ferred to a gene encoding a C1 domain so that a “schlep- Amores, A., Force, A., Yan, Y.-L., Joly, L., Amemiya, C., Fritz, A. Ho, per” molecule could be anchored in the membrane. R.K., Langeland, J., Prince, V., Wang, Y.-W., et al. (1998). Zebrafish Such a proto-MHC molecule, also upregulated in times hox clusters and vertebrate genome evolution. Science 282, 1711– 1714. of stress, perhaps bound to arrays of peptides in virally infected cells that could either lead to NK-like cell lysis Bartl, S. (1998). What sharks can tell us about the evolution of MHC genes. Immunol. Rev. 166, 317–331. by a failure to be recognized by inhibitory receptors Bernstein, R.M., Schluter, S.F., Bernstein, H., and Marchalonis, J.J. (say, if viral peptides were predominantly bound) or rec- (1996). Primordial emergence of the recombination activating gene ognition by activating receptors (say, if peptides from 1 (RAG1): sequence of the complete shark gene indicates homology upregulated self-stress-related proteins were predomi- to microbial integrases. Proc. Natl. Acad. Sci. USA 93, 9454–9459. nantly bound). Note that the proto-MHC molecule itself Cantrell, D. (1998). Lymphocyte signaling: a coordinating role for would not have to be more like class I or class II structur- Vav? Curr. Biol. 8, 535–538. ally, but its function and tissue distribution are obviously Charlemagne, J., Fellah, J.S., De Guerra, A., Kerfourn, F., and Par- proposed to be class I-like—thus, all cells of the body tula, S. (1998). T-cell receptors in ectothermic vertebrates. Immunol. could be protected from viral attack. It is hoped that Rev. 166, 87–102. studies of MHC-like regions in prejawed deuterostomes Clark, M.S., Pontarotti, P., Gilles, A., and Elgar, G. (2000). Identifica- ␤ (Figure 1) may yield class I-like molecules or better yet tion and characterization of a proteasome subunit in the Japanese pufferfish (Fugu rubripes). J. Immunol. 165, 4446–4452. PBR candidates that predated class I/class II (in fact, Clark, M.S., Shaw, L., Kelly, A., Snell, P., and Elgar, G. (2001). Charac- they may already be there in the paralogous regions, but terization of the MHC class I region of the Japanese pufferfish (Fugu we do not recognize them; Abi-Rached and Pontarotti, rubripes). Immunogenetics 52, 174–185. 2000). Courtet, M., Flajnik, M.F., and Du Pasquier, L. (2001). Major histo- As a last step, once the preclass I molecule’s function compatibility complex and immunoglobulin loci visualized by in situ was somehow tied into that of the proteasome, fortu- hybridization on Xenopus chromosomes. Dev. Comp. Immunol. 25, itous events like the postulated en bloc duplications 149–157. early in vertebrate evolution to generate the immunopro- Clow, L.A., Gross, P.S., Shih, C.S., and Smith, L.C. (2001). Expression teasome elements and generally expand the repertoire of SpC3, the sea urchin complement component, in response to of genes involved in defense, and the insertion of a lipopolysaccharide. Immunogenetics 51, 1021–1033. transposable element into one of these V-C1 NK-like Dey, A., Thornton, A.M., Lonergan, M., Weissman, S.M., Chamber- lain, J.W., and Ozato, K. (1992). Occupancy of upstream regulatory receptor genes to provide a mechanism to generate sites in vivo coincides with MHC class I gene expression in mouse somatic diversity in antigen receptor genes, may have tissues. Mol. Cell. Biol. 12, 3590–3599. set off the specific responses that characterize adaptive Davis, M.M., and Bjorkman, P.J. (1988). T-cell antigen receptor immunity. If this scenario has any validity, i.e., if peptides genes and T-cell recognition. Nature 334, 395–402. played a role in recognition by protoantigen receptors, Du Pasquier, L. (2000). The phylogenetic origin of antigen-specific it would support the original Bjorkman/Davis scenario receptors. Curr. Top. Microbiol. Immunol. 248, 160–185. of having the rearrangement break in antigen receptor Du Pasquier, L., Miggiano, V.C., Kobel, H.-R., and Fischberg, M. genes in CDR3 (i.e., in the center of a dimeric antigen (1977). The genetic control of histocompatibility reactions in natural receptor). In the next few years, comprehensive studies and laboratory-made polyploid individuals of the clawed toad, Xeno- of the preduplicated “MHC” in jawless vertebrates and pus. Immunogenetics 5, 129–141. invertebrates may reveal whether the models proposed Flajnik, M.F., Hsu, E., Kaufman, J.F., and Du Pasquier, L. (1987). above have any substance. Changes in the immune system during metamorphosis of Xenopus. Immunol. Today 8, 52–64. Flajnik, M.F., Canel, C., Kramer, J., and Kasahara, M. (1991). Which Supplemental Data came first, class I or class II? Immunogenetics 33, 295–300. Additional Tables (S1–S5) showing the paralogous copies of genes Flajnik, M.F., Ohta, Y., Namikawa-Yamada, C., and Nonaka, M. on human chromosomes 1, 6 (MHC), 9, and 19 are available online (1999a). Insight into the primordial MHC from studies in ectothermic at http://www.immunity.com/cgi/content/full/15/3/351/DC1. vertebrates. Immunol. Rev. 167, 59–67. Flajnik, M.F., Ohta, Y., Greenberg, A.S., Salter-Cid, L., Carrizosa, References A., Du Pasquier, L., and Kasahara, M. (1999b). Two ancient allelic lineages at the single classical class I locus in the Xenopus MHC. Abi-Rached, L., and Pontarotti, P. (2000). The MHC “Big-Bang”: J. Immunol. 163, 3826–3833. duplication and exon shuffling during chordate evolution. A hypo- Groh, V., Rhinehart, R., Randolph-Habecker, J., Topp, M.S., Riddell, thetico-deductive approach. In Major Histocompatibility Complex: S.R., and Spies, T. (2001). Costimulation of CD8 ␣/␤ T cells by Evolution, Structure, and Function, M. Kasahara, ed. (Tokyo, NKG2D via engagement by MIC induced on virus-infected cells. Springer-Verlag), pp. 45–52. Nat. Immunol. 2, 255–260. Abi-Rached, L., McDermott, M.F., and Pontarotti, P. (1999). The Gruen, J.R., and Weissman, S.M. (1997). Evolving views of the major MHC Big Bang. Immunol. Rev. 167, 33–44. histocompatibility complex. Blood 90, 4252–4265. Alcami, A., and Koszinowski, U.H. (2000). Viral mechanisms of im- Hansen, J.D., Strassburger, P., and Du Pasquier, L. (1996). Conser- mune evasion. Trends Microbiol. 8, 410–418. vation of an alpha 2 domain within the teleostean world: MHC class Review 361

I from the rainbow trout Oncorhynchus mykiss. Dev. Comp. Immunol. in paralogous chromosomal regions in man and the house mouse. 20, 417–425. Genomics 16, 1–19. Hansen, J.D., Strassburger, P., Thorgard, G.H., Young, W.P., and The M.H.C. Sequencing Consortium (1999). Complete sequence and Du Pasquier, L. (1999). Expression, linkage, and polymorphisms gene map of a human major histocompatibility complex. Nature 401, of MHC-related genes in rainbow trout, Oncorhynchus mykiss.J. 921–923. Immunol. 163, 774–786. Malaga-Trillo, E., Zaleska-Rutczynska, Z., McAndrew, B., Vincek, V., Horton, J.D., Horton, T.L., Dzialo, R., Gravenor, I., Minter, R., Ritchie, Figueroa, F., Su¨ ltmann, H., and Klein, J. (1998). Linkage relationships P., Gartland, L., Watson, M.D., and Cooper, M.D. (1998). T-cell and and haplotype polymorphism among cichlid Mhc class II B loci. natural killer cell development in thymectomized Xenopus. Immunol. Genetics 149, 1527–1537. Rev. 166, 245–258. Michalova´ , V., Murray, B.W., Su¨ ltmann, H., and Klein, J. (2000). A Hughes, A.L. (1998). Phylogenetic tests of the hypothesis of block contig map of the Mhc class I genomic region in the zebrafish reveals duplication of homologous genes on human chromosomes 6,9,and ancient synteny. J. Immunol. 164, 5296–5305. 1. Mol. Biol. Evol. 15, 854–870. Miller, M.M., Goto, R.M., Taylor, R.L., Zoorob, R., Auffray, C., Briles, Hughes, A.L., and Hughes, M.K. (1995). Small genomes for better R.W., Briles, W.E., and Bloom, S.E. (1996). Assignment of Rfp-Y to flyers. Nature 377, 391. the chicken MHC/NOR microchromosome and evidence for high- Hughes, A.L., and Yeager, M. (1997). Molecular evolution of the frequency recombination associated with the nucleolar organizer vertebrate immune system. BioEssays 19, 777–786. region. Proc. Natl. Acad. Sci. USA 93, 3958–3962. Joly, E., and Butcher, G.W. (1998). Why are there two rat TAPs? Murray, B.W., Nilsson, P., Zaleska-Rutczynska, Z., Su¨ ltmann, H., Immunol. Today 19, 580–585. and Klein, J. (2000). Linkage relationships and haplotype variation of the MHC class I A genes in the cichlid fish Oreochromis niloticus. Kandil, E., Namikawa, C., Nonaka, M., Greenberg, A.S., Flajnik, M.F., Mar. Biotechnol. 2, 437–448. Ishibashi, T., and Kasahara, M. (1996). Isolation of low molecular mass polypeptide complementary DNA clones from primitive verte- Nadeau, J.H., and Kosowsky, M. (1991). Mouse map of paralogous brates: implications for the origin of MHC class I-restricted antigen genes. Mamm. Genome 1, S433–S460. presentation. J. Immunol. 156, 4245–4253. Naruse, K., Shima, A., and Nonaka, M. (2000). MHC gene organization in the bony fish, medaka. In Major Histocompatibility Complex: Kasahara, M. (1999). The chromosomal duplication model of the Evolution, Structure, and Function, M. Kasahara, ed. (Tokyo: major histocompatibility complex. Immunol. Rev. 167, 17–32. Springer-Verlag), pp. 91–109. Kasahara, M., Flajnik, M.F., Ishibashi, T., and Natori, T. (1995). Evolu- Nonaka, M., Namikawa, C., Kato, Y., Sasaki, M., Salter-Cid, L., and tion of the major histocompatibility complex: a current overview. Flajnik, M.F. (1997). Major histocompatibility complex gene mapping Transpl. Immunol. 3, 1–20. in the amphibian Xenopus implies a primordial organization. Proc. Kasahara, M., Hayashi, M., Tanaka, K., Inoko, H., Sugaya, K., Ikem- Natl. Acad. Sci. USA 94, 5789–5791. ura, T., and Ishibashi, T. (1996). Chromosomal localization of the Nonaka, M., Yamada-Namikawa, C., Flajnik, M.F., and Du Pasquier, proteasome Z subunit gene reveals an ancient chromosomal dupli- L. (2000). Trans-species polymorphism of major histocompatibility cation involving the major histocompatibility complex. Proc. Natl. complex-encoded proteasome subunit LMP7 in an amphibian ge- Acad. Sci. USA 93, 9096–9101. nus, Xenopus. Immunogenetics 51, 186–192. Kasahara, M., Nakaya, J., Sata, Y., and Takahata, N. (1997). Chromo- Ohno, S. (1970). Evolution by Gene Duplication. (New York: Springer- somal duplication and the emergence of the adaptive immune sys- Verlag). tem. Trends Genet. 13, 90–92. Ohno, S. (1999). Gene duplication and the uniqueness of vertebrate Katsanis, N., Fitzgibbon, J., and Fisher, E.M. (1996). Paralogy map- genomes circa 1970–1999. Sem. Cell Devel. Biol. 10, 517–522. ping: identification of a region in the human MHC triplicated onto human chromosomes 1 and 9 allows the prediction and isolation Ohta, Y., Okamura, K., McKinney, E.C., Bartl, S., Hashimoto, K., and of novel PBX and NOTCH loci. Genomics 35, 101–108. Flajnik, M.F. (2000). Primitive synteny of vertebrate major histocompatibility complex class I and class II genes. Proc. Natl. Acad. Sci. Kaufman, J. (1989). Vertebrates and the evolution of the major histo- USA 97, 4712–4717. compatibility complex (MHC) class I and class II molecules. Verh. Dtsch. Zool. Ges. 81, 131–144. Persson, A.C., Stet, R.J., and Pilstro¨ m, L. (1999). Characterization of MHC class I and beta2-microglobulin sequences in Atlantic cod Kaufman, J. (1999). Co-evolving genes in MHC haplotypes: the “rule” reveals an unusually high number of expressed class I genes. Immu- for non-mammalian vertebrates? Immunogenetics 50, 228–236. nogenetics 50, 49–59. Kaufman, J., Vo¨ lk, H., and Wallny, H.-J. (1995). A “minimal essential Postlethwait, J.H., Yan, Y.L., Gates, M.A., Horne, S., Amores, A., Mhc” and an “unrecognized MHC”: Two extremes in selection for Brownie, A., Donovan, A., Egan, E.S., Force, A., Gong, Z., et al. polymorphism. Immunol. Rev. 143, 63–88. (1998). Vertebrate genome evolution and the zebrafish gene map. Kaufman, J., Milne, S., Go¨ bel, T.W.F., Walker, B.A., Jacob, J.P., Nat. Genet. 18, 345–349. Auffray, C., Zoorob, R., and Beck, S. (1999). The chicken B locus is Rast, J.P., Oliveri, P., and Davidson, E.H. (2000). Conserved linkage a minimal essential major histocompatibility complex. Nature 401, among sea urchin homologs of genes encoded in the vertebrate 923–925. MHC region. In Major Histocompatibility Complex: Evolution, Struc- Klein, J., and Sato, A. (1998). Birth of the major histocompatibility ture, and Function, M. Kasahara, ed. (Tokyo: Springer-Verlag), pp. complex. Scand. J. Immunol. 47, 199–209. 66–74. Klein, J., Ono, H., and Klein, D. and O’hUigin, C. (1992). The accor- Richards, M.H., and Nelson, J.L. (2000). The evolution of vertebrate dion model of MHC evolution. In Progress in Immunology Vol. VIII, antigen receptors: a phylogenetic approach. Mol. Biol. Evol. 17, J. Gergely et al., eds., (Budapest: Springer-Verlag) pp. 137–143. 146–155. Kumar, S., and Hedges, S.B. (1998). A molecular timescale for verte- Robey, E. (1997). Notch in vertebrates. Curr. Opin. Genet. Dev. 7, brate evolution. Nature 392, 917–920. 551–557. Laird, D.J., De Tomaso, A.W., Cooper, M., and Weissman, I.L. (2000). Salter-Cid, L., and Flajnik, M.F. (1995). Evolution and developmental 50 million years of chordate evolution: seeking the origins of adap- regulation of the major histocompatibility complex. Crit. Rev. Immu- tive immunity. Proc. Natl. Acad. Sci. USA 97, 6924–6926. nol. 15, 31–75. Laurens, V., Chapusot, C., del Rosario Ordon˜ ez, M., Bentrari, F., Salter-Cid, L., Nonaka, M., and Flajnik, M.F. (1998). Expression of Padros, M.R., and Tournefier, A. (2001). Axolotl MHC class II beta MHC class Ia and class Ib during ontogeny: high expression in chain: predominance of one allele and alternative splicing of the epithelia and coregulation of class Ia and lmp7 genes. J. Immunol. beta1 domain. Eur. J. Immunol. 31, 506–515. 160, 2853–2861. Lundin, L.G. (1993). Evolution of the vertebrate genome as reflected Sammut, B., Du Pasquier, L., Ducoroy, P., Laurens, V., Marcuz, A., Immunity 362

and Tournefier, A. (1999). Axolotl MHC architecture and polymorphism. Eur. J. Immunol. 29, 2897–2907. Shum, B.P., Avila, D., Du Pasquier, L., Kasahara, M., and Flajnik, M.F. (1993). Isolation of a classical MHC class I cDNA from an amphibian: evidence for only one class I locus in the Xenopus MHC. J. Immunol. 151, 5376–5386. Shum, B.P., Guethlein, L., Flodin, L.R., Adkison, M.A., Hedrick, R.P., Nehring, R.B., Stet, R.J.M., Secombes, C., and Parham, P. (2001). Modes of salmonid MHC class I and II evolution differ from the primate paradigm. J. Immunol. 166, 3297–3308. Spring, J. (1997). Vertebrate evolution by interspecific hybridization—are we polyploid? FEBS Lett. 400, 2–8. Su¨ ltmann, H., Sato, A., Murray, B.W., Takesaki, N., Geisler, R., Rauch, G.-J., and Klein, J. (2000). Conservation of Mhc class III region synteny between zebrafish and human as determined by radiation hybrid mapping. J. Immunol. 165, 6984–6993. Thorpe, K., Abdullah, S., Kaufman, J., and Beck, S. (1996). Phylogeny and structure of the RING3 gene. Immunogenetics 44, 391–396. Trachtulec, Z., Hamvas, R.M.J., Forejt, J., Lehrach, H.R., Vincek, V., and Klein, J. (1997). Linkage of TATA-binding protein and proteasome subunit C5 genes in mice and humans reveals synteny conserved between mammals and invertebrates. Genomics 44, 1–7. Trowsdale, J. (2001). Genetic and functional relationships between MHC and NK genes. Immunity, in press. Vaux, D.L., Haecker, G., and Strasser, A. (1994). An evolutionary perspective on apoptosis. Cell 76, 777–779. Venter, J.C., et al. (2001). The sequence of the human genome. Science 291, 1304–1351. Volpi, E.V., Chevret, E., Jones, T., Vatcheva, R., Williamson, J., Beck, S., Campbell, R.D., Goldsworthy, M., Powis, S.H., Ragoussis, J., et al. (2000). Large-scale chromatin organization of the major histocompatibility complex and other regions of human chromosome 6 and its response to interferon in interphase nuclei. J. Cell Sci. 113, 1565–1576. Wolfe, K.H. (2001). Yesterday’s polyploids and mystery of diploidization. Nat. Rev. Genet. 2, 333–341. Yokoyama, W.M. (1998). Natural killer cell receptors. Curr. Opin. Immunol. 10, 298–305.