Greater Prairie Chickens Have a Compact MHC-B with a Single Class IA Locus

Author's personal copy

Immunogenetics DOI 10.1007/s00251-012-0664-7

ORIGINAL PAPER

Greater prairie chickens have a compact MHC-B with a single class IA locus

J. A. Eimes & K. M. Reed & K. M. Mendoza & J. L. Bollmer & L. A. Whittingham & Z. W. Bateson & P. O. Dunn

Received: 18 July 2012 /Accepted: 22 October 2012 # Springer-Verlag Berlin Heidelberg 2012

Abstract The major histocompatibility complex (MHC) Introduction plays a central role in innate and adaptive immunity, but relatively little is known about the evolution of the number The major histocompatibility complex (MHC) is a family of and arrangement of MHC genes in birds. Insights into the genes found in all jawed vertebrates (Trowsdale 1988)that evolution of the MHC in birds can be gained by comparing facilitates the adaptive immune response to foreign parasites the genetic architecture of the MHC between closely related and pathogens (Janeway et al. 2005). Comparisons across species. We used a fosmid DNA library to sequence a 60.9-kb distantly related taxa have revealed major differences in the region of the MHC of the greater prairie chicken (Tympanuchus size, number, and arrangement (i.e., architecture) of MHC loci cupido), one of five species of Galliformes with a physically (Kelley et al. 2005). For example, the MHC in galliform birds mapped MHC. Greater prairie chickens have the smallest core appears to be dramatically more compact than that of mam- MHC yet observed in any bird species, and major changes are mals. The core MHC-B of the domestic chicken (Gallus observed in the number and arrangement of MHC loci. In gallus) is 1/20th the size of the MHC in humans (HLA) and particular, the greater prairie chicken differs from other contains just 19 coding genes compared with over 60 in Galliformes in the deletion of an important class I antigen humans (Guillemot et al. 1988;Kelleyetal.2005). binding gene. Analysis of the remaining class IA gene in a The compact nature of the MHC in domestic chickens and population of greater prairie chickens in Wisconsin, USA strong associations between individual haplotypes and disease revealed little evidence for selection at the region responsible resistance led to the description of the chicken MHC as a for antigen binding. “minimal essential MHC” (Kaufman et al. 1995). Kaufman (1999) hypothesized that in birds, fewer antigen presenting molecules will recognize fewer pathogens, but with lessened Keywords MHC . Fosmidlibrary .Galliformes .Comparative risk of autoimmune disorders, while in mammals, the larger genomics . Tympanuchus cupido suite of expressed molecules protects against more pathogens but at the expense of greater risk of autoimmunity. For example, in humans, there are up to six expressed MHC class I genes, and several associations with diseases and specific loci have been described (reviewed by Shiina et al. 2004a). * : : : Domestic chickens, in contrast, have only two class I genes, J. A. Eimes ( :) J. L. Bollmer L. A. Whittingham Z. W. Bateson P. O. Dunn BF1 and BF2,andBF2 is dominantly expressed, and it has Department of Biological Sciences, been hypothesized that BF2 plays a role in resistance to Rous University of Wisconsin-Milwaukee, sarcoma virus and Marek's disease (Kaufman and Venugopal Milwaukee, WI, USA e-mail: [email protected] 1998; Wallny et al. 2006). : In addition to the domestic chicken, the core MHC-B has K. M. Reed K. M. Mendoza been sequenced in four additional species within the Department of Veterinary and Biomedical Sciences, College of Veterinary Medicine, University of Minnesota, Galliformes: domestic turkey (Meleagris gallopavo), golden St. Paul, MN, USA pheasant (Chrysolophus pictus), the black grouse (Tetrao Author's personal copy

Immunogenetics tetrix, Genbank accession number AC:JQ028669), and overnight at 50 V using the Fosmid Control DNA (Epicentre) Japanese quail (Coturnix japonica), and the MHC has many (40 kb) as a size marker. The gDNA was additionally sheared fewer genes, less intergenic space and smaller introns than by vortexing for 40-s intervals until an intact, 40 kb fragment the mammalian MHC (Shiina et al. 2004a, b; Chaves et al. was generated. We increased the efficiency of the insert liga- 2009;Yeetal.2012). Assembled sequences of the Galliform tion (on which all subsequent cloning steps are dependent) by MHC-B suggest that even among closely related taxa the gene omitting further size selection of the gDNA. The MaxPlax number, order and orientation are variable. For example, the Lambda Packaging Extracts, EPI300-T1 Plating Strain MHC-B of the turkey and Japanese quail contain several BG (pCC1FOS, Epicentre) preferentially accepts inserts of 30– genes between Blec2 and BTN2, while the domestic chicken 40 kb and, thus, size selects the fragment during the packaging has just one (Shiina et al. 2004a, b; Chaves et al. 2009). The step. domestic chicken has two MHC-B class IIB genes (Kaufman et al. 1999), whereas the turkey and golden pheasant have at Cloning least three (Chaves et al. 2009;Yeetal.2012), and in the Japanese quail, block duplications have generated up to eight Single-copy gDNA fragments (~40 kb) were cloned using class IIB genes in some individuals (Shiina et al. 2004a, b; the CopyControl™ Fosmid Library Production Kit with Hosomichi et al. 2006). pCC1FOS Vector and Phage T-1 Resistant EPI300™-T1R To further examine variation in the MHC of the locally Escherichia coli Plating Strain (Epicentre). We end-repaired threatened species, the greater prairie chicken (Tympanuchus 15.0 μg of the DNA fragments using the Epicentre Fosmid cupido), we produced a physical map of the MHC-B region. kit according to the manufacturer's instructions, and 0.25 μg Previous work on this species (Eimes et al. 2011) had sug- of that DNA was ligated into 0.5 μg pCC1FOS vector DNA. gested variation in class II gene number between individuals, Transformed fosmids were packaged into MaxPlax Lambda and our efforts to amplify class I and other regions of the core Packaging Extract (Epicentre) and processed according to MHC failed, suggesting a different MHC-B genetic architec- the manufacturer's instructions followed by infection of ture than other Galliformes. The MHC-B assembly presented EPI300-T1R E. coli cells (Epicentre). Infected bacteria were here covers the region extending 60.9 kb from the BTN2 gene spread on 50 LB plates (22 cm) supplemented with 12.5 mg/ to the centromere protein gene CenpA. Our results show that mL chloramphenicol and incubated at 37 °C overnight. the genetic architecture of the MHC-B, while generally conserved in Galliformes, exhibits some important variations, Colony screening which is in part due to major inversions and deletions that have occurred between closely related species. In addition to a Colonies were harvested from each plate by flooding with physical map of the MHC in the greater prairie chicken, we 1.5 ml LBand gently scraping 700 μl of the mixture into investigated genetic variation and natural selection on the 1 ml microcentrifuge tubes supplemented with 20 % glycerol previously undescribed (in this species) class I peptide binding for immediate storage at −80 °C. The remaining mixture from gene. each plate was placed into a separate microcentrifuge tube for DNA extraction using the Epicentre FosmidMax Purification kit. Each extracted fosmid sub-library was then screened for Materials and methods MHC genes using the PCR primers Blex2F (Eimes et al. 2010) and E3R (Chaves et al. 2010), which together amplify a 279-bp Fosmid library construction fragment of the MHC class II gene encoding the β chain, and the primers pcTAP2F1 and pcTAP2R1 which amplify a 380- gDNA insert preparation bp fragment of the TAP2 antigen peptide transporter gene in the MHC class I (Table 1). The following PCR conditions were The library was constructed from DNA extracted from whole used: 50.0–100.0 ng fosmid DNA, 0.5 U of HotStar Taq blood from an adult female captured in Polk County, polymerase (Qiagen, USA), 4.0 μl of Q solution (Qiagen) Minnesota in 2008. High molecular weight genomic DNA and final concentrations of 0.5 μM of each primer, 200 μM was extracted using the Masterpure Complete DNA each dNTP and 1× of HotStar Taq PCR Buffer (Qiagen) and Purification Kit (Epicentre Biotechnologies, USA) and resus- water to a total volume of 25.0 μl. Cycling conditions were: pended in TE buffer at a concentration of 0.5 μg/μl. To 15minat95°Cinitialactivationfollowedby30cyclesof98° generate approximately 40 kb DNA fragments for insertion C for 20 s, 60 °C for 30 s, and 72 °C for 30 s. into the fosmid vector, 40 μg of purified gDNAwas randomly Positive sub-libraries were identified by sequencing the sheared by passing the sample through a 200-μl pipette tip 50 PCR amplicons and confirming homology to MHC genes in times. To confirm correct size of the insert, 1.0 μl of the the domestic turkey and chicken using BLAST. The resulting DNA was run on a 20-cm long, 0.8 % agarose gel corresponding glycerol stock (200 μl) from each positive Author's personal copy

Immunogenetics

Table 1 PCR primers and conditions used to amplify regions Locus/primer Primer sequence Product size (bp) Annealing temp (°C) Extension time of the prairie chicken MHC-B BTN2 F2 cacggctctgggtgctgtgg 2,850 58 4 m BLEC1 R3 cagaccgcttggcttggcga TAPBP F2 agtggggaggtggaggggga 4,250 65 6 m BRD2 R2 tgcagccaacgagaagcccg TAPBP 18 F ccagctgcgcttcccaacga 620 65 1 m TAPBP 18 R tgccgtcggtctcaccccaa IIB-2 15 F cacccggcccacccctatga 850 65 1 m IIB-2 15 R aggggacgccccatggactc DMB1Fa gaaggctgcagtgctcgcca 4,250 63 6 m DMB2Rab cgtcctctgcaacagccggg DMB2U tgggttgctgtggggcctct 2,450 60 3 m dmTAP2-Ra Tacgtgcccaccaagctgcg IAex2F gcggtacttctgcaccgcga 220 62 30 s IAex3R actgcctctggaactgccgc nTAP1-C4-F gggggtgggggcagtaggag 4,850 63 6 m nTAP1-C4-R cacagcgctcagcctgcact pcTAP1 F1 agggccatttttctcctcat 380 60 30 s pcTAP1 R1 tgattgctcagaaggtcac pcBG1F tggagcggcacagggtgagt 354 62 45 s pcBG1R gggctgcaaccaccccagtt pcBlec2F gacagagcaggcaggcagca 156 62 30 s pcBlec2r actgccctgagaccaacggga xTAPBpFalt ggcaggaacacggcgcagat 1,120 63 1.5 m xTAPBPR1 agtggtggttggcggcagtg pc1A ex3F gcttcaaccagagcggcggt 674 62 1 m pc1A ex3R acgatgggccgcgggtagaa

sub-library (one TAP1 and one class II) was diluted with LB E. coli sequences and quality trimming, the remaining reads and plated at a density of 200–300 colonies per plate for (4,357, average length 273 bp) were de novo-assembled and secondary screening. Colonies were harvested and DNA was aligned to the turkey MHC as a reference. The resulting extracted and screened by PCR as described above. A positive seven contigs were manually edited to resolve homopoly- sub-library for both class I and class II was identified and mer ambiguities resulting in a single sequence of 27.3 kb, diluted to 200–300 colonies on a single plate. Whole colony representing approximately 40× coverage (maximum depth PCR (using conditions described above) was performed until 119). The TAP1 positive fragment was sequenced using an single positive colonies for each MHC class were identified. Ilumina HiSeq 2000 at the Biomedical Genomics Center, Single positive colonies were cultured in 7 ml LB+chloram- University of Minnesota. Raw Illumina data (49,976,444 phenicol and 10.5 μl induction solution in a 14-ml culture tube reads representing a subset of a single flow cell) were and shaken overnight at 37 °C at 215 rpm. The fosmid DNA trimmed for vector and Illumina adaptors and process con- was extracted and prepared for sequencing using the Epicentre trol sequences and double filtered to remove fosmid vector FosmidMax Purification Kit. and E. coli sequences using BLAST and a custom Perl scripts. The resulting sequences (35,522,364 reads with an Sequencing and contig construction average length of 76 bp) were then assembled with VELVET (Zerbino and Birney 2008) using standard The MHC class II-positive fosmid was sequenced at defaults and two different k-mer lengths (57 and 63). The Research and Testing Laboratory, LLC (Lubbock, TX) us- assembly from the 63 k-mer resulted in 1,019 sequence ing a Roche 454 FLX Genome Sequencer. Because of the contigs ranging in size from 66 to 5,733 bp. These were relatively small size of the fosmid sequence, only a very aligned to the turkey whole genome sequence using BLAST small fraction of a full 454 plate was used for sequencing. to assign contigs to position. The VELVET contigs were The 33,559 raw 454 sequence reads (average length of assembled using Sequencher and positioned relative to each 513 bp) were initially assembled with Sequencher software other based on alignment to the turkey and chicken MHC (Gene Codes Inc, Ann Arbor, MI.) and contigs compared to assemblies resulting in a final assembly of 30 kb with nine the NCBI sequence databases by BLAST. After removal of gaps averaging 33 bp. Author's personal copy

Immunogenetics

Gaps and ambiguities in the assemblies were resolved by inferring haplotypes from genotype data in a population PCR using the primers and conditions in Table 1.Additional sample. PHASE assigned haplotypes using information long-range PCR was performed to confirm gene presence/ based on heterozygous genotypes, the cloned haplotypes absence and orientation for MHC regions in the prairie chick- and the 13 homozygous individuals that were entered into en that did not align with the turkey or chicken (Table 1). the program. dN/dS ratios (Codon Z test) were calculated gDNA from the same individual used to construct the fosmid using the Nei and Gojobori (Jukes-Cantor correction) meth- library was used to amplify the approximately 4.5 kb region od of overall means in Mega 5.10 (Nei and Gojobori 1986; separating the two fosmid fragments (DMB1 to DMB2)and Tamura et al. 2011). Estimates of nucleotide diversity were generate one contiguous MHC-B fragment of class I, II, and also calculated in Mega 5.10. In addition, an a priori test for III genes. PCR products that exceeded 1.5 kb were sequenced selection on individual codons was performed using the by primer walking. All PCR products were sequenced on program CODEML, found in the PAML 4.0 statistical soft- Applied Biosystems (Carlsbad, CA, USA) 3730XL and ware package (Yang 2007). The CODEML models used in 3130 automated DNA sequencers (Universityof Chicago this study were M7 and M8 which have been shown to Cancer Research Center). provide robust estimates for selection at highly polymorphic loci (Nielsen and Yang 1998; Bos and Waldman 2006; Gene identification, annotation, and sequence analyses Alcaide et al. 2008). In order to compare nucleotide diversity and diversi- Final sequences were aligned with the fully annotated se- fying selection of the class I PBR of prairie chickens quence of the turkey and chicken MHC for comparisons of with other avian species, single locus class I exon 3 predicted gene sequences, gene order and orientation and sequences were obtained from NCBI for the domestic expressed sequence tags (ESTs) using Sequencher and chicken (Acc No. AM282692-AM282700), the Eurasian Geneious version 5.4 (Biomatters, Aukland, NZ) (Figs. 1 kestrel (Falco tinnunculus; Acc No. EU120698– and 2). Gene annotations, CG content analysis, taxonomic EU120722), and the lesser kestrel (Falco naumanni; comparisons of gene nucleotide similarity (Table 2)and Acc. No. JF831087-JF831120). We chose these species sequence similarity dot matrix plots (Fig. 3)weremade because they are the only taxa for which an adequate using Geneious software. MHC nomenclature is consistent number of confirmed, single locus class I sequences were with that used for the turkey and chicken (Chaves et al. available. In this study, the PBR is study-defined as those 2009; Shiina et al. 2004a, b). codons identified from analysis of structural models of In order to estimate polymorphism and test for selection the peptide binding grooves in class I proteins of the at the single class I locus, IA, we sampled 30 unrelated domestic chicken (Bjorkman et al. 1987; Wallny et al. individuals from a locally threatened population (N0 2006; Koch et al. 2007; Promerova et al. 2009). ~1,200) in Portage County, Wisconsin (provenance data available in Johnson et al. (2003)). Individuals were geno- cDNA typed over the entire exon 3, which, along with exon 2, codes for the peptide binding region (PBR) in domestic Expression of class I and II genes was confirmed using chickens (Bjorkman et al. 1987; Wallny et al. 2006). A RNA extracted from the spleens of two adult female 674-bp fragment extending from exon 2 to exon 4 was prairie chickens from Minnesota found recently de- amplified using the PCR and Sanger sequencing conditions ceased in winter of 2009. Spleen tissue (5 mg) was described above and the primers pcIA ex3F and pcIA ex3R preserved in RNAlater (Qiagen, USA) and RNA was (Table 1). We aligned the 273-bp exon 3 sequences using extracted and DNase treated using the RNeasyMini Kit Geneious and identified heterozygous sites from chromato- (Qiagen), and cDNA was synthesized using the grams. For four individuals, some polymorphic sites could QuantiTect Reverse Transcription Kit (Qiagen) using not be unambiguously assigned, and these individuals were the manufacturer's protocol. The IA gene was amplified cloned, and haplotypes were identified by sequencing 12 using the primers IAex2F and IAex3R (Table 1), and colonies from each PCR insert. We cross-checked genotypes the class II B genes (IIB1 and IIB2) were amplified from cloned sequences with those from direct sequencing to using primers RNA F1 (Strand et al. 2007)andE3R eliminate false polymorphisms often associated with cloning (Chaves et al. 2010). The primers RNA F1 and E3R (Alcaide et al. 2011). We then used the program PHASE amplified cDNA to produce a 167-bp fragment that (Stephens et al. 2001) in the DnaSP v.5.0 software package extended from the 108th bp of exon 2 to the 85th bp (Librado and Rozas 2009) to identify specific haplotypes for of exon 3 (minus intron 2). Class II B PCR products each individual. PHASE employs Bayesian algorithms for were cloned using Clonesmart blunt-end high-count Author's personal copy

Immunogenetics

Fig. 1 An annotated map of the B-F/BL region of the MHC of the greater prairie chicken (JX971120). Black=entire genes; Gray=predicted coding DNA sequences; White=fosmid sequences; Black triangles=primer positions kanamycin cloning kits with chemically competent cells tated map (Fig. 1) are deposited online at Genbank (Lucigen, Middleton, WI, USA). Sequences and anno- (accession numbers pending).

Fig. 2 The comparative genomics of the core MHC-B in five species Although not to scale, the maps reflect the relative sizes and densities of Galliformes. The phylogeny is from Kriegs et al. (2007). (Tycu0 of the MHC-B in each species. Note the absence of a second class I Tympanuchus cupido; Chpi0Chrysolophus pictus; Mega0Meleagris peptide binding gene in the Tycu MHC. Maps (other than Tycu) are gallopavo; Gaga0Gallus gallus; Coja0Coturnix japonica). Arrows from Shiina et al. 2004a, b (Coja), Chaves et al. 2009 (Mega), Ye et al. indicate major gene inversions relative to the domestic chicken. 2012 (Chpi), and Jacob et al. 2000 (Gaga) Author's personal copy

Immunogenetics

Table 2 Comparison of MHC-B genes (coding DNA sequence, Gene CDS size Predicted No. Percent nucleotide Percent AA sequence CDS) between prairie chicken (bp) protein exons identity similarity and domestic chicken (Gallus) size (aa) and turkey Gallus Turkey Gallus Turkey

Blec 1 567 189 5 91.7 94.2 87.3 93.1 Class II IIB1 786 262 6 91.0 92.9 82.7 85.8 TAPBP 1,297 431 7 91.9 95.1 79.4 87.7 Class II IIB2 792 264 6 92.9 93.2 87.1 87.5 BRD2 2,343 781 11 91.5 97.6 98.9 99.0 DMA 792 264 4 90.0 93.1 87.9 90.5 DMB1 978 326 5 90.0 95.0 87.9 91.1 DMB2 777 259 6 90.6 95.6 75.2 91.4 TAP2 2,094 697 10 91.9 94.2 92.4 94.7 TAP1 1,856 586 11 92.7 95.1 92.9 96.3 Class I IA 1,080 360 8 85.7 87.7 74.5 77.1 C4 5,538 1,846 40 85.5 96.2 85.7 96.1

Results other Galliformes (Shiina et al. 2004a, b; Chaves et al. 2009; Ye et al. 2012). We tested for the presence of these genes in Gene number and homology other regions of the prairie chicken genome by designing PCR primers from conserved regions of the turkey and domestic The two fosmid clones and intervening sequence spanned chicken BG1 as well as Blec2 genes (Table 1). We success- 60.9 kb from exon 7 of the butyrophilin-like gene (BTN2)to fully amplified both BG1 and Blec2-like fragments in the exon 4 of the centromere protein A gene (CenpA) (Genbank greater prairie chicken outside the B-F/B-L MHC region. Accession number: JX971120), a region that is homologous to The Blec2 fragment was most similar to exon 3 and intron 3 both the domestic chicken (AP011531) and turkey of Blec2 in the turkey (92 % pairwise sequence identity). (DQ993255) (Fig. 1). The MHC B-F/B-L region of the prairie Given the duplicated nature of the BG genes in other chicken is more compact than the domestic chicken or turkey. Galliformes, homology of this amplified fragment is uncer- The region spanning the end of BTN2 to the end of C4 was tain. The amplified fragment aligned to exon 1 and intron 1 of 59,800 compared to 76,190 bp in chicken and 90,270 bp in several Galliformes BG genes but was most similar to BG1 of turkey (Fig. 2). The sequence contained just 14 genes (com- the domestic chicken (81.0 % pairwise sequence identity). pared to 17 and 20 in chicken and turkey, respectively). The 14 When the comparison was restricted to the turkey, the frag- predicted genes included a portion (1,487 bp) of BTN2,oneC- ment was most similar to BG3 (78 % pairwise sequence type lectin-like gene (Blec), two classical MHC class IIB (β identity). chain) loci, the TAP associated glycoprotein (TAPBP), the bromodomain-containing protein 2 gene (BRD2), three non- Loss of class I locus and inversion of TAP genes classical class II domains: α1(DMA), β1(DMB1)andIgC1 β2(DMB2), two class 1 transporter proteins (TAP1 and TAP2), Relative to the domestic chicken and turkey, the TAP1 and one classical MHC class I locus (1A; α1andα2 chains), a TAP2 gene block was inverted in the greater prairie chicken, complement protein (C4)andaportion(863bp)ofCenpA. and the second IA gene was absent (Fig. 2). Long-range PCR between DMB2 and TAP2 confirmed the absence of Rearrangement of BG and Blec2 genes the second IA gene in the prairie chicken core MHC-B, and it was also absent in the gDNA from 12 additional, unrelated All other Galliformes sequenced to date have at least one BG birds from Wisconsin. gene located between BTN2 and Blec1 (Shiina et al. 2004a, b; A single class IA gene, IA, was found between TAP1 and Chaves et al. 2009;Yeetal.2012), but this gene was not C4 (Fig. 2). Based on its location and expression, this gene present in this region of the prairie chicken MHC (Fig. 2). is probably homologous to BF2 in domestic chickens (IA2 Also absent from this region of the MHC-B of the prairie in turkeys). Thirty additional Wisconsin birds were tested chicken was the putative NK cell receptor Blec2 (NK2 in the for the presence of class IA genes outside the MHC-B region golden pheasant), which is conserved in the core MHC-B of by amplifying the polymorphic region of IA (Wallny et al. Author's personal copy

Immunogenetics

Fig. 3 Dot matrix analysis of the 67.2 kb MHC-B region of a DMA TAP1 CenpA BRD2 TAP2 1A2 C4 IIB1 TAPBP IIB2 DMB2 1A1 IIB3 DMB1 BG2 BG3 Blec2 Blec1 the prairie-chicken (Y axis) with BTN2 BG1 the corresponding 93.5 kb region of the turkey (X axis) (a) BTN2 and the 78.5 kb region of the Blec1 IIB1 domestic chicken (Gaga) (b). TAPBP Continuous main diagonal indicates high sequence simi- IIB2 larity. Parallel lines indicate duplications and lines perpen- BRD2 dicular to the diagonal indicate highly similar but inverted DMA regions DMB1 DMB2

TAP2

TAP1

CenpA

b CenpA IIB2 Blec1 BG1 Blec2 TAP1 TAP2 C4 BTN2 IIB1 TAPBP BRD2 DMA DMB1 IAI DMB2 IA2

BTN2 Blec1 IIB1

TAPBP

IIB2

BRD2

DMA DMB1 DMB2

TAP2

TAP1

CenpA

2006; Koch et al. 2007) with universal primers. None of from the four cloned individuals, provided sufficient power these individuals had more than two distinct sequences, for PHASE to identify haplotypes with near 100 % proba- which indicated only a single class IA locus in greater bility (Alcaide et al. 2011). A total of 14 unique sequences prairie chickens. Expression of a single IA locus was con- were found with an average nucleotide diversity of 0.042. firmed by PCR amplification of cDNA from spleen tissue Translation of the nucleotide sequences produced 14 unique (primers IAex2F and IAex3R). amino acid sequences (Fig. 4). At both the exon level and at

peptide binding codons only, dN/dS ratios were not signifi- Polymorphism and selection at the IA locus cantly different than one, indicating strong selection was not acting on the IA locus in prairie chickens. Among bird Of 30 individuals genotyped at exon 3 of the IA locus, 13 species for which single IA locus data are available, the were homozygotes, which, combined with the haplotypes prairie chicken was the only species without a significant Author's personal copy

Immunogenetics

Fig. 4 Alignment of the 14 amino acid sequences from class I exon 3 2007). Residues identified by CODEML as under positive selection found in 30 prairie chickens sampled. Included in the alignment are a are boxed. Genbank Accession numbers for prairie chicken are single turkey and domestic chicken sequence. Gray boxes indicate JX237356-JX237369, for turkey DQ993255, and for domestic chicken peptide binding codons identified in domestic chicken and black boxes AB268588 are putatively conserved residues (Wallny et al. 2006; Koch et al.

level of selection on this region based on dN/dS ratios chicken and turkey were more similar (mean094.2, range0 (Table 3). In the CODEML a priori test for selection, a 87.7–97.6) than those between prairie chicken and domestic maximum likelihood ratio test showed a better fit for model chicken (mean090.4, range085.5–92.7), consistent with the M8, than the neutral model, M7(P<0.001). The M8 model accepted phylogenetic relationships of the taxa (Pereira and did identify positive section at the IA locus; however, only Baker 2006). Amino acid residues were also more similar three codons were so identified: 4, 22, and 61 (>95 % poste- between the prairie chicken and turkey (mean090.9, range0 rior probability), all of which correspond to putative peptide 77.1–99.0) compared to the domestic chicken (mean086.0, binding codons identified in domestic chickens (Fig. 4). range074.5–98.9). Comparative identity matrix plots also indicated that the prairie chicken was more similar to the turkey Class IIB than the domestic chicken (Fig. 3).

Two class IIB loci were present in the prairie chicken core MHC-B, each flanking the TAPBP gene, an arrangement Discussion identical to the domestic chicken (Figs. 1 and 2). TAPBP was oriented in the same direction as in the domestic chicken A highly compact core MHC (the gene is in reverse orientation in the turkey) (Fig. 2). Based on the similarity of gene order between the prairie chicken and We assembled a 60.9-kb region of the core MHC-B of the the domestic chicken, the IIB1 gene is homologous to the greater prairie chicken and found 14 genes that were homolo- BLB1 and IIB2 is homologous to BLB2.Sequencingofthe gous to those in the domestic chicken and turkey. Genes of the PCR amplicons (exon 2 to 3: 167 bp; see Table 1)ofcDNA prairie chicken MHC-B were more similar to the turkey than the from spleen samples from two individuals revealed two domestic chicken at both the nucleotide and protein level, which unique sequences per individual indicating that at least one reflects the closer phylogenetic relationship of the prairie chick- locus was expressed. The IIB3 gene of the turkey is expressed en and turkey compared to the domestic chicken (Kriegs et al. while the IIB2 gene is not (Chaves et al. 2009), suggesting that 2007). The MHC of the prairie chicken appears to be smaller the IIB3 of turkeys and BLB2 of domestic chicken are homol- than the domestic chicken and turkey, in both the number of ogous. These results suggest that the prairie chicken IIB1 is coding DNA sequences and overall size. The core MHC-B has homologous to the turkey IIB1, and the IIB2 gene of the now been sequenced in five species of Galliformes, and a prairie chicken is putatively homologous to the IIB3 gene of surprising amount of variation in gene presence, order, number, the turkey. and orientation has been demonstrated. However, within prairie chickens, we found the smallest core MHC-B ofanybirdyet Gene similarity between species sequenced. Several loci appeared to be absent from the core MHC-B of Nucleotide and amino acid sequences for the fully sequenced the greater prairie chicken. One of these, Blec2, is a putative genes were compared between the three species of Galliformes NK cell receptor found in the MHC of other Galliformes and (Table 2). In all cases, the coding sequences between the prairie has been implicated in resistance to Marek's disease in Author's personal copy

Immunogenetics

Table 3 Polymorphism and tests for selection at exon 3 and the peptide binding codons (PBC) of the IA gene in four bird species

Species N Alleles π Ex 3 S Ex 3 π PBC S PBC dN/dS PBC P

Tympanuchus cupido 30 14 0.044 27 0.121 10 1.02 0.410 Gallus gallus 7 9 0.050 30 0.229 17 3.55 0.006 Falco tinnunculus 25 23 0.041 33 0.146 13 4.35 0.023 Falco naumanni 36 18 0.031 33 0.100 11 5.88 0.022

Additional information for Gallus is in Worley et al. (2008) and for Falco is in Alcaide et al. (2009)

N0individuals; π0nucleotide diversity; S0number of segregating sites; dN/dS0ratio of non-synonymous and synonymous substitutions; P0level of significance for Codon Z test of dN/dS ratio domestic chickens (Kelley and Trowsdale 2005; Rogers and Minimal selection on a single class I locus Kaufman 2008). We found a Blec2-like gene in the gDNA of the individual from which the fosmid library was constructed, The most notable difference between the MHC-B of prairie but not within the core MHC-B, indicating this Blec2-like chickens and other Galliformes was the absence of a second gene is located elsewhere in the MHC or genome. class IA locus in prairie chickens. The prairie chicken class IA Unexpectedly, there was no BG gene between the BTN2 aligns with the BF1 and BF2 genes in the domestic chicken and Blec1 genes in the prairie chicken MHC. The domestic and IA1 and IA2 genes in the turkey and golden pheasant. chicken and golden pheasant have at least one BG gene However, in the prairie chicken, there is no IA locus between located between BTN2 and Blec1, the turkey and Japanese DMB2 and the TAP genes and only a single IA gene is located quail have three in that location (Fig. 2) (Shiina et al. 2004a, between TAP1 and C4, which is the location of the class IA2 b; Bauer and Reed 2011; Ye et al. 2012). Our PCR results gene in the other Galliformes (Fig. 2). These differences in the from gDNA indicate that the prairie chicken has at least one prairie chicken may be associated with the inversion of the BG gene, but its location remains unknown. A possible TAP gene block accompanied by a deletion of the second IA explanation for the absence of the BG and Blec2 genes from locus (Fig. 2). The TAP gene block is also inverted in the the core MHC-B in the prairie chicken is a block inversion golden pheasant; however, unlike the prairie chicken, the of the BTN2-BG-Blec2 gene complex which would agree pheasant has retained the IA2 gene (Ye et al. 2012). Both class with our PCR and sequencing results placing BTN2 IA genes are expressed in the turkey (Reed et al. 2011); ~2,000 bp from Blec1. However, this scenario would require however, in domestic chickens, only the BF2 locus is highly an additional inversion of BTN2 since our PCR results expressed, and it appears to bind to a wider spectrum of require a reverse orientation of BTN2 that a single block antigenic peptides than any single class I gene in mammals inversion would have generated. (Wallny et al. 2006). Studies of other bird species have The results of this study show that even in closely related revealed a similar pattern of expression of a single class I species of birds, the genetic architecture of the core MHC locus among duplicated loci (Moon et al. 2005). can vary substantially. Although gene rearrangements, dele- Given the presence of a single IA locus, another interesting tions, and duplications are observed in the MHC of result of this study was the low level of detectable selection at Galliformes, the general size of the core MHC is relatively the IA locus. Several studies have tested for selection on class I stable within the order. Copy number variation is also ob- loci in birds. With the exception of the domestic chicken served, but it is limited to a few paralogs in both class I and (Worley et al. 2008) and some species in the genus Falco II loci. This contrasts considerably with the picture that is (Alcaide et al. 2009, 2010; Gangoso et al. 2012), these relied emerging in the Order Passeriformes. The only extensive map on sequences from several paralogs (duplicated loci within the of a passerine to date, that of the zebra finch (Taeniopygia same species). However, dN/dS ratios derived from such guttata), suggests that the MHC in this order is dramatically mixed locus samples may not be accurate, because differences expanded relative to Galliformes, with potentially dozens of in synonymous substitutions between loci may be larger than duplications of class I and II genes. Furthermore, in contrast to within loci, resulting in an underestimate of the true dN/dS Galliformes, the MHC in passerines may be scattered on ratio of a specific gene (Hughes and Nei 1989). Furthermore, several chromosomes (Balakrishnan et al. 2010,butsee it is common for only a single class I locus to be expressed in Ekblom et al. (2011). The extensive duplication of class I vertebrate taxa, regardless of the number of paralogs and genes in passerines also occurs in mammals and may be therefore only one locus is presumably under selection caused by the disruption of the tightly linked TAP-class I gene (Kaufman 1999; Flajnik and Kasahara 2001). Comparisons complex that occurs in the domestic chicken (Kaufman 1999; of dN/dS ratios between the minimally expressed chicken BF1 Balakrishnan et al. 2010). and the dominantly expressed BF2 loci support this idea Author's personal copy

Immunogenetics

(Worley et al. 2008). It is not surprising, then, that most studies Another possible explanation for our results is that drift testing for balancing selection at the MHC class I in birds have in this fragmented population has removed most evidence of produced dN/dS greater than one (when combining data from selection at this locus. The Wisconsin study population several loci), although none of them have shown selection at a experienced a bottleneck, resulting in a decline from ap- level of significance. However, using data from species for proximately 50,000 to 1,200 individuals over the last which sequences were available for an individual class I locus, 75 years (Johnson et al. 2003). We previously demonstrated we found a strong, significant signal of selection at the puta- a significant loss of MHC class II variation in this popula- tive PBR for the domestic chicken and two species of Falco tion (Eimes et al. 2011). Tests for positive selection at class (Table 3). Thus, the apparent absence of selection at the IA II loci were significant, but only in historic samples (1950; locus in the greater prairie chicken was unexpected and con- class I was not studied). Earlier studies of this population trasts with our previous study of class II loci in prairie chick- have found a similar loss of genetic variation at neutral ens (Eimes et al. 2011), where we found that the dN/dS ratio in markers (Bellinger et al. 2003; Johnson et al. 2003); how- the PBR (2.02) was significantly larger than one (codon-based ever, the loss of MHC class II variation exceeded that of Z test, Z02.01; P00.047). neutral genetic variation, possibly because of drift acting on Among vertebrates, conclusive evidence for a species with copy number variation (Eimes et al. 2011). Low genetic a single, classical MHC class I locus has only been found in the variation at both MHC and neutral genes combined with a amphibian genus Xenopus (Shum et al. 1993). Similar to our lack of detectable selection at either class I or II (possibly study, the dN/dS within the PBR of exon 3 (α2) in Xenopus was due to drift rather than purifying selection) indicates that the not significantly greater than one; however, dN was higher in “genetic health” of this population may have been reduced Xenopus (36±7) than in prairie chickens (26±1.0) (Flajnik et due to the population bottleneck. Studies of larger popula- al. 1999; Bos and Waldman 2006). Flajnik et al. (1999) argued tions of prairie chickens that have not experienced as severe that selection was acting on the class I locus in Xenopus despite a bottleneck are needed to assess whether drift has erased the low dN/dS because the dS across the entire gene had reached traces of diversifying selection on the IA locus in this pop- saturation due to the age of the allelic lineages and the higher ulation or if strong directional selection may have acted on mutation rate of frogs relative to most other vertebrates. These this locus throughout the range of the species. factors do not explain our results because MHC class I PBR allelic lineages in birds have not been shown to persist for very long periods, (Hess and Edwards 2002), mutation rates in birds are not unusually high and the dS in exon 3 in our sample was Acknowledgments Funding was provided by grants from the Re- relatively low (9.13±1.1). search Growth Initiative, University of Wisconsin-Milwaukee Gradu- ate School, and National Science Foundation (DEB-0948695) to POD, An alternative explanation for the lack of polymorphism JLB and LAW, American Ornithologists' Union, The American Muse- and detectable balancing selection at the IA locus is that strong um of Natural History, and Ruth Walker Research Award from the directional selection has acted on this locus and reduced University of Wisconsin-Milwaukee to JAE, and USDA-CSREES NRI variation at the PBR. Recently, Gangoso et al. (2012)reported Grants (#2005-01326 and 2009-35205-05302) to KMR. We thank J. Toepfer and R. Bellinger for tissue collection and processing, C. relatively low levels of class I polymorphism among several Wimpee for advice and assistance in the laboratory, and R. Settlage, species of Falco and argued that strong directional selection for assistance with bioinformatics. could account for the lower levels of class I variation found in some species relative to others within the same genus. They argued that strong selection from particular parasites might References reduce variation. Nonetheless, Gangoso et al. (2012)found significant evidence of diversifying selection at the putative PBR of a single class I locus (as defined by Alcaide et al. Alcaide M, Edwards SV, Negro JJ, Serrano D, Tella JL (2008) Exten- 2009, 2010), in contrast to our study. One interesting sive polymorphism and geographical variation at a positively selected MHC class IIB gene of the lesser kestrel (Falco nau- finding in our study is that all of the 14 nucleotide sequen- manni). Mol Ecol 17:2652–2665 ces obtained from 30 individuals coded for unique amino Alcaide M, Edwards SV, Cadahia L, Negro JJ (2009) MHC class I acid sequences, which may suggest selection for a wide genes of birds of prey: isolation, polymorphism and diversifying – variety of amino acid sequences. This scenario would selection. Conserv Genet 10:1349 1355 Alcaide M, Lemus JA, Blanco G, Tella JL, Serrano D, Negro JJ, seem to contradict the hypothesis that class I motifs in Rodriguez A, Garcia-Montijano M (2010) MHC diversity and birds are typically less variable than mammals because differential exposure to pathogens in kestrels (Aves: Falconidae). they bind a wider spectrum of peptides (Wallny et al. Mol Ecol 19:691–705 2006;Kochetal.2007) and suggests that, at least in Alcaide M, Rodriguez A, Negro JJ (2011) Sampling strategies for accurate computational inferences of gametic phase across highly greater prairie chickens, a wider diversity of different polymorphic major histocompatibility complex loci. BMC Res sequences increases immunocompetence. Notes 4:151 Author's personal copy

Immunogenetics

Balakrishnan CN, Ekblom R, Voelker M, Westerdahl H, Godinez R, Kaufman J (1999) Co-evolving genes in MHC haplotypes: the “rule” Kotkiewicz H, Burt DW, Graves T, Griffin DK, Warren WC, for nonmammalian vertebrates? Immunogenetics 50:228–236 Edwards SV (2010) Gene duplication and fragmentation in the Kaufman J, Venugopal K (1998) The importance of MHC for Rous zebra finch major histocompatibility complex. BMC Biol 8:29 sarcoma virus and Marek's disease virus—some Payne-ful con- Bauer MM, Reed KM (2011) Extended sequence of the turkey MHC siderations. Avian Pathol 27:S82–S87 B-locus and sequence variation in the highly polymorphic B-G Kaufman J, Volk H, Wallny HJ (1995) A “minimal essential Mhc” and loci. Immunogenetics 63:209–221 an “unrecognized Mhc”: two extremes in selection for polymor- Bellinger MR, Johnson JA, Toepfer J, Dunn PO (2003) Loss of genetic phism. Immunol Rev 143:63–88 variation in Greater Prairie Chickens following a population bot- Kaufman J, Milne S, Gobel TWF, Walker BA, Jacob JP, Auffray C, tleneck in Wisconsin, USA. Conserv Biol 17:717–724 Zoorob R, Beck S (1999) The chicken B locus is a minimal Bjorkman PJ, Saper MA, Samraoui B, Bennett WS, Strominger JL, essential major histocompatibility complex. Nature 401:923–925 Wiley DC (1987) Structure of the human class I histocompatibil- Kelley J, Trowsdale J (2005) Features of MHC and NK gene clusters. ity antigen, HLA-A2. Nature 329:506–512 Transplant Immunol 14:129–134 Bos DH, Waldman B (2006) Polymorphism, natural selection, and Kelley J, Walter L, Trowsdale J (2005) Comparative genomics of structural modeling of class Ia MHC in the African clawed frog major histocompatibility complexes. Immunogenetics 56:683– (Xenopus laevis). Immunogenetics 58:433–442 695 Chaves LD, Krueth SB, Reed KM (2009) Defining the turkey MHC: Koch M, Camp S, Collen T et al (2007) Structures of an MHC class I sequence and genes of the B locus. J Immunol 183:6530–6537 molecule from B21 chickens illustrate promiscuous peptide bind- Chaves LD, Faile GM, Krueth SB, Hendrickson JA, Reed KM (2010) ing. Immunity 27:885–899 Haplotype variation, recombination, and gene conversion within Kriegs JO, Matzke A, Churakov G, Kuritzin A, Mayr G, Brosius J, the turkey MHC-B locus. Immunogenetics 62:465–477 Scmitz J (2007) Waves of genomic hitchhikers shed light on the Eimes JA, Bollmer JL, Dunn PO, Whittingham LA, Wimpee C (2010) evolution of gamebirds (Aves: Galliformes). BMC Evol Biol Mhc class II diversity and balancing selection in greater prairie- 7:190 chickens. Genetica 138:265–271 Librado P, Rozas J (2009) DnaSP v5: a software for comprehensive Eimes JA, Bollmer JL, Whittingham LA, Johnson JA, Van Oosterhout analysis of DNA polymorphism data. Bioinformatics 25:1451– C, Dunn PO (2011) Rapid loss of MHC class II variation in a 1452 bottlenecked population is explained by drift and loss of copy Moon DA, Veniamin SM, Parks-Dely JA, Magor KE (2005) The MHC number variation. J Evol Biol 24:1847–1856 of the duck (Anas platyrhynchos) contains five differentially Ekblom R, Stapley J, Ball AD, Birkhead T, Burke T, Slate J (2011) expressed class I genes. J Immunol 175:6702–6712 Genetic mapping of the major histocompatibility complex in Nei M, Gojobori T (1986) Simple methods for estimating the numbers the zebra finch (Taeniopygia guttata). Immunogenetics 63:523–530 of synonymous and nonsynonymous nucleotide substitutions. Flajnik MF, Kasahara M (2001) Comparative genomics of the MHC: Mol Biol Evol 3:418–426 glimpses into the evolution of the adaptive immune system. Nielsen R, Yang ZH (1998) Likelihood models for detecting positively Immunity 15:351–362 selected amino acid sites and applications to the HIV-1 envelope Flajnik MF, Ohta Y, Greenberg AS, Salter-Cid L, Carrizosa A, Du gene. Genetics 148:929–936 Pasquier L, Kasahara M (1999) Two ancient allelic lineages at the Pereira SL, Baker AJ (2006) A molecular timescale for galliform birds single classical class I locus in the Xenopus MHC. J Immunol accounting for uncertainty in time estimates and heterogeneity 163:3826–3833 of rates of DNA substitutions across lineages and sites. Mol Gangoso L, Alcaide M, Grande JM, Munoz J, Talbot SL, Sonsthagens Phylogenet Evol 38:499–509 SA, Sage GK, Figuerola J (2012) Colonizing the world in spite of Promerova M, Albrecht T, Bryja J (2009) Extremely high MHC class I reduced MHC variation. J Evol Biol 25:1438–1447 variation in a population of a long-distance migrant, the Scarlet Guillemot F, Billault A, Pourquie O, Behar G, Chausse AM, Zoorob R, Rosefinch (Carpodacus erythrinus). Immunogenetics 61:451–461 Kreibich G, Auffray C (1988) A molecular map of the chicken Reed KM, Bauer MM, Monson MS, Benoit B, Chaves LD, O'Hare TH, major histocompatibility complex: the class II beta genes are Delany ME (2011) Defining the Turkey MHC: identification of closely linked to the class I genes and the nucleolar organizer. expressed class I- and class IIB-like genes independent of the EMBO J 7:2775–2785 MHC-B. Immunogenetics 63:753–771 Hess CM, Edwards SV (2002) The evolution of the major histocom- Rogers SL, Kaufman J (2008) High allelic polymorphism, moderate patibility complex in birds. Bioscience 52:423–431 sequence diversity and diversifying selection for B-NK but not B- Hosomichi K, Shiina T, Suzuki S, Tanaka M, Shimizu S, Iwamoto S, lec, the pair of lectin-like receptor genes in the chicken MHC. Hara H, Yoshida Y, Kulski JK, Inoko H, Hanzawa K (2006) The Immunogenetics 60:461–475 major histocompatibility complex (Mhc) class IIB region has Shiina T, Inoko H, Kulski JK (2004a) An update of the HLA genomic greater genomic structural flexibility and diversity in the quail region, locus information and disease associations: 2004. Tissue than the chicken. BMC Genomics 7:322 Antigens 64:631–649 Hughes AL, Nei M (1989) Evolution of the major histocompatibility Shiina T, Shimizu S, Hosomichi K, Kohara S, Watanabe S, Hanzawa complex—independent origin of nonclassical class-I genes in K, Beck S, Kulski JK, Inoko H (2004b) Comparative genomic different groups of mammals. Mol Biol Evol 6:559–579 analysis of two avian (quail and chicken) MHC regions. J Immu- Jacob JP, Milne S, Beck S, Kaufman J (2000) The major and a minor nol 172:6751–6763 class II beta-chain (B-LB) gene flank the Tapasin gene in the B-F/ Shum BP, Avila D, Dupasquier L, Kasahara M, Flajnik MF (1993) B-L region of the chicken major histocompatibility complex. Isolation of a classical Mhc class-I Cdna from an amphibian— Immunogenetics 51:138–147 evidence for only one class-I locus in the Xenopus Mhc. J Immu- Janeway CA, Travers P, Walport M, Schlomchik M (2005) The im- nol 151:5376–5386 mune system in health and disease. Garland Science Publishing, Stephens M, Smith NJ, Donnelly P (2001) A new statistical method for New York, NY; Abingdon, UK haplotype reconstruction from population data. Am J Hum Genet Johnson JA, Toepfer JE, Dunn PO (2003) Contrasting patterns of mito- 68:978–989 chondrial and microsatellite population structure in fragmented pop- Strand T, Westerdahl H, Hoeglund J, Alatalo RV, Siitari H (2007) The ulations of greater prairie-chickens. Mol Ecol 12:3335–3347 Mhc class II of the Black grouse (Tetrao tetrix) consists of low Author's personal copy

Immunogenetics

numbers of B and Y genes with variable diversity and expression. Worley K, Gillingham M, Jensen P, Kennedy LJ, Pizzari T, Kaufman J, Immunogenetics 59:725–734 Richardson DS (2008) Single locus typing of MHC class I and Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S (2011) class IIB loci in a population of red jungle fowl. Immunogenetics MEGA5: molecular evolutionary genetics analysis using maxi- 60:233–247 mum likelihood, evolutionary distance, and maximum parsimony Yang Z (2007) PAML 4: phylogenetic analysis by maximum likeli- methods. Mol Biol Evol 28:2731–2739 hood. Mol Biol Evol 24:1586–1591 Trowsdale J (1988) Molecular genetics of the MHC. Immunology 41:1–17 Ye Q, He K, Wu SY, Wan QH (2012) Isolation of a 97-kb minimal Wallny HJ, Avila D, Hunt LG et al (2006) Peptide motifs of the single essential MHC B locus from a new reverse-4D BAC library of the dominantly expressed class I molecule explain the striking MHC- golden pheasant. PLoS One 7:e32154 determined response to Rous sarcoma virus in chickens. Proc Natl Zerbino DR, Birney E (2008) Velvet: algorithms for de novo short read Acad Sci USA 103:1434–1439 assembly using de Bruijn graphs. Genome Res 18:821–829