ORIGINAL PAPER

Species-specific evolution of class I MHC genes in iguanas(Order: Squamata; Subfamily: Iguaninae)

Scott Glaberman & Adalgisa Caccone

Received: 29 January 2008 /Accepted: 9 April 2008 /Published online: 17 May 2008# Springer-Verlag 2008

Abstract Over the last few decades, the major histocom-patibility complex (MHC) has emerged as a model forunderstanding the influence of natural selection on geneticdiversity in populations as well as for investigating thegenetic basis of host resistance to pathogens. However,many vertebrate taxa remain underrepresented in the field ofMHC research, preventing its application to studies ofdisease, evolution, and conservation genetics in thesegroups. This is particularly true for squamates, which areby far the most diversified order of non-avian reptiles buthave not been the subject of any recent MHC studies. In thispaper, we present MHC class I complementary DNA datafrom three squamate species in the subfamily Iguaninae(iguanas): the Galápagos marine iguana (Amblyrhynchuscristatus), the Galápagos land iguana (Conolophus sub-cristatus), and the green iguana (Iguana iguana). All se-quences obtained are related to the few published classI genes from other squamates. There is evidence formultiple loci in each species, and the conserved α-3domain appears to be evolving in a species-specific manner.Conversely, there is some indication of shared polymor-phism between species in the peptide-binding α-1 and α-2domains, suggesting that these two regions have differentphylogenetic histories. The great similarity between α-3sequences in marine iguanas in particular suggests thatconcerted evolution is acting to homogenize class I loci

within species. However, while less likely, the data are alsocompatible with a birth and death model of evolution.

Keywords Major histocompatibility complex . Squamate .

Reptile . Duplication . Concerted evolution . Classical .

Nonclassical . Polymorphism . Birth and death evolution

Introduction

Class I genes of the major histocompatibility complex(MHC) represent a distinct evolutionary lineage that existsin all vertebrates examined. They encode a single polypep-tide chain whose folding characteristics are structurallyconserved. The two N-terminal domains of the protein,designated α-1 and α-2, interact to form a cleft that iscapable of binding short peptides or, in some cases, lipids(Beckman et al. 1994; Bjorkman et al. 1987; Sieling et al.1995). The adjacent α-3 domain resembles the constantregion of immunoglobulins and often associates covalentlywith a β2-microglobulin molecule (Klein 1986; Rodgersand Cook 2005). Lastly, many class I molecules alsocontain two C-terminal domains that span the cell mem-brane and anchor the protein into the cytoplasm.

Class I genes have traditionally been divided into twogroups commonly referred to as classical or class Ia andnonclassical or class Ib. Typically, classical loci are highlypolymorphic, have strong and wide expression acrossdifferent tissue types, and are located within the MHCregion of the genome. They encode receptor structures thatbind short peptide fragments derived from pathogens foundwithin the host cell. The MHC-antigen complex is thentransported to the cell surface where it is recognized byCD8+ TC cells. This triggers the activation and differenti-ation of naïve T cells into cytotoxic T lymphocytes, which

Immunogenetics (2008) 60:371–382DOI 10.1007/s00251-008-0298-y

All iguana class I sequences were submitted to GenBank under thefollowing accession numbers: EU604308–EU604319.

S. Glaberman (*) :A. CacconeDepartment of Ecology and Evolutionary Biologyand the Yale Institute for Biospheric Studies,Yale University, 21 Sachem Street,New Haven, CT 06520-8105, USAe-mail: scott.glaberman@gmail.com

subsequently mediate the lysis of the infected host cell(Shawar et al. 1994). Nonclassical genes, on the other hand,are characteristically monomorphic or oligomorphic, haveweaker expression that is usually tissue-specific, and mayor may not be located within the MHC region. They take ona broad array of functions that are just beginning to begrasped, but do not necessarily include antigen presentation(Braud et al. 1999; Rodgers and Cook 2005).

The classical/nonclassical distinction does not reflectindependent evolutionary groupings, but has rather devel-oped as a simple means of separating genes based strictlyon like characteristics and function. In fact, class Ia and Ibloci tend to form monophyletic clusters, even within manyshallowly diverged taxa. This suggests that class I genesare continually re-expanding and diversifying from alimited number of ancestral copies in different evolution-ary lines. However, in mammals—by far the best studiedgroup in terms of MHC organization and function—somenonclassical genes are known to have emerged veryrecently, while others indicate a very ancient divergence(Nei et al. 1997; Rodgers and Cook 2005). This is not thecase for classical loci where orthology is rare or absent inmost groups and only exists at the level of closely relatedspecies.

The complexity of the class I lineage is, in part, a productof the dynamic nature of the MHC region as a whole. Acomparison of mapping data from representatives of themajor vertebrate groups has shown that the relative position,copy number, and size of the different MHC gene classeshave all been subject to radical change (Flajnik 2004;Flajnik and Kasahara 2001; Kelley et al. 2005). Still, inmany species, class I loci have a particularly rapid turnoverrate compared to their class II counterparts (Takahashi et al.2000; but see Bos and Waldman 2006 and Shum et al.2001). Nei et al. (1997) ascribed this to a birth-and-deathprocess where genes are frequently duplicated and subse-quently lost or retained as pseudogenes. Such a mode ofevolution could account for the close relationship amongclass Ia and Ib genes in many species and the widevariation in divergence times inferred for nonclassical loci.However, concerted evolution within and between loci haslong been cited as an important generator of similarityamong class I genes within a species as well (Baltimore1981; Hess and Edwards 2002; Rada et al. 1990).

MHC research has not been evenly distributed acrossvertebrate taxa, but studies from many underrepresentedgroups such as amphibians and non-avian reptiles areslowly emerging (Hauswaldt et al. 2007; Miller et al.2005, 2006; Ohta et al. 2006). The order Squamata, whichincludes lizards, snakes, and amphisbaenians, is one of theleast represented groups in the literature. Yet, it is by far themost diversified of the four non-avian reptile orders,containing more than 7,000 species worldwide—more than

all mammals together (Vitt et al. 2003). Characterization ofclass I genes in this group are necessary for severalpurposes. First, it will help establish whether the rapidturnover of class I loci is ubiquitous across vertebrates.Second, it will lead to a greater overall description of thesquamate MHC region and its structure. Squamates, alongwith the tuatara (order: Sphenodontia), are the third majorlineage of amniotes, in addition to archosaurs (birds andcrocodilians) and mammals. A comparison of the organi-zation and composition of MHC genes in these three groupswill ultimately allow us to infer the ancestral structure ofthe amniote MHC. Finally, due to their high level ofpolymorphism and central role in the immune response,MHC genes have become an important tool for studyinggenetic diversity and disease interactions in natural pop-ulations. Development of MHC markers in a squamatespecies would help stimulate and facilitate such studies inthis group.

In this paper, we present class I cDNA data from threespecies in the squamate subfamily Iguaninae (iguanas): theGalápagos marine iguana (Amblyrhynchus cristatus), theGalápagos land iguana (Conolophus subcristatus), andthe green iguana (Iguana iguana). The first two are sistertaxa and are both endemic to the Galápagos archipelago(Rassmann 1997). Marine iguanas have a unique naturalhistory among reptiles, as they inhabit the coastline ofislands, often in large aggregates, and forage exclusively onmarine algae (Wikelski 1999; Wikelski and Trillmich1994). In contrast, land iguanas are completely terrestrialand much more solitary (Snell et al. 1984). The greeniguana is an arboreal species that is common in the pettrade. It occupies much of the new world tropics rangingfrom South America northward to Mexico, including aninvasive population in southern Florida (Burghardt andRand 1982).

Materials and methods

RNA isolation and cDNA synthesis

Blood was taken from a single representative of three differentIguana species: a Galápagos marine iguana (A. cristatus) fromthe island of Santa Cruz, Galápagos; a captive Galápagosland iguana (C. subcristatus) of unknown island origin;and a captive green iguana (I. iguana) from the BeardsleyZoo in Connecticut. Total RNA was extracted from eachspecimen using the Tri Reagent BD kit (Molecular ResearchCenter, Cincinnati, OH, USA) and the protocol provided.Complementary DNA (cDNA) was synthesized using theSuperScript III RT enzyme and reagents provided inthe GeneRacer kit (Invitrogen, Carlsbad, CA, USA). The

372 Immunogenetics (2008) 60:371–382

complete kit protocol was followed for generating first-strand cDNA pools with intact 5′ ends except that bothrandom hexamers and the oligo dT primer provided in the kitwere used in a single reverse transcription reaction to obtainfull-length cDNA fragments.

5′ and 3′ RACE of marine iguana cDNA

To design gene-specific primers for 3′ RACE, a small frag-ment [201 base pairs (bp)] of the class I α-2 domain wasamplified by polymerase chain reaction (PCR) using theforward primer described in Radtkey et al. (1996) and adegenerate primer (MHCI-R3: 5′-ASGTAYYTCBBCAGCCACT-3′) designed from an alignment of a range ofvertebrate class I sequences. Sequences derived from thesmall amplified fragment were used to design two forward-facing gene-specific primers for use in 3′ RACE (IgNC-F1 5 ′ -TGYGAGCTGAGGAAAGATGGGAGCATAG-3′; IgNC-F2 5′-TTACCARTGTGCTTATGATGG-GAGG GAC-3′). These primers were used in a nestedPCR with oligo dT adaptor-specific primers provided inthe GeneRacer kit. The same primers and PCR conditionswere used to amplify a distantly related class I gene(Amcr-UA) gene that is currently being characterized(Glaberman and Du Pasquier, unpublished).

Sequence data derived from 3′ RACE products were thenused to design two gene-specific reverse primers (MHC1-NC-5R-R1 5 ′-ATGCTGCTTCAGGCTCGTTGG-3 ′;MHC1-NC-5R-R2 5′-TGGCATCAGCACTCCCTTTTCCAG-3′) in the 3′ untranslated region (UTR) for use in 5′RACE. In the first round of PCR, an adaptor-specific primerprovided in the GeneRacer kit was used in combination withthe gene-specific primer MHC1-NC-5R-R1. PCR wasperformed in a 50 μl reaction with the following reagentsand concentrations: 2.5 U Platinum Taq DNA PolymeraseHigh Fidelity (Invitrogen), 200 μM of each dNTP, 0.2 μMgene-specific primer, 2 mM MgSO4, 1× High Fidelity PCRbuffer. Cycling conditions were as follows: initial denatur-ation (denat) at 94°C for 2 min followed by five cycles of94°C for 30 s and 72°C for 2 min; five cycles of 94°C for30 s and 70°C for 2 min; 23 cycles of 94°C for 30 s, 59°Cfor 30 s, and 68°C for 2 min; and a final extension (ext)for 10 min at 68°C. The nested 5′ RACE step was carriedout with the nested 5′ adaptor primer and the primerMHC1-NC-5R-R2 using the same reagent concentrations as3′ RACE and the following cycling conditions: initial denatat 94°C for 2 min; 25 cycles with 30 s of denat at 94°C,30 s of annealing at 60°C, 2 min ext at 68°C; and a final extof 68°C for 20 min.

The nested RACE products from both the 5′ and 3′procedures were ligated into the pCR4-TOPO vector,transformed, and grown overnight on Luria–Bertani (LB)plates following the manufacturer’s protocol for the TOPO

TA cloning kit for sequencing (Invitrogen). The nestedRACE products from both the 5′ and 3′ procedures wereligated into the pCR4-TOPO vector, transformed, andgrown overnight on LB plates following the manufacturer’sprotocol for the TOPO TA cloning kit for sequencing(Invitrogen). Approximately ten clones each from 5′ and 3′RACE were added to 25 μl of water and heated at 94°C for10 min to lyse cells. One microliter of this solution wasadded to a 25 μl PCR reaction containing 1.5 U AmpliTaqGold DNA polymerase (Applied Biosystems, Foster City,CA, USA), 200 μM of each dNTP, 1 μM standard T3 andT7 primers, 1.5 mM MgCl2, and 1× PCR buffer withoutMgCl2. Cycling conditions were as follows: initial denat at94°C for 10 min, followed by 30 cycles with 30 s of denatat 94°C, 30 s of annealing at 55°C, and 90 s ext at 72°C anda final ext of 72°C for 10 min. PCR products were purifiedby filtration using either the QIAquick PCR purification kit(Qiagen, Valencia, CA, USA) or Millipore (Billerica, MA,USA) filter plate and sequenced using a 3730 DNAanalyzer (Applied Biosystems).

Sequence data were then used to design primers in boththe 5′ (MHC1-Gr1a-F1 TCTCTGTGGCTCAGCACTTGAACCTG and MHC1-Gr1b-F1 5′-ATACGCAGATAGACGGAGGAGGGAG-3′) and 3′ (MHC1-Gr1-R1TCCAAACACTTCTCAACATCTATACAGGTG) UTRsof the class I RACE fragments. The two 5′ UTR primerswere intended to target two vastly different 5′ regions foundin the 5′ RACE clones. These primers were both used incombination with the conserved MHC1-Gr1-R1 primer onfirst-strand cDNA with the same reagent concentrations asfor RACE and the following cycling conditions: initial denatat 94°C for 2 min; 35 cycles with 30 s of denat at 94°C, 30 sof annealing at 60°C, 2 min ext at 68°C; and a final ext of68°C for 20 min. The resulting fragments were cloned, andat least ten colonies were amplified and sequenced asdescribed above.

Amplification of conolophus and iguana class I sequences

Primers MHC1-Gr1a-F1 and MHC1-Gr1b-F1 were used incombination with MHC1-Gr1-R1 to amplify first-strandcDNA from a green iguana specimen as described above,while only the MHC1-Gr1b-F1 primer successfully ampli-fied the land iguana cDNA. All PCR fragments were clonedand sequenced in the same manner as the amplified marineiguana cDNA.

Data analysis

All class I cDNA sequences were edited in the programSEQUENCHER 4.2.2 (Gene Codes Corporation, Ann Arbor,MI, USA), and both nucleotide and amino acid translatedsequences were aligned using the program MUSCLE 3.6

Immunogenetics (2008) 60:371–382 373

(Edgar 2004). Uncorrected genetic distances (p distance)between all class I iguana sequences were calculatedseparately for each domain using the program MEGA 4.0(Tamura et al. 2007).

Full-length amino acid coding sequences derived fromiguana cDNA data were aligned with the followingvertebrate class I sequences from GenBank: Ameiva lizard(Ameiva ameiva), LC1 M81094; Tuatara (Sphenodonpunctatus) , Sppu-U*01 DQ145788, Sppu-U*02DQ145789; human (Homo sapiens), HLA-A U07161.This alignment was used to identify conserved class Iresidues with a known function according to Kaufmanet al. (1994).

An expanded data set of class I α-3 sequences from arange of vertebrates, including the species listed above,were used in Bayesian, maximum likelihood (ML), andNJ phylogenetic reconstruction. Additional sequences usedin this analysis were obtained from GenBank and are asfollows: Ameiva lizard, LC5 M81095, LC25 M91097;Water snake (Nerodia sipedon), SC1 M81099; Chinesesoft-shelled turtle (Pelodiscus sinensis), AB185243;Chicken (Gallus gallus), B-F10 X12780; Mallard(Anas platyrhynchos), Du2 AB115242; Great reed warbler(Acrocephalus arundinaceus), cN3 AJ005503; Axolotl(Ambystoma mexicanum), Amme-3 U83137; African clawedfrog (Xenopus laevis), UAA-1f L20733; mouse (Musmusculus), H2K L36312, H2-D1 NM_010380, H2-Q1NM_010390, H2-Q10 NM_010391; wallaby (Macropusrufogriseus), Maru-UB*01 L04952; platypus (Ornithorhyn-chus anatinus), Oran2-1 AY112715; possum (Trichosurusvulpecula), Trvu-UB AF359509; rainbow trout (Oncorhyn-chus mykiss), Onmy-UBA AF287487; zebrafish (Daniorerio), Dare-UBA NM131471; human, HLA-B7 U29057,HLA-Cw D50852. This set of sequences is similar to theone used in a study of tuatara class I genes by Miller et al.(2006) and was chosen for consistency.

The program MRMODELTEST v2 (Nylander 2004),which is based on code from the MODELTEST software(Posada and Crandall 1998), was used to compare the fit ofdifferent nucleotide substitution models to the vertebrateclass I α-3 dataset. The general time reversible model(GTR) with additional parameters for gamma distributionand fraction of invariable sites provided the best fit to thedata according to both the hierarchical likelihood ratio testand the Akaike information criterion. This model wasimplemented in a Bayesian framework using the programMRBAYES (Ronquist and Huelsenbeck 2003) as well as inML reconstruction using the TREEFINDER software (Jobbet al. 2004). For Bayesian analysis, the default softwaresettings were used, and the search was run for 2,000,000generations, with the first 10% of parameter samplesdiscarded as burn-in. In the ML analysis, 1,000 bootstrapreplicates were run to assess support for specific nodes. NJ

search and bootstrap analysis were carried out in PAUPusing the GTR substitution model with 500 replicates.

Tests of selection were carried out using the maximumlikelihood framework implemented in the program PAMLv.4 (Yang 2007). This approach allows for estimation ofdN/dS values (also referred to as ω) among multiple classesof amino acid sites without a priori designation of whichsites may be under selection (Nielsen and Yang 1998; Yanget al. 2000). This is particularly useful because the three-dimensional structure and peptide binding activity of class Imolecules have not been established in many vertebrategroups including squamates. Bjorkman et al. (1987)defined peptide-binding residues (PBR) using the humanHLA-A2 protein structure, but the involvement of many ofthese sites in antigen recognition has been shown to beinconsistent across vertebrates. Iguana class I sequenceswere fitted to two different models of codon site evolution,M1a and M2a, which have been recommended by Yang(2007) for comparison using the likelihood ratio test (LRT).The M1a model constrains a portion of amino acid sites tohave ω=1, while the remaining sites are allowed to varybetween 0 and 1 (0<ω<1). The M2a model allows for thesetwo site classes plus a third category where a proportion ofsites can have ω>1. If by LRT comparison M2a provides abetter fit to the data than M1a, it is considered support forpositive selection where the ω value of sites under selectionas well as their proportion over total sites is estimated bythe program. To carry out the LRT, two times the likelihooddifference between the two nested models are compared toa χ2 distribution with degrees of freedom equal to thedifference in the number of parameters between the models(M1a and M2a have two and four parameters, respectively).According to Yang et al. (2000), this test is conservativebecause it requires that sites in the ω>1 category be underselection over all branches of the phylogeny and will fail ifonly a few lineages have undergone positive selection. Ifthe LRT is significant, the Bayes empirical Bayes approach

Fig. 1 Amino acid alignment of iguana class I sequences with othervertebrates. Sequence abbreviations are as follows: Amcr Amblyrhyn-chus cristatus, Cosu Conolophus subcristatus, Igig Iguana iguana,Amam Ameiva ameiva, Sppu Sphenodon punctatus, Hosa Homosapiens. Coding domains are separated according to Koller and Orr(1985). Sequences Amcr UB*0402, Cosu-UB*0102, and Igig-UB*0102 are not shown in the alignment because they are verysimilar to one of the sequences displayed. Shaded columns indicateamino acid positions that are conserved or have expected functions.These amino acid positions also contain the following additionallabels: asterisks: conserved peptide-binding residues of antigen N andC termini; diamonds: salt bridge-forming residues; circles: disulfidebridge-forming cysteines; squares: N-glycosylation site; CD8:expected CD8 binding site. Vertical lines above alignment positionsindicate sites under positive selection. Information and accessionnumbers of vertebrate class I sequences can be found in “Materialsand methods”

b

374 Immunogenetics (2008) 60:371–382

used in the program assigns specific amino acid sites to theω>1 group with a certain posterior probability. One of theassumptions of this method of likelihood estimation is thatthe phylogenetic relationship among taxa is constant across

sites. Anisimova et al. (2003) showed that failure to ac-count for recombination in such tests of selection can leadto high rates of false positives. Due to the few sequencesused in this study, a full examination of recombination

Immunogenetics (2008) 60:371–382 375

between iguana class I sequences was not performed inthis study. However, separate analyses were carried out forα-1 and α-2 data, as their evolutionary histories may verywell be separated by recombination in the interveningintron.

Results

Iguana class I cDNA sequences

Amplification of marine iguana cDNA yielded five full-length class I coding sequences (Fig. 1) that were namedaccording to the nomenclature rules proposed by Kleinet al. (1990). Two of these sequences, Amcr-UB*01 andAmcr-UB*02, have nearly identical 5′ UTR and leadersequences, but differ somewhat in the α-1 (5.1%) and α-2(3.6%) domains which are typically involved in antigenrecognition. The other three sequences have similar 5′ UTRand signal peptides, but are divergent from Amcr-UB*01and Amcr-UB*02 in α-1 (2.2–10.9%) and α-2 (5.1–6.9%)portions of the gene. Two of these sequences—designatedAmcr-UB*0401 and Amcr-UB*0402—only differ fromeach other by a single non-synonymous change in the α-1region that was confirmed by sequencing multiple clones.The last of the three sequences, Amcr-UB*03, is verydifferent from the other two in the α-1 (9.1–9.4%) and α-2(6.5%) domains. All five marine iguana sequences areidentical in the usually conserved Ig-like α-3 domain of theclass I receptor molecule (Fig. 1). This suggests that thesesequences recently diverged from a common ancestor (butsee “Discussion”), and therefore, the name Amcr-UB wasmaintained for all sequences rather than assigning them toseparate groups (Amcr-UB, UC, UD, etc.). As five uniquesequences were derived from a single marine iguana, thisimplies that at least three class I loci exist in this species.Moreover, the two divergent 5′ ends in these sequencessuggest that Amcr-UB*01 and Amcr-UB*02 may belong toone locus or group of loci, while Amcr-UB*03, Amcr-UB*0401, and Amcr-UB*0402 belong to a group of at leasttwo other loci.

Four full-length coding sequences were derived fromGalápagos land iguana cDNA (Fig. 1). Two of these,labeled Cosu-UB*0101 and Cosu-UB*0102, are highlysimilar, differing from each other at only two positions.The other two sequences, Cosu-UB*02 and Cosu-UB*03,are highly divergent both from each other and from Cosu-UB*01 in the α-1 (11.6–13.4%) and α-2 (10.1–14.5%)domains, but not in the α-3 (0% to 0.4%) or Tm/Cyt(0–0.5%) regions. Therefore, in this species, we find ev-idence for at least two class I loci.

Finally, three unique full-length coding sequences wereidentified from green iguana cDNA (Fig. 1). Two of thesesequences, designated Igig-UB*0101 and Igig-UB*0102,differ at only two nucleotide positions. The third sequence,Igig-UB*02, is divergent from the others in the α-1 (20.3%)and α-2 (6.2%) domains, but not in α-3 (0.4%) or Tm/Cyt(0.5%) regions.

Class I relationships within and between iguana species

Conserved α-3 nucleotide data are nearly or completelyidentical within each of the three iguana species, butslightly divergent between species (2.9–8.7%). Despite theassumed sister relationship of Galápagos marine and landiguanas, α-3 sequences are less similar between thesespecies (8.0–8.3%) than between marine and green iguanas(2.9–3.3%). In addition, phylogenetic reconstruction in-cluding class I sequences from other squamates showsstrong support for a basal position of land iguana sequencesrelative to the other two iguana species (Fig. 2). However,the amino acid alignment of α-3 reveals that many of thedivergent sites between land iguanas and the other speciesare located in the portion of the class I molecule which isinvolved in CD8 binding.

Unlike the α-3 domain, the patterns of variation anddivergence in α-1 and α-2 vary greatly both within andbetween the three species. For example, α-1 sequences aremore differentiated within land iguanas (11.6–13.4%) thanwithin marine iguanas (2.2–10.9%); yet, the divergencebetween some pairs of land and marine iguana sequences(7.6–17%) is lower than within land iguanas, suggesting amore complicated evolutionary history for this region. Thesituation is similar for the α-2 domain. However, in both α-1 and α-2, the genetic affinity between marine and greeniguanas is not apparent, as is the case in the α-3 region.There is also evidence of shared polymorphism in theamino acid sequences of the α-1 and α-2 domains. Forexample, in the α-1 alignment, Amcr-UB*03 and Cosu-UB*02 both contain S, L, and K residues at positions 49,50, and 52, respectively, as well as D, M, and K residues atpositions 80, 82, and 83, respectively. In the α-2 alignment,Cosu-UB*02, Cosu-UB*03, and Igig-UB*0101 share D, P,

Fig. 2 Phylogenetic reconstruction using α-3 (exon 4) sequence datafrom the major vertebrate groups. The top number at each node is theBayesian posterior distribution value supporting the topology dis-played. The middle and bottom numbers are bootstrap values frommaximum likelihood and neighbor joining searches, respectively.Branch lengths are represented as expected number of substitutionsper site and were inferred along with topology in the Bayesiananalysis

b

376 Immunogenetics (2008) 60:371–382

Immunogenetics (2008) 60:371–382 377

S, and N residues at positions 155, 156, and 161,respectively. These are the more conspicuous examples ofresidue sharing between genes from different species, butthere are many more.

Class I relationships of iguanas and other vertebrates

The sequences of all three iguana species contain many ofthe residues that are known to be important for class Ireceptor structure and the binding of antigenic peptides.This includes the presence of four cysteine (C) residues atalignment positions 108, 172, 211, and 267 that areinvolved in intradomain disulfide bridge formation, twosalt bridges formed by histidine (H) and aspartic acid (D)residues at positions 5 and 33 in the α-1 domain, and 100and 126 in the α-2 domain, and an NQS glycosylation siteat positions 93–95. In the classical class I molecules ofvertebrates, there are nine amino acid positions that arehighly conserved and known to bind the C- and N-terminalresidues of antigenic peptides (Kaufman et al. 1994; Shumet al. 1999). They correspond to the following amino acidtypes and alignment positions: Y11, Y66, Y91, Y130,T150, K153, W154, Y167, and Y182. Compared to thecommon reference type found in human HLA-A, -B, and-C molecules, all iguana class I sequences possess eithereight or nine of these conserved residues, suggesting thatthey all encode classical class I molecules. However,

official designation of a locus as classical or nonclassicaldepends on the level and diversity of tissue expression,location in the MHC region, and the degree of polymor-phism in populations.

Phylogenetic analysis, which included class I α-3 datafrom a range of vertebrates, reveals that the cDNAsequences from the three iguana species are closely relatedand monophyletic (Fig. 2). This is not surprising given thegreat similarity of their α-3 nucleotide and amino acidsequences (Fig. 1). All three phylogenetic inferencemethods strongly support the monophyly of availablesquamate class I sequences, but their relationship to classI genes of other vertebrates is not well resolved, possibly asa result of the high level of divergence in the α-3 domain.

Selection on iguana class I sequences

The likelihood approach implemented in this study stronglysupports a pattern of diversifying selection in the α-1 andα-2 domains. In both cases, the selection model (M2a),which allows a portion of amino acid sites to have ω>1, fitsthe data better than the neutral model (M1a) based on theLRT (Table 1). This is the case when marine iguanasequences are analyzed separately as well as when alliguana class I sequences are combined. For marine iguanaα-1 sequences, 17% of amino acid sites were assigned tothe positively selected site class, with an estimated ω of

Table 1 Likelihood values and ω (non-synonymous to synonymous rate ratio) estimates of class I data under two different models of codonevolution

Sequences Model lnL dN/dS Parameter estimates

Marine iguanas α-1 Neutral −558.686 0.708 P0=0.292 p1=0.708ω0=0.000 ω1=1.000

Marine iguanas α-1 Selection −541.247a 7.644 p0=0.826 p1=0.003 p2=0.171ω0=1.000 ω1=1.000 ω2=39.796

Marine iguanas α-2 Neutral −550.774 0.272 p0=0.728 p1=0.272ω0=0.000 ω1=1.000

Marine iguanas α-2 Selection −543.762a 1.081 p0=0.898 p1=0.000 p2=0.102ω0=0.130 ω1=1.000 ω2=9.475

All iguanas α-1 Neutral −1138.088 0.593 p0=0.417 p1=0.583ω0=0.023 ω1=1.000

All iguanas α-1 Selection −1106.830a 2.071 p0=0.276 p1=0.585 P2=0.139ω0=0.000 ω1=1.000 ω2=10.727

All iguanas α-2 Neutral −1007.995 0.318 p0=0.713 p1=0.287ω0=0.043 ω1=1.000

All iguanas α-2 Selection −980.551a 1.124 p0=0.605 p1=0.295 p2=0.100ω0=0.039 ω1=1.000 ω2=8.079

Calculations were carried out separately for the α-1 and α-2 domains and for marine iguana sequences alone as well as all iguana sequencescombined. The “neutral” and “selection” labels correspond to the M1a and M2a models respectively in the program PAML (Yang 2007). The dN/dS column represents the average of this value over all sites and branches in the phylogeny. The “p” label reflects the proportion of sites assignedto a particular ω value. See Yang et al. (2000) for detailed method description.a Reflect a significant difference (α=0.05) between the neutral and selection models according to the likelihood ratio test.

378 Immunogenetics (2008) 60:371–382

nearly 40. This value is substantially lower in the combinediguana analysis (ω=10.727), but is still very high comparedto neutral expectations. Overall, the α-2 domain also showsstrong selection at approximately 10% of amino acid sites,but the level of positive selection is lower than for α-1.Codons assigned to the class of positively selected siteswith a posterior probability of >95% in either the separatemarine iguana analysis or complete iguana dataset arelabeled in the protein alignment of Fig. 1. The majority ofthese sites are predicted peptide-binding or upward facingresidues (nine out of 11 in α-1; five out of seven in α-2) asdefined for the human HLA-A2 structure (Bjorkman et al.1987); yet, the fact that some of the selected sites are notknown to be important for antigen recognition supports theidea that a priori designation of PBR positions, as employedin other popular tests of selection (e.g., Z test; Nei andKumar 2000), may lead to underestimates of selectivepressure.

Discussion

MHC class I data were collected from three membersof the Iguaninae subfamily: Galápagos marine and landiguanas and the common green iguana. The sequencesgenerated from each of these species appear to representfunctional genes, as they possess all of the proteindomains necessary to form a complete class I receptormolecule and have no premature termination signals. Inaddition, these genes are expressed, at least at the RNAlevel, as genetic data were derived from cDNA poolsisolated from whole blood. The major structural featuresthat distinguish class I molecules are intact in all of thesequences, including the conserved amino acid residuesthat are known to bind the terminal ends of antigenicpeptides for presentation to CD8+ T cells. Finally, thelarge ω (dN/dS) values at a portion of sites in the twopeptide-binding domains show that the sequences havebeen subject to strong positive selection, which isindicative of an active role in antigen recognition andprocessing. Therefore, we preliminarily classify the locirepresented by these sequences as class Ia. However, otherdescriptive information such as the breadth of tissueexpression and the position of these genes in the MHCregion are important criteria by which to designateclassical genes and therefore must be addressed in thefuture.

In each of the three species, there is evidence of multipleclass I loci: at least three in marine iguanas and two in bothland and green iguanas. Within species, the portion of thesequences that correspond to the α-1 and α-2 domains isquite variable, while the α-3 and Tm/Cyt regions are eitheridentical or nearly so. There is no clear pattern of homology

within or between species in the α-1 and α-2 domains, butthere is slight divergence between species in the α-3 andTm/Cyt gene segments.

Data from the α-3 domain are frequently used to recoverevolutionary relationships among class I sequences acrossspecies because they are relatively conserved in comparisonto the two peptide-binding domains, which are often underbalancing selection. In this study, from α-3 data alone, itappears that class I loci are more closely related within thethree iguana species than between them. Such a pattern ofexclusivity among class I sequences of different species istypical in many vertebrate groups and is thought to be theresult of a birth-and-death mode of evolution. In thisprocess, lineages are continually expanding through dupli-cation, many of which are eventually lost or renderednonfunctional. For example, 11 class I loci were identifiedin the MHC of the gray short-tailed opossum (Monodelphisdomestica), and all of these genes diverged after the split ofmarsupials from eutherian mammals (Belov et al. 2006). Inbirds, a comparative analysis of the chicken and quail MHCregions showed that the latter possesses approximately 14class I loci whose origin postdates the separation of thesetwo species despite the fact that they occupy the samefamily (Shiina et al. 1999, 2004). These are just two of themany cases of broad independent class I expansions that arerepresented in the literature (Hess and Edwards 2002;Hughes and Nei 1989; Nei et al. 1997). However, many ofthese examples involve species that have large or modestdivergence times, and even the split of chickens and quailsis on the order of 80 million years (Shiina et al. 2004). Atthe level of closely related species, as is the case in ourstudy, there are fewer class I studies, and the mechanismsthat lead to independent lineages have been a subject ofintense debate.

Nei et al. (1997) argued that the birth-and-death modeladequately explains the patterns of class I relationshipswithin and between species of mammals. Under this model,the probability of lineage extinction should increase withtime. They suggest that the orthology seen among the A, B,and C loci in humans, chimpanzees, and gorillas is due tothe recent divergence of these species and that there has notbeen sufficient time for the extinction and expansion ofdifferent lineages in any of these taxa. This could explainthe breakdown in orthology at the C locus, which is notpresent in orangutans, whose departure from the human linedates back 13 million years. Finally, the MHC of cotton-toptamarins (Saguinus oedipus) contains a completely inde-pendent class I lineage which diverged before the origin ofthe A, B, and C loci, approximately 40 years ago (Goodmanet al. 1998).

Although land and marine iguanas are sister taxa, theyare estimated to have diverged between 10 and 20 millionyears ago on islands in the Galápagos archipelago that are

Immunogenetics (2008) 60:371–382 379

now submerged (Rassmann 1997). This is already in thetime frame in which orthology begins to dissolve in thegroup of mammals discussed, and therefore, perhaps, it isnot unreasonable that the α-3 sequences in each of the threeiguana species are the product of separately expanding classI lineages. However, the shared polymorphism observedbetween α-1 and α-2 sequences of different iguana specieswould have to be the product of ancestral polymorphism atthe original locus where multiple allelic lineages weremaintained after species divergence. Such cases of trans-species polymorphism are one of the hallmarks of MHCgenes and are an expected consequence of strong balancingselection (Bernatchez and Landry 2003; Hughes and Nei1988; Hughes and Yeager 1998; Piertney and Oliver 2006).The maintenance of ancestral alleles across species in theα-1 and α-2 regions, but not in the α-3 domain, would bepossible if there were one or more recombination break-points separating the evolution of these regions. This isplausible considering that each of the three domains iscoded for by a separate exon. Alternatively, convergentevolution of amino acid residues could also produce theshared polymorphism in the α-1 and α-2 domains and haspreviously been observed at antigen-binding sites inmammalian MHC genes (e.g., Kriener et al. 2000).

Concerted evolution is the other main mechanism thatmay explain the similarity of MHC loci within iguaninespecies and has been convincingly demonstrated in bothclass I (e.g., Joly and Rouillon 2006) and class II (e.g.,Wittzell et al. 1999) genes (reviewed in Hess and Edwards2002). Under this scenario, multiple class I loci must haveexisted before speciation. After species divergence, the α-3domains would have then become homogenized in eachspecies through interlocus gene conversion. In contrast, theα-1 and α-2 domains show no clear signal of concertedevolution, and it is likely that their polymorphism has beenmaintained across species boundaries.

Although neither of the proposed scenarios can be ruledout, a prominent role for concerted evolution seems moreplausible. In the five marine iguana sequences, the α-3nucleotide sequences are identical. Under the birth-and-deathmodel, the duplication of loci would have to have been veryrecent and the level of purifying selection incredibly strongto prevent the accumulation of any differences among loci inthis domain. Therefore, a recent episode of gene conversionbetween loci seems more likely. However, the collection ofclass I data from both exons and introns at the populationlevel is necessary to understand the different evolutionaryforces shaping these genes. Recent studies have used suchdata to evaluate the effects of inter- and intragenic geneconversion and recombination on MHC sequences (Millerand Lambert 2004a; Reusch and Langefors 2005).

The α-3 sequences are more closely related betweenmarine and green iguanas than they are between marine and

land iguanas despite the fact that the latter pair are sistertaxa. This pattern is observed both in terms of geneticdistance and phylogenetic relationships. A closer look atthe amino acid sequences in this region reveals that anumber of these differences occur at or near α-3 residuesthat interact with the CD8 co-receptor on T cells. Thisportion of the gene is known to vary widely amongvertebrates, which is thought to reflect co-evolution of theclass I receptor and the CD8 binding pocket (Kaufmanet al. 1994). Under the birth-and-death model, differentancestral CD8 types may have become fixed in marine andgreen iguanas versus land iguanas before duplication. Onthe other hand, if concerted evolution is acting, differentloci may have provided the source α-3 sequence for theother homogenized gene copies. Either of these scenarioscould cause discordance between the gene and speciestrees. In addition, a period of strong directional selection inthe α-3 domain of land iguanas could also produce such aconflict.

The class I data presented here can serve as an importantlaunching point for studies of conservation genetics, sexualselection, and disease in each of the three iguana species.Over the last several decades, the MHC has emerged as animportant model system for evaluating the relative influ-ence of natural selection versus drift and migration on thelevels of genetic variation in populations (Bernatchez andLandry 2003; Piertney and Oliver 2006). This is importantwhen considering that selection may have its greatest effecton functionally important genes. A well-known study onisland foxes (Urocyon littoralis) showed that balancingselection is the likely cause of high class II MHC diversityin a population that was shown to be monomorphic atneutral loci (Aguilar et al. 2004). Overall, however, there isa great deal of conflicting evidence over the contribution ofselection to the maintenance of genetic diversity in smallpopulations (Campos et al. 2006; Miller and Lambert2004b; Richardson and Westerdahl 2003; Smulders et al.2003; Wenink et al. 1998).

In the past century, Galápagos land iguana populationswere severely reduced in numbers, mainly as a result ofcompetition, predation, and habitat alteration by humansand introduced animals (Snell et al. 1984). A captivebreeding program that began in 1976 eventually led to thesuccessful reintroduction of these animals into part of theirformer range. A study of genetic variability at neutral loci iscurrently being carried out (Gentile et al., unpublished), butMHC data could be used to understand the degree to whichselection maintained variation following these populationreductions.

Marine iguanas are more abundant than their land-dwelling sister taxon, but they too present an interestingsystem in which MHC data could be applied. A number ofstudies have described a lek mating system in this species

380 Immunogenetics (2008) 60:371–382

where males defend territory and compete for femalesbased on behavior as well as body size and condition(Wikelski et al. 1996; Wikelski and Trillmich 1997). The“good genes” hypothesis originally proposed by Hamiltonand Zuk (1982) has inspired many studies that attempt toassociate certain MHC genotypes with phenotypic traitsthat determine female choice (Penn and Potts 1999). Such aconnection would suggest that females use secondarysexual characters to identify males with specific MHCgenotypes that confer greater resistance to disease (vonSchantz et al. 1996). Given the strong element of femalechoice in marine iguanas, studies associating morphology,MHC genotype, and mating success could be veryinformative.

Green iguanas are widespread, and there is a great dealknown about their captive care, husbandry, and medicine(Jacobson 2003). This species could serve as an importantreptile experimental system in which to investigate theinteraction between disease and MHC genotype-dependentresistance. Such studies are rapidly increasing in mammalsand birds, as the number of described genomes and MHCregions continues to expand. But few if any real modelsystems exist in non-avian reptiles.

At the higher taxonomic level, we hope that the datapresented here can serve as a seed for characterizing class Igenes in other squamates. This will provide a broaderpicture of the forces that shape MHC evolution in thisgroup. In turn, an improved understanding of the squamateMHC will enable a more fruitful and comprehensivecomparison of how the immune systems of the majorvertebrate lineages have evolved.

Acknowledgments Rob Tomas and the Beardsley Zoo generouslyprovided the green iguana blood sample. Cruz Marquez aided incollection of the marine and land iguana blood specimens. The facilitiesprovided by the Galapagos Genetics, Epidemiology and PathologyLaboratory in the Galápagos National Park were essential in the timelyextraction of RNA. Maria Moreno and Stephen Dellaporta gave helpfuladvice on the RACE procedure. Louis Du Pasquier, Ylenia Chiari, andMichel Slotman contributed important input and interpretation ofresults. Funding was provided by the Yale Institute for BiosphericStudies as well as Project AWARE and National Geographic Society(Award no 7589-04) grants to AC.

References

Aguilar A, Roemer G, Debenham S, Binns M, Garcelon D, WayneRK (2004) High MHC diversity maintained by balancingselection in an otherwise genetically monomorphic mammal.Proc Natl Acad Sci USA 101:3490–3494

Anisimova M, Nielsen R, Yang ZH (2003) Effect of recombination onthe accuracy of the likelihood method for detecting positiveselection at amino acid sites. Genetics 164:1229–1236

Baltimore D (1981) Gene conversion—some implications for immu-noglobulin genes. Cell 24:592–594

Beckman EM, Porcelli SA, Morita CT, Behar SM, Furlong ST,Brenner MB (1994) Recognition of a lipid antigen by CD1-restricted alpha beta+ T cells. Nature 372:691–694

Belov K, Deakin JE, Papenfuss AT, Baker ML, Melman SD, Siddle HV,Gouin N, Goode DL, Sargeant TJ, Robinson MD, Wakefield MJ,Mahony S, Cross JGR, Benos PV, Samollow PB, Speed TP,Graves JAM, Miller RD (2006) Reconstructing an ancestralmammalian immune supercomplex from a marsupial majorhistocompatibility complex. Plos Biol 4:317–328

Bernatchez L, Landry C (2003) MHC studies in nonmodel vertebrates:what have we learned about natural selection in 15 years? J EvolBiol 16:363–377

Bjorkman PJ, Saper MA, Samraoui B, Bennett WS, Strominger JL,Wiley DC (1987) The foreign antigen-binding site and T-cellrecognition regions of class-i histocompatibility antigens. Nature329:512–518

Bos DH, Waldman B (2006) Evolution by recombination andtransspecies polymorphism in the MHC class I gene of Xenopuslaevis. Mol Biol Evol 23:137–143

Braud VM, Allan DS, McMichael AJ (1999) Functions of nonclas-sical MHC and non-MHC-encoded class I molecules. Curr OpinImmunol 11:100–108

Burghardt GM, Rand AS (1982) Iguanas of the world: their behavior,ecology, and conservation. Noyes Publications, Park Ridge, NJ

Campos JL, Posada D, Moran P (2006) Genetic variation at MHC,mitochondrial and microsatellite loci in isolated populations ofbrown trout (Salmo trutta). Conserv Genet 7:515–530

Edgar RC (2004) MUSCLE: multiple sequence alignment with highaccuracy and high throughput. Nucleic Acids Res 32:1792–1797

Flajnik M (2004) Comparative genomics of the MHC. TissueAntigens 64:328

Flajnik MF, Kasahara M (2001) Comparative genomics of the MHC:glimpses into the evolution of the adaptive immune system.Immunity 15:351–362

Goodman M, Porter CA, Czelusniak J, Page SL, Schneider H,Shoshani J, Gunnell G, Groves CP (1998) Toward a phylogeneticclassification of primates based on DNA evidence complementedby fossil evidence. Mol Phylogenet Evol 9:585–598

Hamilton WD, Zuk M (1982) Heritable true fitness and bright birds—a role for parasites. Science 218:384–387

Hauswaldt JS, Stuckas H, Pfautsch S, Tiedemann R (2007) Molecularcharacterization of MHC class II in a nonmodel anuran species, thefire-bellied toad Bombina bombina. Immunogenetics 59:479–491

Hess CM, Edwards SV (2002) The evolution of the major histocom-patibility complex in birds. Bioscience 52:423–431

Hughes AL, Nei M (1988) Pattern of nucleotide substitution at majorhistocompatibility complex class I loci reveals overdominantselection. Nature 335:167–170

Hughes AL, Nei M (1989) Evolution of the major histocompatibilitycomplex—independent origin of nonclassical class-I genes indifferent groups of mammals. Mol Biol Evol 6:559–579

Hughes AL, Yeager M (1998) Natural selection at major histocompat-ibility complex loci of vertebrates. Annu Rev Genet 32:415–435

Jacobson ER (2003) Biology, husbandry, and medicine of the greeniguana. Krieger, Malabar, FL

Jobb G, von Haeseler A, Strimmer K (2004) TREEFINDER: apowerful graphical analysis environment for molecular phyloge-netics. BMC Evol Biol 4:18

Joly E, Rouillon V (2006) The orthology of HLA-E and H2-Qa1 ishidden by their concerted evolution with other MHC class Imolecules. Biol Direct 1:2

Kaufman J, Salomonsen J, Flajnik M (1994) Evolutionary conserva-tion of MHC class I and class II molecules—different yet thesame. Semin Immunol 6:411–424

Kelley J, Walter L, Trowsdale J (2005) Comparative genomics of majorhistocompatibility complexes. Immunogenetics 56:683–695

Immunogenetics (2008) 60:371–382 381

Klein J (1986) Natural history of the major histocompatibilitycomplex. Wiley, New York

Klein J, Bontrop RE, Dawkins RL, Erlich HA, Gyllensten UB, HeiseER, Jones PP, Parham P, Wakeland EK, Watkins DI (1990)Nomenclature for the major histocompatibility complexes ofdifferent species—a proposal. Immunogenetics 31:217–219

Koller BH, Orr HT (1985) Cloning and complete sequence of anHLA-A2 gene: analysis of two HLA-A alleles at the nucleotidelevel. J Immunol 134:2727–2733

Kriener K, O’hUigin C, Tichy H, Klein J (2000) Convergent evolutionof major histocompatibility complex molecules in humans andNew World monkeys. Immunogenetics 51:169–178

Miller HC, Lambert DM (2004a) Gene duplication and geneconversion in class II MHC genes of New Zealand robins(Petroicidae). Immunogenetics 56:178–191

Miller HC, Lambert DM (2004b) Genetic drift outweighs balancingselection in shaping post-bottleneck major histocompatibilitycomplex variation in New Zealand robins (Petroicidae). Mol Ecol13:3709–3721

Miller HC, Belov K, Daugherty CH (2005) Characterization of MHCclass II genes from an ancient reptile lineage, Sphenodon(tuatara). Immunogenetics 57:883–891

Miller HC, Belov K, Daugherty CH (2006) Proceedings of the SMBETri-National Young Investigators’ Workshop 2005. MHC class Igenes in the tuatara (Sphenodon spp.): evolution of the MHC inan ancient reptilian order. Mol Biol Evol 23:949–956

Nei M, Kumar S (2000) Molecular evolution and phylogenetics.Oxford University Press, Oxford

Nei M, Gu X, Sitnikova T (1997) Evolution by the birth-and-deathprocess in multigene families of the vertebrate immune system.Proc Natl Acad Sci USA 94:7799–7806

Nielsen R, Yang ZH (1998) Likelihood models for detectingpositively selected amino acid sites and applications to theHIV-1 envelope gene. Genetics 148:929–936

Nylander JAA (2004) MrModeltest v2. Program distributed by theauthor. Evolutionary Biology Centre, Uppsala University

Ohta Y, Goetz W, Hossain MZ, Nonaka M, Flajnik MF (2006)Ancestral organization of the MHC revealed in the amphibianXenopus. J Immunol 176:3674–3685

Penn DJ, Potts WK (1999) The evolution of mating preferences andmajor histocompatibility complex genes. Am Nat 153:145–164

Piertney SB, Oliver MK (2006) The evolutionary ecology of the majorhistocompatibility complex. Heredity 96:7–21

Posada D, Crandall KA (1998) MODELTEST: testing the model ofDNA substitution. Bioinformatics 14:817–818

Rada C, Lorenzi R, Powis SJ, Vandenbogaerde J, Parham P, HowardJC (1990) Concerted evolution of class-i genes in the majorhistocompatibility complex of murine rodents. Proc Natl AcadSci USA 87:2167–2171

Radtkey RR, Becker B, Miller RD, Riblet R, Case TJ (1996) Variationand evolution of class I MHC in sexual and parthenogeneticgeckos. Proc R Soc Lond B–Biol Sci 263:1023–1032

Rassmann K (1997) Evolutionary age of the Galapagos iguanaspredates the age of the present Galapagos Islands. MolPhylogenet Evol 7:158–172

Reusch TBH, Langefors A (2005) Inter- and intralocus recombina-tion drive MHC class IIB gene diversification in a teleost, thethree-spined stickleback Gasterosteus aculeatus. J Mol Evol 61:531–U45

Richardson DS, Westerdahl H (2003) MHC diversity in twoAcrocephalus species: the outbred great reed warbler and theinbred Seychelles warbler. Mol Ecol 12:3523–3529

Rodgers JR, Cook RG (2005) MHC class IB molecules bridge innateand acquired immunity. Nat Rev Immunol 5:459–471

Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogeneticinference under mixed models. Bioinformatics 19:1572–1574

Shawar SM, Vyas JM, Rodgers JR, Rich RR (1994) Antigenpresentation by major histocompatibility complex class I-Bmolecules. Annu Rev Immunol 12:839–880

Shiina T, Shimizu C, Oka A, Teraoka Y, Imanishi T, Gojobori T,Hanzawa K, Watanabe S, Inoko H (1999) Gene organization ofthe quail major histocompatibility complex (MhcCoja) class Igene region. Immunogenetics 49:384–394

Shiina T, Shimizu S, Hosomichi K, Kohara S, Watanabe S, Hanzawa K,Beck S, Kulski JK, Inoko H (2004) Comparative genomic analysisof two avian (quail and chicken) MHC regions. J Immunol172:6751–6763

Shum BP, Rajalingam R, Magor KE, Azumi K, Carr WH, Dixon B,Stet RJ, Adkison MA, Hedrick RP, Parham P (1999) A divergentnon-classical class I gene conserved in salmonids. Immunoge-netics 49:479–490

Shum BP, Guethlein L, Flodin LR, Adkison MA, Hedrick RP, NehringRB, Stet RJM, Secombes C, Parham P (2001) Modes ofsalmonid MHC class I and II evolution differ from the primateparadigm. J Immunol 166:3297–3308

Sieling PA, Chatterjee D, Porcelli SA, Prigozy TI, Mazzaccaro RJ,Soriano T, Bloom BR, Brenner MB, Kronenberg M, Brennan PJet al (1995) CD1-restricted T cell recognition of microbiallipoglycan antigens. Science 269:227–230

Smulders MJM, Snoek LB, Booy G, Vosman B (2003) Complete lossof MHC genetic diversity in the common hamster (Cricetuscricetus) population in The Netherlands. Consequences forconservation strategies. Conserv Genet 4:441–451

Snell HL, Snell HM, Tracy CR (1984) Variation among populations ofGalapagos land iguanas (Conolophus)—contrasts of phylogenyand ecology. Biol J Linn Soc 21:185–207

Takahashi K, Rooney AP, Nei M (2000) Origins and divergence timesof mammalian class II MHC gene clusters. J Heredity 91:198–204

Tamura K, Dudley J, Nei M, Kumar S (2007) MEGA4: molecularevolutionary genetics analysis (MEGA) software version 4.0.Mol Biol Evol 24:1596–1599

Vitt LJ, Pianka ER, Cooper WE, Schwenk K (2003) History and theglobal ecology of squamate reptiles. Am Nat 162:44–60

von Schantz T, Wittzell H, Goransson G, Grahn M, Persson K (1996)MHC genotype and male ornamentation: genetic evidence forthe Hamilton–Zuk model. Proc R Soc Lond B–Biol Sci 263:265–271

Wenink PW, Groen AF, Roelke-Parker ME, Prins HHT (1998) Africanbuffalo maintain high genetic diversity in the major histocom-patibility complex in spite of historically known populationbottlenecks. Mol Ecol 7:1315–1322

Wikelski M (1999) Influences of parasites and thermoregulationon grouping tendencies in marine iguanas. Behav Ecol 10:22–29

Wikelski M, Trillmich F (1994) Foraging strategies of the Galápagosmarine iguana (Amblyrhynchus cristatus)—adapting behavioralrules to ontogenic size change. Behaviour 128:255–279

Wikelski M, Trillmich F (1997) Body size and sexual size dimorphismin marine iguanas fluctuate as a result of opposing natural andsexual selection: an island comparison. Evolution 51:922–936

Wikelski M, Carbone C, Trillmich F (1996) Lekking in marineiguanas: female grouping and male reproductive strategies. AnimBehav 52:581–596

Wittzell H, Bernot A, Auffray C, Zoorob R (1999) Concertedevolution of two MHC class II B loci in pheasants and domesticchickens. Mol Biol Evol 16:479–490

Yang ZH (2007) PAML 4: phylogenetic analysis by maximumlikelihood. Mol Biol Evol 24:1586–1591

Yang ZH, Nielsen R, Goldman N, Pedersen AMK (2000) Codon-substitution models for heterogeneous selection pressure at aminoacid sites. Genetics 155:431–449

382 Immunogenetics (2008) 60:371–382