Hyperconservation of the Putative Antigen Recognition Site of the MHC Class I-b Molecule TL in the Subfamily : Evidence That Thymus Leukemia Antigen Is This information is current as an Ancient Mammalian Gene of September 24, 2021. Beckley K. Davis, Richard G. Cook, Robert R. Rich and John R. Rodgers J Immunol 2002; 169:6890-6899; ; doi: 10.4049/jimmunol.169.12.6890 Downloaded from http://www.jimmunol.org/content/169/12/6890

References This article cites 57 articles, 20 of which you can access for free at: http://www.jimmunol.org/ http://www.jimmunol.org/content/169/12/6890.full#ref-list-1

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists by guest on September 24, 2021

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2002 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Hyperconservation of the Putative Antigen Recognition Site of the MHC Class I-b Molecule TL in the Subfamily Murinae: Evidence That Thymus Leukemia Antigen Is an Ancient Mammalian Gene1

Beckley K. Davis, Richard G. Cook, Robert R. Rich,2 and John R. Rodgers3

“Classical” MHC class I (I-a) genes are extraordinarily polymorphic, but “nonclassical” MHC class I (I-b) genes are monomorphic or oligomorphic. Although diversifying (positive) Darwinian selection is thought to explain the origin and maintenance of MHC class I-a polymorphisms, genetic mechanisms underlying MHC class I-b evolution are uncertain. In one extreme model, MHC class I-b loci are derived by gene duplication from MHC class I-a alleles but rapidly drift into functional obsolescence and are eventually deleted. In this model, extant MHC class I-b genes are relatively young, tend to be dysfunctional or pseudogenic, and orthologies are restricted to close taxa. An alternative model proposed that the mouse MHC class I-b gene thymus leukemia Ag (TL) arose Downloaded from ϳ100 million years ago, near the time of the mammalian radiation. To determine the mode of evolution of TL, we cloned TL from genomic DNA of 11 species of subfamily Murinae. Every sample we tested contained TL, suggesting this molecule has been maintained throughout murine evolution. The sequence similarity of TL orthologs ranged from 85–99% and was inversely pro- portional to taxonomic distance. The sequences showed high conservation throughout the entire extracellular domains with exceptional conservation in the putative Ag recognition site. Our results strengthen the hypotheses that TL has evolved a spe-

cialized function and represents an ancient MHC class I-b gene. The Journal of Immunology, 2002, 169: 6890–6899. http://www.jimmunol.org/

he origin and function of MHC-linked class I-b genes Obata et al. (7) studied seven alleles of TL from inbred strains remains unresolved. One view suggests they derive from of musculus and reported it is not closely related to other T “classical” class I-a MHC genes (1), the chief presenters mouse class I genes. They suggested TL originated at or before the of Ag peptides to T cells. Class I-a molecules exhibit extraordinary time of the mammalian radiation (ϳ100 million years ago polymorphism, apparently maintained by positive Darwinian se- (MYA)). Moreover, the ratio of nonsynonymous (dN) to synony- 4 Ͻ lection within the Ag recognition site (ARS) for the ability to mous (dS) substitutions in the putative ARS was 1 and not higher present diverse pathogen peptides (2). In contrast, class I-b genes than for non-ARS residues in TL, consistent with negative or neu- are monomorphic or oligomorphic. Hughes and Nei (1) suggest tral selection (7). The possibility of negative (purifying) selection by guest on September 24, 2021 that many mouse class I-b genes, including thymus leukemia Ags operating on TL, coupled with its apparently ancient origin, sug- (TL), do not have human orthologs but rather were more closely gested that TL may have evolved a specialized and highly con- related to mouse class I-a genes. They proposed that mouse class served function early in mammalian evolution. I-b genes were derived by gene duplication from mouse I-a alleles TLs are encoded by certain MHC class I-b genes within the T but then rapidly drifted into pseudogene status. Indeed, many class subregion (8). They are ϳ45-kDa cell surface glycoproteins non- ␤ ␤ I-b genes are poorly expressed in most cell types, either because of covalently associated with 2-microglobulin ( 2m) (3) and are ex- low transcription or protein instability, and some are clearly pseu- pressed on activated T cells (9), developing thymocytes and small dogenes (3). Moreover, some mouse class I-b genes such as Qa-2 intestinal epithelium, and intraepithelial lymphocytes (10, 11) and (4) and H2-B1 (5) still appear to be more closely related to mouse certain leukemias (12). Cell surface expression of TL is TAP-in- class I-a genes than to any rat MHC gene. This process is consis- dependent (13–15). The observations that TL is expressed at a site tent with Ohno’s (6) model of gene duplication, in which one enriched for ␥␦ T cells and is oligomorphic led to the hypothesis duplicated copy is released from selection pressure and usually that TL presents conserved Ags to ␥␦ T cells in the gut (10, 16, degenerates into a pseudogene. 17). However, no peptides or motifs have been characterized (13, 18), leaving open the possibility that TL does not present peptide. Unlike H2-M3 and Qa-1 (19–21), TL orthologs have not been reported in other species (22, 23). Obata et al. (7) suggested TL is Department of Immunology, Baylor College of Medicine, Houston, TX 77030 not closely related to other mouse H2 genes but their phylogenetic Received for publication August 26, 2002. Accepted for publication October 15, 2002. analysis could not rule out the possibility that TL arose relatively The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance recently from another mouse MHC class I gene. The “young TL ” with 18 U.S.C. Section 1734 solely to indicate this fact. model is consistent with the model of Hughes and Nei (1) that 1 This work was supported by the National Institutes of Health Grants RO1 AI30036 monomorphic MHC class I-b genes arise by duplication from and AI18882 (to R.R.R.) and RO1 AI17897 (to R.G.C. and J.R.R.). MHC class I-a alleles. However, to achieve the degree of diver- 2 Emory University School of Medicine, Atlanta, GA 30322. gence noted by Obata (7), the young TL models require that a 3 Address correspondence and reprint requests to Dr. John R. Rodgers, Department of proto-TL gene would undergo rapid diversifying evolution (posi- Immunology, Baylor College of Medicine, One Baylor Plaza Room M929, Houston, tive selection) for some new function, rather than degenerate under TX 77030. E-mail address: [email protected] neutral evolution. In contrast, the “old TL ” model posits that TL is 4 Abbreviations used in this paper: ARS, Ag recognition site; TL, thymus leukemia as old as it appears, that is, that it has evolved like a molecular Ag; MYA, million years ago; dN, rate of nonsynonymous substitution; dS, rate of ␤ ␤ synonymous substitution; 2m, 2-microglobulin; cytb, cytochrome b. clock. This model leaves unspecified the origin of the proto-TL

Copyright © 2002 by The American Association of Immunologists, Inc. 0022-1767/02/$02.00 The Journal of Immunology 6891 gene itself, for example, whether it arose from an early mammalian Table I. Revision of nomenclature of TL genes MHC class I-a allele. These two models differ chiefly in the time of divergence from non-TL MHC genes, and thus the rapidity with Old Nomenclature Relative which TL must have diverged from a putative MHC class I-a or- (phenotype) Locus/Strain New Nomenclature Expression igin. To distinguish between these models and obtain direct evi- TLa a1 (TLA) A/J MumuTL0101 High dence for the time of divergence, we collected TL sequences from TLa a2 (TLA) A/J MumuTL0102 High the murine genera Mus and . TLa a3 (TLA) A/J MumuTL0103 High a4 A The Old World subfamily Murinae within family in- TLa (TL ) A/J MumuTL0104 High TLa b (TLB) T3b/C3H MumuTL0201 Low cludes several lineages whose relationships are unclear. One lin- TLa c (TLC) T3d/BALB/cJ MumuTL0202 Low eage includes Rattus and its close allies, Tokudaia and Diplothrix TLa d (TLC) T18/BALB/cJ MumuTL0301 Intermediate (24). A second lineage includes Mus and, probably, Hybomys and TLa e (TLE) P/J MumuTL0401 High TLa f (TLF) 129/Boy MumuTL0501 Intermediate Mastomys. These two lineages diverged between 14 and 40 MYA. g G w1 (25, 26). The genus Mus consists of four subgenera, Mus, Nanno- TLa (TL ) TLa /molossinus MumuTL0302 Unknown mys (African pygmy mice), (spiny mice), and Coelymys (shrew mice). These subgenera diverged ϳ9 MYA (27). We characterized the extracellular domains of 27 different TL total of 200 ng of genomic DNA was amplified using platinum Pfx (In- sequences from 10 species of Mus, including 2 subspecies from M. vitrogen, Carlsbad, CA). musculus and 3 strains of Rattus norvegicus. These molecules show hyperconservation of the putative ARS, based on the ratio of Nomenclature Downloaded from dN and dS. The conserved orthology within Murinae strengthens We followed a convention for naming MHC alleles in which the first two the hypothesis that TL evolved a specialized function before the letters of the genus is followed by the first two letters of the species, divergence of mice and rats and likely at the beginning of the followed by the gene name and four digits (31). The first two digits rep- resent a major lineage and the last two represent subtypes. For example, the mammalian radiation. first TL gene described from Mus minutoides would be designated MumiTL1401 because it is the fourteenth major lineage of TL analyzed.

Materials and Methods Subtype digits were applied to newly described TL sequences based on http://www.jimmunol.org/ Cell lines and phylogeny of exons 2 and 3 even though allelism could not be determined. New sequences were deposited in GenBank.5 An older nomenclature exists Cell lines from Mus dunni, Mus abbotti, Mus setulosus, Mus minutoides, for TL of inbred mice (Table I). TLag refers to the protein encoded by the Mus platythrix, Mus shortridgei, and Mus pahari were obtained from Dr. TLaw1 locus (7); in our analysis the name is changed to MumuTL0302. S. Chattopadhyay (National Institutes of Health, Bethesda, MD). Speci- mens of Mus musculus praetextus, M. M. derived from animals captured in DNA sequencing Faiyum and Giza, Egypt, were obtained from R. D. Sage (University of PCR-amplified genomic DNA was cloned into pZero TOPO Blunt II (In- California, Berkeley, CA) (28). Cell lines were generated from primary tail vitrogen) plasmid and multiple clones were selected for DNA sequence cell fibroblasts of M. M. praetextus (Faiyum), and M. M. praetextus (Giza) analysis. DNA sequencing was performed by Lone Star Labs (Houston, using wild-type SV40 virus provided by Dr. J. Butel (Baylor College of

TX) on the ABI Prism Automated DNA sequencer 377XL using Big Dye by guest on September 24, 2021 Medicine, Houston, TX) using the technique of Lander and Chattopadhyay Terminator Ready Reaction Cycle Sequencing kit (Applied Biosystems, (29). Genomic DNA samples of Mus cookii and Mus caroli were provided Foster City, CA). The cytb gene was sequenced with the PCR primers used by C. Kozak (National Institute of Allergy and Infectious Diseases, Be- in the PCR and with an internal forward primer (5Ј-CCCTAGTCGAAT thesda, MD). BALB/cJ, C57BL/6J, and M. pahari/Ei were obtained from GAATTTGAGG) derived from a consensus region of Mus cytb. Overlap- The Jackson Laboratory (Bar Harbor, ME). Outbred Sprague Dawley ping primers were used to sequence the 3-kb fragment of TL. External (Holtzman strain) rats were obtained from Harlan Breeders (Indianapolis, IN). Ϫ primers (M13F and M13R) specific for the plasmid backbone were used Tissues were harvested surgically and stored at 80°C until use. DNA was initially. Exon 2 (5Ј-GTTCTGGGAGGAGGTCGGAGTCTCAC forward extracted by phenol:chloroform from tissues or cell lines. DNA from an out- and 5Ј-TGGGGACAGACTCTTAGATTT reverse), exon 3 (5Ј-GTTTG bred Wistar rat (Harlan Breeders) and the Fischer rat-derived CREF cell line GAGAATTCCTAGGGTGGGCGGG forward and 5Ј-CTGTTGTCACCT (30) were kindly provided by Drs. J. Rosen and S. Marriott, respectively (Bay- TTTAAAATTAAA reverse) and exon 4 (5Ј-TTTTATGTAACCTACT lor College of Medicine). GGGGAAATTTGA forward and 5Ј-CTGGGAAGGGAAGGGTAAGG ACATGATGG reverse) specific primers were used to determine exon Isolation of TL genes by PCR sequences. Intron 3 is ϳ1.8 kb. Intron 4-specific primers (5Ј-GAACA For genomic amplification, oligonucleotide primers were designed from GAAAAAAGACACAGGAGTGCACAGG forward, 5Ј-CACATGTGTT conserved regions of known TL genes. Primers, exon 2–4 forward (5Ј-GT TTTGGAGGATCTGAGGAGAAG reverse, and internal 5Ј-AGGAAC TCTGGGAGGAGGTCGGAGTCTCAC) and exon 2–4 reverse (5Ј-CAT ATGAAGAGGCTGAACCTTGAG, 5Ј-ACMGWTAGAATCKCCACTTG, TGTTCTTTCTCATCCACATCATAAC) were used to generate an ϳ3-kb 5Ј-CCTTTCATCCTGAAGAGA) were also used to sequence the entire intron product encoding the extracellular domain of TL under these conditions: 3 of representative samples. Ambiguities were resequenced in the opposite initial denaturation at 94°C for 2 min, then 35 cycles of 94°C for 1 min, direction or called manually. 60 Ϯ 10°C for 1 min, 68°C for 3 min, and a final extension for 10 min at dS and dN were calculated according to the method of Nei and Gojobori 68°C. Each reaction was optimized for Mg2ϩ concentration, which ranged (32, 33) with the Synonymous/Nonsynonymous Analysis Program from 1–5 ␮M, and temperature, using a Mastercycler gradient thermocy- (SNAP): http://hiv-web.lanl.gov. SEs were calculated by the method of Ota cler (Brinkman Instruments, Hamburg, Germany). This pair of primers and Nei (34). Amino acid residues implicated in the ARS of TL Ags (7) worked for all species of Mus except M. pahari.A3Ј primer for M. pahari were predicted based on sequence alignment with HLA-A2 (35). (5Ј-CTGGGAAGGGAAGGGTAAGGACATGATGG) was complemen- tary to a different region of intron 4. A low stringency search of the rat Phylogenic analysis TRACE archive (http://www.ncbi.nlm.nih.gov/blast/mmtrace.html) dem- DNA sequences were aligned using MEGALIGN (DNAstar) and Clustal X onstrated putative TL sequences. Intronic primers for putative rat TL were software (36). We excluded insertions and deletions from our analysis. Ј designed based on the sequences obtained for exons 2 (5 -GGCTC Trees were constructed using Clustal X and MEGA2 (37) by the neighbor- Ј CCATCGGATTCCACG and 3 -GGCCTGAGTCCTGCTCCCTTCTTG), joining method (38). The overall significance of the branching pattern for Ј Ј 3(5-GGAAACCTCCAGACCATGCTTG and 3 -GAGGAGGCTCCCA each tree was estimated by bootstrapping (39) and by internal branch test Ј Ј TCGGATTCC), and 4 (5 -CAACTTCCACTCTTCTCCTC and 3 -CCAT (40). The murine MHC class I sequences used in this study were H2-Kb, CACCATTATGAATCTGTC). The following primers were used to gen- H2-Kd, H2-Kf, H2-Kj, H2-Kk, H2-Kq, H2-Ks, H2-Kw28, H2-D2d, H2-Dd, erate an ϳ1150-bp fragment of the entire coding sequence of cytochrome b (cytb): a degenerate forward primer (5Ј-TYTYCWTYTTNGGTTTA CAARAC) and reverse primer (5Ј-TGAAAAAYCATCGTTGT) specific 5 The sequences have been deposited in GenBank under accession nos. to flanking tRNA sequences (S. Steppan, unpublished observations). A AY144125–AY144179. 6892 EVOLUTION OF THE MHC CLASS I-b MOLECULE TL

H2-Df, H2-Dp, H2-Dr, H2-Ds, H2-B1, H2-M3, M3-spretus, H2-Q1, H2-Q4, Results H2-Q5, H2-Q7, H2-Qa-1, H2-T24, H2-T10, RT1.M3, RT1.BM1, RT1.P1, Phylogeny of MHC class I-a and I-b genes RT1.P2, RT1.A2n, RT1.A2h, RT1.Af, and RT1.Au. Peromyscus maniculatus sequences used were Pm13, Pm41, Pm52, Pm62, Pm11, and Pm53. cytb The MHC class I gene family based on representative sequences sequences used were Rattus rattus, R. norvegicus, Rattus argentiventer, from human, rat, and mouse illustrates the relationships of MHC Sigmodon hispidus, M. caroli, Mus cookii, Mus poschiavinus, Mus mus- culus domesticus, Mus musculoides, and Mus speciligus. class I-a and I-b genes (Fig. 1). MHC class I-b genes distantly related to the MHC (CD1d, HFE, and FcRn) have orthologs in

Statistical analysis of slopes of dN /dS human, rat, and mouse. T23/Qa-1 has a rat ortholog (RT1.BM1) but HLA-E is apparently not orthologous even though these mol- We used Student’s t statistic to compare the slopes of two lines from linear regression analysis, forcing the y-intercept to zero to place all the error into ecules appear to have a similar functions (42). ␭ the slope. To test the significance of the difference between two slopes 1 The seven TL sequences available before this study fall within a ␭ and 2: significant clade (100%) that is on a long branch representing ϳ20–25% divergence from the nearest non-TL neighbor. These ␭1 Ϫ ␭2 t ϭ results are concordant with those of Obata et al. (7) in which the ␴ˆf TL clade lies outside a cluster containing other murine and human where MHC-linked class I genes. The TL sequences share between 92 and 99% sequence identity at the nucleotide level. TL genes are 1 1 f 2 ϭ ϩ tightly clustered and fall into at least four different groups (Fig. 1, ͑N Ϫ 1͒␴ 2 ͑N Ϫ 1͒␴ 2 1 x1 2 x2 inset): A-strain sequences, T3 sequences, T18-like sequences, and Downloaded from and Tlaf (strain 129) sequences. The putative rat TL-like genes (23), RT1.P1 and RT1.P2, are outside of the TL family and do not rep- ͑ Ϫ ͒␴ 2 ϩ ͑ Ϫ ͒␴ 2 N1 2 ˆ 1 N2 2 ˆ 2 ␴ˆ 2 ϭ resent rat orthologs of mouse TL. N1 Ϫ N2 Ϫ 2

␴ 2 where x is the variance of the dS values and Ni is the number of pairwise Isolation of cytb genomic sequences from mice and rats Ϫ comparisons (41). The degrees of freedom were Ns 2 where Ns is the ϭ Ϫ We determined the entire sequence of cytb to confirm sample iden- http://www.jimmunol.org/ number of sequences. Letting degrees of freedom Ns 2 corrects for the partial nonindependence of multiple comparisons. tity and to create an independent phylogeny among specimens by guest on September 24, 2021

FIGURE 1. The TL genes represent a di- vergent multigene family. Exon 2 sequences from human, rat, and mouse MHC class I genes were aligned and used to construct den- drograms by the neighbor-joining method. The dendrogram is based on nucleotide substitution per site calculated by Jukes and Cantor method (35). A chicken MHC class I sequence was used as an outgroup. Parenthetical citations designate the origin of the orthologous genes: m ϭ mouse, r ϭ rat, and h ϭ human. An as- terisk indicates that only two A-strain genes are used for this dendrogram, TLaa1 and TLaa4,a newly described gene from A-strain derived ASL.1 cell line (12). TLaa2 and TLaa3 exon 2 sequences are not published. A newer nomen- clature for TL genes is given after the older name. Bootstrap values and interior branch tests were calculated by MEGA2 with 1,000 replications and values Ͼ50% are shown. The Journal of Immunology 6893

Fig. 3A, inset reflects the well-conserved nature of TL orthologs. Rattus TL (RanoTL1901) is the most divergent TL gene from any mouse gene but the coding regions of all three strains of rats tested are identical. As expected, it has a degree of divergence (ϳ85%) similar to other divergences among MHC orthologs between these genera (45, 46). Sequences from close relatives of laboratory mice form a highly significant (100%) cluster with T3, T18, Tlaf, and A-strain sequences. As with the cytb tree, the four Mus subgenera are supported, with the exception of M. Pyromys shortridgei noted before. When we constructed phylogenic trees based on the ␣3 region (Fig. 3B), the total branch length of the tree was reduced. This may ␤ reflect the functional constraints of binding to 2m and CD8 co- receptors or might reflect homogenization of this exon (2). How- ever, the ␣3 dendrograms still showed that TL sequences form a separate branch with significant interior branch strength (97%) (the bootstrap values were 32%, data not shown), although locus spec- ificity is not as great as in exons 2 and 3. The M. shortridgei

sequences (MushTL1801 and 1802) grouped significantly away Downloaded from from TL family members. This suggests that M. shortridgei has exchanged exon 4 with another MHC class I gene. Our overall phylogenetic data demonstrate that TL orthologs exist outside of FIGURE 2. cytb orthologs are highly conserved within the subfamily laboratory strains of mice and are most easily defined by exons 2 Murinae. A total of 1,140 nucleotides of cytb were aligned and used to and 3 and not exon 4. create dendrograms as previously described. A single asterisk indicates http://www.jimmunol.org/ GenBank sequences cytb sequence obtained from GenBank. A double as- terisk indicates obtained from GenBank and confirmed by our analysis. A dN and dS in TL orthologs S. hispidus (cotton rat) cytb sequence serves as an outgroup. Both MHC class I-a and II genes are highly polymorphic and show evidence of positive selection in the ARS relative to the non-ARS residues of exons 2 and 3 (2, 47). In contrast, some MHC class I-b sampled (Fig. 2) for comparison with previously published se- genes showed reduced or absent positive selection in the ARS (1). quences within the family Muridae (43). To determine whether the ARS of TL has been under negative

This analysis revealed only one discrepancy with other phylo- selection, we calculated dN and dS values for ARS and non-ARS genetic analyses based on molecular and morphological criteria residues and compared the ratios of each region within species and by guest on September 24, 2021 Ϯ (44). Thus, Rattus and Mus formed distinct clades with high boot- across species (see Table II and Fig. 4). The mean dN (2.2 0.9) strap values (92%). All samples assigned M. musculus fell into a from the ARS of TL orthologs was significantly lower ( p Ͻ 0.05) Ϯ significant clade. The four subgenera (Mus, , , than mean dS values (17.8 3.9) (Table II) in the ARS. In con- Ϯ Ͻ and Pyromys)ofMus were supported. The exceptional sample trast, the mean dN (23.0 3.0) was significantly higher ( p 0.05) Ϯ was M. shortridgei. This species has been included within the than mean dS (15.6 3.3) in the MHC class I-a ARS, and the Ϯ Ͻ subgenus Pyromys on morphological grounds (44), but multiple mean dN (23.0 3.0) was significantly higher ( p 0.05) in the sequences (cytb, TL, and H-2 M3) (C. Doyle, R. Rich, R. Cook, ARS than in the non-ARS and exon 4 (6.6 Ϯ 1.2 and 5.1 Ϯ 1.1, and J. Rodgers) manuscript in preparation) suggest that our respectively). sample of M. shortridgei is more closely allied with Mus Within exon 4 of MHC class I-a genes the ratio of mean dN to Ϯ Ͻ Coelomys pahari (98%). mean dS (1.4 0.02) is significantly ( p 0.05) greater than the ratio in the non-ARS (0.58 Ϯ 0.03). This is mainly due to a de- The TL gene family is well conserved in mice and rats pressed dS value (see Table II) that may result from homogeniza- Using a 5Ј primer from intron 2 and a 3Ј reverse primer from intron tion of this exon within the mouse MHC class I-a genes (2). The d 4ofT18 , we cloned segments encoding the extracellular domains dN:dS ratio for TL in exon 4 is similar to the dN:dS ratio in the Ͻ of putative TL sequences from Mus species. No specific PCR prod- non-ARS of TL but significantly ( p 0.05) lower than the dN:dS ucts were generated with these primers from R. norvegicus ratio of MHC class I-a exon 4. Nonetheless, exon 4 of TL orthologs (Sprague Dawley). We screened the rat TRACE archive and rat appears to have an evolutionary history different from that of exon genome database under relaxed stringency with exons 2–4 from 4 of MHC class I-a genes. This might reflect a lack of homoge- T18d as probes. This screen revealed putative TL sequences for nization of exon 4 of TL orthologs with other MHC class I genes, each exon. 5Ј and 3Ј primers complementary to flanking introns with the exception of M. Pyromys shortridgei. were made for exons 2–4 individually and these permitted isola- The fact that the dN:dS ratio of the ARS in TL are significantly tion of TL from three strains of rats (Sprague Dawley, Wistar, and lower than the dN:dS ratio in the non-ARS and exon 4 suggested Fischer). RT-PCR analysis confirmed that these exons were de- that the ARS is hyperconserved relative to other regions of the rived from a single gene and not isolated fragments (data not molecule. To correct for the high variance in dS due to including shown). sequences from both close and distant taxa, we plotted the dN vs dS A neighbor-joining tree of the coding region of exons 2 and 3 values for all pairwise comparisons of TL (Fig. 4). The slope of TL for 33 different sequences (Fig. 3A) reveals a single TL cluster ARS (m ϭ 0.121 Ϯ 0.003) is significantly less than the slope of the encompassing all 33 sequences with high bootstrap value (100%) non-ARS (m ϭ 0.300 Ϯ 0.005, p Ͻ 0.005) and exon 4 (m ϭ when compared with the outgroups, M. M. domesticus and R. nor- 0.270 Ϯ 0.006, p Ͻ 0.005). Thus, the ARS of TL is hypercon- vegicus class I-a and I-b genes. The shallowness of the TL clade in served relative to the non-ARS and exon 4. 6894 EVOLUTION OF THE MHC CLASS I-b MOLECULE TL Downloaded from http://www.jimmunol.org/ by guest on September 24, 2021

FIGURE 3. Orthologs of TL exist outside of M. musculus. A, Exon 2 to 3 coding sequences were created from genomic DNA sequence based on alignment. These sequences were used to create dendrograms. B, Exon 4 sequences from contiguous DNA segments were used to create a dendrogram. Dendrograms were based on the nucleotide sequence beginning at position ϩ3 to maintain the correct reading frame. H2-K/D/Q/T and RT1 complex exons were used as outgroups Interior branch test values are shown. A double dashed line indicates an artificial compression of the scale introduced to minimize the dendrogram. a, Sequences from a M. M. praetextus cell line derived from an caught near Giza, Egypt. b, Sequences from a M. M. praetextus cell line derived from an animal caught near Faiyum, Egypt. The Journal of Immunology 6895

Table II. Mean Ϯ SE nucleotide substitutions/100 synonymous and nonsynonymous sites in comparison among murid TL genes

ARS (n ϭ 57) Non-ARS (n ϭ 122) Exon 4 (n ϭ 92)

Comparisons dS dN dS dN dS dN TL family members TL vs TL (28) 17.8 Ϯ 3.9 2.2 Ϯ 0.9* 20.7 Ϯ 3.1 6.5 Ϯ 1.1* 13.4 Ϯ 2.5 4.2 Ϯ 0.8 Class I-a family membersa Class I-a vs class I-a (19) 15.6 Ϯ 3.3 23.0 Ϯ 3† 11.3 Ϯ 2.1 6.6 Ϯ 1.2* 3.6 Ϯ 1.3‡ 5.1 Ϯ 1.1†

a Includes 15 mouse and 4 rat MHC class I-a sequences. Ͻ ء , Values of p 0.005, dN is significantly lower than dS. † Ͻ Values of p 0.005, dN is significantly higher than dS. ‡ Ͻ Values of p 0.005, dS (I-a) is significantly lower than dS (TL).

Characterization of TL orthologs from the genera Mus and Discussion Rattus We describe 27 new TL sequences in 11 species of Murinae and A striking feature of the TL sequences is the highly conserved show that they have limited variability. Our data support the model Downloaded from nature of the extracellular domains, especially the ARS. All TL of Obata et al. (7), based on a limited data set, n ϭ 7 in one species, orthologs contain the four conserved cysteine residues needed to which suggested TL is ancient and thus should exist in distantly form the two intramolecular disulfide bonds that are necessary for related genera. They do not support the model of Hughes and Nei MHC class I structure (48). A majority of TL orthologs also con- (1) and Rogers (53) which suggests that mouse MHC class I-b tain N-linked glycosylation motifs (NXS/T, X P) (49) at position genes arise from MHC class I-a genes and rapidly evolve toward 86 and 90. Two natural variants occur: A-strain alleles have a pseudogeny. N86S substitution that destroys the glycosylation site and Rattus Our data show that TL genes arose before the split of Rattus and http://www.jimmunol.org/ TL (RanoTL1901) has only the N86 glycosylation site (Fig. 5). All Mus and thus have been retained for at least 30 million years. The TL orthologs that have a glycosylation motif at N86 use the less estimated time of divergence between mice and rats remains con- efficient (20%) and rarely used motif, NLS (49). Almost all other troversial. The fossil record dates the separation between 12 and 14 MHC class I molecules use NQS (data not shown) as a recognition MYA (54). Molecular studies based on different genes estimate the motif which is more efficiently glycosylated (49). divergence time of Rattus and Mus to between 20 and 40 MYA The MHC class I-a ARS, based on HLA-A2, contains six pock- (25, 26, 55). This range is attributable to different approaches and ets (A–F) that anchor individual residues of an extended peptide “calibration” times (26) based on the divergence of birds and

(35). Pocket A includes 10 residues that anchor the N terminus, of . Extrapolating from our expanded data set of the TL by guest on September 24, 2021 which three tyrosines are conserved in both TL and class I-a mol- gene family, we estimate TL to have diverged from other mam- ecules. Pocket F seals the C terminus of the peptide (50). All of the malian MHC class I genes ϳ100 MYA. This estimated divergence anchoring and sealing positions conserved in class I-a molecules time, at or before the time of the mammalian radiation, suggests are also conserved in TL (Table III), suggesting that if TL binds that TL was present in ancestral mammals. We do not find a TL peptides they could be 8–10 aa in length. The internal pockets of ortholog in the human genome, suggesting that TL was lost in this TL sequences show slight variability within the TL gene family. species. Molecular modeling suggests that the ARS of TL is not as oc- Our data also support the hypothesis of multiple ancient TL cluded or hydrophobic as the ARS of CD1 molecules (51), thus it genes in Mus. As seen in Fig. 3A, the percentage of divergence seems unlikely that TL would bind lipids. (Jukes-Cantor distance) is as great for different loci of TL (T3 and Obata et al. (7) described 17 residues that are unique to seven TL T18) within the same species as the percentage of divergence for sequences in strains and that are rarely found in orthologs from different species. The amount of diversity between other MHC class I molecules. Of these, 13 are in the ARS and 4 T3 and T18 in the ARS and non-ARS is due to an increase in dS are in the non-ARS. Within the ARS, six of these TL signatures are (0.03 and 0.05, respectively) relative to the dN (0.0 and 0.02, re- conserved in our data set: L155 and K165 are not found in 200 spectively). Our extensive data set modifies the original model of other mammalian MHC class I sequences. A61, E65, F73, and Obata et al. (7) by demonstrating that the TL gene family is well M81 are only found in one other sequence (PemaT24 (52), T24d, conserved between mice and rats and contains multiple sequences H2-M9, and H2-M10, respectively). Several other signatures in TL that are ancient. exist outside of the ARS region. Residues 11–13 (Ala, Leu, Ser) TL gene family members showed no signs of positive selection are highly conserved in TL sequences but are also found in several in the ARS. In particular, the slopes of dN/dS for the ARS were H2-M3 and Qa-1 orthologs. significantly lower, approximately one-third, than those of non- Table III shows the limited variability of the TL ARS. Sixteen ARS and exon 4. Thus, the ARS is hyperconserved in TL se- of 57 residues are variable in the ARS. Of the 16 variable residues, quences. These data suggest that the ARS of TL has been func- 5 are conservative changes (L82F, L169F, Y22F, R62K, and tionally conserved for at least 30 million years. This region may V67I), 8 have infrequent but nonconservative changes (24, 61, 69, interact with a conserved ligand such as a TAP-independent pep- 116, 145, 156, 163, and 166), and 3 residues (76, 149, and 150) in tide, or another protein such as a NK cell receptor-like molecule. the ARS are frequently and nonconservatively variable. Nine residues The functions of most MHC class I-b molecules are poorly de- within the ARS are invariant and conserved in other mammalian and fined. Some are obviously pseudogenes but others have been avian MHC class I molecules (1) and might represent residues in- shown to contribute to host defense and nonimmune functions (3). volved in maintaining the structure of the ARS. Of these, all TL mol- The biological role of TL is unknown but the restricted expression ecules have a nonconservative V165K substitution. pattern of TL suggests a role in intestinal immunology and/or T 6896 EVOLUTION OF THE MHC CLASS I-b MOLECULE TL

Table III. ARS Composition

Amino Acid Residue in MHC Position Residue in TL (34)a Class I-a (15)

5Lb ML 7c YY 9 Y EVYH 22 Y33 F1d YF 24 A33 T1 SEA 26 G G 57 P P 58 E E 59c YY 61 A33 Q1 E 62 R32 K1 T1 RQE 63 E EINQ 64 T T 65 E QR 66 I KRIN 67 V30 I4 AS 68 T K 69 S33 N1 GDSR Downloaded from 70 N QNHS 71 A EK 72 Q QE 73 F WSI 74 F FS 75 R R 27 6 1

76 E G R V http://www.jimmunol.org/ 77 N SDN 80 T NTI 81 M LA 82 L33 F1 LQ 84e YY 95 I LIVY 97 V QWRGV 99 Y SAYRF 114 E QWEL 116 H33 Y1 YFHV 143e TT by guest on September 24, 2021 145 S33 L1 RH 146e KK 147e WW 149 Q21 R12 L1 Q 150 A38 D5 T1 SAG 151 G G 152 Y ADE 154 E E 155 L HRTY 156 R33 S1 YDLKF FIGURE 4. The TL gene family has evolved under negative selection. 157 R RK 158 T A dN and dS were calculated from pairwise comparisons of all TL orthologs c Ⅺ ࡗ 159 YY ( ) and MHC class I-a ( ) used in our study and plotted as a scatter 161 E E diagram. Rates were calculated for A, ARS; B, non-ARS; and C, exon 4. 162 G G Slopes were calculated by linear regression. The slope of the linear regres- 163 P33 L1 EATV sion of TL ARS is significantly lower that the slope of the Non-ARS and 165 K V exon 4; p Ͻ 0.005. 166 D32 H2 E 167 S WS 169 L31 F3 HRLS 171c YYH cell effector function. TL transcripts (␣1–␣3) are expressed in the a small intestine of both M. pahari/Ei and Rattus (data not shown). The no. of sequences compared. b Invariant residues. The conserved expression pattern of TL orthologs strengthens the c Residues implicated in sealing pocket A (N terminus). hypothesis that TL has a biological role in the mucosal immunity. d The frequency of the variable residue. e T18d was shown to bind CD8␣␣ homodimers expressed on in- Residues implicated in sealing pocket F (C terminus). testinal intraepithelial lymphocyte with greater affinity than it

FIGURE 5. Predicted amino acid sequences of TL orthologs. A majority sequence of our data set is shown above the samples but does not represent a natural sequence. Identity is indicated by a dash. Dots represent deletions and an asterisk represents a stop codon. a and b, Sequences from two different M. M. praetextus cell lines (see Fig. 3). ∧ represents residues that are rarely found in other MHC class I molecules. Conserved cysteines are indicated by ␣ ␣ bold lettering. Glycosylation sites are underlined. Only one rat TL sequence (RanoTL1901) is shown. A, Predicted 1 domain; B, predicted 2 domain; and ␣ C, predicted 3 domain. The Journal of Immunology 6897 Downloaded from http://www.jimmunol.org/ by guest on September 24, 2021

FIGURE 5. 6898 EVOLUTION OF THE MHC CLASS I-b MOLECULE TL binds CD␣␤ heterodimers (56). The affinity of T18d for CD8␣␣ References b d was 10-fold higher than that of H-2K for CD8␣␣. T18 and 1. Hughes, A. L., and M. Nei. 1989. Evolution of the major histocompatibility H2-Kb probably bind to CD8␣␣ analogously (57). H2-Kb com- complex: independent origin of nonclassical class I genes in different groups of ␣␣ b mammals. Mol. Biol. Evol. 6:559. plexed with CD8 revealed two key contact regions of H2-K , 2. Hughes, A. L., and M. Nei. 1988. Pattern of nucleotide substitution at major the AB and CD loops (58). Only the AB loop differs considerably histocompatibility complex class I loci reveals overdominant selection. Nature in T18d: H2-Kb has PEDK (195–198) while T18d has PEGY. Two 335:167. 3. Shawar, S. M., J. M. Vyas, J. R. Rodgers, and R. R. Rich. 1994. Antigen pre- major motifs are seen in AB loop of Mus in TL sequences: PEGY sentation by major histocompatibility complex class I-B molecules. Annu. Rev. (T18-like) and PEGD (T3-like). TL Ags that contain either motif Immunol. 12:839. are expressed in the small intestine (10, 11). If preferential binding 4. Soloski, M. J., J. W. Uhr, L. Flaherty, and E. S. Vitetta. 1981. Qa-2, H-2K, and H-2D alloantigens evolved from a common ancestral gene. J. Exp. Med. 153: to CD8␣␣ is necessary for TL function, our data suggest some 1080. plasticity in the interaction with the CD8␣␣ homodimers. Alter- 5. Sipes, S. L., M. V. Medaglia, D. L. Stabley, C. S. DeBruyn, M. S. Alden, natively, the TL family members have multiple and complemen- V. Catenacci, and C. P. Landel. 1996. A new major histocompatibility complex class I b gene expressed in the mouse blastocyst and placenta. Immunogenetics tary functions such that only TL molecules that contain PEGY 45:108. interact with CD8␣␣ homodimers preferentially while other TL 6. Ohno, S. 1970. Evolution by Gene Duplication. Springer-Verlag, New York. 7. Obata, Y., Y. Satta, K. Moriwaki, T. Shiroishi, H. Hasegawa, T. Takahashi, and loci perform another function. N. Takahata. 1994. Structure, function, and evolution of mouse TL genes, non- Another functional signature of TL is its cell surface expression classical class I genes of the major histocompatibility complex. Proc. Natl. Acad. in the absence of TAP2 (14). Cells lacking functional TAP show a Sci. USA 91:6589. 8. Amadou, C., A. Kumanovics, E. P. Jones, D. Lambracht-Washington, marked decrease in MHC class I-a expression because peptides are M. Yoshino, and K. Fischer Lindahl. 1999. The mouse major histocompatibility limiting (59). Due to the hyperconserved nature of the ARS of TL, complex: some assembly required. Immunol. Rev. 167:211. Downloaded from TL binds a relatively invariant and TAP-independent Ag or no 9. Cook, R. G., and N. F. Landolfi. 1983. Expression of the thymus leukemia an- tigen by activated peripheral T lymphocytes. J. Exp. Med. 158:1012. peptide at all. No informative Ag-elution studies of TL Ags have 10. Hershberg, R., P. Eghtesady, B. Sydora, K. Brorson, H. Cheroutre, R. Modlin, described specific peptides or motifs (our unpublished observa- and M. Kronenberg. 1990. Expression of the thymus leukemia antigen in mouse tions) (13, 18). intestinal epithelium. Proc. Natl. Acad. Sci. USA 87:9727. 11. Wu, M., L. van Kaer, S. Itohara, and S. Tonegawa. 1991. Highly restricted ex- An alternative hypothesis suggests that TL molecules bypass the pression of the thymus leukemia antigens on intestinal epithelial cells. J. Exp. quality control mechanisms of the peptide loading complex in the Med. 174:213. http://www.jimmunol.org/ endoplasmic reticulum that normally restrict MHC class I-a mat- 12. Albrecht, A. M., J. L. Biedler, D. J. Hutchison, B. A. Spengler, and E. Stockert. 1976. Radiation-induced murine leukemia ERLD in cell culture. Cancer Res. uration. Calreticulin, an endoplasmic reticulum-resident chap- 36:3784. arone, interacts with sugar moieties on N86 glycosylation site of 13. Holcombe, H. R., A. R. Castano, H. Cheroutre, M. Teitell, J. K. Maher, d P. A. Peterson, and M. Kronenberg. 1995. Nonclassical behavior of the thymus H2-L (60). Most TL molecules have two glycosylation sites in leukemia antigen: peptide transporter-independent expression of a nonclassical close proximity, N86 and N90, but are only monoglycosylated class I molecule. J. Exp. Med. 181:1433. ␤ (data not shown). Two natural variants occur. A-strain alleles have 14. Rodgers, J. R., V. Mehta, and R. G. Cook. 1995. Surface expression of 2- microglobulin-associated thymus-leukemia antigen is independent of TAP2. Eur. a N86S substitution that ablates the first glycosylation site. Rattus J. Immunol. 25:1001. TL molecules contain only the first glycosylation site at N86 which 15. Tsujimura, K., Y. Obata, S. Iwase, Y. Matsudaira, S. Ozeki, and T. Takahashi. is the poor acceptor motif, NLS. Another quality control interac- 2000. The epitope detected by cytotoxic T lymphocytes against thymus leukemia by guest on September 24, 2021 (TL) antigen is TAP independent. Int. Immunol. 12:1217. tion involves the TAP-tapasin-calnexin complex. Residues 128– 16. Eghtesady, P., K. A. Brorson, H. Cheroutre, R. E. Tigelaar, L. Hood, and 137 in ␣2 of H-2Ld have been implicated in binding this complex. M. Kronenberg. 1992. Expression of mouse Tla region class I genes in tissues This region is highly conserved in TL gene family members and enriched for ␥␦cells. Immunogenetics 36:377. 17. Sharma, P., M. J. Page, L. S. Poritz, W. A. Koltun, and M. J. Chorney. 1997. An MHC class I genes. There is a A136V substitution in several se- increased ␥␦T cell population in the intestine of thymus-leukemia antigen trans- quences across species boundaries in TL gene family members. genic mice. Cell Immunol. 176:153. Cell surface expression of TL, like that of other MHC class I 18. Sharma, P., S. Joyce, K. A. Chorney, J. W. Griffith, R. H. Bonneau, F. D. Wilson, ␤ C. A. Johnson, R. A. Flavell, and M. J. Chorney. 1996. Thymus-leukemia antigen molecules, depends on 2m association (14). A total of 82 different interacts with T cells and self-peptides. J. Immunol. 156:987. MHC class I molecules were shown to have 19 residues that make 19. Wang, C. R., D. Lambracht, K. Wonigeit, J. C. Howard, and K. Fischer. Lindahl. 44 contacts to 18 ␤ m residues (61). Of these, 37 contact points 1995. Rat RT1 orthologs of mouse H2-M class Ib genes. Immunogenetics 42:63. 2 20. Hughes, A. L. 1991. Independent gene duplications, not concerted evolution, were conserved Ͼ90%. Within laboratory strains of mice, TL mol- explain relationships among class I MHC genes of murine . Immunoge- ␤ netics 33:367. ecules maintain 89% conservation of 2m-contact residues. In the 21. Crew, M. D., L. M. Bates, C. A. Douglass, and J. L. York. 1996. Expressed larger set of 34 TL sequences, 74% of the residues are conserved. Peromyscus maniculatus (Pema) MHC class I genes: evolutionary implications Conservation of specific contact residues suggests that all TL gene and the identification of a gene encoding a Qa1-like antigen. Immunogenetics family members require ␤ m. 44:177. 2 22. Rogers, M. J., R. N. Germain, J. Hare, E. Long, and D. S. Singer. 1985. Com- The evolution of the MHC class I genes in the mouse is char- parison of MHC genes among distantly related members of the genus Mus. J. Im- acterized by a birth and death process (62). Gene duplications are munol. 134:630. thought to generate novel protein functions but little is known 23. Matsuura, A., S. Takayama, M. Kinebuchi, Y. Hashimoto, K. Kasai, D. Kozutsumi, S. Ichimiya, R. Honda, T. Natori, and K. Kikuchi. 1997. RT1.P, about the selective pressures governing this process (63). Ohno (6) rat class Ib genes related to mouse TL: evidence that CD1 molecules but not hypothesized that once a gene duplicated, one copy was freed from authentic TL antigens are expressed by rat thymus. Immunogenetics 46:293. selective pressures to drift and either assume novel function or be 24. Suzuki, H., K. Tsuchiya, and N. Takezaki. 2000. A molecular phylogenetic framework for the Ryukyu endemic rodents Tokudaia osimensis and Diplothrix lost. An alternative hypothesis suggests that the selective pressures legata. Mol. Phylogenet. Evol. 15:15. are lessened on both copies, allowing “subfunctionalization” (64). 25. Adkins, R. M., E. L. Gelke, D. Rowe, and R. L. Honeycutt. 2001. Molecular phylogeny and divergence time estimates for major groups: evidence from Alternatively, “complementation” occurs such that both genes multiple genes. Mol. Biol. Evol. 18:777. complement separate functions (64). The TL gene family provides 26. Nei, M., P. Xu, and G. Glazko. 2001. Estimation of divergence times from mul- an opportunity to study the selective pressures governing a multi- tiprotein sequences for a few mammalian species and several distantly related organisms. Proc. Natl. Acad. Sci. USA 98:2497. gene family in the MHC during the past 30–100 million years. 27. Bonhomme, F. 1986. Evolutionary relationships in the genus Mus. Curr. Top. Microbiol. Immunol. 127:19. Acknowledgments 28. Ferris, S. D., R. D. Sage, E. M. Prager, U. Ritte, and A. C. Wilson. 1983. Mi- tochondrial DNA evolution in mice. Genetics 105:681. We thank D. Brake for technical assistance and C. Doyle and J. Levitt for 29. Lander, M. R., and S. K. Chattopadhyay. 1984. A Mus dunni cell line that lacks useful discussions. sequences closely related to endogenous murine leukemia viruses and can be The Journal of Immunology 6899

infected by ectropic, amphotropic, xenotropic, and mink cell focus-forming vi- 48. Prochnicka-Chalufour, A., J. L. Casanova, P. Kourilsky, and J. M. Claverie. ruses. J. Virol. 52:695. 1989. Extensive structural homology between H-2 K/D/L antigens and non- poly- 30. Fisher, P. B., L. E. Babiss, I. B. Weinstein, and H. S. Ginsberg. 1982. Analysis morphic class I Qa, Tla and “37” molecules suggests they may act as peptide of type 5 adenovirus transformation with a cloned rat embryo cell line (CREF). carriers. Res. Immunol. 140:133. Proc. Natl. Acad. Sci. USA 79:3527. 49. Shakin-Eshleman, S. H., S. L. Spitalnik, and L. Kasturi. 1996. The amino acid at 31. Klein, J., R. E. Bontrop, R. L. Dawkins, H. A. Erlich, U. B. Gyllensten, the X position of an Asn-X-Ser sequon is an important determinant of N-linked E. R. Heise, P. P. Jones, P. Parham, E. K. Wakeland, and D. I. Watkins. 1990. core-glycosylation efficiency. J. Biol. Chem. 271:6363. Nomenclature for the major histocompatibility complexes of different species: a 50. Saper, M. A., P. J. Bjorkman, and D. C. Wiley. 1991. Refined structure of the proposal. Immunogenetics 31:217. human histocompatibility antigen HLA-A2 at 2.6 A resolution. J. Mol. Biol. 32. Nei, M., and T. Gojobori. 1986. Simple methods for estimating the numbers of 219:277. synonymous and nonsynonymous nucleotide substitutions. Mol. Biol. Evol. 51. Natarajan, K., H. Li, R. A. Mariuzza, and D. H. Margulies. 1999. MHC class I 3:418. molecules, structure and function. Rev. Immunogenet. 1:32. 33. Korber, B. 2000. HIV Signature and Sequence Variation Analysis: Computa- 52. Crew, M. D., C. A. Douglass, and K. A. Brorson. 1994. H-2T24 and PemaT24: tional Analysis of HIV Molecular Sequences. Kluwer, Dordrecht, The Nether- orthologous expressed MHC class Ib genes from mouse and Peromyscus man- lands. iculatus. Immunogenetics 39:412. 34. Ota, T., and M. Nei. 1994. Variance and covariances of the numbers of synon- 53. Rogers, J. H. 1985. Mouse histocompatibility-related genes are not conserved in ymous and nonsynonymous substitutions per site. Mol. Biol. Evol. 11.:613. other mammals. EMBO J. 4:749. 35. Bjorkman, P. J., M. A. Saper, B. Samraoui, W. S. Bennett, J. L. Strominger, and 54. Jacobs, L. L., and D. Pilbeam. 1980. Of mice an men: fossil-based divergence D. C. Wiley. 1987. The foreign antigen binding site and T cell recognition regions dates and molecular “clocks.” J. Hum. Evol. 9:551. of class I histocompatibility antigens. Nature 329:512. 55. Kumar, S., and S. B. Hedges. 1998. A molecular timescale for vertebrate evo- 36. Thompson, J. D., T. J. Gibson, F. Plewniak, F. Jeanmougin, and D. G. Higgins. lution. Nature 392:917. 1997. The CLUSTAL_X windows interface: flexible strategies for multiple se- 56. Leishman, A. J., O. V. Naidenko, A. Attinger, F. Koning, C. J. Lena, Y. Xiong, quence alignment aided by quality analysis tools. Nucleic Acids Res. 25:4876. H. C. Chang, E. Reinherz, M. Kronenberg, and H. Cheroutre. 2001. T cell re- 37. Kumar, S., K. Tamura, I. Jakobsen, and M. Nei. 2001. MEGA2: molecular evo- sponses modulated through interaction between CD8␣␣ and the nonclassical lutionary genetics analysis software. Bioinformatics 17:1244.

MHC class I molecule, TL. Science 294:1936. Downloaded from 38. Saitou, N., and M. Nei. 1987. The neighbor-joining method: a new method for 57. Devine, L., L. Rogozinski, O. V. Naidenko, H. Cheroutre, and P. B. Kavathas. reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406. 2002. The complementarity-determining region-like loops of CD8 ␣ interact dif- 39. Felsenstein, J. 1985. Confidence limits on phylogenies: an approach using the ferently with ␤ -microglobulin of the class I molecules H-2Kb and thymic leu- bootstrap. Evolution 39:783. 2 kemia antigen, while similarly with their ␣3 domains. J. Immunol. 168:3881. 40. Sitnikova, T., A. Rzhetsky, and M. Nei. 1995. Interior-branch and bootstrap tests of phylogenetic trees. Mol. Biol. Evol. 12:319. 58. Kern, P. S., M. K. Teng, A. Smolyar, J. H. Liu, J. Liu, R. E. Hussey, R. Spoerl, H. C. Chang, E. L. Reinherz, and J. H. Wang. 1998. Structural basis of CD8 41. Bailey, N. T. 1980. Statistical Methods in Biology. Wiley, New York. ␣␣ 42. Kurepa, Z., C. A. Hasemann, and J. Forman. 1998. Qa-1b binds conserved class coreceptor function revealed by crystallographic analysis of a murine CD8 ectodomain fragment in complex with H-2Kb. Immunity 9:519.

I leader peptides derived from several mammalian species. J. Exp. Med. 188:973. http://www.jimmunol.org/ 43. Martin, Y., G. Gerlach, C. Schlotterer, and A. Meyer. 2000. Molecular phylogeny 59. Attaya, M., S. Jameson, C. K. Martinez, E. Hermel, C. Aldrich, J. Forman, of European muroid rodents based on complete cytochrome b sequences. Mol. K. Fischer Lindahl, M. J. Bevan, and J. J. Monaco. 1992. Ham-2 corrects the Phylogenet. Evol. 16:37. class I antigen-processing defect in RMA-S cells. Nature 355:647. 44. Silver, L. 1995. Mouse Genetics: Concepts and Applications. Oxford Univ. Press, 60. Harris, M. R., Y. Y. Yu, C. S. Kindle, T. H. Hansen, and J. C. Solheim. 1998. London and New York. Calreticulin and calnexin interact with different protein and glycan determinants 45. Parker, K. E., C. A. Carter, G. Murphy, and J. W. Fabre. 1991. The rat RT.BM1 during the assembly of MHC class I. J. Immunol. 160:5404. MHC class I cDNA shows a high level of sequence similarity to the mouse 61. Tysoe-Calnon, V. A., J. E. Grundy, A. S. Nealis, and S. J. Perkins. 1990. High d ␤ H-2T23 gene. Immunogenetics 34:211. conservation of 2-microglobulin contact residues among 82 class I major his- 46. Wang, C.-R., A. Livingston, G. Butcher, E. Hermel, J. C. Howard, and tocompatability complex ␣-chain sequences. Biochem. Soc. Trans. 18:930. K. Fischer Lindahl. 1991. Antigen presentation by neoclassical MHC class I gene 62. Nei, M., X. Gu, and T. Sitnikova. 1997. Evolution by the birth-and-death process products in murine rodents. In Molecular Evolution of the Major Histocompat- in multigene families of the vertebrate immune system. Proc. Natl. Acad. Sci.

ibility Complex, Vol. H59, NATO ASI Series Ed. J. Klein and D. Klein, eds. USA 94:7799. by guest on September 24, 2021 Springer, Heidelberg, p. 441. 63. Otto, S. P., and P. Yong. 2002. The evolution of gene duplicates. Adv. Genet. 47. Hughes, A. L., and M. Nei. 1989. Nucleotide substitution at major histocompat- 46:451. ibility complex class II loci: evidence for overdominant selection. Proc. Natl. 64. Kondrashov, F. A., I. B. Rogozin, Y. I. Wolf, and E. V. Koonin. 2002. Selection Acad. Sci. USA 86:958. in the evolution of gene duplications. Genome Biol. 3:RESEARCH0008.